2023-04-27 12:19:53,966 INFO [train.py:976] (7/8) Training started 2023-04-27 12:19:53,966 INFO [train.py:986] (7/8) Device: cuda:7 2023-04-27 12:19:53,968 INFO [train.py:995] (7/8) {'frame_shift_ms': 10.0, 'allowed_excess_duration_ratio': 0.1, 'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'a23383c5a381713b51e9014f3f05d096f8aceec3', 'k2-git-date': 'Wed Apr 26 15:33:33 2023', 'lhotse-version': '1.14.0.dev+git.b61b917.dirty', 'torch-version': '1.13.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.1', 'icefall-git-branch': 'master', 'icefall-git-sha1': '45c13e9-dirty', 'icefall-git-date': 'Mon Apr 24 15:00:02 2023', 'icefall-path': '/k2-dev/yangyifan/icefall-master', 'k2-path': '/k2-dev/yangyifan/anaconda3/envs/icefall/lib/python3.10/site-packages/k2-1.23.4.dev20230427+cuda11.6.torch1.13.1-py3.10-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/k2-dev/yangyifan/anaconda3/envs/icefall/lib/python3.10/site-packages/lhotse-1.14.0.dev0+git.b61b917.dirty-py3.10.egg/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-3-0423201227-84b4557756-8lx4n', 'IP address': '10.177.6.147'}, 'world_size': 8, 'master_port': 12354, 'tensorboard': True, 'num_epochs': 30, 'start_epoch': 1, 'start_batch': 0, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp_multidataset'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'base_lr': 0.05, 'lr_batches': 5000, 'lr_epochs': 3.5, 'context_size': 2, 'prune_range': 5, 'lm_scale': 0.25, 'am_scale': 0.0, 'simple_loss_scale': 0.5, 'seed': 42, 'print_diagnostics': False, 'inf_check': False, 'save_every_n': 2000, 'keep_last_k': 30, 'average_period': 200, 'use_fp16': True, 'use_multidataset': True, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, 'manifest_dir': PosixPath('data/fbank'), 'cv_manifest_dir': PosixPath('data/en/fbank'), 'max_duration': 700, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'blank_id': 0, 'vocab_size': 500} 2023-04-27 12:19:53,968 INFO [train.py:997] (7/8) About to create model 2023-04-27 12:19:54,631 INFO [zipformer.py:178] (7/8) At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8. 2023-04-27 12:19:54,647 INFO [train.py:1001] (7/8) Number of model parameters: 70369391 2023-04-27 12:19:57,252 INFO [train.py:1016] (7/8) Using DDP 2023-04-27 12:19:58,270 INFO [multidataset.py:46] (7/8) About to get multidataset train cuts 2023-04-27 12:19:58,270 INFO [multidataset.py:49] (7/8) Loading LibriSpeech in lazy mode 2023-04-27 12:19:58,294 INFO [multidataset.py:65] (7/8) Loading GigaSpeech 1998 splits in lazy mode 2023-04-27 12:20:00,752 INFO [multidataset.py:72] (7/8) Loading CommonVoice in lazy mode 2023-04-27 12:20:00,755 INFO [asr_datamodule.py:230] (7/8) Enable MUSAN 2023-04-27 12:20:00,755 INFO [asr_datamodule.py:231] (7/8) About to get Musan cuts 2023-04-27 12:20:02,935 INFO [asr_datamodule.py:255] (7/8) Enable SpecAugment 2023-04-27 12:20:02,935 INFO [asr_datamodule.py:256] (7/8) Time warp factor: 80 2023-04-27 12:20:02,935 INFO [asr_datamodule.py:266] (7/8) Num frame mask: 10 2023-04-27 12:20:02,935 INFO [asr_datamodule.py:279] (7/8) About to create train dataset 2023-04-27 12:20:02,936 INFO [asr_datamodule.py:306] (7/8) Using DynamicBucketingSampler. 2023-04-27 12:20:07,304 INFO [asr_datamodule.py:321] (7/8) About to create train dataloader 2023-04-27 12:20:07,305 INFO [asr_datamodule.py:435] (7/8) About to get dev-clean cuts 2023-04-27 12:20:07,306 INFO [asr_datamodule.py:442] (7/8) About to get dev-other cuts 2023-04-27 12:20:07,307 INFO [asr_datamodule.py:352] (7/8) About to create dev dataset 2023-04-27 12:20:07,542 INFO [asr_datamodule.py:369] (7/8) About to create dev dataloader 2023-04-27 12:20:25,627 INFO [train.py:904] (7/8) Epoch 1, batch 0, loss[loss=7.578, simple_loss=6.867, pruned_loss=7.095, over 17042.00 frames. ], tot_loss[loss=7.578, simple_loss=6.867, pruned_loss=7.095, over 17042.00 frames. ], batch size: 55, lr: 2.50e-02, grad_scale: 2.0 2023-04-27 12:20:25,627 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 12:20:32,882 INFO [train.py:938] (7/8) Epoch 1, validation: loss=6.911, simple_loss=6.238, pruned_loss=6.721, over 944034.00 frames. 2023-04-27 12:20:32,883 INFO [train.py:939] (7/8) Maximum memory allocated so far is 12383MB 2023-04-27 12:20:36,308 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:20:52,251 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:20:55,563 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=88.94 vs. limit=5.0 2023-04-27 12:21:03,486 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=8.11 vs. limit=2.0 2023-04-27 12:21:17,145 INFO [train.py:904] (7/8) Epoch 1, batch 50, loss[loss=1.213, simple_loss=1.076, pruned_loss=1.233, over 16926.00 frames. ], tot_loss[loss=2.17, simple_loss=1.966, pruned_loss=1.954, over 754588.93 frames. ], batch size: 116, lr: 2.75e-02, grad_scale: 2.0 2023-04-27 12:21:21,688 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=15.04 vs. limit=2.0 2023-04-27 12:21:34,874 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=12.93 vs. limit=2.0 2023-04-27 12:21:44,895 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5895, 5.5898, 5.5897, 5.5897, 5.5898, 5.5890, 5.5896, 5.5898], device='cuda:7'), covar=tensor([0.0020, 0.0012, 0.0029, 0.0018, 0.0019, 0.0011, 0.0026, 0.0033], device='cuda:7'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:7'), out_proj_covar=tensor([9.1740e-06, 9.2261e-06, 9.1070e-06, 9.0574e-06, 9.2376e-06, 8.9710e-06, 9.0338e-06, 9.1459e-06], device='cuda:7') 2023-04-27 12:21:46,512 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:21:58,211 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=94.82 vs. limit=5.0 2023-04-27 12:22:01,767 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=8.15 vs. limit=2.0 2023-04-27 12:22:02,804 WARNING [train.py:894] (7/8) Grad scale is small: 0.001953125 2023-04-27 12:22:02,805 INFO [train.py:904] (7/8) Epoch 1, batch 100, loss[loss=1.352, simple_loss=1.168, pruned_loss=1.475, over 11877.00 frames. ], tot_loss[loss=1.637, simple_loss=1.457, pruned_loss=1.611, over 1323156.03 frames. ], batch size: 246, lr: 3.00e-02, grad_scale: 0.00390625 2023-04-27 12:22:13,672 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 5.700e+01 2.326e+02 5.095e+02 1.135e+03 3.099e+06, threshold=1.019e+03, percent-clipped=0.0 2023-04-27 12:22:19,219 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=8.56 vs. limit=2.0 2023-04-27 12:22:19,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=63.41 vs. limit=5.0 2023-04-27 12:22:20,964 WARNING [optim.py:388] (7/8) Scaling gradients by 0.0112030990421772, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:21,103 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.24, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.974e+09, grad_sumsq = 5.165e+10, orig_rms_sq=3.822e-02 2023-04-27 12:22:27,955 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=18.22 vs. limit=2.0 2023-04-27 12:22:43,822 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:22:46,129 WARNING [optim.py:388] (7/8) Scaling gradients by 0.0022801109589636326, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:46,266 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.68, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.358e+11, grad_sumsq = 3.286e+12, orig_rms_sq=4.131e-02 2023-04-27 12:22:49,652 WARNING [optim.py:388] (7/8) Scaling gradients by 0.04246773198246956, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:49,791 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.92, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=5.309e+08, grad_sumsq = 1.285e+10, orig_rms_sq=4.131e-02 2023-04-27 12:22:51,352 WARNING [optim.py:388] (7/8) Scaling gradients by 0.000716241542249918, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:51,489 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.93, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.876e+12, grad_sumsq = 4.542e+13, orig_rms_sq=4.131e-02 2023-04-27 12:22:53,004 INFO [train.py:904] (7/8) Epoch 1, batch 150, loss[loss=1.009, simple_loss=0.8498, pruned_loss=1.141, over 17206.00 frames. ], tot_loss[loss=1.399, simple_loss=1.227, pruned_loss=1.446, over 1761679.04 frames. ], batch size: 44, lr: 3.25e-02, grad_scale: 0.00390625 2023-04-27 12:22:53,729 WARNING [optim.py:388] (7/8) Scaling gradients by 0.049951765686273575, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:53,832 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.92, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=3.845e+08, grad_sumsq = 8.968e+09, orig_rms_sq=4.287e-02 2023-04-27 12:22:58,547 WARNING [optim.py:388] (7/8) Scaling gradients by 0.00609818659722805, model_norm_threshold=1019.0284423828125 2023-04-27 12:22:58,674 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.49, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.379e+10, grad_sumsq = 3.140e+11, orig_rms_sq=4.392e-02 2023-04-27 12:23:02,516 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=4.52 vs. limit=2.0 2023-04-27 12:23:16,874 WARNING [optim.py:388] (7/8) Scaling gradients by 0.059935860335826874, model_norm_threshold=1019.0284423828125 2023-04-27 12:23:17,008 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.63, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=1.808e+08, grad_sumsq = 3.916e+09, orig_rms_sq=4.617e-02 2023-04-27 12:23:21,392 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8030, 3.8175, 3.8268, 3.8128, 3.7615, 3.7223, 3.7679, 3.8367], device='cuda:7'), covar=tensor([0.0219, 0.0142, 0.0153, 0.0068, 0.0181, 0.0397, 0.0355, 0.0208], device='cuda:7'), in_proj_covar=tensor([0.0010, 0.0010, 0.0010, 0.0010, 0.0010, 0.0010, 0.0010, 0.0010], device='cuda:7'), out_proj_covar=tensor([1.0350e-05, 1.0206e-05, 1.0338e-05, 1.0011e-05, 1.0099e-05, 1.0089e-05, 9.9285e-06, 1.0500e-05], device='cuda:7') 2023-04-27 12:23:28,343 WARNING [optim.py:388] (7/8) Scaling gradients by 0.060559310019016266, model_norm_threshold=1019.0284423828125 2023-04-27 12:23:28,455 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.90, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=2.556e+08, grad_sumsq = 6.221e+09, orig_rms_sq=4.108e-02 2023-04-27 12:23:42,818 WARNING [train.py:894] (7/8) Grad scale is small: 0.00390625 2023-04-27 12:23:42,819 INFO [train.py:904] (7/8) Epoch 1, batch 200, loss[loss=0.9795, simple_loss=0.8277, pruned_loss=1.016, over 16450.00 frames. ], tot_loss[loss=1.257, simple_loss=1.091, pruned_loss=1.315, over 2107237.79 frames. ], batch size: 75, lr: 3.50e-02, grad_scale: 0.0078125 2023-04-27 12:23:51,004 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 2.076e+02 2.707e+02 4.762e+02 1.423e+06, threshold=5.415e+02, percent-clipped=11.0 2023-04-27 12:23:51,004 WARNING [optim.py:388] (7/8) Scaling gradients by 0.002041660714894533, model_norm_threshold=541.4743041992188 2023-04-27 12:23:51,124 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.86, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=6.060e+10, grad_sumsq = 1.544e+12, orig_rms_sq=3.924e-02 2023-04-27 12:24:00,577 WARNING [optim.py:388] (7/8) Scaling gradients by 0.02974529005587101, model_norm_threshold=541.4743041992188 2023-04-27 12:24:00,708 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.85, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=2.829e+08, grad_sumsq = 7.259e+09, orig_rms_sq=3.897e-02 2023-04-27 12:24:01,479 WARNING [optim.py:388] (7/8) Scaling gradients by 0.01955481991171837, model_norm_threshold=541.4743041992188 2023-04-27 12:24:01,588 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.84, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=6.426e+08, grad_sumsq = 1.649e+10, orig_rms_sq=3.897e-02 2023-04-27 12:24:12,789 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=5.73 vs. limit=2.0 2023-04-27 12:24:29,720 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=12.19 vs. limit=2.0 2023-04-27 12:24:30,807 INFO [train.py:904] (7/8) Epoch 1, batch 250, loss[loss=0.9011, simple_loss=0.7617, pruned_loss=0.8792, over 16672.00 frames. ], tot_loss[loss=1.165, simple_loss=1.003, pruned_loss=1.211, over 2377693.50 frames. ], batch size: 134, lr: 3.75e-02, grad_scale: 0.0078125 2023-04-27 12:24:33,595 WARNING [optim.py:388] (7/8) Scaling gradients by 0.057925041764974594, model_norm_threshold=541.4743041992188 2023-04-27 12:24:33,704 INFO [optim.py:450] (7/8) Parameter Dominanting tot_sumsq module.encoder.encoder_embed.conv.0.weight with proportion 0.59, where dominant_sumsq=(grad_sumsq*orig_rms_sq)=5.145e+07, grad_sumsq = 1.327e+09, orig_rms_sq=3.876e-02 2023-04-27 12:24:40,405 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=5.54 vs. limit=2.0 2023-04-27 12:25:16,118 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:25:20,633 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:25:21,112 WARNING [train.py:894] (7/8) Grad scale is small: 0.0078125 2023-04-27 12:25:21,112 INFO [train.py:904] (7/8) Epoch 1, batch 300, loss[loss=1.014, simple_loss=0.8439, pruned_loss=0.9938, over 17056.00 frames. ], tot_loss[loss=1.096, simple_loss=0.9368, pruned_loss=1.125, over 2594288.96 frames. ], batch size: 50, lr: 4.00e-02, grad_scale: 0.015625 2023-04-27 12:25:30,080 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 9.163e+01 1.282e+02 1.731e+02 2.635e+02 2.769e+04, threshold=3.463e+02, percent-clipped=7.0 2023-04-27 12:25:58,290 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=3.45 vs. limit=2.0 2023-04-27 12:26:12,610 INFO [train.py:904] (7/8) Epoch 1, batch 350, loss[loss=0.8514, simple_loss=0.7058, pruned_loss=0.8047, over 16755.00 frames. ], tot_loss[loss=1.047, simple_loss=0.8896, pruned_loss=1.059, over 2758214.55 frames. ], batch size: 89, lr: 4.25e-02, grad_scale: 0.015625 2023-04-27 12:26:18,749 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=357.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:26:35,251 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=34.07 vs. limit=5.0 2023-04-27 12:26:52,619 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=387.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:27:06,006 INFO [train.py:904] (7/8) Epoch 1, batch 400, loss[loss=0.8783, simple_loss=0.72, pruned_loss=0.8217, over 16558.00 frames. ], tot_loss[loss=1.012, simple_loss=0.8539, pruned_loss=1.008, over 2875639.27 frames. ], batch size: 68, lr: 4.50e-02, grad_scale: 0.03125 2023-04-27 12:27:17,882 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 7.937e+01 1.110e+02 1.413e+02 1.837e+02 3.136e+02, threshold=2.826e+02, percent-clipped=0.0 2023-04-27 12:27:46,088 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=439.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:27:50,822 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5881, 4.4667, 4.8802, 5.4463, 5.6274, 5.0945, 5.1883, 4.6589], device='cuda:7'), covar=tensor([0.1019, 0.1940, 0.3458, 0.1132, 0.0408, 0.1106, 0.1344, 0.2512], device='cuda:7'), in_proj_covar=tensor([0.0010, 0.0010, 0.0011, 0.0010, 0.0010, 0.0010, 0.0010, 0.0010], device='cuda:7'), out_proj_covar=tensor([9.5390e-06, 9.7773e-06, 9.9844e-06, 9.7799e-06, 9.3359e-06, 9.8582e-06, 9.8515e-06, 9.7698e-06], device='cuda:7') 2023-04-27 12:27:56,261 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=448.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:27:59,545 INFO [train.py:904] (7/8) Epoch 1, batch 450, loss[loss=0.8402, simple_loss=0.6905, pruned_loss=0.7512, over 16218.00 frames. ], tot_loss[loss=0.9858, simple_loss=0.8255, pruned_loss=0.9654, over 2979094.60 frames. ], batch size: 165, lr: 4.75e-02, grad_scale: 0.03125 2023-04-27 12:28:51,194 INFO [train.py:904] (7/8) Epoch 1, batch 500, loss[loss=0.8765, simple_loss=0.7102, pruned_loss=0.7843, over 16839.00 frames. ], tot_loss[loss=0.9651, simple_loss=0.8024, pruned_loss=0.9295, over 3052368.15 frames. ], batch size: 102, lr: 4.99e-02, grad_scale: 0.0625 2023-04-27 12:28:57,933 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=8.53 vs. limit=2.0 2023-04-27 12:29:01,371 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 7.996e+01 1.277e+02 1.626e+02 2.121e+02 5.144e+02, threshold=3.251e+02, percent-clipped=17.0 2023-04-27 12:29:17,577 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=2.08 vs. limit=2.0 2023-04-27 12:29:44,936 INFO [train.py:904] (7/8) Epoch 1, batch 550, loss[loss=0.8194, simple_loss=0.6679, pruned_loss=0.6996, over 12333.00 frames. ], tot_loss[loss=0.9516, simple_loss=0.7861, pruned_loss=0.9001, over 3103952.10 frames. ], batch size: 246, lr: 4.98e-02, grad_scale: 0.0625 2023-04-27 12:29:58,134 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=562.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:30:03,840 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=568.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:30:17,246 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-04-27 12:30:27,489 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=590.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:30:38,232 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=600.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:30:38,413 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.29 vs. limit=2.0 2023-04-27 12:30:38,748 INFO [train.py:904] (7/8) Epoch 1, batch 600, loss[loss=0.9713, simple_loss=0.781, pruned_loss=0.8305, over 17187.00 frames. ], tot_loss[loss=0.9397, simple_loss=0.7717, pruned_loss=0.872, over 3158721.63 frames. ], batch size: 46, lr: 4.98e-02, grad_scale: 0.125 2023-04-27 12:30:48,394 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 8.862e+01 1.447e+02 1.955e+02 2.870e+02 6.309e+02, threshold=3.911e+02, percent-clipped=20.0 2023-04-27 12:31:02,570 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=623.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:31:03,559 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=2.30 vs. limit=2.0 2023-04-27 12:31:07,541 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=629.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:31:18,198 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.84 vs. limit=2.0 2023-04-27 12:31:23,189 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=2.62 vs. limit=2.0 2023-04-27 12:31:28,001 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=648.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:31:31,006 INFO [train.py:904] (7/8) Epoch 1, batch 650, loss[loss=0.9396, simple_loss=0.7548, pruned_loss=0.783, over 17149.00 frames. ], tot_loss[loss=0.9274, simple_loss=0.7583, pruned_loss=0.8428, over 3187038.76 frames. ], batch size: 49, lr: 4.98e-02, grad_scale: 0.125 2023-04-27 12:31:31,373 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=651.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:31:32,151 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=652.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:31:46,555 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=12.35 vs. limit=5.0 2023-04-27 12:32:05,612 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=3.10 vs. limit=2.0 2023-04-27 12:32:14,696 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=6.43 vs. limit=5.0 2023-04-27 12:32:22,404 INFO [train.py:904] (7/8) Epoch 1, batch 700, loss[loss=0.8615, simple_loss=0.6961, pruned_loss=0.6918, over 16816.00 frames. ], tot_loss[loss=0.9189, simple_loss=0.7501, pruned_loss=0.8138, over 3226138.89 frames. ], batch size: 39, lr: 4.98e-02, grad_scale: 0.25 2023-04-27 12:32:31,777 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.097e+02 2.100e+02 3.098e+02 3.988e+02 9.500e+02, threshold=6.196e+02, percent-clipped=26.0 2023-04-27 12:32:49,667 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=7.92 vs. limit=5.0 2023-04-27 12:32:53,122 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=6.61 vs. limit=2.0 2023-04-27 12:32:57,168 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=5.86 vs. limit=2.0 2023-04-27 12:33:00,891 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=739.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:33:01,107 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=2.73 vs. limit=2.0 2023-04-27 12:33:05,975 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=743.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:33:14,015 INFO [train.py:904] (7/8) Epoch 1, batch 750, loss[loss=0.7733, simple_loss=0.6385, pruned_loss=0.5815, over 16661.00 frames. ], tot_loss[loss=0.9012, simple_loss=0.7369, pruned_loss=0.7747, over 3257428.09 frames. ], batch size: 134, lr: 4.97e-02, grad_scale: 0.25 2023-04-27 12:33:14,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-04-27 12:33:51,950 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=787.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:34:06,269 INFO [train.py:904] (7/8) Epoch 1, batch 800, loss[loss=0.8119, simple_loss=0.6717, pruned_loss=0.5965, over 12555.00 frames. ], tot_loss[loss=0.8743, simple_loss=0.7181, pruned_loss=0.7282, over 3273184.01 frames. ], batch size: 246, lr: 4.97e-02, grad_scale: 0.5 2023-04-27 12:34:17,534 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.716e+02 2.855e+02 4.131e+02 5.648e+02 8.889e+02, threshold=8.261e+02, percent-clipped=19.0 2023-04-27 12:34:54,204 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=847.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:34:58,593 INFO [train.py:904] (7/8) Epoch 1, batch 850, loss[loss=0.6786, simple_loss=0.5801, pruned_loss=0.4595, over 16876.00 frames. ], tot_loss[loss=0.841, simple_loss=0.6957, pruned_loss=0.6766, over 3287251.61 frames. ], batch size: 109, lr: 4.96e-02, grad_scale: 0.5 2023-04-27 12:35:17,073 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5850, 3.5687, 3.5218, 3.5876, 3.3617, 3.6121, 3.6075, 3.6734], device='cuda:7'), covar=tensor([0.6407, 0.5354, 0.5665, 0.6226, 0.6021, 0.5369, 0.3818, 0.6250], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0042, 0.0049, 0.0045, 0.0046, 0.0050, 0.0037, 0.0049], device='cuda:7'), out_proj_covar=tensor([4.0025e-05, 3.8420e-05, 4.2232e-05, 4.0703e-05, 4.3209e-05, 4.2262e-05, 3.4836e-05, 4.8189e-05], device='cuda:7') 2023-04-27 12:35:26,994 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9997, 2.7080, 3.0611, 2.9177, 2.9878, 2.9045, 2.8416, 2.6822], device='cuda:7'), covar=tensor([0.3724, 0.6313, 0.4691, 0.5266, 0.3808, 0.6722, 0.7016, 0.6438], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0098, 0.0099, 0.0099, 0.0085, 0.0091, 0.0102, 0.0106], device='cuda:7'), out_proj_covar=tensor([6.7197e-05, 7.0264e-05, 7.4310e-05, 7.2394e-05, 6.2240e-05, 7.3851e-05, 7.9888e-05, 8.9349e-05], device='cuda:7') 2023-04-27 12:35:50,498 INFO [train.py:904] (7/8) Epoch 1, batch 900, loss[loss=0.7221, simple_loss=0.62, pruned_loss=0.4783, over 16641.00 frames. ], tot_loss[loss=0.8043, simple_loss=0.6708, pruned_loss=0.6251, over 3292866.38 frames. ], batch size: 62, lr: 4.96e-02, grad_scale: 1.0 2023-04-27 12:35:57,674 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=908.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:36:00,425 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.021e+02 3.478e+02 3.989e+02 5.953e+02 1.563e+03, threshold=7.979e+02, percent-clipped=10.0 2023-04-27 12:36:08,850 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=918.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:36:14,422 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=924.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:36:37,858 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=946.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:36:42,351 INFO [train.py:904] (7/8) Epoch 1, batch 950, loss[loss=0.6727, simple_loss=0.594, pruned_loss=0.4165, over 16653.00 frames. ], tot_loss[loss=0.7733, simple_loss=0.6507, pruned_loss=0.5802, over 3300727.73 frames. ], batch size: 57, lr: 4.96e-02, grad_scale: 1.0 2023-04-27 12:36:44,170 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=952.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:36:49,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7245, 3.3877, 4.1116, 3.7523, 3.1465, 3.9737, 3.8252, 3.9885], device='cuda:7'), covar=tensor([0.4304, 0.4363, 0.3194, 0.4620, 0.9343, 0.2912, 0.4623, 0.2595], device='cuda:7'), in_proj_covar=tensor([0.0041, 0.0042, 0.0037, 0.0036, 0.0035, 0.0034, 0.0043, 0.0035], device='cuda:7'), out_proj_covar=tensor([3.4954e-05, 3.1741e-05, 3.0293e-05, 3.1070e-05, 3.1720e-05, 2.9764e-05, 3.7869e-05, 3.0113e-05], device='cuda:7') 2023-04-27 12:37:35,474 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1000.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:37:36,005 INFO [train.py:904] (7/8) Epoch 1, batch 1000, loss[loss=0.5902, simple_loss=0.5192, pruned_loss=0.365, over 15497.00 frames. ], tot_loss[loss=0.7407, simple_loss=0.6293, pruned_loss=0.5366, over 3308952.13 frames. ], batch size: 190, lr: 4.95e-02, grad_scale: 1.0 2023-04-27 12:37:46,291 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.400e+02 3.913e+02 4.932e+02 6.154e+02 1.349e+03, threshold=9.864e+02, percent-clipped=6.0 2023-04-27 12:38:20,086 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7759, 3.6746, 3.7925, 3.6686, 3.9808, 3.8173, 3.7558, 3.4502], device='cuda:7'), covar=tensor([0.3091, 0.3985, 0.3372, 0.3216, 0.1902, 0.3036, 0.3726, 0.3622], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0126, 0.0124, 0.0126, 0.0104, 0.0117, 0.0131, 0.0133], device='cuda:7'), out_proj_covar=tensor([8.5095e-05, 9.1558e-05, 9.3629e-05, 9.2339e-05, 7.5093e-05, 9.2981e-05, 1.0138e-04, 1.0993e-04], device='cuda:7') 2023-04-27 12:38:22,304 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1043.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:38:29,484 INFO [train.py:904] (7/8) Epoch 1, batch 1050, loss[loss=0.5692, simple_loss=0.5136, pruned_loss=0.3325, over 16834.00 frames. ], tot_loss[loss=0.7112, simple_loss=0.6103, pruned_loss=0.4979, over 3317503.96 frames. ], batch size: 42, lr: 4.95e-02, grad_scale: 1.0 2023-04-27 12:38:40,920 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.03 vs. limit=5.0 2023-04-27 12:39:12,772 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1091.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:39:21,807 INFO [train.py:904] (7/8) Epoch 1, batch 1100, loss[loss=0.5514, simple_loss=0.5072, pruned_loss=0.3087, over 17007.00 frames. ], tot_loss[loss=0.6816, simple_loss=0.5906, pruned_loss=0.4622, over 3319867.15 frames. ], batch size: 41, lr: 4.94e-02, grad_scale: 1.0 2023-04-27 12:39:33,205 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.538e+02 4.173e+02 5.189e+02 6.744e+02 1.137e+03, threshold=1.038e+03, percent-clipped=1.0 2023-04-27 12:40:17,538 INFO [train.py:904] (7/8) Epoch 1, batch 1150, loss[loss=0.5457, simple_loss=0.4926, pruned_loss=0.3153, over 16879.00 frames. ], tot_loss[loss=0.656, simple_loss=0.5745, pruned_loss=0.4304, over 3324900.56 frames. ], batch size: 116, lr: 4.94e-02, grad_scale: 1.0 2023-04-27 12:40:19,493 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1153.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 12:41:10,410 INFO [train.py:904] (7/8) Epoch 1, batch 1200, loss[loss=0.6013, simple_loss=0.5424, pruned_loss=0.3463, over 17253.00 frames. ], tot_loss[loss=0.6325, simple_loss=0.5597, pruned_loss=0.4024, over 3331469.52 frames. ], batch size: 52, lr: 4.93e-02, grad_scale: 2.0 2023-04-27 12:41:12,655 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1203.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 12:41:21,564 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.465e+02 4.824e+02 5.887e+02 7.169e+02 1.569e+03, threshold=1.177e+03, percent-clipped=1.0 2023-04-27 12:41:25,840 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1214.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:41:30,060 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1218.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:41:34,836 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1224.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 12:41:57,802 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1246.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 12:42:03,022 INFO [train.py:904] (7/8) Epoch 1, batch 1250, loss[loss=0.5356, simple_loss=0.5045, pruned_loss=0.2848, over 17115.00 frames. ], tot_loss[loss=0.6121, simple_loss=0.5464, pruned_loss=0.3793, over 3328139.40 frames. ], batch size: 47, lr: 4.92e-02, grad_scale: 2.0 2023-04-27 12:42:14,345 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2023-04-27 12:42:17,893 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1266.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:42:25,410 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1272.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:42:49,930 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1294.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:42:56,590 INFO [train.py:904] (7/8) Epoch 1, batch 1300, loss[loss=0.5308, simple_loss=0.4862, pruned_loss=0.2957, over 11850.00 frames. ], tot_loss[loss=0.5939, simple_loss=0.5347, pruned_loss=0.3588, over 3328361.84 frames. ], batch size: 246, lr: 4.92e-02, grad_scale: 2.0 2023-04-27 12:43:07,782 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.547e+02 4.841e+02 5.972e+02 7.377e+02 1.990e+03, threshold=1.194e+03, percent-clipped=4.0 2023-04-27 12:43:44,357 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8726, 4.0577, 3.4604, 3.9245, 3.6546, 3.9327, 3.5751, 3.7009], device='cuda:7'), covar=tensor([0.3017, 0.2690, 0.2897, 0.2610, 0.2972, 0.2403, 0.3071, 0.2538], device='cuda:7'), in_proj_covar=tensor([0.0095, 0.0094, 0.0106, 0.0087, 0.0099, 0.0095, 0.0099, 0.0095], device='cuda:7'), out_proj_covar=tensor([8.5570e-05, 7.8651e-05, 8.8020e-05, 7.5836e-05, 8.7755e-05, 8.0817e-05, 8.5623e-05, 8.7015e-05], device='cuda:7') 2023-04-27 12:43:51,775 INFO [train.py:904] (7/8) Epoch 1, batch 1350, loss[loss=0.5365, simple_loss=0.4955, pruned_loss=0.2943, over 16715.00 frames. ], tot_loss[loss=0.5769, simple_loss=0.5246, pruned_loss=0.3399, over 3324791.07 frames. ], batch size: 124, lr: 4.91e-02, grad_scale: 2.0 2023-04-27 12:44:49,667 INFO [train.py:904] (7/8) Epoch 1, batch 1400, loss[loss=0.5126, simple_loss=0.4761, pruned_loss=0.2783, over 15579.00 frames. ], tot_loss[loss=0.5621, simple_loss=0.5147, pruned_loss=0.3248, over 3316611.47 frames. ], batch size: 190, lr: 4.91e-02, grad_scale: 2.0 2023-04-27 12:45:00,105 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.619e+02 5.087e+02 6.130e+02 7.329e+02 1.559e+03, threshold=1.226e+03, percent-clipped=4.0 2023-04-27 12:45:27,779 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1434.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:45:46,443 INFO [train.py:904] (7/8) Epoch 1, batch 1450, loss[loss=0.4916, simple_loss=0.4642, pruned_loss=0.2601, over 16344.00 frames. ], tot_loss[loss=0.547, simple_loss=0.5051, pruned_loss=0.3097, over 3312960.32 frames. ], batch size: 36, lr: 4.90e-02, grad_scale: 2.0 2023-04-27 12:46:34,553 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1495.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:46:41,353 INFO [train.py:904] (7/8) Epoch 1, batch 1500, loss[loss=0.4597, simple_loss=0.4593, pruned_loss=0.2228, over 17123.00 frames. ], tot_loss[loss=0.5336, simple_loss=0.4965, pruned_loss=0.2968, over 3314776.47 frames. ], batch size: 48, lr: 4.89e-02, grad_scale: 2.0 2023-04-27 12:46:44,233 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1503.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 12:46:51,046 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1509.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:46:52,710 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.422e+02 5.181e+02 6.478e+02 8.944e+02 1.260e+03, threshold=1.296e+03, percent-clipped=1.0 2023-04-27 12:47:38,774 INFO [train.py:904] (7/8) Epoch 1, batch 1550, loss[loss=0.515, simple_loss=0.4834, pruned_loss=0.2746, over 16719.00 frames. ], tot_loss[loss=0.5257, simple_loss=0.492, pruned_loss=0.2884, over 3314200.17 frames. ], batch size: 134, lr: 4.89e-02, grad_scale: 2.0 2023-04-27 12:47:39,058 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1551.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:47:56,138 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1566.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:48:29,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5598, 5.5223, 5.3029, 5.4183, 5.2710, 5.6963, 5.6832, 5.2007], device='cuda:7'), covar=tensor([0.0485, 0.0858, 0.1072, 0.1038, 0.1315, 0.0618, 0.0630, 0.1428], device='cuda:7'), in_proj_covar=tensor([0.0096, 0.0112, 0.0110, 0.0120, 0.0121, 0.0095, 0.0092, 0.0120], device='cuda:7'), out_proj_covar=tensor([8.3632e-05, 1.0101e-04, 9.3788e-05, 1.0363e-04, 1.1247e-04, 8.3090e-05, 8.2716e-05, 1.0897e-04], device='cuda:7') 2023-04-27 12:48:35,348 INFO [train.py:904] (7/8) Epoch 1, batch 1600, loss[loss=0.5566, simple_loss=0.5307, pruned_loss=0.2905, over 16719.00 frames. ], tot_loss[loss=0.5211, simple_loss=0.49, pruned_loss=0.2827, over 3305097.14 frames. ], batch size: 62, lr: 4.88e-02, grad_scale: 4.0 2023-04-27 12:48:37,299 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0310, 5.0346, 5.0978, 5.2276, 4.7920, 4.8875, 4.6558, 4.6290], device='cuda:7'), covar=tensor([0.0584, 0.0415, 0.0496, 0.0361, 0.1158, 0.0774, 0.0799, 0.1421], device='cuda:7'), in_proj_covar=tensor([0.0032, 0.0033, 0.0034, 0.0026, 0.0029, 0.0032, 0.0027, 0.0027], device='cuda:7'), out_proj_covar=tensor([2.3310e-05, 2.3969e-05, 2.5520e-05, 1.9927e-05, 2.2750e-05, 2.3390e-05, 2.1559e-05, 2.1932e-05], device='cuda:7') 2023-04-27 12:48:43,883 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4694, 3.8468, 3.3090, 3.4583, 3.0413, 3.4008, 3.3692, 3.8169], device='cuda:7'), covar=tensor([0.0997, 0.0600, 0.0911, 0.0815, 0.1125, 0.1065, 0.2141, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0072, 0.0071, 0.0079, 0.0082, 0.0075, 0.0079, 0.0081, 0.0070], device='cuda:7'), out_proj_covar=tensor([6.7641e-05, 6.9132e-05, 6.9949e-05, 7.0538e-05, 6.5655e-05, 6.9952e-05, 6.9531e-05, 6.1992e-05], device='cuda:7') 2023-04-27 12:48:47,170 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.141e+02 5.212e+02 6.871e+02 8.557e+02 2.137e+03, threshold=1.374e+03, percent-clipped=9.0 2023-04-27 12:48:47,534 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1611.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:49:02,756 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-04-27 12:49:05,726 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1627.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 12:49:32,291 INFO [train.py:904] (7/8) Epoch 1, batch 1650, loss[loss=0.6618, simple_loss=0.6009, pruned_loss=0.3665, over 12192.00 frames. ], tot_loss[loss=0.5145, simple_loss=0.4871, pruned_loss=0.2755, over 3309648.39 frames. ], batch size: 247, lr: 4.87e-02, grad_scale: 4.0 2023-04-27 12:49:57,787 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1672.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:50:29,827 INFO [train.py:904] (7/8) Epoch 1, batch 1700, loss[loss=0.496, simple_loss=0.48, pruned_loss=0.2543, over 16474.00 frames. ], tot_loss[loss=0.5056, simple_loss=0.483, pruned_loss=0.2669, over 3321091.38 frames. ], batch size: 146, lr: 4.86e-02, grad_scale: 4.0 2023-04-27 12:50:41,829 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.189e+02 5.218e+02 6.714e+02 8.049e+02 1.427e+03, threshold=1.343e+03, percent-clipped=1.0 2023-04-27 12:51:28,773 INFO [train.py:904] (7/8) Epoch 1, batch 1750, loss[loss=0.4889, simple_loss=0.5006, pruned_loss=0.2336, over 16658.00 frames. ], tot_loss[loss=0.4975, simple_loss=0.4792, pruned_loss=0.2593, over 3328748.14 frames. ], batch size: 57, lr: 4.86e-02, grad_scale: 4.0 2023-04-27 12:52:15,509 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1790.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 12:52:27,935 INFO [train.py:904] (7/8) Epoch 1, batch 1800, loss[loss=0.4892, simple_loss=0.4826, pruned_loss=0.2459, over 16436.00 frames. ], tot_loss[loss=0.492, simple_loss=0.4769, pruned_loss=0.2541, over 3325879.12 frames. ], batch size: 68, lr: 4.85e-02, grad_scale: 4.0 2023-04-27 12:52:37,421 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1809.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:52:38,894 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.321e+02 5.485e+02 6.556e+02 7.752e+02 2.000e+03, threshold=1.311e+03, percent-clipped=5.0 2023-04-27 12:52:42,681 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7534, 4.8024, 4.5043, 4.6296, 4.5806, 4.8791, 4.8413, 4.5197], device='cuda:7'), covar=tensor([0.0450, 0.0649, 0.0647, 0.0853, 0.0912, 0.0539, 0.0502, 0.1053], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0128, 0.0122, 0.0134, 0.0138, 0.0110, 0.0098, 0.0132], device='cuda:7'), out_proj_covar=tensor([9.5630e-05, 1.1783e-04, 1.0569e-04, 1.1816e-04, 1.3139e-04, 9.8962e-05, 8.8126e-05, 1.2564e-04], device='cuda:7') 2023-04-27 12:53:10,968 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1838.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:53:25,098 INFO [train.py:904] (7/8) Epoch 1, batch 1850, loss[loss=0.4705, simple_loss=0.4591, pruned_loss=0.2399, over 16789.00 frames. ], tot_loss[loss=0.4859, simple_loss=0.4742, pruned_loss=0.2488, over 3327929.34 frames. ], batch size: 124, lr: 4.84e-02, grad_scale: 4.0 2023-04-27 12:53:32,839 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=1857.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:53:32,940 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1857.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:53:33,118 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-04-27 12:53:48,640 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1870.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 12:54:22,051 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1899.0, num_to_drop=2, layers_to_drop={2, 3} 2023-04-27 12:54:23,740 INFO [train.py:904] (7/8) Epoch 1, batch 1900, loss[loss=0.4699, simple_loss=0.454, pruned_loss=0.2425, over 16672.00 frames. ], tot_loss[loss=0.4734, simple_loss=0.4671, pruned_loss=0.2393, over 3322688.16 frames. ], batch size: 134, lr: 4.83e-02, grad_scale: 4.0 2023-04-27 12:54:36,245 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.249e+02 5.423e+02 6.685e+02 8.506e+02 1.861e+03, threshold=1.337e+03, percent-clipped=8.0 2023-04-27 12:54:44,501 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1918.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 12:54:49,133 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1922.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 12:55:00,056 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1931.0, num_to_drop=2, layers_to_drop={1, 3} 2023-04-27 12:55:22,819 INFO [train.py:904] (7/8) Epoch 1, batch 1950, loss[loss=0.4877, simple_loss=0.4856, pruned_loss=0.2443, over 15503.00 frames. ], tot_loss[loss=0.4635, simple_loss=0.4623, pruned_loss=0.2316, over 3323758.30 frames. ], batch size: 191, lr: 4.83e-02, grad_scale: 4.0 2023-04-27 12:55:42,131 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1967.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:56:23,578 INFO [train.py:904] (7/8) Epoch 1, batch 2000, loss[loss=0.3462, simple_loss=0.3876, pruned_loss=0.1524, over 16023.00 frames. ], tot_loss[loss=0.4575, simple_loss=0.4588, pruned_loss=0.2274, over 3311683.82 frames. ], batch size: 35, lr: 4.82e-02, grad_scale: 8.0 2023-04-27 12:56:36,806 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.491e+02 5.646e+02 7.242e+02 8.659e+02 1.834e+03, threshold=1.448e+03, percent-clipped=2.0 2023-04-27 12:57:27,622 INFO [train.py:904] (7/8) Epoch 1, batch 2050, loss[loss=0.3768, simple_loss=0.4311, pruned_loss=0.1613, over 17027.00 frames. ], tot_loss[loss=0.4472, simple_loss=0.4531, pruned_loss=0.2201, over 3303331.09 frames. ], batch size: 50, lr: 4.81e-02, grad_scale: 8.0 2023-04-27 12:57:40,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4837, 5.4865, 4.6745, 4.8937, 5.4998, 5.6192, 5.0025, 5.5872], device='cuda:7'), covar=tensor([0.0122, 0.0180, 0.0248, 0.0231, 0.0083, 0.0100, 0.0142, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0040, 0.0045, 0.0055, 0.0052, 0.0041, 0.0045, 0.0052, 0.0051], device='cuda:7'), out_proj_covar=tensor([3.4103e-05, 4.2071e-05, 5.4222e-05, 4.7555e-05, 3.4469e-05, 4.0034e-05, 4.9487e-05, 4.7082e-05], device='cuda:7') 2023-04-27 12:58:19,061 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2090.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 12:58:32,238 INFO [train.py:904] (7/8) Epoch 1, batch 2100, loss[loss=0.4284, simple_loss=0.4654, pruned_loss=0.1957, over 17086.00 frames. ], tot_loss[loss=0.438, simple_loss=0.4485, pruned_loss=0.2134, over 3307174.54 frames. ], batch size: 53, lr: 4.80e-02, grad_scale: 16.0 2023-04-27 12:58:45,936 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.641e+02 4.108e+02 4.910e+02 5.858e+02 1.001e+03, threshold=9.819e+02, percent-clipped=0.0 2023-04-27 12:58:59,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0175, 4.0438, 3.9441, 4.0391, 3.8523, 4.1992, 3.9829, 3.8594], device='cuda:7'), covar=tensor([0.0457, 0.0369, 0.0236, 0.0318, 0.0775, 0.0197, 0.0351, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0064, 0.0058, 0.0074, 0.0062, 0.0069, 0.0062, 0.0066, 0.0067], device='cuda:7'), out_proj_covar=tensor([5.4771e-05, 5.0262e-05, 6.3136e-05, 5.1432e-05, 6.2606e-05, 5.0149e-05, 6.0527e-05, 5.9405e-05], device='cuda:7') 2023-04-27 12:59:20,452 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2138.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 12:59:35,901 INFO [train.py:904] (7/8) Epoch 1, batch 2150, loss[loss=0.4482, simple_loss=0.4564, pruned_loss=0.22, over 16530.00 frames. ], tot_loss[loss=0.4329, simple_loss=0.4459, pruned_loss=0.2096, over 3309989.57 frames. ], batch size: 75, lr: 4.79e-02, grad_scale: 16.0 2023-04-27 13:00:32,118 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2194.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:00:40,254 INFO [train.py:904] (7/8) Epoch 1, batch 2200, loss[loss=0.4605, simple_loss=0.4608, pruned_loss=0.2301, over 11594.00 frames. ], tot_loss[loss=0.4257, simple_loss=0.4422, pruned_loss=0.2044, over 3306607.02 frames. ], batch size: 246, lr: 4.78e-02, grad_scale: 16.0 2023-04-27 13:00:53,198 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.805e+02 5.075e+02 6.122e+02 8.359e+02 1.658e+03, threshold=1.224e+03, percent-clipped=12.0 2023-04-27 13:00:56,705 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2213.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:01:00,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0546, 3.5760, 3.3833, 3.8739, 4.0117, 1.6052, 3.9228, 2.9794], device='cuda:7'), covar=tensor([0.3935, 0.1314, 0.1898, 0.0517, 0.1319, 0.7313, 0.0550, 0.0951], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0035, 0.0056, 0.0038, 0.0034, 0.0073, 0.0035, 0.0023], device='cuda:7'), out_proj_covar=tensor([5.9294e-05, 3.5717e-05, 5.2140e-05, 2.9602e-05, 3.4489e-05, 6.3716e-05, 2.9456e-05, 2.1893e-05], device='cuda:7') 2023-04-27 13:01:07,671 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2222.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 13:01:11,935 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2226.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:01:44,909 INFO [train.py:904] (7/8) Epoch 1, batch 2250, loss[loss=0.3708, simple_loss=0.4141, pruned_loss=0.1638, over 17209.00 frames. ], tot_loss[loss=0.4217, simple_loss=0.4397, pruned_loss=0.2017, over 3300722.99 frames. ], batch size: 45, lr: 4.77e-02, grad_scale: 16.0 2023-04-27 13:02:05,785 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2267.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 13:02:08,864 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2270.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:02:49,163 INFO [train.py:904] (7/8) Epoch 1, batch 2300, loss[loss=0.396, simple_loss=0.4177, pruned_loss=0.1871, over 16853.00 frames. ], tot_loss[loss=0.4157, simple_loss=0.4358, pruned_loss=0.1977, over 3305895.91 frames. ], batch size: 96, lr: 4.77e-02, grad_scale: 16.0 2023-04-27 13:03:01,868 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.985e+02 4.330e+02 5.680e+02 7.295e+02 1.284e+03, threshold=1.136e+03, percent-clipped=1.0 2023-04-27 13:03:07,887 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2315.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:03:39,673 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2340.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:03:53,145 INFO [train.py:904] (7/8) Epoch 1, batch 2350, loss[loss=0.4451, simple_loss=0.4336, pruned_loss=0.2283, over 16748.00 frames. ], tot_loss[loss=0.4119, simple_loss=0.434, pruned_loss=0.1949, over 3305729.72 frames. ], batch size: 83, lr: 4.76e-02, grad_scale: 16.0 2023-04-27 13:04:00,345 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5823, 5.9471, 5.6961, 5.7580, 5.8273, 6.0613, 6.1694, 5.6714], device='cuda:7'), covar=tensor([0.0325, 0.0462, 0.0566, 0.0837, 0.0903, 0.0333, 0.0369, 0.0954], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0149, 0.0130, 0.0143, 0.0161, 0.0112, 0.0104, 0.0149], device='cuda:7'), out_proj_covar=tensor([1.0793e-04, 1.3963e-04, 1.1822e-04, 1.2971e-04, 1.5587e-04, 1.0723e-04, 9.6676e-05, 1.4716e-04], device='cuda:7') 2023-04-27 13:04:54,412 INFO [train.py:904] (7/8) Epoch 1, batch 2400, loss[loss=0.5124, simple_loss=0.502, pruned_loss=0.2613, over 11675.00 frames. ], tot_loss[loss=0.4103, simple_loss=0.4334, pruned_loss=0.1935, over 3297472.49 frames. ], batch size: 246, lr: 4.75e-02, grad_scale: 16.0 2023-04-27 13:04:55,668 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2401.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 13:05:07,297 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.161e+02 4.808e+02 6.214e+02 8.409e+02 1.581e+03, threshold=1.243e+03, percent-clipped=3.0 2023-04-27 13:05:48,717 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-04-27 13:05:55,896 INFO [train.py:904] (7/8) Epoch 1, batch 2450, loss[loss=0.3342, simple_loss=0.3873, pruned_loss=0.1405, over 17246.00 frames. ], tot_loss[loss=0.4053, simple_loss=0.4314, pruned_loss=0.1895, over 3310300.44 frames. ], batch size: 44, lr: 4.74e-02, grad_scale: 16.0 2023-04-27 13:06:03,758 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-27 13:06:51,102 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2494.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 13:06:58,999 INFO [train.py:904] (7/8) Epoch 1, batch 2500, loss[loss=0.3732, simple_loss=0.4267, pruned_loss=0.1599, over 17130.00 frames. ], tot_loss[loss=0.3995, simple_loss=0.4281, pruned_loss=0.1854, over 3313600.06 frames. ], batch size: 48, lr: 4.73e-02, grad_scale: 16.0 2023-04-27 13:07:11,505 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.680e+02 4.365e+02 5.111e+02 7.039e+02 1.163e+03, threshold=1.022e+03, percent-clipped=0.0 2023-04-27 13:07:14,722 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2513.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:07:29,940 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2526.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:07:50,931 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2542.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:08:01,515 INFO [train.py:904] (7/8) Epoch 1, batch 2550, loss[loss=0.334, simple_loss=0.3823, pruned_loss=0.1428, over 16857.00 frames. ], tot_loss[loss=0.3975, simple_loss=0.4267, pruned_loss=0.1842, over 3303627.02 frames. ], batch size: 42, lr: 4.72e-02, grad_scale: 16.0 2023-04-27 13:08:16,050 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2561.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:08:32,316 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=2574.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:08:40,016 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-27 13:09:07,701 INFO [train.py:904] (7/8) Epoch 1, batch 2600, loss[loss=0.3625, simple_loss=0.4078, pruned_loss=0.1586, over 17009.00 frames. ], tot_loss[loss=0.392, simple_loss=0.4238, pruned_loss=0.18, over 3316513.12 frames. ], batch size: 41, lr: 4.71e-02, grad_scale: 16.0 2023-04-27 13:09:12,286 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-27 13:09:20,093 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.935e+02 4.794e+02 6.148e+02 7.560e+02 1.171e+03, threshold=1.230e+03, percent-clipped=4.0 2023-04-27 13:09:34,653 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.9348, 6.2182, 5.7526, 6.2917, 5.6361, 5.6606, 6.0082, 6.3529], device='cuda:7'), covar=tensor([0.0272, 0.0392, 0.0386, 0.0176, 0.0386, 0.0228, 0.0212, 0.0140], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0131, 0.0114, 0.0090, 0.0118, 0.0104, 0.0108, 0.0089], device='cuda:7'), out_proj_covar=tensor([1.0185e-04, 1.1738e-04, 9.6680e-05, 6.7457e-05, 9.8539e-05, 8.4236e-05, 9.1433e-05, 7.7924e-05], device='cuda:7') 2023-04-27 13:10:11,891 INFO [train.py:904] (7/8) Epoch 1, batch 2650, loss[loss=0.4006, simple_loss=0.4368, pruned_loss=0.1822, over 16687.00 frames. ], tot_loss[loss=0.3862, simple_loss=0.4207, pruned_loss=0.1758, over 3321753.99 frames. ], batch size: 57, lr: 4.70e-02, grad_scale: 16.0 2023-04-27 13:10:55,311 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2685.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:11:10,032 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2696.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:11:15,041 INFO [train.py:904] (7/8) Epoch 1, batch 2700, loss[loss=0.3878, simple_loss=0.4296, pruned_loss=0.173, over 16063.00 frames. ], tot_loss[loss=0.3824, simple_loss=0.4187, pruned_loss=0.173, over 3328501.81 frames. ], batch size: 35, lr: 4.69e-02, grad_scale: 16.0 2023-04-27 13:11:29,647 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.771e+02 4.333e+02 5.264e+02 6.238e+02 1.242e+03, threshold=1.053e+03, percent-clipped=1.0 2023-04-27 13:12:14,277 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2746.0, num_to_drop=2, layers_to_drop={0, 1} 2023-04-27 13:12:20,241 INFO [train.py:904] (7/8) Epoch 1, batch 2750, loss[loss=0.3147, simple_loss=0.3621, pruned_loss=0.1336, over 16973.00 frames. ], tot_loss[loss=0.3774, simple_loss=0.4155, pruned_loss=0.1696, over 3334820.18 frames. ], batch size: 41, lr: 4.68e-02, grad_scale: 16.0 2023-04-27 13:13:23,784 INFO [train.py:904] (7/8) Epoch 1, batch 2800, loss[loss=0.3107, simple_loss=0.3698, pruned_loss=0.1258, over 16791.00 frames. ], tot_loss[loss=0.3762, simple_loss=0.415, pruned_loss=0.1687, over 3336361.71 frames. ], batch size: 39, lr: 4.67e-02, grad_scale: 16.0 2023-04-27 13:13:35,362 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.742e+02 4.329e+02 5.433e+02 6.254e+02 2.106e+03, threshold=1.087e+03, percent-clipped=5.0 2023-04-27 13:13:37,698 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-27 13:13:54,374 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2824.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:13:59,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2202, 4.7862, 5.1143, 5.1707, 4.5323, 4.9534, 5.1721, 4.6722], device='cuda:7'), covar=tensor([0.0329, 0.0263, 0.0258, 0.0145, 0.0987, 0.0347, 0.0150, 0.0283], device='cuda:7'), in_proj_covar=tensor([0.0091, 0.0076, 0.0113, 0.0083, 0.0134, 0.0096, 0.0084, 0.0091], device='cuda:7'), out_proj_covar=tensor([1.1262e-04, 8.3828e-05, 1.3332e-04, 9.4619e-05, 1.6887e-04, 1.1871e-04, 9.9705e-05, 1.1757e-04], device='cuda:7') 2023-04-27 13:14:26,007 INFO [train.py:904] (7/8) Epoch 1, batch 2850, loss[loss=0.4606, simple_loss=0.4654, pruned_loss=0.2279, over 15466.00 frames. ], tot_loss[loss=0.3736, simple_loss=0.4135, pruned_loss=0.1669, over 3335383.92 frames. ], batch size: 191, lr: 4.66e-02, grad_scale: 16.0 2023-04-27 13:14:50,270 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9090, 5.1093, 4.7086, 4.9706, 5.0375, 5.2777, 5.2525, 4.8298], device='cuda:7'), covar=tensor([0.0448, 0.0742, 0.0790, 0.1086, 0.1308, 0.0466, 0.0523, 0.1229], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0167, 0.0142, 0.0151, 0.0181, 0.0124, 0.0118, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 13:15:09,634 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2885.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 13:15:27,294 INFO [train.py:904] (7/8) Epoch 1, batch 2900, loss[loss=0.3573, simple_loss=0.3871, pruned_loss=0.1638, over 16520.00 frames. ], tot_loss[loss=0.3722, simple_loss=0.4107, pruned_loss=0.1668, over 3334090.26 frames. ], batch size: 68, lr: 4.65e-02, grad_scale: 16.0 2023-04-27 13:15:40,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.037e+02 5.015e+02 6.760e+02 8.832e+02 1.641e+03, threshold=1.352e+03, percent-clipped=13.0 2023-04-27 13:16:03,418 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6352, 5.8992, 5.4862, 6.0284, 5.4237, 5.6304, 5.4886, 5.9856], device='cuda:7'), covar=tensor([0.0389, 0.0501, 0.0484, 0.0224, 0.0493, 0.0279, 0.0409, 0.0199], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0143, 0.0125, 0.0095, 0.0125, 0.0110, 0.0127, 0.0095], device='cuda:7'), out_proj_covar=tensor([1.1348e-04, 1.3189e-04, 1.0880e-04, 7.5434e-05, 1.0754e-04, 9.1694e-05, 1.1256e-04, 8.6483e-05], device='cuda:7') 2023-04-27 13:16:13,958 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2937.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:16:32,284 INFO [train.py:904] (7/8) Epoch 1, batch 2950, loss[loss=0.3138, simple_loss=0.3781, pruned_loss=0.1248, over 17103.00 frames. ], tot_loss[loss=0.3701, simple_loss=0.4092, pruned_loss=0.1655, over 3329743.57 frames. ], batch size: 53, lr: 4.64e-02, grad_scale: 16.0 2023-04-27 13:17:09,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0599, 4.1255, 3.9417, 4.5310, 4.1114, 4.0154, 3.9502, 4.5791], device='cuda:7'), covar=tensor([0.0112, 0.0165, 0.0174, 0.0162, 0.0091, 0.0173, 0.0160, 0.0058], device='cuda:7'), in_proj_covar=tensor([0.0021, 0.0020, 0.0021, 0.0020, 0.0022, 0.0019, 0.0023, 0.0018], device='cuda:7'), out_proj_covar=tensor([1.9803e-05, 1.9475e-05, 1.9608e-05, 1.8746e-05, 2.0040e-05, 1.8792e-05, 2.0532e-05, 1.5409e-05], device='cuda:7') 2023-04-27 13:17:29,939 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2996.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:17:32,231 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2998.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 13:17:35,326 INFO [train.py:904] (7/8) Epoch 1, batch 3000, loss[loss=0.3967, simple_loss=0.435, pruned_loss=0.1792, over 16764.00 frames. ], tot_loss[loss=0.3686, simple_loss=0.4076, pruned_loss=0.1648, over 3330469.49 frames. ], batch size: 57, lr: 4.63e-02, grad_scale: 16.0 2023-04-27 13:17:35,327 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 13:17:43,859 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9195, 3.1197, 3.2371, 3.4813, 3.2117, 3.4088, 2.8248, 3.0952], device='cuda:7'), covar=tensor([0.0143, 0.0127, 0.0171, 0.0157, 0.0095, 0.0073, 0.0192, 0.0182], device='cuda:7'), in_proj_covar=tensor([0.0034, 0.0040, 0.0037, 0.0040, 0.0033, 0.0036, 0.0039, 0.0042], device='cuda:7'), out_proj_covar=tensor([3.1851e-05, 3.3212e-05, 3.3575e-05, 3.8117e-05, 2.6969e-05, 2.8965e-05, 3.3864e-05, 3.5142e-05], device='cuda:7') 2023-04-27 13:17:45,055 INFO [train.py:938] (7/8) Epoch 1, validation: loss=0.2847, simple_loss=0.3895, pruned_loss=0.08992, over 944034.00 frames. 2023-04-27 13:17:45,056 INFO [train.py:939] (7/8) Maximum memory allocated so far is 15490MB 2023-04-27 13:17:59,842 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.008e+02 4.355e+02 5.212e+02 6.501e+02 1.071e+03, threshold=1.042e+03, percent-clipped=0.0 2023-04-27 13:18:08,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8746, 2.0112, 1.7167, 1.1507, 1.9039, 1.8840, 2.0470, 2.0006], device='cuda:7'), covar=tensor([0.0132, 0.0171, 0.0212, 0.0419, 0.0126, 0.0141, 0.0115, 0.0168], device='cuda:7'), in_proj_covar=tensor([0.0027, 0.0031, 0.0034, 0.0035, 0.0030, 0.0034, 0.0033, 0.0031], device='cuda:7'), out_proj_covar=tensor([3.4419e-05, 3.3841e-05, 3.8009e-05, 3.7769e-05, 3.3310e-05, 3.6602e-05, 3.4652e-05, 3.5640e-05], device='cuda:7') 2023-04-27 13:18:09,358 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3019.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:18:35,897 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0964, 4.9043, 4.1352, 4.3601, 4.7027, 5.0780, 4.4671, 5.0286], device='cuda:7'), covar=tensor([0.0172, 0.0185, 0.0237, 0.0390, 0.0157, 0.0119, 0.0160, 0.0133], device='cuda:7'), in_proj_covar=tensor([0.0052, 0.0047, 0.0068, 0.0067, 0.0048, 0.0055, 0.0063, 0.0059], device='cuda:7'), out_proj_covar=tensor([6.4535e-05, 5.6786e-05, 9.7552e-05, 8.5199e-05, 5.1914e-05, 6.2667e-05, 8.3183e-05, 7.6721e-05], device='cuda:7') 2023-04-27 13:18:36,944 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3041.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:18:41,580 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3044.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:18:50,101 INFO [train.py:904] (7/8) Epoch 1, batch 3050, loss[loss=0.379, simple_loss=0.4002, pruned_loss=0.1789, over 16924.00 frames. ], tot_loss[loss=0.3655, simple_loss=0.4051, pruned_loss=0.163, over 3328654.13 frames. ], batch size: 109, lr: 4.62e-02, grad_scale: 16.0 2023-04-27 13:19:27,629 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3080.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 13:19:53,986 INFO [train.py:904] (7/8) Epoch 1, batch 3100, loss[loss=0.41, simple_loss=0.4184, pruned_loss=0.2007, over 16698.00 frames. ], tot_loss[loss=0.3633, simple_loss=0.4036, pruned_loss=0.1615, over 3329168.28 frames. ], batch size: 124, lr: 4.61e-02, grad_scale: 16.0 2023-04-27 13:20:07,652 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.721e+02 4.392e+02 5.238e+02 7.570e+02 1.450e+03, threshold=1.048e+03, percent-clipped=8.0 2023-04-27 13:21:00,211 INFO [train.py:904] (7/8) Epoch 1, batch 3150, loss[loss=0.3382, simple_loss=0.4027, pruned_loss=0.1368, over 17101.00 frames. ], tot_loss[loss=0.3587, simple_loss=0.3995, pruned_loss=0.159, over 3335120.58 frames. ], batch size: 48, lr: 4.60e-02, grad_scale: 16.0 2023-04-27 13:21:38,563 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3180.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:21:46,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2535, 5.4593, 5.0240, 5.5055, 4.9556, 5.2479, 5.1898, 5.4229], device='cuda:7'), covar=tensor([0.0330, 0.0490, 0.0460, 0.0230, 0.0559, 0.0298, 0.0338, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0160, 0.0133, 0.0103, 0.0135, 0.0117, 0.0135, 0.0098], device='cuda:7'), out_proj_covar=tensor([1.2380e-04, 1.4922e-04, 1.1627e-04, 8.4110e-05, 1.1723e-04, 1.0150e-04, 1.2303e-04, 9.3061e-05], device='cuda:7') 2023-04-27 13:22:05,298 INFO [train.py:904] (7/8) Epoch 1, batch 3200, loss[loss=0.3043, simple_loss=0.3619, pruned_loss=0.1234, over 16838.00 frames. ], tot_loss[loss=0.3566, simple_loss=0.3984, pruned_loss=0.1574, over 3319784.88 frames. ], batch size: 42, lr: 4.59e-02, grad_scale: 16.0 2023-04-27 13:22:17,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.273e+02 4.612e+02 6.158e+02 7.733e+02 1.150e+03, threshold=1.232e+03, percent-clipped=3.0 2023-04-27 13:23:07,327 INFO [train.py:904] (7/8) Epoch 1, batch 3250, loss[loss=0.3407, simple_loss=0.4037, pruned_loss=0.1388, over 17076.00 frames. ], tot_loss[loss=0.3574, simple_loss=0.3986, pruned_loss=0.1581, over 3323328.62 frames. ], batch size: 49, lr: 4.58e-02, grad_scale: 16.0 2023-04-27 13:23:56,332 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3288.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:24:02,468 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3293.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:24:12,260 INFO [train.py:904] (7/8) Epoch 1, batch 3300, loss[loss=0.404, simple_loss=0.4306, pruned_loss=0.1887, over 16394.00 frames. ], tot_loss[loss=0.3574, simple_loss=0.3993, pruned_loss=0.1577, over 3325299.13 frames. ], batch size: 146, lr: 4.57e-02, grad_scale: 16.0 2023-04-27 13:24:19,258 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5965, 4.5943, 4.7908, 4.9279, 5.1169, 4.6961, 4.5826, 4.8420], device='cuda:7'), covar=tensor([0.0265, 0.0171, 0.0465, 0.0373, 0.0324, 0.0286, 0.0437, 0.0183], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0104, 0.0134, 0.0132, 0.0138, 0.0116, 0.0126, 0.0102], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:7') 2023-04-27 13:24:25,123 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.531e+02 4.292e+02 5.268e+02 6.867e+02 1.392e+03, threshold=1.054e+03, percent-clipped=2.0 2023-04-27 13:25:03,624 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3341.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:25:16,133 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3349.0, num_to_drop=2, layers_to_drop={1, 2} 2023-04-27 13:25:17,759 INFO [train.py:904] (7/8) Epoch 1, batch 3350, loss[loss=0.3483, simple_loss=0.3839, pruned_loss=0.1563, over 16223.00 frames. ], tot_loss[loss=0.3549, simple_loss=0.3979, pruned_loss=0.1559, over 3332837.22 frames. ], batch size: 165, lr: 4.56e-02, grad_scale: 16.0 2023-04-27 13:25:50,433 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3375.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:26:09,275 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3389.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:26:17,419 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-27 13:26:24,816 INFO [train.py:904] (7/8) Epoch 1, batch 3400, loss[loss=0.4412, simple_loss=0.4541, pruned_loss=0.2142, over 12162.00 frames. ], tot_loss[loss=0.3502, simple_loss=0.395, pruned_loss=0.1527, over 3334924.52 frames. ], batch size: 247, lr: 4.55e-02, grad_scale: 16.0 2023-04-27 13:26:39,223 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.681e+02 4.195e+02 5.225e+02 6.801e+02 1.040e+03, threshold=1.045e+03, percent-clipped=0.0 2023-04-27 13:27:32,314 INFO [train.py:904] (7/8) Epoch 1, batch 3450, loss[loss=0.3198, simple_loss=0.3594, pruned_loss=0.1401, over 16550.00 frames. ], tot_loss[loss=0.3482, simple_loss=0.393, pruned_loss=0.1517, over 3322738.50 frames. ], batch size: 75, lr: 4.54e-02, grad_scale: 16.0 2023-04-27 13:28:12,494 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3480.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:28:15,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1584, 5.0891, 4.9397, 5.4847, 5.3315, 5.4629, 5.4326, 5.2791], device='cuda:7'), covar=tensor([0.0290, 0.0258, 0.1092, 0.0362, 0.0440, 0.0183, 0.0307, 0.0287], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0131, 0.0223, 0.0162, 0.0134, 0.0138, 0.0129, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 13:28:39,723 INFO [train.py:904] (7/8) Epoch 1, batch 3500, loss[loss=0.3003, simple_loss=0.3638, pruned_loss=0.1184, over 17185.00 frames. ], tot_loss[loss=0.3442, simple_loss=0.39, pruned_loss=0.1492, over 3326265.57 frames. ], batch size: 45, lr: 4.53e-02, grad_scale: 16.0 2023-04-27 13:28:53,815 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 4.599e+02 5.593e+02 7.423e+02 2.273e+03, threshold=1.119e+03, percent-clipped=10.0 2023-04-27 13:29:16,324 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3528.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:29:34,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4914, 4.0065, 3.5433, 3.4816, 4.0368, 3.6857, 4.0061, 4.0127], device='cuda:7'), covar=tensor([0.0271, 0.0153, 0.0201, 0.0189, 0.0153, 0.0256, 0.0222, 0.0160], device='cuda:7'), in_proj_covar=tensor([0.0047, 0.0034, 0.0034, 0.0043, 0.0036, 0.0040, 0.0045, 0.0039], device='cuda:7'), out_proj_covar=tensor([6.2889e-05, 4.7532e-05, 4.6314e-05, 5.6449e-05, 4.9413e-05, 6.3906e-05, 6.0088e-05, 5.2161e-05], device='cuda:7') 2023-04-27 13:29:40,847 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3546.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:29:46,548 INFO [train.py:904] (7/8) Epoch 1, batch 3550, loss[loss=0.2777, simple_loss=0.3431, pruned_loss=0.1062, over 17196.00 frames. ], tot_loss[loss=0.3424, simple_loss=0.3886, pruned_loss=0.1481, over 3324867.90 frames. ], batch size: 45, lr: 4.51e-02, grad_scale: 16.0 2023-04-27 13:30:44,892 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3593.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:30:54,501 INFO [train.py:904] (7/8) Epoch 1, batch 3600, loss[loss=0.3451, simple_loss=0.3797, pruned_loss=0.1553, over 16909.00 frames. ], tot_loss[loss=0.3382, simple_loss=0.3855, pruned_loss=0.1455, over 3322446.52 frames. ], batch size: 109, lr: 4.50e-02, grad_scale: 16.0 2023-04-27 13:31:04,121 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3607.0, num_to_drop=2, layers_to_drop={0, 3} 2023-04-27 13:31:08,705 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.446e+02 5.845e+02 7.592e+02 1.096e+03 2.095e+03, threshold=1.518e+03, percent-clipped=22.0 2023-04-27 13:31:52,477 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3641.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:31:56,491 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3644.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:32:06,064 INFO [train.py:904] (7/8) Epoch 1, batch 3650, loss[loss=0.2884, simple_loss=0.3288, pruned_loss=0.124, over 16809.00 frames. ], tot_loss[loss=0.3354, simple_loss=0.3819, pruned_loss=0.1444, over 3311188.23 frames. ], batch size: 102, lr: 4.49e-02, grad_scale: 16.0 2023-04-27 13:32:33,870 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2204, 4.0367, 4.1706, 4.2998, 3.7921, 4.2276, 4.0198, 3.9367], device='cuda:7'), covar=tensor([0.0224, 0.0147, 0.0161, 0.0091, 0.0652, 0.0169, 0.0262, 0.0161], device='cuda:7'), in_proj_covar=tensor([0.0096, 0.0072, 0.0121, 0.0089, 0.0150, 0.0097, 0.0085, 0.0095], device='cuda:7'), out_proj_covar=tensor([1.4395e-04, 9.9951e-05, 1.7433e-04, 1.2040e-04, 2.0105e-04, 1.4613e-04, 1.2453e-04, 1.4681e-04], device='cuda:7') 2023-04-27 13:32:42,408 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3675.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:32:43,687 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3676.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:33:20,385 INFO [train.py:904] (7/8) Epoch 1, batch 3700, loss[loss=0.293, simple_loss=0.3382, pruned_loss=0.1239, over 16437.00 frames. ], tot_loss[loss=0.3326, simple_loss=0.3774, pruned_loss=0.1439, over 3308853.00 frames. ], batch size: 75, lr: 4.48e-02, grad_scale: 16.0 2023-04-27 13:33:35,073 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.010e+02 4.933e+02 6.416e+02 8.029e+02 1.327e+03, threshold=1.283e+03, percent-clipped=0.0 2023-04-27 13:33:54,112 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3723.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:34:15,111 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3737.0, num_to_drop=2, layers_to_drop={0, 2} 2023-04-27 13:34:33,715 INFO [train.py:904] (7/8) Epoch 1, batch 3750, loss[loss=0.3747, simple_loss=0.4056, pruned_loss=0.1719, over 15559.00 frames. ], tot_loss[loss=0.3342, simple_loss=0.3773, pruned_loss=0.1456, over 3297513.82 frames. ], batch size: 190, lr: 4.47e-02, grad_scale: 16.0 2023-04-27 13:35:13,041 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.96 vs. limit=5.0 2023-04-27 13:35:46,285 INFO [train.py:904] (7/8) Epoch 1, batch 3800, loss[loss=0.357, simple_loss=0.3861, pruned_loss=0.164, over 16757.00 frames. ], tot_loss[loss=0.3361, simple_loss=0.3779, pruned_loss=0.1472, over 3288619.28 frames. ], batch size: 124, lr: 4.46e-02, grad_scale: 16.0 2023-04-27 13:36:00,504 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.355e+02 5.070e+02 6.208e+02 8.135e+02 1.675e+03, threshold=1.242e+03, percent-clipped=4.0 2023-04-27 13:36:57,588 INFO [train.py:904] (7/8) Epoch 1, batch 3850, loss[loss=0.3226, simple_loss=0.3635, pruned_loss=0.1408, over 16731.00 frames. ], tot_loss[loss=0.3351, simple_loss=0.3767, pruned_loss=0.1468, over 3276134.81 frames. ], batch size: 134, lr: 4.45e-02, grad_scale: 16.0 2023-04-27 13:38:09,802 INFO [train.py:904] (7/8) Epoch 1, batch 3900, loss[loss=0.2922, simple_loss=0.354, pruned_loss=0.1152, over 16654.00 frames. ], tot_loss[loss=0.332, simple_loss=0.3737, pruned_loss=0.1452, over 3277475.65 frames. ], batch size: 62, lr: 4.44e-02, grad_scale: 16.0 2023-04-27 13:38:11,883 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3902.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:38:11,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3717, 4.1646, 4.3650, 4.4113, 3.6317, 4.3519, 4.1260, 3.9788], device='cuda:7'), covar=tensor([0.0394, 0.0187, 0.0192, 0.0150, 0.0959, 0.0206, 0.0280, 0.0245], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0069, 0.0112, 0.0086, 0.0139, 0.0092, 0.0083, 0.0090], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:7') 2023-04-27 13:38:24,115 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0367, 4.1195, 4.2988, 4.3345, 4.4776, 4.2642, 3.9190, 4.3363], device='cuda:7'), covar=tensor([0.0452, 0.0283, 0.0474, 0.0451, 0.0423, 0.0325, 0.0650, 0.0340], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0099, 0.0124, 0.0125, 0.0132, 0.0109, 0.0118, 0.0099], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001], device='cuda:7') 2023-04-27 13:38:24,737 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.787e+02 4.914e+02 5.725e+02 7.504e+02 1.784e+03, threshold=1.145e+03, percent-clipped=3.0 2023-04-27 13:38:46,496 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3926.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:39:11,934 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3944.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:39:21,371 INFO [train.py:904] (7/8) Epoch 1, batch 3950, loss[loss=0.3151, simple_loss=0.3564, pruned_loss=0.1369, over 16487.00 frames. ], tot_loss[loss=0.33, simple_loss=0.3714, pruned_loss=0.1442, over 3277138.74 frames. ], batch size: 146, lr: 4.43e-02, grad_scale: 16.0 2023-04-27 13:39:59,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3968, 2.4955, 2.2991, 2.9260, 2.8184, 2.5295, 2.7845, 2.1851], device='cuda:7'), covar=tensor([0.0240, 0.0418, 0.0527, 0.0167, 0.0140, 0.0331, 0.0219, 0.0223], device='cuda:7'), in_proj_covar=tensor([0.0028, 0.0046, 0.0039, 0.0035, 0.0032, 0.0034, 0.0037, 0.0034], device='cuda:7'), out_proj_covar=tensor([4.2298e-05, 6.7653e-05, 5.7779e-05, 4.3818e-05, 4.3377e-05, 4.7146e-05, 4.6357e-05, 4.4535e-05], device='cuda:7') 2023-04-27 13:40:14,034 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3987.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:40:21,227 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=3992.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:40:25,226 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-27 13:40:37,511 INFO [train.py:904] (7/8) Epoch 1, batch 4000, loss[loss=0.3214, simple_loss=0.368, pruned_loss=0.1374, over 16335.00 frames. ], tot_loss[loss=0.3272, simple_loss=0.3692, pruned_loss=0.1426, over 3277275.42 frames. ], batch size: 165, lr: 4.42e-02, grad_scale: 16.0 2023-04-27 13:40:41,204 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-27 13:40:52,174 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.081e+02 4.378e+02 5.733e+02 7.371e+02 1.613e+03, threshold=1.147e+03, percent-clipped=6.0 2023-04-27 13:41:11,232 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1598, 2.9153, 3.0790, 2.6838, 2.8190, 2.9762, 2.9116, 2.8953], device='cuda:7'), covar=tensor([0.0488, 0.0159, 0.0083, 0.0158, 0.0158, 0.0113, 0.0123, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0050, 0.0030, 0.0031, 0.0041, 0.0032, 0.0034, 0.0039, 0.0037], device='cuda:7'), out_proj_covar=tensor([7.4929e-05, 4.8632e-05, 4.7703e-05, 5.8062e-05, 4.8570e-05, 5.9461e-05, 5.7442e-05, 5.5321e-05], device='cuda:7') 2023-04-27 13:41:16,944 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-27 13:41:21,862 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 13:41:24,005 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4032.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:41:50,492 INFO [train.py:904] (7/8) Epoch 1, batch 4050, loss[loss=0.2481, simple_loss=0.3117, pruned_loss=0.09222, over 16265.00 frames. ], tot_loss[loss=0.3176, simple_loss=0.3636, pruned_loss=0.1358, over 3279704.87 frames. ], batch size: 35, lr: 4.41e-02, grad_scale: 16.0 2023-04-27 13:42:05,081 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-27 13:43:04,576 INFO [train.py:904] (7/8) Epoch 1, batch 4100, loss[loss=0.3108, simple_loss=0.3718, pruned_loss=0.1249, over 16674.00 frames. ], tot_loss[loss=0.3129, simple_loss=0.3617, pruned_loss=0.132, over 3260822.65 frames. ], batch size: 57, lr: 4.40e-02, grad_scale: 32.0 2023-04-27 13:43:05,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1148, 1.6170, 1.7280, 2.4398, 2.4251, 2.2032, 2.1214, 2.1667], device='cuda:7'), covar=tensor([0.0135, 0.0535, 0.0241, 0.0183, 0.0085, 0.0137, 0.0205, 0.0180], device='cuda:7'), in_proj_covar=tensor([0.0023, 0.0037, 0.0028, 0.0028, 0.0022, 0.0027, 0.0027, 0.0031], device='cuda:7'), out_proj_covar=tensor([2.5429e-05, 3.9463e-05, 2.9670e-05, 2.8139e-05, 2.0025e-05, 2.3349e-05, 2.5336e-05, 2.8515e-05], device='cuda:7') 2023-04-27 13:43:18,965 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.292e+02 4.086e+02 5.427e+02 7.425e+02 1.337e+03, threshold=1.085e+03, percent-clipped=4.0 2023-04-27 13:44:05,065 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4141.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:44:20,308 INFO [train.py:904] (7/8) Epoch 1, batch 4150, loss[loss=0.3108, simple_loss=0.3732, pruned_loss=0.1241, over 16688.00 frames. ], tot_loss[loss=0.3248, simple_loss=0.3723, pruned_loss=0.1387, over 3230864.62 frames. ], batch size: 62, lr: 4.39e-02, grad_scale: 32.0 2023-04-27 13:45:24,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-27 13:45:30,139 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-27 13:45:37,095 INFO [train.py:904] (7/8) Epoch 1, batch 4200, loss[loss=0.3847, simple_loss=0.4355, pruned_loss=0.1669, over 16198.00 frames. ], tot_loss[loss=0.3334, simple_loss=0.3816, pruned_loss=0.1426, over 3201549.36 frames. ], batch size: 165, lr: 4.38e-02, grad_scale: 16.0 2023-04-27 13:45:39,677 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4202.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 13:45:39,719 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4202.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:45:52,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.534e+02 4.783e+02 6.571e+02 8.274e+02 1.863e+03, threshold=1.314e+03, percent-clipped=9.0 2023-04-27 13:46:21,667 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.09 vs. limit=2.0 2023-04-27 13:46:45,993 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6332, 2.3368, 2.6355, 2.6755, 3.1383, 2.6206, 2.9068, 2.8974], device='cuda:7'), covar=tensor([0.0102, 0.0588, 0.0219, 0.0168, 0.0078, 0.0259, 0.0152, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0028, 0.0051, 0.0037, 0.0035, 0.0031, 0.0034, 0.0035, 0.0032], device='cuda:7'), out_proj_covar=tensor([3.9508e-05, 8.2007e-05, 5.8181e-05, 4.4231e-05, 4.2190e-05, 4.8754e-05, 4.6476e-05, 4.2469e-05], device='cuda:7') 2023-04-27 13:46:50,184 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4250.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:46:50,870 INFO [train.py:904] (7/8) Epoch 1, batch 4250, loss[loss=0.3229, simple_loss=0.3759, pruned_loss=0.1349, over 16711.00 frames. ], tot_loss[loss=0.3339, simple_loss=0.3836, pruned_loss=0.1421, over 3187170.36 frames. ], batch size: 124, lr: 4.36e-02, grad_scale: 16.0 2023-04-27 13:47:28,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6947, 2.6462, 2.6225, 1.7667, 2.5429, 2.6638, 2.5551, 2.6595], device='cuda:7'), covar=tensor([0.0136, 0.0136, 0.0168, 0.1464, 0.0197, 0.0115, 0.0143, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0048, 0.0051, 0.0052, 0.0110, 0.0047, 0.0049, 0.0046, 0.0056], device='cuda:7'), out_proj_covar=tensor([6.0923e-05, 6.5550e-05, 7.0600e-05, 1.4301e-04, 6.7043e-05, 6.3325e-05, 6.9737e-05, 6.8307e-05], device='cuda:7') 2023-04-27 13:47:37,207 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4282.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:47:46,122 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4288.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:48:04,726 INFO [train.py:904] (7/8) Epoch 1, batch 4300, loss[loss=0.3232, simple_loss=0.3904, pruned_loss=0.128, over 16609.00 frames. ], tot_loss[loss=0.3317, simple_loss=0.3838, pruned_loss=0.1398, over 3191968.38 frames. ], batch size: 68, lr: 4.35e-02, grad_scale: 16.0 2023-04-27 13:48:21,410 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.968e+02 4.766e+02 5.895e+02 7.865e+02 1.445e+03, threshold=1.179e+03, percent-clipped=2.0 2023-04-27 13:48:52,550 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4332.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 13:49:17,794 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4349.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:49:20,285 INFO [train.py:904] (7/8) Epoch 1, batch 4350, loss[loss=0.3554, simple_loss=0.409, pruned_loss=0.1509, over 16917.00 frames. ], tot_loss[loss=0.3357, simple_loss=0.3878, pruned_loss=0.1417, over 3177556.06 frames. ], batch size: 109, lr: 4.34e-02, grad_scale: 16.0 2023-04-27 13:49:34,429 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-27 13:50:04,950 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4380.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 13:50:36,410 INFO [train.py:904] (7/8) Epoch 1, batch 4400, loss[loss=0.3724, simple_loss=0.4089, pruned_loss=0.1679, over 11773.00 frames. ], tot_loss[loss=0.3369, simple_loss=0.3895, pruned_loss=0.1422, over 3174884.45 frames. ], batch size: 247, lr: 4.33e-02, grad_scale: 16.0 2023-04-27 13:50:51,966 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.736e+02 5.030e+02 6.587e+02 8.143e+02 1.430e+03, threshold=1.317e+03, percent-clipped=9.0 2023-04-27 13:51:48,421 INFO [train.py:904] (7/8) Epoch 1, batch 4450, loss[loss=0.3109, simple_loss=0.3902, pruned_loss=0.1158, over 16554.00 frames. ], tot_loss[loss=0.3365, simple_loss=0.391, pruned_loss=0.141, over 3177991.16 frames. ], batch size: 68, lr: 4.32e-02, grad_scale: 16.0 2023-04-27 13:52:19,084 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-27 13:52:34,523 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5210, 5.3367, 5.1873, 5.1706, 5.2680, 5.7489, 5.5308, 5.1527], device='cuda:7'), covar=tensor([0.0678, 0.1081, 0.0662, 0.1183, 0.1514, 0.0531, 0.0628, 0.1604], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0180, 0.0137, 0.0148, 0.0183, 0.0132, 0.0135, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 13:52:57,217 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4497.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:53:02,600 INFO [train.py:904] (7/8) Epoch 1, batch 4500, loss[loss=0.325, simple_loss=0.3793, pruned_loss=0.1353, over 16940.00 frames. ], tot_loss[loss=0.3327, simple_loss=0.3888, pruned_loss=0.1383, over 3187369.82 frames. ], batch size: 116, lr: 4.31e-02, grad_scale: 8.0 2023-04-27 13:53:20,028 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.477e+02 4.034e+02 4.961e+02 6.637e+02 1.457e+03, threshold=9.923e+02, percent-clipped=1.0 2023-04-27 13:53:38,444 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-27 13:54:14,099 INFO [train.py:904] (7/8) Epoch 1, batch 4550, loss[loss=0.3208, simple_loss=0.3822, pruned_loss=0.1297, over 16471.00 frames. ], tot_loss[loss=0.3309, simple_loss=0.3878, pruned_loss=0.137, over 3209417.58 frames. ], batch size: 68, lr: 4.30e-02, grad_scale: 8.0 2023-04-27 13:54:58,002 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4582.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:55:25,726 INFO [train.py:904] (7/8) Epoch 1, batch 4600, loss[loss=0.2922, simple_loss=0.3677, pruned_loss=0.1084, over 16849.00 frames. ], tot_loss[loss=0.3289, simple_loss=0.3874, pruned_loss=0.1352, over 3221075.23 frames. ], batch size: 102, lr: 4.29e-02, grad_scale: 8.0 2023-04-27 13:55:43,315 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.332e+02 3.980e+02 4.876e+02 6.501e+02 1.584e+03, threshold=9.751e+02, percent-clipped=6.0 2023-04-27 13:56:06,521 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1442, 3.8449, 3.7076, 4.5003, 3.8566, 4.0012, 3.8943, 3.7484], device='cuda:7'), covar=tensor([0.0326, 0.0520, 0.0440, 0.0263, 0.1142, 0.0398, 0.0536, 0.0975], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0067, 0.0057, 0.0059, 0.0119, 0.0065, 0.0083, 0.0069], device='cuda:7'), out_proj_covar=tensor([6.8328e-05, 7.2068e-05, 6.0725e-05, 7.0058e-05, 1.3235e-04, 7.1211e-05, 8.3486e-05, 8.4542e-05], device='cuda:7') 2023-04-27 13:56:07,508 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4630.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:56:28,176 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4644.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 13:56:36,586 INFO [train.py:904] (7/8) Epoch 1, batch 4650, loss[loss=0.379, simple_loss=0.4046, pruned_loss=0.1767, over 11937.00 frames. ], tot_loss[loss=0.3272, simple_loss=0.3856, pruned_loss=0.1344, over 3214592.04 frames. ], batch size: 247, lr: 4.28e-02, grad_scale: 8.0 2023-04-27 13:56:49,586 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.42 vs. limit=5.0 2023-04-27 13:57:50,184 INFO [train.py:904] (7/8) Epoch 1, batch 4700, loss[loss=0.2801, simple_loss=0.3428, pruned_loss=0.1088, over 16812.00 frames. ], tot_loss[loss=0.3221, simple_loss=0.3812, pruned_loss=0.1315, over 3218673.27 frames. ], batch size: 116, lr: 4.27e-02, grad_scale: 8.0 2023-04-27 13:58:07,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5435, 3.3648, 3.6268, 3.1941, 3.7301, 3.7972, 3.7132, 3.3310], device='cuda:7'), covar=tensor([0.0618, 0.0182, 0.0084, 0.0177, 0.0106, 0.0145, 0.0194, 0.0198], device='cuda:7'), in_proj_covar=tensor([0.0062, 0.0028, 0.0029, 0.0041, 0.0029, 0.0030, 0.0037, 0.0038], device='cuda:7'), out_proj_covar=tensor([9.9754e-05, 4.8216e-05, 5.0109e-05, 6.6836e-05, 5.0761e-05, 5.3819e-05, 5.9686e-05, 6.1669e-05], device='cuda:7') 2023-04-27 13:58:07,990 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.460e+02 4.210e+02 5.632e+02 6.697e+02 1.082e+03, threshold=1.126e+03, percent-clipped=2.0 2023-04-27 13:59:02,510 INFO [train.py:904] (7/8) Epoch 1, batch 4750, loss[loss=0.2741, simple_loss=0.3458, pruned_loss=0.1012, over 16858.00 frames. ], tot_loss[loss=0.3171, simple_loss=0.3763, pruned_loss=0.129, over 3218088.96 frames. ], batch size: 90, lr: 4.26e-02, grad_scale: 8.0 2023-04-27 14:00:11,420 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4797.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:00:16,567 INFO [train.py:904] (7/8) Epoch 1, batch 4800, loss[loss=0.2979, simple_loss=0.3463, pruned_loss=0.1248, over 16840.00 frames. ], tot_loss[loss=0.3137, simple_loss=0.3729, pruned_loss=0.1273, over 3222645.72 frames. ], batch size: 39, lr: 4.25e-02, grad_scale: 8.0 2023-04-27 14:00:34,037 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.251e+02 4.322e+02 5.200e+02 6.651e+02 1.076e+03, threshold=1.040e+03, percent-clipped=0.0 2023-04-27 14:01:18,146 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3160, 4.1412, 4.5484, 4.6497, 4.8141, 4.2393, 4.4498, 4.5023], device='cuda:7'), covar=tensor([0.0225, 0.0249, 0.0498, 0.0344, 0.0296, 0.0254, 0.0458, 0.0192], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0099, 0.0125, 0.0119, 0.0131, 0.0109, 0.0124, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:7') 2023-04-27 14:01:22,950 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4845.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:01:31,751 INFO [train.py:904] (7/8) Epoch 1, batch 4850, loss[loss=0.2977, simple_loss=0.3698, pruned_loss=0.1128, over 16678.00 frames. ], tot_loss[loss=0.3154, simple_loss=0.3749, pruned_loss=0.128, over 3188216.16 frames. ], batch size: 134, lr: 4.24e-02, grad_scale: 8.0 2023-04-27 14:01:43,621 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.15 vs. limit=5.0 2023-04-27 14:02:49,101 INFO [train.py:904] (7/8) Epoch 1, batch 4900, loss[loss=0.296, simple_loss=0.3605, pruned_loss=0.1158, over 16689.00 frames. ], tot_loss[loss=0.3132, simple_loss=0.3734, pruned_loss=0.1264, over 3175836.18 frames. ], batch size: 62, lr: 4.23e-02, grad_scale: 8.0 2023-04-27 14:03:07,685 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.780e+02 4.613e+02 5.959e+02 1.211e+03, threshold=9.227e+02, percent-clipped=3.0 2023-04-27 14:03:35,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2900, 4.3892, 4.7945, 4.6881, 4.9225, 4.3395, 4.4462, 4.6493], device='cuda:7'), covar=tensor([0.0257, 0.0188, 0.0324, 0.0372, 0.0242, 0.0210, 0.0433, 0.0201], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0098, 0.0124, 0.0121, 0.0131, 0.0106, 0.0126, 0.0095], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:7') 2023-04-27 14:03:47,115 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9592, 4.1269, 3.8785, 4.0069, 3.5760, 3.9658, 3.8597, 3.9778], device='cuda:7'), covar=tensor([0.0407, 0.0580, 0.0580, 0.0346, 0.0752, 0.0514, 0.0509, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0168, 0.0150, 0.0109, 0.0140, 0.0118, 0.0148, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-27 14:03:55,014 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4944.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:04:04,729 INFO [train.py:904] (7/8) Epoch 1, batch 4950, loss[loss=0.3322, simple_loss=0.3924, pruned_loss=0.136, over 15390.00 frames. ], tot_loss[loss=0.3138, simple_loss=0.3738, pruned_loss=0.1268, over 3180036.88 frames. ], batch size: 190, lr: 4.21e-02, grad_scale: 8.0 2023-04-27 14:05:02,202 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2499, 3.2608, 3.2794, 1.5689, 3.1847, 3.2237, 3.0889, 2.9483], device='cuda:7'), covar=tensor([0.0116, 0.0156, 0.0145, 0.1956, 0.0185, 0.0208, 0.0139, 0.0334], device='cuda:7'), in_proj_covar=tensor([0.0054, 0.0060, 0.0056, 0.0126, 0.0053, 0.0055, 0.0054, 0.0069], device='cuda:7'), out_proj_covar=tensor([7.1498e-05, 8.1343e-05, 8.1176e-05, 1.7236e-04, 8.0998e-05, 7.7402e-05, 8.4967e-05, 9.2039e-05], device='cuda:7') 2023-04-27 14:05:04,817 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=4992.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:05:17,370 INFO [train.py:904] (7/8) Epoch 1, batch 5000, loss[loss=0.31, simple_loss=0.3798, pruned_loss=0.1201, over 16820.00 frames. ], tot_loss[loss=0.3145, simple_loss=0.3756, pruned_loss=0.1267, over 3191453.30 frames. ], batch size: 96, lr: 4.20e-02, grad_scale: 8.0 2023-04-27 14:05:35,392 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.744e+02 4.820e+02 5.822e+02 7.344e+02 1.526e+03, threshold=1.164e+03, percent-clipped=12.0 2023-04-27 14:06:09,443 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3510, 3.0946, 3.1263, 3.3157, 2.8682, 3.2050, 3.0885, 3.0240], device='cuda:7'), covar=tensor([0.0262, 0.0173, 0.0203, 0.0150, 0.0763, 0.0225, 0.0748, 0.0272], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0062, 0.0111, 0.0084, 0.0138, 0.0089, 0.0077, 0.0089], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 14:06:15,467 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7749, 3.8878, 3.6065, 1.8027, 2.9019, 2.2378, 3.4017, 4.1741], device='cuda:7'), covar=tensor([0.0313, 0.0329, 0.0281, 0.2023, 0.0918, 0.1162, 0.0781, 0.0138], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0054, 0.0084, 0.0129, 0.0118, 0.0116, 0.0103, 0.0050], device='cuda:7'), out_proj_covar=tensor([1.1320e-04, 9.0460e-05, 1.1236e-04, 1.5894e-04, 1.5553e-04, 1.4524e-04, 1.5068e-04, 8.0700e-05], device='cuda:7') 2023-04-27 14:06:29,425 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-27 14:06:31,099 INFO [train.py:904] (7/8) Epoch 1, batch 5050, loss[loss=0.2985, simple_loss=0.369, pruned_loss=0.114, over 16452.00 frames. ], tot_loss[loss=0.3132, simple_loss=0.375, pruned_loss=0.1257, over 3194653.72 frames. ], batch size: 146, lr: 4.19e-02, grad_scale: 8.0 2023-04-27 14:07:42,591 INFO [train.py:904] (7/8) Epoch 1, batch 5100, loss[loss=0.2752, simple_loss=0.3482, pruned_loss=0.1011, over 16869.00 frames. ], tot_loss[loss=0.3074, simple_loss=0.3704, pruned_loss=0.1222, over 3212966.31 frames. ], batch size: 116, lr: 4.18e-02, grad_scale: 8.0 2023-04-27 14:07:47,646 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-27 14:07:59,831 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.617e+02 4.246e+02 5.535e+02 6.678e+02 1.197e+03, threshold=1.107e+03, percent-clipped=1.0 2023-04-27 14:08:58,110 INFO [train.py:904] (7/8) Epoch 1, batch 5150, loss[loss=0.2907, simple_loss=0.3587, pruned_loss=0.1113, over 17000.00 frames. ], tot_loss[loss=0.3061, simple_loss=0.3703, pruned_loss=0.121, over 3208984.21 frames. ], batch size: 50, lr: 4.17e-02, grad_scale: 8.0 2023-04-27 14:10:12,912 INFO [train.py:904] (7/8) Epoch 1, batch 5200, loss[loss=0.299, simple_loss=0.3543, pruned_loss=0.1218, over 17109.00 frames. ], tot_loss[loss=0.3057, simple_loss=0.3694, pruned_loss=0.121, over 3204819.18 frames. ], batch size: 47, lr: 4.16e-02, grad_scale: 8.0 2023-04-27 14:10:30,247 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.582e+02 4.001e+02 4.794e+02 6.093e+02 1.086e+03, threshold=9.588e+02, percent-clipped=0.0 2023-04-27 14:11:26,087 INFO [train.py:904] (7/8) Epoch 1, batch 5250, loss[loss=0.2972, simple_loss=0.3676, pruned_loss=0.1133, over 16429.00 frames. ], tot_loss[loss=0.3056, simple_loss=0.3679, pruned_loss=0.1216, over 3199149.45 frames. ], batch size: 146, lr: 4.15e-02, grad_scale: 8.0 2023-04-27 14:12:01,721 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-27 14:12:37,176 INFO [train.py:904] (7/8) Epoch 1, batch 5300, loss[loss=0.2769, simple_loss=0.3432, pruned_loss=0.1053, over 15407.00 frames. ], tot_loss[loss=0.3018, simple_loss=0.3639, pruned_loss=0.1198, over 3199806.11 frames. ], batch size: 190, lr: 4.14e-02, grad_scale: 8.0 2023-04-27 14:12:54,710 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.864e+02 4.537e+02 5.350e+02 6.217e+02 1.130e+03, threshold=1.070e+03, percent-clipped=3.0 2023-04-27 14:13:49,537 INFO [train.py:904] (7/8) Epoch 1, batch 5350, loss[loss=0.2962, simple_loss=0.3623, pruned_loss=0.1151, over 16836.00 frames. ], tot_loss[loss=0.2989, simple_loss=0.3616, pruned_loss=0.1181, over 3207817.64 frames. ], batch size: 116, lr: 4.13e-02, grad_scale: 8.0 2023-04-27 14:14:11,725 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5277, 3.6367, 3.6237, 1.4927, 3.4875, 3.7008, 3.5070, 3.5448], device='cuda:7'), covar=tensor([0.0136, 0.0207, 0.0137, 0.2037, 0.0217, 0.0192, 0.0121, 0.0227], device='cuda:7'), in_proj_covar=tensor([0.0059, 0.0064, 0.0057, 0.0134, 0.0058, 0.0057, 0.0060, 0.0076], device='cuda:7'), out_proj_covar=tensor([8.2746e-05, 9.0430e-05, 8.4843e-05, 1.8717e-04, 9.0452e-05, 8.3417e-05, 9.6333e-05, 1.0526e-04], device='cuda:7') 2023-04-27 14:14:40,196 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8455, 1.4248, 1.4565, 1.6818, 1.5515, 1.6175, 2.0642, 1.9008], device='cuda:7'), covar=tensor([0.0116, 0.0479, 0.0231, 0.0252, 0.0183, 0.0256, 0.0165, 0.0309], device='cuda:7'), in_proj_covar=tensor([0.0029, 0.0052, 0.0033, 0.0031, 0.0032, 0.0036, 0.0029, 0.0032], device='cuda:7'), out_proj_covar=tensor([3.1174e-05, 7.0668e-05, 3.8648e-05, 3.8638e-05, 3.3952e-05, 3.8905e-05, 3.4292e-05, 3.6737e-05], device='cuda:7') 2023-04-27 14:15:00,992 INFO [train.py:904] (7/8) Epoch 1, batch 5400, loss[loss=0.3564, simple_loss=0.41, pruned_loss=0.1514, over 16913.00 frames. ], tot_loss[loss=0.3031, simple_loss=0.3656, pruned_loss=0.1203, over 3213364.11 frames. ], batch size: 116, lr: 4.12e-02, grad_scale: 8.0 2023-04-27 14:15:18,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 4.427e+02 5.350e+02 6.399e+02 9.942e+02, threshold=1.070e+03, percent-clipped=0.0 2023-04-27 14:16:19,555 INFO [train.py:904] (7/8) Epoch 1, batch 5450, loss[loss=0.3507, simple_loss=0.4063, pruned_loss=0.1475, over 16851.00 frames. ], tot_loss[loss=0.3108, simple_loss=0.3716, pruned_loss=0.125, over 3214525.68 frames. ], batch size: 116, lr: 4.11e-02, grad_scale: 8.0 2023-04-27 14:17:37,168 INFO [train.py:904] (7/8) Epoch 1, batch 5500, loss[loss=0.3706, simple_loss=0.4178, pruned_loss=0.1617, over 16742.00 frames. ], tot_loss[loss=0.3278, simple_loss=0.3838, pruned_loss=0.1359, over 3168563.86 frames. ], batch size: 124, lr: 4.10e-02, grad_scale: 8.0 2023-04-27 14:17:46,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8960, 3.5464, 3.7796, 3.9500, 3.3940, 3.7909, 3.5785, 3.4897], device='cuda:7'), covar=tensor([0.0262, 0.0182, 0.0189, 0.0112, 0.0690, 0.0171, 0.0385, 0.0235], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0061, 0.0110, 0.0084, 0.0135, 0.0087, 0.0075, 0.0089], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 14:17:56,202 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.545e+02 5.862e+02 7.575e+02 9.403e+02 2.285e+03, threshold=1.515e+03, percent-clipped=16.0 2023-04-27 14:18:57,395 INFO [train.py:904] (7/8) Epoch 1, batch 5550, loss[loss=0.3904, simple_loss=0.4332, pruned_loss=0.1738, over 16267.00 frames. ], tot_loss[loss=0.3424, simple_loss=0.3942, pruned_loss=0.1453, over 3164699.95 frames. ], batch size: 165, lr: 4.09e-02, grad_scale: 8.0 2023-04-27 14:20:17,853 INFO [train.py:904] (7/8) Epoch 1, batch 5600, loss[loss=0.4726, simple_loss=0.4681, pruned_loss=0.2385, over 10834.00 frames. ], tot_loss[loss=0.3511, simple_loss=0.4, pruned_loss=0.1511, over 3144445.98 frames. ], batch size: 246, lr: 4.08e-02, grad_scale: 8.0 2023-04-27 14:20:37,627 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.538e+02 5.637e+02 6.821e+02 8.546e+02 1.933e+03, threshold=1.364e+03, percent-clipped=3.0 2023-04-27 14:21:39,929 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5650.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:21:40,598 INFO [train.py:904] (7/8) Epoch 1, batch 5650, loss[loss=0.3556, simple_loss=0.4121, pruned_loss=0.1496, over 16777.00 frames. ], tot_loss[loss=0.3669, simple_loss=0.4097, pruned_loss=0.162, over 3066123.44 frames. ], batch size: 83, lr: 4.07e-02, grad_scale: 8.0 2023-04-27 14:22:59,475 INFO [train.py:904] (7/8) Epoch 1, batch 5700, loss[loss=0.3896, simple_loss=0.4363, pruned_loss=0.1714, over 16476.00 frames. ], tot_loss[loss=0.37, simple_loss=0.412, pruned_loss=0.164, over 3060482.19 frames. ], batch size: 75, lr: 4.06e-02, grad_scale: 8.0 2023-04-27 14:23:16,240 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5711.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:23:17,262 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6458, 3.7169, 3.6104, 3.6343, 3.6076, 4.0280, 4.0000, 3.6550], device='cuda:7'), covar=tensor([0.1270, 0.1291, 0.1087, 0.1718, 0.2805, 0.0867, 0.0814, 0.1858], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0186, 0.0156, 0.0168, 0.0206, 0.0150, 0.0154, 0.0220], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:23:17,962 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.964e+02 6.202e+02 7.607e+02 9.700e+02 2.079e+03, threshold=1.521e+03, percent-clipped=5.0 2023-04-27 14:23:33,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5188, 4.4053, 4.4219, 4.9248, 4.7773, 4.6725, 4.8693, 4.6310], device='cuda:7'), covar=tensor([0.0374, 0.0318, 0.1002, 0.0283, 0.0419, 0.0304, 0.0331, 0.0296], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0139, 0.0228, 0.0161, 0.0137, 0.0141, 0.0122, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:24:21,188 INFO [train.py:904] (7/8) Epoch 1, batch 5750, loss[loss=0.4174, simple_loss=0.4303, pruned_loss=0.2023, over 11353.00 frames. ], tot_loss[loss=0.3725, simple_loss=0.4147, pruned_loss=0.1652, over 3040462.73 frames. ], batch size: 246, lr: 4.05e-02, grad_scale: 8.0 2023-04-27 14:24:26,951 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9173, 3.0507, 3.3878, 3.3893, 3.4316, 3.0894, 3.2786, 3.3378], device='cuda:7'), covar=tensor([0.0341, 0.0322, 0.0443, 0.0361, 0.0331, 0.0356, 0.0526, 0.0279], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0099, 0.0127, 0.0122, 0.0133, 0.0107, 0.0137, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 14:24:28,931 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5755.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:25:12,863 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-04-27 14:25:42,940 INFO [train.py:904] (7/8) Epoch 1, batch 5800, loss[loss=0.3867, simple_loss=0.4143, pruned_loss=0.1796, over 12206.00 frames. ], tot_loss[loss=0.3711, simple_loss=0.4139, pruned_loss=0.1642, over 3005992.19 frames. ], batch size: 247, lr: 4.04e-02, grad_scale: 8.0 2023-04-27 14:25:56,705 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3916, 2.9126, 2.8682, 3.7671, 2.7417, 3.3885, 2.9480, 2.7117], device='cuda:7'), covar=tensor([0.0283, 0.0364, 0.0318, 0.0205, 0.1073, 0.0222, 0.0504, 0.0947], device='cuda:7'), in_proj_covar=tensor([0.0079, 0.0082, 0.0072, 0.0082, 0.0151, 0.0079, 0.0102, 0.0094], device='cuda:7'), out_proj_covar=tensor([1.0076e-04, 9.9304e-05, 8.6868e-05, 1.0589e-04, 1.8702e-04, 9.5908e-05, 1.1363e-04, 1.2513e-04], device='cuda:7') 2023-04-27 14:26:01,841 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.312e+02 5.567e+02 6.621e+02 8.933e+02 1.804e+03, threshold=1.324e+03, percent-clipped=2.0 2023-04-27 14:26:06,973 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5816.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:27:02,352 INFO [train.py:904] (7/8) Epoch 1, batch 5850, loss[loss=0.3523, simple_loss=0.4122, pruned_loss=0.1462, over 16929.00 frames. ], tot_loss[loss=0.3667, simple_loss=0.4113, pruned_loss=0.161, over 3033573.80 frames. ], batch size: 109, lr: 4.03e-02, grad_scale: 8.0 2023-04-27 14:27:08,052 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4181, 3.1900, 3.2468, 3.4167, 3.0446, 3.2608, 3.2310, 3.1739], device='cuda:7'), covar=tensor([0.0188, 0.0144, 0.0163, 0.0112, 0.0563, 0.0180, 0.0508, 0.0189], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0060, 0.0105, 0.0084, 0.0134, 0.0083, 0.0075, 0.0087], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 14:27:15,032 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5858.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:27:38,533 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9802, 4.1193, 3.8101, 2.8570, 3.9361, 4.0437, 4.1688, 2.7894], device='cuda:7'), covar=tensor([0.1096, 0.0073, 0.0107, 0.0325, 0.0081, 0.0064, 0.0095, 0.0344], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0036, 0.0038, 0.0055, 0.0035, 0.0036, 0.0041, 0.0051], device='cuda:7'), out_proj_covar=tensor([1.5791e-04, 7.0778e-05, 7.4765e-05, 1.0108e-04, 6.6877e-05, 7.6181e-05, 7.8028e-05, 9.6734e-05], device='cuda:7') 2023-04-27 14:28:25,819 INFO [train.py:904] (7/8) Epoch 1, batch 5900, loss[loss=0.2827, simple_loss=0.36, pruned_loss=0.1027, over 16664.00 frames. ], tot_loss[loss=0.3628, simple_loss=0.409, pruned_loss=0.1583, over 3043021.59 frames. ], batch size: 89, lr: 4.02e-02, grad_scale: 8.0 2023-04-27 14:28:48,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.490e+02 5.425e+02 6.564e+02 8.517e+02 1.690e+03, threshold=1.313e+03, percent-clipped=2.0 2023-04-27 14:28:58,989 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5919.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:29:18,025 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8084, 2.0888, 2.1363, 3.2987, 3.3045, 3.0520, 2.9724, 3.2132], device='cuda:7'), covar=tensor([0.0091, 0.0702, 0.0345, 0.0097, 0.0050, 0.0163, 0.0126, 0.0100], device='cuda:7'), in_proj_covar=tensor([0.0036, 0.0079, 0.0057, 0.0042, 0.0036, 0.0037, 0.0045, 0.0037], device='cuda:7'), out_proj_covar=tensor([5.8146e-05, 1.4169e-04, 1.0270e-04, 6.9244e-05, 5.8577e-05, 6.0002e-05, 6.8127e-05, 6.1364e-05], device='cuda:7') 2023-04-27 14:29:49,235 INFO [train.py:904] (7/8) Epoch 1, batch 5950, loss[loss=0.2991, simple_loss=0.3634, pruned_loss=0.1174, over 16615.00 frames. ], tot_loss[loss=0.3599, simple_loss=0.4085, pruned_loss=0.1557, over 3036147.35 frames. ], batch size: 76, lr: 4.01e-02, grad_scale: 8.0 2023-04-27 14:30:52,666 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5990.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:31:14,071 INFO [train.py:904] (7/8) Epoch 1, batch 6000, loss[loss=0.3951, simple_loss=0.4192, pruned_loss=0.1855, over 11964.00 frames. ], tot_loss[loss=0.3601, simple_loss=0.4081, pruned_loss=0.156, over 3038234.98 frames. ], batch size: 248, lr: 4.00e-02, grad_scale: 8.0 2023-04-27 14:31:14,072 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 14:31:23,947 INFO [train.py:938] (7/8) Epoch 1, validation: loss=0.2762, simple_loss=0.3752, pruned_loss=0.08863, over 944034.00 frames. 2023-04-27 14:31:23,948 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 14:31:31,602 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6006.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 14:31:41,461 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.294e+02 5.611e+02 7.114e+02 9.006e+02 1.900e+03, threshold=1.423e+03, percent-clipped=2.0 2023-04-27 14:32:30,101 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6043.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:32:42,921 INFO [train.py:904] (7/8) Epoch 1, batch 6050, loss[loss=0.3321, simple_loss=0.3998, pruned_loss=0.1322, over 16976.00 frames. ], tot_loss[loss=0.3567, simple_loss=0.4056, pruned_loss=0.1539, over 3055765.96 frames. ], batch size: 55, lr: 3.99e-02, grad_scale: 8.0 2023-04-27 14:32:44,847 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6051.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:34:03,531 INFO [train.py:904] (7/8) Epoch 1, batch 6100, loss[loss=0.3546, simple_loss=0.4009, pruned_loss=0.1541, over 15469.00 frames. ], tot_loss[loss=0.3517, simple_loss=0.4025, pruned_loss=0.1505, over 3069526.28 frames. ], batch size: 191, lr: 3.98e-02, grad_scale: 8.0 2023-04-27 14:34:09,268 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6104.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:34:21,419 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6111.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:34:24,017 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.820e+02 4.604e+02 6.189e+02 8.357e+02 1.892e+03, threshold=1.238e+03, percent-clipped=2.0 2023-04-27 14:34:45,863 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6062, 3.5519, 3.9908, 4.0433, 4.1140, 3.5395, 3.8172, 3.9548], device='cuda:7'), covar=tensor([0.0329, 0.0310, 0.0432, 0.0437, 0.0376, 0.0382, 0.0573, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0110, 0.0138, 0.0135, 0.0144, 0.0119, 0.0154, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 14:34:59,880 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6135.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:35:26,097 INFO [train.py:904] (7/8) Epoch 1, batch 6150, loss[loss=0.2968, simple_loss=0.3654, pruned_loss=0.1141, over 16885.00 frames. ], tot_loss[loss=0.3484, simple_loss=0.3993, pruned_loss=0.1488, over 3073911.09 frames. ], batch size: 96, lr: 3.97e-02, grad_scale: 8.0 2023-04-27 14:35:46,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6351, 1.3725, 1.4718, 1.3644, 1.7589, 1.5361, 1.6326, 1.8449], device='cuda:7'), covar=tensor([0.0115, 0.0417, 0.0205, 0.0176, 0.0118, 0.0203, 0.0155, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0031, 0.0057, 0.0037, 0.0034, 0.0036, 0.0038, 0.0030, 0.0031], device='cuda:7'), out_proj_covar=tensor([3.5539e-05, 8.0076e-05, 4.6248e-05, 4.3958e-05, 4.3139e-05, 4.5980e-05, 4.0098e-05, 3.8806e-05], device='cuda:7') 2023-04-27 14:35:57,312 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6170.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:36:15,029 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-27 14:36:39,088 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6196.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:36:46,267 INFO [train.py:904] (7/8) Epoch 1, batch 6200, loss[loss=0.4546, simple_loss=0.4525, pruned_loss=0.2283, over 11575.00 frames. ], tot_loss[loss=0.3463, simple_loss=0.397, pruned_loss=0.1478, over 3092285.03 frames. ], batch size: 250, lr: 3.96e-02, grad_scale: 8.0 2023-04-27 14:36:49,306 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6203.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:36:54,700 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.4333, 3.4323, 3.3390, 2.0093, 3.1580, 3.3674, 3.3627, 1.9891], device='cuda:7'), covar=tensor([0.1284, 0.0079, 0.0080, 0.0467, 0.0114, 0.0106, 0.0139, 0.0513], device='cuda:7'), in_proj_covar=tensor([0.0097, 0.0037, 0.0038, 0.0064, 0.0037, 0.0037, 0.0044, 0.0060], device='cuda:7'), out_proj_covar=tensor([1.7861e-04, 7.4627e-05, 7.8480e-05, 1.2177e-04, 7.5011e-05, 8.0708e-05, 8.5598e-05, 1.1627e-04], device='cuda:7') 2023-04-27 14:37:03,980 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6211.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:37:05,931 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.990e+02 5.160e+02 6.786e+02 8.603e+02 1.824e+03, threshold=1.357e+03, percent-clipped=7.0 2023-04-27 14:37:08,096 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6214.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:37:10,790 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0733, 3.6981, 3.6430, 3.2812, 3.8680, 3.1867, 3.6577, 3.8340], device='cuda:7'), covar=tensor([0.0094, 0.0107, 0.0096, 0.0380, 0.0081, 0.0338, 0.0090, 0.0109], device='cuda:7'), in_proj_covar=tensor([0.0046, 0.0036, 0.0051, 0.0071, 0.0039, 0.0062, 0.0050, 0.0048], device='cuda:7'), out_proj_covar=tensor([1.0865e-04, 8.2792e-05, 1.2730e-04, 1.5447e-04, 8.7756e-05, 1.4125e-04, 1.2612e-04, 1.2639e-04], device='cuda:7') 2023-04-27 14:37:33,325 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6231.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:37:38,248 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-27 14:37:57,629 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-27 14:38:02,380 INFO [train.py:904] (7/8) Epoch 1, batch 6250, loss[loss=0.3337, simple_loss=0.3943, pruned_loss=0.1366, over 16318.00 frames. ], tot_loss[loss=0.3445, simple_loss=0.3961, pruned_loss=0.1464, over 3103581.80 frames. ], batch size: 165, lr: 3.95e-02, grad_scale: 8.0 2023-04-27 14:38:20,951 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6264.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:24,226 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6266.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:32,100 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6272.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:38:45,255 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.3740, 1.4858, 1.4732, 1.1770, 1.6297, 1.4948, 1.2826, 1.5529], device='cuda:7'), covar=tensor([0.0093, 0.0271, 0.0153, 0.0176, 0.0113, 0.0145, 0.0166, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0029, 0.0057, 0.0038, 0.0036, 0.0035, 0.0039, 0.0031, 0.0030], device='cuda:7'), out_proj_covar=tensor([3.4777e-05, 8.1400e-05, 4.7597e-05, 4.8235e-05, 4.3225e-05, 4.7120e-05, 4.2471e-05, 3.9248e-05], device='cuda:7') 2023-04-27 14:39:17,033 INFO [train.py:904] (7/8) Epoch 1, batch 6300, loss[loss=0.3527, simple_loss=0.4038, pruned_loss=0.1508, over 16239.00 frames. ], tot_loss[loss=0.3447, simple_loss=0.3965, pruned_loss=0.1464, over 3095582.48 frames. ], batch size: 165, lr: 3.94e-02, grad_scale: 8.0 2023-04-27 14:39:24,414 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-27 14:39:26,273 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6306.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 14:39:36,720 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.400e+02 5.487e+02 6.666e+02 8.343e+02 1.856e+03, threshold=1.333e+03, percent-clipped=2.0 2023-04-27 14:39:57,711 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7073, 3.6599, 4.0249, 4.1238, 4.2403, 3.8431, 3.7377, 4.1137], device='cuda:7'), covar=tensor([0.0434, 0.0440, 0.0615, 0.0578, 0.0487, 0.0422, 0.0872, 0.0334], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0113, 0.0134, 0.0134, 0.0144, 0.0118, 0.0157, 0.0111], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:39:59,658 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6327.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:40:29,464 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6346.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:40:36,334 INFO [train.py:904] (7/8) Epoch 1, batch 6350, loss[loss=0.3122, simple_loss=0.3773, pruned_loss=0.1235, over 16846.00 frames. ], tot_loss[loss=0.3489, simple_loss=0.3988, pruned_loss=0.1495, over 3097243.20 frames. ], batch size: 42, lr: 3.93e-02, grad_scale: 8.0 2023-04-27 14:40:41,032 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6354.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 14:41:15,050 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0787, 3.8917, 3.2499, 3.9962, 3.2206, 2.3629, 4.3036, 4.8394], device='cuda:7'), covar=tensor([0.1487, 0.0536, 0.0959, 0.0224, 0.1595, 0.1184, 0.0222, 0.0038], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0134, 0.0168, 0.0100, 0.0176, 0.0136, 0.0104, 0.0059], device='cuda:7'), out_proj_covar=tensor([2.1618e-04, 1.6561e-04, 1.8316e-04, 1.1697e-04, 2.1474e-04, 1.5908e-04, 1.2583e-04, 7.3933e-05], device='cuda:7') 2023-04-27 14:41:33,847 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2023-04-27 14:41:39,811 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2023-04-27 14:41:50,868 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6399.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:41:53,496 INFO [train.py:904] (7/8) Epoch 1, batch 6400, loss[loss=0.2953, simple_loss=0.3626, pruned_loss=0.114, over 16820.00 frames. ], tot_loss[loss=0.3521, simple_loss=0.4003, pruned_loss=0.152, over 3080194.94 frames. ], batch size: 102, lr: 3.92e-02, grad_scale: 8.0 2023-04-27 14:42:08,622 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6411.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:42:08,773 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9777, 2.5845, 2.3111, 3.2144, 2.5432, 2.9396, 2.6247, 2.3218], device='cuda:7'), covar=tensor([0.0265, 0.0290, 0.0287, 0.0234, 0.0780, 0.0209, 0.0460, 0.0814], device='cuda:7'), in_proj_covar=tensor([0.0092, 0.0094, 0.0079, 0.0097, 0.0168, 0.0091, 0.0116, 0.0114], device='cuda:7'), out_proj_covar=tensor([1.2122e-04, 1.1772e-04, 9.8905e-05, 1.2858e-04, 2.1577e-04, 1.1388e-04, 1.3135e-04, 1.5217e-04], device='cuda:7') 2023-04-27 14:42:10,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 4.039e+02 5.999e+02 7.400e+02 9.043e+02 1.587e+03, threshold=1.480e+03, percent-clipped=2.0 2023-04-27 14:43:09,623 INFO [train.py:904] (7/8) Epoch 1, batch 6450, loss[loss=0.315, simple_loss=0.3501, pruned_loss=0.14, over 11688.00 frames. ], tot_loss[loss=0.3482, simple_loss=0.3979, pruned_loss=0.1493, over 3085441.52 frames. ], batch size: 248, lr: 3.91e-02, grad_scale: 8.0 2023-04-27 14:43:22,745 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6459.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:43:24,510 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-27 14:43:26,570 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6462.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:43:33,824 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6467.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:44:12,822 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6491.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:44:26,980 INFO [train.py:904] (7/8) Epoch 1, batch 6500, loss[loss=0.3146, simple_loss=0.3693, pruned_loss=0.1299, over 16996.00 frames. ], tot_loss[loss=0.3434, simple_loss=0.3936, pruned_loss=0.1466, over 3094861.32 frames. ], batch size: 41, lr: 3.90e-02, grad_scale: 16.0 2023-04-27 14:44:45,187 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.861e+02 4.801e+02 5.896e+02 8.038e+02 1.836e+03, threshold=1.179e+03, percent-clipped=2.0 2023-04-27 14:44:46,926 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6514.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:45:00,573 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6523.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:45:05,063 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6526.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:45:05,436 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-27 14:45:07,737 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6528.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:45:42,991 INFO [train.py:904] (7/8) Epoch 1, batch 6550, loss[loss=0.3103, simple_loss=0.3947, pruned_loss=0.1129, over 16794.00 frames. ], tot_loss[loss=0.3487, simple_loss=0.3982, pruned_loss=0.1496, over 3055434.21 frames. ], batch size: 96, lr: 3.89e-02, grad_scale: 16.0 2023-04-27 14:45:55,682 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6559.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:46:00,537 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6562.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:46:08,812 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6567.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:46:12,817 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5833, 5.0132, 4.9332, 4.8007, 4.9429, 5.3169, 5.1435, 4.7564], device='cuda:7'), covar=tensor([0.0674, 0.0851, 0.0708, 0.1147, 0.1541, 0.0539, 0.0600, 0.1474], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0193, 0.0159, 0.0169, 0.0208, 0.0156, 0.0151, 0.0224], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 14:46:23,878 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.04 vs. limit=5.0 2023-04-27 14:46:27,258 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0194, 4.1088, 3.5144, 1.8057, 3.0316, 2.3998, 3.4717, 4.1742], device='cuda:7'), covar=tensor([0.0319, 0.0211, 0.0302, 0.1972, 0.0829, 0.1141, 0.0738, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0090, 0.0063, 0.0100, 0.0144, 0.0138, 0.0133, 0.0125, 0.0060], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 14:46:59,698 INFO [train.py:904] (7/8) Epoch 1, batch 6600, loss[loss=0.3897, simple_loss=0.4308, pruned_loss=0.1743, over 16671.00 frames. ], tot_loss[loss=0.3506, simple_loss=0.401, pruned_loss=0.1501, over 3072284.23 frames. ], batch size: 57, lr: 3.89e-02, grad_scale: 16.0 2023-04-27 14:47:10,169 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-27 14:47:18,179 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.822e+02 5.266e+02 6.465e+02 8.044e+02 1.550e+03, threshold=1.293e+03, percent-clipped=3.0 2023-04-27 14:47:32,888 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6622.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:47:43,255 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6629.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:47:51,557 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2252, 3.8308, 4.0562, 4.2134, 3.5548, 4.0498, 3.9531, 3.7060], device='cuda:7'), covar=tensor([0.0239, 0.0190, 0.0191, 0.0135, 0.0856, 0.0186, 0.0272, 0.0259], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0065, 0.0116, 0.0091, 0.0141, 0.0091, 0.0081, 0.0096], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:48:10,820 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6646.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:48:18,091 INFO [train.py:904] (7/8) Epoch 1, batch 6650, loss[loss=0.3199, simple_loss=0.3874, pruned_loss=0.1262, over 16350.00 frames. ], tot_loss[loss=0.3487, simple_loss=0.3998, pruned_loss=0.1488, over 3104556.10 frames. ], batch size: 146, lr: 3.88e-02, grad_scale: 16.0 2023-04-27 14:49:18,705 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6690.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:49:19,275 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-04-27 14:49:24,166 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6694.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:49:32,245 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6699.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:49:34,326 INFO [train.py:904] (7/8) Epoch 1, batch 6700, loss[loss=0.4166, simple_loss=0.4331, pruned_loss=0.2001, over 11446.00 frames. ], tot_loss[loss=0.348, simple_loss=0.3983, pruned_loss=0.1488, over 3113784.67 frames. ], batch size: 247, lr: 3.87e-02, grad_scale: 16.0 2023-04-27 14:49:52,617 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.128e+02 5.771e+02 7.025e+02 8.722e+02 1.711e+03, threshold=1.405e+03, percent-clipped=7.0 2023-04-27 14:50:14,902 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2086, 4.1815, 4.0252, 1.5296, 4.1497, 4.2000, 3.5387, 3.5841], device='cuda:7'), covar=tensor([0.0240, 0.0074, 0.0167, 0.2220, 0.0079, 0.0045, 0.0168, 0.0202], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0066, 0.0062, 0.0146, 0.0061, 0.0055, 0.0067, 0.0083], device='cuda:7'), out_proj_covar=tensor([1.1784e-04, 1.0262e-04, 1.0282e-04, 2.1876e-04, 9.9517e-05, 8.9419e-05, 1.1818e-04, 1.2725e-04], device='cuda:7') 2023-04-27 14:50:38,883 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4283, 4.9547, 4.7885, 4.9419, 4.9149, 5.2933, 5.1303, 4.8148], device='cuda:7'), covar=tensor([0.0618, 0.0762, 0.0667, 0.0875, 0.1414, 0.0520, 0.0630, 0.1318], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0198, 0.0166, 0.0173, 0.0216, 0.0165, 0.0160, 0.0230], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:50:45,228 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6747.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:50:50,990 INFO [train.py:904] (7/8) Epoch 1, batch 6750, loss[loss=0.2988, simple_loss=0.3595, pruned_loss=0.119, over 16781.00 frames. ], tot_loss[loss=0.3467, simple_loss=0.3971, pruned_loss=0.1482, over 3123811.95 frames. ], batch size: 102, lr: 3.86e-02, grad_scale: 16.0 2023-04-27 14:51:26,871 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7961, 3.5829, 3.6581, 3.9054, 3.3099, 3.6577, 3.5962, 3.5090], device='cuda:7'), covar=tensor([0.0254, 0.0162, 0.0198, 0.0110, 0.0665, 0.0177, 0.0440, 0.0198], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0064, 0.0117, 0.0090, 0.0141, 0.0092, 0.0082, 0.0097], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:51:51,534 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6791.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:52:05,972 INFO [train.py:904] (7/8) Epoch 1, batch 6800, loss[loss=0.3322, simple_loss=0.3947, pruned_loss=0.1349, over 16587.00 frames. ], tot_loss[loss=0.3453, simple_loss=0.3959, pruned_loss=0.1473, over 3113143.47 frames. ], batch size: 75, lr: 3.85e-02, grad_scale: 16.0 2023-04-27 14:52:24,955 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.031e+02 5.369e+02 6.197e+02 7.365e+02 1.417e+03, threshold=1.239e+03, percent-clipped=1.0 2023-04-27 14:52:33,898 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6818.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 14:52:40,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2279, 1.5745, 1.6515, 1.3941, 2.1087, 2.0442, 2.4880, 2.1517], device='cuda:7'), covar=tensor([0.0074, 0.0390, 0.0181, 0.0277, 0.0117, 0.0155, 0.0207, 0.0206], device='cuda:7'), in_proj_covar=tensor([0.0030, 0.0062, 0.0043, 0.0042, 0.0037, 0.0043, 0.0031, 0.0033], device='cuda:7'), out_proj_covar=tensor([3.6224e-05, 8.9983e-05, 5.7946e-05, 5.8272e-05, 4.9927e-05, 5.6036e-05, 4.5352e-05, 4.5217e-05], device='cuda:7') 2023-04-27 14:52:41,653 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6823.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:52:46,935 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6826.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:52:59,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9193, 2.9876, 3.3278, 3.3235, 3.4153, 2.9778, 3.2431, 3.3306], device='cuda:7'), covar=tensor([0.0383, 0.0379, 0.0532, 0.0495, 0.0436, 0.0463, 0.0609, 0.0316], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0108, 0.0134, 0.0128, 0.0145, 0.0116, 0.0161, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 14:53:06,063 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6839.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:23,177 INFO [train.py:904] (7/8) Epoch 1, batch 6850, loss[loss=0.4166, simple_loss=0.4334, pruned_loss=0.1999, over 11772.00 frames. ], tot_loss[loss=0.3495, simple_loss=0.3992, pruned_loss=0.1499, over 3091416.40 frames. ], batch size: 247, lr: 3.84e-02, grad_scale: 16.0 2023-04-27 14:53:35,364 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6859.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:46,373 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6867.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:53:57,789 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6874.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:54:03,470 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6878.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:54:37,400 INFO [train.py:904] (7/8) Epoch 1, batch 6900, loss[loss=0.4313, simple_loss=0.454, pruned_loss=0.2043, over 15409.00 frames. ], tot_loss[loss=0.3483, simple_loss=0.4002, pruned_loss=0.1482, over 3086565.79 frames. ], batch size: 191, lr: 3.83e-02, grad_scale: 16.0 2023-04-27 14:54:46,299 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-27 14:54:47,070 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6907.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:54:50,407 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6909.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:54:55,481 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.157e+02 5.195e+02 6.018e+02 7.768e+02 1.319e+03, threshold=1.204e+03, percent-clipped=1.0 2023-04-27 14:54:59,971 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6915.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:55:10,024 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6922.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:55:36,559 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6939.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:55:54,253 INFO [train.py:904] (7/8) Epoch 1, batch 6950, loss[loss=0.3162, simple_loss=0.3857, pruned_loss=0.1234, over 16721.00 frames. ], tot_loss[loss=0.3551, simple_loss=0.4049, pruned_loss=0.1526, over 3068421.49 frames. ], batch size: 89, lr: 3.82e-02, grad_scale: 16.0 2023-04-27 14:56:25,068 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:56:25,204 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:56:47,702 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6985.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 14:57:12,240 INFO [train.py:904] (7/8) Epoch 1, batch 7000, loss[loss=0.4176, simple_loss=0.4318, pruned_loss=0.2016, over 11601.00 frames. ], tot_loss[loss=0.3534, simple_loss=0.4045, pruned_loss=0.1511, over 3063286.44 frames. ], batch size: 248, lr: 3.81e-02, grad_scale: 16.0 2023-04-27 14:57:25,748 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4751, 2.9399, 2.6307, 3.8657, 2.6018, 3.4122, 2.8895, 2.8129], device='cuda:7'), covar=tensor([0.0265, 0.0336, 0.0314, 0.0194, 0.1145, 0.0203, 0.0502, 0.0778], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0100, 0.0085, 0.0109, 0.0178, 0.0097, 0.0122, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 14:57:30,864 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.210e+02 5.287e+02 6.266e+02 8.014e+02 1.368e+03, threshold=1.253e+03, percent-clipped=3.0 2023-04-27 14:58:27,053 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7048.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 14:58:31,408 INFO [train.py:904] (7/8) Epoch 1, batch 7050, loss[loss=0.4042, simple_loss=0.4263, pruned_loss=0.1911, over 11466.00 frames. ], tot_loss[loss=0.3528, simple_loss=0.4046, pruned_loss=0.1506, over 3070365.31 frames. ], batch size: 248, lr: 3.80e-02, grad_scale: 16.0 2023-04-27 14:59:45,975 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4460, 4.9603, 4.7679, 5.0001, 5.0452, 5.3305, 5.2135, 4.8587], device='cuda:7'), covar=tensor([0.0619, 0.0986, 0.0782, 0.1158, 0.1427, 0.0706, 0.0650, 0.1658], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0195, 0.0165, 0.0171, 0.0206, 0.0163, 0.0152, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 14:59:51,875 INFO [train.py:904] (7/8) Epoch 1, batch 7100, loss[loss=0.3478, simple_loss=0.3736, pruned_loss=0.161, over 11426.00 frames. ], tot_loss[loss=0.352, simple_loss=0.4026, pruned_loss=0.1507, over 3026755.35 frames. ], batch size: 247, lr: 3.79e-02, grad_scale: 16.0 2023-04-27 15:00:05,911 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7109.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 15:00:11,240 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.812e+02 5.623e+02 6.703e+02 8.193e+02 2.007e+03, threshold=1.341e+03, percent-clipped=3.0 2023-04-27 15:00:20,588 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7118.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:00:27,655 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7123.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:01:11,094 INFO [train.py:904] (7/8) Epoch 1, batch 7150, loss[loss=0.4049, simple_loss=0.4198, pruned_loss=0.195, over 11469.00 frames. ], tot_loss[loss=0.348, simple_loss=0.3991, pruned_loss=0.1484, over 3057133.67 frames. ], batch size: 246, lr: 3.78e-02, grad_scale: 8.0 2023-04-27 15:01:30,112 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7163.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:01:31,563 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8073, 3.9035, 3.3047, 1.7222, 2.8524, 2.1691, 3.1998, 3.8926], device='cuda:7'), covar=tensor([0.0297, 0.0244, 0.0309, 0.2181, 0.0907, 0.1293, 0.1056, 0.0338], device='cuda:7'), in_proj_covar=tensor([0.0098, 0.0071, 0.0110, 0.0151, 0.0146, 0.0140, 0.0136, 0.0067], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 15:01:34,862 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7166.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:01:42,497 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7171.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:02:27,369 INFO [train.py:904] (7/8) Epoch 1, batch 7200, loss[loss=0.2977, simple_loss=0.3648, pruned_loss=0.1153, over 17206.00 frames. ], tot_loss[loss=0.3445, simple_loss=0.3966, pruned_loss=0.1462, over 3031152.64 frames. ], batch size: 45, lr: 3.78e-02, grad_scale: 8.0 2023-04-27 15:02:46,730 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.108e+02 4.516e+02 5.501e+02 7.177e+02 1.508e+03, threshold=1.100e+03, percent-clipped=3.0 2023-04-27 15:03:03,392 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7224.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 15:03:19,583 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7234.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:03:46,521 INFO [train.py:904] (7/8) Epoch 1, batch 7250, loss[loss=0.3034, simple_loss=0.3595, pruned_loss=0.1236, over 16386.00 frames. ], tot_loss[loss=0.3406, simple_loss=0.3933, pruned_loss=0.144, over 3037698.67 frames. ], batch size: 146, lr: 3.77e-02, grad_scale: 8.0 2023-04-27 15:04:06,711 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7265.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:04:14,406 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-27 15:04:21,337 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7275.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:04:35,929 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7285.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:04:58,728 INFO [train.py:904] (7/8) Epoch 1, batch 7300, loss[loss=0.354, simple_loss=0.4048, pruned_loss=0.1515, over 16240.00 frames. ], tot_loss[loss=0.3393, simple_loss=0.392, pruned_loss=0.1433, over 3047434.36 frames. ], batch size: 165, lr: 3.76e-02, grad_scale: 8.0 2023-04-27 15:05:19,386 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.541e+02 5.301e+02 6.672e+02 8.069e+02 1.507e+03, threshold=1.334e+03, percent-clipped=6.0 2023-04-27 15:05:49,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7333.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:05:50,512 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9378, 3.8601, 4.2806, 4.3410, 4.3793, 3.9190, 4.0356, 4.1795], device='cuda:7'), covar=tensor([0.0230, 0.0259, 0.0311, 0.0295, 0.0293, 0.0243, 0.0585, 0.0246], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0110, 0.0127, 0.0126, 0.0138, 0.0112, 0.0157, 0.0111], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:05:53,668 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7336.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:06:15,338 INFO [train.py:904] (7/8) Epoch 1, batch 7350, loss[loss=0.3084, simple_loss=0.355, pruned_loss=0.1309, over 16498.00 frames. ], tot_loss[loss=0.3365, simple_loss=0.3898, pruned_loss=0.1416, over 3039995.76 frames. ], batch size: 68, lr: 3.75e-02, grad_scale: 8.0 2023-04-27 15:06:15,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6036, 2.5607, 1.6442, 2.6041, 1.8233, 2.5410, 1.7727, 2.1533], device='cuda:7'), covar=tensor([0.0119, 0.0192, 0.1451, 0.0103, 0.0775, 0.0254, 0.1279, 0.0564], device='cuda:7'), in_proj_covar=tensor([0.0059, 0.0064, 0.0136, 0.0061, 0.0117, 0.0067, 0.0146, 0.0106], device='cuda:7'), out_proj_covar=tensor([1.0280e-04, 1.1773e-04, 2.1315e-04, 9.9820e-05, 1.8673e-04, 1.3022e-04, 2.2808e-04, 1.8035e-04], device='cuda:7') 2023-04-27 15:07:03,531 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1934, 2.8284, 2.6933, 3.5524, 2.5133, 3.2435, 2.6149, 2.5670], device='cuda:7'), covar=tensor([0.0292, 0.0306, 0.0262, 0.0209, 0.0963, 0.0192, 0.0532, 0.0898], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0108, 0.0090, 0.0121, 0.0190, 0.0104, 0.0130, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 15:07:30,974 INFO [train.py:904] (7/8) Epoch 1, batch 7400, loss[loss=0.3254, simple_loss=0.3894, pruned_loss=0.1307, over 17018.00 frames. ], tot_loss[loss=0.3381, simple_loss=0.3914, pruned_loss=0.1424, over 3056455.39 frames. ], batch size: 50, lr: 3.74e-02, grad_scale: 8.0 2023-04-27 15:07:36,110 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7404.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:07:40,515 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7407.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:07:50,857 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.431e+02 4.999e+02 6.259e+02 7.546e+02 1.554e+03, threshold=1.252e+03, percent-clipped=1.0 2023-04-27 15:08:25,284 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-27 15:08:52,569 INFO [train.py:904] (7/8) Epoch 1, batch 7450, loss[loss=0.4309, simple_loss=0.4379, pruned_loss=0.2119, over 11583.00 frames. ], tot_loss[loss=0.3416, simple_loss=0.3937, pruned_loss=0.1448, over 3045074.77 frames. ], batch size: 247, lr: 3.73e-02, grad_scale: 8.0 2023-04-27 15:09:22,462 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7468.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:10:15,208 INFO [train.py:904] (7/8) Epoch 1, batch 7500, loss[loss=0.2853, simple_loss=0.36, pruned_loss=0.1053, over 17196.00 frames. ], tot_loss[loss=0.3421, simple_loss=0.3948, pruned_loss=0.1447, over 3048864.67 frames. ], batch size: 45, lr: 3.72e-02, grad_scale: 8.0 2023-04-27 15:10:33,341 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5839, 2.5260, 1.5788, 2.6134, 1.8702, 2.5727, 1.6912, 2.1924], device='cuda:7'), covar=tensor([0.0081, 0.0156, 0.1135, 0.0091, 0.0604, 0.0192, 0.1059, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0060, 0.0064, 0.0136, 0.0061, 0.0117, 0.0070, 0.0145, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:10:35,040 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.564e+02 5.230e+02 6.451e+02 8.258e+02 1.814e+03, threshold=1.290e+03, percent-clipped=2.0 2023-04-27 15:10:43,194 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7519.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:11:06,798 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7534.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:11:24,017 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6927, 5.8605, 5.4845, 5.8361, 5.1092, 5.1730, 5.3881, 5.9766], device='cuda:7'), covar=tensor([0.0293, 0.0526, 0.0706, 0.0231, 0.0550, 0.0292, 0.0348, 0.0372], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0209, 0.0199, 0.0126, 0.0163, 0.0130, 0.0181, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-27 15:11:31,899 INFO [train.py:904] (7/8) Epoch 1, batch 7550, loss[loss=0.3785, simple_loss=0.4161, pruned_loss=0.1705, over 15432.00 frames. ], tot_loss[loss=0.3418, simple_loss=0.3941, pruned_loss=0.1448, over 3055239.86 frames. ], batch size: 191, lr: 3.72e-02, grad_scale: 4.0 2023-04-27 15:11:54,893 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7565.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:12:21,505 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7582.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:12:50,420 INFO [train.py:904] (7/8) Epoch 1, batch 7600, loss[loss=0.3491, simple_loss=0.3916, pruned_loss=0.1533, over 16329.00 frames. ], tot_loss[loss=0.3415, simple_loss=0.393, pruned_loss=0.145, over 3050216.05 frames. ], batch size: 165, lr: 3.71e-02, grad_scale: 8.0 2023-04-27 15:13:10,074 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7613.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:13:12,790 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.798e+02 5.452e+02 6.510e+02 8.243e+02 1.443e+03, threshold=1.302e+03, percent-clipped=3.0 2023-04-27 15:13:26,786 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-04-27 15:13:30,325 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7624.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:13:35,989 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7832, 4.3977, 4.6351, 4.8166, 4.0833, 4.6299, 4.5022, 4.2682], device='cuda:7'), covar=tensor([0.0236, 0.0121, 0.0166, 0.0103, 0.0718, 0.0186, 0.0150, 0.0175], device='cuda:7'), in_proj_covar=tensor([0.0090, 0.0064, 0.0118, 0.0093, 0.0145, 0.0096, 0.0086, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:13:40,336 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7631.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:13:48,382 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-04-27 15:14:13,726 INFO [train.py:904] (7/8) Epoch 1, batch 7650, loss[loss=0.4026, simple_loss=0.4222, pruned_loss=0.1915, over 11479.00 frames. ], tot_loss[loss=0.3428, simple_loss=0.3937, pruned_loss=0.146, over 3033777.00 frames. ], batch size: 247, lr: 3.70e-02, grad_scale: 8.0 2023-04-27 15:14:39,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4496, 4.3563, 1.7754, 4.5628, 2.4388, 4.4327, 2.1689, 3.1586], device='cuda:7'), covar=tensor([0.0041, 0.0092, 0.1763, 0.0043, 0.0891, 0.0112, 0.1328, 0.0582], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0064, 0.0137, 0.0060, 0.0118, 0.0070, 0.0145, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:14:53,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0738, 3.9767, 4.0894, 4.4984, 4.4486, 4.1900, 4.5015, 4.2611], device='cuda:7'), covar=tensor([0.0442, 0.0384, 0.0950, 0.0280, 0.0367, 0.0321, 0.0229, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0167, 0.0258, 0.0178, 0.0155, 0.0160, 0.0139, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:14:56,128 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9608, 2.7458, 2.1422, 3.1632, 3.1271, 3.1742, 2.1583, 2.8283], device='cuda:7'), covar=tensor([0.2047, 0.0278, 0.1589, 0.0114, 0.0182, 0.0285, 0.0942, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0091, 0.0158, 0.0055, 0.0062, 0.0078, 0.0129, 0.0110], device='cuda:7'), out_proj_covar=tensor([2.2616e-04, 1.4248e-04, 2.2173e-04, 9.5849e-05, 1.1183e-04, 1.4728e-04, 1.8940e-04, 1.6787e-04], device='cuda:7') 2023-04-27 15:15:15,266 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7685.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:15:39,930 INFO [train.py:904] (7/8) Epoch 1, batch 7700, loss[loss=0.3381, simple_loss=0.3945, pruned_loss=0.1408, over 16380.00 frames. ], tot_loss[loss=0.3436, simple_loss=0.3942, pruned_loss=0.1466, over 3061239.81 frames. ], batch size: 35, lr: 3.69e-02, grad_scale: 8.0 2023-04-27 15:15:45,470 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7704.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:16:00,938 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.630e+02 5.474e+02 6.817e+02 8.632e+02 3.010e+03, threshold=1.363e+03, percent-clipped=1.0 2023-04-27 15:16:45,717 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7743.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:16:57,069 INFO [train.py:904] (7/8) Epoch 1, batch 7750, loss[loss=0.3154, simple_loss=0.3747, pruned_loss=0.128, over 16472.00 frames. ], tot_loss[loss=0.3435, simple_loss=0.3942, pruned_loss=0.1464, over 3057587.65 frames. ], batch size: 68, lr: 3.68e-02, grad_scale: 8.0 2023-04-27 15:16:59,264 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7752.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 15:17:16,061 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7763.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:18:14,293 INFO [train.py:904] (7/8) Epoch 1, batch 7800, loss[loss=0.2922, simple_loss=0.36, pruned_loss=0.1122, over 16560.00 frames. ], tot_loss[loss=0.3457, simple_loss=0.3956, pruned_loss=0.1479, over 3056158.99 frames. ], batch size: 68, lr: 3.67e-02, grad_scale: 8.0 2023-04-27 15:18:16,723 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7802.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:18:20,004 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7804.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:18:36,279 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.181e+02 4.983e+02 6.289e+02 7.590e+02 1.248e+03, threshold=1.258e+03, percent-clipped=0.0 2023-04-27 15:18:43,896 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7819.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 15:18:51,868 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5888, 3.6521, 2.9102, 3.2971, 2.7103, 2.0236, 3.7910, 4.2806], device='cuda:7'), covar=tensor([0.2037, 0.0688, 0.1274, 0.0478, 0.2097, 0.1731, 0.0280, 0.0057], device='cuda:7'), in_proj_covar=tensor([0.0211, 0.0169, 0.0196, 0.0119, 0.0210, 0.0152, 0.0133, 0.0071], device='cuda:7'), out_proj_covar=tensor([2.5268e-04, 2.0216e-04, 2.1529e-04, 1.4027e-04, 2.5306e-04, 1.8149e-04, 1.6202e-04, 8.9692e-05], device='cuda:7') 2023-04-27 15:19:31,867 INFO [train.py:904] (7/8) Epoch 1, batch 7850, loss[loss=0.3566, simple_loss=0.4102, pruned_loss=0.1515, over 16944.00 frames. ], tot_loss[loss=0.345, simple_loss=0.3958, pruned_loss=0.1471, over 3062855.06 frames. ], batch size: 116, lr: 3.66e-02, grad_scale: 8.0 2023-04-27 15:19:51,322 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7863.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:19:56,901 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7867.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:20:49,089 INFO [train.py:904] (7/8) Epoch 1, batch 7900, loss[loss=0.4108, simple_loss=0.4249, pruned_loss=0.1984, over 11362.00 frames. ], tot_loss[loss=0.3422, simple_loss=0.3938, pruned_loss=0.1453, over 3059196.19 frames. ], batch size: 246, lr: 3.66e-02, grad_scale: 8.0 2023-04-27 15:20:50,802 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-27 15:21:11,997 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.293e+02 4.952e+02 6.401e+02 7.784e+02 1.850e+03, threshold=1.280e+03, percent-clipped=3.0 2023-04-27 15:21:12,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6103, 3.4915, 3.6734, 3.5739, 3.7177, 4.0361, 4.0362, 3.6815], device='cuda:7'), covar=tensor([0.1609, 0.1664, 0.1088, 0.1914, 0.2268, 0.0971, 0.0782, 0.1813], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0209, 0.0180, 0.0189, 0.0224, 0.0184, 0.0163, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:21:39,339 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7931.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:22:08,702 INFO [train.py:904] (7/8) Epoch 1, batch 7950, loss[loss=0.3393, simple_loss=0.3854, pruned_loss=0.1466, over 16366.00 frames. ], tot_loss[loss=0.3419, simple_loss=0.3933, pruned_loss=0.1453, over 3062434.70 frames. ], batch size: 146, lr: 3.65e-02, grad_scale: 8.0 2023-04-27 15:22:54,878 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=7979.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:22:56,201 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7980.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:23:30,885 INFO [train.py:904] (7/8) Epoch 1, batch 8000, loss[loss=0.3455, simple_loss=0.4034, pruned_loss=0.1438, over 16701.00 frames. ], tot_loss[loss=0.3417, simple_loss=0.3931, pruned_loss=0.1452, over 3071949.92 frames. ], batch size: 89, lr: 3.64e-02, grad_scale: 8.0 2023-04-27 15:23:51,334 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.520e+02 5.080e+02 6.216e+02 7.580e+02 1.182e+03, threshold=1.243e+03, percent-clipped=0.0 2023-04-27 15:24:45,925 INFO [train.py:904] (7/8) Epoch 1, batch 8050, loss[loss=0.3408, simple_loss=0.3965, pruned_loss=0.1426, over 16231.00 frames. ], tot_loss[loss=0.3417, simple_loss=0.3934, pruned_loss=0.145, over 3073870.14 frames. ], batch size: 165, lr: 3.63e-02, grad_scale: 8.0 2023-04-27 15:25:04,704 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8063.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:25:08,294 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5928, 3.5690, 3.7092, 3.5293, 3.6795, 4.0477, 4.0483, 3.6581], device='cuda:7'), covar=tensor([0.1790, 0.1511, 0.1019, 0.1813, 0.2196, 0.1024, 0.0739, 0.1695], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0213, 0.0184, 0.0189, 0.0228, 0.0185, 0.0163, 0.0239], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:25:57,866 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8099.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:26:00,727 INFO [train.py:904] (7/8) Epoch 1, batch 8100, loss[loss=0.3924, simple_loss=0.4285, pruned_loss=0.1782, over 15553.00 frames. ], tot_loss[loss=0.3408, simple_loss=0.3929, pruned_loss=0.1444, over 3054644.74 frames. ], batch size: 190, lr: 3.62e-02, grad_scale: 4.0 2023-04-27 15:26:15,157 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8111.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:26:22,783 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.844e+02 5.411e+02 6.350e+02 7.571e+02 1.310e+03, threshold=1.270e+03, percent-clipped=1.0 2023-04-27 15:27:16,454 INFO [train.py:904] (7/8) Epoch 1, batch 8150, loss[loss=0.3085, simple_loss=0.3668, pruned_loss=0.1251, over 16686.00 frames. ], tot_loss[loss=0.3362, simple_loss=0.3889, pruned_loss=0.1418, over 3083731.42 frames. ], batch size: 76, lr: 3.62e-02, grad_scale: 4.0 2023-04-27 15:27:24,585 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7134, 3.5158, 2.6989, 4.2526, 3.0400, 4.1977, 3.3942, 2.8143], device='cuda:7'), covar=tensor([0.0279, 0.0271, 0.0320, 0.0234, 0.0970, 0.0147, 0.0429, 0.0949], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0115, 0.0095, 0.0131, 0.0200, 0.0114, 0.0136, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:27:28,063 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8158.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:28:33,144 INFO [train.py:904] (7/8) Epoch 1, batch 8200, loss[loss=0.3412, simple_loss=0.3938, pruned_loss=0.1443, over 16694.00 frames. ], tot_loss[loss=0.3331, simple_loss=0.3859, pruned_loss=0.1401, over 3091282.00 frames. ], batch size: 134, lr: 3.61e-02, grad_scale: 4.0 2023-04-27 15:28:38,818 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-27 15:28:56,623 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.505e+02 5.388e+02 6.507e+02 7.835e+02 3.023e+03, threshold=1.301e+03, percent-clipped=3.0 2023-04-27 15:29:17,153 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-27 15:29:50,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7590, 1.4461, 1.4605, 1.2152, 1.6617, 1.6001, 1.6845, 1.7935], device='cuda:7'), covar=tensor([0.0046, 0.0212, 0.0137, 0.0164, 0.0087, 0.0131, 0.0075, 0.0067], device='cuda:7'), in_proj_covar=tensor([0.0032, 0.0069, 0.0053, 0.0055, 0.0046, 0.0056, 0.0034, 0.0038], device='cuda:7'), out_proj_covar=tensor([4.3010e-05, 1.0225e-04, 7.6450e-05, 8.2346e-05, 6.9034e-05, 8.1303e-05, 5.5165e-05, 6.0426e-05], device='cuda:7') 2023-04-27 15:29:53,209 INFO [train.py:904] (7/8) Epoch 1, batch 8250, loss[loss=0.3145, simple_loss=0.3848, pruned_loss=0.122, over 16672.00 frames. ], tot_loss[loss=0.33, simple_loss=0.3849, pruned_loss=0.1375, over 3075902.46 frames. ], batch size: 134, lr: 3.60e-02, grad_scale: 4.0 2023-04-27 15:30:15,008 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-27 15:30:16,068 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8265.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:30:41,229 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8280.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:30:50,851 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4021, 4.2460, 3.9377, 3.6411, 4.1883, 2.3680, 3.8491, 4.2078], device='cuda:7'), covar=tensor([0.0104, 0.0064, 0.0089, 0.0340, 0.0066, 0.0983, 0.0092, 0.0103], device='cuda:7'), in_proj_covar=tensor([0.0048, 0.0038, 0.0056, 0.0076, 0.0041, 0.0086, 0.0055, 0.0053], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:31:14,616 INFO [train.py:904] (7/8) Epoch 1, batch 8300, loss[loss=0.2557, simple_loss=0.3474, pruned_loss=0.08201, over 16739.00 frames. ], tot_loss[loss=0.3216, simple_loss=0.3792, pruned_loss=0.132, over 3049837.08 frames. ], batch size: 89, lr: 3.59e-02, grad_scale: 4.0 2023-04-27 15:31:15,883 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4683, 1.4940, 1.5410, 1.3343, 2.1541, 2.1339, 2.2942, 2.2965], device='cuda:7'), covar=tensor([0.0030, 0.0390, 0.0171, 0.0234, 0.0086, 0.0121, 0.0070, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0031, 0.0070, 0.0054, 0.0057, 0.0046, 0.0056, 0.0033, 0.0038], device='cuda:7'), out_proj_covar=tensor([4.1570e-05, 1.0411e-04, 7.8492e-05, 8.4345e-05, 6.9781e-05, 8.0566e-05, 5.4366e-05, 6.1283e-05], device='cuda:7') 2023-04-27 15:31:40,150 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.807e+02 4.177e+02 5.020e+02 6.033e+02 1.438e+03, threshold=1.004e+03, percent-clipped=1.0 2023-04-27 15:31:56,904 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8326.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:31:59,952 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8328.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:32:05,129 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6219, 3.5290, 3.4380, 1.6581, 3.6546, 3.7573, 3.0884, 3.4244], device='cuda:7'), covar=tensor([0.0427, 0.0066, 0.0181, 0.1805, 0.0083, 0.0049, 0.0239, 0.0137], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0063, 0.0065, 0.0145, 0.0061, 0.0055, 0.0074, 0.0084], device='cuda:7'), out_proj_covar=tensor([1.4611e-04, 1.0998e-04, 1.1452e-04, 2.3186e-04, 1.0909e-04, 9.8767e-05, 1.3880e-04, 1.4027e-04], device='cuda:7') 2023-04-27 15:32:11,906 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2488, 3.1909, 3.2067, 3.4902, 3.3951, 3.3645, 3.3539, 3.4490], device='cuda:7'), covar=tensor([0.0347, 0.0360, 0.0798, 0.0336, 0.0467, 0.0593, 0.0451, 0.0306], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0162, 0.0246, 0.0174, 0.0148, 0.0154, 0.0137, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:32:36,049 INFO [train.py:904] (7/8) Epoch 1, batch 8350, loss[loss=0.2985, simple_loss=0.3728, pruned_loss=0.1121, over 16271.00 frames. ], tot_loss[loss=0.3149, simple_loss=0.3758, pruned_loss=0.127, over 3063507.49 frames. ], batch size: 165, lr: 3.58e-02, grad_scale: 4.0 2023-04-27 15:32:48,131 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.55 vs. limit=5.0 2023-04-27 15:33:07,482 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8646, 4.1289, 3.8252, 4.0027, 3.5969, 3.8195, 3.8341, 3.9556], device='cuda:7'), covar=tensor([0.0481, 0.0648, 0.0794, 0.0325, 0.0612, 0.0586, 0.0467, 0.0760], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0212, 0.0198, 0.0130, 0.0162, 0.0133, 0.0182, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:33:38,529 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9334, 4.1788, 3.9502, 4.0313, 3.5377, 3.8517, 3.8764, 4.0162], device='cuda:7'), covar=tensor([0.0454, 0.0758, 0.0760, 0.0323, 0.0657, 0.0576, 0.0536, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0208, 0.0193, 0.0128, 0.0158, 0.0132, 0.0179, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-27 15:33:41,814 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4400, 3.4522, 3.0320, 3.4088, 2.5822, 2.0399, 3.7870, 4.0087], device='cuda:7'), covar=tensor([0.1740, 0.0557, 0.0846, 0.0292, 0.1651, 0.1377, 0.0194, 0.0058], device='cuda:7'), in_proj_covar=tensor([0.0202, 0.0165, 0.0190, 0.0116, 0.0185, 0.0151, 0.0126, 0.0068], device='cuda:7'), out_proj_covar=tensor([2.4134e-04, 1.9629e-04, 2.0859e-04, 1.3724e-04, 2.2328e-04, 1.8171e-04, 1.5237e-04, 8.7425e-05], device='cuda:7') 2023-04-27 15:33:43,557 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-04-27 15:33:55,697 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8399.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:33:55,818 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1148, 4.1012, 1.6377, 4.1170, 2.4295, 4.0002, 1.8910, 2.9665], device='cuda:7'), covar=tensor([0.0036, 0.0077, 0.1897, 0.0041, 0.0899, 0.0153, 0.1520, 0.0609], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0068, 0.0149, 0.0064, 0.0123, 0.0073, 0.0149, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:33:57,935 INFO [train.py:904] (7/8) Epoch 1, batch 8400, loss[loss=0.2726, simple_loss=0.3469, pruned_loss=0.0991, over 16416.00 frames. ], tot_loss[loss=0.3069, simple_loss=0.3702, pruned_loss=0.1218, over 3064524.78 frames. ], batch size: 146, lr: 3.58e-02, grad_scale: 8.0 2023-04-27 15:34:21,546 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.723e+02 4.252e+02 5.243e+02 6.006e+02 1.213e+03, threshold=1.049e+03, percent-clipped=1.0 2023-04-27 15:35:12,951 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8447.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:35:18,235 INFO [train.py:904] (7/8) Epoch 1, batch 8450, loss[loss=0.2538, simple_loss=0.3407, pruned_loss=0.08346, over 16406.00 frames. ], tot_loss[loss=0.3022, simple_loss=0.367, pruned_loss=0.1188, over 3053204.81 frames. ], batch size: 146, lr: 3.57e-02, grad_scale: 8.0 2023-04-27 15:35:30,881 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8458.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:35:38,562 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4651, 4.2112, 4.2174, 4.6851, 4.7512, 4.4745, 4.8152, 4.6311], device='cuda:7'), covar=tensor([0.0447, 0.0414, 0.1123, 0.0436, 0.0387, 0.0392, 0.0287, 0.0364], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0161, 0.0243, 0.0174, 0.0146, 0.0152, 0.0138, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:35:52,890 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 15:36:16,378 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-27 15:36:39,482 INFO [train.py:904] (7/8) Epoch 1, batch 8500, loss[loss=0.2431, simple_loss=0.3211, pruned_loss=0.08256, over 16511.00 frames. ], tot_loss[loss=0.2947, simple_loss=0.3613, pruned_loss=0.114, over 3073004.46 frames. ], batch size: 68, lr: 3.56e-02, grad_scale: 8.0 2023-04-27 15:36:49,164 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8506.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:37:04,607 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.878e+02 4.840e+02 6.056e+02 7.383e+02 1.593e+03, threshold=1.211e+03, percent-clipped=7.0 2023-04-27 15:37:15,402 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4450, 3.2872, 3.1588, 3.0507, 3.2851, 2.2445, 3.1413, 3.1229], device='cuda:7'), covar=tensor([0.0073, 0.0065, 0.0095, 0.0227, 0.0071, 0.0841, 0.0097, 0.0091], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0038, 0.0055, 0.0070, 0.0041, 0.0087, 0.0054, 0.0051], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:37:53,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3761, 3.1113, 2.8159, 2.6926, 2.2186, 1.8964, 2.9523, 3.3126], device='cuda:7'), covar=tensor([0.1252, 0.0423, 0.0675, 0.0283, 0.1571, 0.1354, 0.0278, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0208, 0.0170, 0.0193, 0.0117, 0.0181, 0.0156, 0.0129, 0.0068], device='cuda:7'), out_proj_covar=tensor([2.4786e-04, 1.9985e-04, 2.1224e-04, 1.3818e-04, 2.1915e-04, 1.8866e-04, 1.5581e-04, 8.7689e-05], device='cuda:7') 2023-04-27 15:38:03,992 INFO [train.py:904] (7/8) Epoch 1, batch 8550, loss[loss=0.2724, simple_loss=0.3492, pruned_loss=0.09782, over 16867.00 frames. ], tot_loss[loss=0.2908, simple_loss=0.3575, pruned_loss=0.112, over 3041033.49 frames. ], batch size: 96, lr: 3.55e-02, grad_scale: 8.0 2023-04-27 15:39:44,574 INFO [train.py:904] (7/8) Epoch 1, batch 8600, loss[loss=0.3092, simple_loss=0.3768, pruned_loss=0.1208, over 15274.00 frames. ], tot_loss[loss=0.29, simple_loss=0.358, pruned_loss=0.111, over 3039674.53 frames. ], batch size: 191, lr: 3.54e-02, grad_scale: 8.0 2023-04-27 15:40:17,766 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-27 15:40:17,878 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.023e+02 4.189e+02 5.220e+02 6.302e+02 1.296e+03, threshold=1.044e+03, percent-clipped=1.0 2023-04-27 15:40:28,416 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8621.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:41:26,182 INFO [train.py:904] (7/8) Epoch 1, batch 8650, loss[loss=0.2671, simple_loss=0.3447, pruned_loss=0.09477, over 16213.00 frames. ], tot_loss[loss=0.2858, simple_loss=0.3552, pruned_loss=0.1082, over 3045690.29 frames. ], batch size: 165, lr: 3.54e-02, grad_scale: 8.0 2023-04-27 15:42:45,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1316, 1.5536, 1.9723, 2.1950, 2.3773, 2.3450, 1.4031, 2.4296], device='cuda:7'), covar=tensor([0.0068, 0.0561, 0.0230, 0.0157, 0.0082, 0.0170, 0.0368, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0094, 0.0073, 0.0053, 0.0042, 0.0045, 0.0069, 0.0041], device='cuda:7'), out_proj_covar=tensor([8.4127e-05, 1.7062e-04, 1.4111e-04, 9.9472e-05, 7.6867e-05, 8.4940e-05, 1.1865e-04, 7.6701e-05], device='cuda:7') 2023-04-27 15:43:13,577 INFO [train.py:904] (7/8) Epoch 1, batch 8700, loss[loss=0.2898, simple_loss=0.3457, pruned_loss=0.1169, over 12081.00 frames. ], tot_loss[loss=0.2807, simple_loss=0.3503, pruned_loss=0.1055, over 3040566.47 frames. ], batch size: 247, lr: 3.53e-02, grad_scale: 8.0 2023-04-27 15:43:20,686 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 15:43:32,092 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3125, 4.1216, 4.3605, 4.7571, 4.7149, 4.4189, 4.7704, 4.6092], device='cuda:7'), covar=tensor([0.0394, 0.0360, 0.0742, 0.0211, 0.0367, 0.0332, 0.0207, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0162, 0.0244, 0.0175, 0.0150, 0.0155, 0.0137, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:43:41,226 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.978e+02 4.653e+02 5.603e+02 6.924e+02 1.986e+03, threshold=1.121e+03, percent-clipped=4.0 2023-04-27 15:43:54,264 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4687, 3.5645, 3.0801, 3.4336, 2.6577, 2.0828, 3.6938, 4.0105], device='cuda:7'), covar=tensor([0.1682, 0.0519, 0.0893, 0.0275, 0.1325, 0.1362, 0.0229, 0.0056], device='cuda:7'), in_proj_covar=tensor([0.0209, 0.0174, 0.0197, 0.0119, 0.0172, 0.0157, 0.0131, 0.0070], device='cuda:7'), out_proj_covar=tensor([2.4823e-04, 2.0434e-04, 2.1670e-04, 1.3908e-04, 2.0885e-04, 1.8998e-04, 1.5929e-04, 9.0644e-05], device='cuda:7') 2023-04-27 15:44:42,147 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6812, 2.5722, 1.6379, 2.6770, 2.0002, 2.7080, 1.9162, 2.4129], device='cuda:7'), covar=tensor([0.0094, 0.0150, 0.1397, 0.0089, 0.0755, 0.0232, 0.1037, 0.0470], device='cuda:7'), in_proj_covar=tensor([0.0063, 0.0068, 0.0151, 0.0066, 0.0127, 0.0076, 0.0154, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-27 15:44:49,931 INFO [train.py:904] (7/8) Epoch 1, batch 8750, loss[loss=0.233, simple_loss=0.3143, pruned_loss=0.07592, over 12553.00 frames. ], tot_loss[loss=0.2771, simple_loss=0.3481, pruned_loss=0.1031, over 3040911.89 frames. ], batch size: 248, lr: 3.52e-02, grad_scale: 8.0 2023-04-27 15:45:14,505 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8760.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:46:42,415 INFO [train.py:904] (7/8) Epoch 1, batch 8800, loss[loss=0.3018, simple_loss=0.3555, pruned_loss=0.1241, over 12704.00 frames. ], tot_loss[loss=0.2739, simple_loss=0.3457, pruned_loss=0.1011, over 3056047.17 frames. ], batch size: 247, lr: 3.51e-02, grad_scale: 8.0 2023-04-27 15:47:13,493 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.267e+02 4.433e+02 5.797e+02 7.279e+02 2.542e+03, threshold=1.159e+03, percent-clipped=4.0 2023-04-27 15:47:25,134 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8821.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:48:03,500 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5071, 1.3636, 1.6513, 1.8568, 1.6136, 1.7089, 1.4338, 1.8731], device='cuda:7'), covar=tensor([0.0102, 0.0421, 0.0227, 0.0127, 0.0106, 0.0163, 0.0282, 0.0084], device='cuda:7'), in_proj_covar=tensor([0.0046, 0.0091, 0.0070, 0.0053, 0.0042, 0.0044, 0.0068, 0.0039], device='cuda:7'), out_proj_covar=tensor([8.4636e-05, 1.6613e-04, 1.3574e-04, 9.8507e-05, 7.5608e-05, 8.2070e-05, 1.1738e-04, 7.2536e-05], device='cuda:7') 2023-04-27 15:48:27,347 INFO [train.py:904] (7/8) Epoch 1, batch 8850, loss[loss=0.265, simple_loss=0.3474, pruned_loss=0.09131, over 16307.00 frames. ], tot_loss[loss=0.2744, simple_loss=0.3481, pruned_loss=0.1003, over 3048251.90 frames. ], batch size: 146, lr: 3.51e-02, grad_scale: 8.0 2023-04-27 15:50:13,549 INFO [train.py:904] (7/8) Epoch 1, batch 8900, loss[loss=0.3222, simple_loss=0.3874, pruned_loss=0.1285, over 15614.00 frames. ], tot_loss[loss=0.274, simple_loss=0.3485, pruned_loss=0.0997, over 3053777.75 frames. ], batch size: 190, lr: 3.50e-02, grad_scale: 8.0 2023-04-27 15:50:42,919 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.973e+02 4.392e+02 5.224e+02 6.190e+02 1.359e+03, threshold=1.045e+03, percent-clipped=1.0 2023-04-27 15:50:56,764 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8921.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:51:13,746 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8927.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:52:17,949 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2023-04-27 15:52:19,960 INFO [train.py:904] (7/8) Epoch 1, batch 8950, loss[loss=0.2572, simple_loss=0.3388, pruned_loss=0.08779, over 16167.00 frames. ], tot_loss[loss=0.2734, simple_loss=0.348, pruned_loss=0.0994, over 3077190.21 frames. ], batch size: 165, lr: 3.49e-02, grad_scale: 8.0 2023-04-27 15:53:00,568 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=8969.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:53:01,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.54 vs. limit=5.0 2023-04-27 15:53:15,930 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8608, 5.4774, 5.5647, 5.3579, 5.5511, 6.0005, 5.7897, 5.3338], device='cuda:7'), covar=tensor([0.0450, 0.1100, 0.0640, 0.1415, 0.1771, 0.0694, 0.0792, 0.1667], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0208, 0.0174, 0.0183, 0.0218, 0.0181, 0.0147, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 15:53:38,404 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8986.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:53:42,891 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8988.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:53:45,093 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.13 vs. limit=5.0 2023-04-27 15:54:08,291 INFO [train.py:904] (7/8) Epoch 1, batch 9000, loss[loss=0.2699, simple_loss=0.3285, pruned_loss=0.1057, over 12145.00 frames. ], tot_loss[loss=0.2701, simple_loss=0.3444, pruned_loss=0.09788, over 3062039.11 frames. ], batch size: 247, lr: 3.48e-02, grad_scale: 8.0 2023-04-27 15:54:08,291 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 15:54:17,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2725, 5.4751, 5.1277, 5.5340, 5.0389, 5.0539, 5.1893, 5.5111], device='cuda:7'), covar=tensor([0.0484, 0.0672, 0.0656, 0.0225, 0.0438, 0.0266, 0.0454, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0210, 0.0182, 0.0121, 0.0153, 0.0125, 0.0172, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-27 15:54:19,192 INFO [train.py:938] (7/8) Epoch 1, validation: loss=0.2299, simple_loss=0.3267, pruned_loss=0.06658, over 944034.00 frames. 2023-04-27 15:54:19,193 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 15:54:46,208 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9013.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:54:51,601 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.333e+02 3.931e+02 4.892e+02 5.908e+02 1.148e+03, threshold=9.783e+02, percent-clipped=2.0 2023-04-27 15:55:56,664 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9047.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:56:02,939 INFO [train.py:904] (7/8) Epoch 1, batch 9050, loss[loss=0.2372, simple_loss=0.3139, pruned_loss=0.08024, over 16863.00 frames. ], tot_loss[loss=0.2722, simple_loss=0.3461, pruned_loss=0.09916, over 3079300.22 frames. ], batch size: 102, lr: 3.48e-02, grad_scale: 8.0 2023-04-27 15:56:48,924 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9074.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:56:59,105 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8506, 3.7346, 3.3855, 3.2741, 3.7998, 1.8799, 3.4635, 3.4974], device='cuda:7'), covar=tensor([0.0080, 0.0064, 0.0110, 0.0238, 0.0050, 0.1187, 0.0086, 0.0122], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0039, 0.0054, 0.0066, 0.0039, 0.0090, 0.0053, 0.0051], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 15:57:45,081 INFO [train.py:904] (7/8) Epoch 1, batch 9100, loss[loss=0.2565, simple_loss=0.3412, pruned_loss=0.08587, over 16834.00 frames. ], tot_loss[loss=0.2715, simple_loss=0.3447, pruned_loss=0.09916, over 3067054.13 frames. ], batch size: 83, lr: 3.47e-02, grad_scale: 8.0 2023-04-27 15:58:14,298 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.937e+02 4.687e+02 5.823e+02 7.250e+02 1.575e+03, threshold=1.165e+03, percent-clipped=5.0 2023-04-27 15:58:14,908 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9116.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 15:59:13,716 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-04-27 15:59:21,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0461, 2.5021, 2.1807, 3.0658, 3.1969, 3.2171, 2.1632, 2.9877], device='cuda:7'), covar=tensor([0.1722, 0.0307, 0.1451, 0.0104, 0.0134, 0.0335, 0.0913, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0093, 0.0163, 0.0061, 0.0069, 0.0083, 0.0137, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 15:59:42,647 INFO [train.py:904] (7/8) Epoch 1, batch 9150, loss[loss=0.2937, simple_loss=0.352, pruned_loss=0.1177, over 12100.00 frames. ], tot_loss[loss=0.2702, simple_loss=0.3447, pruned_loss=0.09786, over 3065819.91 frames. ], batch size: 248, lr: 3.46e-02, grad_scale: 8.0 2023-04-27 16:01:27,202 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0889, 3.0901, 3.1924, 3.4300, 3.3520, 3.1644, 3.3163, 3.3518], device='cuda:7'), covar=tensor([0.0405, 0.0363, 0.0859, 0.0340, 0.0457, 0.0836, 0.0434, 0.0302], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0161, 0.0239, 0.0168, 0.0141, 0.0148, 0.0133, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:01:27,888 INFO [train.py:904] (7/8) Epoch 1, batch 9200, loss[loss=0.2496, simple_loss=0.3123, pruned_loss=0.09345, over 11739.00 frames. ], tot_loss[loss=0.2669, simple_loss=0.34, pruned_loss=0.09692, over 3048268.32 frames. ], batch size: 248, lr: 3.45e-02, grad_scale: 8.0 2023-04-27 16:01:54,930 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.438e+02 5.266e+02 6.436e+02 8.220e+02 2.129e+03, threshold=1.287e+03, percent-clipped=5.0 2023-04-27 16:02:01,588 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.42 vs. limit=5.0 2023-04-27 16:02:20,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8402, 3.8111, 3.8125, 2.9905, 3.7753, 3.6873, 3.8904, 2.0353], device='cuda:7'), covar=tensor([0.1228, 0.0066, 0.0059, 0.0330, 0.0052, 0.0084, 0.0048, 0.0766], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0046, 0.0051, 0.0087, 0.0048, 0.0050, 0.0050, 0.0095], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 16:03:04,223 INFO [train.py:904] (7/8) Epoch 1, batch 9250, loss[loss=0.2377, simple_loss=0.3065, pruned_loss=0.08449, over 12632.00 frames. ], tot_loss[loss=0.266, simple_loss=0.339, pruned_loss=0.09644, over 3038779.47 frames. ], batch size: 250, lr: 3.45e-02, grad_scale: 8.0 2023-04-27 16:04:17,164 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9283.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:04:58,124 INFO [train.py:904] (7/8) Epoch 1, batch 9300, loss[loss=0.2463, simple_loss=0.3224, pruned_loss=0.08513, over 16162.00 frames. ], tot_loss[loss=0.2626, simple_loss=0.3359, pruned_loss=0.09465, over 3020153.32 frames. ], batch size: 165, lr: 3.44e-02, grad_scale: 8.0 2023-04-27 16:05:33,419 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.238e+02 3.939e+02 4.601e+02 5.465e+02 1.094e+03, threshold=9.201e+02, percent-clipped=0.0 2023-04-27 16:05:42,561 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-27 16:06:27,895 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9342.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:06:44,339 INFO [train.py:904] (7/8) Epoch 1, batch 9350, loss[loss=0.2724, simple_loss=0.3432, pruned_loss=0.1008, over 15208.00 frames. ], tot_loss[loss=0.2629, simple_loss=0.3363, pruned_loss=0.09474, over 3049068.41 frames. ], batch size: 191, lr: 3.43e-02, grad_scale: 8.0 2023-04-27 16:07:24,449 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9369.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:08:26,600 INFO [train.py:904] (7/8) Epoch 1, batch 9400, loss[loss=0.2688, simple_loss=0.3519, pruned_loss=0.09286, over 15534.00 frames. ], tot_loss[loss=0.2621, simple_loss=0.3361, pruned_loss=0.09407, over 3050309.24 frames. ], batch size: 190, lr: 3.43e-02, grad_scale: 8.0 2023-04-27 16:08:54,872 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9414.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:08:56,916 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.254e+02 4.024e+02 4.732e+02 5.961e+02 1.474e+03, threshold=9.465e+02, percent-clipped=2.0 2023-04-27 16:08:58,060 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9416.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:09:46,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0755, 3.0827, 1.4964, 3.0916, 2.0077, 2.9854, 1.7815, 2.5051], device='cuda:7'), covar=tensor([0.0098, 0.0159, 0.1852, 0.0101, 0.1088, 0.0451, 0.1547, 0.0668], device='cuda:7'), in_proj_covar=tensor([0.0067, 0.0072, 0.0154, 0.0068, 0.0136, 0.0084, 0.0160, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-27 16:10:08,182 INFO [train.py:904] (7/8) Epoch 1, batch 9450, loss[loss=0.231, simple_loss=0.3171, pruned_loss=0.07243, over 16908.00 frames. ], tot_loss[loss=0.2638, simple_loss=0.3387, pruned_loss=0.09444, over 3066322.22 frames. ], batch size: 102, lr: 3.42e-02, grad_scale: 8.0 2023-04-27 16:10:24,739 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-27 16:10:33,963 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9464.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:10:57,230 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9238, 3.2027, 2.4223, 3.6946, 3.7230, 3.7742, 1.8923, 3.0894], device='cuda:7'), covar=tensor([0.2105, 0.0305, 0.1572, 0.0108, 0.0152, 0.0251, 0.1235, 0.0538], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0099, 0.0167, 0.0063, 0.0074, 0.0086, 0.0141, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:10:59,098 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9475.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:11:07,346 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2732, 1.6147, 1.8020, 1.5053, 1.9249, 1.9202, 2.1115, 2.2896], device='cuda:7'), covar=tensor([0.0036, 0.0298, 0.0169, 0.0269, 0.0113, 0.0231, 0.0068, 0.0089], device='cuda:7'), in_proj_covar=tensor([0.0033, 0.0075, 0.0060, 0.0067, 0.0053, 0.0064, 0.0035, 0.0043], device='cuda:7'), out_proj_covar=tensor([4.5317e-05, 1.1580e-04, 9.0255e-05, 1.0336e-04, 8.0304e-05, 9.6923e-05, 5.4334e-05, 6.9812e-05], device='cuda:7') 2023-04-27 16:11:35,924 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5276, 1.6797, 1.7237, 1.4569, 2.2721, 2.1898, 2.5627, 2.6528], device='cuda:7'), covar=tensor([0.0031, 0.0298, 0.0192, 0.0304, 0.0097, 0.0165, 0.0039, 0.0088], device='cuda:7'), in_proj_covar=tensor([0.0033, 0.0074, 0.0059, 0.0066, 0.0053, 0.0063, 0.0035, 0.0042], device='cuda:7'), out_proj_covar=tensor([4.4799e-05, 1.1387e-04, 8.9698e-05, 1.0275e-04, 8.0229e-05, 9.5332e-05, 5.3981e-05, 6.8648e-05], device='cuda:7') 2023-04-27 16:11:50,048 INFO [train.py:904] (7/8) Epoch 1, batch 9500, loss[loss=0.2589, simple_loss=0.3379, pruned_loss=0.08994, over 16651.00 frames. ], tot_loss[loss=0.2621, simple_loss=0.3377, pruned_loss=0.09324, over 3087892.55 frames. ], batch size: 134, lr: 3.41e-02, grad_scale: 8.0 2023-04-27 16:11:58,786 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7229, 3.6259, 4.1041, 4.1391, 4.2075, 3.8428, 3.9419, 3.9963], device='cuda:7'), covar=tensor([0.0270, 0.0356, 0.0427, 0.0426, 0.0443, 0.0257, 0.0555, 0.0292], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0111, 0.0125, 0.0127, 0.0137, 0.0113, 0.0159, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:12:09,022 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 16:12:21,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.089e+02 4.155e+02 5.314e+02 6.779e+02 1.064e+03, threshold=1.063e+03, percent-clipped=4.0 2023-04-27 16:13:37,019 INFO [train.py:904] (7/8) Epoch 1, batch 9550, loss[loss=0.2649, simple_loss=0.343, pruned_loss=0.09343, over 16484.00 frames. ], tot_loss[loss=0.2627, simple_loss=0.3379, pruned_loss=0.09373, over 3104086.21 frames. ], batch size: 68, lr: 3.41e-02, grad_scale: 8.0 2023-04-27 16:14:46,702 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9583.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:14:48,672 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9584.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:14:52,085 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7993, 3.2399, 3.1018, 1.2239, 3.1556, 3.2983, 3.1234, 2.7420], device='cuda:7'), covar=tensor([0.0940, 0.0141, 0.0219, 0.2255, 0.0150, 0.0116, 0.0207, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0071, 0.0071, 0.0149, 0.0063, 0.0066, 0.0077, 0.0092], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:15:18,627 INFO [train.py:904] (7/8) Epoch 1, batch 9600, loss[loss=0.2632, simple_loss=0.351, pruned_loss=0.0877, over 16203.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3407, pruned_loss=0.09576, over 3092897.11 frames. ], batch size: 165, lr: 3.40e-02, grad_scale: 8.0 2023-04-27 16:15:48,621 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 4.688e+02 5.707e+02 6.655e+02 1.542e+03, threshold=1.141e+03, percent-clipped=4.0 2023-04-27 16:15:50,361 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-04-27 16:16:20,303 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9631.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:16:20,394 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8566, 3.1338, 3.3430, 3.4330, 3.4712, 3.1032, 3.2272, 3.3504], device='cuda:7'), covar=tensor([0.0308, 0.0294, 0.0382, 0.0359, 0.0361, 0.0330, 0.0610, 0.0246], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0110, 0.0125, 0.0126, 0.0138, 0.0114, 0.0164, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:16:45,917 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9642.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:16:53,448 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9645.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:16:53,561 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9996, 2.5774, 2.2038, 3.1734, 2.3695, 2.9801, 2.5047, 2.0049], device='cuda:7'), covar=tensor([0.0262, 0.0293, 0.0266, 0.0247, 0.1006, 0.0193, 0.0476, 0.1144], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0129, 0.0104, 0.0145, 0.0202, 0.0123, 0.0147, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:17:08,563 INFO [train.py:904] (7/8) Epoch 1, batch 9650, loss[loss=0.2765, simple_loss=0.3489, pruned_loss=0.102, over 15308.00 frames. ], tot_loss[loss=0.2684, simple_loss=0.3428, pruned_loss=0.09701, over 3059178.10 frames. ], batch size: 191, lr: 3.39e-02, grad_scale: 8.0 2023-04-27 16:17:26,923 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9658.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:17:29,427 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-04-27 16:17:52,765 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9669.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:18:01,084 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7516, 4.6784, 4.1532, 4.1482, 4.6586, 2.5383, 4.1277, 4.3412], device='cuda:7'), covar=tensor([0.0067, 0.0043, 0.0083, 0.0161, 0.0037, 0.0941, 0.0072, 0.0101], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0040, 0.0057, 0.0066, 0.0041, 0.0094, 0.0053, 0.0052], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:18:01,558 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-04-27 16:18:35,171 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9690.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:18:56,675 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-27 16:18:57,053 INFO [train.py:904] (7/8) Epoch 1, batch 9700, loss[loss=0.2808, simple_loss=0.3485, pruned_loss=0.1065, over 16538.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3408, pruned_loss=0.09574, over 3062860.26 frames. ], batch size: 68, lr: 3.38e-02, grad_scale: 8.0 2023-04-27 16:19:19,940 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-27 16:19:24,742 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.013e+02 4.098e+02 5.059e+02 6.165e+02 1.399e+03, threshold=1.012e+03, percent-clipped=2.0 2023-04-27 16:19:28,779 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=9717.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:19:33,572 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9719.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:20:27,164 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9743.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:20:40,575 INFO [train.py:904] (7/8) Epoch 1, batch 9750, loss[loss=0.272, simple_loss=0.3489, pruned_loss=0.09752, over 16311.00 frames. ], tot_loss[loss=0.2654, simple_loss=0.3395, pruned_loss=0.09567, over 3066185.57 frames. ], batch size: 165, lr: 3.38e-02, grad_scale: 8.0 2023-04-27 16:21:17,333 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9770.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:22:00,695 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5320, 4.4001, 4.5930, 4.9787, 4.8889, 4.4839, 4.9229, 4.7729], device='cuda:7'), covar=tensor([0.0416, 0.0307, 0.0828, 0.0223, 0.0419, 0.0353, 0.0293, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0163, 0.0240, 0.0167, 0.0144, 0.0145, 0.0135, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:22:19,297 INFO [train.py:904] (7/8) Epoch 1, batch 9800, loss[loss=0.2429, simple_loss=0.3352, pruned_loss=0.07532, over 16694.00 frames. ], tot_loss[loss=0.2636, simple_loss=0.3389, pruned_loss=0.09414, over 3076892.27 frames. ], batch size: 89, lr: 3.37e-02, grad_scale: 8.0 2023-04-27 16:22:26,992 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9804.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:22:47,907 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.893e+02 4.491e+02 5.387e+02 6.683e+02 1.234e+03, threshold=1.077e+03, percent-clipped=5.0 2023-04-27 16:24:05,880 INFO [train.py:904] (7/8) Epoch 1, batch 9850, loss[loss=0.2813, simple_loss=0.3604, pruned_loss=0.1011, over 15375.00 frames. ], tot_loss[loss=0.2645, simple_loss=0.3405, pruned_loss=0.09422, over 3067418.80 frames. ], batch size: 191, lr: 3.36e-02, grad_scale: 8.0 2023-04-27 16:25:34,463 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-27 16:25:40,667 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9764, 3.8800, 3.6250, 3.3756, 3.8020, 1.6371, 3.4709, 3.7617], device='cuda:7'), covar=tensor([0.0072, 0.0065, 0.0085, 0.0216, 0.0060, 0.1453, 0.0088, 0.0106], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0040, 0.0056, 0.0065, 0.0040, 0.0096, 0.0053, 0.0052], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:25:58,734 INFO [train.py:904] (7/8) Epoch 1, batch 9900, loss[loss=0.2699, simple_loss=0.3522, pruned_loss=0.09378, over 15238.00 frames. ], tot_loss[loss=0.2639, simple_loss=0.3406, pruned_loss=0.0936, over 3074097.09 frames. ], batch size: 190, lr: 3.36e-02, grad_scale: 8.0 2023-04-27 16:26:31,898 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.967e+02 4.199e+02 5.185e+02 6.475e+02 1.120e+03, threshold=1.037e+03, percent-clipped=1.0 2023-04-27 16:27:32,792 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9940.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:27:57,279 INFO [train.py:904] (7/8) Epoch 1, batch 9950, loss[loss=0.2979, simple_loss=0.3594, pruned_loss=0.1182, over 12770.00 frames. ], tot_loss[loss=0.2656, simple_loss=0.3429, pruned_loss=0.09417, over 3082371.22 frames. ], batch size: 248, lr: 3.35e-02, grad_scale: 8.0 2023-04-27 16:28:13,922 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.59 vs. limit=5.0 2023-04-27 16:29:18,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3014, 3.3639, 2.6430, 2.8049, 2.4139, 2.0278, 3.2816, 3.7501], device='cuda:7'), covar=tensor([0.1759, 0.0546, 0.1064, 0.0495, 0.1176, 0.1369, 0.0345, 0.0079], device='cuda:7'), in_proj_covar=tensor([0.0213, 0.0189, 0.0209, 0.0131, 0.0169, 0.0162, 0.0148, 0.0079], device='cuda:7'), out_proj_covar=tensor([2.4841e-04, 2.1937e-04, 2.2780e-04, 1.5513e-04, 2.0119e-04, 1.9623e-04, 1.7507e-04, 9.7006e-05], device='cuda:7') 2023-04-27 16:30:02,401 INFO [train.py:904] (7/8) Epoch 1, batch 10000, loss[loss=0.2286, simple_loss=0.3212, pruned_loss=0.06794, over 16862.00 frames. ], tot_loss[loss=0.263, simple_loss=0.3403, pruned_loss=0.09281, over 3094754.88 frames. ], batch size: 90, lr: 3.34e-02, grad_scale: 8.0 2023-04-27 16:30:30,599 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10014.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:30:32,719 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.207e+02 4.345e+02 5.531e+02 6.703e+02 1.367e+03, threshold=1.106e+03, percent-clipped=3.0 2023-04-27 16:31:22,540 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10039.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:31:43,673 INFO [train.py:904] (7/8) Epoch 1, batch 10050, loss[loss=0.32, simple_loss=0.3879, pruned_loss=0.126, over 16220.00 frames. ], tot_loss[loss=0.2623, simple_loss=0.3403, pruned_loss=0.09213, over 3117333.77 frames. ], batch size: 165, lr: 3.34e-02, grad_scale: 8.0 2023-04-27 16:32:06,748 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10062.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:32:19,119 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5610, 4.8372, 4.8442, 4.7863, 4.7512, 5.2179, 5.0879, 4.6773], device='cuda:7'), covar=tensor([0.0597, 0.0728, 0.0644, 0.1198, 0.1719, 0.0635, 0.0649, 0.1599], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0208, 0.0177, 0.0177, 0.0220, 0.0180, 0.0154, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-27 16:32:23,113 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10070.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:32:27,878 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.84 vs. limit=5.0 2023-04-27 16:32:52,016 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5555, 4.3145, 4.2796, 4.4657, 4.0170, 4.2684, 4.3333, 4.0419], device='cuda:7'), covar=tensor([0.0145, 0.0093, 0.0121, 0.0083, 0.0392, 0.0141, 0.0200, 0.0164], device='cuda:7'), in_proj_covar=tensor([0.0088, 0.0065, 0.0113, 0.0088, 0.0136, 0.0091, 0.0082, 0.0099], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:33:16,695 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10099.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:33:18,946 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10100.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 16:33:19,591 INFO [train.py:904] (7/8) Epoch 1, batch 10100, loss[loss=0.2701, simple_loss=0.3364, pruned_loss=0.1019, over 12457.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.3415, pruned_loss=0.09359, over 3106666.66 frames. ], batch size: 246, lr: 3.33e-02, grad_scale: 16.0 2023-04-27 16:33:28,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0327, 4.1073, 2.1715, 4.1473, 2.4722, 3.9203, 2.3034, 3.0713], device='cuda:7'), covar=tensor([0.0036, 0.0072, 0.1374, 0.0031, 0.0876, 0.0218, 0.1190, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0066, 0.0077, 0.0159, 0.0069, 0.0136, 0.0090, 0.0162, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 16:33:49,436 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.152e+02 4.595e+02 5.786e+02 6.822e+02 2.551e+03, threshold=1.157e+03, percent-clipped=1.0 2023-04-27 16:33:53,662 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10118.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:34:06,074 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10123.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 16:35:05,150 INFO [train.py:904] (7/8) Epoch 2, batch 0, loss[loss=0.4446, simple_loss=0.4408, pruned_loss=0.2242, over 16405.00 frames. ], tot_loss[loss=0.4446, simple_loss=0.4408, pruned_loss=0.2242, over 16405.00 frames. ], batch size: 146, lr: 3.26e-02, grad_scale: 8.0 2023-04-27 16:35:05,150 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 16:35:12,743 INFO [train.py:938] (7/8) Epoch 2, validation: loss=0.2235, simple_loss=0.3221, pruned_loss=0.06242, over 944034.00 frames. 2023-04-27 16:35:12,743 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 16:35:37,572 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-27 16:35:47,182 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3913, 1.6339, 1.7498, 1.3408, 2.0495, 1.9942, 1.9489, 2.3180], device='cuda:7'), covar=tensor([0.0029, 0.0204, 0.0124, 0.0182, 0.0080, 0.0127, 0.0053, 0.0060], device='cuda:7'), in_proj_covar=tensor([0.0036, 0.0077, 0.0065, 0.0071, 0.0062, 0.0069, 0.0039, 0.0046], device='cuda:7'), out_proj_covar=tensor([5.0514e-05, 1.1865e-04, 9.9956e-05, 1.1038e-04, 9.5770e-05, 1.0476e-04, 5.9811e-05, 7.5628e-05], device='cuda:7') 2023-04-27 16:36:09,236 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-04-27 16:36:19,060 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-27 16:36:22,867 INFO [train.py:904] (7/8) Epoch 2, batch 50, loss[loss=0.3589, simple_loss=0.3836, pruned_loss=0.1671, over 16247.00 frames. ], tot_loss[loss=0.3194, simple_loss=0.3692, pruned_loss=0.1348, over 753677.65 frames. ], batch size: 164, lr: 3.25e-02, grad_scale: 4.0 2023-04-27 16:36:45,682 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.307e+02 5.068e+02 6.475e+02 7.712e+02 1.660e+03, threshold=1.295e+03, percent-clipped=4.0 2023-04-27 16:36:56,752 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1416, 3.5582, 2.7619, 4.5536, 2.6340, 4.1635, 3.0512, 2.4465], device='cuda:7'), covar=tensor([0.0220, 0.0253, 0.0283, 0.0182, 0.1223, 0.0150, 0.0465, 0.1299], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0132, 0.0106, 0.0153, 0.0210, 0.0126, 0.0148, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:37:13,526 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2023-04-27 16:37:16,896 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10240.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:37:26,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7371, 4.3768, 4.5805, 5.0520, 5.0749, 4.5718, 5.1601, 4.9345], device='cuda:7'), covar=tensor([0.0381, 0.0447, 0.1061, 0.0345, 0.0312, 0.0515, 0.0303, 0.0284], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0199, 0.0299, 0.0207, 0.0170, 0.0172, 0.0158, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:37:30,911 INFO [train.py:904] (7/8) Epoch 2, batch 100, loss[loss=0.2689, simple_loss=0.3453, pruned_loss=0.09621, over 17245.00 frames. ], tot_loss[loss=0.3051, simple_loss=0.3598, pruned_loss=0.1252, over 1326643.14 frames. ], batch size: 52, lr: 3.25e-02, grad_scale: 4.0 2023-04-27 16:37:42,399 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0041, 4.1778, 4.1480, 3.2681, 4.1311, 3.7958, 4.2826, 2.2738], device='cuda:7'), covar=tensor([0.0845, 0.0043, 0.0066, 0.0289, 0.0047, 0.0097, 0.0037, 0.0674], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0049, 0.0057, 0.0094, 0.0050, 0.0053, 0.0055, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 16:37:56,344 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2023-04-27 16:38:22,580 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10288.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:38:38,650 INFO [train.py:904] (7/8) Epoch 2, batch 150, loss[loss=0.2568, simple_loss=0.3216, pruned_loss=0.09605, over 16853.00 frames. ], tot_loss[loss=0.3052, simple_loss=0.3594, pruned_loss=0.1255, over 1761587.17 frames. ], batch size: 42, lr: 3.24e-02, grad_scale: 4.0 2023-04-27 16:38:56,674 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10314.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:39:01,495 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.010e+02 4.543e+02 5.695e+02 6.778e+02 1.541e+03, threshold=1.139e+03, percent-clipped=2.0 2023-04-27 16:39:47,449 INFO [train.py:904] (7/8) Epoch 2, batch 200, loss[loss=0.3175, simple_loss=0.3585, pruned_loss=0.1383, over 16488.00 frames. ], tot_loss[loss=0.3038, simple_loss=0.3586, pruned_loss=0.1245, over 2108608.32 frames. ], batch size: 146, lr: 3.23e-02, grad_scale: 4.0 2023-04-27 16:40:03,410 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10362.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:40:45,259 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0383, 4.0615, 1.7375, 4.2314, 2.4769, 4.1862, 1.8769, 2.9059], device='cuda:7'), covar=tensor([0.0050, 0.0215, 0.1617, 0.0039, 0.0913, 0.0202, 0.1573, 0.0650], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0083, 0.0162, 0.0073, 0.0140, 0.0098, 0.0168, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0003, 0.0001, 0.0003, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 16:40:50,474 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10395.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:40:55,720 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10399.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:40:58,056 INFO [train.py:904] (7/8) Epoch 2, batch 250, loss[loss=0.2838, simple_loss=0.341, pruned_loss=0.1133, over 16551.00 frames. ], tot_loss[loss=0.2978, simple_loss=0.3534, pruned_loss=0.121, over 2381015.28 frames. ], batch size: 68, lr: 3.23e-02, grad_scale: 4.0 2023-04-27 16:41:11,001 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10409.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:41:22,656 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.115e+02 4.151e+02 5.077e+02 6.310e+02 1.196e+03, threshold=1.015e+03, percent-clipped=1.0 2023-04-27 16:41:22,993 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10418.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 16:41:39,136 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3694, 3.5511, 2.6651, 2.8620, 2.4424, 1.9889, 3.5333, 3.9817], device='cuda:7'), covar=tensor([0.2017, 0.0586, 0.1160, 0.0487, 0.1947, 0.1639, 0.0353, 0.0210], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0202, 0.0221, 0.0138, 0.0202, 0.0167, 0.0157, 0.0090], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 16:42:03,851 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10447.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:42:09,621 INFO [train.py:904] (7/8) Epoch 2, batch 300, loss[loss=0.2301, simple_loss=0.3008, pruned_loss=0.0797, over 16762.00 frames. ], tot_loss[loss=0.2939, simple_loss=0.3503, pruned_loss=0.1187, over 2593556.25 frames. ], batch size: 39, lr: 3.22e-02, grad_scale: 4.0 2023-04-27 16:42:14,769 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10454.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:42:36,324 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10470.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:43:19,862 INFO [train.py:904] (7/8) Epoch 2, batch 350, loss[loss=0.2478, simple_loss=0.3138, pruned_loss=0.09088, over 17224.00 frames. ], tot_loss[loss=0.2885, simple_loss=0.3456, pruned_loss=0.1157, over 2759246.62 frames. ], batch size: 46, lr: 3.21e-02, grad_scale: 4.0 2023-04-27 16:43:39,913 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10515.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:43:42,894 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.580e+02 4.016e+02 4.975e+02 5.908e+02 1.216e+03, threshold=9.949e+02, percent-clipped=2.0 2023-04-27 16:44:27,071 INFO [train.py:904] (7/8) Epoch 2, batch 400, loss[loss=0.2813, simple_loss=0.3547, pruned_loss=0.1039, over 16664.00 frames. ], tot_loss[loss=0.2856, simple_loss=0.3436, pruned_loss=0.1138, over 2890771.00 frames. ], batch size: 57, lr: 3.21e-02, grad_scale: 8.0 2023-04-27 16:45:11,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7876, 2.8027, 2.8472, 2.3021, 2.9641, 2.8874, 3.1224, 1.8325], device='cuda:7'), covar=tensor([0.0930, 0.0119, 0.0115, 0.0427, 0.0079, 0.0098, 0.0057, 0.0689], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0050, 0.0056, 0.0093, 0.0051, 0.0053, 0.0057, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 16:45:16,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6145, 3.8305, 3.9352, 3.9972, 3.9896, 3.6148, 3.3752, 3.9783], device='cuda:7'), covar=tensor([0.0429, 0.0359, 0.0460, 0.0483, 0.0517, 0.0414, 0.0944, 0.0369], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0133, 0.0154, 0.0154, 0.0175, 0.0143, 0.0215, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-27 16:45:25,938 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1555, 4.8600, 4.7085, 4.3618, 4.9699, 2.1418, 4.6471, 5.0356], device='cuda:7'), covar=tensor([0.0071, 0.0063, 0.0077, 0.0271, 0.0052, 0.1271, 0.0083, 0.0091], device='cuda:7'), in_proj_covar=tensor([0.0055, 0.0047, 0.0068, 0.0089, 0.0050, 0.0100, 0.0064, 0.0066], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:45:36,495 INFO [train.py:904] (7/8) Epoch 2, batch 450, loss[loss=0.2546, simple_loss=0.3146, pruned_loss=0.09723, over 16678.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3406, pruned_loss=0.1112, over 2979965.33 frames. ], batch size: 89, lr: 3.20e-02, grad_scale: 8.0 2023-04-27 16:45:59,588 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.527e+02 4.134e+02 5.028e+02 5.858e+02 1.204e+03, threshold=1.006e+03, percent-clipped=2.0 2023-04-27 16:46:11,023 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-27 16:46:21,636 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.87 vs. limit=5.0 2023-04-27 16:46:44,109 INFO [train.py:904] (7/8) Epoch 2, batch 500, loss[loss=0.2603, simple_loss=0.3379, pruned_loss=0.09136, over 17119.00 frames. ], tot_loss[loss=0.2785, simple_loss=0.3374, pruned_loss=0.1098, over 3049073.32 frames. ], batch size: 48, lr: 3.20e-02, grad_scale: 8.0 2023-04-27 16:46:47,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8529, 4.1195, 2.8998, 4.9581, 5.0849, 4.6312, 2.7277, 3.9061], device='cuda:7'), covar=tensor([0.1977, 0.0329, 0.1354, 0.0093, 0.0087, 0.0287, 0.1029, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0105, 0.0164, 0.0062, 0.0079, 0.0092, 0.0141, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:47:45,062 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10695.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:47:52,721 INFO [train.py:904] (7/8) Epoch 2, batch 550, loss[loss=0.2829, simple_loss=0.3433, pruned_loss=0.1113, over 16740.00 frames. ], tot_loss[loss=0.2759, simple_loss=0.3362, pruned_loss=0.1078, over 3116475.22 frames. ], batch size: 62, lr: 3.19e-02, grad_scale: 8.0 2023-04-27 16:47:56,343 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0423, 5.3694, 5.0528, 5.2747, 4.6485, 4.7441, 4.8257, 5.4140], device='cuda:7'), covar=tensor([0.0461, 0.0662, 0.0868, 0.0324, 0.0549, 0.0386, 0.0486, 0.0540], device='cuda:7'), in_proj_covar=tensor([0.0202, 0.0280, 0.0249, 0.0165, 0.0196, 0.0159, 0.0224, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:48:17,046 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.652e+02 3.972e+02 4.860e+02 6.091e+02 1.021e+03, threshold=9.720e+02, percent-clipped=1.0 2023-04-27 16:48:17,386 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10718.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:48:51,499 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10743.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:48:53,583 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10744.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:49:02,286 INFO [train.py:904] (7/8) Epoch 2, batch 600, loss[loss=0.2726, simple_loss=0.3242, pruned_loss=0.1105, over 15613.00 frames. ], tot_loss[loss=0.2774, simple_loss=0.3361, pruned_loss=0.1094, over 3155765.94 frames. ], batch size: 190, lr: 3.18e-02, grad_scale: 8.0 2023-04-27 16:49:22,718 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10765.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:49:24,354 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=10766.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:49:33,837 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10774.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:50:11,102 INFO [train.py:904] (7/8) Epoch 2, batch 650, loss[loss=0.2557, simple_loss=0.3141, pruned_loss=0.09867, over 16724.00 frames. ], tot_loss[loss=0.276, simple_loss=0.3354, pruned_loss=0.1083, over 3202098.54 frames. ], batch size: 89, lr: 3.18e-02, grad_scale: 8.0 2023-04-27 16:50:17,154 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10805.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:50:24,082 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10810.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 16:50:33,027 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 4.152e+02 4.979e+02 6.118e+02 1.196e+03, threshold=9.959e+02, percent-clipped=5.0 2023-04-27 16:50:41,622 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-27 16:50:42,916 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2023-04-27 16:50:57,128 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10835.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:51:19,284 INFO [train.py:904] (7/8) Epoch 2, batch 700, loss[loss=0.2794, simple_loss=0.329, pruned_loss=0.1149, over 16884.00 frames. ], tot_loss[loss=0.2743, simple_loss=0.3346, pruned_loss=0.107, over 3225511.37 frames. ], batch size: 116, lr: 3.17e-02, grad_scale: 8.0 2023-04-27 16:52:06,339 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10885.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 16:52:26,671 INFO [train.py:904] (7/8) Epoch 2, batch 750, loss[loss=0.3163, simple_loss=0.3561, pruned_loss=0.1382, over 16779.00 frames. ], tot_loss[loss=0.2758, simple_loss=0.3355, pruned_loss=0.108, over 3251849.68 frames. ], batch size: 124, lr: 3.17e-02, grad_scale: 8.0 2023-04-27 16:52:50,210 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.205e+02 3.722e+02 4.836e+02 5.954e+02 8.013e+02, threshold=9.671e+02, percent-clipped=0.0 2023-04-27 16:53:28,023 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10946.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 16:53:33,387 INFO [train.py:904] (7/8) Epoch 2, batch 800, loss[loss=0.2676, simple_loss=0.3369, pruned_loss=0.09914, over 16656.00 frames. ], tot_loss[loss=0.2741, simple_loss=0.3341, pruned_loss=0.1071, over 3274381.94 frames. ], batch size: 62, lr: 3.16e-02, grad_scale: 8.0 2023-04-27 16:54:01,654 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.14 vs. limit=5.0 2023-04-27 16:54:43,066 INFO [train.py:904] (7/8) Epoch 2, batch 850, loss[loss=0.2129, simple_loss=0.2932, pruned_loss=0.06628, over 16849.00 frames. ], tot_loss[loss=0.2731, simple_loss=0.3333, pruned_loss=0.1065, over 3281423.74 frames. ], batch size: 42, lr: 3.15e-02, grad_scale: 8.0 2023-04-27 16:54:48,105 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0433, 3.8604, 3.9698, 4.4007, 4.3892, 4.1127, 4.2122, 4.2748], device='cuda:7'), covar=tensor([0.0451, 0.0477, 0.1218, 0.0349, 0.0386, 0.0556, 0.0497, 0.0361], device='cuda:7'), in_proj_covar=tensor([0.0210, 0.0235, 0.0355, 0.0249, 0.0203, 0.0203, 0.0184, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 16:55:06,012 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.379e+02 3.969e+02 4.868e+02 6.050e+02 1.097e+03, threshold=9.736e+02, percent-clipped=1.0 2023-04-27 16:55:30,537 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-27 16:55:49,296 INFO [train.py:904] (7/8) Epoch 2, batch 900, loss[loss=0.2386, simple_loss=0.3168, pruned_loss=0.08018, over 16869.00 frames. ], tot_loss[loss=0.2707, simple_loss=0.3316, pruned_loss=0.105, over 3284220.94 frames. ], batch size: 42, lr: 3.15e-02, grad_scale: 8.0 2023-04-27 16:56:10,048 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11065.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:56:14,903 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3722, 4.0230, 2.9736, 4.8086, 2.6955, 4.4078, 3.2538, 2.9484], device='cuda:7'), covar=tensor([0.0244, 0.0214, 0.0285, 0.0144, 0.1252, 0.0157, 0.0449, 0.1266], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0148, 0.0122, 0.0171, 0.0229, 0.0139, 0.0159, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 16:56:57,478 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11100.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:56:58,325 INFO [train.py:904] (7/8) Epoch 2, batch 950, loss[loss=0.2948, simple_loss=0.3359, pruned_loss=0.1269, over 16406.00 frames. ], tot_loss[loss=0.271, simple_loss=0.3321, pruned_loss=0.105, over 3296066.12 frames. ], batch size: 146, lr: 3.14e-02, grad_scale: 8.0 2023-04-27 16:57:10,451 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11110.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 16:57:14,733 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11113.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:57:21,475 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.464e+02 3.927e+02 4.944e+02 5.832e+02 1.325e+03, threshold=9.889e+02, percent-clipped=4.0 2023-04-27 16:57:38,034 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11130.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:58:06,962 INFO [train.py:904] (7/8) Epoch 2, batch 1000, loss[loss=0.2766, simple_loss=0.3534, pruned_loss=0.0999, over 17077.00 frames. ], tot_loss[loss=0.2695, simple_loss=0.3298, pruned_loss=0.1046, over 3304739.63 frames. ], batch size: 55, lr: 3.14e-02, grad_scale: 8.0 2023-04-27 16:58:08,687 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3692, 2.7767, 2.6921, 3.6772, 2.2421, 3.3960, 2.5561, 2.4529], device='cuda:7'), covar=tensor([0.0305, 0.0331, 0.0257, 0.0227, 0.1255, 0.0174, 0.0545, 0.1048], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0148, 0.0121, 0.0170, 0.0228, 0.0140, 0.0160, 0.0203], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 16:58:16,907 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11158.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:58:37,529 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11174.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 16:59:13,990 INFO [train.py:904] (7/8) Epoch 2, batch 1050, loss[loss=0.272, simple_loss=0.346, pruned_loss=0.09903, over 17068.00 frames. ], tot_loss[loss=0.2684, simple_loss=0.3289, pruned_loss=0.104, over 3299392.42 frames. ], batch size: 53, lr: 3.13e-02, grad_scale: 8.0 2023-04-27 16:59:22,118 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 16:59:28,450 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5040, 4.9217, 4.7349, 4.8390, 4.6943, 5.2928, 5.0348, 4.6969], device='cuda:7'), covar=tensor([0.0898, 0.0962, 0.0956, 0.1227, 0.2227, 0.0686, 0.0743, 0.1815], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0268, 0.0234, 0.0229, 0.0290, 0.0235, 0.0201, 0.0291], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 16:59:36,192 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.907e+02 3.907e+02 4.719e+02 5.702e+02 1.083e+03, threshold=9.439e+02, percent-clipped=2.0 2023-04-27 17:00:00,172 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11235.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:00:08,462 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11241.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 17:00:21,907 INFO [train.py:904] (7/8) Epoch 2, batch 1100, loss[loss=0.2752, simple_loss=0.3458, pruned_loss=0.1023, over 17101.00 frames. ], tot_loss[loss=0.2667, simple_loss=0.328, pruned_loss=0.1027, over 3307563.71 frames. ], batch size: 47, lr: 3.12e-02, grad_scale: 8.0 2023-04-27 17:00:28,767 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 17:01:28,367 INFO [train.py:904] (7/8) Epoch 2, batch 1150, loss[loss=0.2413, simple_loss=0.3215, pruned_loss=0.08055, over 17171.00 frames. ], tot_loss[loss=0.2651, simple_loss=0.3269, pruned_loss=0.1017, over 3304882.68 frames. ], batch size: 46, lr: 3.12e-02, grad_scale: 8.0 2023-04-27 17:01:52,667 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.647e+02 4.017e+02 4.974e+02 5.619e+02 1.017e+03, threshold=9.949e+02, percent-clipped=1.0 2023-04-27 17:02:39,337 INFO [train.py:904] (7/8) Epoch 2, batch 1200, loss[loss=0.2626, simple_loss=0.336, pruned_loss=0.09464, over 17034.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.3264, pruned_loss=0.1011, over 3310839.45 frames. ], batch size: 55, lr: 3.11e-02, grad_scale: 8.0 2023-04-27 17:02:54,695 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-27 17:03:00,966 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4968, 3.8218, 4.2921, 3.2844, 4.0629, 3.9542, 4.3659, 2.0450], device='cuda:7'), covar=tensor([0.0790, 0.0070, 0.0050, 0.0335, 0.0055, 0.0100, 0.0063, 0.0748], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0052, 0.0057, 0.0098, 0.0050, 0.0056, 0.0061, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 17:03:45,279 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5096, 5.1013, 5.2430, 5.3817, 4.6896, 5.2047, 5.2105, 4.9208], device='cuda:7'), covar=tensor([0.0191, 0.0155, 0.0149, 0.0098, 0.0873, 0.0146, 0.0120, 0.0224], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0088, 0.0162, 0.0129, 0.0197, 0.0125, 0.0112, 0.0139], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 17:03:46,871 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11400.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:03:47,490 INFO [train.py:904] (7/8) Epoch 2, batch 1250, loss[loss=0.3028, simple_loss=0.3503, pruned_loss=0.1276, over 16910.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3268, pruned_loss=0.1019, over 3309518.87 frames. ], batch size: 90, lr: 3.11e-02, grad_scale: 8.0 2023-04-27 17:04:10,335 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.776e+02 4.393e+02 5.372e+02 6.605e+02 1.355e+03, threshold=1.074e+03, percent-clipped=5.0 2023-04-27 17:04:17,560 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-27 17:04:26,861 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11430.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:04:47,903 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-27 17:04:49,933 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11448.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:04:52,455 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 17:04:53,736 INFO [train.py:904] (7/8) Epoch 2, batch 1300, loss[loss=0.2679, simple_loss=0.3398, pruned_loss=0.09804, over 17192.00 frames. ], tot_loss[loss=0.2642, simple_loss=0.3263, pruned_loss=0.101, over 3316960.60 frames. ], batch size: 46, lr: 3.10e-02, grad_scale: 8.0 2023-04-27 17:05:30,775 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11478.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:05:59,713 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1191, 4.2264, 3.6274, 3.4171, 3.0773, 2.3421, 4.5816, 5.2313], device='cuda:7'), covar=tensor([0.1569, 0.0516, 0.0845, 0.0483, 0.2022, 0.1230, 0.0257, 0.0054], device='cuda:7'), in_proj_covar=tensor([0.0234, 0.0215, 0.0224, 0.0150, 0.0232, 0.0171, 0.0169, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 17:06:01,559 INFO [train.py:904] (7/8) Epoch 2, batch 1350, loss[loss=0.2873, simple_loss=0.3379, pruned_loss=0.1183, over 16873.00 frames. ], tot_loss[loss=0.2641, simple_loss=0.3266, pruned_loss=0.1008, over 3315451.17 frames. ], batch size: 116, lr: 3.10e-02, grad_scale: 8.0 2023-04-27 17:06:24,758 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.664e+02 4.249e+02 5.319e+02 6.551e+02 1.548e+03, threshold=1.064e+03, percent-clipped=4.0 2023-04-27 17:06:41,822 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11530.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:06:45,104 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11532.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:06:57,573 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11541.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 17:07:09,714 INFO [train.py:904] (7/8) Epoch 2, batch 1400, loss[loss=0.2368, simple_loss=0.3163, pruned_loss=0.07866, over 17099.00 frames. ], tot_loss[loss=0.2644, simple_loss=0.3267, pruned_loss=0.101, over 3317462.69 frames. ], batch size: 47, lr: 3.09e-02, grad_scale: 8.0 2023-04-27 17:08:03,732 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11589.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 17:08:09,144 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11593.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:08:19,426 INFO [train.py:904] (7/8) Epoch 2, batch 1450, loss[loss=0.2578, simple_loss=0.3291, pruned_loss=0.09326, over 16581.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.326, pruned_loss=0.1013, over 3315088.02 frames. ], batch size: 62, lr: 3.08e-02, grad_scale: 8.0 2023-04-27 17:08:43,793 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.475e+02 3.985e+02 4.876e+02 5.970e+02 8.943e+02, threshold=9.752e+02, percent-clipped=0.0 2023-04-27 17:08:56,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4327, 4.8128, 4.4453, 4.6872, 4.1062, 4.4149, 4.3244, 4.8189], device='cuda:7'), covar=tensor([0.0495, 0.0607, 0.0749, 0.0361, 0.0624, 0.0474, 0.0514, 0.0564], device='cuda:7'), in_proj_covar=tensor([0.0210, 0.0292, 0.0254, 0.0173, 0.0199, 0.0166, 0.0232, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 17:09:06,492 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11635.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:09:23,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-04-27 17:09:26,119 INFO [train.py:904] (7/8) Epoch 2, batch 1500, loss[loss=0.2663, simple_loss=0.3368, pruned_loss=0.09794, over 16777.00 frames. ], tot_loss[loss=0.264, simple_loss=0.3253, pruned_loss=0.1014, over 3312543.15 frames. ], batch size: 57, lr: 3.08e-02, grad_scale: 8.0 2023-04-27 17:09:57,854 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9809, 3.1328, 3.5900, 2.8380, 3.6784, 3.6763, 3.8462, 2.0085], device='cuda:7'), covar=tensor([0.0826, 0.0173, 0.0098, 0.0282, 0.0058, 0.0078, 0.0047, 0.0670], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0050, 0.0056, 0.0097, 0.0050, 0.0054, 0.0061, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:7') 2023-04-27 17:10:19,989 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-27 17:10:30,136 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11696.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:10:33,681 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11699.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:10:36,165 INFO [train.py:904] (7/8) Epoch 2, batch 1550, loss[loss=0.3368, simple_loss=0.3773, pruned_loss=0.1481, over 16705.00 frames. ], tot_loss[loss=0.2663, simple_loss=0.327, pruned_loss=0.1028, over 3310554.76 frames. ], batch size: 134, lr: 3.07e-02, grad_scale: 8.0 2023-04-27 17:10:58,964 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.956e+02 4.303e+02 5.262e+02 6.472e+02 2.309e+03, threshold=1.052e+03, percent-clipped=9.0 2023-04-27 17:11:44,746 INFO [train.py:904] (7/8) Epoch 2, batch 1600, loss[loss=0.3853, simple_loss=0.4155, pruned_loss=0.1775, over 12195.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3292, pruned_loss=0.1042, over 3313349.34 frames. ], batch size: 246, lr: 3.07e-02, grad_scale: 8.0 2023-04-27 17:11:57,714 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11760.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:12:53,456 INFO [train.py:904] (7/8) Epoch 2, batch 1650, loss[loss=0.2899, simple_loss=0.3607, pruned_loss=0.1096, over 17065.00 frames. ], tot_loss[loss=0.2704, simple_loss=0.331, pruned_loss=0.1049, over 3316024.10 frames. ], batch size: 53, lr: 3.06e-02, grad_scale: 8.0 2023-04-27 17:13:15,802 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11816.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:13:17,809 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.782e+02 3.949e+02 4.848e+02 6.012e+02 1.253e+03, threshold=9.697e+02, percent-clipped=3.0 2023-04-27 17:13:35,505 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11830.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:13:36,884 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5621, 4.7646, 4.8324, 1.8537, 3.4235, 2.6905, 4.1093, 4.7759], device='cuda:7'), covar=tensor([0.0331, 0.0271, 0.0212, 0.1893, 0.0686, 0.1095, 0.0692, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0097, 0.0140, 0.0157, 0.0145, 0.0142, 0.0149, 0.0095], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 17:13:37,139 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2023-04-27 17:13:56,977 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11846.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:14:03,725 INFO [train.py:904] (7/8) Epoch 2, batch 1700, loss[loss=0.2143, simple_loss=0.2918, pruned_loss=0.06843, over 16787.00 frames. ], tot_loss[loss=0.2728, simple_loss=0.3338, pruned_loss=0.1059, over 3305514.23 frames. ], batch size: 39, lr: 3.06e-02, grad_scale: 8.0 2023-04-27 17:14:40,682 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11877.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:14:42,249 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=11878.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:14:55,512 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11888.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:15:11,417 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0629, 1.7367, 1.4457, 1.6439, 2.2285, 2.0743, 2.1787, 2.2477], device='cuda:7'), covar=tensor([0.0052, 0.0149, 0.0143, 0.0172, 0.0066, 0.0125, 0.0057, 0.0065], device='cuda:7'), in_proj_covar=tensor([0.0042, 0.0084, 0.0077, 0.0086, 0.0074, 0.0083, 0.0048, 0.0057], device='cuda:7'), out_proj_covar=tensor([6.7245e-05, 1.2949e-04, 1.1972e-04, 1.3557e-04, 1.1764e-04, 1.3400e-04, 7.5675e-05, 9.8075e-05], device='cuda:7') 2023-04-27 17:15:13,200 INFO [train.py:904] (7/8) Epoch 2, batch 1750, loss[loss=0.3602, simple_loss=0.4025, pruned_loss=0.1589, over 11780.00 frames. ], tot_loss[loss=0.2732, simple_loss=0.3347, pruned_loss=0.1058, over 3297840.19 frames. ], batch size: 246, lr: 3.05e-02, grad_scale: 8.0 2023-04-27 17:15:21,983 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11907.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:15:36,932 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.686e+02 4.328e+02 5.055e+02 5.950e+02 1.641e+03, threshold=1.011e+03, percent-clipped=5.0 2023-04-27 17:15:40,273 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11920.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:16:17,843 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11948.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:16:21,488 INFO [train.py:904] (7/8) Epoch 2, batch 1800, loss[loss=0.2575, simple_loss=0.335, pruned_loss=0.09002, over 17098.00 frames. ], tot_loss[loss=0.2742, simple_loss=0.3366, pruned_loss=0.1059, over 3304328.34 frames. ], batch size: 49, lr: 3.05e-02, grad_scale: 8.0 2023-04-27 17:17:03,321 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11981.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:06,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3979, 4.4389, 4.3490, 1.6824, 3.1216, 2.3488, 3.6678, 4.3353], device='cuda:7'), covar=tensor([0.0299, 0.0351, 0.0227, 0.1750, 0.0722, 0.0998, 0.0716, 0.0445], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0097, 0.0135, 0.0153, 0.0141, 0.0137, 0.0144, 0.0094], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 17:17:07,219 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11984.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:16,286 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11991.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:33,765 INFO [train.py:904] (7/8) Epoch 2, batch 1850, loss[loss=0.2717, simple_loss=0.3555, pruned_loss=0.09395, over 16738.00 frames. ], tot_loss[loss=0.2758, simple_loss=0.3385, pruned_loss=0.1065, over 3303048.63 frames. ], batch size: 57, lr: 3.04e-02, grad_scale: 8.0 2023-04-27 17:17:45,627 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12009.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:17:57,890 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.566e+02 4.018e+02 4.936e+02 6.102e+02 1.131e+03, threshold=9.871e+02, percent-clipped=1.0 2023-04-27 17:18:35,052 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12045.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:18:43,206 INFO [train.py:904] (7/8) Epoch 2, batch 1900, loss[loss=0.2649, simple_loss=0.3391, pruned_loss=0.09536, over 17038.00 frames. ], tot_loss[loss=0.2735, simple_loss=0.3367, pruned_loss=0.1051, over 3307613.46 frames. ], batch size: 50, lr: 3.04e-02, grad_scale: 8.0 2023-04-27 17:18:49,566 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12055.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:19:17,259 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12076.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:19:34,505 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-04-27 17:19:51,459 INFO [train.py:904] (7/8) Epoch 2, batch 1950, loss[loss=0.2576, simple_loss=0.3343, pruned_loss=0.09041, over 17137.00 frames. ], tot_loss[loss=0.2724, simple_loss=0.3363, pruned_loss=0.1042, over 3308592.04 frames. ], batch size: 48, lr: 3.03e-02, grad_scale: 8.0 2023-04-27 17:20:14,622 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.438e+02 4.259e+02 5.225e+02 6.323e+02 1.109e+03, threshold=1.045e+03, percent-clipped=2.0 2023-04-27 17:20:40,752 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12137.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:20:59,931 INFO [train.py:904] (7/8) Epoch 2, batch 2000, loss[loss=0.2987, simple_loss=0.3478, pruned_loss=0.1248, over 16796.00 frames. ], tot_loss[loss=0.2716, simple_loss=0.3349, pruned_loss=0.1041, over 3315536.84 frames. ], batch size: 90, lr: 3.02e-02, grad_scale: 8.0 2023-04-27 17:21:05,212 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2372, 3.6088, 2.9071, 4.6095, 2.7244, 4.7090, 3.1474, 2.6450], device='cuda:7'), covar=tensor([0.0221, 0.0281, 0.0283, 0.0163, 0.1165, 0.0114, 0.0511, 0.1352], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0156, 0.0130, 0.0184, 0.0235, 0.0147, 0.0166, 0.0216], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 17:21:28,830 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12172.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:21:41,447 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4852, 3.3905, 2.5119, 2.7907, 2.5019, 1.9513, 3.3247, 3.7659], device='cuda:7'), covar=tensor([0.1693, 0.0592, 0.1159, 0.0510, 0.1745, 0.1480, 0.0394, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0240, 0.0217, 0.0224, 0.0154, 0.0243, 0.0173, 0.0169, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-27 17:21:51,853 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12188.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:21:53,633 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-27 17:22:09,012 INFO [train.py:904] (7/8) Epoch 2, batch 2050, loss[loss=0.3044, simple_loss=0.3516, pruned_loss=0.1286, over 16355.00 frames. ], tot_loss[loss=0.2712, simple_loss=0.3343, pruned_loss=0.1041, over 3319874.28 frames. ], batch size: 146, lr: 3.02e-02, grad_scale: 16.0 2023-04-27 17:22:11,155 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12202.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:22:32,915 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.569e+02 4.228e+02 5.159e+02 6.169e+02 1.128e+03, threshold=1.032e+03, percent-clipped=1.0 2023-04-27 17:22:59,035 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12236.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:23:20,407 INFO [train.py:904] (7/8) Epoch 2, batch 2100, loss[loss=0.2335, simple_loss=0.3149, pruned_loss=0.07607, over 17027.00 frames. ], tot_loss[loss=0.2725, simple_loss=0.3354, pruned_loss=0.1048, over 3311606.89 frames. ], batch size: 50, lr: 3.01e-02, grad_scale: 16.0 2023-04-27 17:23:45,066 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9722, 3.7808, 3.8641, 4.2838, 4.2338, 3.9766, 4.1222, 4.1609], device='cuda:7'), covar=tensor([0.0369, 0.0382, 0.0954, 0.0292, 0.0356, 0.0589, 0.0397, 0.0307], device='cuda:7'), in_proj_covar=tensor([0.0227, 0.0254, 0.0377, 0.0269, 0.0217, 0.0209, 0.0193, 0.0216], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 17:23:54,374 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12276.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:24:15,564 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12291.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:24:28,734 INFO [train.py:904] (7/8) Epoch 2, batch 2150, loss[loss=0.3255, simple_loss=0.3742, pruned_loss=0.1385, over 16508.00 frames. ], tot_loss[loss=0.2734, simple_loss=0.3365, pruned_loss=0.1051, over 3321441.85 frames. ], batch size: 146, lr: 3.01e-02, grad_scale: 16.0 2023-04-27 17:24:33,555 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12304.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:24:51,838 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.440e+02 4.527e+02 5.287e+02 6.148e+02 1.100e+03, threshold=1.057e+03, percent-clipped=4.0 2023-04-27 17:25:16,318 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8994, 5.5004, 5.4507, 5.3169, 5.4070, 5.9194, 5.7138, 5.4277], device='cuda:7'), covar=tensor([0.0543, 0.0974, 0.0920, 0.1362, 0.2012, 0.0655, 0.0679, 0.1532], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0273, 0.0248, 0.0234, 0.0304, 0.0247, 0.0211, 0.0314], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 17:25:22,226 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12339.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:25:23,463 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12340.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:25:31,945 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3301, 3.7449, 2.9699, 4.7281, 2.7740, 4.6668, 3.1852, 2.7305], device='cuda:7'), covar=tensor([0.0256, 0.0302, 0.0300, 0.0181, 0.1320, 0.0133, 0.0529, 0.1385], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0154, 0.0127, 0.0177, 0.0232, 0.0144, 0.0160, 0.0211], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 17:25:38,428 INFO [train.py:904] (7/8) Epoch 2, batch 2200, loss[loss=0.3493, simple_loss=0.3733, pruned_loss=0.1626, over 16902.00 frames. ], tot_loss[loss=0.2752, simple_loss=0.3384, pruned_loss=0.106, over 3327907.83 frames. ], batch size: 109, lr: 3.00e-02, grad_scale: 16.0 2023-04-27 17:25:44,397 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12355.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:25:56,684 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7689, 4.1981, 2.4065, 5.0674, 5.0673, 4.4827, 2.7433, 3.3658], device='cuda:7'), covar=tensor([0.1933, 0.0313, 0.1514, 0.0086, 0.0142, 0.0348, 0.0994, 0.0619], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0109, 0.0164, 0.0068, 0.0096, 0.0101, 0.0147, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-27 17:26:12,220 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.30 vs. limit=2.0 2023-04-27 17:26:26,032 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.19 vs. limit=5.0 2023-04-27 17:26:48,789 INFO [train.py:904] (7/8) Epoch 2, batch 2250, loss[loss=0.2679, simple_loss=0.3232, pruned_loss=0.1063, over 16536.00 frames. ], tot_loss[loss=0.2763, simple_loss=0.3393, pruned_loss=0.1066, over 3314258.41 frames. ], batch size: 146, lr: 3.00e-02, grad_scale: 16.0 2023-04-27 17:26:52,141 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12403.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:27:12,084 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.514e+02 4.351e+02 5.087e+02 6.619e+02 9.905e+02, threshold=1.017e+03, percent-clipped=0.0 2023-04-27 17:27:32,334 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12432.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:27:51,444 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12446.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:27:57,828 INFO [train.py:904] (7/8) Epoch 2, batch 2300, loss[loss=0.2779, simple_loss=0.3287, pruned_loss=0.1136, over 16813.00 frames. ], tot_loss[loss=0.2751, simple_loss=0.3387, pruned_loss=0.1058, over 3317122.78 frames. ], batch size: 124, lr: 2.99e-02, grad_scale: 8.0 2023-04-27 17:28:01,645 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-27 17:28:12,016 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0028, 1.9233, 2.4247, 2.7975, 3.1304, 3.0042, 1.8972, 3.0945], device='cuda:7'), covar=tensor([0.0048, 0.0287, 0.0147, 0.0112, 0.0028, 0.0074, 0.0205, 0.0045], device='cuda:7'), in_proj_covar=tensor([0.0069, 0.0107, 0.0090, 0.0078, 0.0053, 0.0054, 0.0088, 0.0054], device='cuda:7'), out_proj_covar=tensor([1.2605e-04, 1.9312e-04, 1.7248e-04, 1.4965e-04, 9.6205e-05, 1.0301e-04, 1.5362e-04, 1.0152e-04], device='cuda:7') 2023-04-27 17:28:27,185 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12472.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:06,510 INFO [train.py:904] (7/8) Epoch 2, batch 2350, loss[loss=0.318, simple_loss=0.36, pruned_loss=0.138, over 16865.00 frames. ], tot_loss[loss=0.2763, simple_loss=0.3395, pruned_loss=0.1066, over 3320841.85 frames. ], batch size: 90, lr: 2.99e-02, grad_scale: 8.0 2023-04-27 17:29:07,919 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12502.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:15,031 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12507.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:29:31,566 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.397e+02 4.650e+02 5.724e+02 7.343e+02 1.924e+03, threshold=1.145e+03, percent-clipped=10.0 2023-04-27 17:29:33,940 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12520.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:30:14,756 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12550.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:30:15,612 INFO [train.py:904] (7/8) Epoch 2, batch 2400, loss[loss=0.277, simple_loss=0.3304, pruned_loss=0.1118, over 16833.00 frames. ], tot_loss[loss=0.278, simple_loss=0.3404, pruned_loss=0.1078, over 3310800.08 frames. ], batch size: 96, lr: 2.98e-02, grad_scale: 8.0 2023-04-27 17:30:52,395 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12576.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:31:26,492 INFO [train.py:904] (7/8) Epoch 2, batch 2450, loss[loss=0.2652, simple_loss=0.3444, pruned_loss=0.09304, over 17139.00 frames. ], tot_loss[loss=0.2782, simple_loss=0.3413, pruned_loss=0.1076, over 3315112.32 frames. ], batch size: 49, lr: 2.98e-02, grad_scale: 8.0 2023-04-27 17:31:31,793 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12604.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:31:49,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9605, 4.2678, 3.5428, 3.7576, 3.2110, 2.4628, 4.6415, 5.1735], device='cuda:7'), covar=tensor([0.1534, 0.0482, 0.0805, 0.0368, 0.1703, 0.1087, 0.0174, 0.0078], device='cuda:7'), in_proj_covar=tensor([0.0245, 0.0226, 0.0229, 0.0160, 0.0256, 0.0177, 0.0174, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 17:31:51,023 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.702e+02 4.143e+02 4.943e+02 6.147e+02 1.117e+03, threshold=9.887e+02, percent-clipped=0.0 2023-04-27 17:31:59,247 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12624.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:32:20,253 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12640.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:32:35,555 INFO [train.py:904] (7/8) Epoch 2, batch 2500, loss[loss=0.2391, simple_loss=0.3219, pruned_loss=0.07819, over 17127.00 frames. ], tot_loss[loss=0.2773, simple_loss=0.3408, pruned_loss=0.1069, over 3318511.07 frames. ], batch size: 49, lr: 2.97e-02, grad_scale: 8.0 2023-04-27 17:32:37,514 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12652.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:33:27,072 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12688.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:33:43,146 INFO [train.py:904] (7/8) Epoch 2, batch 2550, loss[loss=0.3725, simple_loss=0.4067, pruned_loss=0.1691, over 11821.00 frames. ], tot_loss[loss=0.2773, simple_loss=0.3408, pruned_loss=0.1069, over 3318777.20 frames. ], batch size: 246, lr: 2.97e-02, grad_scale: 8.0 2023-04-27 17:33:54,915 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-04-27 17:34:08,156 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.860e+02 4.439e+02 5.440e+02 6.842e+02 1.130e+03, threshold=1.088e+03, percent-clipped=2.0 2023-04-27 17:34:27,597 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12732.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:34:53,771 INFO [train.py:904] (7/8) Epoch 2, batch 2600, loss[loss=0.2768, simple_loss=0.3523, pruned_loss=0.1007, over 17249.00 frames. ], tot_loss[loss=0.2772, simple_loss=0.3408, pruned_loss=0.1069, over 3309546.60 frames. ], batch size: 52, lr: 2.96e-02, grad_scale: 8.0 2023-04-27 17:35:08,293 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-04-27 17:35:08,917 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12763.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:35:33,524 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=12780.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:36:01,530 INFO [train.py:904] (7/8) Epoch 2, batch 2650, loss[loss=0.253, simple_loss=0.3366, pruned_loss=0.08472, over 17221.00 frames. ], tot_loss[loss=0.2765, simple_loss=0.341, pruned_loss=0.106, over 3305141.41 frames. ], batch size: 52, lr: 2.96e-02, grad_scale: 8.0 2023-04-27 17:36:03,347 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12802.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:36:27,048 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.569e+02 4.221e+02 4.762e+02 5.639e+02 1.002e+03, threshold=9.524e+02, percent-clipped=0.0 2023-04-27 17:36:33,938 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12824.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 17:37:09,361 INFO [train.py:904] (7/8) Epoch 2, batch 2700, loss[loss=0.2674, simple_loss=0.3271, pruned_loss=0.1039, over 16852.00 frames. ], tot_loss[loss=0.2749, simple_loss=0.3402, pruned_loss=0.1048, over 3307495.81 frames. ], batch size: 96, lr: 2.95e-02, grad_scale: 8.0 2023-04-27 17:37:21,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7346, 1.6195, 2.1312, 2.4423, 2.7739, 2.5375, 1.6914, 2.5619], device='cuda:7'), covar=tensor([0.0068, 0.0312, 0.0183, 0.0157, 0.0035, 0.0114, 0.0222, 0.0063], device='cuda:7'), in_proj_covar=tensor([0.0072, 0.0109, 0.0092, 0.0082, 0.0055, 0.0055, 0.0090, 0.0055], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-27 17:37:29,357 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6954, 4.6881, 2.1021, 4.5839, 2.6606, 4.4631, 2.3831, 3.1215], device='cuda:7'), covar=tensor([0.0036, 0.0086, 0.1376, 0.0030, 0.0824, 0.0199, 0.1284, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0084, 0.0102, 0.0166, 0.0075, 0.0152, 0.0125, 0.0170, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 17:38:19,515 INFO [train.py:904] (7/8) Epoch 2, batch 2750, loss[loss=0.286, simple_loss=0.3417, pruned_loss=0.1151, over 16880.00 frames. ], tot_loss[loss=0.2738, simple_loss=0.34, pruned_loss=0.1037, over 3312634.28 frames. ], batch size: 109, lr: 2.95e-02, grad_scale: 8.0 2023-04-27 17:38:41,653 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.401e+02 4.125e+02 4.810e+02 5.955e+02 1.093e+03, threshold=9.620e+02, percent-clipped=2.0 2023-04-27 17:39:26,265 INFO [train.py:904] (7/8) Epoch 2, batch 2800, loss[loss=0.2523, simple_loss=0.3319, pruned_loss=0.08637, over 17086.00 frames. ], tot_loss[loss=0.2739, simple_loss=0.3398, pruned_loss=0.104, over 3314732.43 frames. ], batch size: 47, lr: 2.94e-02, grad_scale: 8.0 2023-04-27 17:40:33,621 INFO [train.py:904] (7/8) Epoch 2, batch 2850, loss[loss=0.4073, simple_loss=0.4267, pruned_loss=0.1939, over 12203.00 frames. ], tot_loss[loss=0.2729, simple_loss=0.3392, pruned_loss=0.1033, over 3312135.22 frames. ], batch size: 247, lr: 2.94e-02, grad_scale: 8.0 2023-04-27 17:40:39,106 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0554, 3.9854, 3.3114, 3.6779, 3.0351, 2.1495, 4.4015, 4.8913], device='cuda:7'), covar=tensor([0.1404, 0.0530, 0.0794, 0.0389, 0.1647, 0.1303, 0.0200, 0.0101], device='cuda:7'), in_proj_covar=tensor([0.0237, 0.0223, 0.0225, 0.0157, 0.0247, 0.0175, 0.0176, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 17:40:57,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.754e+02 4.789e+02 5.762e+02 6.825e+02 1.913e+03, threshold=1.152e+03, percent-clipped=8.0 2023-04-27 17:41:41,141 INFO [train.py:904] (7/8) Epoch 2, batch 2900, loss[loss=0.3068, simple_loss=0.3752, pruned_loss=0.1192, over 16743.00 frames. ], tot_loss[loss=0.2726, simple_loss=0.3378, pruned_loss=0.1037, over 3315005.41 frames. ], batch size: 62, lr: 2.93e-02, grad_scale: 8.0 2023-04-27 17:41:49,104 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.62 vs. limit=5.0 2023-04-27 17:41:51,111 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5879, 4.7046, 4.5716, 4.7035, 4.5478, 5.1823, 5.0162, 4.6726], device='cuda:7'), covar=tensor([0.0856, 0.1182, 0.1045, 0.1462, 0.2378, 0.0751, 0.0870, 0.1903], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0267, 0.0238, 0.0232, 0.0301, 0.0242, 0.0210, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 17:42:49,017 INFO [train.py:904] (7/8) Epoch 2, batch 2950, loss[loss=0.2886, simple_loss=0.3415, pruned_loss=0.1179, over 16709.00 frames. ], tot_loss[loss=0.2743, simple_loss=0.3374, pruned_loss=0.1056, over 3306074.17 frames. ], batch size: 89, lr: 2.93e-02, grad_scale: 8.0 2023-04-27 17:42:50,483 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13102.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:43:13,805 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.070e+02 4.352e+02 4.979e+02 5.894e+02 1.092e+03, threshold=9.958e+02, percent-clipped=0.0 2023-04-27 17:43:14,168 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=13119.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 17:43:53,710 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=13150.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:43:54,571 INFO [train.py:904] (7/8) Epoch 2, batch 3000, loss[loss=0.2453, simple_loss=0.3218, pruned_loss=0.08436, over 17043.00 frames. ], tot_loss[loss=0.2756, simple_loss=0.3381, pruned_loss=0.1065, over 3309009.84 frames. ], batch size: 50, lr: 2.92e-02, grad_scale: 8.0 2023-04-27 17:43:54,571 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 17:44:03,915 INFO [train.py:938] (7/8) Epoch 2, validation: loss=0.1858, simple_loss=0.2917, pruned_loss=0.04, over 944034.00 frames. 2023-04-27 17:44:03,916 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 17:44:53,947 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6660, 4.6441, 4.7239, 1.9610, 3.6078, 2.4948, 4.1294, 4.6976], device='cuda:7'), covar=tensor([0.0283, 0.0408, 0.0261, 0.1903, 0.0664, 0.1164, 0.0722, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0110, 0.0146, 0.0156, 0.0147, 0.0140, 0.0148, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 17:45:09,428 INFO [train.py:904] (7/8) Epoch 2, batch 3050, loss[loss=0.2544, simple_loss=0.3154, pruned_loss=0.09669, over 16852.00 frames. ], tot_loss[loss=0.2739, simple_loss=0.3369, pruned_loss=0.1055, over 3311068.48 frames. ], batch size: 83, lr: 2.92e-02, grad_scale: 8.0 2023-04-27 17:45:33,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.792e+02 4.585e+02 5.657e+02 6.871e+02 1.163e+03, threshold=1.131e+03, percent-clipped=2.0 2023-04-27 17:45:42,155 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-27 17:46:15,662 INFO [train.py:904] (7/8) Epoch 2, batch 3100, loss[loss=0.369, simple_loss=0.4012, pruned_loss=0.1684, over 12163.00 frames. ], tot_loss[loss=0.2742, simple_loss=0.3373, pruned_loss=0.1056, over 3310106.09 frames. ], batch size: 247, lr: 2.91e-02, grad_scale: 8.0 2023-04-27 17:46:48,768 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-27 17:47:22,159 INFO [train.py:904] (7/8) Epoch 2, batch 3150, loss[loss=0.2738, simple_loss=0.3276, pruned_loss=0.11, over 16711.00 frames. ], tot_loss[loss=0.2723, simple_loss=0.3357, pruned_loss=0.1044, over 3317413.77 frames. ], batch size: 89, lr: 2.91e-02, grad_scale: 8.0 2023-04-27 17:47:44,593 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.682e+02 3.986e+02 4.862e+02 6.258e+02 1.253e+03, threshold=9.723e+02, percent-clipped=2.0 2023-04-27 17:47:54,199 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8649, 3.6404, 3.2891, 1.5820, 2.5760, 1.9275, 3.4418, 3.6611], device='cuda:7'), covar=tensor([0.0301, 0.0419, 0.0367, 0.1897, 0.0865, 0.1141, 0.0660, 0.0420], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0112, 0.0145, 0.0157, 0.0147, 0.0141, 0.0152, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 17:48:28,285 INFO [train.py:904] (7/8) Epoch 2, batch 3200, loss[loss=0.2719, simple_loss=0.3449, pruned_loss=0.09939, over 17249.00 frames. ], tot_loss[loss=0.2707, simple_loss=0.3349, pruned_loss=0.1032, over 3326537.12 frames. ], batch size: 52, lr: 2.90e-02, grad_scale: 8.0 2023-04-27 17:49:34,233 INFO [train.py:904] (7/8) Epoch 2, batch 3250, loss[loss=0.3053, simple_loss=0.3549, pruned_loss=0.1278, over 16909.00 frames. ], tot_loss[loss=0.2701, simple_loss=0.334, pruned_loss=0.1031, over 3326002.48 frames. ], batch size: 109, lr: 2.90e-02, grad_scale: 8.0 2023-04-27 17:49:58,160 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.960e+02 4.105e+02 4.918e+02 6.860e+02 1.329e+03, threshold=9.835e+02, percent-clipped=3.0 2023-04-27 17:49:58,647 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13419.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:50:40,889 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9261, 4.0721, 2.2027, 5.1583, 4.9582, 4.3923, 2.3950, 3.3877], device='cuda:7'), covar=tensor([0.1769, 0.0359, 0.1725, 0.0065, 0.0235, 0.0368, 0.1143, 0.0626], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0114, 0.0168, 0.0070, 0.0110, 0.0110, 0.0151, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 17:50:42,674 INFO [train.py:904] (7/8) Epoch 2, batch 3300, loss[loss=0.2952, simple_loss=0.36, pruned_loss=0.1152, over 16714.00 frames. ], tot_loss[loss=0.2704, simple_loss=0.3349, pruned_loss=0.103, over 3319203.18 frames. ], batch size: 62, lr: 2.89e-02, grad_scale: 8.0 2023-04-27 17:50:56,007 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8689, 2.7280, 2.7488, 1.6190, 2.8110, 2.8500, 2.5098, 2.5061], device='cuda:7'), covar=tensor([0.0750, 0.0125, 0.0205, 0.1236, 0.0105, 0.0079, 0.0281, 0.0307], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0082, 0.0081, 0.0150, 0.0075, 0.0070, 0.0097, 0.0111], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 17:51:03,261 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=13467.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:51:48,096 INFO [train.py:904] (7/8) Epoch 2, batch 3350, loss[loss=0.3102, simple_loss=0.362, pruned_loss=0.1292, over 16836.00 frames. ], tot_loss[loss=0.2705, simple_loss=0.3354, pruned_loss=0.1028, over 3319175.84 frames. ], batch size: 102, lr: 2.89e-02, grad_scale: 8.0 2023-04-27 17:51:50,403 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0264, 3.4508, 2.1075, 4.2927, 4.1953, 4.0347, 1.7006, 3.1944], device='cuda:7'), covar=tensor([0.1357, 0.0300, 0.1384, 0.0064, 0.0170, 0.0280, 0.1052, 0.0463], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0114, 0.0167, 0.0070, 0.0109, 0.0110, 0.0150, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 17:52:13,286 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.954e+02 4.311e+02 5.188e+02 6.478e+02 1.182e+03, threshold=1.038e+03, percent-clipped=6.0 2023-04-27 17:52:56,206 INFO [train.py:904] (7/8) Epoch 2, batch 3400, loss[loss=0.2716, simple_loss=0.3279, pruned_loss=0.1077, over 16843.00 frames. ], tot_loss[loss=0.2695, simple_loss=0.3351, pruned_loss=0.102, over 3309269.81 frames. ], batch size: 90, lr: 2.88e-02, grad_scale: 8.0 2023-04-27 17:54:05,274 INFO [train.py:904] (7/8) Epoch 2, batch 3450, loss[loss=0.3164, simple_loss=0.3621, pruned_loss=0.1353, over 16522.00 frames. ], tot_loss[loss=0.2668, simple_loss=0.3325, pruned_loss=0.1005, over 3321078.67 frames. ], batch size: 68, lr: 2.88e-02, grad_scale: 8.0 2023-04-27 17:54:15,449 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4835, 2.3475, 2.2692, 2.1884, 2.9232, 2.8480, 3.9032, 3.3442], device='cuda:7'), covar=tensor([0.0022, 0.0146, 0.0133, 0.0158, 0.0090, 0.0132, 0.0045, 0.0064], device='cuda:7'), in_proj_covar=tensor([0.0049, 0.0091, 0.0087, 0.0091, 0.0081, 0.0091, 0.0056, 0.0068], device='cuda:7'), out_proj_covar=tensor([8.2709e-05, 1.4233e-04, 1.3445e-04, 1.4492e-04, 1.3405e-04, 1.5002e-04, 9.0806e-05, 1.1458e-04], device='cuda:7') 2023-04-27 17:54:29,795 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.037e+02 4.547e+02 5.342e+02 6.587e+02 1.202e+03, threshold=1.068e+03, percent-clipped=2.0 2023-04-27 17:55:11,522 INFO [train.py:904] (7/8) Epoch 2, batch 3500, loss[loss=0.2271, simple_loss=0.29, pruned_loss=0.08209, over 16812.00 frames. ], tot_loss[loss=0.2646, simple_loss=0.3303, pruned_loss=0.0994, over 3314556.03 frames. ], batch size: 39, lr: 2.88e-02, grad_scale: 8.0 2023-04-27 17:56:21,455 INFO [train.py:904] (7/8) Epoch 2, batch 3550, loss[loss=0.2261, simple_loss=0.3146, pruned_loss=0.06882, over 17141.00 frames. ], tot_loss[loss=0.2618, simple_loss=0.328, pruned_loss=0.09782, over 3325478.99 frames. ], batch size: 47, lr: 2.87e-02, grad_scale: 8.0 2023-04-27 17:56:45,025 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.316e+02 4.227e+02 5.059e+02 5.919e+02 1.365e+03, threshold=1.012e+03, percent-clipped=3.0 2023-04-27 17:57:11,516 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=13737.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 17:57:15,745 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9370, 4.6008, 4.4829, 5.1706, 5.2382, 4.7961, 5.2682, 5.1380], device='cuda:7'), covar=tensor([0.0443, 0.0574, 0.1737, 0.0531, 0.0522, 0.0418, 0.0343, 0.0419], device='cuda:7'), in_proj_covar=tensor([0.0246, 0.0284, 0.0409, 0.0306, 0.0239, 0.0226, 0.0223, 0.0234], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 17:57:15,880 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0146, 4.0879, 2.7089, 5.1917, 5.2535, 4.8610, 2.2513, 3.9508], device='cuda:7'), covar=tensor([0.1569, 0.0335, 0.1357, 0.0058, 0.0145, 0.0230, 0.1235, 0.0444], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0112, 0.0164, 0.0069, 0.0108, 0.0108, 0.0150, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 17:57:29,827 INFO [train.py:904] (7/8) Epoch 2, batch 3600, loss[loss=0.2726, simple_loss=0.347, pruned_loss=0.09915, over 17101.00 frames. ], tot_loss[loss=0.261, simple_loss=0.3272, pruned_loss=0.09741, over 3319508.76 frames. ], batch size: 48, lr: 2.87e-02, grad_scale: 8.0 2023-04-27 17:58:37,333 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=13798.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 17:58:40,998 INFO [train.py:904] (7/8) Epoch 2, batch 3650, loss[loss=0.2673, simple_loss=0.3502, pruned_loss=0.09219, over 17078.00 frames. ], tot_loss[loss=0.2612, simple_loss=0.3259, pruned_loss=0.09828, over 3299849.07 frames. ], batch size: 53, lr: 2.86e-02, grad_scale: 8.0 2023-04-27 17:59:08,122 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.959e+02 4.047e+02 4.833e+02 5.657e+02 1.025e+03, threshold=9.667e+02, percent-clipped=2.0 2023-04-27 17:59:16,886 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-27 17:59:54,884 INFO [train.py:904] (7/8) Epoch 2, batch 3700, loss[loss=0.2681, simple_loss=0.3205, pruned_loss=0.1078, over 16854.00 frames. ], tot_loss[loss=0.261, simple_loss=0.3231, pruned_loss=0.0994, over 3289903.20 frames. ], batch size: 116, lr: 2.86e-02, grad_scale: 8.0 2023-04-27 18:00:20,432 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0487, 1.8286, 1.6585, 1.5580, 2.2892, 2.3093, 2.4383, 2.3257], device='cuda:7'), covar=tensor([0.0045, 0.0144, 0.0133, 0.0171, 0.0062, 0.0096, 0.0063, 0.0073], device='cuda:7'), in_proj_covar=tensor([0.0050, 0.0090, 0.0087, 0.0092, 0.0083, 0.0091, 0.0057, 0.0067], device='cuda:7'), out_proj_covar=tensor([8.5822e-05, 1.4144e-04, 1.3402e-04, 1.4876e-04, 1.3647e-04, 1.4932e-04, 9.4612e-05, 1.1478e-04], device='cuda:7') 2023-04-27 18:01:07,711 INFO [train.py:904] (7/8) Epoch 2, batch 3750, loss[loss=0.3182, simple_loss=0.3599, pruned_loss=0.1383, over 11506.00 frames. ], tot_loss[loss=0.2629, simple_loss=0.3236, pruned_loss=0.1011, over 3275923.88 frames. ], batch size: 247, lr: 2.85e-02, grad_scale: 8.0 2023-04-27 18:01:33,976 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.812e+02 4.240e+02 4.945e+02 6.119e+02 1.020e+03, threshold=9.889e+02, percent-clipped=4.0 2023-04-27 18:01:41,195 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6942, 4.4155, 4.5366, 4.6246, 4.0593, 4.5446, 4.4519, 4.3460], device='cuda:7'), covar=tensor([0.0217, 0.0122, 0.0149, 0.0099, 0.0682, 0.0146, 0.0175, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0089, 0.0164, 0.0127, 0.0194, 0.0134, 0.0112, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:01:46,175 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8643, 3.8422, 2.9245, 3.2176, 2.9123, 1.9901, 3.9762, 4.4128], device='cuda:7'), covar=tensor([0.1457, 0.0456, 0.0957, 0.0437, 0.1586, 0.1451, 0.0247, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0245, 0.0219, 0.0235, 0.0162, 0.0249, 0.0180, 0.0184, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:01:48,351 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-04-27 18:02:24,617 INFO [train.py:904] (7/8) Epoch 2, batch 3800, loss[loss=0.2678, simple_loss=0.3297, pruned_loss=0.103, over 16791.00 frames. ], tot_loss[loss=0.2649, simple_loss=0.3245, pruned_loss=0.1027, over 3279422.96 frames. ], batch size: 96, lr: 2.85e-02, grad_scale: 8.0 2023-04-27 18:03:40,206 INFO [train.py:904] (7/8) Epoch 2, batch 3850, loss[loss=0.2445, simple_loss=0.3008, pruned_loss=0.09414, over 16783.00 frames. ], tot_loss[loss=0.2641, simple_loss=0.3236, pruned_loss=0.1023, over 3280557.73 frames. ], batch size: 124, lr: 2.84e-02, grad_scale: 8.0 2023-04-27 18:04:07,045 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.317e+02 4.007e+02 4.656e+02 5.451e+02 1.276e+03, threshold=9.312e+02, percent-clipped=3.0 2023-04-27 18:04:53,522 INFO [train.py:904] (7/8) Epoch 2, batch 3900, loss[loss=0.2731, simple_loss=0.3238, pruned_loss=0.1112, over 16556.00 frames. ], tot_loss[loss=0.2647, simple_loss=0.3234, pruned_loss=0.1031, over 3276996.56 frames. ], batch size: 146, lr: 2.84e-02, grad_scale: 8.0 2023-04-27 18:05:25,209 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14072.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:05:31,178 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7060, 4.0871, 3.3695, 3.2543, 3.1258, 2.6401, 4.3434, 4.9903], device='cuda:7'), covar=tensor([0.1890, 0.0504, 0.0949, 0.0493, 0.1794, 0.1022, 0.0280, 0.0065], device='cuda:7'), in_proj_covar=tensor([0.0246, 0.0216, 0.0234, 0.0163, 0.0257, 0.0180, 0.0186, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:05:39,938 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9623, 2.7539, 2.2441, 3.3368, 3.3180, 3.3662, 1.7737, 2.6079], device='cuda:7'), covar=tensor([0.1491, 0.0356, 0.1286, 0.0078, 0.0150, 0.0196, 0.1269, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0116, 0.0169, 0.0068, 0.0107, 0.0109, 0.0154, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:05:48,197 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0220, 3.4220, 2.3391, 4.0220, 3.8760, 3.9744, 1.6473, 3.0132], device='cuda:7'), covar=tensor([0.1437, 0.0296, 0.1274, 0.0066, 0.0200, 0.0223, 0.1192, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0115, 0.0167, 0.0068, 0.0106, 0.0108, 0.0152, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:05:55,879 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14093.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 18:06:05,410 INFO [train.py:904] (7/8) Epoch 2, batch 3950, loss[loss=0.2383, simple_loss=0.2999, pruned_loss=0.08832, over 16560.00 frames. ], tot_loss[loss=0.2651, simple_loss=0.3229, pruned_loss=0.1037, over 3258958.84 frames. ], batch size: 68, lr: 2.83e-02, grad_scale: 8.0 2023-04-27 18:06:09,655 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14103.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:06:31,645 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 3.936e+02 4.888e+02 5.915e+02 1.002e+03, threshold=9.776e+02, percent-clipped=1.0 2023-04-27 18:06:34,607 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14120.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 18:06:44,520 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7272, 3.7270, 1.6411, 3.8030, 2.5388, 3.7190, 1.7540, 2.7861], device='cuda:7'), covar=tensor([0.0072, 0.0172, 0.1696, 0.0056, 0.0768, 0.0350, 0.1561, 0.0610], device='cuda:7'), in_proj_covar=tensor([0.0082, 0.0103, 0.0172, 0.0080, 0.0158, 0.0130, 0.0178, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:06:52,569 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14133.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:07:17,643 INFO [train.py:904] (7/8) Epoch 2, batch 4000, loss[loss=0.2476, simple_loss=0.3096, pruned_loss=0.09284, over 16792.00 frames. ], tot_loss[loss=0.2655, simple_loss=0.3227, pruned_loss=0.1042, over 3268105.83 frames. ], batch size: 102, lr: 2.83e-02, grad_scale: 8.0 2023-04-27 18:07:37,798 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14164.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:08:02,161 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14181.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 18:08:31,015 INFO [train.py:904] (7/8) Epoch 2, batch 4050, loss[loss=0.2456, simple_loss=0.3183, pruned_loss=0.08643, over 16626.00 frames. ], tot_loss[loss=0.2603, simple_loss=0.3202, pruned_loss=0.1002, over 3269896.88 frames. ], batch size: 68, lr: 2.83e-02, grad_scale: 8.0 2023-04-27 18:08:55,491 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0958, 4.1933, 4.1809, 4.1761, 4.2288, 4.6845, 4.4286, 4.1561], device='cuda:7'), covar=tensor([0.1048, 0.0875, 0.0846, 0.1248, 0.1627, 0.0658, 0.0707, 0.1658], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0267, 0.0245, 0.0235, 0.0297, 0.0254, 0.0211, 0.0313], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:08:56,325 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.111e+02 3.521e+02 4.461e+02 5.736e+02 1.012e+03, threshold=8.922e+02, percent-clipped=2.0 2023-04-27 18:09:01,265 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1008, 4.1499, 1.7747, 4.1621, 2.6950, 4.1647, 2.1858, 2.8310], device='cuda:7'), covar=tensor([0.0034, 0.0075, 0.1692, 0.0037, 0.0745, 0.0164, 0.1327, 0.0607], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0097, 0.0165, 0.0078, 0.0154, 0.0124, 0.0171, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:09:05,121 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-27 18:09:18,601 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7346, 4.2730, 3.6811, 3.1060, 3.0824, 2.3441, 4.6040, 5.3674], device='cuda:7'), covar=tensor([0.1919, 0.0491, 0.0786, 0.0617, 0.1997, 0.1201, 0.0235, 0.0047], device='cuda:7'), in_proj_covar=tensor([0.0250, 0.0222, 0.0236, 0.0167, 0.0267, 0.0181, 0.0189, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:09:43,236 INFO [train.py:904] (7/8) Epoch 2, batch 4100, loss[loss=0.2834, simple_loss=0.3622, pruned_loss=0.1023, over 16895.00 frames. ], tot_loss[loss=0.2579, simple_loss=0.3196, pruned_loss=0.09807, over 3252750.04 frames. ], batch size: 96, lr: 2.82e-02, grad_scale: 8.0 2023-04-27 18:10:25,762 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6854, 4.8833, 4.5215, 4.7327, 4.2641, 4.2676, 4.4579, 4.8741], device='cuda:7'), covar=tensor([0.0395, 0.0521, 0.0809, 0.0305, 0.0451, 0.0479, 0.0422, 0.0511], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0289, 0.0257, 0.0182, 0.0202, 0.0177, 0.0235, 0.0205], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:10:44,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8616, 5.9898, 5.8073, 5.6116, 5.7607, 6.1010, 5.8208, 5.7177], device='cuda:7'), covar=tensor([0.0564, 0.0729, 0.0706, 0.1274, 0.1499, 0.0613, 0.0617, 0.1451], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0257, 0.0235, 0.0228, 0.0285, 0.0246, 0.0201, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:10:50,060 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1045, 1.5738, 1.4515, 1.4766, 1.7673, 1.6700, 1.9820, 1.9444], device='cuda:7'), covar=tensor([0.0023, 0.0096, 0.0108, 0.0116, 0.0065, 0.0110, 0.0033, 0.0046], device='cuda:7'), in_proj_covar=tensor([0.0047, 0.0093, 0.0091, 0.0093, 0.0084, 0.0092, 0.0054, 0.0066], device='cuda:7'), out_proj_covar=tensor([7.9725e-05, 1.4603e-04, 1.4123e-04, 1.5141e-04, 1.3842e-04, 1.5083e-04, 8.9156e-05, 1.1234e-04], device='cuda:7') 2023-04-27 18:10:59,420 INFO [train.py:904] (7/8) Epoch 2, batch 4150, loss[loss=0.337, simple_loss=0.3988, pruned_loss=0.1376, over 16317.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3304, pruned_loss=0.1036, over 3212454.58 frames. ], batch size: 165, lr: 2.82e-02, grad_scale: 8.0 2023-04-27 18:11:25,252 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.427e+02 3.836e+02 4.498e+02 5.328e+02 1.203e+03, threshold=8.995e+02, percent-clipped=3.0 2023-04-27 18:11:31,573 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2698, 4.6474, 4.4490, 4.5271, 4.5798, 5.0189, 4.8046, 4.5697], device='cuda:7'), covar=tensor([0.0742, 0.0807, 0.0816, 0.1250, 0.1667, 0.0665, 0.0636, 0.1525], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0257, 0.0234, 0.0229, 0.0289, 0.0244, 0.0201, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:12:14,212 INFO [train.py:904] (7/8) Epoch 2, batch 4200, loss[loss=0.2904, simple_loss=0.358, pruned_loss=0.1114, over 17193.00 frames. ], tot_loss[loss=0.2763, simple_loss=0.3393, pruned_loss=0.1067, over 3193433.17 frames. ], batch size: 44, lr: 2.81e-02, grad_scale: 8.0 2023-04-27 18:12:15,895 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8524, 3.6300, 3.4251, 1.5959, 2.8614, 2.0733, 3.2856, 3.7229], device='cuda:7'), covar=tensor([0.0327, 0.0472, 0.0335, 0.1836, 0.0682, 0.1106, 0.0646, 0.0468], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0102, 0.0145, 0.0155, 0.0146, 0.0137, 0.0149, 0.0096], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 18:13:15,620 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14393.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:13:26,264 INFO [train.py:904] (7/8) Epoch 2, batch 4250, loss[loss=0.2693, simple_loss=0.3478, pruned_loss=0.09545, over 16499.00 frames. ], tot_loss[loss=0.2781, simple_loss=0.3423, pruned_loss=0.1069, over 3178861.86 frames. ], batch size: 68, lr: 2.81e-02, grad_scale: 8.0 2023-04-27 18:13:44,198 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14413.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:13:52,835 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.326e+02 4.073e+02 4.856e+02 5.593e+02 1.302e+03, threshold=9.713e+02, percent-clipped=3.0 2023-04-27 18:14:05,697 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14428.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:14:24,146 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14441.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:14:38,556 INFO [train.py:904] (7/8) Epoch 2, batch 4300, loss[loss=0.274, simple_loss=0.351, pruned_loss=0.09849, over 17231.00 frames. ], tot_loss[loss=0.2761, simple_loss=0.3425, pruned_loss=0.1048, over 3193890.18 frames. ], batch size: 44, lr: 2.80e-02, grad_scale: 16.0 2023-04-27 18:14:49,280 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14459.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:14:53,301 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-27 18:15:10,459 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14474.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:15:12,751 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14476.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 18:15:29,986 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14487.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:15:49,382 INFO [train.py:904] (7/8) Epoch 2, batch 4350, loss[loss=0.2571, simple_loss=0.3374, pruned_loss=0.08842, over 17120.00 frames. ], tot_loss[loss=0.2792, simple_loss=0.3462, pruned_loss=0.1061, over 3198367.66 frames. ], batch size: 47, lr: 2.80e-02, grad_scale: 16.0 2023-04-27 18:15:51,328 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-27 18:15:58,423 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-27 18:16:13,776 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.262e+02 4.011e+02 4.650e+02 6.201e+02 1.293e+03, threshold=9.301e+02, percent-clipped=5.0 2023-04-27 18:16:30,799 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1836, 4.0894, 4.0128, 4.1260, 3.6348, 4.2088, 3.9425, 3.7842], device='cuda:7'), covar=tensor([0.0208, 0.0096, 0.0144, 0.0096, 0.0590, 0.0108, 0.0247, 0.0194], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0080, 0.0143, 0.0112, 0.0173, 0.0117, 0.0101, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:16:56,177 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14548.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:16:59,407 INFO [train.py:904] (7/8) Epoch 2, batch 4400, loss[loss=0.2705, simple_loss=0.35, pruned_loss=0.09551, over 16493.00 frames. ], tot_loss[loss=0.2808, simple_loss=0.3482, pruned_loss=0.1066, over 3203754.31 frames. ], batch size: 35, lr: 2.79e-02, grad_scale: 16.0 2023-04-27 18:18:09,497 INFO [train.py:904] (7/8) Epoch 2, batch 4450, loss[loss=0.258, simple_loss=0.3329, pruned_loss=0.09155, over 16997.00 frames. ], tot_loss[loss=0.2814, simple_loss=0.3502, pruned_loss=0.1063, over 3202679.10 frames. ], batch size: 53, lr: 2.79e-02, grad_scale: 16.0 2023-04-27 18:18:34,421 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.338e+02 3.415e+02 4.231e+02 5.024e+02 9.561e+02, threshold=8.461e+02, percent-clipped=1.0 2023-04-27 18:18:38,130 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4664, 4.5245, 2.3215, 5.5134, 5.2119, 4.2083, 2.7575, 3.3764], device='cuda:7'), covar=tensor([0.1310, 0.0247, 0.1562, 0.0032, 0.0083, 0.0204, 0.0961, 0.0568], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0114, 0.0164, 0.0064, 0.0095, 0.0105, 0.0151, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:19:17,781 INFO [train.py:904] (7/8) Epoch 2, batch 4500, loss[loss=0.246, simple_loss=0.3184, pruned_loss=0.08682, over 16349.00 frames. ], tot_loss[loss=0.2785, simple_loss=0.3482, pruned_loss=0.1044, over 3212163.18 frames. ], batch size: 35, lr: 2.79e-02, grad_scale: 16.0 2023-04-27 18:19:36,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4331, 4.1096, 3.8611, 1.7888, 2.9879, 2.4831, 3.7574, 4.3631], device='cuda:7'), covar=tensor([0.0244, 0.0386, 0.0397, 0.1944, 0.0803, 0.1114, 0.0713, 0.0333], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0104, 0.0149, 0.0158, 0.0148, 0.0139, 0.0152, 0.0095], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 18:20:24,613 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2023-04-27 18:20:28,470 INFO [train.py:904] (7/8) Epoch 2, batch 4550, loss[loss=0.2815, simple_loss=0.3629, pruned_loss=0.1001, over 17014.00 frames. ], tot_loss[loss=0.2798, simple_loss=0.3492, pruned_loss=0.1052, over 3203666.34 frames. ], batch size: 50, lr: 2.78e-02, grad_scale: 16.0 2023-04-27 18:20:54,777 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.252e+02 3.646e+02 4.280e+02 5.239e+02 1.001e+03, threshold=8.560e+02, percent-clipped=1.0 2023-04-27 18:21:08,882 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14728.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:21:25,086 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14740.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:21:43,179 INFO [train.py:904] (7/8) Epoch 2, batch 4600, loss[loss=0.2776, simple_loss=0.3548, pruned_loss=0.1002, over 16646.00 frames. ], tot_loss[loss=0.2792, simple_loss=0.3493, pruned_loss=0.1045, over 3193803.23 frames. ], batch size: 134, lr: 2.78e-02, grad_scale: 16.0 2023-04-27 18:21:54,091 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14759.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:22:10,210 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14769.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:22:18,632 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14776.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:22:18,690 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14776.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 18:22:40,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6574, 4.9237, 4.5336, 4.7142, 4.2417, 4.2424, 4.3610, 4.9808], device='cuda:7'), covar=tensor([0.0333, 0.0433, 0.0656, 0.0297, 0.0443, 0.0421, 0.0429, 0.0484], device='cuda:7'), in_proj_covar=tensor([0.0205, 0.0269, 0.0254, 0.0177, 0.0190, 0.0168, 0.0220, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:22:54,489 INFO [train.py:904] (7/8) Epoch 2, batch 4650, loss[loss=0.2608, simple_loss=0.3346, pruned_loss=0.09348, over 16498.00 frames. ], tot_loss[loss=0.2766, simple_loss=0.3467, pruned_loss=0.1033, over 3195508.67 frames. ], batch size: 68, lr: 2.77e-02, grad_scale: 16.0 2023-04-27 18:22:55,636 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14801.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:23:03,204 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14807.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:23:15,519 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6539, 1.9405, 1.5518, 1.7224, 2.2414, 2.1744, 2.5106, 2.5320], device='cuda:7'), covar=tensor([0.0018, 0.0160, 0.0199, 0.0191, 0.0086, 0.0146, 0.0036, 0.0058], device='cuda:7'), in_proj_covar=tensor([0.0041, 0.0092, 0.0091, 0.0095, 0.0082, 0.0093, 0.0052, 0.0065], device='cuda:7'), out_proj_covar=tensor([6.6840e-05, 1.4355e-04, 1.4092e-04, 1.5446e-04, 1.3550e-04, 1.5098e-04, 8.5875e-05, 1.0918e-04], device='cuda:7') 2023-04-27 18:23:20,767 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.700e+02 3.221e+02 3.729e+02 4.286e+02 9.218e+02, threshold=7.459e+02, percent-clipped=2.0 2023-04-27 18:23:27,861 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=14824.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 18:23:54,907 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14843.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:24:06,751 INFO [train.py:904] (7/8) Epoch 2, batch 4700, loss[loss=0.2664, simple_loss=0.3309, pruned_loss=0.1009, over 16554.00 frames. ], tot_loss[loss=0.2746, simple_loss=0.3444, pruned_loss=0.1024, over 3193221.19 frames. ], batch size: 62, lr: 2.77e-02, grad_scale: 16.0 2023-04-27 18:25:20,209 INFO [train.py:904] (7/8) Epoch 2, batch 4750, loss[loss=0.2241, simple_loss=0.3034, pruned_loss=0.07236, over 16807.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.3414, pruned_loss=0.1011, over 3190827.00 frames. ], batch size: 83, lr: 2.77e-02, grad_scale: 16.0 2023-04-27 18:25:45,845 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.318e+02 3.402e+02 4.045e+02 5.011e+02 1.217e+03, threshold=8.089e+02, percent-clipped=2.0 2023-04-27 18:26:30,659 INFO [train.py:904] (7/8) Epoch 2, batch 4800, loss[loss=0.2396, simple_loss=0.3246, pruned_loss=0.07726, over 16726.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3365, pruned_loss=0.09763, over 3211717.98 frames. ], batch size: 83, lr: 2.76e-02, grad_scale: 16.0 2023-04-27 18:27:43,246 INFO [train.py:904] (7/8) Epoch 2, batch 4850, loss[loss=0.2274, simple_loss=0.3162, pruned_loss=0.06936, over 16920.00 frames. ], tot_loss[loss=0.2656, simple_loss=0.3377, pruned_loss=0.09675, over 3222040.47 frames. ], batch size: 90, lr: 2.76e-02, grad_scale: 16.0 2023-04-27 18:27:48,107 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5378, 1.8007, 1.5858, 1.4398, 2.2928, 1.9702, 2.4180, 2.3916], device='cuda:7'), covar=tensor([0.0020, 0.0140, 0.0156, 0.0205, 0.0061, 0.0147, 0.0035, 0.0060], device='cuda:7'), in_proj_covar=tensor([0.0041, 0.0091, 0.0090, 0.0096, 0.0081, 0.0095, 0.0051, 0.0065], device='cuda:7'), out_proj_covar=tensor([6.5077e-05, 1.4354e-04, 1.3939e-04, 1.5612e-04, 1.3311e-04, 1.5579e-04, 8.3446e-05, 1.0968e-04], device='cuda:7') 2023-04-27 18:28:11,116 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.137e+02 3.566e+02 4.230e+02 5.018e+02 9.509e+02, threshold=8.461e+02, percent-clipped=3.0 2023-04-27 18:28:58,611 INFO [train.py:904] (7/8) Epoch 2, batch 4900, loss[loss=0.2571, simple_loss=0.3271, pruned_loss=0.09353, over 16350.00 frames. ], tot_loss[loss=0.2642, simple_loss=0.3368, pruned_loss=0.09582, over 3207511.57 frames. ], batch size: 35, lr: 2.75e-02, grad_scale: 16.0 2023-04-27 18:29:24,612 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15069.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:30:04,942 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15096.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:30:10,704 INFO [train.py:904] (7/8) Epoch 2, batch 4950, loss[loss=0.2389, simple_loss=0.3164, pruned_loss=0.08075, over 17006.00 frames. ], tot_loss[loss=0.2645, simple_loss=0.3371, pruned_loss=0.09596, over 3214485.54 frames. ], batch size: 55, lr: 2.75e-02, grad_scale: 16.0 2023-04-27 18:30:24,073 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2271, 4.1936, 4.6314, 4.6455, 4.6415, 4.2488, 4.2340, 4.3056], device='cuda:7'), covar=tensor([0.0192, 0.0212, 0.0257, 0.0272, 0.0356, 0.0198, 0.0578, 0.0232], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0141, 0.0164, 0.0158, 0.0184, 0.0151, 0.0230, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-27 18:30:31,192 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1640, 1.4267, 2.1755, 2.7143, 3.0878, 2.8506, 1.4765, 3.1289], device='cuda:7'), covar=tensor([0.0035, 0.0268, 0.0140, 0.0100, 0.0021, 0.0059, 0.0208, 0.0024], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0109, 0.0092, 0.0080, 0.0055, 0.0057, 0.0091, 0.0052], device='cuda:7'), out_proj_covar=tensor([1.2439e-04, 1.9456e-04, 1.7215e-04, 1.5140e-04, 9.4729e-05, 1.0458e-04, 1.5835e-04, 9.2852e-05], device='cuda:7') 2023-04-27 18:30:34,535 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15117.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:30:37,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.417e+02 3.788e+02 4.518e+02 5.433e+02 9.876e+02, threshold=9.036e+02, percent-clipped=3.0 2023-04-27 18:31:12,982 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15143.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:31:13,111 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9354, 3.8107, 3.2256, 1.5190, 2.8089, 2.1193, 3.2093, 4.0482], device='cuda:7'), covar=tensor([0.0305, 0.0359, 0.0536, 0.1748, 0.0750, 0.1035, 0.0751, 0.0268], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0104, 0.0154, 0.0155, 0.0149, 0.0141, 0.0152, 0.0097], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 18:31:15,458 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3017, 3.1072, 1.4712, 3.1597, 2.1433, 3.1161, 1.6259, 2.4832], device='cuda:7'), covar=tensor([0.0035, 0.0173, 0.1503, 0.0065, 0.0755, 0.0289, 0.1265, 0.0538], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0095, 0.0162, 0.0076, 0.0151, 0.0118, 0.0169, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:31:22,569 INFO [train.py:904] (7/8) Epoch 2, batch 5000, loss[loss=0.2661, simple_loss=0.3452, pruned_loss=0.09348, over 16704.00 frames. ], tot_loss[loss=0.2662, simple_loss=0.3391, pruned_loss=0.09665, over 3211238.37 frames. ], batch size: 134, lr: 2.74e-02, grad_scale: 16.0 2023-04-27 18:32:08,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0387, 3.6048, 2.3740, 4.8247, 4.7104, 4.1898, 2.4817, 3.1036], device='cuda:7'), covar=tensor([0.1190, 0.0312, 0.1201, 0.0039, 0.0080, 0.0227, 0.0817, 0.0512], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0118, 0.0164, 0.0067, 0.0099, 0.0108, 0.0147, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:32:13,418 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-27 18:32:21,458 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15191.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:32:26,384 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15195.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:32:34,519 INFO [train.py:904] (7/8) Epoch 2, batch 5050, loss[loss=0.3217, simple_loss=0.3741, pruned_loss=0.1346, over 12024.00 frames. ], tot_loss[loss=0.2656, simple_loss=0.3393, pruned_loss=0.09599, over 3203687.01 frames. ], batch size: 247, lr: 2.74e-02, grad_scale: 16.0 2023-04-27 18:33:00,585 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.434e+02 3.699e+02 4.750e+02 5.849e+02 1.293e+03, threshold=9.501e+02, percent-clipped=2.0 2023-04-27 18:33:45,527 INFO [train.py:904] (7/8) Epoch 2, batch 5100, loss[loss=0.2511, simple_loss=0.3242, pruned_loss=0.08899, over 16569.00 frames. ], tot_loss[loss=0.2616, simple_loss=0.3356, pruned_loss=0.09382, over 3218973.60 frames. ], batch size: 75, lr: 2.74e-02, grad_scale: 16.0 2023-04-27 18:33:53,409 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15256.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 18:34:56,473 INFO [train.py:904] (7/8) Epoch 2, batch 5150, loss[loss=0.2783, simple_loss=0.3538, pruned_loss=0.1014, over 16836.00 frames. ], tot_loss[loss=0.2609, simple_loss=0.3359, pruned_loss=0.09297, over 3213181.53 frames. ], batch size: 116, lr: 2.73e-02, grad_scale: 16.0 2023-04-27 18:35:03,392 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6400, 1.8122, 1.6296, 1.6545, 2.3287, 2.1790, 2.6384, 2.5339], device='cuda:7'), covar=tensor([0.0019, 0.0166, 0.0153, 0.0185, 0.0066, 0.0119, 0.0030, 0.0056], device='cuda:7'), in_proj_covar=tensor([0.0043, 0.0095, 0.0093, 0.0098, 0.0084, 0.0098, 0.0053, 0.0067], device='cuda:7'), out_proj_covar=tensor([6.5883e-05, 1.5022e-04, 1.4339e-04, 1.5917e-04, 1.3811e-04, 1.5981e-04, 8.6336e-05, 1.1259e-04], device='cuda:7') 2023-04-27 18:35:22,261 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.454e+02 3.560e+02 4.122e+02 5.032e+02 1.152e+03, threshold=8.244e+02, percent-clipped=1.0 2023-04-27 18:36:05,913 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7218, 3.2968, 3.2163, 2.4109, 3.0033, 3.0738, 3.2910, 1.7414], device='cuda:7'), covar=tensor([0.0659, 0.0032, 0.0060, 0.0320, 0.0053, 0.0121, 0.0049, 0.0500], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0050, 0.0059, 0.0110, 0.0053, 0.0058, 0.0058, 0.0102], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:36:10,029 INFO [train.py:904] (7/8) Epoch 2, batch 5200, loss[loss=0.2627, simple_loss=0.3368, pruned_loss=0.09432, over 16741.00 frames. ], tot_loss[loss=0.2617, simple_loss=0.3359, pruned_loss=0.09369, over 3208316.90 frames. ], batch size: 89, lr: 2.73e-02, grad_scale: 16.0 2023-04-27 18:36:15,919 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8748, 3.6959, 3.1552, 1.5701, 2.6816, 2.1754, 3.3259, 3.8202], device='cuda:7'), covar=tensor([0.0219, 0.0384, 0.0480, 0.1767, 0.0762, 0.1021, 0.0693, 0.0235], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0104, 0.0151, 0.0154, 0.0145, 0.0140, 0.0149, 0.0097], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 18:37:16,393 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15396.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:37:23,470 INFO [train.py:904] (7/8) Epoch 2, batch 5250, loss[loss=0.3147, simple_loss=0.3607, pruned_loss=0.1343, over 12382.00 frames. ], tot_loss[loss=0.2598, simple_loss=0.333, pruned_loss=0.09326, over 3212603.79 frames. ], batch size: 247, lr: 2.72e-02, grad_scale: 16.0 2023-04-27 18:37:37,031 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0588, 4.1776, 4.1507, 4.0710, 4.1481, 4.6252, 4.4749, 4.0282], device='cuda:7'), covar=tensor([0.0953, 0.0893, 0.0909, 0.1120, 0.1595, 0.0630, 0.0574, 0.1527], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0264, 0.0243, 0.0232, 0.0299, 0.0246, 0.0194, 0.0307], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:37:48,867 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.244e+02 3.713e+02 4.444e+02 5.638e+02 1.103e+03, threshold=8.887e+02, percent-clipped=2.0 2023-04-27 18:38:23,204 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15444.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:38:32,791 INFO [train.py:904] (7/8) Epoch 2, batch 5300, loss[loss=0.2249, simple_loss=0.303, pruned_loss=0.07341, over 16456.00 frames. ], tot_loss[loss=0.256, simple_loss=0.3288, pruned_loss=0.09156, over 3214378.09 frames. ], batch size: 146, lr: 2.72e-02, grad_scale: 16.0 2023-04-27 18:39:12,943 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-27 18:39:43,360 INFO [train.py:904] (7/8) Epoch 2, batch 5350, loss[loss=0.2589, simple_loss=0.3433, pruned_loss=0.08722, over 16770.00 frames. ], tot_loss[loss=0.2543, simple_loss=0.327, pruned_loss=0.09081, over 3207418.10 frames. ], batch size: 83, lr: 2.72e-02, grad_scale: 16.0 2023-04-27 18:40:09,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.455e+02 3.880e+02 4.934e+02 5.714e+02 1.287e+03, threshold=9.868e+02, percent-clipped=1.0 2023-04-27 18:40:56,838 INFO [train.py:904] (7/8) Epoch 2, batch 5400, loss[loss=0.2684, simple_loss=0.3379, pruned_loss=0.09948, over 16525.00 frames. ], tot_loss[loss=0.2586, simple_loss=0.3315, pruned_loss=0.09288, over 3194801.01 frames. ], batch size: 68, lr: 2.71e-02, grad_scale: 16.0 2023-04-27 18:40:57,232 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15551.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 18:41:22,558 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3831, 4.1941, 4.1753, 3.5889, 4.1449, 1.6304, 3.8980, 4.1853], device='cuda:7'), covar=tensor([0.0059, 0.0054, 0.0056, 0.0280, 0.0054, 0.1251, 0.0072, 0.0089], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0051, 0.0074, 0.0098, 0.0060, 0.0107, 0.0070, 0.0075], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:42:14,673 INFO [train.py:904] (7/8) Epoch 2, batch 5450, loss[loss=0.3458, simple_loss=0.3901, pruned_loss=0.1508, over 11875.00 frames. ], tot_loss[loss=0.2635, simple_loss=0.3358, pruned_loss=0.09562, over 3189017.29 frames. ], batch size: 246, lr: 2.71e-02, grad_scale: 16.0 2023-04-27 18:42:43,048 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.524e+02 4.170e+02 4.882e+02 6.060e+02 1.199e+03, threshold=9.765e+02, percent-clipped=3.0 2023-04-27 18:43:34,041 INFO [train.py:904] (7/8) Epoch 2, batch 5500, loss[loss=0.2925, simple_loss=0.3635, pruned_loss=0.1107, over 16844.00 frames. ], tot_loss[loss=0.2778, simple_loss=0.3465, pruned_loss=0.1045, over 3162740.11 frames. ], batch size: 102, lr: 2.70e-02, grad_scale: 16.0 2023-04-27 18:43:51,328 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7167, 4.6826, 4.5118, 3.9297, 4.5310, 2.0556, 4.3519, 4.5857], device='cuda:7'), covar=tensor([0.0076, 0.0042, 0.0064, 0.0306, 0.0056, 0.1164, 0.0064, 0.0081], device='cuda:7'), in_proj_covar=tensor([0.0062, 0.0051, 0.0076, 0.0099, 0.0060, 0.0108, 0.0070, 0.0077], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:44:50,011 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4572, 4.3393, 4.2465, 3.5828, 4.2912, 2.0159, 4.1825, 4.2584], device='cuda:7'), covar=tensor([0.0071, 0.0056, 0.0061, 0.0341, 0.0056, 0.1227, 0.0060, 0.0102], device='cuda:7'), in_proj_covar=tensor([0.0062, 0.0051, 0.0075, 0.0099, 0.0059, 0.0107, 0.0070, 0.0077], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:44:50,703 INFO [train.py:904] (7/8) Epoch 2, batch 5550, loss[loss=0.2727, simple_loss=0.357, pruned_loss=0.09421, over 16809.00 frames. ], tot_loss[loss=0.2909, simple_loss=0.3562, pruned_loss=0.1129, over 3149304.62 frames. ], batch size: 102, lr: 2.70e-02, grad_scale: 16.0 2023-04-27 18:45:19,405 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.692e+02 5.827e+02 6.984e+02 8.601e+02 1.757e+03, threshold=1.397e+03, percent-clipped=15.0 2023-04-27 18:45:25,285 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-04-27 18:45:59,865 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9015, 3.8130, 3.6767, 2.7853, 3.5270, 3.6848, 3.7907, 1.8087], device='cuda:7'), covar=tensor([0.0643, 0.0031, 0.0058, 0.0292, 0.0049, 0.0110, 0.0038, 0.0526], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0052, 0.0058, 0.0109, 0.0053, 0.0058, 0.0058, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 18:46:11,061 INFO [train.py:904] (7/8) Epoch 2, batch 5600, loss[loss=0.4179, simple_loss=0.434, pruned_loss=0.2009, over 11121.00 frames. ], tot_loss[loss=0.3004, simple_loss=0.3632, pruned_loss=0.1188, over 3134076.40 frames. ], batch size: 248, lr: 2.70e-02, grad_scale: 16.0 2023-04-27 18:46:38,649 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2589, 1.6300, 2.4189, 3.0551, 2.9854, 3.3035, 1.4558, 3.2608], device='cuda:7'), covar=tensor([0.0033, 0.0227, 0.0115, 0.0062, 0.0030, 0.0040, 0.0205, 0.0033], device='cuda:7'), in_proj_covar=tensor([0.0072, 0.0110, 0.0094, 0.0081, 0.0059, 0.0057, 0.0093, 0.0054], device='cuda:7'), out_proj_covar=tensor([1.2745e-04, 1.9540e-04, 1.7351e-04, 1.5243e-04, 1.0194e-04, 1.0412e-04, 1.6058e-04, 9.5611e-05], device='cuda:7') 2023-04-27 18:47:17,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5642, 3.6282, 1.4266, 3.5759, 2.5148, 3.6341, 1.9431, 2.7190], device='cuda:7'), covar=tensor([0.0035, 0.0102, 0.1741, 0.0045, 0.0656, 0.0217, 0.1281, 0.0550], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0096, 0.0160, 0.0073, 0.0151, 0.0121, 0.0167, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:47:34,301 INFO [train.py:904] (7/8) Epoch 2, batch 5650, loss[loss=0.4441, simple_loss=0.4551, pruned_loss=0.2165, over 11140.00 frames. ], tot_loss[loss=0.31, simple_loss=0.3703, pruned_loss=0.1248, over 3116983.98 frames. ], batch size: 247, lr: 2.69e-02, grad_scale: 16.0 2023-04-27 18:48:01,999 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.886e+02 5.674e+02 6.824e+02 8.519e+02 2.118e+03, threshold=1.365e+03, percent-clipped=2.0 2023-04-27 18:48:05,164 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7041, 2.8510, 2.3614, 4.0383, 2.1299, 3.8246, 2.3629, 2.4999], device='cuda:7'), covar=tensor([0.0259, 0.0348, 0.0301, 0.0159, 0.1187, 0.0169, 0.0572, 0.0914], device='cuda:7'), in_proj_covar=tensor([0.0200, 0.0170, 0.0145, 0.0200, 0.0254, 0.0160, 0.0177, 0.0231], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:48:50,808 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5972, 4.3755, 4.6131, 4.9797, 4.9773, 4.3625, 5.0237, 4.8548], device='cuda:7'), covar=tensor([0.0461, 0.0574, 0.0897, 0.0233, 0.0281, 0.0423, 0.0200, 0.0285], device='cuda:7'), in_proj_covar=tensor([0.0225, 0.0262, 0.0360, 0.0269, 0.0204, 0.0191, 0.0198, 0.0209], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:48:53,343 INFO [train.py:904] (7/8) Epoch 2, batch 5700, loss[loss=0.4086, simple_loss=0.4237, pruned_loss=0.1967, over 11178.00 frames. ], tot_loss[loss=0.3135, simple_loss=0.3726, pruned_loss=0.1272, over 3120912.10 frames. ], batch size: 248, lr: 2.69e-02, grad_scale: 16.0 2023-04-27 18:48:53,951 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15851.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:49:44,691 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15883.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:50:08,490 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-27 18:50:09,242 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=15899.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:50:12,124 INFO [train.py:904] (7/8) Epoch 2, batch 5750, loss[loss=0.286, simple_loss=0.361, pruned_loss=0.1055, over 16663.00 frames. ], tot_loss[loss=0.3174, simple_loss=0.3759, pruned_loss=0.1295, over 3081488.01 frames. ], batch size: 134, lr: 2.69e-02, grad_scale: 8.0 2023-04-27 18:50:17,382 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4460, 3.4104, 3.2693, 2.9118, 3.3552, 2.0662, 3.1900, 3.1627], device='cuda:7'), covar=tensor([0.0066, 0.0049, 0.0080, 0.0284, 0.0064, 0.1137, 0.0077, 0.0098], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0050, 0.0074, 0.0097, 0.0057, 0.0108, 0.0069, 0.0074], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:50:42,009 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.285e+02 4.919e+02 6.542e+02 8.153e+02 1.932e+03, threshold=1.308e+03, percent-clipped=2.0 2023-04-27 18:51:22,908 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15944.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:51:33,510 INFO [train.py:904] (7/8) Epoch 2, batch 5800, loss[loss=0.2731, simple_loss=0.3465, pruned_loss=0.09983, over 16431.00 frames. ], tot_loss[loss=0.3166, simple_loss=0.3761, pruned_loss=0.1285, over 3067358.82 frames. ], batch size: 146, lr: 2.68e-02, grad_scale: 8.0 2023-04-27 18:51:40,528 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1544, 3.6248, 2.6573, 4.9288, 4.8781, 4.4048, 2.4263, 3.5186], device='cuda:7'), covar=tensor([0.1391, 0.0391, 0.1220, 0.0054, 0.0147, 0.0282, 0.1026, 0.0541], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0118, 0.0166, 0.0067, 0.0103, 0.0114, 0.0152, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:52:22,829 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0393, 3.8280, 3.5738, 4.3103, 4.3456, 3.9852, 4.3358, 4.2287], device='cuda:7'), covar=tensor([0.0545, 0.0548, 0.1711, 0.0501, 0.0487, 0.0543, 0.0437, 0.0507], device='cuda:7'), in_proj_covar=tensor([0.0226, 0.0261, 0.0362, 0.0269, 0.0206, 0.0192, 0.0202, 0.0211], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:52:49,529 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-04-27 18:52:56,030 INFO [train.py:904] (7/8) Epoch 2, batch 5850, loss[loss=0.329, simple_loss=0.3832, pruned_loss=0.1374, over 15324.00 frames. ], tot_loss[loss=0.3143, simple_loss=0.3743, pruned_loss=0.1271, over 3059504.64 frames. ], batch size: 190, lr: 2.68e-02, grad_scale: 8.0 2023-04-27 18:52:57,414 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16001.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:53:11,372 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7279, 3.1500, 2.1556, 4.0411, 3.9100, 3.8050, 1.9194, 2.7556], device='cuda:7'), covar=tensor([0.1552, 0.0385, 0.1395, 0.0058, 0.0163, 0.0240, 0.1117, 0.0742], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0118, 0.0166, 0.0067, 0.0103, 0.0113, 0.0151, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:53:25,313 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.722e+02 4.866e+02 5.967e+02 7.152e+02 1.262e+03, threshold=1.193e+03, percent-clipped=0.0 2023-04-27 18:54:09,115 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-27 18:54:18,256 INFO [train.py:904] (7/8) Epoch 2, batch 5900, loss[loss=0.2622, simple_loss=0.3405, pruned_loss=0.092, over 16697.00 frames. ], tot_loss[loss=0.3122, simple_loss=0.3731, pruned_loss=0.1256, over 3073634.69 frames. ], batch size: 57, lr: 2.67e-02, grad_scale: 8.0 2023-04-27 18:54:27,720 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 18:54:40,602 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16062.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:55:02,363 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2267, 1.5264, 2.4245, 3.0398, 3.1061, 3.2901, 1.5255, 3.2257], device='cuda:7'), covar=tensor([0.0034, 0.0235, 0.0118, 0.0079, 0.0027, 0.0044, 0.0201, 0.0039], device='cuda:7'), in_proj_covar=tensor([0.0067, 0.0105, 0.0089, 0.0078, 0.0055, 0.0056, 0.0088, 0.0051], device='cuda:7'), out_proj_covar=tensor([1.1931e-04, 1.8483e-04, 1.6436e-04, 1.4519e-04, 9.4044e-05, 1.0028e-04, 1.5184e-04, 8.9439e-05], device='cuda:7') 2023-04-27 18:55:17,964 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16085.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:55:23,702 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1424, 5.5245, 5.1242, 5.3867, 4.7958, 4.6708, 5.0980, 5.6104], device='cuda:7'), covar=tensor([0.0528, 0.0565, 0.0874, 0.0320, 0.0551, 0.0435, 0.0406, 0.0557], device='cuda:7'), in_proj_covar=tensor([0.0212, 0.0291, 0.0272, 0.0188, 0.0201, 0.0184, 0.0240, 0.0207], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 18:55:42,173 INFO [train.py:904] (7/8) Epoch 2, batch 5950, loss[loss=0.3061, simple_loss=0.3721, pruned_loss=0.12, over 16929.00 frames. ], tot_loss[loss=0.3094, simple_loss=0.3725, pruned_loss=0.1232, over 3070245.13 frames. ], batch size: 109, lr: 2.67e-02, grad_scale: 8.0 2023-04-27 18:56:13,608 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.088e+02 4.747e+02 6.406e+02 8.221e+02 1.979e+03, threshold=1.281e+03, percent-clipped=5.0 2023-04-27 18:56:56,702 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16146.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:57:01,960 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16149.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:57:04,569 INFO [train.py:904] (7/8) Epoch 2, batch 6000, loss[loss=0.2842, simple_loss=0.3505, pruned_loss=0.1089, over 16590.00 frames. ], tot_loss[loss=0.3084, simple_loss=0.3718, pruned_loss=0.1225, over 3100326.76 frames. ], batch size: 75, lr: 2.67e-02, grad_scale: 8.0 2023-04-27 18:57:04,570 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 18:57:15,927 INFO [train.py:938] (7/8) Epoch 2, validation: loss=0.2302, simple_loss=0.3372, pruned_loss=0.06166, over 944034.00 frames. 2023-04-27 18:57:15,928 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 18:58:05,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2932, 1.4656, 1.7685, 2.1178, 2.2128, 2.0409, 1.4659, 2.2022], device='cuda:7'), covar=tensor([0.0049, 0.0236, 0.0112, 0.0105, 0.0040, 0.0080, 0.0171, 0.0044], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0108, 0.0092, 0.0081, 0.0058, 0.0058, 0.0091, 0.0053], device='cuda:7'), out_proj_covar=tensor([1.2274e-04, 1.9124e-04, 1.6812e-04, 1.5142e-04, 9.9042e-05, 1.0449e-04, 1.5573e-04, 9.2256e-05], device='cuda:7') 2023-04-27 18:58:12,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6965, 3.6035, 2.4081, 4.5653, 4.5602, 4.1184, 2.1700, 3.2078], device='cuda:7'), covar=tensor([0.1659, 0.0365, 0.1363, 0.0062, 0.0140, 0.0261, 0.1052, 0.0636], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0119, 0.0164, 0.0067, 0.0104, 0.0113, 0.0149, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 18:58:34,801 INFO [train.py:904] (7/8) Epoch 2, batch 6050, loss[loss=0.2562, simple_loss=0.338, pruned_loss=0.08721, over 16456.00 frames. ], tot_loss[loss=0.3057, simple_loss=0.3695, pruned_loss=0.121, over 3118495.60 frames. ], batch size: 68, lr: 2.66e-02, grad_scale: 8.0 2023-04-27 18:58:48,631 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16210.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:59:04,138 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.028e+02 4.577e+02 5.718e+02 7.421e+02 1.323e+03, threshold=1.144e+03, percent-clipped=1.0 2023-04-27 18:59:12,772 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 18:59:24,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5237, 4.2984, 1.6059, 4.3594, 2.7945, 4.3570, 2.0526, 3.0040], device='cuda:7'), covar=tensor([0.0027, 0.0114, 0.1849, 0.0030, 0.0733, 0.0224, 0.1506, 0.0603], device='cuda:7'), in_proj_covar=tensor([0.0074, 0.0103, 0.0170, 0.0076, 0.0160, 0.0130, 0.0176, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 18:59:33,680 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16239.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 18:59:52,056 INFO [train.py:904] (7/8) Epoch 2, batch 6100, loss[loss=0.276, simple_loss=0.3496, pruned_loss=0.1012, over 16587.00 frames. ], tot_loss[loss=0.3023, simple_loss=0.3676, pruned_loss=0.1185, over 3135629.49 frames. ], batch size: 68, lr: 2.66e-02, grad_scale: 8.0 2023-04-27 19:01:08,905 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2023-04-27 19:01:17,158 INFO [train.py:904] (7/8) Epoch 2, batch 6150, loss[loss=0.3143, simple_loss=0.3802, pruned_loss=0.1242, over 15320.00 frames. ], tot_loss[loss=0.3007, simple_loss=0.3657, pruned_loss=0.1178, over 3138081.36 frames. ], batch size: 190, lr: 2.66e-02, grad_scale: 8.0 2023-04-27 19:01:25,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6980, 3.6367, 3.2344, 2.9570, 2.9232, 2.3552, 3.9879, 4.6824], device='cuda:7'), covar=tensor([0.1781, 0.0694, 0.0998, 0.0690, 0.1656, 0.1092, 0.0342, 0.0095], device='cuda:7'), in_proj_covar=tensor([0.0253, 0.0223, 0.0239, 0.0180, 0.0276, 0.0180, 0.0196, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:01:45,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.137e+02 4.737e+02 5.827e+02 7.174e+02 1.739e+03, threshold=1.165e+03, percent-clipped=5.0 2023-04-27 19:02:19,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8949, 3.8685, 2.7682, 4.9336, 4.8215, 4.3553, 1.9899, 3.2472], device='cuda:7'), covar=tensor([0.1509, 0.0301, 0.1158, 0.0050, 0.0143, 0.0246, 0.1167, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0119, 0.0167, 0.0068, 0.0105, 0.0116, 0.0152, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 19:02:34,536 INFO [train.py:904] (7/8) Epoch 2, batch 6200, loss[loss=0.3085, simple_loss=0.3657, pruned_loss=0.1257, over 15514.00 frames. ], tot_loss[loss=0.2978, simple_loss=0.3626, pruned_loss=0.1165, over 3145661.73 frames. ], batch size: 191, lr: 2.65e-02, grad_scale: 8.0 2023-04-27 19:02:45,607 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16357.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:03:06,821 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16371.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:03:12,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9796, 3.7554, 3.6672, 2.9677, 3.5717, 3.7187, 3.9358, 1.9997], device='cuda:7'), covar=tensor([0.0576, 0.0025, 0.0055, 0.0262, 0.0060, 0.0092, 0.0027, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0050, 0.0058, 0.0110, 0.0051, 0.0058, 0.0059, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:03:51,369 INFO [train.py:904] (7/8) Epoch 2, batch 6250, loss[loss=0.3485, simple_loss=0.3875, pruned_loss=0.1547, over 11566.00 frames. ], tot_loss[loss=0.2984, simple_loss=0.3628, pruned_loss=0.1169, over 3122797.54 frames. ], batch size: 246, lr: 2.65e-02, grad_scale: 8.0 2023-04-27 19:04:18,709 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.383e+02 4.905e+02 5.910e+02 7.685e+02 1.899e+03, threshold=1.182e+03, percent-clipped=6.0 2023-04-27 19:04:37,399 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16432.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:04:51,251 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16441.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:05:05,100 INFO [train.py:904] (7/8) Epoch 2, batch 6300, loss[loss=0.3393, simple_loss=0.3905, pruned_loss=0.144, over 17098.00 frames. ], tot_loss[loss=0.2973, simple_loss=0.3624, pruned_loss=0.1161, over 3122847.72 frames. ], batch size: 49, lr: 2.64e-02, grad_scale: 8.0 2023-04-27 19:05:17,007 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5808, 1.3844, 1.9338, 2.3608, 2.3678, 2.5117, 1.4121, 2.5033], device='cuda:7'), covar=tensor([0.0041, 0.0260, 0.0133, 0.0087, 0.0043, 0.0064, 0.0185, 0.0030], device='cuda:7'), in_proj_covar=tensor([0.0068, 0.0108, 0.0093, 0.0079, 0.0059, 0.0057, 0.0092, 0.0053], device='cuda:7'), out_proj_covar=tensor([1.1886e-04, 1.9012e-04, 1.6968e-04, 1.4645e-04, 1.0062e-04, 1.0387e-04, 1.5775e-04, 9.2858e-05], device='cuda:7') 2023-04-27 19:06:22,090 INFO [train.py:904] (7/8) Epoch 2, batch 6350, loss[loss=0.3616, simple_loss=0.3891, pruned_loss=0.1671, over 11226.00 frames. ], tot_loss[loss=0.3009, simple_loss=0.3645, pruned_loss=0.1187, over 3106153.95 frames. ], batch size: 248, lr: 2.64e-02, grad_scale: 8.0 2023-04-27 19:06:29,789 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16505.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:06:36,048 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-27 19:06:48,139 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4090, 4.2096, 4.4086, 3.4101, 4.2237, 4.2316, 4.5815, 2.3492], device='cuda:7'), covar=tensor([0.0479, 0.0022, 0.0041, 0.0223, 0.0035, 0.0082, 0.0019, 0.0426], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0052, 0.0058, 0.0110, 0.0052, 0.0059, 0.0060, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:06:52,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.677e+02 5.217e+02 6.547e+02 8.022e+02 1.954e+03, threshold=1.309e+03, percent-clipped=7.0 2023-04-27 19:07:21,323 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16539.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:07:39,358 INFO [train.py:904] (7/8) Epoch 2, batch 6400, loss[loss=0.2846, simple_loss=0.3545, pruned_loss=0.1074, over 16832.00 frames. ], tot_loss[loss=0.3033, simple_loss=0.3653, pruned_loss=0.1207, over 3084974.58 frames. ], batch size: 116, lr: 2.64e-02, grad_scale: 8.0 2023-04-27 19:08:34,560 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16587.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:08:55,791 INFO [train.py:904] (7/8) Epoch 2, batch 6450, loss[loss=0.2764, simple_loss=0.3471, pruned_loss=0.1028, over 16377.00 frames. ], tot_loss[loss=0.3013, simple_loss=0.3645, pruned_loss=0.1191, over 3072123.20 frames. ], batch size: 146, lr: 2.63e-02, grad_scale: 8.0 2023-04-27 19:09:25,707 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.052e+02 4.859e+02 5.740e+02 6.998e+02 1.216e+03, threshold=1.148e+03, percent-clipped=0.0 2023-04-27 19:09:35,164 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4219, 2.2281, 1.6755, 1.6801, 2.8344, 2.5949, 3.4646, 3.1015], device='cuda:7'), covar=tensor([0.0009, 0.0139, 0.0174, 0.0190, 0.0072, 0.0119, 0.0025, 0.0056], device='cuda:7'), in_proj_covar=tensor([0.0040, 0.0094, 0.0094, 0.0097, 0.0085, 0.0095, 0.0052, 0.0069], device='cuda:7'), out_proj_covar=tensor([5.8591e-05, 1.4639e-04, 1.4295e-04, 1.5227e-04, 1.3998e-04, 1.5280e-04, 8.3458e-05, 1.1477e-04], device='cuda:7') 2023-04-27 19:10:13,794 INFO [train.py:904] (7/8) Epoch 2, batch 6500, loss[loss=0.2952, simple_loss=0.3614, pruned_loss=0.1145, over 16513.00 frames. ], tot_loss[loss=0.2985, simple_loss=0.3615, pruned_loss=0.1177, over 3073815.98 frames. ], batch size: 35, lr: 2.63e-02, grad_scale: 8.0 2023-04-27 19:10:14,554 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2566, 2.5834, 2.1188, 3.6337, 1.9013, 3.4547, 2.2154, 2.2234], device='cuda:7'), covar=tensor([0.0300, 0.0435, 0.0372, 0.0203, 0.1455, 0.0204, 0.0717, 0.1058], device='cuda:7'), in_proj_covar=tensor([0.0206, 0.0180, 0.0151, 0.0211, 0.0261, 0.0167, 0.0184, 0.0240], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:10:24,419 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16657.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:11:32,130 INFO [train.py:904] (7/8) Epoch 2, batch 6550, loss[loss=0.2944, simple_loss=0.3846, pruned_loss=0.1021, over 16911.00 frames. ], tot_loss[loss=0.3017, simple_loss=0.365, pruned_loss=0.1192, over 3073437.44 frames. ], batch size: 90, lr: 2.63e-02, grad_scale: 8.0 2023-04-27 19:11:37,739 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16705.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:11:59,596 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.469e+02 4.692e+02 5.928e+02 7.455e+02 1.589e+03, threshold=1.186e+03, percent-clipped=4.0 2023-04-27 19:12:11,257 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16727.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:12:33,078 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16741.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:12:33,388 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-27 19:12:47,624 INFO [train.py:904] (7/8) Epoch 2, batch 6600, loss[loss=0.2961, simple_loss=0.3674, pruned_loss=0.1125, over 16895.00 frames. ], tot_loss[loss=0.3045, simple_loss=0.3678, pruned_loss=0.1206, over 3071119.89 frames. ], batch size: 116, lr: 2.62e-02, grad_scale: 8.0 2023-04-27 19:13:29,076 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-04-27 19:13:36,578 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1103, 4.1599, 4.1500, 4.1205, 4.1665, 4.5991, 4.3773, 4.0642], device='cuda:7'), covar=tensor([0.1124, 0.0952, 0.0928, 0.1268, 0.1729, 0.0639, 0.0774, 0.1521], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0271, 0.0250, 0.0237, 0.0306, 0.0266, 0.0208, 0.0315], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:13:47,419 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16789.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:14:06,685 INFO [train.py:904] (7/8) Epoch 2, batch 6650, loss[loss=0.3033, simple_loss=0.3696, pruned_loss=0.1185, over 15487.00 frames. ], tot_loss[loss=0.3062, simple_loss=0.3687, pruned_loss=0.1218, over 3078652.68 frames. ], batch size: 191, lr: 2.62e-02, grad_scale: 8.0 2023-04-27 19:14:13,271 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16805.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:14:35,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.480e+02 5.041e+02 6.003e+02 7.722e+02 1.259e+03, threshold=1.201e+03, percent-clipped=2.0 2023-04-27 19:15:23,524 INFO [train.py:904] (7/8) Epoch 2, batch 6700, loss[loss=0.2639, simple_loss=0.3363, pruned_loss=0.09574, over 16993.00 frames. ], tot_loss[loss=0.3064, simple_loss=0.3682, pruned_loss=0.1223, over 3065092.12 frames. ], batch size: 55, lr: 2.61e-02, grad_scale: 8.0 2023-04-27 19:15:26,815 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=16853.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:15:30,516 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2023-04-27 19:15:50,680 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16868.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:16:03,718 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-27 19:16:38,379 INFO [train.py:904] (7/8) Epoch 2, batch 6750, loss[loss=0.2603, simple_loss=0.3317, pruned_loss=0.09449, over 16738.00 frames. ], tot_loss[loss=0.3057, simple_loss=0.3672, pruned_loss=0.1222, over 3083128.00 frames. ], batch size: 124, lr: 2.61e-02, grad_scale: 8.0 2023-04-27 19:16:47,187 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6754, 4.5627, 4.3993, 3.6456, 4.4530, 1.6541, 4.2139, 4.3987], device='cuda:7'), covar=tensor([0.0046, 0.0053, 0.0056, 0.0290, 0.0047, 0.1358, 0.0057, 0.0088], device='cuda:7'), in_proj_covar=tensor([0.0064, 0.0053, 0.0078, 0.0098, 0.0058, 0.0112, 0.0071, 0.0078], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:16:47,212 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16906.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:17:07,559 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.622e+02 5.251e+02 6.489e+02 8.459e+02 1.507e+03, threshold=1.298e+03, percent-clipped=4.0 2023-04-27 19:17:22,305 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16929.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:17:32,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7629, 1.3843, 1.5868, 1.7522, 1.6006, 1.7884, 1.4396, 1.8546], device='cuda:7'), covar=tensor([0.0048, 0.0145, 0.0073, 0.0076, 0.0033, 0.0044, 0.0103, 0.0030], device='cuda:7'), in_proj_covar=tensor([0.0068, 0.0107, 0.0092, 0.0081, 0.0058, 0.0057, 0.0091, 0.0052], device='cuda:7'), out_proj_covar=tensor([1.1874e-04, 1.8704e-04, 1.6717e-04, 1.4756e-04, 1.0031e-04, 1.0273e-04, 1.5504e-04, 9.0011e-05], device='cuda:7') 2023-04-27 19:17:53,314 INFO [train.py:904] (7/8) Epoch 2, batch 6800, loss[loss=0.333, simple_loss=0.3754, pruned_loss=0.1453, over 11327.00 frames. ], tot_loss[loss=0.3047, simple_loss=0.3659, pruned_loss=0.1217, over 3066670.85 frames. ], batch size: 246, lr: 2.61e-02, grad_scale: 8.0 2023-04-27 19:18:00,683 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-27 19:18:18,725 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16967.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:19:10,226 INFO [train.py:904] (7/8) Epoch 2, batch 6850, loss[loss=0.2874, simple_loss=0.3811, pruned_loss=0.09691, over 16874.00 frames. ], tot_loss[loss=0.3081, simple_loss=0.3689, pruned_loss=0.1236, over 3054410.56 frames. ], batch size: 96, lr: 2.60e-02, grad_scale: 8.0 2023-04-27 19:19:38,154 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-27 19:19:38,487 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.045e+02 4.695e+02 5.738e+02 7.185e+02 1.728e+03, threshold=1.148e+03, percent-clipped=4.0 2023-04-27 19:19:49,820 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17027.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:19:52,742 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-04-27 19:20:24,246 INFO [train.py:904] (7/8) Epoch 2, batch 6900, loss[loss=0.2604, simple_loss=0.3345, pruned_loss=0.09311, over 16957.00 frames. ], tot_loss[loss=0.3092, simple_loss=0.3714, pruned_loss=0.1235, over 3062555.11 frames. ], batch size: 41, lr: 2.60e-02, grad_scale: 8.0 2023-04-27 19:20:51,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0154, 4.2129, 4.0999, 4.2431, 4.1712, 4.6145, 4.4460, 4.1651], device='cuda:7'), covar=tensor([0.1212, 0.1172, 0.1086, 0.1247, 0.1899, 0.0783, 0.0754, 0.1803], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0267, 0.0245, 0.0234, 0.0304, 0.0269, 0.0205, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:21:01,235 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17075.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:21:40,742 INFO [train.py:904] (7/8) Epoch 2, batch 6950, loss[loss=0.2787, simple_loss=0.3451, pruned_loss=0.1061, over 16582.00 frames. ], tot_loss[loss=0.3146, simple_loss=0.3748, pruned_loss=0.1272, over 3043040.89 frames. ], batch size: 62, lr: 2.60e-02, grad_scale: 8.0 2023-04-27 19:22:09,953 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.451e+02 5.662e+02 6.855e+02 8.747e+02 1.724e+03, threshold=1.371e+03, percent-clipped=6.0 2023-04-27 19:22:31,471 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7413, 3.6800, 1.5406, 3.7694, 2.4581, 3.7371, 1.6865, 2.5263], device='cuda:7'), covar=tensor([0.0039, 0.0160, 0.1758, 0.0035, 0.0804, 0.0381, 0.1564, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0074, 0.0108, 0.0172, 0.0074, 0.0163, 0.0137, 0.0179, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 19:22:54,975 INFO [train.py:904] (7/8) Epoch 2, batch 7000, loss[loss=0.2892, simple_loss=0.3767, pruned_loss=0.1009, over 16472.00 frames. ], tot_loss[loss=0.3134, simple_loss=0.3751, pruned_loss=0.1258, over 3046518.61 frames. ], batch size: 68, lr: 2.59e-02, grad_scale: 8.0 2023-04-27 19:23:00,192 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2821, 4.5377, 4.2592, 4.3799, 3.9197, 3.9371, 4.1507, 4.5465], device='cuda:7'), covar=tensor([0.0473, 0.0663, 0.0739, 0.0344, 0.0490, 0.0739, 0.0476, 0.0582], device='cuda:7'), in_proj_covar=tensor([0.0224, 0.0309, 0.0284, 0.0196, 0.0208, 0.0193, 0.0256, 0.0215], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:23:53,772 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17189.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:24:10,880 INFO [train.py:904] (7/8) Epoch 2, batch 7050, loss[loss=0.3001, simple_loss=0.3706, pruned_loss=0.1148, over 16610.00 frames. ], tot_loss[loss=0.3137, simple_loss=0.3758, pruned_loss=0.1258, over 3043695.78 frames. ], batch size: 57, lr: 2.59e-02, grad_scale: 8.0 2023-04-27 19:24:25,432 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6450, 2.9405, 2.9758, 2.2043, 2.9468, 2.9704, 3.0848, 1.6190], device='cuda:7'), covar=tensor([0.0599, 0.0049, 0.0064, 0.0318, 0.0063, 0.0094, 0.0043, 0.0505], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0055, 0.0058, 0.0112, 0.0056, 0.0060, 0.0060, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:24:37,726 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.062e+02 5.191e+02 6.345e+02 7.855e+02 1.482e+03, threshold=1.269e+03, percent-clipped=3.0 2023-04-27 19:24:44,021 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17224.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:24:56,681 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9170, 3.8954, 1.3373, 3.9010, 2.4647, 3.9666, 1.9225, 2.7635], device='cuda:7'), covar=tensor([0.0027, 0.0085, 0.1265, 0.0042, 0.0501, 0.0129, 0.1006, 0.0421], device='cuda:7'), in_proj_covar=tensor([0.0073, 0.0105, 0.0168, 0.0072, 0.0161, 0.0136, 0.0176, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 19:25:23,022 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17250.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:25:23,670 INFO [train.py:904] (7/8) Epoch 2, batch 7100, loss[loss=0.275, simple_loss=0.3414, pruned_loss=0.1043, over 16651.00 frames. ], tot_loss[loss=0.3098, simple_loss=0.3723, pruned_loss=0.1236, over 3067202.13 frames. ], batch size: 57, lr: 2.59e-02, grad_scale: 8.0 2023-04-27 19:25:41,141 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17262.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:26:38,672 INFO [train.py:904] (7/8) Epoch 2, batch 7150, loss[loss=0.3176, simple_loss=0.3813, pruned_loss=0.1269, over 16627.00 frames. ], tot_loss[loss=0.3079, simple_loss=0.3703, pruned_loss=0.1228, over 3065263.83 frames. ], batch size: 62, lr: 2.58e-02, grad_scale: 8.0 2023-04-27 19:26:42,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-27 19:26:44,907 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5658, 3.4119, 3.6209, 3.5176, 3.6643, 4.0375, 3.8457, 3.5798], device='cuda:7'), covar=tensor([0.1693, 0.1596, 0.1162, 0.2096, 0.2014, 0.1101, 0.1019, 0.2199], device='cuda:7'), in_proj_covar=tensor([0.0198, 0.0270, 0.0249, 0.0238, 0.0306, 0.0269, 0.0211, 0.0319], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:27:07,390 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.740e+02 5.248e+02 6.301e+02 7.700e+02 1.780e+03, threshold=1.260e+03, percent-clipped=1.0 2023-04-27 19:27:52,979 INFO [train.py:904] (7/8) Epoch 2, batch 7200, loss[loss=0.2653, simple_loss=0.3487, pruned_loss=0.09093, over 16770.00 frames. ], tot_loss[loss=0.3031, simple_loss=0.3668, pruned_loss=0.1197, over 3055822.93 frames. ], batch size: 89, lr: 2.58e-02, grad_scale: 8.0 2023-04-27 19:27:59,334 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4476, 2.6394, 2.3955, 3.9152, 2.0725, 3.7003, 2.3771, 2.3315], device='cuda:7'), covar=tensor([0.0313, 0.0482, 0.0350, 0.0186, 0.1444, 0.0200, 0.0699, 0.1156], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0187, 0.0159, 0.0217, 0.0268, 0.0173, 0.0190, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:28:10,370 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17362.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:28:18,175 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6633, 1.9827, 1.4714, 1.7594, 2.3918, 2.1536, 2.5131, 2.4608], device='cuda:7'), covar=tensor([0.0013, 0.0122, 0.0166, 0.0153, 0.0063, 0.0115, 0.0030, 0.0053], device='cuda:7'), in_proj_covar=tensor([0.0045, 0.0099, 0.0100, 0.0102, 0.0088, 0.0102, 0.0053, 0.0071], device='cuda:7'), out_proj_covar=tensor([6.4223e-05, 1.5234e-04, 1.5120e-04, 1.5869e-04, 1.4351e-04, 1.6309e-04, 8.3308e-05, 1.1602e-04], device='cuda:7') 2023-04-27 19:28:45,093 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17383.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:29:13,603 INFO [train.py:904] (7/8) Epoch 2, batch 7250, loss[loss=0.2497, simple_loss=0.3195, pruned_loss=0.08995, over 16826.00 frames. ], tot_loss[loss=0.2985, simple_loss=0.363, pruned_loss=0.1171, over 3066987.42 frames. ], batch size: 116, lr: 2.58e-02, grad_scale: 8.0 2023-04-27 19:29:15,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7735, 3.5908, 3.1942, 1.6892, 2.7156, 1.9898, 3.3369, 3.5580], device='cuda:7'), covar=tensor([0.0301, 0.0405, 0.0498, 0.1823, 0.0796, 0.1119, 0.0659, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0110, 0.0154, 0.0156, 0.0148, 0.0140, 0.0153, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 19:29:31,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0078, 3.7733, 3.9261, 2.5593, 3.9301, 3.8201, 3.9837, 1.7733], device='cuda:7'), covar=tensor([0.0515, 0.0029, 0.0035, 0.0289, 0.0033, 0.0082, 0.0020, 0.0470], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0054, 0.0057, 0.0110, 0.0053, 0.0058, 0.0058, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:29:42,543 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.618e+02 4.777e+02 5.726e+02 7.027e+02 1.873e+03, threshold=1.145e+03, percent-clipped=4.0 2023-04-27 19:29:47,284 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17423.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:30:05,856 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9017, 3.6249, 3.7049, 3.8161, 3.3312, 3.7739, 3.4918, 3.5860], device='cuda:7'), covar=tensor([0.0298, 0.0194, 0.0201, 0.0139, 0.0643, 0.0189, 0.0553, 0.0252], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0092, 0.0147, 0.0117, 0.0171, 0.0125, 0.0105, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:30:19,153 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17444.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:30:29,425 INFO [train.py:904] (7/8) Epoch 2, batch 7300, loss[loss=0.2873, simple_loss=0.3595, pruned_loss=0.1075, over 16714.00 frames. ], tot_loss[loss=0.2967, simple_loss=0.3615, pruned_loss=0.116, over 3072484.81 frames. ], batch size: 134, lr: 2.57e-02, grad_scale: 8.0 2023-04-27 19:30:48,651 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 19:31:06,627 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2023-04-27 19:31:46,484 INFO [train.py:904] (7/8) Epoch 2, batch 7350, loss[loss=0.2641, simple_loss=0.347, pruned_loss=0.09056, over 16777.00 frames. ], tot_loss[loss=0.2952, simple_loss=0.3603, pruned_loss=0.1151, over 3071457.97 frames. ], batch size: 116, lr: 2.57e-02, grad_scale: 8.0 2023-04-27 19:32:16,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.647e+02 4.932e+02 5.634e+02 7.018e+02 1.459e+03, threshold=1.127e+03, percent-clipped=2.0 2023-04-27 19:32:22,558 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17524.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:32:57,427 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17545.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:33:06,156 INFO [train.py:904] (7/8) Epoch 2, batch 7400, loss[loss=0.2873, simple_loss=0.3608, pruned_loss=0.1069, over 16705.00 frames. ], tot_loss[loss=0.296, simple_loss=0.361, pruned_loss=0.1155, over 3079809.86 frames. ], batch size: 76, lr: 2.57e-02, grad_scale: 8.0 2023-04-27 19:33:25,285 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17562.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:33:41,414 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17572.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:33:50,525 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-27 19:34:27,197 INFO [train.py:904] (7/8) Epoch 2, batch 7450, loss[loss=0.2849, simple_loss=0.3611, pruned_loss=0.1044, over 16844.00 frames. ], tot_loss[loss=0.2989, simple_loss=0.363, pruned_loss=0.1173, over 3084567.43 frames. ], batch size: 102, lr: 2.56e-02, grad_scale: 8.0 2023-04-27 19:34:43,855 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17610.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:34:59,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.824e+02 5.406e+02 6.493e+02 7.703e+02 1.630e+03, threshold=1.299e+03, percent-clipped=5.0 2023-04-27 19:35:18,728 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-27 19:35:49,112 INFO [train.py:904] (7/8) Epoch 2, batch 7500, loss[loss=0.3626, simple_loss=0.3994, pruned_loss=0.1629, over 11406.00 frames. ], tot_loss[loss=0.2997, simple_loss=0.3641, pruned_loss=0.1176, over 3083340.02 frames. ], batch size: 248, lr: 2.56e-02, grad_scale: 8.0 2023-04-27 19:35:53,365 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9087, 3.2796, 3.2811, 1.4815, 3.4342, 3.4589, 2.8876, 2.8008], device='cuda:7'), covar=tensor([0.0864, 0.0114, 0.0150, 0.1461, 0.0087, 0.0060, 0.0298, 0.0322], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0083, 0.0081, 0.0150, 0.0074, 0.0070, 0.0104, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:36:17,006 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-27 19:36:53,167 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5601, 3.4127, 3.0247, 2.6346, 2.8338, 2.0952, 3.7668, 4.2544], device='cuda:7'), covar=tensor([0.1729, 0.0671, 0.0980, 0.0723, 0.1718, 0.1176, 0.0316, 0.0158], device='cuda:7'), in_proj_covar=tensor([0.0254, 0.0223, 0.0238, 0.0182, 0.0273, 0.0181, 0.0198, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:37:05,101 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17700.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:37:05,684 INFO [train.py:904] (7/8) Epoch 2, batch 7550, loss[loss=0.3025, simple_loss=0.363, pruned_loss=0.121, over 16899.00 frames. ], tot_loss[loss=0.2987, simple_loss=0.3629, pruned_loss=0.1172, over 3094372.78 frames. ], batch size: 109, lr: 2.56e-02, grad_scale: 8.0 2023-04-27 19:37:32,665 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17718.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:37:34,651 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.677e+02 5.137e+02 6.469e+02 8.298e+02 1.927e+03, threshold=1.294e+03, percent-clipped=3.0 2023-04-27 19:37:43,795 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7994, 4.0661, 3.8126, 3.9246, 3.4093, 3.6412, 3.7468, 3.9557], device='cuda:7'), covar=tensor([0.0442, 0.0618, 0.0697, 0.0328, 0.0515, 0.0735, 0.0458, 0.0681], device='cuda:7'), in_proj_covar=tensor([0.0220, 0.0308, 0.0278, 0.0194, 0.0203, 0.0191, 0.0243, 0.0206], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:37:51,096 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17730.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:38:05,276 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17739.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 19:38:20,262 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-27 19:38:22,226 INFO [train.py:904] (7/8) Epoch 2, batch 7600, loss[loss=0.271, simple_loss=0.3385, pruned_loss=0.1017, over 16299.00 frames. ], tot_loss[loss=0.2972, simple_loss=0.361, pruned_loss=0.1167, over 3094511.94 frames. ], batch size: 35, lr: 2.55e-02, grad_scale: 8.0 2023-04-27 19:38:36,593 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7023, 2.4809, 2.5905, 1.8391, 2.5868, 2.4054, 2.6145, 1.6477], device='cuda:7'), covar=tensor([0.0398, 0.0049, 0.0073, 0.0273, 0.0063, 0.0084, 0.0056, 0.0412], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0055, 0.0061, 0.0114, 0.0055, 0.0060, 0.0060, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:38:37,874 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17761.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:39:18,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7849, 4.0498, 3.7558, 3.9409, 3.5329, 3.6514, 3.8055, 3.9237], device='cuda:7'), covar=tensor([0.0521, 0.0558, 0.0722, 0.0313, 0.0499, 0.0820, 0.0409, 0.0724], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0302, 0.0277, 0.0191, 0.0201, 0.0191, 0.0244, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:39:25,097 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17791.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:39:39,715 INFO [train.py:904] (7/8) Epoch 2, batch 7650, loss[loss=0.3675, simple_loss=0.3981, pruned_loss=0.1685, over 11479.00 frames. ], tot_loss[loss=0.2992, simple_loss=0.3624, pruned_loss=0.1179, over 3097397.57 frames. ], batch size: 246, lr: 2.55e-02, grad_scale: 8.0 2023-04-27 19:40:10,729 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.673e+02 5.378e+02 6.384e+02 7.645e+02 1.933e+03, threshold=1.277e+03, percent-clipped=1.0 2023-04-27 19:40:50,329 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17845.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:40:58,871 INFO [train.py:904] (7/8) Epoch 2, batch 7700, loss[loss=0.3164, simple_loss=0.3729, pruned_loss=0.1299, over 16852.00 frames. ], tot_loss[loss=0.3022, simple_loss=0.3638, pruned_loss=0.1203, over 3067100.18 frames. ], batch size: 116, lr: 2.55e-02, grad_scale: 8.0 2023-04-27 19:40:59,426 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9171, 3.7763, 3.7350, 3.8479, 3.3630, 3.8129, 3.5427, 3.5603], device='cuda:7'), covar=tensor([0.0274, 0.0213, 0.0180, 0.0110, 0.0664, 0.0223, 0.0505, 0.0235], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0096, 0.0148, 0.0119, 0.0178, 0.0128, 0.0108, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:42:03,250 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2376, 4.8471, 4.9004, 5.0758, 4.4523, 5.0882, 4.9480, 4.7262], device='cuda:7'), covar=tensor([0.0209, 0.0224, 0.0143, 0.0087, 0.0639, 0.0147, 0.0110, 0.0187], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0096, 0.0146, 0.0118, 0.0174, 0.0125, 0.0106, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:42:04,330 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=17893.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:42:16,833 INFO [train.py:904] (7/8) Epoch 2, batch 7750, loss[loss=0.2895, simple_loss=0.362, pruned_loss=0.1085, over 16709.00 frames. ], tot_loss[loss=0.3022, simple_loss=0.3645, pruned_loss=0.12, over 3074238.33 frames. ], batch size: 134, lr: 2.54e-02, grad_scale: 16.0 2023-04-27 19:42:46,218 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-27 19:42:46,608 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.308e+02 4.989e+02 6.319e+02 7.239e+02 1.280e+03, threshold=1.264e+03, percent-clipped=2.0 2023-04-27 19:43:31,915 INFO [train.py:904] (7/8) Epoch 2, batch 7800, loss[loss=0.284, simple_loss=0.3593, pruned_loss=0.1044, over 16623.00 frames. ], tot_loss[loss=0.3038, simple_loss=0.3657, pruned_loss=0.1209, over 3079601.10 frames. ], batch size: 76, lr: 2.54e-02, grad_scale: 16.0 2023-04-27 19:43:47,109 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17960.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:44:33,685 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-27 19:44:47,349 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-27 19:44:52,431 INFO [train.py:904] (7/8) Epoch 2, batch 7850, loss[loss=0.3123, simple_loss=0.3734, pruned_loss=0.1256, over 15224.00 frames. ], tot_loss[loss=0.3038, simple_loss=0.3662, pruned_loss=0.1207, over 3073505.87 frames. ], batch size: 190, lr: 2.54e-02, grad_scale: 16.0 2023-04-27 19:45:18,630 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18018.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:45:21,253 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.805e+02 5.114e+02 6.121e+02 7.560e+02 1.271e+03, threshold=1.224e+03, percent-clipped=1.0 2023-04-27 19:45:21,833 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0693, 3.9062, 4.0083, 1.4900, 4.1627, 4.1065, 3.0967, 3.2414], device='cuda:7'), covar=tensor([0.0879, 0.0099, 0.0095, 0.1414, 0.0050, 0.0039, 0.0252, 0.0292], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0085, 0.0082, 0.0145, 0.0073, 0.0069, 0.0105, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:45:22,942 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18021.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:45:33,581 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 19:45:48,913 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18039.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:46:05,621 INFO [train.py:904] (7/8) Epoch 2, batch 7900, loss[loss=0.3138, simple_loss=0.3758, pruned_loss=0.1259, over 16777.00 frames. ], tot_loss[loss=0.3003, simple_loss=0.3636, pruned_loss=0.1185, over 3098396.23 frames. ], batch size: 124, lr: 2.53e-02, grad_scale: 8.0 2023-04-27 19:46:14,080 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18056.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:46:29,413 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18066.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:47:03,305 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18086.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:47:05,057 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18087.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 19:47:25,404 INFO [train.py:904] (7/8) Epoch 2, batch 7950, loss[loss=0.2922, simple_loss=0.3513, pruned_loss=0.1165, over 16730.00 frames. ], tot_loss[loss=0.301, simple_loss=0.3639, pruned_loss=0.1191, over 3095752.47 frames. ], batch size: 89, lr: 2.53e-02, grad_scale: 8.0 2023-04-27 19:47:56,300 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.877e+02 5.212e+02 6.244e+02 7.914e+02 2.379e+03, threshold=1.249e+03, percent-clipped=3.0 2023-04-27 19:48:07,656 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 19:48:25,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1506, 3.0663, 1.4501, 3.1424, 2.3945, 3.2138, 1.7470, 2.4619], device='cuda:7'), covar=tensor([0.0055, 0.0198, 0.1443, 0.0050, 0.0560, 0.0290, 0.1234, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0074, 0.0113, 0.0174, 0.0075, 0.0159, 0.0140, 0.0179, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 19:48:42,381 INFO [train.py:904] (7/8) Epoch 2, batch 8000, loss[loss=0.3357, simple_loss=0.3869, pruned_loss=0.1422, over 16764.00 frames. ], tot_loss[loss=0.3016, simple_loss=0.3642, pruned_loss=0.1195, over 3089149.74 frames. ], batch size: 39, lr: 2.53e-02, grad_scale: 8.0 2023-04-27 19:49:03,433 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7564, 5.0396, 4.7032, 4.8276, 4.3227, 4.3128, 4.5262, 5.0841], device='cuda:7'), covar=tensor([0.0423, 0.0511, 0.0874, 0.0310, 0.0475, 0.0541, 0.0475, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0222, 0.0310, 0.0281, 0.0197, 0.0208, 0.0199, 0.0254, 0.0213], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:49:16,819 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18174.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:49:18,139 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18175.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:49:28,726 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7313, 3.6976, 3.0860, 2.8843, 2.7873, 2.2329, 3.8908, 4.2878], device='cuda:7'), covar=tensor([0.1813, 0.0554, 0.0964, 0.0673, 0.1925, 0.1190, 0.0292, 0.0236], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0226, 0.0241, 0.0183, 0.0286, 0.0186, 0.0201, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 19:49:34,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8306, 3.7440, 2.4924, 4.8515, 4.6519, 4.0361, 2.2403, 3.0060], device='cuda:7'), covar=tensor([0.1589, 0.0362, 0.1329, 0.0073, 0.0148, 0.0293, 0.1154, 0.0745], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0122, 0.0167, 0.0070, 0.0112, 0.0121, 0.0154, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 19:49:35,916 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6676, 2.6130, 2.5060, 2.0524, 2.4812, 2.4784, 2.6877, 1.8144], device='cuda:7'), covar=tensor([0.0384, 0.0045, 0.0066, 0.0215, 0.0057, 0.0069, 0.0040, 0.0348], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0054, 0.0058, 0.0109, 0.0054, 0.0059, 0.0057, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:49:40,577 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1613, 4.2164, 4.0612, 3.1155, 4.0819, 4.0129, 4.4107, 2.0833], device='cuda:7'), covar=tensor([0.0454, 0.0023, 0.0042, 0.0211, 0.0029, 0.0083, 0.0021, 0.0444], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0054, 0.0059, 0.0109, 0.0054, 0.0059, 0.0057, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 19:49:56,237 INFO [train.py:904] (7/8) Epoch 2, batch 8050, loss[loss=0.2645, simple_loss=0.3455, pruned_loss=0.09175, over 16685.00 frames. ], tot_loss[loss=0.3003, simple_loss=0.3634, pruned_loss=0.1186, over 3087407.77 frames. ], batch size: 62, lr: 2.52e-02, grad_scale: 8.0 2023-04-27 19:50:10,908 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.63 vs. limit=5.0 2023-04-27 19:50:24,940 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.258e+02 4.729e+02 5.978e+02 7.201e+02 1.227e+03, threshold=1.196e+03, percent-clipped=0.0 2023-04-27 19:50:47,464 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18235.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:50:48,753 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18236.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:51:11,137 INFO [train.py:904] (7/8) Epoch 2, batch 8100, loss[loss=0.3358, simple_loss=0.3794, pruned_loss=0.1461, over 11451.00 frames. ], tot_loss[loss=0.2986, simple_loss=0.363, pruned_loss=0.1171, over 3102643.88 frames. ], batch size: 246, lr: 2.52e-02, grad_scale: 8.0 2023-04-27 19:52:29,028 INFO [train.py:904] (7/8) Epoch 2, batch 8150, loss[loss=0.2361, simple_loss=0.3173, pruned_loss=0.07745, over 16872.00 frames. ], tot_loss[loss=0.2971, simple_loss=0.361, pruned_loss=0.1166, over 3096888.84 frames. ], batch size: 102, lr: 2.52e-02, grad_scale: 8.0 2023-04-27 19:52:35,799 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4859, 3.8094, 4.0899, 4.0704, 4.0731, 3.6987, 3.3175, 3.9499], device='cuda:7'), covar=tensor([0.0488, 0.0387, 0.0448, 0.0562, 0.0692, 0.0503, 0.1686, 0.0371], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0153, 0.0166, 0.0163, 0.0201, 0.0169, 0.0259, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 19:52:53,558 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18316.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:53:00,201 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.20 vs. limit=5.0 2023-04-27 19:53:00,711 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.225e+02 5.223e+02 6.693e+02 9.034e+02 1.543e+03, threshold=1.339e+03, percent-clipped=7.0 2023-04-27 19:53:48,026 INFO [train.py:904] (7/8) Epoch 2, batch 8200, loss[loss=0.2835, simple_loss=0.3501, pruned_loss=0.1085, over 16653.00 frames. ], tot_loss[loss=0.2935, simple_loss=0.3577, pruned_loss=0.1147, over 3122204.01 frames. ], batch size: 62, lr: 2.51e-02, grad_scale: 4.0 2023-04-27 19:53:57,201 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18356.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:54:46,796 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18386.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:55:10,277 INFO [train.py:904] (7/8) Epoch 2, batch 8250, loss[loss=0.2371, simple_loss=0.325, pruned_loss=0.07463, over 16450.00 frames. ], tot_loss[loss=0.2916, simple_loss=0.3576, pruned_loss=0.1128, over 3107561.07 frames. ], batch size: 75, lr: 2.51e-02, grad_scale: 4.0 2023-04-27 19:55:15,797 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18404.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:55:44,644 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.019e+02 4.548e+02 5.368e+02 6.842e+02 2.128e+03, threshold=1.074e+03, percent-clipped=3.0 2023-04-27 19:55:54,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5085, 3.5878, 3.9405, 3.9523, 3.9975, 3.5480, 3.6729, 3.7759], device='cuda:7'), covar=tensor([0.0260, 0.0370, 0.0300, 0.0314, 0.0336, 0.0316, 0.0680, 0.0271], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0150, 0.0163, 0.0161, 0.0195, 0.0168, 0.0253, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 19:56:05,978 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18434.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:56:33,573 INFO [train.py:904] (7/8) Epoch 2, batch 8300, loss[loss=0.2573, simple_loss=0.3138, pruned_loss=0.1005, over 12446.00 frames. ], tot_loss[loss=0.2847, simple_loss=0.3528, pruned_loss=0.1083, over 3087045.15 frames. ], batch size: 247, lr: 2.51e-02, grad_scale: 4.0 2023-04-27 19:57:37,590 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18490.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:57:54,226 INFO [train.py:904] (7/8) Epoch 2, batch 8350, loss[loss=0.2921, simple_loss=0.3468, pruned_loss=0.1187, over 12283.00 frames. ], tot_loss[loss=0.2789, simple_loss=0.3496, pruned_loss=0.1041, over 3073726.37 frames. ], batch size: 246, lr: 2.50e-02, grad_scale: 4.0 2023-04-27 19:58:06,781 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18508.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 19:58:28,819 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.204e+02 3.865e+02 4.910e+02 6.166e+02 1.131e+03, threshold=9.820e+02, percent-clipped=1.0 2023-04-27 19:58:42,172 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18530.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:58:44,086 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18531.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:59:15,569 INFO [train.py:904] (7/8) Epoch 2, batch 8400, loss[loss=0.2871, simple_loss=0.3437, pruned_loss=0.1152, over 12159.00 frames. ], tot_loss[loss=0.2742, simple_loss=0.3457, pruned_loss=0.1013, over 3044689.98 frames. ], batch size: 248, lr: 2.50e-02, grad_scale: 8.0 2023-04-27 19:59:16,246 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18551.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 19:59:45,596 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18569.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 20:00:35,066 INFO [train.py:904] (7/8) Epoch 2, batch 8450, loss[loss=0.2441, simple_loss=0.3233, pruned_loss=0.08239, over 16695.00 frames. ], tot_loss[loss=0.2706, simple_loss=0.3431, pruned_loss=0.09904, over 3046620.95 frames. ], batch size: 134, lr: 2.50e-02, grad_scale: 8.0 2023-04-27 20:01:00,364 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18616.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:01:09,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.749e+02 4.009e+02 5.012e+02 6.279e+02 1.629e+03, threshold=1.002e+03, percent-clipped=8.0 2023-04-27 20:01:54,987 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4006, 4.2727, 4.1089, 1.6960, 4.3405, 4.2673, 3.5191, 3.5011], device='cuda:7'), covar=tensor([0.0893, 0.0063, 0.0133, 0.1514, 0.0060, 0.0067, 0.0215, 0.0279], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0082, 0.0078, 0.0146, 0.0071, 0.0068, 0.0104, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:01:55,567 INFO [train.py:904] (7/8) Epoch 2, batch 8500, loss[loss=0.2376, simple_loss=0.3165, pruned_loss=0.07934, over 16469.00 frames. ], tot_loss[loss=0.2636, simple_loss=0.3375, pruned_loss=0.09484, over 3038187.37 frames. ], batch size: 68, lr: 2.49e-02, grad_scale: 8.0 2023-04-27 20:02:07,211 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7519, 3.5881, 2.5396, 4.4823, 4.4071, 4.2845, 1.8170, 3.3216], device='cuda:7'), covar=tensor([0.1640, 0.0392, 0.1244, 0.0093, 0.0166, 0.0254, 0.1408, 0.0542], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0123, 0.0167, 0.0071, 0.0111, 0.0121, 0.0155, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 20:02:17,872 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18664.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:02:34,590 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18674.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:02:34,706 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3694, 2.8572, 2.4808, 2.3983, 2.2340, 1.8882, 2.7995, 3.1513], device='cuda:7'), covar=tensor([0.1616, 0.0538, 0.0964, 0.0693, 0.1696, 0.1565, 0.0303, 0.0201], device='cuda:7'), in_proj_covar=tensor([0.0247, 0.0215, 0.0230, 0.0175, 0.0235, 0.0184, 0.0191, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:03:19,434 INFO [train.py:904] (7/8) Epoch 2, batch 8550, loss[loss=0.2783, simple_loss=0.3516, pruned_loss=0.1026, over 16751.00 frames. ], tot_loss[loss=0.2599, simple_loss=0.3342, pruned_loss=0.09279, over 3023839.94 frames. ], batch size: 124, lr: 2.49e-02, grad_scale: 4.0 2023-04-27 20:04:03,006 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.327e+02 4.014e+02 4.987e+02 6.688e+02 1.627e+03, threshold=9.973e+02, percent-clipped=3.0 2023-04-27 20:04:26,414 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18735.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:04:58,645 INFO [train.py:904] (7/8) Epoch 2, batch 8600, loss[loss=0.286, simple_loss=0.3556, pruned_loss=0.1082, over 16653.00 frames. ], tot_loss[loss=0.2599, simple_loss=0.3354, pruned_loss=0.09215, over 3042755.01 frames. ], batch size: 134, lr: 2.49e-02, grad_scale: 2.0 2023-04-27 20:06:17,747 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6579, 1.8303, 1.6411, 1.6454, 2.3142, 2.0468, 2.4167, 2.3737], device='cuda:7'), covar=tensor([0.0019, 0.0161, 0.0159, 0.0181, 0.0089, 0.0139, 0.0035, 0.0066], device='cuda:7'), in_proj_covar=tensor([0.0047, 0.0103, 0.0103, 0.0107, 0.0094, 0.0103, 0.0055, 0.0076], device='cuda:7'), out_proj_covar=tensor([6.6678e-05, 1.5707e-04, 1.5336e-04, 1.6509e-04, 1.4975e-04, 1.6357e-04, 8.2400e-05, 1.2128e-04], device='cuda:7') 2023-04-27 20:06:25,128 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-27 20:06:32,671 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9252, 1.2775, 1.4741, 1.7333, 1.8305, 1.6895, 1.4516, 1.8872], device='cuda:7'), covar=tensor([0.0055, 0.0178, 0.0093, 0.0101, 0.0042, 0.0065, 0.0139, 0.0055], device='cuda:7'), in_proj_covar=tensor([0.0071, 0.0111, 0.0095, 0.0083, 0.0062, 0.0058, 0.0098, 0.0056], device='cuda:7'), out_proj_covar=tensor([1.1974e-04, 1.9125e-04, 1.6939e-04, 1.4814e-04, 1.0436e-04, 9.9109e-05, 1.6569e-04, 9.3700e-05], device='cuda:7') 2023-04-27 20:06:37,155 INFO [train.py:904] (7/8) Epoch 2, batch 8650, loss[loss=0.2326, simple_loss=0.3107, pruned_loss=0.0772, over 12084.00 frames. ], tot_loss[loss=0.2569, simple_loss=0.3334, pruned_loss=0.09024, over 3039137.42 frames. ], batch size: 246, lr: 2.49e-02, grad_scale: 2.0 2023-04-27 20:07:17,790 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5321, 1.8280, 1.7050, 1.6706, 2.2977, 2.0527, 2.2885, 2.3316], device='cuda:7'), covar=tensor([0.0016, 0.0130, 0.0128, 0.0145, 0.0064, 0.0124, 0.0032, 0.0055], device='cuda:7'), in_proj_covar=tensor([0.0046, 0.0102, 0.0102, 0.0106, 0.0093, 0.0103, 0.0054, 0.0075], device='cuda:7'), out_proj_covar=tensor([6.5068e-05, 1.5626e-04, 1.5209e-04, 1.6265e-04, 1.4829e-04, 1.6263e-04, 8.0896e-05, 1.1955e-04], device='cuda:7') 2023-04-27 20:07:26,311 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9641, 5.8455, 5.7466, 5.7382, 5.7003, 6.1798, 6.0496, 5.7030], device='cuda:7'), covar=tensor([0.0548, 0.0995, 0.0898, 0.1292, 0.2035, 0.0820, 0.0714, 0.1602], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0253, 0.0236, 0.0226, 0.0283, 0.0257, 0.0201, 0.0296], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:07:29,118 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.199e+02 3.846e+02 4.801e+02 6.827e+02 6.621e+03, threshold=9.601e+02, percent-clipped=16.0 2023-04-27 20:07:44,463 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18830.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:07:46,501 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18831.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:07:47,013 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-27 20:07:57,238 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1562, 3.9219, 3.7703, 2.6202, 3.7309, 3.6871, 3.9084, 1.8274], device='cuda:7'), covar=tensor([0.0428, 0.0016, 0.0030, 0.0253, 0.0023, 0.0035, 0.0016, 0.0468], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0053, 0.0058, 0.0106, 0.0053, 0.0059, 0.0057, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 20:08:14,589 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18846.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:08:23,073 INFO [train.py:904] (7/8) Epoch 2, batch 8700, loss[loss=0.2524, simple_loss=0.3303, pruned_loss=0.08721, over 16882.00 frames. ], tot_loss[loss=0.2518, simple_loss=0.3291, pruned_loss=0.08721, over 3072039.89 frames. ], batch size: 116, lr: 2.48e-02, grad_scale: 2.0 2023-04-27 20:08:49,561 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18864.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:09:14,854 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18878.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:09:17,252 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=18879.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:09:23,649 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3345, 2.9312, 2.5747, 2.3505, 2.2582, 1.9457, 2.9744, 3.2455], device='cuda:7'), covar=tensor([0.1423, 0.0597, 0.0876, 0.0661, 0.1447, 0.1312, 0.0350, 0.0173], device='cuda:7'), in_proj_covar=tensor([0.0246, 0.0215, 0.0231, 0.0176, 0.0220, 0.0182, 0.0190, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:10:01,031 INFO [train.py:904] (7/8) Epoch 2, batch 8750, loss[loss=0.2048, simple_loss=0.2846, pruned_loss=0.06251, over 11898.00 frames. ], tot_loss[loss=0.2493, simple_loss=0.3274, pruned_loss=0.08557, over 3067600.47 frames. ], batch size: 246, lr: 2.48e-02, grad_scale: 2.0 2023-04-27 20:10:57,519 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.382e+02 3.853e+02 4.840e+02 6.218e+02 1.090e+03, threshold=9.680e+02, percent-clipped=4.0 2023-04-27 20:11:05,636 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-27 20:11:10,312 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18930.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:11:53,576 INFO [train.py:904] (7/8) Epoch 2, batch 8800, loss[loss=0.2115, simple_loss=0.3019, pruned_loss=0.06061, over 16595.00 frames. ], tot_loss[loss=0.2463, simple_loss=0.325, pruned_loss=0.08377, over 3074929.11 frames. ], batch size: 62, lr: 2.48e-02, grad_scale: 4.0 2023-04-27 20:13:18,255 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18991.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:13:38,738 INFO [train.py:904] (7/8) Epoch 2, batch 8850, loss[loss=0.2296, simple_loss=0.3204, pruned_loss=0.0694, over 16174.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.3269, pruned_loss=0.08235, over 3072661.80 frames. ], batch size: 165, lr: 2.47e-02, grad_scale: 4.0 2023-04-27 20:14:28,282 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.418e+02 3.744e+02 4.791e+02 5.972e+02 1.134e+03, threshold=9.582e+02, percent-clipped=2.0 2023-04-27 20:14:43,744 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19030.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:15:27,235 INFO [train.py:904] (7/8) Epoch 2, batch 8900, loss[loss=0.252, simple_loss=0.3345, pruned_loss=0.08477, over 16369.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3271, pruned_loss=0.08197, over 3074835.86 frames. ], batch size: 146, lr: 2.47e-02, grad_scale: 4.0 2023-04-27 20:17:32,227 INFO [train.py:904] (7/8) Epoch 2, batch 8950, loss[loss=0.2407, simple_loss=0.3189, pruned_loss=0.08126, over 16488.00 frames. ], tot_loss[loss=0.2472, simple_loss=0.3277, pruned_loss=0.08334, over 3074369.85 frames. ], batch size: 68, lr: 2.47e-02, grad_scale: 4.0 2023-04-27 20:18:11,548 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0133, 2.3906, 2.2073, 3.2218, 2.0666, 3.0217, 2.2705, 1.9793], device='cuda:7'), covar=tensor([0.0276, 0.0498, 0.0329, 0.0203, 0.1261, 0.0200, 0.0633, 0.1199], device='cuda:7'), in_proj_covar=tensor([0.0211, 0.0189, 0.0158, 0.0211, 0.0260, 0.0169, 0.0191, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:18:20,861 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 4.048e+02 4.879e+02 5.848e+02 1.298e+03, threshold=9.758e+02, percent-clipped=2.0 2023-04-27 20:18:32,664 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19129.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 20:19:11,866 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19146.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:19:21,330 INFO [train.py:904] (7/8) Epoch 2, batch 9000, loss[loss=0.2397, simple_loss=0.3223, pruned_loss=0.07851, over 16156.00 frames. ], tot_loss[loss=0.2427, simple_loss=0.3233, pruned_loss=0.08106, over 3056999.36 frames. ], batch size: 165, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:19:21,331 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 20:19:31,138 INFO [train.py:938] (7/8) Epoch 2, validation: loss=0.2044, simple_loss=0.3047, pruned_loss=0.05205, over 944034.00 frames. 2023-04-27 20:19:31,139 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 20:20:00,862 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19164.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:20:52,964 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19190.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 20:21:02,824 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19194.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:21:13,875 INFO [train.py:904] (7/8) Epoch 2, batch 9050, loss[loss=0.204, simple_loss=0.2929, pruned_loss=0.05754, over 16893.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.3253, pruned_loss=0.0819, over 3078935.96 frames. ], batch size: 96, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:21:38,415 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19212.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 20:22:01,201 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.755e+02 4.212e+02 4.597e+02 5.989e+02 9.578e+02, threshold=9.193e+02, percent-clipped=0.0 2023-04-27 20:22:07,315 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4416, 4.3437, 4.2507, 3.7373, 4.3194, 1.8705, 3.9668, 4.2339], device='cuda:7'), covar=tensor([0.0073, 0.0063, 0.0070, 0.0204, 0.0052, 0.1292, 0.0081, 0.0105], device='cuda:7'), in_proj_covar=tensor([0.0060, 0.0049, 0.0076, 0.0084, 0.0056, 0.0109, 0.0070, 0.0074], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:22:58,689 INFO [train.py:904] (7/8) Epoch 2, batch 9100, loss[loss=0.25, simple_loss=0.3163, pruned_loss=0.09184, over 12471.00 frames. ], tot_loss[loss=0.2443, simple_loss=0.3242, pruned_loss=0.08218, over 3074990.27 frames. ], batch size: 247, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:24:23,848 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19286.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:24:55,905 INFO [train.py:904] (7/8) Epoch 2, batch 9150, loss[loss=0.2359, simple_loss=0.3126, pruned_loss=0.07964, over 16670.00 frames. ], tot_loss[loss=0.2438, simple_loss=0.3244, pruned_loss=0.08158, over 3073926.02 frames. ], batch size: 57, lr: 2.46e-02, grad_scale: 2.0 2023-04-27 20:25:05,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5858, 4.1414, 4.1978, 1.8417, 4.3266, 4.4015, 3.2762, 3.4693], device='cuda:7'), covar=tensor([0.0872, 0.0103, 0.0165, 0.1502, 0.0053, 0.0041, 0.0343, 0.0339], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0082, 0.0078, 0.0145, 0.0069, 0.0068, 0.0105, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0003], device='cuda:7') 2023-04-27 20:25:49,339 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.714e+02 4.631e+02 5.471e+02 6.935e+02 2.154e+03, threshold=1.094e+03, percent-clipped=5.0 2023-04-27 20:25:54,638 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.25 vs. limit=5.0 2023-04-27 20:25:59,968 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19330.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:26:11,295 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5449, 3.3937, 3.0317, 1.7134, 2.4747, 2.1347, 2.9311, 3.4113], device='cuda:7'), covar=tensor([0.0420, 0.0526, 0.0539, 0.1747, 0.0969, 0.0987, 0.1025, 0.0455], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0101, 0.0155, 0.0152, 0.0143, 0.0135, 0.0147, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 20:26:40,136 INFO [train.py:904] (7/8) Epoch 2, batch 9200, loss[loss=0.227, simple_loss=0.2961, pruned_loss=0.07901, over 11857.00 frames. ], tot_loss[loss=0.2386, simple_loss=0.3184, pruned_loss=0.0794, over 3072520.60 frames. ], batch size: 246, lr: 2.45e-02, grad_scale: 4.0 2023-04-27 20:27:30,641 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19378.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:28:03,632 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19394.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:28:16,038 INFO [train.py:904] (7/8) Epoch 2, batch 9250, loss[loss=0.2094, simple_loss=0.3, pruned_loss=0.05943, over 15365.00 frames. ], tot_loss[loss=0.2386, simple_loss=0.3183, pruned_loss=0.07946, over 3091170.28 frames. ], batch size: 191, lr: 2.45e-02, grad_scale: 4.0 2023-04-27 20:29:05,882 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.723e+02 4.011e+02 4.767e+02 6.782e+02 2.707e+03, threshold=9.534e+02, percent-clipped=7.0 2023-04-27 20:29:07,816 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4689, 3.4177, 1.5921, 3.4258, 2.2256, 3.4522, 1.8657, 2.7676], device='cuda:7'), covar=tensor([0.0045, 0.0162, 0.1664, 0.0041, 0.0798, 0.0258, 0.1311, 0.0513], device='cuda:7'), in_proj_covar=tensor([0.0074, 0.0111, 0.0168, 0.0073, 0.0151, 0.0132, 0.0172, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 20:30:06,017 INFO [train.py:904] (7/8) Epoch 2, batch 9300, loss[loss=0.2309, simple_loss=0.2965, pruned_loss=0.08267, over 12527.00 frames. ], tot_loss[loss=0.2359, simple_loss=0.3157, pruned_loss=0.07798, over 3080487.43 frames. ], batch size: 250, lr: 2.45e-02, grad_scale: 4.0 2023-04-27 20:30:14,340 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19455.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:30:21,743 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5049, 4.3101, 4.2472, 3.7882, 4.3491, 1.4246, 4.1029, 4.2419], device='cuda:7'), covar=tensor([0.0056, 0.0044, 0.0077, 0.0187, 0.0046, 0.1450, 0.0059, 0.0082], device='cuda:7'), in_proj_covar=tensor([0.0060, 0.0049, 0.0074, 0.0083, 0.0057, 0.0109, 0.0067, 0.0073], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:31:20,845 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19485.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 20:31:49,022 INFO [train.py:904] (7/8) Epoch 2, batch 9350, loss[loss=0.2424, simple_loss=0.3169, pruned_loss=0.08392, over 16838.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.3161, pruned_loss=0.07836, over 3071101.59 frames. ], batch size: 124, lr: 2.44e-02, grad_scale: 4.0 2023-04-27 20:32:30,645 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-04-27 20:32:37,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.597e+02 3.883e+02 4.622e+02 6.098e+02 2.142e+03, threshold=9.245e+02, percent-clipped=3.0 2023-04-27 20:32:40,488 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9661, 4.6586, 4.5438, 2.2344, 4.5814, 4.6859, 3.8143, 3.8986], device='cuda:7'), covar=tensor([0.0673, 0.0035, 0.0096, 0.1142, 0.0047, 0.0029, 0.0192, 0.0218], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0080, 0.0079, 0.0145, 0.0071, 0.0068, 0.0104, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:33:28,468 INFO [train.py:904] (7/8) Epoch 2, batch 9400, loss[loss=0.2357, simple_loss=0.333, pruned_loss=0.06923, over 15396.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.3162, pruned_loss=0.07818, over 3060471.74 frames. ], batch size: 191, lr: 2.44e-02, grad_scale: 4.0 2023-04-27 20:33:38,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8665, 2.9335, 2.6830, 4.4108, 1.9880, 4.2757, 2.5380, 2.2776], device='cuda:7'), covar=tensor([0.0246, 0.0472, 0.0330, 0.0125, 0.1486, 0.0121, 0.0625, 0.1199], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0195, 0.0159, 0.0217, 0.0267, 0.0173, 0.0194, 0.0239], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:34:12,656 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-04-27 20:34:39,214 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19586.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:35:08,268 INFO [train.py:904] (7/8) Epoch 2, batch 9450, loss[loss=0.2191, simple_loss=0.3014, pruned_loss=0.06845, over 16491.00 frames. ], tot_loss[loss=0.2382, simple_loss=0.3184, pruned_loss=0.07903, over 3048986.18 frames. ], batch size: 68, lr: 2.44e-02, grad_scale: 4.0 2023-04-27 20:35:42,918 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19618.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:35:56,428 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.451e+02 4.122e+02 5.164e+02 6.390e+02 1.240e+03, threshold=1.033e+03, percent-clipped=6.0 2023-04-27 20:35:57,235 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1725, 3.2710, 1.5907, 3.2014, 2.1912, 3.2438, 1.8274, 2.6894], device='cuda:7'), covar=tensor([0.0052, 0.0148, 0.1387, 0.0039, 0.0734, 0.0284, 0.1379, 0.0558], device='cuda:7'), in_proj_covar=tensor([0.0075, 0.0110, 0.0170, 0.0073, 0.0154, 0.0135, 0.0176, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 20:36:16,219 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19634.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:36:22,318 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3025, 1.2946, 1.6165, 1.9550, 2.0925, 2.2536, 1.3726, 2.0740], device='cuda:7'), covar=tensor([0.0046, 0.0199, 0.0128, 0.0105, 0.0049, 0.0058, 0.0167, 0.0045], device='cuda:7'), in_proj_covar=tensor([0.0072, 0.0109, 0.0095, 0.0082, 0.0063, 0.0059, 0.0097, 0.0054], device='cuda:7'), out_proj_covar=tensor([1.2153e-04, 1.8504e-04, 1.6733e-04, 1.4396e-04, 1.0542e-04, 9.7564e-05, 1.6190e-04, 8.7953e-05], device='cuda:7') 2023-04-27 20:36:48,224 INFO [train.py:904] (7/8) Epoch 2, batch 9500, loss[loss=0.2472, simple_loss=0.3259, pruned_loss=0.08419, over 16336.00 frames. ], tot_loss[loss=0.2355, simple_loss=0.3164, pruned_loss=0.07734, over 3065829.36 frames. ], batch size: 146, lr: 2.43e-02, grad_scale: 4.0 2023-04-27 20:37:21,144 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-04-27 20:37:46,314 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19679.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:38:34,563 INFO [train.py:904] (7/8) Epoch 2, batch 9550, loss[loss=0.2415, simple_loss=0.3258, pruned_loss=0.07853, over 16727.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.3162, pruned_loss=0.07764, over 3060743.51 frames. ], batch size: 76, lr: 2.43e-02, grad_scale: 4.0 2023-04-27 20:39:23,646 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 3.908e+02 4.753e+02 5.956e+02 1.328e+03, threshold=9.507e+02, percent-clipped=2.0 2023-04-27 20:40:12,718 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19750.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:40:13,482 INFO [train.py:904] (7/8) Epoch 2, batch 9600, loss[loss=0.2419, simple_loss=0.3301, pruned_loss=0.07682, over 16859.00 frames. ], tot_loss[loss=0.2385, simple_loss=0.3184, pruned_loss=0.07928, over 3049296.19 frames. ], batch size: 102, lr: 2.43e-02, grad_scale: 8.0 2023-04-27 20:40:59,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0387, 4.2171, 4.2121, 4.2293, 4.2134, 4.6171, 4.4291, 4.1015], device='cuda:7'), covar=tensor([0.0999, 0.0929, 0.0729, 0.1141, 0.1607, 0.0768, 0.0694, 0.1397], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0247, 0.0235, 0.0222, 0.0276, 0.0255, 0.0190, 0.0286], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:40:59,985 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-27 20:41:22,086 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19785.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 20:41:48,644 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-27 20:42:00,152 INFO [train.py:904] (7/8) Epoch 2, batch 9650, loss[loss=0.2284, simple_loss=0.3175, pruned_loss=0.06963, over 16915.00 frames. ], tot_loss[loss=0.241, simple_loss=0.3218, pruned_loss=0.08008, over 3074562.68 frames. ], batch size: 102, lr: 2.43e-02, grad_scale: 8.0 2023-04-27 20:42:54,376 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.395e+02 4.451e+02 5.149e+02 6.499e+02 1.556e+03, threshold=1.030e+03, percent-clipped=6.0 2023-04-27 20:43:10,775 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=19833.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 20:43:48,664 INFO [train.py:904] (7/8) Epoch 2, batch 9700, loss[loss=0.2446, simple_loss=0.3224, pruned_loss=0.08336, over 16772.00 frames. ], tot_loss[loss=0.24, simple_loss=0.3205, pruned_loss=0.0797, over 3084911.92 frames. ], batch size: 124, lr: 2.42e-02, grad_scale: 8.0 2023-04-27 20:45:33,548 INFO [train.py:904] (7/8) Epoch 2, batch 9750, loss[loss=0.2301, simple_loss=0.3017, pruned_loss=0.07922, over 12405.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3191, pruned_loss=0.07977, over 3072384.39 frames. ], batch size: 248, lr: 2.42e-02, grad_scale: 8.0 2023-04-27 20:46:21,275 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.379e+02 3.927e+02 4.879e+02 5.766e+02 1.284e+03, threshold=9.758e+02, percent-clipped=1.0 2023-04-27 20:46:27,206 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-27 20:47:15,374 INFO [train.py:904] (7/8) Epoch 2, batch 9800, loss[loss=0.2163, simple_loss=0.2942, pruned_loss=0.06923, over 12441.00 frames. ], tot_loss[loss=0.2369, simple_loss=0.3178, pruned_loss=0.07802, over 3065406.52 frames. ], batch size: 248, lr: 2.42e-02, grad_scale: 8.0 2023-04-27 20:47:33,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4097, 4.2496, 4.4183, 4.7271, 4.7113, 4.2633, 4.7884, 4.7532], device='cuda:7'), covar=tensor([0.0528, 0.0508, 0.0809, 0.0325, 0.0399, 0.0478, 0.0279, 0.0280], device='cuda:7'), in_proj_covar=tensor([0.0244, 0.0275, 0.0366, 0.0276, 0.0213, 0.0189, 0.0220, 0.0220], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:48:00,427 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19974.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:49:05,361 INFO [train.py:904] (7/8) Epoch 2, batch 9850, loss[loss=0.234, simple_loss=0.3208, pruned_loss=0.07358, over 15439.00 frames. ], tot_loss[loss=0.2366, simple_loss=0.3189, pruned_loss=0.07722, over 3048227.41 frames. ], batch size: 191, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:49:46,197 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20020.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:49:55,608 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.468e+02 4.069e+02 4.622e+02 5.694e+02 1.429e+03, threshold=9.244e+02, percent-clipped=4.0 2023-04-27 20:50:55,104 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6154, 3.0973, 2.9659, 2.0308, 2.8876, 2.9041, 3.1657, 1.4875], device='cuda:7'), covar=tensor([0.0482, 0.0037, 0.0056, 0.0280, 0.0054, 0.0060, 0.0029, 0.0495], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0054, 0.0058, 0.0106, 0.0053, 0.0057, 0.0057, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 20:50:58,204 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20050.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:50:58,956 INFO [train.py:904] (7/8) Epoch 2, batch 9900, loss[loss=0.2338, simple_loss=0.3046, pruned_loss=0.08154, over 12510.00 frames. ], tot_loss[loss=0.2373, simple_loss=0.3194, pruned_loss=0.07757, over 3028625.37 frames. ], batch size: 248, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:51:30,618 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-27 20:52:02,661 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2500, 3.1736, 1.3872, 3.3089, 2.2315, 3.1901, 1.9052, 2.6224], device='cuda:7'), covar=tensor([0.0053, 0.0168, 0.1465, 0.0036, 0.0667, 0.0366, 0.1159, 0.0509], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0112, 0.0171, 0.0073, 0.0153, 0.0136, 0.0176, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 20:52:09,596 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8295, 3.7712, 3.6191, 3.7107, 3.2878, 3.7514, 3.5909, 3.4571], device='cuda:7'), covar=tensor([0.0284, 0.0158, 0.0201, 0.0143, 0.0591, 0.0186, 0.0462, 0.0305], device='cuda:7'), in_proj_covar=tensor([0.0114, 0.0092, 0.0139, 0.0116, 0.0164, 0.0123, 0.0099, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 20:52:12,606 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20081.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:52:31,309 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20089.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:52:51,147 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20098.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:52:56,917 INFO [train.py:904] (7/8) Epoch 2, batch 9950, loss[loss=0.2301, simple_loss=0.3177, pruned_loss=0.07122, over 17007.00 frames. ], tot_loss[loss=0.239, simple_loss=0.3216, pruned_loss=0.07823, over 3029281.14 frames. ], batch size: 109, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:53:24,087 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20112.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:53:55,582 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 4.201e+02 4.972e+02 6.369e+02 1.989e+03, threshold=9.943e+02, percent-clipped=8.0 2023-04-27 20:54:20,877 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-27 20:54:56,072 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20150.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:54:56,745 INFO [train.py:904] (7/8) Epoch 2, batch 10000, loss[loss=0.1932, simple_loss=0.2841, pruned_loss=0.05112, over 17023.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.3199, pruned_loss=0.07717, over 3055050.29 frames. ], batch size: 55, lr: 2.41e-02, grad_scale: 8.0 2023-04-27 20:55:41,364 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20173.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:56:37,814 INFO [train.py:904] (7/8) Epoch 2, batch 10050, loss[loss=0.2327, simple_loss=0.3112, pruned_loss=0.07705, over 11879.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.3194, pruned_loss=0.07654, over 3059909.55 frames. ], batch size: 248, lr: 2.40e-02, grad_scale: 8.0 2023-04-27 20:56:44,610 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20204.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:57:24,817 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.701e+02 4.201e+02 4.991e+02 6.585e+02 1.392e+03, threshold=9.982e+02, percent-clipped=3.0 2023-04-27 20:58:10,197 INFO [train.py:904] (7/8) Epoch 2, batch 10100, loss[loss=0.2163, simple_loss=0.2974, pruned_loss=0.06764, over 16292.00 frames. ], tot_loss[loss=0.2376, simple_loss=0.3203, pruned_loss=0.07752, over 3062222.83 frames. ], batch size: 146, lr: 2.40e-02, grad_scale: 8.0 2023-04-27 20:58:38,003 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20265.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:58:58,840 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20274.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 20:59:54,874 INFO [train.py:904] (7/8) Epoch 3, batch 0, loss[loss=0.4152, simple_loss=0.4236, pruned_loss=0.2034, over 16744.00 frames. ], tot_loss[loss=0.4152, simple_loss=0.4236, pruned_loss=0.2034, over 16744.00 frames. ], batch size: 124, lr: 2.28e-02, grad_scale: 8.0 2023-04-27 20:59:54,875 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 21:00:02,300 INFO [train.py:938] (7/8) Epoch 3, validation: loss=0.2012, simple_loss=0.3019, pruned_loss=0.05024, over 944034.00 frames. 2023-04-27 21:00:02,301 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 21:00:18,029 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-27 21:00:31,874 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20322.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:00:38,800 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.937e+02 4.805e+02 5.776e+02 7.144e+02 1.042e+03, threshold=1.155e+03, percent-clipped=3.0 2023-04-27 21:00:59,233 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2131, 4.2698, 4.7933, 4.7697, 4.8383, 4.3006, 4.3160, 4.3142], device='cuda:7'), covar=tensor([0.0215, 0.0310, 0.0374, 0.0380, 0.0343, 0.0291, 0.0652, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0150, 0.0167, 0.0166, 0.0195, 0.0172, 0.0252, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-27 21:01:12,900 INFO [train.py:904] (7/8) Epoch 3, batch 50, loss[loss=0.2915, simple_loss=0.3437, pruned_loss=0.1196, over 16438.00 frames. ], tot_loss[loss=0.2924, simple_loss=0.3498, pruned_loss=0.1175, over 749169.59 frames. ], batch size: 75, lr: 2.28e-02, grad_scale: 2.0 2023-04-27 21:01:16,012 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 21:01:45,795 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20376.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:01:58,199 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-04-27 21:02:03,786 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20388.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:02:19,625 INFO [train.py:904] (7/8) Epoch 3, batch 100, loss[loss=0.2584, simple_loss=0.3397, pruned_loss=0.08852, over 17066.00 frames. ], tot_loss[loss=0.2771, simple_loss=0.3398, pruned_loss=0.1073, over 1328746.09 frames. ], batch size: 50, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:02:43,285 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20418.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:02:54,678 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.729e+02 4.335e+02 5.443e+02 6.351e+02 1.481e+03, threshold=1.089e+03, percent-clipped=3.0 2023-04-27 21:03:18,756 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20445.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:03:24,419 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20449.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:03:26,375 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8559, 4.6963, 4.5370, 4.7032, 3.9553, 4.6446, 4.7240, 4.2520], device='cuda:7'), covar=tensor([0.0318, 0.0152, 0.0195, 0.0147, 0.0815, 0.0199, 0.0193, 0.0253], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0105, 0.0158, 0.0131, 0.0193, 0.0142, 0.0111, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:03:27,033 INFO [train.py:904] (7/8) Epoch 3, batch 150, loss[loss=0.2742, simple_loss=0.3431, pruned_loss=0.1027, over 16786.00 frames. ], tot_loss[loss=0.2754, simple_loss=0.3384, pruned_loss=0.1062, over 1760932.55 frames. ], batch size: 62, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:03:49,817 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20468.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:04:05,932 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20479.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:04:35,440 INFO [train.py:904] (7/8) Epoch 3, batch 200, loss[loss=0.2174, simple_loss=0.2955, pruned_loss=0.06965, over 16815.00 frames. ], tot_loss[loss=0.2708, simple_loss=0.3351, pruned_loss=0.1033, over 2109930.48 frames. ], batch size: 42, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:04:40,010 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.03 vs. limit=5.0 2023-04-27 21:05:09,761 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.891e+02 4.616e+02 5.990e+02 1.220e+03, threshold=9.233e+02, percent-clipped=2.0 2023-04-27 21:05:16,714 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3269, 4.7084, 4.3746, 4.5532, 4.1440, 4.0514, 4.1545, 4.7609], device='cuda:7'), covar=tensor([0.0543, 0.0627, 0.0765, 0.0400, 0.0552, 0.0788, 0.0544, 0.0606], device='cuda:7'), in_proj_covar=tensor([0.0249, 0.0350, 0.0309, 0.0215, 0.0235, 0.0218, 0.0278, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:05:27,353 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-27 21:05:43,635 INFO [train.py:904] (7/8) Epoch 3, batch 250, loss[loss=0.228, simple_loss=0.2955, pruned_loss=0.0802, over 17181.00 frames. ], tot_loss[loss=0.2694, simple_loss=0.3317, pruned_loss=0.1036, over 2380160.21 frames. ], batch size: 46, lr: 2.27e-02, grad_scale: 2.0 2023-04-27 21:05:58,366 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20560.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:06:10,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2104, 2.2759, 1.6661, 2.0326, 2.8913, 2.8647, 3.6276, 3.2451], device='cuda:7'), covar=tensor([0.0019, 0.0141, 0.0176, 0.0157, 0.0076, 0.0108, 0.0039, 0.0056], device='cuda:7'), in_proj_covar=tensor([0.0056, 0.0111, 0.0109, 0.0110, 0.0102, 0.0109, 0.0063, 0.0081], device='cuda:7'), out_proj_covar=tensor([7.6883e-05, 1.6821e-04, 1.5892e-04, 1.6531e-04, 1.5611e-04, 1.6782e-04, 9.3578e-05, 1.2659e-04], device='cuda:7') 2023-04-27 21:06:35,646 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20588.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:06:54,935 INFO [train.py:904] (7/8) Epoch 3, batch 300, loss[loss=0.2372, simple_loss=0.324, pruned_loss=0.07522, over 17082.00 frames. ], tot_loss[loss=0.265, simple_loss=0.3284, pruned_loss=0.1008, over 2585798.50 frames. ], batch size: 55, lr: 2.26e-02, grad_scale: 2.0 2023-04-27 21:07:29,014 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.643e+02 3.919e+02 4.746e+02 5.726e+02 1.058e+03, threshold=9.491e+02, percent-clipped=1.0 2023-04-27 21:07:51,079 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-27 21:07:59,199 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20649.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:08:01,107 INFO [train.py:904] (7/8) Epoch 3, batch 350, loss[loss=0.1859, simple_loss=0.2608, pruned_loss=0.0555, over 16779.00 frames. ], tot_loss[loss=0.2583, simple_loss=0.3234, pruned_loss=0.09661, over 2750728.02 frames. ], batch size: 39, lr: 2.26e-02, grad_scale: 2.0 2023-04-27 21:08:36,873 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20676.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:08:59,912 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6079, 3.5451, 2.7606, 2.5463, 2.6383, 1.9599, 3.5138, 4.0483], device='cuda:7'), covar=tensor([0.1640, 0.0487, 0.0938, 0.0747, 0.1539, 0.1178, 0.0361, 0.0324], device='cuda:7'), in_proj_covar=tensor([0.0261, 0.0233, 0.0251, 0.0192, 0.0251, 0.0187, 0.0207, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:09:09,382 INFO [train.py:904] (7/8) Epoch 3, batch 400, loss[loss=0.2278, simple_loss=0.3043, pruned_loss=0.07571, over 16819.00 frames. ], tot_loss[loss=0.2582, simple_loss=0.3225, pruned_loss=0.09695, over 2878530.02 frames. ], batch size: 42, lr: 2.26e-02, grad_scale: 4.0 2023-04-27 21:09:17,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1111, 3.6855, 3.9632, 2.9707, 3.8744, 3.8671, 4.0351, 1.9753], device='cuda:7'), covar=tensor([0.0386, 0.0073, 0.0048, 0.0202, 0.0043, 0.0071, 0.0027, 0.0380], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0056, 0.0059, 0.0105, 0.0054, 0.0060, 0.0059, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:09:41,231 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20724.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:09:45,247 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.989e+02 4.408e+02 5.242e+02 6.457e+02 1.269e+03, threshold=1.048e+03, percent-clipped=5.0 2023-04-27 21:10:09,418 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20744.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:10:10,768 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20745.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:10:17,856 INFO [train.py:904] (7/8) Epoch 3, batch 450, loss[loss=0.2595, simple_loss=0.3111, pruned_loss=0.1039, over 16877.00 frames. ], tot_loss[loss=0.2533, simple_loss=0.3189, pruned_loss=0.09385, over 2986292.42 frames. ], batch size: 109, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:10:29,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6911, 3.5333, 3.5873, 3.9235, 3.9490, 3.5182, 3.8724, 3.9659], device='cuda:7'), covar=tensor([0.0460, 0.0524, 0.1035, 0.0383, 0.0369, 0.1240, 0.0458, 0.0326], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0332, 0.0451, 0.0334, 0.0258, 0.0234, 0.0249, 0.0262], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:10:42,443 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20768.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:10:49,455 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20774.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:11:14,839 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20793.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:11:24,056 INFO [train.py:904] (7/8) Epoch 3, batch 500, loss[loss=0.2437, simple_loss=0.3232, pruned_loss=0.08216, over 16707.00 frames. ], tot_loss[loss=0.2517, simple_loss=0.3173, pruned_loss=0.09306, over 3066267.97 frames. ], batch size: 57, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:11:46,180 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20816.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:12:00,590 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.032e+02 4.223e+02 4.982e+02 5.898e+02 1.253e+03, threshold=9.964e+02, percent-clipped=1.0 2023-04-27 21:12:34,381 INFO [train.py:904] (7/8) Epoch 3, batch 550, loss[loss=0.2423, simple_loss=0.3218, pruned_loss=0.08135, over 17079.00 frames. ], tot_loss[loss=0.2499, simple_loss=0.3157, pruned_loss=0.09203, over 3109646.76 frames. ], batch size: 53, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:12:38,424 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9339, 3.9724, 2.9825, 5.1536, 5.1009, 4.4284, 1.8834, 3.4560], device='cuda:7'), covar=tensor([0.1438, 0.0381, 0.1029, 0.0079, 0.0154, 0.0269, 0.1260, 0.0598], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0128, 0.0164, 0.0071, 0.0132, 0.0130, 0.0155, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:12:45,343 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20860.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:12:48,435 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9660, 4.8969, 4.7127, 2.4422, 4.8397, 4.9972, 3.7640, 4.2323], device='cuda:7'), covar=tensor([0.0749, 0.0085, 0.0210, 0.1181, 0.0045, 0.0034, 0.0304, 0.0227], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0082, 0.0080, 0.0148, 0.0074, 0.0072, 0.0114, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:12:59,717 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-27 21:13:40,846 INFO [train.py:904] (7/8) Epoch 3, batch 600, loss[loss=0.2344, simple_loss=0.314, pruned_loss=0.07741, over 17165.00 frames. ], tot_loss[loss=0.2493, simple_loss=0.3146, pruned_loss=0.09205, over 3153154.84 frames. ], batch size: 46, lr: 2.25e-02, grad_scale: 4.0 2023-04-27 21:13:50,655 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=20908.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:13:55,234 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-04-27 21:14:01,160 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8829, 4.5852, 4.4760, 5.0326, 5.1907, 4.5104, 5.1612, 5.0834], device='cuda:7'), covar=tensor([0.0451, 0.0582, 0.1364, 0.0542, 0.0357, 0.0406, 0.0392, 0.0320], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0333, 0.0448, 0.0338, 0.0256, 0.0235, 0.0254, 0.0267], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:14:01,699 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-04-27 21:14:15,036 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3811, 1.6616, 2.6037, 3.0712, 3.0004, 3.3125, 1.6910, 3.1644], device='cuda:7'), covar=tensor([0.0041, 0.0229, 0.0111, 0.0096, 0.0053, 0.0063, 0.0188, 0.0040], device='cuda:7'), in_proj_covar=tensor([0.0079, 0.0115, 0.0102, 0.0093, 0.0073, 0.0064, 0.0104, 0.0058], device='cuda:7'), out_proj_covar=tensor([1.3159e-04, 1.9256e-04, 1.7433e-04, 1.6099e-04, 1.1743e-04, 1.0732e-04, 1.6946e-04, 9.6007e-05], device='cuda:7') 2023-04-27 21:14:15,691 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.677e+02 4.207e+02 5.314e+02 6.523e+02 1.813e+03, threshold=1.063e+03, percent-clipped=4.0 2023-04-27 21:14:38,957 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20944.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:14:39,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7482, 4.6020, 2.0314, 4.7222, 2.8364, 4.7672, 2.1795, 3.2327], device='cuda:7'), covar=tensor([0.0032, 0.0124, 0.1381, 0.0028, 0.0715, 0.0174, 0.1278, 0.0513], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0128, 0.0172, 0.0079, 0.0159, 0.0157, 0.0180, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 21:14:48,040 INFO [train.py:904] (7/8) Epoch 3, batch 650, loss[loss=0.2618, simple_loss=0.3234, pruned_loss=0.1001, over 16229.00 frames. ], tot_loss[loss=0.2465, simple_loss=0.313, pruned_loss=0.09003, over 3185709.93 frames. ], batch size: 165, lr: 2.24e-02, grad_scale: 4.0 2023-04-27 21:15:03,848 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6079, 3.4255, 3.1794, 3.8362, 3.8397, 3.4589, 3.6748, 3.8520], device='cuda:7'), covar=tensor([0.0605, 0.0663, 0.1673, 0.0567, 0.0599, 0.1543, 0.0723, 0.0504], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0335, 0.0450, 0.0339, 0.0260, 0.0238, 0.0255, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:15:04,072 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-27 21:15:40,714 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8911, 4.5314, 4.7858, 5.1387, 5.2937, 4.4802, 5.2947, 5.1327], device='cuda:7'), covar=tensor([0.0602, 0.0673, 0.1166, 0.0416, 0.0325, 0.0512, 0.0341, 0.0356], device='cuda:7'), in_proj_covar=tensor([0.0289, 0.0338, 0.0454, 0.0343, 0.0262, 0.0240, 0.0257, 0.0271], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:15:57,313 INFO [train.py:904] (7/8) Epoch 3, batch 700, loss[loss=0.2043, simple_loss=0.2829, pruned_loss=0.06282, over 16851.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3127, pruned_loss=0.08915, over 3218412.21 frames. ], batch size: 42, lr: 2.24e-02, grad_scale: 4.0 2023-04-27 21:16:30,871 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.136e+02 3.749e+02 4.303e+02 5.331e+02 1.033e+03, threshold=8.606e+02, percent-clipped=0.0 2023-04-27 21:16:36,035 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-27 21:16:55,561 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21044.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:17:03,843 INFO [train.py:904] (7/8) Epoch 3, batch 750, loss[loss=0.2496, simple_loss=0.3101, pruned_loss=0.09453, over 16786.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3127, pruned_loss=0.08915, over 3246700.76 frames. ], batch size: 102, lr: 2.24e-02, grad_scale: 4.0 2023-04-27 21:17:06,034 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21052.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:17:35,041 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21074.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:17:58,572 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21092.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:10,822 INFO [train.py:904] (7/8) Epoch 3, batch 800, loss[loss=0.2602, simple_loss=0.3123, pruned_loss=0.1041, over 16864.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.313, pruned_loss=0.08937, over 3266159.64 frames. ], batch size: 116, lr: 2.24e-02, grad_scale: 8.0 2023-04-27 21:18:27,393 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21113.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:34,945 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21118.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:40,094 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21122.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:18:47,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.745e+02 4.034e+02 4.669e+02 5.861e+02 1.175e+03, threshold=9.338e+02, percent-clipped=5.0 2023-04-27 21:19:11,810 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-27 21:19:20,166 INFO [train.py:904] (7/8) Epoch 3, batch 850, loss[loss=0.2209, simple_loss=0.29, pruned_loss=0.07584, over 16822.00 frames. ], tot_loss[loss=0.2435, simple_loss=0.311, pruned_loss=0.08798, over 3281621.64 frames. ], batch size: 102, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:19:58,215 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21179.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:20:27,600 INFO [train.py:904] (7/8) Epoch 3, batch 900, loss[loss=0.2225, simple_loss=0.3054, pruned_loss=0.06976, over 17048.00 frames. ], tot_loss[loss=0.2414, simple_loss=0.3102, pruned_loss=0.08625, over 3297052.41 frames. ], batch size: 55, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:21:03,016 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.660e+02 3.821e+02 4.761e+02 5.619e+02 1.173e+03, threshold=9.521e+02, percent-clipped=3.0 2023-04-27 21:21:19,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2545, 3.2130, 1.4076, 3.3049, 2.1729, 3.2824, 1.8528, 2.4518], device='cuda:7'), covar=tensor([0.0071, 0.0216, 0.1595, 0.0073, 0.0717, 0.0417, 0.1164, 0.0560], device='cuda:7'), in_proj_covar=tensor([0.0082, 0.0129, 0.0172, 0.0081, 0.0158, 0.0160, 0.0180, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 21:21:26,737 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21244.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:21:35,858 INFO [train.py:904] (7/8) Epoch 3, batch 950, loss[loss=0.2378, simple_loss=0.2926, pruned_loss=0.09149, over 16904.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3106, pruned_loss=0.08672, over 3301593.51 frames. ], batch size: 116, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:22:34,705 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21292.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:22:45,624 INFO [train.py:904] (7/8) Epoch 3, batch 1000, loss[loss=0.2031, simple_loss=0.2824, pruned_loss=0.06185, over 16793.00 frames. ], tot_loss[loss=0.2391, simple_loss=0.3081, pruned_loss=0.08509, over 3303990.79 frames. ], batch size: 39, lr: 2.23e-02, grad_scale: 8.0 2023-04-27 21:23:01,345 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21312.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:23:21,768 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.770e+02 3.937e+02 4.742e+02 6.159e+02 9.838e+02, threshold=9.485e+02, percent-clipped=2.0 2023-04-27 21:23:24,004 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21328.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:23:55,384 INFO [train.py:904] (7/8) Epoch 3, batch 1050, loss[loss=0.2479, simple_loss=0.3043, pruned_loss=0.09573, over 16833.00 frames. ], tot_loss[loss=0.2395, simple_loss=0.3079, pruned_loss=0.08557, over 3311262.46 frames. ], batch size: 116, lr: 2.22e-02, grad_scale: 8.0 2023-04-27 21:24:25,452 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21373.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:24:46,856 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21389.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:24:57,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3779, 4.4980, 5.0724, 5.1247, 5.0907, 4.6194, 4.6237, 4.5710], device='cuda:7'), covar=tensor([0.0262, 0.0233, 0.0260, 0.0269, 0.0314, 0.0247, 0.0597, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0174, 0.0194, 0.0190, 0.0224, 0.0197, 0.0290, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 21:25:02,822 INFO [train.py:904] (7/8) Epoch 3, batch 1100, loss[loss=0.2381, simple_loss=0.3152, pruned_loss=0.0805, over 17051.00 frames. ], tot_loss[loss=0.2398, simple_loss=0.3079, pruned_loss=0.0859, over 3315582.81 frames. ], batch size: 53, lr: 2.22e-02, grad_scale: 4.0 2023-04-27 21:25:12,313 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21408.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:25:38,416 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.639e+02 4.233e+02 5.080e+02 6.558e+02 1.311e+03, threshold=1.016e+03, percent-clipped=3.0 2023-04-27 21:26:10,407 INFO [train.py:904] (7/8) Epoch 3, batch 1150, loss[loss=0.2353, simple_loss=0.2962, pruned_loss=0.08717, over 16826.00 frames. ], tot_loss[loss=0.238, simple_loss=0.3067, pruned_loss=0.08467, over 3321016.34 frames. ], batch size: 102, lr: 2.22e-02, grad_scale: 4.0 2023-04-27 21:26:42,693 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21474.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:27:19,796 INFO [train.py:904] (7/8) Epoch 3, batch 1200, loss[loss=0.2514, simple_loss=0.3052, pruned_loss=0.09874, over 16457.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.305, pruned_loss=0.08373, over 3327451.75 frames. ], batch size: 75, lr: 2.22e-02, grad_scale: 8.0 2023-04-27 21:27:45,610 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3527, 4.4290, 4.0543, 1.6640, 2.9457, 2.3197, 3.9289, 4.3176], device='cuda:7'), covar=tensor([0.0282, 0.0377, 0.0410, 0.1686, 0.0772, 0.1006, 0.0663, 0.0586], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0118, 0.0155, 0.0146, 0.0141, 0.0133, 0.0149, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 21:27:56,784 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.205e+02 3.899e+02 4.622e+02 5.754e+02 9.760e+02, threshold=9.243e+02, percent-clipped=0.0 2023-04-27 21:28:27,751 INFO [train.py:904] (7/8) Epoch 3, batch 1250, loss[loss=0.2007, simple_loss=0.2774, pruned_loss=0.06205, over 15878.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3056, pruned_loss=0.08465, over 3325498.85 frames. ], batch size: 35, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:28:33,307 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3161, 5.1494, 5.1567, 4.0931, 5.0992, 2.1064, 4.7126, 5.1560], device='cuda:7'), covar=tensor([0.0058, 0.0072, 0.0057, 0.0360, 0.0047, 0.1467, 0.0080, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0063, 0.0097, 0.0115, 0.0071, 0.0121, 0.0087, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:29:30,785 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21595.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:29:39,772 INFO [train.py:904] (7/8) Epoch 3, batch 1300, loss[loss=0.2421, simple_loss=0.3193, pruned_loss=0.08244, over 17049.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.3047, pruned_loss=0.08387, over 3334215.19 frames. ], batch size: 50, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:29:56,131 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21613.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:30:05,718 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8944, 3.2331, 2.6360, 4.3906, 4.2582, 4.1746, 1.8759, 3.0934], device='cuda:7'), covar=tensor([0.1229, 0.0345, 0.1026, 0.0061, 0.0221, 0.0219, 0.1014, 0.0580], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0129, 0.0165, 0.0074, 0.0141, 0.0135, 0.0154, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:30:17,382 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.341e+02 3.774e+02 4.586e+02 5.366e+02 8.286e+02, threshold=9.173e+02, percent-clipped=0.0 2023-04-27 21:30:49,956 INFO [train.py:904] (7/8) Epoch 3, batch 1350, loss[loss=0.2509, simple_loss=0.3039, pruned_loss=0.09891, over 16702.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.3049, pruned_loss=0.08382, over 3331921.17 frames. ], batch size: 89, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:30:56,175 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21656.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:13,100 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21668.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:20,966 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21674.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:35,894 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21684.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:31:58,967 INFO [train.py:904] (7/8) Epoch 3, batch 1400, loss[loss=0.2318, simple_loss=0.3009, pruned_loss=0.08136, over 16423.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.3048, pruned_loss=0.0838, over 3327325.20 frames. ], batch size: 68, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:32:09,242 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21708.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:32:35,079 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.932e+02 4.012e+02 4.714e+02 6.047e+02 1.220e+03, threshold=9.428e+02, percent-clipped=2.0 2023-04-27 21:32:48,059 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2626, 5.1136, 5.0277, 4.3819, 5.0484, 2.1623, 4.7376, 5.1775], device='cuda:7'), covar=tensor([0.0053, 0.0050, 0.0056, 0.0272, 0.0044, 0.1166, 0.0073, 0.0089], device='cuda:7'), in_proj_covar=tensor([0.0076, 0.0062, 0.0096, 0.0112, 0.0070, 0.0117, 0.0086, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:33:07,175 INFO [train.py:904] (7/8) Epoch 3, batch 1450, loss[loss=0.2168, simple_loss=0.2974, pruned_loss=0.06815, over 16741.00 frames. ], tot_loss[loss=0.2362, simple_loss=0.3044, pruned_loss=0.08398, over 3329118.52 frames. ], batch size: 62, lr: 2.21e-02, grad_scale: 8.0 2023-04-27 21:33:13,879 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21756.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:33:23,943 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21763.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:33:39,355 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21774.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:34:14,646 INFO [train.py:904] (7/8) Epoch 3, batch 1500, loss[loss=0.2763, simple_loss=0.3276, pruned_loss=0.1125, over 16895.00 frames. ], tot_loss[loss=0.2384, simple_loss=0.3055, pruned_loss=0.08561, over 3320424.03 frames. ], batch size: 96, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:34:43,997 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=21822.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:34:47,238 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21824.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:34:52,714 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.241e+02 3.626e+02 4.234e+02 5.507e+02 9.675e+02, threshold=8.468e+02, percent-clipped=1.0 2023-04-27 21:35:23,653 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21850.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:35:24,403 INFO [train.py:904] (7/8) Epoch 3, batch 1550, loss[loss=0.2521, simple_loss=0.3061, pruned_loss=0.09909, over 16896.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3063, pruned_loss=0.08658, over 3325174.21 frames. ], batch size: 96, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:36:31,485 INFO [train.py:904] (7/8) Epoch 3, batch 1600, loss[loss=0.2441, simple_loss=0.3026, pruned_loss=0.09284, over 16880.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3089, pruned_loss=0.08749, over 3323847.52 frames. ], batch size: 109, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:36:46,737 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21911.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:37:07,923 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.625e+02 3.946e+02 4.752e+02 6.203e+02 1.123e+03, threshold=9.504e+02, percent-clipped=2.0 2023-04-27 21:37:36,199 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6709, 4.7840, 4.7495, 4.8081, 4.6159, 5.2642, 4.9306, 4.6059], device='cuda:7'), covar=tensor([0.0890, 0.1267, 0.1092, 0.1343, 0.2736, 0.0924, 0.1167, 0.2210], device='cuda:7'), in_proj_covar=tensor([0.0224, 0.0323, 0.0296, 0.0277, 0.0365, 0.0314, 0.0255, 0.0374], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:37:38,901 INFO [train.py:904] (7/8) Epoch 3, batch 1650, loss[loss=0.2349, simple_loss=0.2931, pruned_loss=0.08832, over 16720.00 frames. ], tot_loss[loss=0.2431, simple_loss=0.3104, pruned_loss=0.08789, over 3328671.86 frames. ], batch size: 124, lr: 2.20e-02, grad_scale: 8.0 2023-04-27 21:37:39,890 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21951.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:02,256 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21968.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:03,304 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21969.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:04,758 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21970.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:15,817 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-27 21:38:24,252 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21984.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:38:50,901 INFO [train.py:904] (7/8) Epoch 3, batch 1700, loss[loss=0.2837, simple_loss=0.3381, pruned_loss=0.1147, over 16749.00 frames. ], tot_loss[loss=0.2469, simple_loss=0.3144, pruned_loss=0.08973, over 3311288.01 frames. ], batch size: 124, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:39:11,658 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22016.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:39:15,150 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-27 21:39:28,012 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.555e+02 4.242e+02 5.121e+02 6.013e+02 1.262e+03, threshold=1.024e+03, percent-clipped=2.0 2023-04-27 21:39:31,654 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22031.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:39:33,270 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22032.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:39:57,293 INFO [train.py:904] (7/8) Epoch 3, batch 1750, loss[loss=0.2068, simple_loss=0.2839, pruned_loss=0.06487, over 17212.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3157, pruned_loss=0.08939, over 3324498.63 frames. ], batch size: 43, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:41:06,921 INFO [train.py:904] (7/8) Epoch 3, batch 1800, loss[loss=0.2365, simple_loss=0.3086, pruned_loss=0.0822, over 16521.00 frames. ], tot_loss[loss=0.2481, simple_loss=0.3166, pruned_loss=0.08981, over 3317746.39 frames. ], batch size: 75, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:41:30,485 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22119.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:41:35,774 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0398, 5.6217, 5.5898, 5.4390, 5.4718, 6.0670, 5.5844, 5.3789], device='cuda:7'), covar=tensor([0.0553, 0.1096, 0.0942, 0.1262, 0.1875, 0.0670, 0.0993, 0.2033], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0338, 0.0304, 0.0286, 0.0374, 0.0324, 0.0259, 0.0386], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:41:44,206 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.299e+02 3.837e+02 4.558e+02 5.944e+02 1.243e+03, threshold=9.115e+02, percent-clipped=4.0 2023-04-27 21:41:45,919 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22130.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:41:57,987 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4166, 3.7017, 3.7547, 1.6548, 3.8313, 3.8026, 3.3565, 3.0939], device='cuda:7'), covar=tensor([0.0646, 0.0076, 0.0097, 0.1112, 0.0054, 0.0045, 0.0218, 0.0268], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0083, 0.0083, 0.0150, 0.0076, 0.0076, 0.0114, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:42:04,508 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2126, 4.1103, 1.8790, 4.2015, 2.4811, 4.1683, 1.9934, 3.0199], device='cuda:7'), covar=tensor([0.0045, 0.0175, 0.1378, 0.0046, 0.0690, 0.0282, 0.1241, 0.0515], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0135, 0.0171, 0.0082, 0.0155, 0.0163, 0.0178, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 21:42:14,103 INFO [train.py:904] (7/8) Epoch 3, batch 1850, loss[loss=0.2566, simple_loss=0.3222, pruned_loss=0.09551, over 16801.00 frames. ], tot_loss[loss=0.2471, simple_loss=0.3164, pruned_loss=0.08889, over 3328161.59 frames. ], batch size: 102, lr: 2.19e-02, grad_scale: 4.0 2023-04-27 21:42:47,502 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22176.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:42:58,827 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-27 21:43:09,329 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22191.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:43:22,353 INFO [train.py:904] (7/8) Epoch 3, batch 1900, loss[loss=0.2701, simple_loss=0.3257, pruned_loss=0.1072, over 16898.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.3158, pruned_loss=0.08847, over 3308787.70 frames. ], batch size: 109, lr: 2.18e-02, grad_scale: 4.0 2023-04-27 21:43:31,119 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22206.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:43:41,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2882, 4.5804, 4.3161, 4.4168, 3.9810, 3.9616, 4.1328, 4.5810], device='cuda:7'), covar=tensor([0.0476, 0.0565, 0.0783, 0.0364, 0.0550, 0.0947, 0.0543, 0.0634], device='cuda:7'), in_proj_covar=tensor([0.0270, 0.0379, 0.0329, 0.0233, 0.0246, 0.0227, 0.0297, 0.0258], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:43:42,701 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22215.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:43:48,027 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6375, 2.5007, 2.0968, 2.5096, 3.1788, 3.0431, 4.0320, 3.4315], device='cuda:7'), covar=tensor([0.0016, 0.0130, 0.0166, 0.0140, 0.0066, 0.0108, 0.0034, 0.0075], device='cuda:7'), in_proj_covar=tensor([0.0060, 0.0116, 0.0114, 0.0117, 0.0110, 0.0117, 0.0074, 0.0093], device='cuda:7'), out_proj_covar=tensor([8.5944e-05, 1.7267e-04, 1.6293e-04, 1.7241e-04, 1.6630e-04, 1.7690e-04, 1.1007e-04, 1.4370e-04], device='cuda:7') 2023-04-27 21:44:02,291 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.402e+02 3.901e+02 4.998e+02 5.919e+02 1.840e+03, threshold=9.997e+02, percent-clipped=6.0 2023-04-27 21:44:12,160 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:12,202 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:32,430 INFO [train.py:904] (7/8) Epoch 3, batch 1950, loss[loss=0.2199, simple_loss=0.2939, pruned_loss=0.07294, over 16773.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.3161, pruned_loss=0.08866, over 3304210.88 frames. ], batch size: 39, lr: 2.18e-02, grad_scale: 4.0 2023-04-27 21:44:32,855 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22251.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:44:44,011 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4848, 4.2890, 4.3096, 4.4029, 3.9716, 4.3140, 4.2003, 4.0419], device='cuda:7'), covar=tensor([0.0288, 0.0241, 0.0158, 0.0123, 0.0650, 0.0168, 0.0267, 0.0262], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0132, 0.0190, 0.0159, 0.0227, 0.0171, 0.0140, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:44:55,257 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1459, 2.1839, 1.8697, 2.0678, 2.7509, 2.5592, 3.5438, 3.0568], device='cuda:7'), covar=tensor([0.0024, 0.0151, 0.0172, 0.0179, 0.0085, 0.0136, 0.0046, 0.0074], device='cuda:7'), in_proj_covar=tensor([0.0062, 0.0118, 0.0115, 0.0118, 0.0111, 0.0119, 0.0076, 0.0093], device='cuda:7'), out_proj_covar=tensor([8.8900e-05, 1.7593e-04, 1.6485e-04, 1.7311e-04, 1.6784e-04, 1.8004e-04, 1.1256e-04, 1.4320e-04], device='cuda:7') 2023-04-27 21:44:58,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3294, 5.6717, 5.4213, 5.6203, 4.9001, 4.7093, 5.1190, 5.8071], device='cuda:7'), covar=tensor([0.0516, 0.0690, 0.0735, 0.0337, 0.0525, 0.0488, 0.0506, 0.0572], device='cuda:7'), in_proj_covar=tensor([0.0272, 0.0384, 0.0334, 0.0235, 0.0249, 0.0230, 0.0303, 0.0261], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:44:58,151 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22269.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:06,156 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22276.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:35,383 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22298.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:36,287 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22299.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:45:38,320 INFO [train.py:904] (7/8) Epoch 3, batch 2000, loss[loss=0.2191, simple_loss=0.3062, pruned_loss=0.06599, over 17095.00 frames. ], tot_loss[loss=0.2448, simple_loss=0.3147, pruned_loss=0.08741, over 3317297.82 frames. ], batch size: 47, lr: 2.18e-02, grad_scale: 8.0 2023-04-27 21:46:01,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22317.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:46:13,079 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22326.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:46:17,544 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.887e+02 3.938e+02 4.847e+02 6.098e+02 1.059e+03, threshold=9.693e+02, percent-clipped=1.0 2023-04-27 21:46:35,328 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6300, 3.6196, 4.0782, 4.0658, 4.0054, 3.6868, 3.7327, 3.7375], device='cuda:7'), covar=tensor([0.0267, 0.0344, 0.0367, 0.0374, 0.0421, 0.0315, 0.0644, 0.0328], device='cuda:7'), in_proj_covar=tensor([0.0189, 0.0177, 0.0199, 0.0191, 0.0234, 0.0197, 0.0298, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 21:46:46,966 INFO [train.py:904] (7/8) Epoch 3, batch 2050, loss[loss=0.2933, simple_loss=0.3415, pruned_loss=0.1225, over 16898.00 frames. ], tot_loss[loss=0.247, simple_loss=0.316, pruned_loss=0.08901, over 3318634.90 frames. ], batch size: 109, lr: 2.18e-02, grad_scale: 8.0 2023-04-27 21:47:12,749 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6162, 3.6376, 2.5998, 2.3528, 2.7262, 1.9815, 3.6426, 4.0742], device='cuda:7'), covar=tensor([0.1744, 0.0537, 0.1177, 0.1032, 0.1772, 0.1402, 0.0387, 0.0430], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0237, 0.0248, 0.0202, 0.0276, 0.0188, 0.0211, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:47:23,922 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5688, 4.8790, 4.4788, 4.7268, 4.2976, 4.2329, 4.4203, 4.8611], device='cuda:7'), covar=tensor([0.0520, 0.0641, 0.0949, 0.0360, 0.0530, 0.0780, 0.0547, 0.0647], device='cuda:7'), in_proj_covar=tensor([0.0272, 0.0387, 0.0336, 0.0234, 0.0251, 0.0234, 0.0303, 0.0264], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:47:33,580 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22385.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:47:55,918 INFO [train.py:904] (7/8) Epoch 3, batch 2100, loss[loss=0.2187, simple_loss=0.3115, pruned_loss=0.063, over 17116.00 frames. ], tot_loss[loss=0.2483, simple_loss=0.3174, pruned_loss=0.08963, over 3315428.63 frames. ], batch size: 47, lr: 2.17e-02, grad_scale: 8.0 2023-04-27 21:48:20,740 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22419.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:48:31,789 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.621e+02 3.997e+02 4.789e+02 6.119e+02 1.821e+03, threshold=9.579e+02, percent-clipped=3.0 2023-04-27 21:48:43,957 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2119, 2.1993, 1.8897, 1.9829, 2.8431, 2.5903, 3.4418, 3.1563], device='cuda:7'), covar=tensor([0.0022, 0.0138, 0.0165, 0.0151, 0.0070, 0.0104, 0.0068, 0.0053], device='cuda:7'), in_proj_covar=tensor([0.0061, 0.0117, 0.0115, 0.0116, 0.0108, 0.0115, 0.0075, 0.0093], device='cuda:7'), out_proj_covar=tensor([8.8944e-05, 1.7335e-04, 1.6374e-04, 1.7087e-04, 1.6332e-04, 1.7326e-04, 1.1172e-04, 1.4293e-04], device='cuda:7') 2023-04-27 21:48:53,873 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7947, 4.0771, 4.1671, 1.6261, 4.4315, 4.3394, 3.2589, 3.3753], device='cuda:7'), covar=tensor([0.0656, 0.0124, 0.0169, 0.1372, 0.0057, 0.0054, 0.0297, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0082, 0.0084, 0.0146, 0.0078, 0.0075, 0.0114, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:48:55,010 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22446.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:48:59,769 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6737, 3.5301, 3.5766, 3.5679, 3.4983, 4.0477, 3.8612, 3.4877], device='cuda:7'), covar=tensor([0.1732, 0.1732, 0.1492, 0.2001, 0.2774, 0.1361, 0.1178, 0.2666], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0335, 0.0303, 0.0288, 0.0370, 0.0323, 0.0263, 0.0386], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:49:00,607 INFO [train.py:904] (7/8) Epoch 3, batch 2150, loss[loss=0.3758, simple_loss=0.4042, pruned_loss=0.1738, over 12023.00 frames. ], tot_loss[loss=0.2511, simple_loss=0.3196, pruned_loss=0.09128, over 3302539.45 frames. ], batch size: 246, lr: 2.17e-02, grad_scale: 8.0 2023-04-27 21:49:22,529 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22467.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:49:47,741 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22486.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:49:51,474 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2513, 2.1559, 1.7237, 1.7169, 2.7344, 2.4206, 3.5430, 3.1136], device='cuda:7'), covar=tensor([0.0025, 0.0150, 0.0177, 0.0198, 0.0079, 0.0139, 0.0047, 0.0062], device='cuda:7'), in_proj_covar=tensor([0.0063, 0.0118, 0.0116, 0.0118, 0.0110, 0.0118, 0.0076, 0.0094], device='cuda:7'), out_proj_covar=tensor([9.1196e-05, 1.7490e-04, 1.6565e-04, 1.7436e-04, 1.6745e-04, 1.7699e-04, 1.1311e-04, 1.4423e-04], device='cuda:7') 2023-04-27 21:50:08,038 INFO [train.py:904] (7/8) Epoch 3, batch 2200, loss[loss=0.2495, simple_loss=0.3321, pruned_loss=0.08342, over 17053.00 frames. ], tot_loss[loss=0.2519, simple_loss=0.3202, pruned_loss=0.09182, over 3302187.53 frames. ], batch size: 53, lr: 2.17e-02, grad_scale: 4.0 2023-04-27 21:50:14,188 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22506.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:50:46,036 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.515e+02 4.185e+02 4.921e+02 5.938e+02 1.054e+03, threshold=9.842e+02, percent-clipped=2.0 2023-04-27 21:50:48,914 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22532.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:51:15,030 INFO [train.py:904] (7/8) Epoch 3, batch 2250, loss[loss=0.26, simple_loss=0.3206, pruned_loss=0.09973, over 16511.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3191, pruned_loss=0.09071, over 3303680.93 frames. ], batch size: 75, lr: 2.17e-02, grad_scale: 4.0 2023-04-27 21:51:18,640 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22554.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:51:42,424 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22571.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:51:49,284 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0216, 5.7466, 5.6799, 5.6251, 5.7175, 6.1497, 5.9430, 5.7302], device='cuda:7'), covar=tensor([0.0543, 0.1163, 0.1018, 0.1212, 0.1893, 0.0671, 0.0697, 0.1522], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0334, 0.0309, 0.0286, 0.0375, 0.0323, 0.0261, 0.0383], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 21:51:55,860 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-27 21:52:09,946 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22593.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:52:20,014 INFO [train.py:904] (7/8) Epoch 3, batch 2300, loss[loss=0.3429, simple_loss=0.3853, pruned_loss=0.1502, over 11569.00 frames. ], tot_loss[loss=0.2501, simple_loss=0.3191, pruned_loss=0.09054, over 3301491.48 frames. ], batch size: 247, lr: 2.17e-02, grad_scale: 4.0 2023-04-27 21:52:57,355 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22626.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:53:01,631 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.100e+02 3.931e+02 4.776e+02 5.724e+02 1.077e+03, threshold=9.553e+02, percent-clipped=1.0 2023-04-27 21:53:12,089 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22639.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:53:30,352 INFO [train.py:904] (7/8) Epoch 3, batch 2350, loss[loss=0.2511, simple_loss=0.3325, pruned_loss=0.08485, over 17021.00 frames. ], tot_loss[loss=0.2503, simple_loss=0.3194, pruned_loss=0.09057, over 3298163.82 frames. ], batch size: 53, lr: 2.16e-02, grad_scale: 4.0 2023-04-27 21:54:01,266 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22674.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:54:35,120 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22700.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 21:54:35,743 INFO [train.py:904] (7/8) Epoch 3, batch 2400, loss[loss=0.2827, simple_loss=0.3381, pruned_loss=0.1136, over 16907.00 frames. ], tot_loss[loss=0.2514, simple_loss=0.3207, pruned_loss=0.09106, over 3296928.54 frames. ], batch size: 109, lr: 2.16e-02, grad_scale: 8.0 2023-04-27 21:54:49,860 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1280, 3.5241, 2.9831, 5.1870, 5.0716, 4.5244, 2.2669, 3.3364], device='cuda:7'), covar=tensor([0.1249, 0.0470, 0.1066, 0.0065, 0.0174, 0.0298, 0.1129, 0.0660], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0131, 0.0162, 0.0076, 0.0143, 0.0137, 0.0153, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 21:55:17,393 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.811e+02 3.861e+02 4.414e+02 5.506e+02 1.152e+03, threshold=8.829e+02, percent-clipped=5.0 2023-04-27 21:55:24,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4395, 5.7717, 5.1749, 5.8546, 5.0957, 4.9374, 5.4752, 5.8027], device='cuda:7'), covar=tensor([0.1151, 0.0925, 0.1598, 0.0507, 0.0937, 0.0629, 0.0663, 0.1066], device='cuda:7'), in_proj_covar=tensor([0.0270, 0.0385, 0.0335, 0.0235, 0.0251, 0.0231, 0.0301, 0.0260], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 21:55:32,329 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22741.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:55:46,917 INFO [train.py:904] (7/8) Epoch 3, batch 2450, loss[loss=0.2308, simple_loss=0.3186, pruned_loss=0.07151, over 17058.00 frames. ], tot_loss[loss=0.2513, simple_loss=0.3213, pruned_loss=0.09067, over 3312136.68 frames. ], batch size: 55, lr: 2.16e-02, grad_scale: 8.0 2023-04-27 21:56:33,367 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22786.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:56:37,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4387, 4.2805, 1.9554, 4.3011, 2.6505, 4.3743, 2.2850, 3.0646], device='cuda:7'), covar=tensor([0.0041, 0.0145, 0.1242, 0.0040, 0.0681, 0.0312, 0.1122, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0134, 0.0168, 0.0086, 0.0158, 0.0168, 0.0178, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 21:56:46,194 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22794.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:56:55,077 INFO [train.py:904] (7/8) Epoch 3, batch 2500, loss[loss=0.2195, simple_loss=0.2951, pruned_loss=0.07189, over 16991.00 frames. ], tot_loss[loss=0.2497, simple_loss=0.3201, pruned_loss=0.08963, over 3315037.15 frames. ], batch size: 41, lr: 2.16e-02, grad_scale: 8.0 2023-04-27 21:56:59,215 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-27 21:57:27,756 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22825.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:57:33,677 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.850e+02 4.203e+02 4.842e+02 6.402e+02 1.699e+03, threshold=9.683e+02, percent-clipped=7.0 2023-04-27 21:57:37,638 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22832.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:57:42,088 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22834.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:00,345 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0769, 1.5739, 1.3645, 1.3580, 1.7467, 1.5860, 1.7410, 1.9238], device='cuda:7'), covar=tensor([0.0030, 0.0108, 0.0130, 0.0129, 0.0070, 0.0127, 0.0060, 0.0071], device='cuda:7'), in_proj_covar=tensor([0.0059, 0.0118, 0.0114, 0.0116, 0.0109, 0.0116, 0.0077, 0.0096], device='cuda:7'), out_proj_covar=tensor([8.5433e-05, 1.7440e-04, 1.6301e-04, 1.7102e-04, 1.6530e-04, 1.7431e-04, 1.1400e-04, 1.4769e-04], device='cuda:7') 2023-04-27 21:58:03,440 INFO [train.py:904] (7/8) Epoch 3, batch 2550, loss[loss=0.2125, simple_loss=0.2891, pruned_loss=0.06796, over 16973.00 frames. ], tot_loss[loss=0.2493, simple_loss=0.3204, pruned_loss=0.08907, over 3308051.03 frames. ], batch size: 41, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 21:58:11,159 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22855.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:31,422 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22871.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:44,551 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22880.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:58:52,655 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22886.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:59:01,526 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22893.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:59:13,846 INFO [train.py:904] (7/8) Epoch 3, batch 2600, loss[loss=0.2418, simple_loss=0.3129, pruned_loss=0.08535, over 16459.00 frames. ], tot_loss[loss=0.247, simple_loss=0.3183, pruned_loss=0.08786, over 3316745.81 frames. ], batch size: 146, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 21:59:39,355 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22919.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 21:59:50,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-27 21:59:53,533 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 3.680e+02 4.418e+02 5.171e+02 1.031e+03, threshold=8.837e+02, percent-clipped=1.0 2023-04-27 22:00:09,799 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=22941.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:00:22,561 INFO [train.py:904] (7/8) Epoch 3, batch 2650, loss[loss=0.2345, simple_loss=0.3056, pruned_loss=0.0817, over 16986.00 frames. ], tot_loss[loss=0.2474, simple_loss=0.319, pruned_loss=0.08786, over 3314193.86 frames. ], batch size: 41, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 22:01:03,733 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-27 22:01:22,764 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22995.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 22:01:30,284 INFO [train.py:904] (7/8) Epoch 3, batch 2700, loss[loss=0.3108, simple_loss=0.3556, pruned_loss=0.1331, over 12567.00 frames. ], tot_loss[loss=0.2465, simple_loss=0.3188, pruned_loss=0.0871, over 3318824.44 frames. ], batch size: 246, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 22:01:37,927 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6472, 4.4510, 1.7708, 4.6158, 2.6855, 4.5925, 2.5036, 3.2586], device='cuda:7'), covar=tensor([0.0031, 0.0146, 0.1533, 0.0036, 0.0734, 0.0197, 0.1133, 0.0504], device='cuda:7'), in_proj_covar=tensor([0.0090, 0.0137, 0.0173, 0.0085, 0.0161, 0.0170, 0.0179, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 22:01:48,334 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-27 22:01:58,415 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-27 22:02:00,070 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-27 22:02:09,774 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.756e+02 4.226e+02 4.976e+02 6.032e+02 3.495e+03, threshold=9.952e+02, percent-clipped=7.0 2023-04-27 22:02:24,105 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23041.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:02:37,956 INFO [train.py:904] (7/8) Epoch 3, batch 2750, loss[loss=0.2161, simple_loss=0.2959, pruned_loss=0.06816, over 16842.00 frames. ], tot_loss[loss=0.2457, simple_loss=0.3184, pruned_loss=0.08651, over 3321323.07 frames. ], batch size: 42, lr: 2.15e-02, grad_scale: 8.0 2023-04-27 22:02:40,328 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1310, 4.2273, 3.3313, 5.2833, 5.2588, 4.4536, 2.1106, 3.6457], device='cuda:7'), covar=tensor([0.1285, 0.0320, 0.0861, 0.0075, 0.0183, 0.0318, 0.1157, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0130, 0.0163, 0.0077, 0.0145, 0.0138, 0.0154, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 22:03:28,954 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23089.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:03:45,518 INFO [train.py:904] (7/8) Epoch 3, batch 2800, loss[loss=0.2486, simple_loss=0.3163, pruned_loss=0.09048, over 16857.00 frames. ], tot_loss[loss=0.2439, simple_loss=0.3175, pruned_loss=0.08516, over 3327224.97 frames. ], batch size: 96, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:03:55,488 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23108.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:04:25,544 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.361e+02 3.403e+02 4.548e+02 5.612e+02 1.011e+03, threshold=9.095e+02, percent-clipped=1.0 2023-04-27 22:04:54,744 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23150.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:04:55,609 INFO [train.py:904] (7/8) Epoch 3, batch 2850, loss[loss=0.2494, simple_loss=0.334, pruned_loss=0.0824, over 17074.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.3182, pruned_loss=0.08673, over 3320241.47 frames. ], batch size: 53, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:05:20,426 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23169.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:05:37,570 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23181.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:06:03,320 INFO [train.py:904] (7/8) Epoch 3, batch 2900, loss[loss=0.2411, simple_loss=0.3185, pruned_loss=0.08185, over 17144.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3166, pruned_loss=0.08678, over 3328988.51 frames. ], batch size: 47, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:06:25,839 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23217.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:06:42,333 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23229.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:06:43,092 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.672e+02 4.009e+02 4.959e+02 5.927e+02 1.397e+03, threshold=9.918e+02, percent-clipped=5.0 2023-04-27 22:07:12,201 INFO [train.py:904] (7/8) Epoch 3, batch 2950, loss[loss=0.2092, simple_loss=0.2887, pruned_loss=0.06491, over 16854.00 frames. ], tot_loss[loss=0.2472, simple_loss=0.3171, pruned_loss=0.08863, over 3315775.71 frames. ], batch size: 42, lr: 2.14e-02, grad_scale: 8.0 2023-04-27 22:07:26,620 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-27 22:07:32,189 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3572, 3.8718, 3.9010, 1.8111, 3.9747, 3.9355, 3.0375, 3.2648], device='cuda:7'), covar=tensor([0.0798, 0.0085, 0.0127, 0.1220, 0.0077, 0.0054, 0.0355, 0.0278], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0082, 0.0082, 0.0143, 0.0078, 0.0076, 0.0115, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 22:07:49,224 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23278.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:08:07,184 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23290.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:08:12,723 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23295.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:08:19,869 INFO [train.py:904] (7/8) Epoch 3, batch 3000, loss[loss=0.2078, simple_loss=0.2897, pruned_loss=0.06296, over 17242.00 frames. ], tot_loss[loss=0.2477, simple_loss=0.3172, pruned_loss=0.08915, over 3322785.14 frames. ], batch size: 43, lr: 2.13e-02, grad_scale: 8.0 2023-04-27 22:08:19,869 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 22:08:30,495 INFO [train.py:938] (7/8) Epoch 3, validation: loss=0.1721, simple_loss=0.279, pruned_loss=0.03262, over 944034.00 frames. 2023-04-27 22:08:30,496 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 22:08:42,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0562, 1.6111, 1.3800, 1.5435, 1.9004, 1.7056, 1.9837, 1.9245], device='cuda:7'), covar=tensor([0.0023, 0.0077, 0.0094, 0.0089, 0.0049, 0.0083, 0.0041, 0.0048], device='cuda:7'), in_proj_covar=tensor([0.0060, 0.0116, 0.0113, 0.0114, 0.0107, 0.0116, 0.0077, 0.0093], device='cuda:7'), out_proj_covar=tensor([8.7994e-05, 1.7035e-04, 1.6065e-04, 1.6659e-04, 1.6088e-04, 1.7172e-04, 1.1476e-04, 1.4183e-04], device='cuda:7') 2023-04-27 22:09:10,135 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.558e+02 3.966e+02 4.777e+02 5.754e+02 1.756e+03, threshold=9.554e+02, percent-clipped=1.0 2023-04-27 22:09:26,139 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23343.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:09:37,547 INFO [train.py:904] (7/8) Epoch 3, batch 3050, loss[loss=0.2293, simple_loss=0.3087, pruned_loss=0.07499, over 17125.00 frames. ], tot_loss[loss=0.248, simple_loss=0.3176, pruned_loss=0.08915, over 3328592.01 frames. ], batch size: 48, lr: 2.13e-02, grad_scale: 4.0 2023-04-27 22:10:44,554 INFO [train.py:904] (7/8) Epoch 3, batch 3100, loss[loss=0.2329, simple_loss=0.3121, pruned_loss=0.07684, over 17226.00 frames. ], tot_loss[loss=0.2467, simple_loss=0.3162, pruned_loss=0.08859, over 3334704.89 frames. ], batch size: 45, lr: 2.13e-02, grad_scale: 4.0 2023-04-27 22:11:28,257 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.500e+02 3.880e+02 4.394e+02 5.487e+02 1.419e+03, threshold=8.788e+02, percent-clipped=5.0 2023-04-27 22:11:53,315 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23450.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:11:54,098 INFO [train.py:904] (7/8) Epoch 3, batch 3150, loss[loss=0.2393, simple_loss=0.3261, pruned_loss=0.07621, over 17249.00 frames. ], tot_loss[loss=0.2457, simple_loss=0.3155, pruned_loss=0.08791, over 3335914.94 frames. ], batch size: 52, lr: 2.13e-02, grad_scale: 4.0 2023-04-27 22:12:12,242 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23464.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:12:21,696 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-27 22:12:35,073 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23481.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:12:44,207 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23488.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:12:58,215 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23498.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:13:01,359 INFO [train.py:904] (7/8) Epoch 3, batch 3200, loss[loss=0.2514, simple_loss=0.3257, pruned_loss=0.08853, over 16671.00 frames. ], tot_loss[loss=0.244, simple_loss=0.314, pruned_loss=0.08704, over 3328364.59 frames. ], batch size: 62, lr: 2.13e-02, grad_scale: 8.0 2023-04-27 22:13:28,324 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2026, 4.1423, 4.1124, 3.6416, 4.1931, 1.7667, 3.9393, 4.1778], device='cuda:7'), covar=tensor([0.0072, 0.0050, 0.0071, 0.0263, 0.0052, 0.1195, 0.0075, 0.0081], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0069, 0.0106, 0.0124, 0.0078, 0.0119, 0.0096, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:13:39,249 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23529.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:13:42,225 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.397e+02 3.601e+02 4.370e+02 5.311e+02 9.274e+02, threshold=8.739e+02, percent-clipped=1.0 2023-04-27 22:14:06,468 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23549.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 22:14:08,944 INFO [train.py:904] (7/8) Epoch 3, batch 3250, loss[loss=0.2579, simple_loss=0.3179, pruned_loss=0.09894, over 16791.00 frames. ], tot_loss[loss=0.2442, simple_loss=0.3137, pruned_loss=0.08737, over 3331840.61 frames. ], batch size: 124, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:14:38,935 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23573.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:14:56,671 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23585.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:14:56,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8301, 4.0008, 2.9444, 5.2829, 5.1848, 4.6043, 2.2266, 3.5665], device='cuda:7'), covar=tensor([0.1366, 0.0369, 0.1107, 0.0054, 0.0187, 0.0287, 0.1174, 0.0584], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0133, 0.0165, 0.0078, 0.0153, 0.0143, 0.0157, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 22:15:18,380 INFO [train.py:904] (7/8) Epoch 3, batch 3300, loss[loss=0.2858, simple_loss=0.3493, pruned_loss=0.1112, over 16271.00 frames. ], tot_loss[loss=0.2437, simple_loss=0.3138, pruned_loss=0.08678, over 3329415.82 frames. ], batch size: 165, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:15:30,964 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-27 22:15:57,203 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.549e+02 3.983e+02 4.910e+02 5.801e+02 1.123e+03, threshold=9.819e+02, percent-clipped=5.0 2023-04-27 22:16:24,694 INFO [train.py:904] (7/8) Epoch 3, batch 3350, loss[loss=0.2786, simple_loss=0.3372, pruned_loss=0.11, over 15518.00 frames. ], tot_loss[loss=0.2449, simple_loss=0.3151, pruned_loss=0.08733, over 3330942.04 frames. ], batch size: 190, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:17:33,464 INFO [train.py:904] (7/8) Epoch 3, batch 3400, loss[loss=0.2651, simple_loss=0.3453, pruned_loss=0.09246, over 17268.00 frames. ], tot_loss[loss=0.2442, simple_loss=0.3148, pruned_loss=0.0868, over 3328191.20 frames. ], batch size: 52, lr: 2.12e-02, grad_scale: 8.0 2023-04-27 22:18:13,370 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 3.881e+02 4.662e+02 5.457e+02 1.013e+03, threshold=9.324e+02, percent-clipped=1.0 2023-04-27 22:18:32,250 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1343, 5.0305, 5.0001, 3.9318, 4.9527, 1.9791, 4.6301, 5.0025], device='cuda:7'), covar=tensor([0.0061, 0.0068, 0.0072, 0.0435, 0.0053, 0.1329, 0.0095, 0.0117], device='cuda:7'), in_proj_covar=tensor([0.0078, 0.0069, 0.0104, 0.0121, 0.0077, 0.0116, 0.0094, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:18:40,211 INFO [train.py:904] (7/8) Epoch 3, batch 3450, loss[loss=0.1999, simple_loss=0.2754, pruned_loss=0.06217, over 16819.00 frames. ], tot_loss[loss=0.2422, simple_loss=0.3127, pruned_loss=0.08588, over 3331034.78 frames. ], batch size: 42, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:18:45,177 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-27 22:18:59,363 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23764.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:19:40,826 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9600, 4.0058, 4.4274, 4.4607, 4.3955, 3.9387, 4.1133, 4.0089], device='cuda:7'), covar=tensor([0.0275, 0.0303, 0.0282, 0.0289, 0.0407, 0.0318, 0.0666, 0.0367], device='cuda:7'), in_proj_covar=tensor([0.0204, 0.0190, 0.0207, 0.0206, 0.0255, 0.0211, 0.0315, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 22:19:47,210 INFO [train.py:904] (7/8) Epoch 3, batch 3500, loss[loss=0.2359, simple_loss=0.312, pruned_loss=0.07992, over 17119.00 frames. ], tot_loss[loss=0.2416, simple_loss=0.3123, pruned_loss=0.08549, over 3330682.09 frames. ], batch size: 48, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:20:04,595 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23812.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:20:20,034 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-27 22:20:31,311 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.876e+02 4.693e+02 5.719e+02 1.171e+03, threshold=9.385e+02, percent-clipped=5.0 2023-04-27 22:20:48,798 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23844.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 22:20:59,090 INFO [train.py:904] (7/8) Epoch 3, batch 3550, loss[loss=0.208, simple_loss=0.2814, pruned_loss=0.0673, over 16312.00 frames. ], tot_loss[loss=0.2397, simple_loss=0.3107, pruned_loss=0.0844, over 3327043.94 frames. ], batch size: 36, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:21:12,193 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-27 22:21:29,389 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23873.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:21:45,170 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23885.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:22:05,758 INFO [train.py:904] (7/8) Epoch 3, batch 3600, loss[loss=0.2492, simple_loss=0.3009, pruned_loss=0.09872, over 16858.00 frames. ], tot_loss[loss=0.2388, simple_loss=0.3096, pruned_loss=0.08406, over 3334416.48 frames. ], batch size: 116, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:22:29,313 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4330, 4.2333, 3.0239, 5.4440, 5.2448, 4.6047, 2.5509, 3.6683], device='cuda:7'), covar=tensor([0.1116, 0.0321, 0.0963, 0.0055, 0.0173, 0.0284, 0.1001, 0.0490], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0130, 0.0162, 0.0077, 0.0151, 0.0142, 0.0154, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 22:22:33,415 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23921.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:22:38,267 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1621, 1.3946, 1.9419, 2.1885, 2.2550, 2.3102, 1.5267, 2.1006], device='cuda:7'), covar=tensor([0.0049, 0.0170, 0.0098, 0.0082, 0.0044, 0.0055, 0.0144, 0.0043], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0126, 0.0114, 0.0107, 0.0088, 0.0071, 0.0114, 0.0069], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-27 22:22:47,020 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.441e+02 3.888e+02 4.838e+02 6.309e+02 1.139e+03, threshold=9.677e+02, percent-clipped=4.0 2023-04-27 22:22:50,343 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=23933.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:23:14,443 INFO [train.py:904] (7/8) Epoch 3, batch 3650, loss[loss=0.2615, simple_loss=0.3124, pruned_loss=0.1053, over 16328.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3074, pruned_loss=0.0837, over 3333003.08 frames. ], batch size: 165, lr: 2.11e-02, grad_scale: 8.0 2023-04-27 22:24:29,890 INFO [train.py:904] (7/8) Epoch 3, batch 3700, loss[loss=0.2657, simple_loss=0.3278, pruned_loss=0.1018, over 16529.00 frames. ], tot_loss[loss=0.239, simple_loss=0.3063, pruned_loss=0.08582, over 3302587.63 frames. ], batch size: 68, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:24:36,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9157, 5.5063, 5.6044, 5.4465, 5.3097, 5.9939, 5.6688, 5.2752], device='cuda:7'), covar=tensor([0.0654, 0.1097, 0.0840, 0.1663, 0.2559, 0.0812, 0.0816, 0.1795], device='cuda:7'), in_proj_covar=tensor([0.0232, 0.0322, 0.0302, 0.0286, 0.0365, 0.0316, 0.0259, 0.0379], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:25:13,796 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.162e+02 3.909e+02 4.503e+02 5.530e+02 1.035e+03, threshold=9.006e+02, percent-clipped=1.0 2023-04-27 22:25:22,658 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24037.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:25:27,629 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3220, 3.9924, 3.3220, 1.8302, 2.8328, 2.1234, 3.7327, 4.0418], device='cuda:7'), covar=tensor([0.0188, 0.0356, 0.0466, 0.1528, 0.0642, 0.0989, 0.0442, 0.0388], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0122, 0.0153, 0.0144, 0.0136, 0.0128, 0.0145, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 22:25:42,971 INFO [train.py:904] (7/8) Epoch 3, batch 3750, loss[loss=0.2213, simple_loss=0.2851, pruned_loss=0.07878, over 16872.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.3069, pruned_loss=0.08774, over 3287006.22 frames. ], batch size: 90, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:26:47,021 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24098.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:26:50,666 INFO [train.py:904] (7/8) Epoch 3, batch 3800, loss[loss=0.2622, simple_loss=0.3159, pruned_loss=0.1043, over 16860.00 frames. ], tot_loss[loss=0.2427, simple_loss=0.3076, pruned_loss=0.0889, over 3290307.25 frames. ], batch size: 116, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:27:34,022 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.430e+02 3.668e+02 4.510e+02 5.659e+02 1.360e+03, threshold=9.019e+02, percent-clipped=5.0 2023-04-27 22:27:52,793 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24144.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:28:01,893 INFO [train.py:904] (7/8) Epoch 3, batch 3850, loss[loss=0.2292, simple_loss=0.2921, pruned_loss=0.08314, over 16312.00 frames. ], tot_loss[loss=0.2438, simple_loss=0.3081, pruned_loss=0.0897, over 3278111.70 frames. ], batch size: 165, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:28:09,560 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7712, 2.4837, 2.6676, 4.2014, 1.9707, 3.7960, 2.3324, 2.4510], device='cuda:7'), covar=tensor([0.0294, 0.0734, 0.0395, 0.0168, 0.1686, 0.0245, 0.0955, 0.1130], device='cuda:7'), in_proj_covar=tensor([0.0248, 0.0224, 0.0187, 0.0248, 0.0294, 0.0199, 0.0215, 0.0283], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 22:28:37,194 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 22:28:37,433 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-27 22:28:44,152 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-27 22:28:54,179 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2908, 4.2483, 4.7735, 4.7512, 4.7238, 4.1783, 4.2323, 4.2995], device='cuda:7'), covar=tensor([0.0203, 0.0273, 0.0249, 0.0287, 0.0335, 0.0279, 0.0677, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0181, 0.0194, 0.0195, 0.0236, 0.0199, 0.0296, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 22:29:01,358 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=24192.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:29:13,635 INFO [train.py:904] (7/8) Epoch 3, batch 3900, loss[loss=0.2243, simple_loss=0.2968, pruned_loss=0.07589, over 16632.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3064, pruned_loss=0.08877, over 3284584.66 frames. ], batch size: 62, lr: 2.10e-02, grad_scale: 8.0 2023-04-27 22:29:56,952 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.402e+02 3.854e+02 4.579e+02 5.596e+02 1.788e+03, threshold=9.157e+02, percent-clipped=5.0 2023-04-27 22:30:25,207 INFO [train.py:904] (7/8) Epoch 3, batch 3950, loss[loss=0.2228, simple_loss=0.2857, pruned_loss=0.08001, over 16794.00 frames. ], tot_loss[loss=0.2419, simple_loss=0.3057, pruned_loss=0.08901, over 3290964.14 frames. ], batch size: 83, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:31:34,861 INFO [train.py:904] (7/8) Epoch 3, batch 4000, loss[loss=0.2527, simple_loss=0.3164, pruned_loss=0.0945, over 16811.00 frames. ], tot_loss[loss=0.241, simple_loss=0.305, pruned_loss=0.08844, over 3299129.45 frames. ], batch size: 124, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:32:17,085 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.388e+02 3.319e+02 4.445e+02 5.556e+02 1.324e+03, threshold=8.889e+02, percent-clipped=2.0 2023-04-27 22:32:45,212 INFO [train.py:904] (7/8) Epoch 3, batch 4050, loss[loss=0.2311, simple_loss=0.3, pruned_loss=0.08105, over 16582.00 frames. ], tot_loss[loss=0.2377, simple_loss=0.3035, pruned_loss=0.08596, over 3291388.84 frames. ], batch size: 62, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:33:46,363 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24393.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:33:50,961 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7863, 4.8620, 5.3877, 5.4245, 5.3447, 4.7933, 4.8615, 4.7118], device='cuda:7'), covar=tensor([0.0174, 0.0219, 0.0153, 0.0212, 0.0250, 0.0162, 0.0514, 0.0203], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0182, 0.0197, 0.0189, 0.0232, 0.0196, 0.0297, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-27 22:33:58,259 INFO [train.py:904] (7/8) Epoch 3, batch 4100, loss[loss=0.2539, simple_loss=0.328, pruned_loss=0.08989, over 16708.00 frames. ], tot_loss[loss=0.2368, simple_loss=0.304, pruned_loss=0.08481, over 3279379.69 frames. ], batch size: 124, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:34:42,997 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.218e+02 2.972e+02 3.823e+02 4.924e+02 9.607e+02, threshold=7.646e+02, percent-clipped=2.0 2023-04-27 22:35:09,922 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3782, 5.5937, 5.3042, 5.4859, 5.0921, 4.5855, 5.2300, 5.7683], device='cuda:7'), covar=tensor([0.0426, 0.0530, 0.0744, 0.0355, 0.0491, 0.0452, 0.0438, 0.0396], device='cuda:7'), in_proj_covar=tensor([0.0263, 0.0370, 0.0321, 0.0232, 0.0242, 0.0232, 0.0295, 0.0262], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 22:35:13,095 INFO [train.py:904] (7/8) Epoch 3, batch 4150, loss[loss=0.2689, simple_loss=0.3492, pruned_loss=0.09434, over 15391.00 frames. ], tot_loss[loss=0.246, simple_loss=0.3134, pruned_loss=0.08929, over 3237671.56 frames. ], batch size: 190, lr: 2.09e-02, grad_scale: 8.0 2023-04-27 22:35:46,714 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-27 22:36:27,613 INFO [train.py:904] (7/8) Epoch 3, batch 4200, loss[loss=0.2441, simple_loss=0.323, pruned_loss=0.08261, over 16812.00 frames. ], tot_loss[loss=0.2535, simple_loss=0.3219, pruned_loss=0.09249, over 3198824.73 frames. ], batch size: 39, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:37:10,951 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.707e+02 3.802e+02 4.415e+02 5.397e+02 1.092e+03, threshold=8.829e+02, percent-clipped=9.0 2023-04-27 22:37:14,874 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-27 22:37:21,723 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8252, 5.5392, 5.5010, 5.4225, 5.4299, 6.0154, 5.7550, 5.5565], device='cuda:7'), covar=tensor([0.0557, 0.0978, 0.0797, 0.1606, 0.2144, 0.0868, 0.1010, 0.2176], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0298, 0.0276, 0.0265, 0.0341, 0.0297, 0.0245, 0.0354], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:37:35,699 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24547.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:37:40,132 INFO [train.py:904] (7/8) Epoch 3, batch 4250, loss[loss=0.2262, simple_loss=0.3157, pruned_loss=0.06833, over 16652.00 frames. ], tot_loss[loss=0.2549, simple_loss=0.3247, pruned_loss=0.09251, over 3175379.72 frames. ], batch size: 62, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:38:07,380 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24569.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:38:45,691 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7154, 2.9189, 2.4892, 4.1235, 1.8567, 4.0358, 2.4724, 2.6347], device='cuda:7'), covar=tensor([0.0307, 0.0639, 0.0435, 0.0184, 0.1792, 0.0185, 0.0800, 0.1219], device='cuda:7'), in_proj_covar=tensor([0.0252, 0.0229, 0.0193, 0.0249, 0.0301, 0.0202, 0.0219, 0.0292], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 22:38:53,675 INFO [train.py:904] (7/8) Epoch 3, batch 4300, loss[loss=0.295, simple_loss=0.3642, pruned_loss=0.1129, over 15323.00 frames. ], tot_loss[loss=0.2533, simple_loss=0.3255, pruned_loss=0.09052, over 3202393.86 frames. ], batch size: 190, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:39:05,078 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24608.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:39:33,588 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-27 22:39:36,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5970, 4.9287, 4.7358, 3.3906, 4.5226, 4.6602, 4.8160, 2.3000], device='cuda:7'), covar=tensor([0.0337, 0.0007, 0.0016, 0.0187, 0.0019, 0.0027, 0.0008, 0.0296], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0049, 0.0056, 0.0108, 0.0052, 0.0060, 0.0055, 0.0102], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:39:37,575 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24630.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:39:38,219 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.269e+02 3.576e+02 4.392e+02 5.357e+02 9.860e+02, threshold=8.785e+02, percent-clipped=3.0 2023-04-27 22:40:06,096 INFO [train.py:904] (7/8) Epoch 3, batch 4350, loss[loss=0.2401, simple_loss=0.3212, pruned_loss=0.07949, over 16772.00 frames. ], tot_loss[loss=0.257, simple_loss=0.3299, pruned_loss=0.09206, over 3213271.83 frames. ], batch size: 83, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:40:59,098 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4101, 4.0334, 3.6744, 1.6956, 3.2805, 2.0561, 3.9297, 3.8877], device='cuda:7'), covar=tensor([0.0163, 0.0325, 0.0359, 0.1623, 0.0558, 0.0938, 0.0415, 0.0476], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0117, 0.0156, 0.0146, 0.0138, 0.0131, 0.0146, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 22:41:11,332 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24693.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:41:21,570 INFO [train.py:904] (7/8) Epoch 3, batch 4400, loss[loss=0.2523, simple_loss=0.3309, pruned_loss=0.08687, over 16916.00 frames. ], tot_loss[loss=0.2597, simple_loss=0.3326, pruned_loss=0.09337, over 3195203.73 frames. ], batch size: 109, lr: 2.08e-02, grad_scale: 8.0 2023-04-27 22:42:02,399 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-27 22:42:05,392 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.377e+02 3.698e+02 4.447e+02 5.386e+02 9.920e+02, threshold=8.895e+02, percent-clipped=4.0 2023-04-27 22:42:21,153 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=24741.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:42:34,123 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9568, 2.3657, 2.0131, 2.5147, 3.2279, 2.8626, 3.9576, 3.5492], device='cuda:7'), covar=tensor([0.0007, 0.0142, 0.0177, 0.0134, 0.0060, 0.0109, 0.0020, 0.0044], device='cuda:7'), in_proj_covar=tensor([0.0053, 0.0116, 0.0119, 0.0116, 0.0109, 0.0118, 0.0074, 0.0094], device='cuda:7'), out_proj_covar=tensor([7.7046e-05, 1.6803e-04, 1.6774e-04, 1.6762e-04, 1.6128e-04, 1.7320e-04, 1.0805e-04, 1.4249e-04], device='cuda:7') 2023-04-27 22:42:34,836 INFO [train.py:904] (7/8) Epoch 3, batch 4450, loss[loss=0.2458, simple_loss=0.3184, pruned_loss=0.08655, over 16443.00 frames. ], tot_loss[loss=0.2618, simple_loss=0.3358, pruned_loss=0.09394, over 3204202.00 frames. ], batch size: 35, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:42:55,268 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24765.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:43:10,989 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0311, 3.5471, 3.6776, 1.5068, 3.9715, 3.9417, 2.9083, 2.8755], device='cuda:7'), covar=tensor([0.0832, 0.0116, 0.0155, 0.1245, 0.0037, 0.0035, 0.0309, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0080, 0.0079, 0.0143, 0.0072, 0.0071, 0.0112, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 22:43:47,147 INFO [train.py:904] (7/8) Epoch 3, batch 4500, loss[loss=0.2711, simple_loss=0.3332, pruned_loss=0.1045, over 11705.00 frames. ], tot_loss[loss=0.2601, simple_loss=0.3346, pruned_loss=0.09285, over 3207973.62 frames. ], batch size: 247, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:44:22,485 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24826.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:44:29,091 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.148e+02 3.848e+02 4.477e+02 7.288e+02, threshold=7.696e+02, percent-clipped=0.0 2023-04-27 22:44:52,856 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24848.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:44:56,548 INFO [train.py:904] (7/8) Epoch 3, batch 4550, loss[loss=0.2638, simple_loss=0.3422, pruned_loss=0.09275, over 16649.00 frames. ], tot_loss[loss=0.2601, simple_loss=0.3346, pruned_loss=0.09282, over 3215103.61 frames. ], batch size: 75, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:46:07,325 INFO [train.py:904] (7/8) Epoch 3, batch 4600, loss[loss=0.2671, simple_loss=0.3429, pruned_loss=0.09562, over 16914.00 frames. ], tot_loss[loss=0.2594, simple_loss=0.3345, pruned_loss=0.0921, over 3223315.94 frames. ], batch size: 116, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:46:10,658 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24903.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:46:21,343 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24909.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:46:44,275 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24925.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:46:52,472 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.022e+02 3.683e+02 4.480e+02 7.885e+02, threshold=7.365e+02, percent-clipped=1.0 2023-04-27 22:47:20,346 INFO [train.py:904] (7/8) Epoch 3, batch 4650, loss[loss=0.2521, simple_loss=0.3192, pruned_loss=0.09246, over 17214.00 frames. ], tot_loss[loss=0.2579, simple_loss=0.3331, pruned_loss=0.09137, over 3218865.25 frames. ], batch size: 45, lr: 2.07e-02, grad_scale: 8.0 2023-04-27 22:47:29,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3627, 4.3252, 4.0983, 2.7726, 3.5579, 4.0422, 3.9518, 2.2527], device='cuda:7'), covar=tensor([0.0300, 0.0010, 0.0017, 0.0225, 0.0039, 0.0041, 0.0014, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0049, 0.0056, 0.0108, 0.0051, 0.0060, 0.0056, 0.0100], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:47:43,053 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3848, 4.2519, 4.0406, 4.1677, 3.7029, 4.2951, 4.1100, 3.8868], device='cuda:7'), covar=tensor([0.0241, 0.0145, 0.0194, 0.0125, 0.0638, 0.0151, 0.0307, 0.0247], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0115, 0.0164, 0.0135, 0.0193, 0.0146, 0.0122, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 22:48:24,073 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24995.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:48:32,758 INFO [train.py:904] (7/8) Epoch 3, batch 4700, loss[loss=0.225, simple_loss=0.2974, pruned_loss=0.07634, over 16561.00 frames. ], tot_loss[loss=0.2542, simple_loss=0.3292, pruned_loss=0.08962, over 3213709.68 frames. ], batch size: 35, lr: 2.06e-02, grad_scale: 4.0 2023-04-27 22:49:17,309 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.399e+02 3.338e+02 4.097e+02 4.784e+02 1.007e+03, threshold=8.194e+02, percent-clipped=3.0 2023-04-27 22:49:45,111 INFO [train.py:904] (7/8) Epoch 3, batch 4750, loss[loss=0.2086, simple_loss=0.2893, pruned_loss=0.06396, over 16628.00 frames. ], tot_loss[loss=0.2518, simple_loss=0.3266, pruned_loss=0.08852, over 3197270.73 frames. ], batch size: 68, lr: 2.06e-02, grad_scale: 4.0 2023-04-27 22:49:52,680 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25056.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 22:50:58,611 INFO [train.py:904] (7/8) Epoch 3, batch 4800, loss[loss=0.2452, simple_loss=0.3248, pruned_loss=0.08282, over 16730.00 frames. ], tot_loss[loss=0.2469, simple_loss=0.3219, pruned_loss=0.08602, over 3198279.61 frames. ], batch size: 134, lr: 2.06e-02, grad_scale: 8.0 2023-04-27 22:51:28,796 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25121.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:51:47,436 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.867e+02 2.955e+02 3.547e+02 4.621e+02 1.014e+03, threshold=7.094e+02, percent-clipped=1.0 2023-04-27 22:51:59,450 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2416, 4.1062, 4.0098, 2.7680, 3.7523, 3.9266, 3.8405, 2.0677], device='cuda:7'), covar=tensor([0.0329, 0.0016, 0.0019, 0.0209, 0.0031, 0.0068, 0.0030, 0.0330], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0050, 0.0059, 0.0109, 0.0053, 0.0062, 0.0058, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:52:13,128 INFO [train.py:904] (7/8) Epoch 3, batch 4850, loss[loss=0.2402, simple_loss=0.3184, pruned_loss=0.08107, over 16667.00 frames. ], tot_loss[loss=0.2486, simple_loss=0.3243, pruned_loss=0.08648, over 3173173.07 frames. ], batch size: 76, lr: 2.06e-02, grad_scale: 2.0 2023-04-27 22:53:25,043 INFO [train.py:904] (7/8) Epoch 3, batch 4900, loss[loss=0.2651, simple_loss=0.344, pruned_loss=0.09308, over 16878.00 frames. ], tot_loss[loss=0.2489, simple_loss=0.3248, pruned_loss=0.08648, over 3138092.17 frames. ], batch size: 116, lr: 2.06e-02, grad_scale: 2.0 2023-04-27 22:53:28,506 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25203.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:53:29,456 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25204.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:53:59,388 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25225.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:54:10,580 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.158e+02 3.416e+02 4.003e+02 5.207e+02 1.200e+03, threshold=8.007e+02, percent-clipped=10.0 2023-04-27 22:54:35,319 INFO [train.py:904] (7/8) Epoch 3, batch 4950, loss[loss=0.2807, simple_loss=0.3571, pruned_loss=0.1022, over 16397.00 frames. ], tot_loss[loss=0.2491, simple_loss=0.3252, pruned_loss=0.08646, over 3156973.51 frames. ], batch size: 146, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:54:36,194 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25251.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:55:08,077 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25273.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:55:22,660 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.50 vs. limit=5.0 2023-04-27 22:55:44,469 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6133, 3.7707, 3.4953, 3.6135, 2.7676, 3.6554, 3.5148, 3.3489], device='cuda:7'), covar=tensor([0.0492, 0.0250, 0.0348, 0.0226, 0.1303, 0.0319, 0.0854, 0.0374], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0122, 0.0173, 0.0142, 0.0199, 0.0157, 0.0126, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:55:48,346 INFO [train.py:904] (7/8) Epoch 3, batch 5000, loss[loss=0.2322, simple_loss=0.3208, pruned_loss=0.07184, over 16784.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3268, pruned_loss=0.08674, over 3164609.75 frames. ], batch size: 83, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:56:35,251 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.358e+02 3.469e+02 4.106e+02 5.107e+02 1.084e+03, threshold=8.211e+02, percent-clipped=3.0 2023-04-27 22:56:59,654 INFO [train.py:904] (7/8) Epoch 3, batch 5050, loss[loss=0.2392, simple_loss=0.3162, pruned_loss=0.08104, over 16639.00 frames. ], tot_loss[loss=0.2488, simple_loss=0.3263, pruned_loss=0.08571, over 3171278.33 frames. ], batch size: 62, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:56:59,967 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25351.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 22:58:08,638 INFO [train.py:904] (7/8) Epoch 3, batch 5100, loss[loss=0.2101, simple_loss=0.2975, pruned_loss=0.06135, over 16811.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3231, pruned_loss=0.08357, over 3187217.75 frames. ], batch size: 83, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:58:34,281 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3296, 3.0076, 2.3548, 2.3025, 2.2585, 1.9659, 3.0067, 3.1743], device='cuda:7'), covar=tensor([0.1652, 0.0600, 0.1066, 0.0964, 0.1654, 0.1354, 0.0389, 0.0333], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0235, 0.0247, 0.0204, 0.0282, 0.0184, 0.0211, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 22:58:38,793 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25421.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 22:58:48,446 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25428.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 22:58:57,534 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 3.320e+02 3.999e+02 5.208e+02 7.552e+02, threshold=7.999e+02, percent-clipped=0.0 2023-04-27 22:59:23,215 INFO [train.py:904] (7/8) Epoch 3, batch 5150, loss[loss=0.2421, simple_loss=0.3152, pruned_loss=0.08447, over 17025.00 frames. ], tot_loss[loss=0.2444, simple_loss=0.3232, pruned_loss=0.08281, over 3193180.53 frames. ], batch size: 55, lr: 2.05e-02, grad_scale: 2.0 2023-04-27 22:59:48,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4501, 3.3684, 3.3831, 2.8511, 3.3495, 1.9094, 3.2068, 3.1933], device='cuda:7'), covar=tensor([0.0082, 0.0071, 0.0072, 0.0269, 0.0065, 0.1217, 0.0095, 0.0118], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0060, 0.0092, 0.0108, 0.0069, 0.0114, 0.0081, 0.0092], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 22:59:50,322 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25469.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:00:20,042 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25489.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:00:35,981 INFO [train.py:904] (7/8) Epoch 3, batch 5200, loss[loss=0.2307, simple_loss=0.3062, pruned_loss=0.07758, over 16783.00 frames. ], tot_loss[loss=0.2442, simple_loss=0.3225, pruned_loss=0.08299, over 3193107.93 frames. ], batch size: 124, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:00:40,364 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25504.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:01:22,146 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.363e+02 3.525e+02 4.137e+02 5.123e+02 8.624e+02, threshold=8.275e+02, percent-clipped=4.0 2023-04-27 23:01:45,801 INFO [train.py:904] (7/8) Epoch 3, batch 5250, loss[loss=0.2215, simple_loss=0.3008, pruned_loss=0.07105, over 16464.00 frames. ], tot_loss[loss=0.2409, simple_loss=0.3184, pruned_loss=0.08173, over 3215544.84 frames. ], batch size: 68, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:01:47,893 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25552.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:02:28,381 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6400, 4.4372, 3.7975, 1.8261, 3.2472, 2.4767, 3.8477, 4.1925], device='cuda:7'), covar=tensor([0.0181, 0.0379, 0.0443, 0.1617, 0.0634, 0.0893, 0.0528, 0.0518], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0112, 0.0157, 0.0146, 0.0140, 0.0130, 0.0145, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 23:02:56,109 INFO [train.py:904] (7/8) Epoch 3, batch 5300, loss[loss=0.2033, simple_loss=0.2842, pruned_loss=0.06116, over 16743.00 frames. ], tot_loss[loss=0.2368, simple_loss=0.3142, pruned_loss=0.07971, over 3220677.66 frames. ], batch size: 124, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:03:14,654 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-27 23:03:21,489 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5121, 4.4529, 2.0031, 4.7297, 3.0495, 4.5612, 2.2106, 3.3189], device='cuda:7'), covar=tensor([0.0038, 0.0125, 0.1703, 0.0022, 0.0635, 0.0179, 0.1391, 0.0558], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0130, 0.0172, 0.0080, 0.0162, 0.0159, 0.0179, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 23:03:43,234 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.802e+02 3.089e+02 3.787e+02 4.480e+02 7.662e+02, threshold=7.574e+02, percent-clipped=0.0 2023-04-27 23:03:46,949 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5005, 2.5195, 2.2186, 3.9580, 1.8933, 3.6215, 2.2152, 2.2065], device='cuda:7'), covar=tensor([0.0334, 0.0777, 0.0524, 0.0180, 0.1739, 0.0256, 0.1037, 0.1346], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0234, 0.0196, 0.0256, 0.0307, 0.0208, 0.0226, 0.0301], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:03:50,322 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4585, 4.5774, 4.6578, 4.5634, 4.5392, 5.1294, 4.8350, 4.5524], device='cuda:7'), covar=tensor([0.0794, 0.1280, 0.0867, 0.1487, 0.1904, 0.0678, 0.0821, 0.1833], device='cuda:7'), in_proj_covar=tensor([0.0220, 0.0297, 0.0272, 0.0263, 0.0341, 0.0296, 0.0239, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:04:08,033 INFO [train.py:904] (7/8) Epoch 3, batch 5350, loss[loss=0.2695, simple_loss=0.3403, pruned_loss=0.09937, over 16235.00 frames. ], tot_loss[loss=0.2352, simple_loss=0.3126, pruned_loss=0.0789, over 3225123.78 frames. ], batch size: 165, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:04:08,441 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25651.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:04:54,970 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-27 23:05:17,181 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=25699.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:05:19,289 INFO [train.py:904] (7/8) Epoch 3, batch 5400, loss[loss=0.2813, simple_loss=0.3529, pruned_loss=0.1048, over 12330.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3159, pruned_loss=0.08032, over 3209136.53 frames. ], batch size: 250, lr: 2.04e-02, grad_scale: 4.0 2023-04-27 23:06:07,789 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 3.733e+02 4.548e+02 5.512e+02 9.876e+02, threshold=9.097e+02, percent-clipped=3.0 2023-04-27 23:06:13,348 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8727, 4.6628, 4.6686, 4.7194, 4.1237, 4.7236, 4.6350, 4.3900], device='cuda:7'), covar=tensor([0.0310, 0.0262, 0.0143, 0.0109, 0.0771, 0.0220, 0.0173, 0.0285], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0133, 0.0178, 0.0147, 0.0208, 0.0169, 0.0130, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:06:34,322 INFO [train.py:904] (7/8) Epoch 3, batch 5450, loss[loss=0.2573, simple_loss=0.3288, pruned_loss=0.09287, over 17207.00 frames. ], tot_loss[loss=0.2427, simple_loss=0.3198, pruned_loss=0.08282, over 3204210.70 frames. ], batch size: 44, lr: 2.03e-02, grad_scale: 4.0 2023-04-27 23:07:11,616 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-27 23:07:24,789 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25784.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:07:36,780 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-27 23:07:40,493 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7632, 3.4987, 3.5431, 1.2880, 3.8050, 3.9028, 2.9493, 2.5720], device='cuda:7'), covar=tensor([0.1151, 0.0110, 0.0201, 0.1459, 0.0092, 0.0044, 0.0283, 0.0556], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0081, 0.0079, 0.0146, 0.0075, 0.0073, 0.0112, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 23:07:49,291 INFO [train.py:904] (7/8) Epoch 3, batch 5500, loss[loss=0.32, simple_loss=0.3772, pruned_loss=0.1314, over 16809.00 frames. ], tot_loss[loss=0.2571, simple_loss=0.3305, pruned_loss=0.09183, over 3159863.41 frames. ], batch size: 39, lr: 2.03e-02, grad_scale: 4.0 2023-04-27 23:07:54,963 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-04-27 23:08:39,228 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.596e+02 5.216e+02 6.145e+02 8.695e+02 2.860e+03, threshold=1.229e+03, percent-clipped=22.0 2023-04-27 23:09:06,207 INFO [train.py:904] (7/8) Epoch 3, batch 5550, loss[loss=0.3803, simple_loss=0.4179, pruned_loss=0.1714, over 16381.00 frames. ], tot_loss[loss=0.2709, simple_loss=0.3408, pruned_loss=0.1004, over 3143473.47 frames. ], batch size: 146, lr: 2.03e-02, grad_scale: 4.0 2023-04-27 23:09:18,176 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25858.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:09:52,761 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-27 23:10:25,196 INFO [train.py:904] (7/8) Epoch 3, batch 5600, loss[loss=0.388, simple_loss=0.4152, pruned_loss=0.1804, over 11429.00 frames. ], tot_loss[loss=0.2809, simple_loss=0.348, pruned_loss=0.1069, over 3098820.57 frames. ], batch size: 248, lr: 2.03e-02, grad_scale: 8.0 2023-04-27 23:10:56,779 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25919.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 23:11:06,715 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-27 23:11:21,473 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.427e+02 5.585e+02 6.873e+02 8.559e+02 2.132e+03, threshold=1.375e+03, percent-clipped=5.0 2023-04-27 23:11:48,522 INFO [train.py:904] (7/8) Epoch 3, batch 5650, loss[loss=0.4161, simple_loss=0.4325, pruned_loss=0.1999, over 10997.00 frames. ], tot_loss[loss=0.2901, simple_loss=0.3545, pruned_loss=0.1129, over 3070873.08 frames. ], batch size: 247, lr: 2.03e-02, grad_scale: 8.0 2023-04-27 23:12:33,889 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1618, 3.0289, 2.7734, 2.0612, 2.6681, 2.1864, 2.8030, 3.0296], device='cuda:7'), covar=tensor([0.0286, 0.0384, 0.0345, 0.1144, 0.0559, 0.0808, 0.0474, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0114, 0.0155, 0.0143, 0.0137, 0.0132, 0.0144, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 23:13:02,676 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0385, 4.1287, 4.1028, 4.2137, 4.1825, 4.6434, 4.4394, 4.1858], device='cuda:7'), covar=tensor([0.1175, 0.1418, 0.1229, 0.1617, 0.2149, 0.0873, 0.0882, 0.2137], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0310, 0.0291, 0.0277, 0.0355, 0.0312, 0.0255, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:13:10,043 INFO [train.py:904] (7/8) Epoch 3, batch 5700, loss[loss=0.33, simple_loss=0.3787, pruned_loss=0.1406, over 11462.00 frames. ], tot_loss[loss=0.2948, simple_loss=0.3573, pruned_loss=0.1162, over 3028656.72 frames. ], batch size: 248, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:13:21,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3725, 4.1050, 4.0616, 1.8950, 4.2999, 4.3590, 3.1355, 3.2780], device='cuda:7'), covar=tensor([0.0862, 0.0092, 0.0182, 0.1250, 0.0050, 0.0030, 0.0293, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0081, 0.0078, 0.0146, 0.0075, 0.0071, 0.0114, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-27 23:14:00,538 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.120e+02 5.064e+02 6.111e+02 7.661e+02 1.195e+03, threshold=1.222e+03, percent-clipped=0.0 2023-04-27 23:14:27,448 INFO [train.py:904] (7/8) Epoch 3, batch 5750, loss[loss=0.3603, simple_loss=0.3913, pruned_loss=0.1646, over 11305.00 frames. ], tot_loss[loss=0.2986, simple_loss=0.3607, pruned_loss=0.1182, over 3003434.09 frames. ], batch size: 248, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:15:22,014 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26084.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:15:47,186 INFO [train.py:904] (7/8) Epoch 3, batch 5800, loss[loss=0.2978, simple_loss=0.3673, pruned_loss=0.1141, over 16909.00 frames. ], tot_loss[loss=0.2959, simple_loss=0.3599, pruned_loss=0.1159, over 3015910.65 frames. ], batch size: 116, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:15:49,066 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26102.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:16:08,032 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-27 23:16:14,428 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0895, 3.8701, 3.3325, 1.8169, 2.7915, 2.2245, 3.3905, 3.8005], device='cuda:7'), covar=tensor([0.0262, 0.0370, 0.0478, 0.1601, 0.0682, 0.0913, 0.0625, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0115, 0.0156, 0.0144, 0.0138, 0.0131, 0.0145, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-27 23:16:36,166 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26132.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:16:38,591 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.751e+02 5.049e+02 6.109e+02 8.114e+02 1.629e+03, threshold=1.222e+03, percent-clipped=2.0 2023-04-27 23:16:41,690 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2459, 3.2414, 1.6255, 3.2790, 2.2773, 3.2163, 1.7462, 2.4551], device='cuda:7'), covar=tensor([0.0059, 0.0185, 0.1365, 0.0042, 0.0609, 0.0330, 0.1244, 0.0522], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0131, 0.0170, 0.0078, 0.0159, 0.0161, 0.0180, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-27 23:17:05,089 INFO [train.py:904] (7/8) Epoch 3, batch 5850, loss[loss=0.2951, simple_loss=0.3623, pruned_loss=0.114, over 16911.00 frames. ], tot_loss[loss=0.2921, simple_loss=0.3574, pruned_loss=0.1134, over 3038088.90 frames. ], batch size: 116, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:17:09,961 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8577, 4.7969, 4.6353, 4.0285, 4.6377, 2.0260, 4.3621, 4.6974], device='cuda:7'), covar=tensor([0.0043, 0.0047, 0.0057, 0.0257, 0.0047, 0.1263, 0.0063, 0.0077], device='cuda:7'), in_proj_covar=tensor([0.0070, 0.0060, 0.0093, 0.0108, 0.0069, 0.0118, 0.0080, 0.0091], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:17:24,128 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26163.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:17:31,414 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26168.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:18:26,966 INFO [train.py:904] (7/8) Epoch 3, batch 5900, loss[loss=0.2612, simple_loss=0.3364, pruned_loss=0.09301, over 16661.00 frames. ], tot_loss[loss=0.2904, simple_loss=0.356, pruned_loss=0.1124, over 3045377.68 frames. ], batch size: 134, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:18:51,897 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26214.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:19:15,336 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26229.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:19:22,629 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.765e+02 4.492e+02 5.249e+02 6.969e+02 1.603e+03, threshold=1.050e+03, percent-clipped=2.0 2023-04-27 23:19:44,166 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6971, 3.5729, 3.6959, 3.9508, 3.9344, 3.5533, 3.8964, 3.9480], device='cuda:7'), covar=tensor([0.0679, 0.0553, 0.0940, 0.0381, 0.0401, 0.0922, 0.0479, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0336, 0.0442, 0.0334, 0.0251, 0.0234, 0.0275, 0.0275], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:19:47,900 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26250.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:19:48,575 INFO [train.py:904] (7/8) Epoch 3, batch 5950, loss[loss=0.3139, simple_loss=0.364, pruned_loss=0.1319, over 11640.00 frames. ], tot_loss[loss=0.2893, simple_loss=0.3567, pruned_loss=0.111, over 3043676.05 frames. ], batch size: 248, lr: 2.02e-02, grad_scale: 8.0 2023-04-27 23:21:07,936 INFO [train.py:904] (7/8) Epoch 3, batch 6000, loss[loss=0.3297, simple_loss=0.3696, pruned_loss=0.1449, over 11421.00 frames. ], tot_loss[loss=0.2887, simple_loss=0.3557, pruned_loss=0.1109, over 3037892.63 frames. ], batch size: 248, lr: 2.01e-02, grad_scale: 8.0 2023-04-27 23:21:07,936 INFO [train.py:929] (7/8) Computing validation loss 2023-04-27 23:21:18,889 INFO [train.py:938] (7/8) Epoch 3, validation: loss=0.2097, simple_loss=0.3184, pruned_loss=0.05055, over 944034.00 frames. 2023-04-27 23:21:18,890 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-27 23:21:34,629 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26311.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:22:07,343 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.939e+02 4.548e+02 5.519e+02 7.491e+02 1.813e+03, threshold=1.104e+03, percent-clipped=4.0 2023-04-27 23:22:36,223 INFO [train.py:904] (7/8) Epoch 3, batch 6050, loss[loss=0.2449, simple_loss=0.3406, pruned_loss=0.0746, over 16878.00 frames. ], tot_loss[loss=0.2841, simple_loss=0.3525, pruned_loss=0.1079, over 3082719.00 frames. ], batch size: 102, lr: 2.01e-02, grad_scale: 8.0 2023-04-27 23:22:51,797 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-27 23:23:51,482 INFO [train.py:904] (7/8) Epoch 3, batch 6100, loss[loss=0.2423, simple_loss=0.3244, pruned_loss=0.08014, over 16508.00 frames. ], tot_loss[loss=0.2828, simple_loss=0.3521, pruned_loss=0.1068, over 3086761.32 frames. ], batch size: 68, lr: 2.01e-02, grad_scale: 8.0 2023-04-27 23:23:53,706 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8099, 5.1921, 5.2867, 5.1816, 5.2420, 5.7538, 5.3768, 5.1296], device='cuda:7'), covar=tensor([0.0662, 0.1214, 0.0899, 0.1507, 0.1895, 0.0709, 0.0861, 0.1849], device='cuda:7'), in_proj_covar=tensor([0.0229, 0.0315, 0.0293, 0.0277, 0.0366, 0.0317, 0.0256, 0.0369], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:24:12,857 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26414.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:24:26,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-27 23:24:42,459 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.827e+02 4.226e+02 5.750e+02 6.713e+02 1.587e+03, threshold=1.150e+03, percent-clipped=3.0 2023-04-27 23:25:00,128 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-27 23:25:11,827 INFO [train.py:904] (7/8) Epoch 3, batch 6150, loss[loss=0.3148, simple_loss=0.3666, pruned_loss=0.1315, over 11876.00 frames. ], tot_loss[loss=0.2819, simple_loss=0.3506, pruned_loss=0.1066, over 3060213.25 frames. ], batch size: 247, lr: 2.01e-02, grad_scale: 4.0 2023-04-27 23:25:23,336 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26458.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:25:50,225 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26475.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:26:00,516 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4492, 4.5382, 5.0448, 5.0110, 4.9871, 4.5291, 4.5645, 4.4390], device='cuda:7'), covar=tensor([0.0215, 0.0212, 0.0215, 0.0276, 0.0346, 0.0207, 0.0534, 0.0265], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0168, 0.0187, 0.0186, 0.0226, 0.0194, 0.0291, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-27 23:26:28,220 INFO [train.py:904] (7/8) Epoch 3, batch 6200, loss[loss=0.3003, simple_loss=0.3723, pruned_loss=0.1141, over 15234.00 frames. ], tot_loss[loss=0.2803, simple_loss=0.3484, pruned_loss=0.1061, over 3056816.51 frames. ], batch size: 191, lr: 2.01e-02, grad_scale: 4.0 2023-04-27 23:26:48,684 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26514.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:27:03,499 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26524.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:27:18,521 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.570e+02 4.817e+02 5.994e+02 8.159e+02 2.733e+03, threshold=1.199e+03, percent-clipped=9.0 2023-04-27 23:27:34,692 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-04-27 23:27:41,895 INFO [train.py:904] (7/8) Epoch 3, batch 6250, loss[loss=0.3287, simple_loss=0.3713, pruned_loss=0.1431, over 11548.00 frames. ], tot_loss[loss=0.2797, simple_loss=0.3479, pruned_loss=0.1057, over 3061738.76 frames. ], batch size: 247, lr: 2.00e-02, grad_scale: 4.0 2023-04-27 23:27:58,190 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26562.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:28:29,896 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0691, 2.1102, 1.8173, 1.8805, 2.8076, 2.4161, 3.0607, 2.9527], device='cuda:7'), covar=tensor([0.0013, 0.0140, 0.0167, 0.0184, 0.0074, 0.0117, 0.0045, 0.0066], device='cuda:7'), in_proj_covar=tensor([0.0052, 0.0118, 0.0121, 0.0123, 0.0113, 0.0123, 0.0078, 0.0099], device='cuda:7'), out_proj_covar=tensor([7.0967e-05, 1.6790e-04, 1.6770e-04, 1.7556e-04, 1.6521e-04, 1.7701e-04, 1.1044e-04, 1.4703e-04], device='cuda:7') 2023-04-27 23:28:40,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9804, 3.3187, 2.4599, 4.5952, 4.4500, 4.1948, 2.0393, 2.9692], device='cuda:7'), covar=tensor([0.1270, 0.0447, 0.1196, 0.0069, 0.0192, 0.0249, 0.1114, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0134, 0.0164, 0.0071, 0.0137, 0.0140, 0.0154, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 23:28:54,859 INFO [train.py:904] (7/8) Epoch 3, batch 6300, loss[loss=0.2705, simple_loss=0.3365, pruned_loss=0.1022, over 17063.00 frames. ], tot_loss[loss=0.2782, simple_loss=0.3472, pruned_loss=0.1046, over 3085538.10 frames. ], batch size: 53, lr: 2.00e-02, grad_scale: 4.0 2023-04-27 23:28:55,988 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-27 23:28:57,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5448, 1.5279, 1.6626, 2.3259, 2.3497, 2.5321, 1.5895, 2.4696], device='cuda:7'), covar=tensor([0.0045, 0.0205, 0.0138, 0.0105, 0.0065, 0.0056, 0.0168, 0.0037], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0122, 0.0108, 0.0102, 0.0092, 0.0068, 0.0114, 0.0063], device='cuda:7'), out_proj_covar=tensor([1.3146e-04, 1.9288e-04, 1.7567e-04, 1.6549e-04, 1.4469e-04, 1.0482e-04, 1.7723e-04, 9.6858e-05], device='cuda:7') 2023-04-27 23:29:02,922 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26606.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:29:48,257 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.782e+02 4.322e+02 5.254e+02 6.822e+02 1.499e+03, threshold=1.051e+03, percent-clipped=4.0 2023-04-27 23:29:50,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3849, 1.8757, 1.5560, 1.5568, 2.2000, 1.9640, 2.2114, 2.2712], device='cuda:7'), covar=tensor([0.0020, 0.0126, 0.0148, 0.0170, 0.0074, 0.0127, 0.0049, 0.0082], device='cuda:7'), in_proj_covar=tensor([0.0052, 0.0118, 0.0120, 0.0121, 0.0112, 0.0122, 0.0077, 0.0097], device='cuda:7'), out_proj_covar=tensor([7.0893e-05, 1.6735e-04, 1.6500e-04, 1.7250e-04, 1.6281e-04, 1.7463e-04, 1.0877e-04, 1.4310e-04], device='cuda:7') 2023-04-27 23:30:11,917 INFO [train.py:904] (7/8) Epoch 3, batch 6350, loss[loss=0.2821, simple_loss=0.3416, pruned_loss=0.1113, over 16232.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3489, pruned_loss=0.107, over 3072523.35 frames. ], batch size: 35, lr: 2.00e-02, grad_scale: 4.0 2023-04-27 23:31:11,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2023-04-27 23:31:25,037 INFO [train.py:904] (7/8) Epoch 3, batch 6400, loss[loss=0.2658, simple_loss=0.3418, pruned_loss=0.09483, over 16611.00 frames. ], tot_loss[loss=0.2808, simple_loss=0.3481, pruned_loss=0.1068, over 3089309.59 frames. ], batch size: 75, lr: 2.00e-02, grad_scale: 8.0 2023-04-27 23:31:31,332 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2023-04-27 23:32:01,280 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-27 23:32:12,498 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.974e+02 5.048e+02 6.146e+02 7.949e+02 2.287e+03, threshold=1.229e+03, percent-clipped=11.0 2023-04-27 23:32:35,293 INFO [train.py:904] (7/8) Epoch 3, batch 6450, loss[loss=0.2504, simple_loss=0.3337, pruned_loss=0.08358, over 16494.00 frames. ], tot_loss[loss=0.2803, simple_loss=0.3482, pruned_loss=0.1062, over 3073999.09 frames. ], batch size: 68, lr: 2.00e-02, grad_scale: 8.0 2023-04-27 23:32:47,298 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26758.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:33:05,704 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26770.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:33:28,332 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3849, 3.7204, 3.9567, 3.9199, 3.8787, 3.5736, 3.2543, 3.6601], device='cuda:7'), covar=tensor([0.0547, 0.0409, 0.0528, 0.0645, 0.0754, 0.0582, 0.1494, 0.0525], device='cuda:7'), in_proj_covar=tensor([0.0189, 0.0172, 0.0192, 0.0186, 0.0230, 0.0196, 0.0299, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-27 23:33:53,767 INFO [train.py:904] (7/8) Epoch 3, batch 6500, loss[loss=0.284, simple_loss=0.3321, pruned_loss=0.1179, over 11156.00 frames. ], tot_loss[loss=0.2781, simple_loss=0.3455, pruned_loss=0.1053, over 3063278.19 frames. ], batch size: 248, lr: 2.00e-02, grad_scale: 8.0 2023-04-27 23:34:01,039 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26806.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:34:26,887 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-04-27 23:34:29,502 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26824.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:34:44,551 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 4.621e+02 5.604e+02 6.997e+02 1.424e+03, threshold=1.121e+03, percent-clipped=2.0 2023-04-27 23:35:11,377 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1298, 4.1090, 4.5829, 4.5530, 4.5846, 4.1907, 4.2085, 4.1535], device='cuda:7'), covar=tensor([0.0214, 0.0256, 0.0327, 0.0326, 0.0296, 0.0239, 0.0651, 0.0279], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0167, 0.0186, 0.0181, 0.0222, 0.0190, 0.0287, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-27 23:35:11,668 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.02 vs. limit=5.0 2023-04-27 23:35:12,124 INFO [train.py:904] (7/8) Epoch 3, batch 6550, loss[loss=0.3016, simple_loss=0.3895, pruned_loss=0.1069, over 16907.00 frames. ], tot_loss[loss=0.2803, simple_loss=0.3485, pruned_loss=0.106, over 3065014.34 frames. ], batch size: 109, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:35:26,469 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-27 23:35:45,521 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26872.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:35:46,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6608, 3.1886, 2.0384, 4.3581, 4.0655, 4.0178, 1.6151, 2.6190], device='cuda:7'), covar=tensor([0.1705, 0.0446, 0.1511, 0.0081, 0.0252, 0.0291, 0.1427, 0.0915], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0132, 0.0163, 0.0070, 0.0136, 0.0142, 0.0152, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 23:36:28,907 INFO [train.py:904] (7/8) Epoch 3, batch 6600, loss[loss=0.2945, simple_loss=0.3647, pruned_loss=0.1122, over 16803.00 frames. ], tot_loss[loss=0.2816, simple_loss=0.3507, pruned_loss=0.1062, over 3076336.97 frames. ], batch size: 124, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:36:34,581 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26904.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:36:37,805 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26906.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:37:08,147 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5848, 4.8673, 4.6193, 4.6201, 4.3463, 4.2766, 4.4633, 4.9252], device='cuda:7'), covar=tensor([0.0428, 0.0633, 0.0770, 0.0389, 0.0458, 0.0683, 0.0463, 0.0570], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0388, 0.0340, 0.0246, 0.0247, 0.0245, 0.0305, 0.0271], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:37:20,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.349e+02 4.809e+02 5.887e+02 7.395e+02 2.123e+03, threshold=1.177e+03, percent-clipped=7.0 2023-04-27 23:37:45,755 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26950.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:37:46,409 INFO [train.py:904] (7/8) Epoch 3, batch 6650, loss[loss=0.3785, simple_loss=0.4008, pruned_loss=0.1781, over 11278.00 frames. ], tot_loss[loss=0.2839, simple_loss=0.352, pruned_loss=0.108, over 3076936.94 frames. ], batch size: 248, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:37:51,273 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=26954.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:38:09,070 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26965.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:38:14,214 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-27 23:38:22,193 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-27 23:39:03,590 INFO [train.py:904] (7/8) Epoch 3, batch 6700, loss[loss=0.3426, simple_loss=0.3836, pruned_loss=0.1508, over 11461.00 frames. ], tot_loss[loss=0.2824, simple_loss=0.3501, pruned_loss=0.1074, over 3076083.45 frames. ], batch size: 247, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:39:19,054 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27011.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:39:57,111 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.121e+02 4.872e+02 5.906e+02 7.141e+02 1.703e+03, threshold=1.181e+03, percent-clipped=1.0 2023-04-27 23:40:21,188 INFO [train.py:904] (7/8) Epoch 3, batch 6750, loss[loss=0.2425, simple_loss=0.33, pruned_loss=0.07751, over 16881.00 frames. ], tot_loss[loss=0.2821, simple_loss=0.3492, pruned_loss=0.1075, over 3085266.58 frames. ], batch size: 102, lr: 1.99e-02, grad_scale: 8.0 2023-04-27 23:40:51,977 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27070.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:41:37,192 INFO [train.py:904] (7/8) Epoch 3, batch 6800, loss[loss=0.2837, simple_loss=0.3529, pruned_loss=0.1072, over 17029.00 frames. ], tot_loss[loss=0.2805, simple_loss=0.3478, pruned_loss=0.1066, over 3083318.05 frames. ], batch size: 55, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:42:03,932 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27118.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:42:04,243 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0183, 4.0560, 3.3744, 2.6644, 3.1962, 2.3775, 4.5520, 4.7689], device='cuda:7'), covar=tensor([0.1854, 0.0627, 0.1064, 0.1002, 0.1892, 0.1104, 0.0303, 0.0276], device='cuda:7'), in_proj_covar=tensor([0.0268, 0.0237, 0.0254, 0.0212, 0.0296, 0.0188, 0.0217, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:42:31,276 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.734e+02 4.279e+02 5.417e+02 7.296e+02 1.152e+03, threshold=1.083e+03, percent-clipped=0.0 2023-04-27 23:42:51,704 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27148.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:42:55,396 INFO [train.py:904] (7/8) Epoch 3, batch 6850, loss[loss=0.2974, simple_loss=0.3912, pruned_loss=0.1018, over 16498.00 frames. ], tot_loss[loss=0.2821, simple_loss=0.3496, pruned_loss=0.1073, over 3083887.19 frames. ], batch size: 68, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:44:04,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6095, 4.9783, 4.9863, 5.0280, 4.9070, 5.5004, 5.1194, 4.9305], device='cuda:7'), covar=tensor([0.0795, 0.1218, 0.1193, 0.1428, 0.2481, 0.0911, 0.1019, 0.1999], device='cuda:7'), in_proj_covar=tensor([0.0235, 0.0316, 0.0299, 0.0270, 0.0358, 0.0320, 0.0260, 0.0371], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:44:10,177 INFO [train.py:904] (7/8) Epoch 3, batch 6900, loss[loss=0.2602, simple_loss=0.3425, pruned_loss=0.08898, over 16835.00 frames. ], tot_loss[loss=0.2824, simple_loss=0.3517, pruned_loss=0.1066, over 3085408.79 frames. ], batch size: 96, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:44:23,350 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27209.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:45:02,583 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.150e+02 4.514e+02 5.203e+02 6.904e+02 1.170e+03, threshold=1.041e+03, percent-clipped=2.0 2023-04-27 23:45:28,468 INFO [train.py:904] (7/8) Epoch 3, batch 6950, loss[loss=0.2577, simple_loss=0.3356, pruned_loss=0.08989, over 16851.00 frames. ], tot_loss[loss=0.286, simple_loss=0.3536, pruned_loss=0.1092, over 3074138.93 frames. ], batch size: 42, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:45:35,322 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-27 23:45:42,190 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27260.0, num_to_drop=1, layers_to_drop={3} 2023-04-27 23:45:52,686 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6612, 4.9634, 5.1128, 5.0612, 5.0181, 5.5657, 5.1856, 5.0217], device='cuda:7'), covar=tensor([0.0736, 0.1178, 0.1138, 0.1323, 0.2085, 0.0669, 0.0870, 0.1621], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0313, 0.0299, 0.0275, 0.0360, 0.0321, 0.0259, 0.0373], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:46:43,379 INFO [train.py:904] (7/8) Epoch 3, batch 7000, loss[loss=0.2792, simple_loss=0.3641, pruned_loss=0.09714, over 16781.00 frames. ], tot_loss[loss=0.2866, simple_loss=0.3548, pruned_loss=0.1091, over 3051911.30 frames. ], batch size: 62, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:46:51,691 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27306.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:47:03,695 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-27 23:47:13,798 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-27 23:47:35,872 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.986e+02 5.041e+02 6.145e+02 8.860e+02 1.565e+03, threshold=1.229e+03, percent-clipped=7.0 2023-04-27 23:47:47,085 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1654, 2.5755, 2.4567, 3.2847, 3.1909, 3.2366, 1.8303, 2.8007], device='cuda:7'), covar=tensor([0.0941, 0.0315, 0.0869, 0.0082, 0.0237, 0.0296, 0.0971, 0.0589], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0132, 0.0162, 0.0070, 0.0136, 0.0142, 0.0151, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-27 23:48:01,115 INFO [train.py:904] (7/8) Epoch 3, batch 7050, loss[loss=0.2444, simple_loss=0.3278, pruned_loss=0.08047, over 16737.00 frames. ], tot_loss[loss=0.2873, simple_loss=0.3562, pruned_loss=0.1092, over 3054018.96 frames. ], batch size: 39, lr: 1.98e-02, grad_scale: 8.0 2023-04-27 23:48:34,993 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7254, 2.6911, 2.5020, 4.0709, 1.8341, 3.9095, 2.3439, 2.4042], device='cuda:7'), covar=tensor([0.0363, 0.0799, 0.0499, 0.0221, 0.2054, 0.0253, 0.0994, 0.1590], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0237, 0.0200, 0.0262, 0.0317, 0.0214, 0.0229, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:48:51,772 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8841, 3.8965, 4.3731, 4.3681, 4.3606, 3.8954, 4.0870, 4.0208], device='cuda:7'), covar=tensor([0.0253, 0.0331, 0.0281, 0.0327, 0.0351, 0.0287, 0.0630, 0.0336], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0172, 0.0189, 0.0182, 0.0227, 0.0195, 0.0293, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-27 23:49:19,634 INFO [train.py:904] (7/8) Epoch 3, batch 7100, loss[loss=0.2738, simple_loss=0.35, pruned_loss=0.0988, over 16710.00 frames. ], tot_loss[loss=0.2881, simple_loss=0.3558, pruned_loss=0.1102, over 3039765.83 frames. ], batch size: 57, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:50:12,108 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.055e+02 4.788e+02 6.087e+02 7.690e+02 2.114e+03, threshold=1.217e+03, percent-clipped=3.0 2023-04-27 23:50:15,552 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7313, 3.6581, 3.7038, 3.6515, 3.7043, 4.0715, 3.9502, 3.6882], device='cuda:7'), covar=tensor([0.1512, 0.1371, 0.1156, 0.1771, 0.2251, 0.1159, 0.0969, 0.2149], device='cuda:7'), in_proj_covar=tensor([0.0227, 0.0309, 0.0293, 0.0268, 0.0351, 0.0320, 0.0252, 0.0369], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:50:36,348 INFO [train.py:904] (7/8) Epoch 3, batch 7150, loss[loss=0.2412, simple_loss=0.3206, pruned_loss=0.08096, over 16435.00 frames. ], tot_loss[loss=0.2869, simple_loss=0.3541, pruned_loss=0.1099, over 3039412.20 frames. ], batch size: 68, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:51:21,466 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0043, 1.5769, 1.3409, 1.4456, 1.6888, 1.6222, 1.7373, 1.8421], device='cuda:7'), covar=tensor([0.0017, 0.0092, 0.0116, 0.0120, 0.0061, 0.0110, 0.0040, 0.0059], device='cuda:7'), in_proj_covar=tensor([0.0054, 0.0120, 0.0122, 0.0122, 0.0114, 0.0125, 0.0078, 0.0100], device='cuda:7'), out_proj_covar=tensor([7.2054e-05, 1.7108e-04, 1.6734e-04, 1.7281e-04, 1.6498e-04, 1.7889e-04, 1.0925e-04, 1.4417e-04], device='cuda:7') 2023-04-27 23:51:31,454 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-04-27 23:51:51,162 INFO [train.py:904] (7/8) Epoch 3, batch 7200, loss[loss=0.2274, simple_loss=0.3108, pruned_loss=0.07201, over 16564.00 frames. ], tot_loss[loss=0.2833, simple_loss=0.3511, pruned_loss=0.1077, over 3033194.48 frames. ], batch size: 75, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:51:55,803 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27504.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:52:45,516 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.403e+02 3.669e+02 4.524e+02 6.083e+02 1.066e+03, threshold=9.047e+02, percent-clipped=1.0 2023-04-27 23:53:11,027 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7566, 3.4595, 3.6859, 2.4466, 3.4931, 3.5183, 3.6273, 1.9287], device='cuda:7'), covar=tensor([0.0354, 0.0026, 0.0026, 0.0217, 0.0026, 0.0055, 0.0018, 0.0301], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0052, 0.0057, 0.0110, 0.0052, 0.0060, 0.0057, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:53:12,431 INFO [train.py:904] (7/8) Epoch 3, batch 7250, loss[loss=0.2777, simple_loss=0.3401, pruned_loss=0.1076, over 16814.00 frames. ], tot_loss[loss=0.2793, simple_loss=0.3478, pruned_loss=0.1054, over 3029312.48 frames. ], batch size: 116, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:53:26,516 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27560.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:54:26,529 INFO [train.py:904] (7/8) Epoch 3, batch 7300, loss[loss=0.32, simple_loss=0.3652, pruned_loss=0.1374, over 11534.00 frames. ], tot_loss[loss=0.2784, simple_loss=0.3469, pruned_loss=0.1049, over 3040354.83 frames. ], batch size: 248, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:54:35,543 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27606.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:54:38,058 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27608.0, num_to_drop=1, layers_to_drop={0} 2023-04-27 23:55:17,370 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.622e+02 4.400e+02 5.943e+02 7.946e+02 1.369e+03, threshold=1.189e+03, percent-clipped=13.0 2023-04-27 23:55:40,976 INFO [train.py:904] (7/8) Epoch 3, batch 7350, loss[loss=0.246, simple_loss=0.3235, pruned_loss=0.08424, over 17155.00 frames. ], tot_loss[loss=0.278, simple_loss=0.3464, pruned_loss=0.1048, over 3029472.58 frames. ], batch size: 46, lr: 1.97e-02, grad_scale: 8.0 2023-04-27 23:55:46,793 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27654.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:56:18,852 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1267, 4.1059, 4.6678, 4.6291, 4.6450, 4.1233, 4.2916, 4.1254], device='cuda:7'), covar=tensor([0.0258, 0.0259, 0.0266, 0.0350, 0.0364, 0.0275, 0.0671, 0.0356], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0167, 0.0184, 0.0183, 0.0221, 0.0192, 0.0287, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-27 23:56:31,309 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1108, 2.9540, 2.2187, 3.0113, 2.9902, 2.9340, 3.0949, 3.0299], device='cuda:7'), covar=tensor([0.0719, 0.0973, 0.2951, 0.1232, 0.1200, 0.2402, 0.1242, 0.1376], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0346, 0.0445, 0.0343, 0.0260, 0.0244, 0.0276, 0.0279], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:56:59,432 INFO [train.py:904] (7/8) Epoch 3, batch 7400, loss[loss=0.2629, simple_loss=0.3419, pruned_loss=0.09199, over 16616.00 frames. ], tot_loss[loss=0.2803, simple_loss=0.3483, pruned_loss=0.1062, over 3035102.29 frames. ], batch size: 57, lr: 1.96e-02, grad_scale: 4.0 2023-04-27 23:57:20,431 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27714.0, num_to_drop=1, layers_to_drop={1} 2023-04-27 23:57:55,129 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.894e+02 4.696e+02 5.728e+02 6.981e+02 1.392e+03, threshold=1.146e+03, percent-clipped=2.0 2023-04-27 23:58:18,294 INFO [train.py:904] (7/8) Epoch 3, batch 7450, loss[loss=0.2498, simple_loss=0.3399, pruned_loss=0.07983, over 16838.00 frames. ], tot_loss[loss=0.2809, simple_loss=0.3491, pruned_loss=0.1063, over 3056490.29 frames. ], batch size: 83, lr: 1.96e-02, grad_scale: 4.0 2023-04-27 23:58:51,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6988, 4.3905, 4.4909, 4.4431, 3.9956, 4.5029, 4.4650, 4.2675], device='cuda:7'), covar=tensor([0.0287, 0.0219, 0.0164, 0.0136, 0.0631, 0.0240, 0.0217, 0.0285], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0131, 0.0169, 0.0143, 0.0199, 0.0160, 0.0127, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-27 23:58:58,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27775.0, num_to_drop=1, layers_to_drop={2} 2023-04-27 23:59:25,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2509, 3.1897, 3.2181, 3.4488, 3.4115, 3.1810, 3.4079, 3.4481], device='cuda:7'), covar=tensor([0.0623, 0.0543, 0.1038, 0.0450, 0.0530, 0.1483, 0.0593, 0.0518], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0360, 0.0459, 0.0353, 0.0268, 0.0253, 0.0289, 0.0291], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-27 23:59:39,395 INFO [train.py:904] (7/8) Epoch 3, batch 7500, loss[loss=0.3285, simple_loss=0.3891, pruned_loss=0.134, over 15416.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3503, pruned_loss=0.1064, over 3051322.73 frames. ], batch size: 191, lr: 1.96e-02, grad_scale: 4.0 2023-04-27 23:59:44,296 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27804.0, num_to_drop=0, layers_to_drop=set() 2023-04-27 23:59:52,349 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2515, 3.0586, 2.4740, 2.2822, 2.3918, 2.0766, 3.1694, 3.4300], device='cuda:7'), covar=tensor([0.1920, 0.0768, 0.1280, 0.1160, 0.1506, 0.1147, 0.0414, 0.0424], device='cuda:7'), in_proj_covar=tensor([0.0264, 0.0239, 0.0254, 0.0213, 0.0293, 0.0191, 0.0216, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:00:33,136 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.883e+02 4.957e+02 5.924e+02 7.629e+02 1.550e+03, threshold=1.185e+03, percent-clipped=6.0 2023-04-28 00:00:55,940 INFO [train.py:904] (7/8) Epoch 3, batch 7550, loss[loss=0.2454, simple_loss=0.3245, pruned_loss=0.08316, over 16481.00 frames. ], tot_loss[loss=0.2804, simple_loss=0.3489, pruned_loss=0.1059, over 3062712.78 frames. ], batch size: 75, lr: 1.96e-02, grad_scale: 4.0 2023-04-28 00:00:58,760 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=27852.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:02:13,361 INFO [train.py:904] (7/8) Epoch 3, batch 7600, loss[loss=0.3218, simple_loss=0.3831, pruned_loss=0.1303, over 16457.00 frames. ], tot_loss[loss=0.28, simple_loss=0.3481, pruned_loss=0.106, over 3053299.13 frames. ], batch size: 146, lr: 1.96e-02, grad_scale: 8.0 2023-04-28 00:02:57,580 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5552, 4.4959, 4.3784, 4.5013, 3.7024, 4.5210, 4.4686, 4.1104], device='cuda:7'), covar=tensor([0.0386, 0.0237, 0.0214, 0.0162, 0.0830, 0.0252, 0.0224, 0.0317], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0132, 0.0170, 0.0142, 0.0198, 0.0161, 0.0126, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 00:03:02,869 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-28 00:03:04,615 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2046, 4.0495, 3.8961, 1.8737, 2.8979, 2.5277, 3.5339, 4.0084], device='cuda:7'), covar=tensor([0.0276, 0.0444, 0.0428, 0.1672, 0.0751, 0.0867, 0.0663, 0.0504], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0121, 0.0159, 0.0148, 0.0142, 0.0135, 0.0149, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-28 00:03:08,899 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.035e+02 4.816e+02 6.068e+02 7.688e+02 1.978e+03, threshold=1.214e+03, percent-clipped=6.0 2023-04-28 00:03:31,750 INFO [train.py:904] (7/8) Epoch 3, batch 7650, loss[loss=0.3213, simple_loss=0.3774, pruned_loss=0.1326, over 15305.00 frames. ], tot_loss[loss=0.2817, simple_loss=0.3493, pruned_loss=0.107, over 3053693.53 frames. ], batch size: 190, lr: 1.96e-02, grad_scale: 8.0 2023-04-28 00:04:52,693 INFO [train.py:904] (7/8) Epoch 3, batch 7700, loss[loss=0.2971, simple_loss=0.3615, pruned_loss=0.1163, over 15526.00 frames. ], tot_loss[loss=0.2835, simple_loss=0.3503, pruned_loss=0.1083, over 3072388.86 frames. ], batch size: 191, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:05:46,980 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.821e+02 4.680e+02 5.750e+02 6.748e+02 1.214e+03, threshold=1.150e+03, percent-clipped=1.0 2023-04-28 00:06:00,671 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-28 00:06:02,217 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1496, 3.0759, 3.1262, 1.6066, 3.2895, 3.2996, 2.7730, 2.7159], device='cuda:7'), covar=tensor([0.0778, 0.0134, 0.0158, 0.1189, 0.0082, 0.0087, 0.0309, 0.0386], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0082, 0.0078, 0.0142, 0.0072, 0.0073, 0.0113, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 00:06:11,253 INFO [train.py:904] (7/8) Epoch 3, batch 7750, loss[loss=0.241, simple_loss=0.3175, pruned_loss=0.08228, over 17135.00 frames. ], tot_loss[loss=0.2818, simple_loss=0.3495, pruned_loss=0.1071, over 3087163.37 frames. ], batch size: 47, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:06:11,932 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3725, 3.1462, 2.6592, 2.3057, 2.3827, 1.9913, 3.2640, 3.4284], device='cuda:7'), covar=tensor([0.1861, 0.0625, 0.1092, 0.1022, 0.1834, 0.1277, 0.0392, 0.0473], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0242, 0.0258, 0.0218, 0.0302, 0.0193, 0.0219, 0.0211], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:06:41,668 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28070.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 00:07:28,346 INFO [train.py:904] (7/8) Epoch 3, batch 7800, loss[loss=0.3697, simple_loss=0.3949, pruned_loss=0.1723, over 11550.00 frames. ], tot_loss[loss=0.2844, simple_loss=0.3511, pruned_loss=0.1089, over 3078264.33 frames. ], batch size: 248, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:08:22,850 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.157e+02 4.949e+02 5.902e+02 7.570e+02 1.555e+03, threshold=1.180e+03, percent-clipped=4.0 2023-04-28 00:08:45,075 INFO [train.py:904] (7/8) Epoch 3, batch 7850, loss[loss=0.2616, simple_loss=0.3415, pruned_loss=0.09084, over 16501.00 frames. ], tot_loss[loss=0.2832, simple_loss=0.3512, pruned_loss=0.1076, over 3094148.48 frames. ], batch size: 75, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:09:18,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2138, 3.5849, 3.7519, 1.7258, 3.9367, 3.9271, 3.1688, 3.1339], device='cuda:7'), covar=tensor([0.0761, 0.0103, 0.0109, 0.1123, 0.0044, 0.0044, 0.0238, 0.0309], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0083, 0.0079, 0.0145, 0.0073, 0.0072, 0.0113, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 00:10:00,841 INFO [train.py:904] (7/8) Epoch 3, batch 7900, loss[loss=0.2888, simple_loss=0.3565, pruned_loss=0.1105, over 15228.00 frames. ], tot_loss[loss=0.2815, simple_loss=0.3497, pruned_loss=0.1066, over 3092444.33 frames. ], batch size: 190, lr: 1.95e-02, grad_scale: 8.0 2023-04-28 00:10:16,872 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-28 00:10:21,660 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-28 00:10:45,274 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28229.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:10:48,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3096, 4.1638, 4.1488, 3.5039, 4.1226, 1.8129, 3.9166, 4.1069], device='cuda:7'), covar=tensor([0.0057, 0.0057, 0.0068, 0.0280, 0.0058, 0.1396, 0.0074, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0074, 0.0062, 0.0095, 0.0110, 0.0071, 0.0121, 0.0083, 0.0094], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:10:55,797 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 4.334e+02 4.942e+02 5.999e+02 2.073e+03, threshold=9.884e+02, percent-clipped=3.0 2023-04-28 00:10:58,848 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3754, 4.3014, 4.9527, 4.9119, 4.8993, 4.5305, 4.4566, 4.3104], device='cuda:7'), covar=tensor([0.0188, 0.0285, 0.0206, 0.0258, 0.0300, 0.0202, 0.0586, 0.0280], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0176, 0.0190, 0.0190, 0.0226, 0.0197, 0.0295, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 00:11:02,119 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 00:11:18,514 INFO [train.py:904] (7/8) Epoch 3, batch 7950, loss[loss=0.2845, simple_loss=0.3591, pruned_loss=0.1049, over 16608.00 frames. ], tot_loss[loss=0.2813, simple_loss=0.3496, pruned_loss=0.1065, over 3087355.75 frames. ], batch size: 57, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:11:26,047 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28255.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:11:53,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7013, 3.4328, 3.0916, 1.6976, 2.5617, 2.1915, 3.1366, 3.2734], device='cuda:7'), covar=tensor([0.0302, 0.0453, 0.0495, 0.1676, 0.0748, 0.0884, 0.0687, 0.0680], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0118, 0.0155, 0.0144, 0.0136, 0.0129, 0.0144, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 00:11:55,240 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4047, 3.4478, 3.2291, 3.3179, 3.0131, 3.4088, 3.1402, 3.2394], device='cuda:7'), covar=tensor([0.0380, 0.0203, 0.0173, 0.0151, 0.0514, 0.0221, 0.0888, 0.0260], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0133, 0.0170, 0.0142, 0.0203, 0.0163, 0.0127, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 00:12:13,967 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28287.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:12:18,742 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28290.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:12:33,514 INFO [train.py:904] (7/8) Epoch 3, batch 8000, loss[loss=0.2611, simple_loss=0.3368, pruned_loss=0.09272, over 16284.00 frames. ], tot_loss[loss=0.2826, simple_loss=0.3502, pruned_loss=0.1075, over 3078360.24 frames. ], batch size: 35, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:12:56,278 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28316.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 00:13:03,592 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4941, 4.5109, 5.0542, 5.1019, 5.0804, 4.6333, 4.6832, 4.4995], device='cuda:7'), covar=tensor([0.0213, 0.0255, 0.0254, 0.0314, 0.0346, 0.0264, 0.0626, 0.0272], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0182, 0.0195, 0.0197, 0.0233, 0.0203, 0.0302, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 00:13:27,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.455e+02 4.065e+02 5.202e+02 6.886e+02 1.574e+03, threshold=1.040e+03, percent-clipped=4.0 2023-04-28 00:13:32,350 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 00:13:46,403 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28348.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:13:49,308 INFO [train.py:904] (7/8) Epoch 3, batch 8050, loss[loss=0.2434, simple_loss=0.3201, pruned_loss=0.08339, over 16658.00 frames. ], tot_loss[loss=0.2814, simple_loss=0.3496, pruned_loss=0.1066, over 3095497.04 frames. ], batch size: 57, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:14:18,787 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28370.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 00:15:05,413 INFO [train.py:904] (7/8) Epoch 3, batch 8100, loss[loss=0.2568, simple_loss=0.3296, pruned_loss=0.09203, over 16609.00 frames. ], tot_loss[loss=0.2792, simple_loss=0.3481, pruned_loss=0.1052, over 3098058.11 frames. ], batch size: 62, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:15:30,654 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28418.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 00:15:57,045 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.864e+02 4.688e+02 5.875e+02 7.578e+02 1.406e+03, threshold=1.175e+03, percent-clipped=5.0 2023-04-28 00:16:20,003 INFO [train.py:904] (7/8) Epoch 3, batch 8150, loss[loss=0.253, simple_loss=0.3231, pruned_loss=0.09148, over 16902.00 frames. ], tot_loss[loss=0.2763, simple_loss=0.3449, pruned_loss=0.1038, over 3100458.27 frames. ], batch size: 96, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:16:34,999 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2734, 4.2255, 3.5774, 3.1071, 3.4974, 3.0169, 4.7632, 4.8921], device='cuda:7'), covar=tensor([0.1597, 0.0559, 0.0904, 0.0847, 0.1787, 0.0894, 0.0245, 0.0267], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0242, 0.0255, 0.0217, 0.0305, 0.0194, 0.0224, 0.0215], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:16:37,993 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2047, 4.0277, 4.2260, 2.8355, 4.0350, 4.1118, 4.1557, 2.2115], device='cuda:7'), covar=tensor([0.0329, 0.0018, 0.0022, 0.0215, 0.0029, 0.0056, 0.0017, 0.0281], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0054, 0.0056, 0.0109, 0.0052, 0.0062, 0.0057, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 00:17:15,242 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28486.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:17:23,245 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:17:36,824 INFO [train.py:904] (7/8) Epoch 3, batch 8200, loss[loss=0.2443, simple_loss=0.3265, pruned_loss=0.08107, over 16668.00 frames. ], tot_loss[loss=0.2736, simple_loss=0.3419, pruned_loss=0.1026, over 3105719.11 frames. ], batch size: 134, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:17:53,107 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 00:18:31,950 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.657e+02 4.728e+02 5.816e+02 7.305e+02 1.524e+03, threshold=1.163e+03, percent-clipped=3.0 2023-04-28 00:18:50,525 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28547.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:18:55,864 INFO [train.py:904] (7/8) Epoch 3, batch 8250, loss[loss=0.2455, simple_loss=0.3287, pruned_loss=0.08116, over 16642.00 frames. ], tot_loss[loss=0.2711, simple_loss=0.3413, pruned_loss=0.1004, over 3102550.74 frames. ], batch size: 134, lr: 1.94e-02, grad_scale: 8.0 2023-04-28 00:19:00,276 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28553.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:19:51,782 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28585.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:20:16,989 INFO [train.py:904] (7/8) Epoch 3, batch 8300, loss[loss=0.2319, simple_loss=0.3004, pruned_loss=0.08169, over 12238.00 frames. ], tot_loss[loss=0.2647, simple_loss=0.3373, pruned_loss=0.09606, over 3095277.77 frames. ], batch size: 246, lr: 1.93e-02, grad_scale: 8.0 2023-04-28 00:20:34,453 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28611.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 00:20:44,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8799, 1.8369, 1.6557, 1.6420, 2.4342, 2.2225, 2.8501, 2.7631], device='cuda:7'), covar=tensor([0.0015, 0.0158, 0.0182, 0.0200, 0.0090, 0.0123, 0.0042, 0.0066], device='cuda:7'), in_proj_covar=tensor([0.0054, 0.0120, 0.0124, 0.0126, 0.0115, 0.0124, 0.0078, 0.0100], device='cuda:7'), out_proj_covar=tensor([7.2300e-05, 1.6812e-04, 1.6876e-04, 1.7570e-04, 1.6349e-04, 1.7446e-04, 1.0992e-04, 1.4168e-04], device='cuda:7') 2023-04-28 00:21:14,650 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.491e+02 3.724e+02 4.538e+02 5.868e+02 1.330e+03, threshold=9.076e+02, percent-clipped=2.0 2023-04-28 00:21:27,816 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28643.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:21:39,710 INFO [train.py:904] (7/8) Epoch 3, batch 8350, loss[loss=0.2564, simple_loss=0.3371, pruned_loss=0.0879, over 15318.00 frames. ], tot_loss[loss=0.2595, simple_loss=0.3348, pruned_loss=0.0921, over 3096186.36 frames. ], batch size: 191, lr: 1.93e-02, grad_scale: 8.0 2023-04-28 00:23:00,537 INFO [train.py:904] (7/8) Epoch 3, batch 8400, loss[loss=0.2281, simple_loss=0.3108, pruned_loss=0.07273, over 16449.00 frames. ], tot_loss[loss=0.2549, simple_loss=0.3311, pruned_loss=0.08932, over 3088704.80 frames. ], batch size: 146, lr: 1.93e-02, grad_scale: 8.0 2023-04-28 00:23:58,230 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.191e+02 3.936e+02 4.649e+02 5.360e+02 9.184e+02, threshold=9.299e+02, percent-clipped=1.0 2023-04-28 00:24:20,126 INFO [train.py:904] (7/8) Epoch 3, batch 8450, loss[loss=0.2645, simple_loss=0.3285, pruned_loss=0.1003, over 12545.00 frames. ], tot_loss[loss=0.2523, simple_loss=0.3292, pruned_loss=0.08768, over 3070866.72 frames. ], batch size: 248, lr: 1.93e-02, grad_scale: 4.0 2023-04-28 00:24:22,648 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2056, 3.2820, 1.6045, 3.2353, 2.2549, 3.2505, 1.9630, 2.6393], device='cuda:7'), covar=tensor([0.0085, 0.0193, 0.1405, 0.0051, 0.0737, 0.0375, 0.1284, 0.0536], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0132, 0.0172, 0.0078, 0.0158, 0.0157, 0.0182, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 00:25:17,336 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9190, 4.1491, 3.9660, 3.9884, 3.6501, 3.7063, 3.9234, 4.0933], device='cuda:7'), covar=tensor([0.0619, 0.0751, 0.0779, 0.0406, 0.0564, 0.1088, 0.0439, 0.0781], device='cuda:7'), in_proj_covar=tensor([0.0270, 0.0378, 0.0323, 0.0244, 0.0241, 0.0247, 0.0296, 0.0260], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:25:21,144 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6157, 4.5092, 4.4585, 3.9205, 4.3170, 1.8581, 4.1659, 4.4091], device='cuda:7'), covar=tensor([0.0049, 0.0040, 0.0053, 0.0190, 0.0054, 0.1319, 0.0064, 0.0076], device='cuda:7'), in_proj_covar=tensor([0.0073, 0.0063, 0.0097, 0.0104, 0.0071, 0.0124, 0.0084, 0.0093], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:25:42,011 INFO [train.py:904] (7/8) Epoch 3, batch 8500, loss[loss=0.2062, simple_loss=0.2788, pruned_loss=0.06678, over 11882.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.3237, pruned_loss=0.08349, over 3064059.44 frames. ], batch size: 248, lr: 1.93e-02, grad_scale: 4.0 2023-04-28 00:26:35,160 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-04-28 00:26:40,692 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.668e+02 4.634e+02 5.862e+02 2.485e+03, threshold=9.268e+02, percent-clipped=6.0 2023-04-28 00:26:50,007 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28842.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:27:01,011 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:27:05,706 INFO [train.py:904] (7/8) Epoch 3, batch 8550, loss[loss=0.24, simple_loss=0.3222, pruned_loss=0.07893, over 16887.00 frames. ], tot_loss[loss=0.2424, simple_loss=0.3209, pruned_loss=0.08193, over 3043267.51 frames. ], batch size: 109, lr: 1.93e-02, grad_scale: 4.0 2023-04-28 00:27:25,977 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28861.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:28:13,745 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28885.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:28:46,471 INFO [train.py:904] (7/8) Epoch 3, batch 8600, loss[loss=0.2199, simple_loss=0.2986, pruned_loss=0.07061, over 12419.00 frames. ], tot_loss[loss=0.2423, simple_loss=0.3215, pruned_loss=0.08155, over 3015544.36 frames. ], batch size: 247, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:29:07,474 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28911.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 00:29:30,915 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28922.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:29:50,840 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28933.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:29:57,571 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.629e+02 3.755e+02 4.530e+02 5.746e+02 8.888e+02, threshold=9.061e+02, percent-clipped=0.0 2023-04-28 00:30:11,316 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28943.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:30:25,449 INFO [train.py:904] (7/8) Epoch 3, batch 8650, loss[loss=0.212, simple_loss=0.2999, pruned_loss=0.06208, over 16808.00 frames. ], tot_loss[loss=0.2391, simple_loss=0.3189, pruned_loss=0.07969, over 2993180.27 frames. ], batch size: 124, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:30:45,828 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28959.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 00:31:12,814 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 00:31:52,875 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=28991.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:32:12,032 INFO [train.py:904] (7/8) Epoch 3, batch 8700, loss[loss=0.2325, simple_loss=0.3137, pruned_loss=0.07561, over 15465.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.3148, pruned_loss=0.07696, over 3017986.40 frames. ], batch size: 191, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:32:15,197 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29002.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:32:44,809 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-28 00:33:18,502 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 3.379e+02 3.929e+02 4.901e+02 7.493e+02, threshold=7.858e+02, percent-clipped=0.0 2023-04-28 00:33:22,243 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6551, 3.6541, 1.5774, 3.7175, 2.3780, 3.7231, 2.0093, 2.8210], device='cuda:7'), covar=tensor([0.0072, 0.0231, 0.1535, 0.0034, 0.0773, 0.0236, 0.1298, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0127, 0.0169, 0.0075, 0.0155, 0.0148, 0.0177, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 00:33:46,759 INFO [train.py:904] (7/8) Epoch 3, batch 8750, loss[loss=0.2608, simple_loss=0.3433, pruned_loss=0.08913, over 16865.00 frames. ], tot_loss[loss=0.2327, simple_loss=0.3139, pruned_loss=0.07579, over 3030982.35 frames. ], batch size: 116, lr: 1.92e-02, grad_scale: 4.0 2023-04-28 00:34:19,017 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29063.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:35:03,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7597, 4.5520, 4.5532, 4.4997, 4.0840, 4.6214, 4.6333, 4.3123], device='cuda:7'), covar=tensor([0.0281, 0.0211, 0.0156, 0.0132, 0.0695, 0.0177, 0.0170, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0123, 0.0162, 0.0133, 0.0185, 0.0152, 0.0115, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:35:38,459 INFO [train.py:904] (7/8) Epoch 3, batch 8800, loss[loss=0.2471, simple_loss=0.3389, pruned_loss=0.07765, over 16773.00 frames. ], tot_loss[loss=0.2299, simple_loss=0.3115, pruned_loss=0.07412, over 3034418.94 frames. ], batch size: 124, lr: 1.92e-02, grad_scale: 8.0 2023-04-28 00:36:29,355 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9428, 2.6487, 2.5491, 4.5268, 1.7922, 3.9305, 2.3665, 2.3935], device='cuda:7'), covar=tensor([0.0343, 0.0995, 0.0521, 0.0160, 0.2305, 0.0302, 0.1093, 0.1740], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0241, 0.0202, 0.0259, 0.0314, 0.0212, 0.0229, 0.0291], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:36:52,046 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.420e+02 3.673e+02 4.669e+02 5.868e+02 1.100e+03, threshold=9.339e+02, percent-clipped=8.0 2023-04-28 00:36:57,493 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7269, 2.7086, 1.6174, 2.7613, 1.9064, 2.8108, 1.9481, 2.4302], device='cuda:7'), covar=tensor([0.0104, 0.0280, 0.1375, 0.0067, 0.0863, 0.0364, 0.1293, 0.0716], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0127, 0.0170, 0.0077, 0.0156, 0.0150, 0.0178, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 00:37:03,968 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29142.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:37:09,956 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2488, 4.2562, 3.8607, 1.9207, 3.0641, 2.4180, 3.6606, 4.2386], device='cuda:7'), covar=tensor([0.0349, 0.0362, 0.0370, 0.1553, 0.0702, 0.0930, 0.0738, 0.0561], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0108, 0.0148, 0.0143, 0.0132, 0.0129, 0.0139, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 00:37:16,984 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29148.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:37:21,274 INFO [train.py:904] (7/8) Epoch 3, batch 8850, loss[loss=0.2273, simple_loss=0.3187, pruned_loss=0.06794, over 16941.00 frames. ], tot_loss[loss=0.2299, simple_loss=0.3136, pruned_loss=0.07307, over 3036745.59 frames. ], batch size: 116, lr: 1.92e-02, grad_scale: 8.0 2023-04-28 00:38:45,757 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29190.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:38:57,899 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29196.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:39:06,875 INFO [train.py:904] (7/8) Epoch 3, batch 8900, loss[loss=0.2272, simple_loss=0.3141, pruned_loss=0.07018, over 16550.00 frames. ], tot_loss[loss=0.2288, simple_loss=0.3136, pruned_loss=0.07195, over 3053055.39 frames. ], batch size: 62, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:39:27,921 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 00:39:40,945 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29217.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:40:35,546 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29236.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:40:36,248 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.258e+02 3.766e+02 4.506e+02 5.745e+02 1.186e+03, threshold=9.012e+02, percent-clipped=1.0 2023-04-28 00:41:11,860 INFO [train.py:904] (7/8) Epoch 3, batch 8950, loss[loss=0.1994, simple_loss=0.2881, pruned_loss=0.05539, over 15526.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3143, pruned_loss=0.07289, over 3069038.03 frames. ], batch size: 192, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:41:42,812 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-28 00:42:54,222 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29297.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:43:00,675 INFO [train.py:904] (7/8) Epoch 3, batch 9000, loss[loss=0.2352, simple_loss=0.299, pruned_loss=0.08573, over 12174.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3104, pruned_loss=0.0708, over 3070793.78 frames. ], batch size: 247, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:43:00,676 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 00:43:11,838 INFO [train.py:938] (7/8) Epoch 3, validation: loss=0.1886, simple_loss=0.2904, pruned_loss=0.04341, over 944034.00 frames. 2023-04-28 00:43:11,839 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 00:43:48,501 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2023-04-28 00:44:26,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.245e+02 3.595e+02 4.170e+02 5.160e+02 8.461e+02, threshold=8.339e+02, percent-clipped=0.0 2023-04-28 00:44:57,080 INFO [train.py:904] (7/8) Epoch 3, batch 9050, loss[loss=0.2004, simple_loss=0.283, pruned_loss=0.0589, over 16895.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3129, pruned_loss=0.07213, over 3091436.47 frames. ], batch size: 96, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:45:12,887 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29358.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:46:41,637 INFO [train.py:904] (7/8) Epoch 3, batch 9100, loss[loss=0.2183, simple_loss=0.3108, pruned_loss=0.06289, over 16611.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3122, pruned_loss=0.0729, over 3079151.58 frames. ], batch size: 62, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:47:54,636 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3838, 4.0626, 4.0802, 1.8427, 4.2907, 4.2175, 3.3756, 3.2882], device='cuda:7'), covar=tensor([0.0726, 0.0070, 0.0106, 0.1155, 0.0034, 0.0034, 0.0197, 0.0301], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0080, 0.0074, 0.0144, 0.0070, 0.0069, 0.0108, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 00:48:08,509 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.097e+02 4.371e+02 5.293e+02 6.986e+02 1.215e+03, threshold=1.059e+03, percent-clipped=12.0 2023-04-28 00:48:40,652 INFO [train.py:904] (7/8) Epoch 3, batch 9150, loss[loss=0.2088, simple_loss=0.3033, pruned_loss=0.05712, over 17084.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.3124, pruned_loss=0.07221, over 3077553.57 frames. ], batch size: 97, lr: 1.91e-02, grad_scale: 8.0 2023-04-28 00:50:24,884 INFO [train.py:904] (7/8) Epoch 3, batch 9200, loss[loss=0.2351, simple_loss=0.3136, pruned_loss=0.07826, over 16691.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3071, pruned_loss=0.07068, over 3077016.81 frames. ], batch size: 134, lr: 1.90e-02, grad_scale: 8.0 2023-04-28 00:50:55,584 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29517.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:51:32,425 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.365e+02 4.043e+02 4.899e+02 6.737e+02 1.438e+03, threshold=9.797e+02, percent-clipped=1.0 2023-04-28 00:51:39,776 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0814, 3.8231, 3.7375, 2.9095, 3.7282, 3.7765, 3.6920, 2.3140], device='cuda:7'), covar=tensor([0.0341, 0.0016, 0.0035, 0.0191, 0.0028, 0.0032, 0.0021, 0.0282], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0052, 0.0056, 0.0106, 0.0052, 0.0060, 0.0055, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 00:52:01,392 INFO [train.py:904] (7/8) Epoch 3, batch 9250, loss[loss=0.1887, simple_loss=0.2828, pruned_loss=0.04727, over 16834.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3062, pruned_loss=0.07026, over 3082847.33 frames. ], batch size: 90, lr: 1.90e-02, grad_scale: 8.0 2023-04-28 00:52:02,460 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4312, 4.5127, 4.5223, 4.5270, 4.6079, 4.9938, 4.7185, 4.4152], device='cuda:7'), covar=tensor([0.0724, 0.1326, 0.1243, 0.1453, 0.2010, 0.0882, 0.0864, 0.1764], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0301, 0.0290, 0.0269, 0.0343, 0.0319, 0.0247, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:52:02,595 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29551.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:52:29,839 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:53:31,951 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29592.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:53:50,314 INFO [train.py:904] (7/8) Epoch 3, batch 9300, loss[loss=0.214, simple_loss=0.283, pruned_loss=0.07247, over 12368.00 frames. ], tot_loss[loss=0.221, simple_loss=0.3038, pruned_loss=0.06915, over 3073339.62 frames. ], batch size: 248, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 00:54:15,997 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29612.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:55:11,316 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.534e+02 4.207e+02 4.979e+02 1.086e+03, threshold=8.414e+02, percent-clipped=1.0 2023-04-28 00:55:35,085 INFO [train.py:904] (7/8) Epoch 3, batch 9350, loss[loss=0.2293, simple_loss=0.3134, pruned_loss=0.07261, over 16887.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.3046, pruned_loss=0.06949, over 3073506.37 frames. ], batch size: 116, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 00:55:51,518 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29658.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:56:00,165 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29662.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:57:16,970 INFO [train.py:904] (7/8) Epoch 3, batch 9400, loss[loss=0.2241, simple_loss=0.3209, pruned_loss=0.06367, over 16881.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.3046, pruned_loss=0.069, over 3072971.61 frames. ], batch size: 102, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 00:57:27,968 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29706.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 00:57:57,826 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9106, 2.9053, 2.3504, 3.8275, 3.6609, 3.6755, 1.5398, 2.8843], device='cuda:7'), covar=tensor([0.1178, 0.0415, 0.1052, 0.0074, 0.0171, 0.0323, 0.1238, 0.0594], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0132, 0.0162, 0.0068, 0.0130, 0.0142, 0.0153, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-04-28 00:58:02,211 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29723.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 00:58:16,029 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3379, 5.2390, 5.0673, 4.5367, 5.0817, 2.2331, 4.8680, 5.0719], device='cuda:7'), covar=tensor([0.0033, 0.0036, 0.0049, 0.0170, 0.0036, 0.1324, 0.0048, 0.0070], device='cuda:7'), in_proj_covar=tensor([0.0071, 0.0060, 0.0094, 0.0094, 0.0068, 0.0120, 0.0082, 0.0090], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:58:32,712 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.403e+02 3.993e+02 5.001e+02 6.201e+02 1.324e+03, threshold=1.000e+03, percent-clipped=7.0 2023-04-28 00:58:46,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5863, 3.5787, 3.0658, 2.2873, 2.5327, 2.1653, 3.7312, 3.9609], device='cuda:7'), covar=tensor([0.1751, 0.0543, 0.0864, 0.1067, 0.1683, 0.1187, 0.0265, 0.0307], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0239, 0.0251, 0.0211, 0.0231, 0.0192, 0.0213, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 00:58:58,188 INFO [train.py:904] (7/8) Epoch 3, batch 9450, loss[loss=0.2138, simple_loss=0.3005, pruned_loss=0.06352, over 16336.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3074, pruned_loss=0.06958, over 3088597.69 frames. ], batch size: 165, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 01:00:37,301 INFO [train.py:904] (7/8) Epoch 3, batch 9500, loss[loss=0.2192, simple_loss=0.294, pruned_loss=0.07222, over 12931.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.3056, pruned_loss=0.06855, over 3077417.07 frames. ], batch size: 247, lr: 1.90e-02, grad_scale: 4.0 2023-04-28 01:00:41,470 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29802.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:00:41,809 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 01:00:55,962 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1113, 2.3845, 1.7191, 1.9773, 3.0117, 2.6778, 3.2971, 3.1427], device='cuda:7'), covar=tensor([0.0014, 0.0116, 0.0162, 0.0143, 0.0056, 0.0101, 0.0028, 0.0046], device='cuda:7'), in_proj_covar=tensor([0.0053, 0.0122, 0.0120, 0.0121, 0.0112, 0.0121, 0.0075, 0.0097], device='cuda:7'), out_proj_covar=tensor([6.8239e-05, 1.6961e-04, 1.6136e-04, 1.6579e-04, 1.5670e-04, 1.6864e-04, 1.0170e-04, 1.3459e-04], device='cuda:7') 2023-04-28 01:01:50,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.435e+02 3.682e+02 4.369e+02 5.544e+02 1.182e+03, threshold=8.738e+02, percent-clipped=2.0 2023-04-28 01:02:22,322 INFO [train.py:904] (7/8) Epoch 3, batch 9550, loss[loss=0.2164, simple_loss=0.2939, pruned_loss=0.06944, over 12766.00 frames. ], tot_loss[loss=0.2222, simple_loss=0.3058, pruned_loss=0.06931, over 3072562.58 frames. ], batch size: 248, lr: 1.89e-02, grad_scale: 4.0 2023-04-28 01:02:50,246 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29863.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:03:49,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29892.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:04:04,158 INFO [train.py:904] (7/8) Epoch 3, batch 9600, loss[loss=0.2697, simple_loss=0.3485, pruned_loss=0.09543, over 16698.00 frames. ], tot_loss[loss=0.2249, simple_loss=0.3078, pruned_loss=0.07096, over 3058841.99 frames. ], batch size: 134, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:04:16,861 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29907.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:04:43,465 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-28 01:05:18,577 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.324e+02 3.589e+02 4.321e+02 5.133e+02 1.002e+03, threshold=8.641e+02, percent-clipped=3.0 2023-04-28 01:05:23,702 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=29940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:05:51,217 INFO [train.py:904] (7/8) Epoch 3, batch 9650, loss[loss=0.219, simple_loss=0.2891, pruned_loss=0.07441, over 12170.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3091, pruned_loss=0.07066, over 3062364.58 frames. ], batch size: 248, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:07:41,663 INFO [train.py:904] (7/8) Epoch 3, batch 9700, loss[loss=0.2203, simple_loss=0.3067, pruned_loss=0.06699, over 16395.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.3077, pruned_loss=0.07028, over 3054695.46 frames. ], batch size: 146, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:08:16,061 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30018.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:08:16,525 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 01:08:59,979 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.947e+02 3.446e+02 4.381e+02 5.437e+02 9.929e+02, threshold=8.763e+02, percent-clipped=3.0 2023-04-28 01:09:24,301 INFO [train.py:904] (7/8) Epoch 3, batch 9750, loss[loss=0.2063, simple_loss=0.2977, pruned_loss=0.05746, over 16795.00 frames. ], tot_loss[loss=0.2235, simple_loss=0.3063, pruned_loss=0.07034, over 3047012.71 frames. ], batch size: 124, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:09:36,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2519, 2.1526, 2.2221, 3.6012, 1.7734, 3.2347, 2.1035, 1.9208], device='cuda:7'), covar=tensor([0.0371, 0.1198, 0.0573, 0.0213, 0.2212, 0.0336, 0.1253, 0.1761], device='cuda:7'), in_proj_covar=tensor([0.0260, 0.0251, 0.0207, 0.0267, 0.0314, 0.0220, 0.0235, 0.0296], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:09:40,707 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6134, 3.6913, 3.0971, 2.2978, 2.6783, 2.2493, 4.0005, 4.0778], device='cuda:7'), covar=tensor([0.1905, 0.0650, 0.1075, 0.1179, 0.1530, 0.1202, 0.0320, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0236, 0.0254, 0.0212, 0.0224, 0.0192, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:09:42,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3700, 4.2562, 4.1652, 3.7027, 4.1677, 1.5838, 3.9967, 4.1558], device='cuda:7'), covar=tensor([0.0063, 0.0055, 0.0078, 0.0217, 0.0063, 0.1667, 0.0074, 0.0112], device='cuda:7'), in_proj_covar=tensor([0.0071, 0.0061, 0.0094, 0.0095, 0.0068, 0.0120, 0.0082, 0.0090], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:11:00,982 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9420, 2.1335, 2.1893, 3.1702, 1.9926, 2.9128, 2.2297, 1.9048], device='cuda:7'), covar=tensor([0.0372, 0.1041, 0.0520, 0.0235, 0.1855, 0.0370, 0.1056, 0.1767], device='cuda:7'), in_proj_covar=tensor([0.0261, 0.0251, 0.0207, 0.0268, 0.0316, 0.0221, 0.0236, 0.0295], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:11:02,940 INFO [train.py:904] (7/8) Epoch 3, batch 9800, loss[loss=0.2018, simple_loss=0.2803, pruned_loss=0.0616, over 12478.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.306, pruned_loss=0.06914, over 3038174.64 frames. ], batch size: 249, lr: 1.89e-02, grad_scale: 8.0 2023-04-28 01:11:48,254 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30125.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:12:14,773 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.212e+02 3.552e+02 4.452e+02 5.972e+02 1.192e+03, threshold=8.904e+02, percent-clipped=4.0 2023-04-28 01:12:47,472 INFO [train.py:904] (7/8) Epoch 3, batch 9850, loss[loss=0.2185, simple_loss=0.3083, pruned_loss=0.06439, over 16827.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.3075, pruned_loss=0.06866, over 3061606.47 frames. ], batch size: 83, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:13:02,461 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30158.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:14:06,763 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30186.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:14:39,101 INFO [train.py:904] (7/8) Epoch 3, batch 9900, loss[loss=0.2183, simple_loss=0.2942, pruned_loss=0.0712, over 12988.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3078, pruned_loss=0.06841, over 3067147.85 frames. ], batch size: 248, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:14:54,539 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30207.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:14:59,640 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30209.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:16:05,827 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.521e+02 3.778e+02 4.995e+02 6.247e+02 1.716e+03, threshold=9.990e+02, percent-clipped=10.0 2023-04-28 01:16:33,053 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7410, 3.5780, 3.7433, 3.6009, 3.8014, 4.1441, 3.9526, 3.6506], device='cuda:7'), covar=tensor([0.1459, 0.1588, 0.1084, 0.2035, 0.2193, 0.1106, 0.0899, 0.2128], device='cuda:7'), in_proj_covar=tensor([0.0207, 0.0298, 0.0281, 0.0259, 0.0335, 0.0307, 0.0239, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:16:35,563 INFO [train.py:904] (7/8) Epoch 3, batch 9950, loss[loss=0.2227, simple_loss=0.3112, pruned_loss=0.06716, over 16671.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3107, pruned_loss=0.06981, over 3061932.49 frames. ], batch size: 134, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:16:47,153 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30255.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:17:23,165 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30270.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:18:00,103 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4009, 3.5095, 2.9606, 2.2748, 2.4749, 2.1293, 3.6869, 3.8365], device='cuda:7'), covar=tensor([0.2040, 0.0655, 0.1079, 0.1152, 0.1507, 0.1256, 0.0389, 0.0345], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0238, 0.0251, 0.0211, 0.0224, 0.0194, 0.0212, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:18:37,076 INFO [train.py:904] (7/8) Epoch 3, batch 10000, loss[loss=0.2221, simple_loss=0.309, pruned_loss=0.06756, over 16197.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.3081, pruned_loss=0.06803, over 3087807.62 frames. ], batch size: 166, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:19:12,457 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30318.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:19:55,685 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.113e+02 3.684e+02 4.714e+02 5.759e+02 1.067e+03, threshold=9.428e+02, percent-clipped=2.0 2023-04-28 01:20:05,759 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9539, 5.3904, 5.0795, 5.1136, 4.6331, 4.4951, 4.8364, 5.4304], device='cuda:7'), covar=tensor([0.0685, 0.0650, 0.0859, 0.0376, 0.0661, 0.0714, 0.0510, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0265, 0.0366, 0.0316, 0.0242, 0.0241, 0.0245, 0.0296, 0.0262], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:20:19,882 INFO [train.py:904] (7/8) Epoch 3, batch 10050, loss[loss=0.243, simple_loss=0.3242, pruned_loss=0.08086, over 16956.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.3077, pruned_loss=0.06795, over 3083143.52 frames. ], batch size: 109, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:20:38,447 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:20:50,750 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30366.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:21:54,412 INFO [train.py:904] (7/8) Epoch 3, batch 10100, loss[loss=0.2265, simple_loss=0.2941, pruned_loss=0.07945, over 12481.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3085, pruned_loss=0.06908, over 3077839.19 frames. ], batch size: 248, lr: 1.88e-02, grad_scale: 8.0 2023-04-28 01:22:37,217 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30421.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:22:55,003 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 01:22:56,718 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9652, 2.7734, 2.6872, 1.6398, 2.8861, 2.8443, 2.4167, 2.3828], device='cuda:7'), covar=tensor([0.0758, 0.0131, 0.0165, 0.1113, 0.0083, 0.0099, 0.0414, 0.0414], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0082, 0.0079, 0.0149, 0.0073, 0.0072, 0.0114, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 01:23:00,423 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.870e+02 3.919e+02 4.872e+02 6.072e+02 1.469e+03, threshold=9.744e+02, percent-clipped=6.0 2023-04-28 01:23:38,636 INFO [train.py:904] (7/8) Epoch 4, batch 0, loss[loss=0.2966, simple_loss=0.3615, pruned_loss=0.1159, over 16739.00 frames. ], tot_loss[loss=0.2966, simple_loss=0.3615, pruned_loss=0.1159, over 16739.00 frames. ], batch size: 57, lr: 1.75e-02, grad_scale: 8.0 2023-04-28 01:23:38,637 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 01:23:46,518 INFO [train.py:938] (7/8) Epoch 4, validation: loss=0.188, simple_loss=0.2904, pruned_loss=0.04284, over 944034.00 frames. 2023-04-28 01:23:46,519 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 01:23:57,023 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5818, 4.2850, 3.8665, 1.9752, 3.0822, 2.3526, 3.7323, 4.2357], device='cuda:7'), covar=tensor([0.0251, 0.0543, 0.0476, 0.1591, 0.0687, 0.1034, 0.0656, 0.0714], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0109, 0.0152, 0.0143, 0.0133, 0.0129, 0.0136, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 01:23:58,004 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30458.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:24:12,727 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6360, 4.0418, 4.0429, 2.1140, 4.2757, 4.2620, 3.2596, 3.2367], device='cuda:7'), covar=tensor([0.0707, 0.0103, 0.0140, 0.1169, 0.0041, 0.0043, 0.0294, 0.0374], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0081, 0.0079, 0.0147, 0.0072, 0.0072, 0.0112, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 01:24:19,365 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 01:24:29,165 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30481.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:24:56,031 INFO [train.py:904] (7/8) Epoch 4, batch 50, loss[loss=0.3417, simple_loss=0.3901, pruned_loss=0.1467, over 11936.00 frames. ], tot_loss[loss=0.2675, simple_loss=0.3337, pruned_loss=0.1006, over 748122.00 frames. ], batch size: 248, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:25:02,271 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30506.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:25:49,219 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2002, 4.3786, 2.3083, 4.7917, 2.8546, 4.6292, 2.1050, 3.2140], device='cuda:7'), covar=tensor([0.0094, 0.0214, 0.1373, 0.0020, 0.0736, 0.0302, 0.1465, 0.0556], device='cuda:7'), in_proj_covar=tensor([0.0098, 0.0138, 0.0173, 0.0081, 0.0160, 0.0162, 0.0181, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 01:25:49,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.617e+02 4.176e+02 5.221e+02 6.159e+02 1.340e+03, threshold=1.044e+03, percent-clipped=3.0 2023-04-28 01:26:02,164 INFO [train.py:904] (7/8) Epoch 4, batch 100, loss[loss=0.2667, simple_loss=0.3339, pruned_loss=0.09975, over 16509.00 frames. ], tot_loss[loss=0.2589, simple_loss=0.3265, pruned_loss=0.09565, over 1324453.18 frames. ], batch size: 68, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:26:23,727 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:27:11,711 INFO [train.py:904] (7/8) Epoch 4, batch 150, loss[loss=0.2277, simple_loss=0.3057, pruned_loss=0.07486, over 17225.00 frames. ], tot_loss[loss=0.2575, simple_loss=0.3243, pruned_loss=0.09531, over 1757309.06 frames. ], batch size: 45, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:28:01,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 01:28:04,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.524e+02 3.830e+02 4.808e+02 5.813e+02 1.621e+03, threshold=9.616e+02, percent-clipped=3.0 2023-04-28 01:28:19,184 INFO [train.py:904] (7/8) Epoch 4, batch 200, loss[loss=0.2616, simple_loss=0.3157, pruned_loss=0.1038, over 16843.00 frames. ], tot_loss[loss=0.255, simple_loss=0.322, pruned_loss=0.09394, over 2104889.02 frames. ], batch size: 83, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:28:21,012 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 01:28:24,007 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30655.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:28:56,912 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7536, 3.9773, 2.9716, 2.4003, 2.9712, 2.1014, 4.3895, 4.3135], device='cuda:7'), covar=tensor([0.2217, 0.0794, 0.1361, 0.1435, 0.2333, 0.1615, 0.0395, 0.0487], device='cuda:7'), in_proj_covar=tensor([0.0270, 0.0247, 0.0261, 0.0221, 0.0266, 0.0201, 0.0223, 0.0220], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:29:26,983 INFO [train.py:904] (7/8) Epoch 4, batch 250, loss[loss=0.2608, simple_loss=0.3198, pruned_loss=0.1009, over 16720.00 frames. ], tot_loss[loss=0.25, simple_loss=0.3181, pruned_loss=0.09093, over 2379847.54 frames. ], batch size: 124, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:29:48,577 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30716.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:29:48,779 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30716.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 01:29:50,014 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9271, 4.6052, 4.3705, 1.8613, 4.7067, 4.7200, 3.4410, 3.8649], device='cuda:7'), covar=tensor([0.0687, 0.0067, 0.0176, 0.1188, 0.0035, 0.0039, 0.0268, 0.0279], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0080, 0.0080, 0.0142, 0.0072, 0.0073, 0.0110, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 01:30:17,625 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9521, 4.2948, 3.4249, 2.4374, 3.0674, 2.4132, 4.5636, 4.4705], device='cuda:7'), covar=tensor([0.2058, 0.0593, 0.1048, 0.1339, 0.2196, 0.1344, 0.0301, 0.0474], device='cuda:7'), in_proj_covar=tensor([0.0268, 0.0244, 0.0258, 0.0220, 0.0269, 0.0199, 0.0221, 0.0220], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:30:21,642 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.145e+02 3.686e+02 4.713e+02 5.844e+02 8.594e+02, threshold=9.427e+02, percent-clipped=0.0 2023-04-28 01:30:36,166 INFO [train.py:904] (7/8) Epoch 4, batch 300, loss[loss=0.2269, simple_loss=0.3048, pruned_loss=0.07447, over 17104.00 frames. ], tot_loss[loss=0.2444, simple_loss=0.3142, pruned_loss=0.08728, over 2595425.97 frames. ], batch size: 53, lr: 1.75e-02, grad_scale: 1.0 2023-04-28 01:31:14,492 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7082, 2.8625, 2.3529, 4.0037, 3.6613, 3.8315, 1.6331, 2.8434], device='cuda:7'), covar=tensor([0.1344, 0.0452, 0.1142, 0.0072, 0.0244, 0.0315, 0.1254, 0.0680], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0133, 0.0161, 0.0070, 0.0149, 0.0148, 0.0152, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 01:31:16,719 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30781.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:31:17,971 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6489, 4.4935, 4.1151, 1.7350, 3.1070, 2.4115, 3.9733, 4.2763], device='cuda:7'), covar=tensor([0.0258, 0.0462, 0.0439, 0.1768, 0.0727, 0.1050, 0.0726, 0.0790], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0119, 0.0155, 0.0144, 0.0135, 0.0130, 0.0139, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-28 01:31:43,553 INFO [train.py:904] (7/8) Epoch 4, batch 350, loss[loss=0.2338, simple_loss=0.2989, pruned_loss=0.08434, over 16751.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.3106, pruned_loss=0.08526, over 2758167.72 frames. ], batch size: 62, lr: 1.74e-02, grad_scale: 1.0 2023-04-28 01:32:20,667 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30829.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:32:38,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.414e+02 3.784e+02 4.816e+02 5.903e+02 1.455e+03, threshold=9.632e+02, percent-clipped=4.0 2023-04-28 01:32:51,074 INFO [train.py:904] (7/8) Epoch 4, batch 400, loss[loss=0.1855, simple_loss=0.2661, pruned_loss=0.0525, over 16973.00 frames. ], tot_loss[loss=0.2375, simple_loss=0.3077, pruned_loss=0.08366, over 2889083.98 frames. ], batch size: 41, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:33:01,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8721, 5.4122, 5.3535, 5.2358, 5.3083, 5.8010, 5.5707, 5.3026], device='cuda:7'), covar=tensor([0.0755, 0.1430, 0.1269, 0.2050, 0.2375, 0.0932, 0.0903, 0.2099], device='cuda:7'), in_proj_covar=tensor([0.0248, 0.0356, 0.0327, 0.0304, 0.0401, 0.0368, 0.0282, 0.0407], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:33:04,250 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4943, 3.6898, 4.2624, 3.2672, 4.1107, 4.1400, 4.2195, 2.3427], device='cuda:7'), covar=tensor([0.0290, 0.0038, 0.0022, 0.0180, 0.0026, 0.0055, 0.0017, 0.0278], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0057, 0.0058, 0.0111, 0.0057, 0.0063, 0.0056, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:33:11,891 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30865.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:01,538 INFO [train.py:904] (7/8) Epoch 4, batch 450, loss[loss=0.2824, simple_loss=0.3312, pruned_loss=0.1168, over 12386.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.3047, pruned_loss=0.08182, over 2984449.73 frames. ], batch size: 248, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:34:05,035 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30903.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:18,178 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=30913.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:35,355 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30925.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:46,501 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30934.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:34:56,475 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.382e+02 3.455e+02 4.330e+02 5.569e+02 1.727e+03, threshold=8.660e+02, percent-clipped=4.0 2023-04-28 01:35:09,308 INFO [train.py:904] (7/8) Epoch 4, batch 500, loss[loss=0.236, simple_loss=0.3006, pruned_loss=0.08572, over 16436.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3034, pruned_loss=0.08091, over 3059325.29 frames. ], batch size: 75, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:35:28,338 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:35:58,994 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30986.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:36:00,296 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0156, 4.3852, 2.1655, 4.6563, 2.7924, 4.5156, 1.9896, 3.2807], device='cuda:7'), covar=tensor([0.0075, 0.0156, 0.1298, 0.0020, 0.0657, 0.0250, 0.1397, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0100, 0.0141, 0.0171, 0.0082, 0.0159, 0.0168, 0.0179, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 01:36:03,760 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4262, 2.7954, 2.4861, 4.9845, 2.1785, 4.2983, 2.6615, 2.5898], device='cuda:7'), covar=tensor([0.0321, 0.1153, 0.0653, 0.0168, 0.2226, 0.0373, 0.1166, 0.2091], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0267, 0.0218, 0.0282, 0.0329, 0.0236, 0.0246, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:36:09,653 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30995.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:36:17,830 INFO [train.py:904] (7/8) Epoch 4, batch 550, loss[loss=0.2282, simple_loss=0.294, pruned_loss=0.08122, over 16776.00 frames. ], tot_loss[loss=0.2323, simple_loss=0.3031, pruned_loss=0.08076, over 3112928.91 frames. ], batch size: 102, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:36:29,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3852, 5.7647, 5.4758, 5.5994, 4.9787, 4.7400, 5.2970, 5.8653], device='cuda:7'), covar=tensor([0.0565, 0.0711, 0.0935, 0.0387, 0.0634, 0.0627, 0.0565, 0.0675], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0436, 0.0375, 0.0272, 0.0279, 0.0280, 0.0341, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:36:33,163 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31011.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:36:33,276 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31011.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:36:33,442 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1126, 2.6291, 2.2992, 4.5738, 1.9422, 4.1611, 2.4534, 2.5552], device='cuda:7'), covar=tensor([0.0360, 0.1123, 0.0648, 0.0240, 0.2256, 0.0362, 0.1197, 0.2021], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0268, 0.0219, 0.0283, 0.0330, 0.0237, 0.0247, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:36:39,305 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31016.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:37:13,490 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.455e+02 3.696e+02 4.486e+02 5.421e+02 1.042e+03, threshold=8.971e+02, percent-clipped=3.0 2023-04-28 01:37:28,712 INFO [train.py:904] (7/8) Epoch 4, batch 600, loss[loss=0.2363, simple_loss=0.2967, pruned_loss=0.08797, over 16157.00 frames. ], tot_loss[loss=0.2316, simple_loss=0.3028, pruned_loss=0.08023, over 3143364.27 frames. ], batch size: 164, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:37:46,907 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31064.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:37:58,233 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31072.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 01:37:59,272 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1663, 4.9744, 4.9242, 4.1864, 4.9310, 1.7801, 4.6641, 4.9910], device='cuda:7'), covar=tensor([0.0045, 0.0060, 0.0072, 0.0325, 0.0052, 0.1672, 0.0077, 0.0089], device='cuda:7'), in_proj_covar=tensor([0.0078, 0.0070, 0.0107, 0.0115, 0.0077, 0.0126, 0.0094, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:38:36,824 INFO [train.py:904] (7/8) Epoch 4, batch 650, loss[loss=0.2056, simple_loss=0.2943, pruned_loss=0.05844, over 17102.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3008, pruned_loss=0.07972, over 3186060.34 frames. ], batch size: 47, lr: 1.74e-02, grad_scale: 2.0 2023-04-28 01:39:30,539 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 3.618e+02 4.311e+02 5.742e+02 1.399e+03, threshold=8.622e+02, percent-clipped=6.0 2023-04-28 01:39:45,332 INFO [train.py:904] (7/8) Epoch 4, batch 700, loss[loss=0.2161, simple_loss=0.3018, pruned_loss=0.0652, over 17131.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.3007, pruned_loss=0.07909, over 3209107.62 frames. ], batch size: 48, lr: 1.73e-02, grad_scale: 2.0 2023-04-28 01:39:48,414 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.89 vs. limit=5.0 2023-04-28 01:40:25,999 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8318, 5.2110, 5.2846, 5.2848, 5.2806, 5.8684, 5.6035, 5.2563], device='cuda:7'), covar=tensor([0.0704, 0.1410, 0.1130, 0.1659, 0.2352, 0.0815, 0.0835, 0.1833], device='cuda:7'), in_proj_covar=tensor([0.0251, 0.0360, 0.0330, 0.0300, 0.0402, 0.0353, 0.0282, 0.0401], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:40:40,943 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2914, 3.0137, 3.8901, 2.6733, 3.7345, 3.9582, 3.8511, 2.1153], device='cuda:7'), covar=tensor([0.0277, 0.0095, 0.0029, 0.0207, 0.0038, 0.0040, 0.0027, 0.0258], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0057, 0.0058, 0.0110, 0.0057, 0.0063, 0.0057, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:40:46,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6507, 4.4934, 4.1730, 1.9764, 3.1286, 2.5944, 3.8136, 4.2331], device='cuda:7'), covar=tensor([0.0224, 0.0388, 0.0366, 0.1435, 0.0664, 0.0877, 0.0619, 0.0804], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0126, 0.0156, 0.0144, 0.0136, 0.0128, 0.0139, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 01:40:53,749 INFO [train.py:904] (7/8) Epoch 4, batch 750, loss[loss=0.2131, simple_loss=0.2836, pruned_loss=0.07132, over 16858.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.3012, pruned_loss=0.07999, over 3230471.84 frames. ], batch size: 42, lr: 1.73e-02, grad_scale: 2.0 2023-04-28 01:41:19,593 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8775, 5.2777, 4.9293, 5.0316, 4.6419, 4.5392, 4.7900, 5.3320], device='cuda:7'), covar=tensor([0.0517, 0.0559, 0.0800, 0.0399, 0.0507, 0.0608, 0.0532, 0.0592], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0447, 0.0384, 0.0279, 0.0288, 0.0287, 0.0352, 0.0310], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:41:19,662 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5970, 4.4784, 4.4457, 4.4732, 4.0009, 4.5183, 4.4526, 4.1906], device='cuda:7'), covar=tensor([0.0422, 0.0291, 0.0202, 0.0175, 0.0876, 0.0266, 0.0285, 0.0369], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0152, 0.0199, 0.0165, 0.0236, 0.0187, 0.0142, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:41:22,166 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3055, 3.9889, 3.2638, 1.9548, 2.8728, 2.3259, 3.6870, 3.7485], device='cuda:7'), covar=tensor([0.0209, 0.0413, 0.0540, 0.1519, 0.0655, 0.0949, 0.0476, 0.0621], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0126, 0.0156, 0.0144, 0.0137, 0.0129, 0.0140, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 01:41:48,308 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.232e+02 3.368e+02 4.224e+02 5.033e+02 8.537e+02, threshold=8.448e+02, percent-clipped=0.0 2023-04-28 01:42:01,632 INFO [train.py:904] (7/8) Epoch 4, batch 800, loss[loss=0.2347, simple_loss=0.2985, pruned_loss=0.08541, over 16803.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.3002, pruned_loss=0.0797, over 3258339.13 frames. ], batch size: 124, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:42:14,349 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31259.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:42:44,482 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31281.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:42:56,310 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31290.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:43:11,603 INFO [train.py:904] (7/8) Epoch 4, batch 850, loss[loss=0.1865, simple_loss=0.2667, pruned_loss=0.05318, over 16810.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.2996, pruned_loss=0.07939, over 3254700.05 frames. ], batch size: 42, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:43:15,597 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4154, 1.4130, 1.9602, 2.2849, 2.5137, 2.5135, 1.3511, 2.4343], device='cuda:7'), covar=tensor([0.0075, 0.0185, 0.0121, 0.0100, 0.0058, 0.0078, 0.0181, 0.0048], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0131, 0.0116, 0.0112, 0.0106, 0.0081, 0.0124, 0.0070], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 01:43:18,676 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 01:43:24,794 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31311.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:43:38,534 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2452, 4.4226, 1.6809, 4.6972, 2.6338, 4.4868, 2.2500, 3.1796], device='cuda:7'), covar=tensor([0.0080, 0.0162, 0.1606, 0.0023, 0.0858, 0.0294, 0.1186, 0.0488], device='cuda:7'), in_proj_covar=tensor([0.0101, 0.0143, 0.0170, 0.0081, 0.0160, 0.0171, 0.0178, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 01:44:07,362 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.752e+02 3.699e+02 4.562e+02 5.706e+02 1.340e+03, threshold=9.123e+02, percent-clipped=5.0 2023-04-28 01:44:19,707 INFO [train.py:904] (7/8) Epoch 4, batch 900, loss[loss=0.1778, simple_loss=0.2555, pruned_loss=0.05008, over 15836.00 frames. ], tot_loss[loss=0.2278, simple_loss=0.2985, pruned_loss=0.07852, over 3272451.01 frames. ], batch size: 35, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:44:26,942 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-28 01:44:32,836 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31359.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:44:42,772 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31367.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:44:51,468 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7794, 3.3983, 2.8357, 5.0856, 4.8880, 4.2756, 1.7560, 3.3922], device='cuda:7'), covar=tensor([0.1407, 0.0462, 0.1122, 0.0067, 0.0217, 0.0356, 0.1322, 0.0616], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0135, 0.0165, 0.0072, 0.0160, 0.0154, 0.0155, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 01:45:31,318 INFO [train.py:904] (7/8) Epoch 4, batch 950, loss[loss=0.243, simple_loss=0.3003, pruned_loss=0.09285, over 16899.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.2989, pruned_loss=0.07762, over 3295010.71 frames. ], batch size: 116, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:45:41,810 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31409.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:46:26,072 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.766e+02 3.401e+02 4.040e+02 4.698e+02 9.697e+02, threshold=8.080e+02, percent-clipped=2.0 2023-04-28 01:46:38,680 INFO [train.py:904] (7/8) Epoch 4, batch 1000, loss[loss=0.2021, simple_loss=0.2807, pruned_loss=0.06178, over 17179.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.2973, pruned_loss=0.07765, over 3305287.01 frames. ], batch size: 46, lr: 1.73e-02, grad_scale: 4.0 2023-04-28 01:47:05,203 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9633, 3.3022, 2.6784, 4.9261, 4.5954, 4.3338, 1.8254, 3.3096], device='cuda:7'), covar=tensor([0.1261, 0.0489, 0.1048, 0.0053, 0.0230, 0.0275, 0.1239, 0.0589], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0137, 0.0166, 0.0074, 0.0163, 0.0156, 0.0156, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 01:47:06,315 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31470.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:47:26,771 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4753, 4.5899, 4.9761, 4.9652, 4.9482, 4.6269, 4.4805, 4.4038], device='cuda:7'), covar=tensor([0.0265, 0.0434, 0.0349, 0.0415, 0.0387, 0.0281, 0.0841, 0.0366], device='cuda:7'), in_proj_covar=tensor([0.0212, 0.0212, 0.0219, 0.0216, 0.0259, 0.0231, 0.0323, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 01:47:36,297 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31491.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:47:37,912 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-28 01:47:39,629 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9050, 3.7182, 3.8769, 4.1562, 4.1721, 3.7057, 4.0363, 4.2025], device='cuda:7'), covar=tensor([0.0715, 0.0690, 0.1146, 0.0426, 0.0448, 0.1281, 0.0851, 0.0373], device='cuda:7'), in_proj_covar=tensor([0.0353, 0.0424, 0.0569, 0.0438, 0.0330, 0.0315, 0.0345, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:47:49,005 INFO [train.py:904] (7/8) Epoch 4, batch 1050, loss[loss=0.2208, simple_loss=0.2991, pruned_loss=0.07124, over 16703.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.2968, pruned_loss=0.07721, over 3318663.46 frames. ], batch size: 57, lr: 1.72e-02, grad_scale: 4.0 2023-04-28 01:48:45,343 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 3.437e+02 4.188e+02 5.032e+02 1.579e+03, threshold=8.377e+02, percent-clipped=2.0 2023-04-28 01:49:00,858 INFO [train.py:904] (7/8) Epoch 4, batch 1100, loss[loss=0.2164, simple_loss=0.2901, pruned_loss=0.07134, over 16244.00 frames. ], tot_loss[loss=0.224, simple_loss=0.2953, pruned_loss=0.07639, over 3312709.56 frames. ], batch size: 36, lr: 1.72e-02, grad_scale: 4.0 2023-04-28 01:49:02,459 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31552.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 01:49:11,820 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31559.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:49:41,980 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31581.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:49:55,774 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31590.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:50:09,609 INFO [train.py:904] (7/8) Epoch 4, batch 1150, loss[loss=0.2245, simple_loss=0.2868, pruned_loss=0.08113, over 16895.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2947, pruned_loss=0.07531, over 3315775.22 frames. ], batch size: 109, lr: 1.72e-02, grad_scale: 4.0 2023-04-28 01:50:19,073 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31607.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:50:49,049 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31629.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:51:00,558 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31638.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:51:04,258 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.256e+02 3.893e+02 4.649e+02 5.891e+02 1.405e+03, threshold=9.297e+02, percent-clipped=5.0 2023-04-28 01:51:19,081 INFO [train.py:904] (7/8) Epoch 4, batch 1200, loss[loss=0.2455, simple_loss=0.3033, pruned_loss=0.09388, over 16477.00 frames. ], tot_loss[loss=0.222, simple_loss=0.2942, pruned_loss=0.07486, over 3314016.43 frames. ], batch size: 146, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:51:40,991 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31667.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 01:51:43,875 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 01:51:56,167 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4704, 5.7826, 5.5091, 5.6624, 5.1203, 4.7810, 5.3200, 5.8416], device='cuda:7'), covar=tensor([0.0480, 0.0649, 0.0807, 0.0336, 0.0563, 0.0516, 0.0551, 0.0571], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0439, 0.0375, 0.0274, 0.0282, 0.0279, 0.0347, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 01:52:27,377 INFO [train.py:904] (7/8) Epoch 4, batch 1250, loss[loss=0.239, simple_loss=0.3063, pruned_loss=0.08584, over 15433.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.295, pruned_loss=0.07568, over 3321072.48 frames. ], batch size: 190, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:52:31,896 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31704.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:52:47,967 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=31715.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 01:53:22,527 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.448e+02 3.415e+02 4.132e+02 5.466e+02 7.512e+02, threshold=8.265e+02, percent-clipped=0.0 2023-04-28 01:53:35,323 INFO [train.py:904] (7/8) Epoch 4, batch 1300, loss[loss=0.2313, simple_loss=0.3076, pruned_loss=0.07744, over 17124.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.2953, pruned_loss=0.07608, over 3314495.65 frames. ], batch size: 47, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:53:43,930 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2391, 5.1593, 4.9363, 4.8864, 4.5220, 5.0129, 5.0172, 4.5969], device='cuda:7'), covar=tensor([0.0302, 0.0188, 0.0160, 0.0154, 0.0716, 0.0207, 0.0187, 0.0369], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0158, 0.0207, 0.0174, 0.0245, 0.0190, 0.0148, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 01:53:55,322 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:53:55,492 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:54:26,647 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31788.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:54:38,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1301, 4.2517, 1.8066, 4.5542, 2.7581, 4.4561, 1.8887, 3.1139], device='cuda:7'), covar=tensor([0.0085, 0.0201, 0.1507, 0.0032, 0.0688, 0.0267, 0.1419, 0.0525], device='cuda:7'), in_proj_covar=tensor([0.0107, 0.0147, 0.0175, 0.0084, 0.0160, 0.0178, 0.0183, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 01:54:44,576 INFO [train.py:904] (7/8) Epoch 4, batch 1350, loss[loss=0.1931, simple_loss=0.2632, pruned_loss=0.06147, over 16971.00 frames. ], tot_loss[loss=0.224, simple_loss=0.2956, pruned_loss=0.07622, over 3306013.03 frames. ], batch size: 41, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:54:45,469 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 01:55:02,656 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-28 01:55:14,561 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31823.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:55:39,674 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.154e+02 3.282e+02 4.037e+02 5.139e+02 9.544e+02, threshold=8.075e+02, percent-clipped=2.0 2023-04-28 01:55:48,228 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31847.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 01:55:50,742 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:55:52,766 INFO [train.py:904] (7/8) Epoch 4, batch 1400, loss[loss=0.1882, simple_loss=0.276, pruned_loss=0.05024, over 17225.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.2944, pruned_loss=0.07527, over 3311803.43 frames. ], batch size: 46, lr: 1.72e-02, grad_scale: 8.0 2023-04-28 01:55:56,027 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3981, 3.9745, 3.8041, 1.6535, 4.0555, 3.9461, 3.2108, 2.9124], device='cuda:7'), covar=tensor([0.0774, 0.0060, 0.0145, 0.1232, 0.0049, 0.0062, 0.0269, 0.0420], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0080, 0.0079, 0.0143, 0.0073, 0.0076, 0.0109, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 01:56:32,807 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31878.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:56:40,551 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31884.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:57:02,754 INFO [train.py:904] (7/8) Epoch 4, batch 1450, loss[loss=0.2477, simple_loss=0.2995, pruned_loss=0.09799, over 16286.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.2931, pruned_loss=0.07517, over 3302017.03 frames. ], batch size: 165, lr: 1.71e-02, grad_scale: 4.0 2023-04-28 01:57:22,471 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31915.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:57:57,397 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31939.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:58:00,354 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.224e+02 3.023e+02 3.906e+02 5.046e+02 9.760e+02, threshold=7.812e+02, percent-clipped=3.0 2023-04-28 01:58:12,750 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31950.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:58:13,464 INFO [train.py:904] (7/8) Epoch 4, batch 1500, loss[loss=0.2386, simple_loss=0.2952, pruned_loss=0.09097, over 16782.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2935, pruned_loss=0.07584, over 3304383.41 frames. ], batch size: 83, lr: 1.71e-02, grad_scale: 4.0 2023-04-28 01:58:48,072 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 01:59:26,629 INFO [train.py:904] (7/8) Epoch 4, batch 1550, loss[loss=0.2622, simple_loss=0.3215, pruned_loss=0.1015, over 12000.00 frames. ], tot_loss[loss=0.2257, simple_loss=0.2965, pruned_loss=0.07742, over 3306131.80 frames. ], batch size: 248, lr: 1.71e-02, grad_scale: 4.0 2023-04-28 01:59:40,583 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=32011.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:00:21,779 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.322e+02 3.491e+02 4.515e+02 5.293e+02 8.780e+02, threshold=9.030e+02, percent-clipped=2.0 2023-04-28 02:00:30,555 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0092, 2.0122, 2.3408, 2.8264, 2.5841, 3.3052, 2.1535, 3.2252], device='cuda:7'), covar=tensor([0.0059, 0.0174, 0.0133, 0.0109, 0.0099, 0.0064, 0.0160, 0.0044], device='cuda:7'), in_proj_covar=tensor([0.0103, 0.0128, 0.0116, 0.0113, 0.0108, 0.0081, 0.0124, 0.0070], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 02:00:34,395 INFO [train.py:904] (7/8) Epoch 4, batch 1600, loss[loss=0.2599, simple_loss=0.3224, pruned_loss=0.09875, over 15545.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.299, pruned_loss=0.07876, over 3307673.45 frames. ], batch size: 190, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:00:37,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5887, 4.3773, 4.0891, 2.0807, 2.9555, 2.5753, 3.8912, 4.3323], device='cuda:7'), covar=tensor([0.0298, 0.0453, 0.0427, 0.1467, 0.0691, 0.0934, 0.0633, 0.0733], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0131, 0.0153, 0.0143, 0.0135, 0.0127, 0.0140, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 02:00:46,084 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32060.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:00:52,813 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32065.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:01:41,678 INFO [train.py:904] (7/8) Epoch 4, batch 1650, loss[loss=0.2547, simple_loss=0.3152, pruned_loss=0.09714, over 16442.00 frames. ], tot_loss[loss=0.2279, simple_loss=0.2996, pruned_loss=0.07817, over 3310501.33 frames. ], batch size: 146, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:01:58,773 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32113.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:02:37,465 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.433e+02 3.425e+02 4.282e+02 5.685e+02 1.161e+03, threshold=8.563e+02, percent-clipped=3.0 2023-04-28 02:02:40,716 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:02:42,134 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0286, 4.9590, 4.7490, 4.1379, 4.8889, 1.6686, 4.5959, 4.9522], device='cuda:7'), covar=tensor([0.0050, 0.0051, 0.0088, 0.0325, 0.0058, 0.1525, 0.0085, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0076, 0.0114, 0.0125, 0.0084, 0.0128, 0.0102, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:02:45,698 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32147.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 02:02:50,065 INFO [train.py:904] (7/8) Epoch 4, batch 1700, loss[loss=0.2489, simple_loss=0.3177, pruned_loss=0.09001, over 17187.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3014, pruned_loss=0.07927, over 3316141.36 frames. ], batch size: 46, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:03:27,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8235, 3.0013, 2.5718, 4.1796, 3.9546, 3.9626, 1.6066, 2.9637], device='cuda:7'), covar=tensor([0.1127, 0.0420, 0.1010, 0.0054, 0.0230, 0.0287, 0.1178, 0.0617], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0136, 0.0163, 0.0075, 0.0169, 0.0156, 0.0154, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 02:03:29,683 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32179.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:03:37,104 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8908, 3.7572, 2.8266, 5.2729, 5.0621, 4.5432, 1.8222, 3.4556], device='cuda:7'), covar=tensor([0.1240, 0.0414, 0.1111, 0.0049, 0.0217, 0.0296, 0.1333, 0.0584], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0136, 0.0164, 0.0075, 0.0169, 0.0156, 0.0154, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 02:03:53,202 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32195.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 02:04:01,139 INFO [train.py:904] (7/8) Epoch 4, batch 1750, loss[loss=0.2685, simple_loss=0.3314, pruned_loss=0.1027, over 15417.00 frames. ], tot_loss[loss=0.2309, simple_loss=0.3028, pruned_loss=0.07951, over 3315725.12 frames. ], batch size: 191, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:04:48,293 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32234.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:04:58,763 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.044e+02 3.415e+02 4.164e+02 5.127e+02 8.958e+02, threshold=8.329e+02, percent-clipped=1.0 2023-04-28 02:05:00,246 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 02:05:01,174 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3109, 3.8756, 3.7500, 1.8651, 3.9847, 3.9949, 3.0548, 2.9860], device='cuda:7'), covar=tensor([0.0804, 0.0076, 0.0133, 0.1150, 0.0066, 0.0060, 0.0292, 0.0397], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0083, 0.0081, 0.0146, 0.0075, 0.0078, 0.0113, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 02:05:11,465 INFO [train.py:904] (7/8) Epoch 4, batch 1800, loss[loss=0.2394, simple_loss=0.3138, pruned_loss=0.08249, over 16440.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.303, pruned_loss=0.07859, over 3323591.51 frames. ], batch size: 146, lr: 1.71e-02, grad_scale: 8.0 2023-04-28 02:05:36,993 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2023-04-28 02:05:39,584 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32271.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:06:18,945 INFO [train.py:904] (7/8) Epoch 4, batch 1850, loss[loss=0.223, simple_loss=0.3019, pruned_loss=0.07206, over 17049.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.303, pruned_loss=0.07764, over 3327902.02 frames. ], batch size: 55, lr: 1.70e-02, grad_scale: 4.0 2023-04-28 02:06:26,661 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32306.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:07:18,155 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.462e+02 3.616e+02 4.272e+02 5.393e+02 1.345e+03, threshold=8.544e+02, percent-clipped=8.0 2023-04-28 02:07:29,373 INFO [train.py:904] (7/8) Epoch 4, batch 1900, loss[loss=0.2091, simple_loss=0.2796, pruned_loss=0.06931, over 16673.00 frames. ], tot_loss[loss=0.2276, simple_loss=0.3017, pruned_loss=0.0767, over 3334870.22 frames. ], batch size: 89, lr: 1.70e-02, grad_scale: 4.0 2023-04-28 02:07:43,024 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:07:59,489 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 02:08:41,345 INFO [train.py:904] (7/8) Epoch 4, batch 1950, loss[loss=0.22, simple_loss=0.3062, pruned_loss=0.0669, over 17134.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.3018, pruned_loss=0.07622, over 3324061.81 frames. ], batch size: 49, lr: 1.70e-02, grad_scale: 4.0 2023-04-28 02:08:50,797 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32408.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:09:19,250 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-28 02:09:37,983 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.345e+02 3.206e+02 3.800e+02 4.937e+02 1.195e+03, threshold=7.601e+02, percent-clipped=2.0 2023-04-28 02:09:39,394 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32444.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:09:47,997 INFO [train.py:904] (7/8) Epoch 4, batch 2000, loss[loss=0.2597, simple_loss=0.3146, pruned_loss=0.1024, over 16670.00 frames. ], tot_loss[loss=0.228, simple_loss=0.3031, pruned_loss=0.07643, over 3314481.81 frames. ], batch size: 134, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:10:03,128 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0887, 4.5748, 4.6456, 3.5592, 4.3604, 4.6570, 4.2547, 2.7642], device='cuda:7'), covar=tensor([0.0227, 0.0019, 0.0022, 0.0170, 0.0026, 0.0035, 0.0025, 0.0250], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0056, 0.0059, 0.0110, 0.0060, 0.0064, 0.0061, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:10:06,228 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5590, 3.4439, 2.7885, 2.1099, 2.5785, 2.0519, 3.4605, 3.4926], device='cuda:7'), covar=tensor([0.1789, 0.0580, 0.1024, 0.1370, 0.1951, 0.1363, 0.0404, 0.0620], device='cuda:7'), in_proj_covar=tensor([0.0273, 0.0251, 0.0260, 0.0227, 0.0299, 0.0198, 0.0228, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:10:28,244 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32479.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:10:45,039 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:10:58,297 INFO [train.py:904] (7/8) Epoch 4, batch 2050, loss[loss=0.2716, simple_loss=0.3287, pruned_loss=0.1072, over 16357.00 frames. ], tot_loss[loss=0.2288, simple_loss=0.3029, pruned_loss=0.07735, over 3314031.36 frames. ], batch size: 165, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:11:34,837 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32527.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:11:44,391 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32534.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:11:57,877 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.075e+02 3.479e+02 4.226e+02 5.349e+02 1.056e+03, threshold=8.453e+02, percent-clipped=7.0 2023-04-28 02:12:07,692 INFO [train.py:904] (7/8) Epoch 4, batch 2100, loss[loss=0.2339, simple_loss=0.3198, pruned_loss=0.07405, over 17134.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3033, pruned_loss=0.07737, over 3311491.78 frames. ], batch size: 48, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:12:36,522 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32571.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:12:51,051 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32582.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:13:17,659 INFO [train.py:904] (7/8) Epoch 4, batch 2150, loss[loss=0.1882, simple_loss=0.266, pruned_loss=0.05522, over 15891.00 frames. ], tot_loss[loss=0.2288, simple_loss=0.3029, pruned_loss=0.07738, over 3320979.80 frames. ], batch size: 35, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:13:25,183 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32606.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:13:29,949 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7379, 2.2653, 1.7740, 1.9208, 2.8003, 2.6406, 3.0889, 2.9718], device='cuda:7'), covar=tensor([0.0042, 0.0140, 0.0189, 0.0178, 0.0077, 0.0120, 0.0070, 0.0086], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0138, 0.0137, 0.0136, 0.0132, 0.0140, 0.0104, 0.0119], device='cuda:7'), out_proj_covar=tensor([9.9993e-05, 1.8563e-04, 1.7884e-04, 1.8110e-04, 1.8045e-04, 1.9034e-04, 1.4055e-04, 1.6348e-04], device='cuda:7') 2023-04-28 02:13:42,065 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32619.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:14:15,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.513e+02 3.522e+02 4.278e+02 5.461e+02 9.146e+02, threshold=8.556e+02, percent-clipped=3.0 2023-04-28 02:14:26,831 INFO [train.py:904] (7/8) Epoch 4, batch 2200, loss[loss=0.2222, simple_loss=0.3056, pruned_loss=0.06943, over 17031.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.3034, pruned_loss=0.07796, over 3321761.48 frames. ], batch size: 50, lr: 1.70e-02, grad_scale: 8.0 2023-04-28 02:14:30,449 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=32654.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:14:35,813 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3369, 4.1530, 4.3199, 4.5906, 4.6299, 4.1358, 4.3652, 4.5726], device='cuda:7'), covar=tensor([0.0667, 0.0561, 0.0983, 0.0403, 0.0369, 0.0811, 0.1015, 0.0363], device='cuda:7'), in_proj_covar=tensor([0.0357, 0.0431, 0.0584, 0.0455, 0.0336, 0.0330, 0.0352, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:14:37,000 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4141, 5.7644, 5.5316, 5.6663, 5.0912, 4.7400, 5.2895, 5.9134], device='cuda:7'), covar=tensor([0.0582, 0.0575, 0.0736, 0.0344, 0.0506, 0.0586, 0.0512, 0.0596], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0450, 0.0382, 0.0282, 0.0287, 0.0284, 0.0353, 0.0311], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:15:34,032 INFO [train.py:904] (7/8) Epoch 4, batch 2250, loss[loss=0.2427, simple_loss=0.3255, pruned_loss=0.07992, over 17122.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3044, pruned_loss=0.07876, over 3321300.17 frames. ], batch size: 48, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:16:32,023 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.152e+02 3.558e+02 4.554e+02 5.446e+02 8.366e+02, threshold=9.108e+02, percent-clipped=0.0 2023-04-28 02:16:42,208 INFO [train.py:904] (7/8) Epoch 4, batch 2300, loss[loss=0.2117, simple_loss=0.2945, pruned_loss=0.06448, over 17137.00 frames. ], tot_loss[loss=0.2319, simple_loss=0.3054, pruned_loss=0.07925, over 3320354.03 frames. ], batch size: 48, lr: 1.69e-02, grad_scale: 4.0 2023-04-28 02:17:10,595 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 02:17:20,773 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 02:17:34,762 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-28 02:17:51,384 INFO [train.py:904] (7/8) Epoch 4, batch 2350, loss[loss=0.2198, simple_loss=0.301, pruned_loss=0.06924, over 17108.00 frames. ], tot_loss[loss=0.2337, simple_loss=0.3063, pruned_loss=0.08055, over 3320324.92 frames. ], batch size: 47, lr: 1.69e-02, grad_scale: 4.0 2023-04-28 02:18:01,859 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5624, 4.5728, 4.4910, 3.9540, 4.4907, 2.0117, 4.2997, 4.4426], device='cuda:7'), covar=tensor([0.0058, 0.0050, 0.0067, 0.0263, 0.0052, 0.1254, 0.0076, 0.0115], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0078, 0.0115, 0.0125, 0.0085, 0.0126, 0.0102, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:18:48,984 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.340e+02 3.303e+02 3.984e+02 5.110e+02 1.375e+03, threshold=7.969e+02, percent-clipped=3.0 2023-04-28 02:18:57,211 INFO [train.py:904] (7/8) Epoch 4, batch 2400, loss[loss=0.2557, simple_loss=0.3164, pruned_loss=0.09746, over 16860.00 frames. ], tot_loss[loss=0.2337, simple_loss=0.3071, pruned_loss=0.08016, over 3330454.58 frames. ], batch size: 96, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:20:06,101 INFO [train.py:904] (7/8) Epoch 4, batch 2450, loss[loss=0.247, simple_loss=0.3243, pruned_loss=0.08483, over 16720.00 frames. ], tot_loss[loss=0.2332, simple_loss=0.307, pruned_loss=0.07973, over 3321673.67 frames. ], batch size: 57, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:20:17,574 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 02:21:03,695 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 3.644e+02 4.364e+02 5.421e+02 9.049e+02, threshold=8.728e+02, percent-clipped=5.0 2023-04-28 02:21:13,291 INFO [train.py:904] (7/8) Epoch 4, batch 2500, loss[loss=0.2148, simple_loss=0.2929, pruned_loss=0.06833, over 16749.00 frames. ], tot_loss[loss=0.2332, simple_loss=0.3072, pruned_loss=0.07965, over 3314196.69 frames. ], batch size: 39, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:21:34,416 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=32966.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:21:43,865 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8869, 4.6025, 4.8376, 5.1750, 5.3376, 4.5712, 5.2678, 5.2066], device='cuda:7'), covar=tensor([0.0878, 0.0805, 0.1385, 0.0510, 0.0404, 0.0638, 0.0390, 0.0372], device='cuda:7'), in_proj_covar=tensor([0.0352, 0.0420, 0.0572, 0.0445, 0.0328, 0.0320, 0.0348, 0.0360], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:21:49,092 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 02:22:21,149 INFO [train.py:904] (7/8) Epoch 4, batch 2550, loss[loss=0.2097, simple_loss=0.2986, pruned_loss=0.06043, over 17135.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.3081, pruned_loss=0.07989, over 3313481.62 frames. ], batch size: 48, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:22:26,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0757, 5.3519, 5.0706, 5.2117, 4.7610, 4.4823, 4.8583, 5.4772], device='cuda:7'), covar=tensor([0.0587, 0.0571, 0.0755, 0.0352, 0.0499, 0.0660, 0.0511, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0317, 0.0443, 0.0378, 0.0277, 0.0280, 0.0277, 0.0348, 0.0305], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:22:57,704 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33027.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:23:20,062 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.536e+02 3.626e+02 4.489e+02 5.750e+02 1.218e+03, threshold=8.978e+02, percent-clipped=4.0 2023-04-28 02:23:30,994 INFO [train.py:904] (7/8) Epoch 4, batch 2600, loss[loss=0.2033, simple_loss=0.2946, pruned_loss=0.05597, over 17106.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.3081, pruned_loss=0.0793, over 3315439.43 frames. ], batch size: 48, lr: 1.69e-02, grad_scale: 8.0 2023-04-28 02:23:55,332 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33069.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:24:33,418 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-28 02:24:39,074 INFO [train.py:904] (7/8) Epoch 4, batch 2650, loss[loss=0.2749, simple_loss=0.3339, pruned_loss=0.108, over 16862.00 frames. ], tot_loss[loss=0.2327, simple_loss=0.3081, pruned_loss=0.07868, over 3316431.68 frames. ], batch size: 109, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:24:42,000 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-04-28 02:25:19,554 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33130.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:25:38,560 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.285e+02 4.083e+02 4.956e+02 1.136e+03, threshold=8.166e+02, percent-clipped=3.0 2023-04-28 02:25:46,697 INFO [train.py:904] (7/8) Epoch 4, batch 2700, loss[loss=0.2068, simple_loss=0.2909, pruned_loss=0.0614, over 16852.00 frames. ], tot_loss[loss=0.2315, simple_loss=0.3074, pruned_loss=0.07784, over 3320025.61 frames. ], batch size: 42, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:25:54,308 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33155.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:26:19,202 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8964, 4.5773, 4.8577, 5.1091, 5.2575, 4.5066, 5.2672, 5.2146], device='cuda:7'), covar=tensor([0.0670, 0.0626, 0.1212, 0.0497, 0.0419, 0.0728, 0.0360, 0.0369], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0426, 0.0580, 0.0451, 0.0337, 0.0324, 0.0351, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:26:56,826 INFO [train.py:904] (7/8) Epoch 4, batch 2750, loss[loss=0.2231, simple_loss=0.2908, pruned_loss=0.07767, over 16702.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3067, pruned_loss=0.07666, over 3324832.36 frames. ], batch size: 134, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:27:01,034 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 02:27:13,913 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5681, 3.4969, 3.9766, 3.9554, 3.9474, 3.6427, 3.7366, 3.6173], device='cuda:7'), covar=tensor([0.0272, 0.0459, 0.0322, 0.0411, 0.0427, 0.0341, 0.0644, 0.0392], device='cuda:7'), in_proj_covar=tensor([0.0222, 0.0220, 0.0224, 0.0228, 0.0277, 0.0236, 0.0343, 0.0206], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 02:27:16,327 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33216.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:27:54,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.476e+02 3.299e+02 3.913e+02 4.982e+02 9.564e+02, threshold=7.826e+02, percent-clipped=3.0 2023-04-28 02:28:04,364 INFO [train.py:904] (7/8) Epoch 4, batch 2800, loss[loss=0.2222, simple_loss=0.3059, pruned_loss=0.06927, over 16377.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.306, pruned_loss=0.07661, over 3326651.78 frames. ], batch size: 68, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:28:46,304 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8762, 3.9304, 3.1042, 2.3420, 3.0052, 2.3459, 4.0625, 4.1510], device='cuda:7'), covar=tensor([0.1687, 0.0518, 0.0980, 0.1250, 0.1909, 0.1158, 0.0358, 0.0500], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0248, 0.0257, 0.0228, 0.0297, 0.0194, 0.0229, 0.0244], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:29:10,596 INFO [train.py:904] (7/8) Epoch 4, batch 2850, loss[loss=0.21, simple_loss=0.2991, pruned_loss=0.06044, over 17291.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.3065, pruned_loss=0.07727, over 3323518.81 frames. ], batch size: 52, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:29:18,158 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9806, 3.0490, 2.6913, 4.5280, 4.2867, 4.2836, 1.6618, 3.0143], device='cuda:7'), covar=tensor([0.1147, 0.0432, 0.0917, 0.0060, 0.0201, 0.0198, 0.1124, 0.0628], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0136, 0.0162, 0.0078, 0.0168, 0.0156, 0.0152, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 02:29:39,690 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33322.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:30:09,815 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.527e+02 4.279e+02 5.315e+02 1.559e+03, threshold=8.559e+02, percent-clipped=4.0 2023-04-28 02:30:20,491 INFO [train.py:904] (7/8) Epoch 4, batch 2900, loss[loss=0.2174, simple_loss=0.2925, pruned_loss=0.07119, over 16772.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.305, pruned_loss=0.0767, over 3332571.49 frames. ], batch size: 39, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:30:47,028 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6126, 3.7157, 2.8326, 2.4404, 2.8148, 2.0788, 3.7378, 3.8536], device='cuda:7'), covar=tensor([0.1722, 0.0487, 0.0902, 0.1139, 0.1814, 0.1239, 0.0354, 0.0501], device='cuda:7'), in_proj_covar=tensor([0.0265, 0.0247, 0.0257, 0.0225, 0.0296, 0.0193, 0.0226, 0.0241], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:31:20,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3760, 3.4928, 3.9254, 2.7045, 3.5850, 3.9039, 3.8533, 2.0475], device='cuda:7'), covar=tensor([0.0257, 0.0060, 0.0018, 0.0179, 0.0036, 0.0031, 0.0023, 0.0265], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0054, 0.0057, 0.0106, 0.0059, 0.0063, 0.0058, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:31:29,007 INFO [train.py:904] (7/8) Epoch 4, batch 2950, loss[loss=0.1917, simple_loss=0.2629, pruned_loss=0.06025, over 16817.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3044, pruned_loss=0.0768, over 3342361.14 frames. ], batch size: 39, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:31:46,400 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.91 vs. limit=5.0 2023-04-28 02:32:01,383 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33425.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:32:27,167 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.112e+02 3.646e+02 4.580e+02 6.032e+02 1.054e+03, threshold=9.160e+02, percent-clipped=6.0 2023-04-28 02:32:35,895 INFO [train.py:904] (7/8) Epoch 4, batch 3000, loss[loss=0.1878, simple_loss=0.2754, pruned_loss=0.05011, over 17221.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3047, pruned_loss=0.07739, over 3341067.78 frames. ], batch size: 44, lr: 1.68e-02, grad_scale: 8.0 2023-04-28 02:32:35,895 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 02:32:45,556 INFO [train.py:938] (7/8) Epoch 4, validation: loss=0.1627, simple_loss=0.2694, pruned_loss=0.02796, over 944034.00 frames. 2023-04-28 02:32:45,557 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 02:32:47,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9979, 3.4468, 3.0185, 5.1671, 4.9520, 4.3382, 1.7178, 3.2828], device='cuda:7'), covar=tensor([0.1182, 0.0425, 0.0926, 0.0044, 0.0200, 0.0319, 0.1213, 0.0629], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0137, 0.0164, 0.0079, 0.0171, 0.0159, 0.0155, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 02:33:13,581 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33472.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:33:18,321 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0887, 1.5599, 2.4340, 2.8721, 2.7700, 3.3055, 1.7581, 2.9909], device='cuda:7'), covar=tensor([0.0050, 0.0205, 0.0108, 0.0094, 0.0067, 0.0055, 0.0176, 0.0067], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0133, 0.0121, 0.0114, 0.0112, 0.0082, 0.0128, 0.0073], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 02:33:54,601 INFO [train.py:904] (7/8) Epoch 4, batch 3050, loss[loss=0.2068, simple_loss=0.2973, pruned_loss=0.05819, over 17068.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3038, pruned_loss=0.07672, over 3348657.10 frames. ], batch size: 50, lr: 1.67e-02, grad_scale: 4.0 2023-04-28 02:34:07,910 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33511.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:34:09,317 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2160, 3.9189, 3.0317, 5.3700, 5.1101, 4.2912, 2.1600, 3.2204], device='cuda:7'), covar=tensor([0.0937, 0.0327, 0.0791, 0.0032, 0.0204, 0.0290, 0.0923, 0.0576], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0137, 0.0165, 0.0079, 0.0171, 0.0156, 0.0155, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 02:34:38,080 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33533.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:34:54,166 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 3.585e+02 4.279e+02 5.215e+02 1.682e+03, threshold=8.559e+02, percent-clipped=1.0 2023-04-28 02:34:56,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0409, 4.9516, 4.8264, 4.7850, 4.3852, 4.8674, 4.8426, 4.5017], device='cuda:7'), covar=tensor([0.0329, 0.0182, 0.0144, 0.0136, 0.0751, 0.0221, 0.0198, 0.0335], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0170, 0.0215, 0.0182, 0.0254, 0.0205, 0.0156, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:35:02,728 INFO [train.py:904] (7/8) Epoch 4, batch 3100, loss[loss=0.2099, simple_loss=0.2994, pruned_loss=0.06022, over 17035.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3035, pruned_loss=0.07723, over 3340474.27 frames. ], batch size: 50, lr: 1.67e-02, grad_scale: 4.0 2023-04-28 02:35:52,375 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3399, 4.0906, 4.3067, 4.6261, 4.6443, 4.1667, 4.4111, 4.6052], device='cuda:7'), covar=tensor([0.0727, 0.0650, 0.1232, 0.0437, 0.0470, 0.0786, 0.1125, 0.0397], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0435, 0.0583, 0.0458, 0.0342, 0.0321, 0.0354, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:36:10,682 INFO [train.py:904] (7/8) Epoch 4, batch 3150, loss[loss=0.214, simple_loss=0.2845, pruned_loss=0.07179, over 16879.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.3033, pruned_loss=0.07645, over 3346418.85 frames. ], batch size: 96, lr: 1.67e-02, grad_scale: 4.0 2023-04-28 02:36:39,503 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33622.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:37:08,964 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 02:37:11,332 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.300e+02 3.211e+02 3.811e+02 4.570e+02 1.076e+03, threshold=7.622e+02, percent-clipped=2.0 2023-04-28 02:37:18,486 INFO [train.py:904] (7/8) Epoch 4, batch 3200, loss[loss=0.1977, simple_loss=0.2816, pruned_loss=0.05688, over 17046.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3016, pruned_loss=0.07531, over 3340898.08 frames. ], batch size: 41, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:37:42,785 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1333, 3.8134, 3.6545, 1.7617, 3.8651, 3.9830, 3.1973, 2.8881], device='cuda:7'), covar=tensor([0.0851, 0.0081, 0.0165, 0.1199, 0.0073, 0.0071, 0.0287, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0086, 0.0086, 0.0146, 0.0076, 0.0078, 0.0116, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 02:37:44,876 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=33670.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:37:45,243 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.87 vs. limit=5.0 2023-04-28 02:37:49,881 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 02:37:54,592 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:38:13,241 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4939, 3.3534, 3.8341, 3.0152, 3.4886, 3.8360, 3.7740, 1.9222], device='cuda:7'), covar=tensor([0.0238, 0.0056, 0.0022, 0.0131, 0.0038, 0.0034, 0.0024, 0.0269], device='cuda:7'), in_proj_covar=tensor([0.0106, 0.0052, 0.0056, 0.0104, 0.0057, 0.0062, 0.0057, 0.0102], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:38:14,645 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 02:38:25,340 INFO [train.py:904] (7/8) Epoch 4, batch 3250, loss[loss=0.2292, simple_loss=0.3337, pruned_loss=0.06235, over 17141.00 frames. ], tot_loss[loss=0.2265, simple_loss=0.302, pruned_loss=0.07547, over 3345606.00 frames. ], batch size: 48, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:38:38,378 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6082, 2.1676, 1.7053, 2.0574, 2.7118, 2.6251, 2.7614, 2.9062], device='cuda:7'), covar=tensor([0.0042, 0.0153, 0.0180, 0.0168, 0.0071, 0.0114, 0.0073, 0.0074], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0139, 0.0137, 0.0136, 0.0132, 0.0141, 0.0107, 0.0118], device='cuda:7'), out_proj_covar=tensor([9.9984e-05, 1.8616e-04, 1.7752e-04, 1.7929e-04, 1.7893e-04, 1.9021e-04, 1.4266e-04, 1.6016e-04], device='cuda:7') 2023-04-28 02:38:58,300 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33725.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:39:06,888 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6183, 5.9826, 5.5751, 5.7236, 5.1038, 4.8452, 5.3949, 6.0001], device='cuda:7'), covar=tensor([0.0599, 0.0603, 0.0895, 0.0435, 0.0568, 0.0549, 0.0510, 0.0634], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0460, 0.0393, 0.0290, 0.0291, 0.0288, 0.0363, 0.0320], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:39:16,430 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33738.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 02:39:18,838 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9678, 4.5433, 4.8066, 5.1295, 5.2484, 4.5634, 5.2247, 5.1832], device='cuda:7'), covar=tensor([0.0772, 0.0931, 0.1448, 0.0540, 0.0435, 0.0598, 0.0417, 0.0393], device='cuda:7'), in_proj_covar=tensor([0.0362, 0.0441, 0.0588, 0.0464, 0.0348, 0.0326, 0.0361, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:39:26,584 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.718e+02 3.624e+02 4.331e+02 5.289e+02 1.384e+03, threshold=8.662e+02, percent-clipped=5.0 2023-04-28 02:39:36,478 INFO [train.py:904] (7/8) Epoch 4, batch 3300, loss[loss=0.3228, simple_loss=0.383, pruned_loss=0.1313, over 12041.00 frames. ], tot_loss[loss=0.2276, simple_loss=0.3033, pruned_loss=0.07592, over 3337683.25 frames. ], batch size: 246, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:40:07,477 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=33773.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:40:11,320 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2071, 1.7359, 2.3657, 2.8675, 2.8195, 3.4061, 2.0005, 3.4297], device='cuda:7'), covar=tensor([0.0051, 0.0189, 0.0106, 0.0094, 0.0076, 0.0051, 0.0152, 0.0036], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0134, 0.0120, 0.0116, 0.0114, 0.0084, 0.0128, 0.0073], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 02:40:45,514 INFO [train.py:904] (7/8) Epoch 4, batch 3350, loss[loss=0.2088, simple_loss=0.2902, pruned_loss=0.06369, over 17136.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.303, pruned_loss=0.07571, over 3332919.58 frames. ], batch size: 49, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:40:58,985 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33811.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:41:22,129 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33828.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:41:26,950 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9242, 5.3945, 5.3767, 5.3567, 5.2421, 5.9246, 5.6110, 5.3936], device='cuda:7'), covar=tensor([0.0770, 0.1566, 0.1463, 0.1559, 0.2728, 0.0862, 0.1106, 0.2006], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0377, 0.0352, 0.0317, 0.0429, 0.0384, 0.0304, 0.0431], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:41:44,267 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.269e+02 3.302e+02 3.975e+02 4.780e+02 1.300e+03, threshold=7.950e+02, percent-clipped=1.0 2023-04-28 02:41:52,911 INFO [train.py:904] (7/8) Epoch 4, batch 3400, loss[loss=0.2341, simple_loss=0.3124, pruned_loss=0.07789, over 16595.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.3026, pruned_loss=0.07588, over 3335370.25 frames. ], batch size: 62, lr: 1.67e-02, grad_scale: 8.0 2023-04-28 02:42:04,836 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=33859.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:43:03,525 INFO [train.py:904] (7/8) Epoch 4, batch 3450, loss[loss=0.2358, simple_loss=0.2976, pruned_loss=0.08695, over 16474.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3016, pruned_loss=0.07542, over 3337528.19 frames. ], batch size: 146, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:43:58,435 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33939.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:44:05,159 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.338e+02 4.153e+02 5.120e+02 1.104e+03, threshold=8.306e+02, percent-clipped=4.0 2023-04-28 02:44:13,351 INFO [train.py:904] (7/8) Epoch 4, batch 3500, loss[loss=0.2408, simple_loss=0.3087, pruned_loss=0.08642, over 16238.00 frames. ], tot_loss[loss=0.2267, simple_loss=0.3011, pruned_loss=0.07615, over 3308114.26 frames. ], batch size: 165, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:45:26,863 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34000.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:45:27,582 INFO [train.py:904] (7/8) Epoch 4, batch 3550, loss[loss=0.2114, simple_loss=0.2954, pruned_loss=0.06366, over 16675.00 frames. ], tot_loss[loss=0.2253, simple_loss=0.2992, pruned_loss=0.07567, over 3312167.90 frames. ], batch size: 62, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:45:38,577 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 02:46:11,066 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34033.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 02:46:27,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34044.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:46:27,925 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.116e+02 3.324e+02 4.237e+02 5.025e+02 1.251e+03, threshold=8.474e+02, percent-clipped=3.0 2023-04-28 02:46:35,601 INFO [train.py:904] (7/8) Epoch 4, batch 3600, loss[loss=0.2125, simple_loss=0.2941, pruned_loss=0.06546, over 16515.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.2976, pruned_loss=0.07445, over 3309582.98 frames. ], batch size: 62, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:47:29,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8828, 4.7882, 4.6906, 4.6657, 4.2873, 4.7620, 4.7044, 4.4873], device='cuda:7'), covar=tensor([0.0424, 0.0269, 0.0190, 0.0156, 0.0771, 0.0262, 0.0243, 0.0329], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0174, 0.0219, 0.0187, 0.0258, 0.0207, 0.0158, 0.0221], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:47:31,056 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9105, 5.3431, 5.0362, 5.2155, 4.7198, 4.5818, 4.9275, 5.5051], device='cuda:7'), covar=tensor([0.0686, 0.0730, 0.1021, 0.0391, 0.0606, 0.0753, 0.0625, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0462, 0.0392, 0.0290, 0.0295, 0.0289, 0.0364, 0.0317], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:47:48,229 INFO [train.py:904] (7/8) Epoch 4, batch 3650, loss[loss=0.2486, simple_loss=0.3102, pruned_loss=0.09353, over 16383.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.2976, pruned_loss=0.07551, over 3303411.10 frames. ], batch size: 146, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:47:51,686 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3776, 4.3006, 4.3088, 3.7987, 4.3183, 1.8322, 4.1246, 4.3191], device='cuda:7'), covar=tensor([0.0067, 0.0065, 0.0077, 0.0253, 0.0056, 0.1383, 0.0080, 0.0095], device='cuda:7'), in_proj_covar=tensor([0.0092, 0.0081, 0.0122, 0.0130, 0.0091, 0.0130, 0.0108, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:47:54,764 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34105.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:48:29,144 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34128.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:48:53,743 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.336e+02 3.266e+02 3.802e+02 4.883e+02 8.817e+02, threshold=7.604e+02, percent-clipped=1.0 2023-04-28 02:48:54,266 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34145.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:49:02,650 INFO [train.py:904] (7/8) Epoch 4, batch 3700, loss[loss=0.2312, simple_loss=0.291, pruned_loss=0.08566, over 16750.00 frames. ], tot_loss[loss=0.2257, simple_loss=0.2965, pruned_loss=0.07745, over 3288653.16 frames. ], batch size: 134, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:49:41,506 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34176.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:50:17,554 INFO [train.py:904] (7/8) Epoch 4, batch 3750, loss[loss=0.2438, simple_loss=0.326, pruned_loss=0.08079, over 17034.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.2974, pruned_loss=0.07883, over 3262580.08 frames. ], batch size: 41, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:50:25,293 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:50:39,079 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8643, 1.2586, 1.5161, 1.7802, 1.7368, 1.9769, 1.3528, 1.8335], device='cuda:7'), covar=tensor([0.0058, 0.0153, 0.0087, 0.0093, 0.0077, 0.0048, 0.0149, 0.0037], device='cuda:7'), in_proj_covar=tensor([0.0106, 0.0133, 0.0122, 0.0115, 0.0114, 0.0084, 0.0128, 0.0073], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 02:51:21,424 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.239e+02 3.358e+02 4.060e+02 5.073e+02 1.417e+03, threshold=8.120e+02, percent-clipped=4.0 2023-04-28 02:51:29,938 INFO [train.py:904] (7/8) Epoch 4, batch 3800, loss[loss=0.2327, simple_loss=0.2956, pruned_loss=0.08488, over 16468.00 frames. ], tot_loss[loss=0.2295, simple_loss=0.2983, pruned_loss=0.08037, over 3261111.52 frames. ], batch size: 68, lr: 1.66e-02, grad_scale: 8.0 2023-04-28 02:52:01,061 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 02:52:36,322 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34295.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:52:38,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6006, 4.4923, 4.4961, 4.0020, 4.4609, 1.9016, 4.2874, 4.5521], device='cuda:7'), covar=tensor([0.0072, 0.0055, 0.0082, 0.0314, 0.0059, 0.1481, 0.0085, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0090, 0.0080, 0.0119, 0.0129, 0.0089, 0.0129, 0.0105, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 02:52:44,490 INFO [train.py:904] (7/8) Epoch 4, batch 3850, loss[loss=0.2101, simple_loss=0.2873, pruned_loss=0.0665, over 16572.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.2984, pruned_loss=0.08113, over 3253735.16 frames. ], batch size: 62, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:53:31,199 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:53:49,433 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.155e+02 3.189e+02 3.757e+02 4.739e+02 9.284e+02, threshold=7.515e+02, percent-clipped=3.0 2023-04-28 02:53:57,017 INFO [train.py:904] (7/8) Epoch 4, batch 3900, loss[loss=0.2094, simple_loss=0.2781, pruned_loss=0.07041, over 16875.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.2973, pruned_loss=0.08068, over 3255826.26 frames. ], batch size: 96, lr: 1.65e-02, grad_scale: 4.0 2023-04-28 02:54:40,998 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34381.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:55:09,565 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34400.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:55:10,264 INFO [train.py:904] (7/8) Epoch 4, batch 3950, loss[loss=0.2579, simple_loss=0.2952, pruned_loss=0.1103, over 16884.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.2966, pruned_loss=0.08176, over 3258694.76 frames. ], batch size: 109, lr: 1.65e-02, grad_scale: 4.0 2023-04-28 02:55:26,174 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 02:55:37,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 02:56:16,760 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.579e+02 4.210e+02 5.438e+02 1.110e+03, threshold=8.420e+02, percent-clipped=9.0 2023-04-28 02:56:24,485 INFO [train.py:904] (7/8) Epoch 4, batch 4000, loss[loss=0.256, simple_loss=0.316, pruned_loss=0.09799, over 16463.00 frames. ], tot_loss[loss=0.2295, simple_loss=0.296, pruned_loss=0.08154, over 3268543.12 frames. ], batch size: 146, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:57:08,979 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7677, 2.6774, 2.2489, 3.9809, 3.5667, 3.7084, 1.5464, 2.8447], device='cuda:7'), covar=tensor([0.1244, 0.0512, 0.1163, 0.0066, 0.0202, 0.0268, 0.1350, 0.0706], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0135, 0.0161, 0.0077, 0.0163, 0.0155, 0.0151, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 02:57:36,440 INFO [train.py:904] (7/8) Epoch 4, batch 4050, loss[loss=0.2177, simple_loss=0.2955, pruned_loss=0.07002, over 16758.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.2954, pruned_loss=0.07969, over 3263407.39 frames. ], batch size: 89, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:57:36,863 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34501.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:58:41,515 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.624e+02 2.984e+02 3.642e+02 7.677e+02, threshold=5.968e+02, percent-clipped=0.0 2023-04-28 02:58:48,891 INFO [train.py:904] (7/8) Epoch 4, batch 4100, loss[loss=0.2579, simple_loss=0.3342, pruned_loss=0.09075, over 16683.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.2955, pruned_loss=0.07802, over 3256688.67 frames. ], batch size: 124, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 02:59:06,952 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2345, 5.4647, 5.0758, 5.2867, 4.7996, 4.5903, 4.9413, 5.5215], device='cuda:7'), covar=tensor([0.0631, 0.0589, 0.0928, 0.0369, 0.0552, 0.0555, 0.0520, 0.0622], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0443, 0.0372, 0.0278, 0.0285, 0.0279, 0.0350, 0.0308], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 02:59:53,978 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34593.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 02:59:57,546 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34595.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:00:06,228 INFO [train.py:904] (7/8) Epoch 4, batch 4150, loss[loss=0.2646, simple_loss=0.3438, pruned_loss=0.0927, over 15456.00 frames. ], tot_loss[loss=0.2352, simple_loss=0.3049, pruned_loss=0.08271, over 3208380.41 frames. ], batch size: 190, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 03:00:38,957 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34622.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:00:44,978 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34626.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:00:58,972 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8880, 3.7309, 3.9182, 4.1178, 4.1756, 3.7908, 4.1404, 4.1845], device='cuda:7'), covar=tensor([0.0716, 0.0649, 0.1050, 0.0440, 0.0424, 0.0932, 0.0402, 0.0329], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0398, 0.0525, 0.0418, 0.0312, 0.0301, 0.0324, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:01:10,674 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34643.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:01:14,880 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-28 03:01:15,086 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.947e+02 3.447e+02 4.221e+02 5.425e+02 1.080e+03, threshold=8.441e+02, percent-clipped=15.0 2023-04-28 03:01:22,994 INFO [train.py:904] (7/8) Epoch 4, batch 4200, loss[loss=0.2204, simple_loss=0.307, pruned_loss=0.0669, over 16905.00 frames. ], tot_loss[loss=0.2415, simple_loss=0.3131, pruned_loss=0.08495, over 3199130.28 frames. ], batch size: 109, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 03:01:28,562 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34654.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:02:13,325 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34683.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:02:19,939 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34687.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:02:39,698 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34700.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:02:40,351 INFO [train.py:904] (7/8) Epoch 4, batch 4250, loss[loss=0.2083, simple_loss=0.2998, pruned_loss=0.05841, over 16837.00 frames. ], tot_loss[loss=0.2441, simple_loss=0.3167, pruned_loss=0.08576, over 3165939.91 frames. ], batch size: 102, lr: 1.65e-02, grad_scale: 8.0 2023-04-28 03:02:53,683 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34710.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:03:47,981 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 2.926e+02 3.520e+02 4.399e+02 1.246e+03, threshold=7.039e+02, percent-clipped=2.0 2023-04-28 03:03:52,078 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:03:56,217 INFO [train.py:904] (7/8) Epoch 4, batch 4300, loss[loss=0.2453, simple_loss=0.3353, pruned_loss=0.07766, over 16885.00 frames. ], tot_loss[loss=0.2423, simple_loss=0.3171, pruned_loss=0.0837, over 3178353.54 frames. ], batch size: 96, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:04:27,337 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34771.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:04:56,401 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 03:05:00,317 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1866, 3.7767, 3.6950, 1.5361, 4.0561, 4.0642, 2.9943, 2.7897], device='cuda:7'), covar=tensor([0.0736, 0.0092, 0.0168, 0.1122, 0.0030, 0.0027, 0.0328, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0081, 0.0081, 0.0140, 0.0069, 0.0071, 0.0112, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 03:05:10,256 INFO [train.py:904] (7/8) Epoch 4, batch 4350, loss[loss=0.2418, simple_loss=0.3192, pruned_loss=0.08215, over 16761.00 frames. ], tot_loss[loss=0.2461, simple_loss=0.3212, pruned_loss=0.08546, over 3175387.03 frames. ], batch size: 39, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:05:10,591 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34801.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:05:29,884 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2630, 4.5772, 2.0778, 4.8480, 3.1036, 4.8018, 2.4937, 3.0119], device='cuda:7'), covar=tensor([0.0086, 0.0134, 0.1516, 0.0021, 0.0595, 0.0178, 0.1229, 0.0598], device='cuda:7'), in_proj_covar=tensor([0.0105, 0.0143, 0.0172, 0.0081, 0.0158, 0.0171, 0.0179, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:05:36,397 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-28 03:06:04,223 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1387, 4.1356, 4.1518, 4.2445, 4.2305, 4.7026, 4.4245, 4.1129], device='cuda:7'), covar=tensor([0.1230, 0.1577, 0.1167, 0.1207, 0.1886, 0.0785, 0.0914, 0.1982], device='cuda:7'), in_proj_covar=tensor([0.0245, 0.0339, 0.0318, 0.0290, 0.0386, 0.0349, 0.0274, 0.0398], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 03:06:15,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.361e+02 3.334e+02 4.019e+02 4.887e+02 8.589e+02, threshold=8.038e+02, percent-clipped=2.0 2023-04-28 03:06:20,220 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=34849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:06:22,179 INFO [train.py:904] (7/8) Epoch 4, batch 4400, loss[loss=0.2422, simple_loss=0.3172, pruned_loss=0.08367, over 16520.00 frames. ], tot_loss[loss=0.2486, simple_loss=0.3235, pruned_loss=0.08691, over 3159589.55 frames. ], batch size: 68, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:06:54,530 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-28 03:07:09,741 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 03:07:32,094 INFO [train.py:904] (7/8) Epoch 4, batch 4450, loss[loss=0.2395, simple_loss=0.3171, pruned_loss=0.08094, over 16634.00 frames. ], tot_loss[loss=0.2498, simple_loss=0.3262, pruned_loss=0.08675, over 3181267.99 frames. ], batch size: 62, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:08:36,350 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.921e+02 3.471e+02 4.143e+02 7.527e+02, threshold=6.942e+02, percent-clipped=0.0 2023-04-28 03:08:40,977 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34949.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:08:43,015 INFO [train.py:904] (7/8) Epoch 4, batch 4500, loss[loss=0.2455, simple_loss=0.3149, pruned_loss=0.08808, over 11735.00 frames. ], tot_loss[loss=0.2475, simple_loss=0.3241, pruned_loss=0.08541, over 3188451.56 frames. ], batch size: 246, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:09:02,803 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0674, 5.0581, 4.7926, 4.7806, 4.4666, 4.9564, 4.7902, 4.6135], device='cuda:7'), covar=tensor([0.0321, 0.0093, 0.0159, 0.0112, 0.0637, 0.0114, 0.0159, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0147, 0.0193, 0.0158, 0.0223, 0.0174, 0.0138, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 03:09:13,078 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 03:09:23,017 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34978.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:09:28,664 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34982.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:09:54,038 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0915, 2.9272, 2.7144, 1.8394, 2.4681, 2.0133, 2.6514, 2.7910], device='cuda:7'), covar=tensor([0.0299, 0.0474, 0.0430, 0.1578, 0.0641, 0.0889, 0.0551, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0127, 0.0155, 0.0144, 0.0136, 0.0128, 0.0142, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:09:54,754 INFO [train.py:904] (7/8) Epoch 4, batch 4550, loss[loss=0.2977, simple_loss=0.3445, pruned_loss=0.1254, over 12054.00 frames. ], tot_loss[loss=0.2477, simple_loss=0.3241, pruned_loss=0.0856, over 3191369.65 frames. ], batch size: 246, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:10:25,368 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9027, 2.7412, 2.5697, 1.7615, 2.7773, 2.7772, 2.4155, 2.2677], device='cuda:7'), covar=tensor([0.0679, 0.0097, 0.0140, 0.0845, 0.0083, 0.0088, 0.0367, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0082, 0.0080, 0.0140, 0.0069, 0.0072, 0.0113, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 03:10:39,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1916, 2.2567, 2.1187, 3.8097, 1.8028, 3.2122, 2.2004, 2.1166], device='cuda:7'), covar=tensor([0.0511, 0.1347, 0.0790, 0.0245, 0.2545, 0.0499, 0.1353, 0.1957], device='cuda:7'), in_proj_covar=tensor([0.0293, 0.0280, 0.0231, 0.0291, 0.0349, 0.0254, 0.0253, 0.0350], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:10:42,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4233, 4.0385, 3.7616, 1.9974, 2.7982, 2.5380, 3.5874, 3.7218], device='cuda:7'), covar=tensor([0.0225, 0.0437, 0.0436, 0.1620, 0.0764, 0.0863, 0.0554, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0127, 0.0154, 0.0143, 0.0136, 0.0128, 0.0141, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:10:57,630 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.863e+02 3.417e+02 4.239e+02 9.930e+02, threshold=6.834e+02, percent-clipped=3.0 2023-04-28 03:11:04,054 INFO [train.py:904] (7/8) Epoch 4, batch 4600, loss[loss=0.2373, simple_loss=0.3198, pruned_loss=0.07734, over 16722.00 frames. ], tot_loss[loss=0.2474, simple_loss=0.3244, pruned_loss=0.08526, over 3192138.67 frames. ], batch size: 57, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:11:11,291 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7956, 6.0259, 5.6966, 6.0079, 5.4711, 4.7814, 5.6889, 6.2120], device='cuda:7'), covar=tensor([0.0389, 0.0452, 0.0794, 0.0274, 0.0443, 0.0544, 0.0353, 0.0507], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0415, 0.0352, 0.0267, 0.0264, 0.0268, 0.0329, 0.0293], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:11:25,504 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35066.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 03:12:02,679 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35092.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:12:05,717 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 03:12:15,721 INFO [train.py:904] (7/8) Epoch 4, batch 4650, loss[loss=0.2336, simple_loss=0.3131, pruned_loss=0.07704, over 16936.00 frames. ], tot_loss[loss=0.2456, simple_loss=0.3226, pruned_loss=0.08427, over 3214940.89 frames. ], batch size: 96, lr: 1.64e-02, grad_scale: 8.0 2023-04-28 03:12:29,940 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6619, 4.6664, 4.5473, 4.4604, 3.9656, 4.6518, 4.5914, 4.2599], device='cuda:7'), covar=tensor([0.0399, 0.0152, 0.0189, 0.0154, 0.0892, 0.0188, 0.0199, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0149, 0.0192, 0.0158, 0.0224, 0.0175, 0.0141, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 03:12:48,790 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0198, 2.0912, 2.0448, 3.4900, 1.7384, 2.9458, 2.1045, 1.9682], device='cuda:7'), covar=tensor([0.0509, 0.1422, 0.0801, 0.0298, 0.2490, 0.0499, 0.1389, 0.1902], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0278, 0.0232, 0.0293, 0.0350, 0.0252, 0.0254, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:13:20,509 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.712e+02 3.466e+02 4.003e+02 6.879e+02, threshold=6.932e+02, percent-clipped=1.0 2023-04-28 03:13:28,258 INFO [train.py:904] (7/8) Epoch 4, batch 4700, loss[loss=0.2161, simple_loss=0.2981, pruned_loss=0.06705, over 16526.00 frames. ], tot_loss[loss=0.2426, simple_loss=0.3197, pruned_loss=0.08279, over 3212247.86 frames. ], batch size: 68, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:13:32,488 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35153.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:13:36,613 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 03:14:41,641 INFO [train.py:904] (7/8) Epoch 4, batch 4750, loss[loss=0.2366, simple_loss=0.3199, pruned_loss=0.07667, over 16460.00 frames. ], tot_loss[loss=0.2382, simple_loss=0.3154, pruned_loss=0.0805, over 3211720.05 frames. ], batch size: 75, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:15:45,876 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.092e+02 2.820e+02 3.474e+02 4.196e+02 7.562e+02, threshold=6.948e+02, percent-clipped=1.0 2023-04-28 03:15:51,955 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35249.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:15:53,756 INFO [train.py:904] (7/8) Epoch 4, batch 4800, loss[loss=0.2589, simple_loss=0.3382, pruned_loss=0.08979, over 16742.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.3119, pruned_loss=0.07858, over 3218448.76 frames. ], batch size: 124, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:16:33,477 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35278.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:16:38,634 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35282.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:16:52,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3555, 3.5285, 1.7400, 3.6120, 2.3506, 3.5690, 1.8373, 2.6971], device='cuda:7'), covar=tensor([0.0092, 0.0191, 0.1419, 0.0044, 0.0733, 0.0301, 0.1386, 0.0551], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0142, 0.0173, 0.0079, 0.0158, 0.0166, 0.0180, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:16:59,986 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35297.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:17:05,715 INFO [train.py:904] (7/8) Epoch 4, batch 4850, loss[loss=0.224, simple_loss=0.3127, pruned_loss=0.06761, over 16651.00 frames. ], tot_loss[loss=0.2355, simple_loss=0.3139, pruned_loss=0.07858, over 3201275.28 frames. ], batch size: 134, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:17:41,859 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35326.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:17:48,535 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35330.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:18:05,264 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-28 03:18:11,909 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35345.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:18:12,515 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 3.003e+02 3.470e+02 4.055e+02 1.018e+03, threshold=6.940e+02, percent-clipped=2.0 2023-04-28 03:18:19,086 INFO [train.py:904] (7/8) Epoch 4, batch 4900, loss[loss=0.2339, simple_loss=0.312, pruned_loss=0.07791, over 16773.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.3129, pruned_loss=0.07742, over 3182836.49 frames. ], batch size: 124, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:18:42,056 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35366.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:18:54,630 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35374.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:19:24,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0113, 3.8388, 3.9729, 2.8085, 3.8365, 1.5325, 3.6840, 3.8313], device='cuda:7'), covar=tensor([0.0110, 0.0099, 0.0101, 0.0569, 0.0088, 0.1980, 0.0124, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0079, 0.0070, 0.0103, 0.0116, 0.0079, 0.0122, 0.0093, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:19:32,912 INFO [train.py:904] (7/8) Epoch 4, batch 4950, loss[loss=0.2383, simple_loss=0.3395, pruned_loss=0.06853, over 16895.00 frames. ], tot_loss[loss=0.2349, simple_loss=0.3138, pruned_loss=0.07798, over 3173568.72 frames. ], batch size: 96, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:19:41,941 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35406.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:19:53,437 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35414.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:20:12,475 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 03:20:21,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0808, 4.8087, 5.0366, 5.3199, 5.4421, 4.6531, 5.3686, 5.3949], device='cuda:7'), covar=tensor([0.0754, 0.0661, 0.1061, 0.0356, 0.0311, 0.0573, 0.0394, 0.0279], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0399, 0.0534, 0.0412, 0.0307, 0.0296, 0.0323, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:20:22,854 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35435.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:20:37,433 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 3.403e+02 3.998e+02 5.018e+02 8.259e+02, threshold=7.996e+02, percent-clipped=7.0 2023-04-28 03:20:41,807 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35448.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:20:45,639 INFO [train.py:904] (7/8) Epoch 4, batch 5000, loss[loss=0.2462, simple_loss=0.3302, pruned_loss=0.08111, over 15261.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.3156, pruned_loss=0.07806, over 3198924.13 frames. ], batch size: 190, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:20:50,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6259, 5.9481, 5.5914, 5.7532, 5.2291, 4.7710, 5.4110, 6.0849], device='cuda:7'), covar=tensor([0.0511, 0.0497, 0.0808, 0.0344, 0.0502, 0.0537, 0.0528, 0.0536], device='cuda:7'), in_proj_covar=tensor([0.0313, 0.0420, 0.0364, 0.0272, 0.0271, 0.0275, 0.0342, 0.0297], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:20:52,646 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3156, 4.2217, 3.7295, 2.0255, 3.2271, 2.4900, 3.9370, 3.8769], device='cuda:7'), covar=tensor([0.0202, 0.0403, 0.0457, 0.1563, 0.0622, 0.0845, 0.0440, 0.0598], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0123, 0.0154, 0.0144, 0.0138, 0.0128, 0.0143, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:21:57,710 INFO [train.py:904] (7/8) Epoch 4, batch 5050, loss[loss=0.2371, simple_loss=0.3186, pruned_loss=0.07777, over 16664.00 frames. ], tot_loss[loss=0.2347, simple_loss=0.3154, pruned_loss=0.07704, over 3209231.37 frames. ], batch size: 124, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:22:38,868 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2419, 2.3009, 1.5576, 2.2079, 2.9215, 2.6563, 3.4356, 3.1912], device='cuda:7'), covar=tensor([0.0022, 0.0167, 0.0250, 0.0180, 0.0080, 0.0140, 0.0032, 0.0070], device='cuda:7'), in_proj_covar=tensor([0.0069, 0.0139, 0.0141, 0.0138, 0.0131, 0.0146, 0.0100, 0.0117], device='cuda:7'), out_proj_covar=tensor([8.9978e-05, 1.8355e-04, 1.8051e-04, 1.7776e-04, 1.7463e-04, 1.9442e-04, 1.3066e-04, 1.5684e-04], device='cuda:7') 2023-04-28 03:22:40,046 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35530.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:22:41,138 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9391, 5.2728, 4.9742, 5.0151, 4.7369, 4.5282, 4.7802, 5.3429], device='cuda:7'), covar=tensor([0.0524, 0.0524, 0.0710, 0.0343, 0.0417, 0.0550, 0.0464, 0.0464], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0415, 0.0360, 0.0270, 0.0268, 0.0271, 0.0336, 0.0293], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:23:03,497 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.171e+02 3.027e+02 3.520e+02 4.379e+02 8.983e+02, threshold=7.040e+02, percent-clipped=1.0 2023-04-28 03:23:10,568 INFO [train.py:904] (7/8) Epoch 4, batch 5100, loss[loss=0.2199, simple_loss=0.3063, pruned_loss=0.06677, over 16890.00 frames. ], tot_loss[loss=0.2331, simple_loss=0.3137, pruned_loss=0.07623, over 3209846.97 frames. ], batch size: 96, lr: 1.63e-02, grad_scale: 8.0 2023-04-28 03:23:11,152 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3059, 1.4910, 2.3137, 3.0391, 2.9240, 3.3916, 1.5828, 3.3539], device='cuda:7'), covar=tensor([0.0043, 0.0234, 0.0137, 0.0097, 0.0083, 0.0061, 0.0230, 0.0030], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0132, 0.0120, 0.0116, 0.0118, 0.0082, 0.0132, 0.0075], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 03:23:51,077 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7775, 3.2380, 3.1693, 2.1587, 3.1049, 3.0908, 2.9750, 1.7704], device='cuda:7'), covar=tensor([0.0359, 0.0024, 0.0024, 0.0230, 0.0044, 0.0059, 0.0048, 0.0309], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0051, 0.0055, 0.0110, 0.0058, 0.0064, 0.0058, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 03:24:08,078 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35591.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:24:22,835 INFO [train.py:904] (7/8) Epoch 4, batch 5150, loss[loss=0.2422, simple_loss=0.3165, pruned_loss=0.08394, over 16627.00 frames. ], tot_loss[loss=0.2323, simple_loss=0.3132, pruned_loss=0.0757, over 3188137.75 frames. ], batch size: 62, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:24:27,758 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-28 03:25:29,064 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.054e+02 2.805e+02 3.383e+02 4.164e+02 1.002e+03, threshold=6.766e+02, percent-clipped=7.0 2023-04-28 03:25:36,103 INFO [train.py:904] (7/8) Epoch 4, batch 5200, loss[loss=0.1961, simple_loss=0.2819, pruned_loss=0.05515, over 16763.00 frames. ], tot_loss[loss=0.232, simple_loss=0.3128, pruned_loss=0.07563, over 3180492.18 frames. ], batch size: 83, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:26:24,227 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 03:26:46,334 INFO [train.py:904] (7/8) Epoch 4, batch 5250, loss[loss=0.2431, simple_loss=0.3315, pruned_loss=0.07737, over 16411.00 frames. ], tot_loss[loss=0.2304, simple_loss=0.3103, pruned_loss=0.07525, over 3191040.55 frames. ], batch size: 146, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:26:47,358 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35701.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 03:27:28,650 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35730.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:27:52,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.886e+02 3.406e+02 4.102e+02 7.546e+02, threshold=6.813e+02, percent-clipped=1.0 2023-04-28 03:27:55,791 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:28:00,147 INFO [train.py:904] (7/8) Epoch 4, batch 5300, loss[loss=0.1849, simple_loss=0.2577, pruned_loss=0.05601, over 16686.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3048, pruned_loss=0.0728, over 3213386.04 frames. ], batch size: 57, lr: 1.62e-02, grad_scale: 8.0 2023-04-28 03:28:28,788 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35771.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:29:02,757 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=35796.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:29:11,339 INFO [train.py:904] (7/8) Epoch 4, batch 5350, loss[loss=0.2587, simple_loss=0.328, pruned_loss=0.09475, over 12494.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3029, pruned_loss=0.07197, over 3195956.77 frames. ], batch size: 246, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:29:24,973 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35811.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:29:57,749 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35832.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:30:17,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 3.081e+02 3.703e+02 4.509e+02 8.729e+02, threshold=7.406e+02, percent-clipped=5.0 2023-04-28 03:30:23,348 INFO [train.py:904] (7/8) Epoch 4, batch 5400, loss[loss=0.255, simple_loss=0.3372, pruned_loss=0.08643, over 16486.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.3061, pruned_loss=0.07323, over 3197164.25 frames. ], batch size: 68, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:30:53,743 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5713, 1.3805, 1.9804, 2.4850, 2.3787, 2.7371, 1.4296, 2.5845], device='cuda:7'), covar=tensor([0.0047, 0.0219, 0.0116, 0.0092, 0.0083, 0.0047, 0.0196, 0.0045], device='cuda:7'), in_proj_covar=tensor([0.0100, 0.0129, 0.0118, 0.0112, 0.0114, 0.0080, 0.0130, 0.0075], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 03:30:53,762 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35872.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:31:06,543 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 03:31:13,889 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35886.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:31:38,581 INFO [train.py:904] (7/8) Epoch 4, batch 5450, loss[loss=0.2394, simple_loss=0.322, pruned_loss=0.07838, over 16758.00 frames. ], tot_loss[loss=0.2321, simple_loss=0.3113, pruned_loss=0.07646, over 3190691.71 frames. ], batch size: 83, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:32:38,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3581, 4.3181, 4.1726, 3.5450, 4.1992, 1.6835, 4.0289, 4.1194], device='cuda:7'), covar=tensor([0.0058, 0.0046, 0.0080, 0.0309, 0.0054, 0.1585, 0.0080, 0.0119], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0072, 0.0109, 0.0122, 0.0082, 0.0129, 0.0096, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:32:50,255 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.590e+02 3.796e+02 5.362e+02 7.541e+02 3.116e+03, threshold=1.072e+03, percent-clipped=25.0 2023-04-28 03:32:56,440 INFO [train.py:904] (7/8) Epoch 4, batch 5500, loss[loss=0.294, simple_loss=0.3669, pruned_loss=0.1105, over 16757.00 frames. ], tot_loss[loss=0.2453, simple_loss=0.3217, pruned_loss=0.08443, over 3166507.66 frames. ], batch size: 83, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:34:09,867 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 03:34:16,685 INFO [train.py:904] (7/8) Epoch 4, batch 5550, loss[loss=0.364, simple_loss=0.3947, pruned_loss=0.1666, over 10850.00 frames. ], tot_loss[loss=0.2579, simple_loss=0.3312, pruned_loss=0.09226, over 3117922.72 frames. ], batch size: 246, lr: 1.62e-02, grad_scale: 4.0 2023-04-28 03:34:17,793 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36001.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:35:02,207 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36030.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:35:29,647 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.087e+02 4.484e+02 5.536e+02 7.020e+02 1.224e+03, threshold=1.107e+03, percent-clipped=3.0 2023-04-28 03:35:33,034 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36049.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 03:35:35,563 INFO [train.py:904] (7/8) Epoch 4, batch 5600, loss[loss=0.2755, simple_loss=0.3505, pruned_loss=0.1003, over 16807.00 frames. ], tot_loss[loss=0.2674, simple_loss=0.3378, pruned_loss=0.09849, over 3092490.18 frames. ], batch size: 124, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:35:36,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-28 03:35:44,940 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-28 03:36:20,418 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:36:57,495 INFO [train.py:904] (7/8) Epoch 4, batch 5650, loss[loss=0.3638, simple_loss=0.3934, pruned_loss=0.1672, over 11169.00 frames. ], tot_loss[loss=0.2755, simple_loss=0.3439, pruned_loss=0.1036, over 3085744.78 frames. ], batch size: 250, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:37:41,228 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36127.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:37:45,907 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0699, 1.3286, 1.8074, 1.8923, 2.1627, 2.2761, 1.4789, 2.0562], device='cuda:7'), covar=tensor([0.0069, 0.0191, 0.0102, 0.0107, 0.0077, 0.0048, 0.0162, 0.0043], device='cuda:7'), in_proj_covar=tensor([0.0097, 0.0127, 0.0114, 0.0108, 0.0111, 0.0078, 0.0126, 0.0073], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 03:38:09,924 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.856e+02 5.092e+02 6.431e+02 7.893e+02 1.649e+03, threshold=1.286e+03, percent-clipped=5.0 2023-04-28 03:38:17,755 INFO [train.py:904] (7/8) Epoch 4, batch 5700, loss[loss=0.3424, simple_loss=0.3824, pruned_loss=0.1512, over 11135.00 frames. ], tot_loss[loss=0.2787, simple_loss=0.3459, pruned_loss=0.1058, over 3067295.17 frames. ], batch size: 247, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:38:42,692 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36167.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:39:05,536 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36181.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:39:13,411 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36186.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:39:36,195 INFO [train.py:904] (7/8) Epoch 4, batch 5750, loss[loss=0.3077, simple_loss=0.3659, pruned_loss=0.1247, over 15406.00 frames. ], tot_loss[loss=0.2819, simple_loss=0.3486, pruned_loss=0.1076, over 3043931.28 frames. ], batch size: 190, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:40:19,615 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2548, 5.6006, 5.2115, 5.3769, 4.9165, 4.7002, 5.0082, 5.6354], device='cuda:7'), covar=tensor([0.0645, 0.0571, 0.0918, 0.0430, 0.0572, 0.0571, 0.0612, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0418, 0.0372, 0.0277, 0.0272, 0.0278, 0.0340, 0.0297], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:40:29,645 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36234.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:40:42,512 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36242.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:40:49,467 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 4.215e+02 5.340e+02 6.535e+02 1.180e+03, threshold=1.068e+03, percent-clipped=0.0 2023-04-28 03:40:55,818 INFO [train.py:904] (7/8) Epoch 4, batch 5800, loss[loss=0.2581, simple_loss=0.3424, pruned_loss=0.08694, over 16834.00 frames. ], tot_loss[loss=0.278, simple_loss=0.3467, pruned_loss=0.1046, over 3047391.06 frames. ], batch size: 102, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:42:05,480 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.40 vs. limit=5.0 2023-04-28 03:42:12,368 INFO [train.py:904] (7/8) Epoch 4, batch 5850, loss[loss=0.2546, simple_loss=0.3326, pruned_loss=0.08831, over 16472.00 frames. ], tot_loss[loss=0.2736, simple_loss=0.3439, pruned_loss=0.1017, over 3071451.10 frames. ], batch size: 146, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:42:42,491 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6646, 3.0259, 3.0747, 1.9813, 2.8788, 2.9236, 2.8358, 1.7252], device='cuda:7'), covar=tensor([0.0333, 0.0027, 0.0033, 0.0248, 0.0048, 0.0051, 0.0041, 0.0290], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0049, 0.0054, 0.0109, 0.0058, 0.0063, 0.0057, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 03:43:14,554 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 03:43:29,068 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 4.315e+02 5.493e+02 6.918e+02 1.909e+03, threshold=1.099e+03, percent-clipped=3.0 2023-04-28 03:43:33,999 INFO [train.py:904] (7/8) Epoch 4, batch 5900, loss[loss=0.3208, simple_loss=0.3619, pruned_loss=0.1399, over 11825.00 frames. ], tot_loss[loss=0.2742, simple_loss=0.3436, pruned_loss=0.1024, over 3045007.06 frames. ], batch size: 246, lr: 1.61e-02, grad_scale: 4.0 2023-04-28 03:43:39,179 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 03:44:56,229 INFO [train.py:904] (7/8) Epoch 4, batch 5950, loss[loss=0.3057, simple_loss=0.361, pruned_loss=0.1253, over 11861.00 frames. ], tot_loss[loss=0.2721, simple_loss=0.3441, pruned_loss=0.09999, over 3076872.38 frames. ], batch size: 246, lr: 1.61e-02, grad_scale: 4.0 2023-04-28 03:45:37,611 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36427.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:45:51,490 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-28 03:46:10,128 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.426e+02 4.337e+02 5.242e+02 6.578e+02 1.706e+03, threshold=1.048e+03, percent-clipped=3.0 2023-04-28 03:46:14,575 INFO [train.py:904] (7/8) Epoch 4, batch 6000, loss[loss=0.2495, simple_loss=0.3285, pruned_loss=0.08525, over 16894.00 frames. ], tot_loss[loss=0.2692, simple_loss=0.3422, pruned_loss=0.09811, over 3119240.51 frames. ], batch size: 96, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:46:14,575 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 03:46:25,214 INFO [train.py:938] (7/8) Epoch 4, validation: loss=0.1966, simple_loss=0.3069, pruned_loss=0.04309, over 944034.00 frames. 2023-04-28 03:46:25,215 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 03:46:49,553 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36467.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:47:01,424 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36475.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:47:44,059 INFO [train.py:904] (7/8) Epoch 4, batch 6050, loss[loss=0.2667, simple_loss=0.3392, pruned_loss=0.09715, over 15234.00 frames. ], tot_loss[loss=0.2677, simple_loss=0.3406, pruned_loss=0.09738, over 3117805.03 frames. ], batch size: 190, lr: 1.61e-02, grad_scale: 8.0 2023-04-28 03:48:06,439 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36515.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:48:13,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1034, 2.9599, 2.5913, 1.9563, 2.4904, 2.0284, 2.7200, 2.9211], device='cuda:7'), covar=tensor([0.0264, 0.0420, 0.0487, 0.1336, 0.0659, 0.0857, 0.0501, 0.0492], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0123, 0.0154, 0.0143, 0.0137, 0.0128, 0.0142, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:48:38,731 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 03:48:42,919 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36537.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:48:58,312 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.180e+02 4.133e+02 5.686e+02 7.938e+02 1.625e+03, threshold=1.137e+03, percent-clipped=9.0 2023-04-28 03:49:03,049 INFO [train.py:904] (7/8) Epoch 4, batch 6100, loss[loss=0.2903, simple_loss=0.3595, pruned_loss=0.1105, over 16250.00 frames. ], tot_loss[loss=0.2665, simple_loss=0.3398, pruned_loss=0.09665, over 3105731.79 frames. ], batch size: 165, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:50:22,086 INFO [train.py:904] (7/8) Epoch 4, batch 6150, loss[loss=0.2435, simple_loss=0.3148, pruned_loss=0.08609, over 16693.00 frames. ], tot_loss[loss=0.2654, simple_loss=0.3379, pruned_loss=0.09641, over 3094891.09 frames. ], batch size: 62, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:50:29,602 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36605.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:50:31,941 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6939, 4.4508, 4.6415, 4.9049, 5.0571, 4.4544, 5.0005, 4.9086], device='cuda:7'), covar=tensor([0.0708, 0.0716, 0.1082, 0.0386, 0.0348, 0.0570, 0.0366, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0415, 0.0532, 0.0412, 0.0310, 0.0300, 0.0333, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:50:34,968 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36608.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:50:41,783 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36612.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:50:43,850 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.54 vs. limit=5.0 2023-04-28 03:50:47,983 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9116, 4.6222, 4.8236, 5.0724, 5.2695, 4.5921, 5.2290, 5.1260], device='cuda:7'), covar=tensor([0.0652, 0.0745, 0.1188, 0.0411, 0.0318, 0.0542, 0.0319, 0.0357], device='cuda:7'), in_proj_covar=tensor([0.0337, 0.0416, 0.0534, 0.0412, 0.0311, 0.0300, 0.0335, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:51:38,885 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.478e+02 3.823e+02 4.755e+02 6.026e+02 1.290e+03, threshold=9.509e+02, percent-clipped=1.0 2023-04-28 03:51:44,331 INFO [train.py:904] (7/8) Epoch 4, batch 6200, loss[loss=0.2656, simple_loss=0.3367, pruned_loss=0.09723, over 16659.00 frames. ], tot_loss[loss=0.263, simple_loss=0.3353, pruned_loss=0.09533, over 3105283.08 frames. ], batch size: 134, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:52:07,265 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36666.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:52:11,537 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36669.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:52:17,438 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36673.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:52:51,894 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9564, 3.7550, 3.3874, 1.7194, 2.8149, 2.3765, 3.4324, 3.6600], device='cuda:7'), covar=tensor([0.0291, 0.0461, 0.0444, 0.1669, 0.0715, 0.0875, 0.0598, 0.0571], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0125, 0.0155, 0.0144, 0.0139, 0.0130, 0.0145, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 03:52:58,550 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36700.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:52:59,119 INFO [train.py:904] (7/8) Epoch 4, batch 6250, loss[loss=0.2652, simple_loss=0.3576, pruned_loss=0.08638, over 16814.00 frames. ], tot_loss[loss=0.264, simple_loss=0.336, pruned_loss=0.09596, over 3086697.60 frames. ], batch size: 96, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:54:09,138 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.662e+02 4.271e+02 5.056e+02 6.233e+02 1.402e+03, threshold=1.011e+03, percent-clipped=5.0 2023-04-28 03:54:10,684 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 03:54:14,296 INFO [train.py:904] (7/8) Epoch 4, batch 6300, loss[loss=0.2486, simple_loss=0.3219, pruned_loss=0.08766, over 16703.00 frames. ], tot_loss[loss=0.2624, simple_loss=0.3353, pruned_loss=0.09471, over 3102388.82 frames. ], batch size: 62, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:54:31,277 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36761.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 03:55:11,489 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9369, 3.1183, 3.4667, 3.4179, 3.4229, 3.1222, 3.2420, 3.2846], device='cuda:7'), covar=tensor([0.0429, 0.0571, 0.0398, 0.0517, 0.0496, 0.0434, 0.0796, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0204, 0.0213, 0.0218, 0.0258, 0.0224, 0.0325, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 03:55:32,015 INFO [train.py:904] (7/8) Epoch 4, batch 6350, loss[loss=0.2542, simple_loss=0.3253, pruned_loss=0.09148, over 16927.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3373, pruned_loss=0.09669, over 3092769.06 frames. ], batch size: 109, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:55:38,762 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36805.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:56:28,087 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36837.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:56:43,688 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.626e+02 4.373e+02 5.288e+02 7.014e+02 2.020e+03, threshold=1.058e+03, percent-clipped=6.0 2023-04-28 03:56:48,104 INFO [train.py:904] (7/8) Epoch 4, batch 6400, loss[loss=0.2385, simple_loss=0.3183, pruned_loss=0.07938, over 16797.00 frames. ], tot_loss[loss=0.2664, simple_loss=0.3372, pruned_loss=0.09778, over 3075423.91 frames. ], batch size: 76, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:56:51,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0150, 4.9792, 4.8291, 3.2665, 4.7587, 1.4347, 4.5708, 4.7494], device='cuda:7'), covar=tensor([0.0102, 0.0075, 0.0109, 0.0586, 0.0079, 0.2373, 0.0102, 0.0194], device='cuda:7'), in_proj_covar=tensor([0.0082, 0.0070, 0.0107, 0.0117, 0.0079, 0.0130, 0.0095, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 03:57:10,851 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36866.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 03:57:40,052 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=36885.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:58:04,377 INFO [train.py:904] (7/8) Epoch 4, batch 6450, loss[loss=0.2809, simple_loss=0.3536, pruned_loss=0.1041, over 16363.00 frames. ], tot_loss[loss=0.2651, simple_loss=0.3369, pruned_loss=0.09666, over 3071684.35 frames. ], batch size: 35, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:58:20,100 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36911.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:18,069 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.485e+02 3.772e+02 4.534e+02 6.144e+02 1.455e+03, threshold=9.067e+02, percent-clipped=1.0 2023-04-28 03:59:23,794 INFO [train.py:904] (7/8) Epoch 4, batch 6500, loss[loss=0.3255, simple_loss=0.3681, pruned_loss=0.1415, over 11904.00 frames. ], tot_loss[loss=0.2621, simple_loss=0.3341, pruned_loss=0.09505, over 3088926.04 frames. ], batch size: 247, lr: 1.60e-02, grad_scale: 8.0 2023-04-28 03:59:35,250 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36958.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:39,451 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36961.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:39,510 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3055, 4.6375, 5.0331, 5.1000, 4.9183, 4.4750, 4.0913, 4.1989], device='cuda:7'), covar=tensor([0.0582, 0.0431, 0.0483, 0.0487, 0.0697, 0.0506, 0.1311, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0203, 0.0213, 0.0217, 0.0256, 0.0221, 0.0325, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 03:59:44,080 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:50,181 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36968.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 03:59:56,721 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:00:28,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7613, 4.7019, 5.3637, 5.3674, 5.3063, 4.8759, 4.8048, 4.5864], device='cuda:7'), covar=tensor([0.0241, 0.0350, 0.0317, 0.0277, 0.0331, 0.0254, 0.0720, 0.0325], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0205, 0.0212, 0.0218, 0.0257, 0.0223, 0.0325, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 04:00:44,414 INFO [train.py:904] (7/8) Epoch 4, batch 6550, loss[loss=0.2472, simple_loss=0.3372, pruned_loss=0.07861, over 16532.00 frames. ], tot_loss[loss=0.2653, simple_loss=0.3372, pruned_loss=0.09674, over 3070774.52 frames. ], batch size: 75, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:00:50,704 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37004.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:01:13,115 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37019.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:01:31,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0321, 3.4442, 3.4647, 1.4525, 3.5767, 3.6664, 2.8219, 2.7772], device='cuda:7'), covar=tensor([0.0766, 0.0097, 0.0154, 0.1208, 0.0058, 0.0056, 0.0350, 0.0358], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0084, 0.0080, 0.0140, 0.0067, 0.0072, 0.0112, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 04:01:56,567 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.603e+02 4.116e+02 5.104e+02 6.571e+02 1.425e+03, threshold=1.021e+03, percent-clipped=7.0 2023-04-28 04:02:01,488 INFO [train.py:904] (7/8) Epoch 4, batch 6600, loss[loss=0.2471, simple_loss=0.3268, pruned_loss=0.08367, over 16869.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3392, pruned_loss=0.09647, over 3097961.25 frames. ], batch size: 96, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:02:09,610 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37056.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:02:23,406 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37065.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:02:28,779 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 04:02:54,048 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 04:03:05,786 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 04:03:19,959 INFO [train.py:904] (7/8) Epoch 4, batch 6650, loss[loss=0.2238, simple_loss=0.303, pruned_loss=0.07233, over 16855.00 frames. ], tot_loss[loss=0.267, simple_loss=0.3396, pruned_loss=0.09721, over 3108574.90 frames. ], batch size: 42, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:03:20,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8080, 3.2391, 3.2503, 1.4164, 3.3484, 3.4408, 2.7810, 2.4705], device='cuda:7'), covar=tensor([0.0909, 0.0115, 0.0171, 0.1259, 0.0072, 0.0066, 0.0324, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0084, 0.0080, 0.0141, 0.0068, 0.0072, 0.0113, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 04:04:33,024 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.678e+02 4.226e+02 5.482e+02 6.973e+02 1.276e+03, threshold=1.096e+03, percent-clipped=4.0 2023-04-28 04:04:37,791 INFO [train.py:904] (7/8) Epoch 4, batch 6700, loss[loss=0.2584, simple_loss=0.336, pruned_loss=0.09042, over 16901.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3381, pruned_loss=0.09704, over 3109181.18 frames. ], batch size: 116, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:04:53,202 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37161.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:05:53,503 INFO [train.py:904] (7/8) Epoch 4, batch 6750, loss[loss=0.2354, simple_loss=0.3149, pruned_loss=0.07798, over 16603.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.3362, pruned_loss=0.09616, over 3133181.27 frames. ], batch size: 62, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:06:02,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2043, 4.2202, 4.1768, 1.4400, 4.4049, 4.5337, 3.1519, 2.9346], device='cuda:7'), covar=tensor([0.1040, 0.0080, 0.0143, 0.1458, 0.0044, 0.0030, 0.0320, 0.0531], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0083, 0.0079, 0.0141, 0.0067, 0.0070, 0.0112, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 04:06:12,863 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.67 vs. limit=5.0 2023-04-28 04:06:59,841 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9501, 5.4573, 5.5696, 5.5115, 5.5460, 6.0474, 5.5801, 5.3496], device='cuda:7'), covar=tensor([0.0610, 0.1217, 0.0945, 0.1432, 0.1844, 0.0606, 0.0876, 0.1650], device='cuda:7'), in_proj_covar=tensor([0.0255, 0.0354, 0.0341, 0.0312, 0.0406, 0.0367, 0.0286, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:07:06,102 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.556e+02 4.104e+02 5.227e+02 6.545e+02 1.583e+03, threshold=1.045e+03, percent-clipped=5.0 2023-04-28 04:07:10,550 INFO [train.py:904] (7/8) Epoch 4, batch 6800, loss[loss=0.2304, simple_loss=0.3167, pruned_loss=0.0721, over 16860.00 frames. ], tot_loss[loss=0.2643, simple_loss=0.3361, pruned_loss=0.09628, over 3118265.24 frames. ], batch size: 116, lr: 1.59e-02, grad_scale: 8.0 2023-04-28 04:07:27,147 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37261.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:31,626 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37264.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:35,442 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37267.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:37,359 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:37,428 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:07:54,390 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9840, 4.0205, 3.8966, 3.8685, 3.4376, 3.9459, 3.7329, 3.7145], device='cuda:7'), covar=tensor([0.0475, 0.0233, 0.0228, 0.0182, 0.0832, 0.0289, 0.0551, 0.0471], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0155, 0.0192, 0.0159, 0.0221, 0.0185, 0.0145, 0.0206], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:08:10,911 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 04:08:27,077 INFO [train.py:904] (7/8) Epoch 4, batch 6850, loss[loss=0.2702, simple_loss=0.3705, pruned_loss=0.08493, over 16721.00 frames. ], tot_loss[loss=0.2657, simple_loss=0.3379, pruned_loss=0.09677, over 3122686.25 frames. ], batch size: 89, lr: 1.59e-02, grad_scale: 4.0 2023-04-28 04:08:39,473 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37309.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:08:43,642 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37312.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:08:46,877 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37314.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:08:49,186 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37316.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:09:08,875 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37329.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:09:29,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7214, 1.1974, 1.4614, 1.5520, 1.7751, 1.8372, 1.4158, 1.7526], device='cuda:7'), covar=tensor([0.0077, 0.0177, 0.0099, 0.0131, 0.0097, 0.0063, 0.0146, 0.0044], device='cuda:7'), in_proj_covar=tensor([0.0106, 0.0136, 0.0122, 0.0115, 0.0122, 0.0084, 0.0136, 0.0077], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 04:09:36,163 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.986e+02 4.056e+02 4.810e+02 6.733e+02 1.178e+03, threshold=9.620e+02, percent-clipped=3.0 2023-04-28 04:09:39,832 INFO [train.py:904] (7/8) Epoch 4, batch 6900, loss[loss=0.2362, simple_loss=0.3205, pruned_loss=0.07592, over 16840.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3401, pruned_loss=0.09606, over 3131002.75 frames. ], batch size: 102, lr: 1.59e-02, grad_scale: 4.0 2023-04-28 04:09:48,115 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37356.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 04:09:51,885 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8539, 1.6123, 1.3424, 1.4242, 1.8146, 1.5714, 1.7048, 1.8118], device='cuda:7'), covar=tensor([0.0028, 0.0109, 0.0166, 0.0139, 0.0072, 0.0116, 0.0054, 0.0078], device='cuda:7'), in_proj_covar=tensor([0.0065, 0.0135, 0.0141, 0.0135, 0.0129, 0.0140, 0.0100, 0.0117], device='cuda:7'), out_proj_covar=tensor([8.3404e-05, 1.7589e-04, 1.7916e-04, 1.7204e-04, 1.6817e-04, 1.8330e-04, 1.2641e-04, 1.5355e-04], device='cuda:7') 2023-04-28 04:09:53,442 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:09:59,888 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37364.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:10:06,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6718, 5.1294, 5.1965, 5.2785, 5.0846, 5.6594, 5.2260, 5.0742], device='cuda:7'), covar=tensor([0.0695, 0.1138, 0.1037, 0.1176, 0.1836, 0.0688, 0.0963, 0.1598], device='cuda:7'), in_proj_covar=tensor([0.0250, 0.0348, 0.0335, 0.0306, 0.0397, 0.0361, 0.0281, 0.0407], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:10:55,355 INFO [train.py:904] (7/8) Epoch 4, batch 6950, loss[loss=0.26, simple_loss=0.3492, pruned_loss=0.08538, over 16820.00 frames. ], tot_loss[loss=0.272, simple_loss=0.344, pruned_loss=0.09995, over 3101778.60 frames. ], batch size: 102, lr: 1.59e-02, grad_scale: 4.0 2023-04-28 04:11:00,665 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37404.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:11:32,955 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37425.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:12:10,359 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.259e+02 4.534e+02 5.752e+02 7.141e+02 1.491e+03, threshold=1.150e+03, percent-clipped=11.0 2023-04-28 04:12:11,791 INFO [train.py:904] (7/8) Epoch 4, batch 7000, loss[loss=0.2573, simple_loss=0.3433, pruned_loss=0.08562, over 16247.00 frames. ], tot_loss[loss=0.2718, simple_loss=0.3444, pruned_loss=0.09958, over 3111023.93 frames. ], batch size: 165, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:12:28,107 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37461.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 04:12:47,972 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0613, 3.7585, 3.8821, 2.6777, 3.6201, 3.8422, 3.6847, 1.9413], device='cuda:7'), covar=tensor([0.0330, 0.0026, 0.0026, 0.0197, 0.0030, 0.0046, 0.0027, 0.0287], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0052, 0.0056, 0.0113, 0.0058, 0.0065, 0.0060, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:13:26,313 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 04:13:29,624 INFO [train.py:904] (7/8) Epoch 4, batch 7050, loss[loss=0.2577, simple_loss=0.3358, pruned_loss=0.08984, over 16931.00 frames. ], tot_loss[loss=0.2712, simple_loss=0.3446, pruned_loss=0.09896, over 3107071.43 frames. ], batch size: 109, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:13:43,151 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37509.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 04:14:45,397 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 3.107e+02 4.211e+02 5.076e+02 6.594e+02 1.566e+03, threshold=1.015e+03, percent-clipped=2.0 2023-04-28 04:14:46,717 INFO [train.py:904] (7/8) Epoch 4, batch 7100, loss[loss=0.2578, simple_loss=0.333, pruned_loss=0.09132, over 15214.00 frames. ], tot_loss[loss=0.2688, simple_loss=0.3422, pruned_loss=0.09771, over 3109818.44 frames. ], batch size: 191, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:15:13,071 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:02,665 INFO [train.py:904] (7/8) Epoch 4, batch 7150, loss[loss=0.3206, simple_loss=0.3591, pruned_loss=0.1411, over 11024.00 frames. ], tot_loss[loss=0.2666, simple_loss=0.3396, pruned_loss=0.09681, over 3115319.98 frames. ], batch size: 247, lr: 1.58e-02, grad_scale: 2.0 2023-04-28 04:16:18,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2637, 2.1261, 2.0864, 3.7591, 1.7140, 3.0867, 2.1943, 2.0170], device='cuda:7'), covar=tensor([0.0532, 0.1479, 0.0867, 0.0276, 0.2677, 0.0575, 0.1461, 0.2172], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0283, 0.0235, 0.0297, 0.0352, 0.0263, 0.0261, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 04:16:23,209 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37614.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:24,834 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37615.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:16:34,266 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0013, 1.2678, 1.6134, 1.7910, 2.1193, 2.1645, 1.3904, 2.0262], device='cuda:7'), covar=tensor([0.0067, 0.0211, 0.0121, 0.0121, 0.0087, 0.0069, 0.0193, 0.0051], device='cuda:7'), in_proj_covar=tensor([0.0102, 0.0132, 0.0119, 0.0112, 0.0118, 0.0082, 0.0132, 0.0075], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 04:16:38,280 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37624.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:17:17,126 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.158e+02 3.757e+02 4.855e+02 6.210e+02 1.451e+03, threshold=9.710e+02, percent-clipped=1.0 2023-04-28 04:17:18,898 INFO [train.py:904] (7/8) Epoch 4, batch 7200, loss[loss=0.1985, simple_loss=0.2871, pruned_loss=0.05495, over 16515.00 frames. ], tot_loss[loss=0.2628, simple_loss=0.3366, pruned_loss=0.09451, over 3104384.77 frames. ], batch size: 68, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:17:33,276 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37660.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:17:36,208 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37662.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:18:40,025 INFO [train.py:904] (7/8) Epoch 4, batch 7250, loss[loss=0.2598, simple_loss=0.3381, pruned_loss=0.09077, over 16443.00 frames. ], tot_loss[loss=0.2592, simple_loss=0.3335, pruned_loss=0.0925, over 3097374.58 frames. ], batch size: 146, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:18:52,049 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37708.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:19:10,340 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37720.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:19:14,636 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6102, 4.5469, 4.5245, 2.0726, 4.8095, 4.8767, 3.3688, 3.6538], device='cuda:7'), covar=tensor([0.0762, 0.0065, 0.0105, 0.1078, 0.0028, 0.0029, 0.0262, 0.0309], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0082, 0.0081, 0.0140, 0.0068, 0.0071, 0.0112, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 04:19:40,446 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8081, 3.0313, 2.3845, 4.0913, 3.6396, 3.8288, 1.6025, 2.9560], device='cuda:7'), covar=tensor([0.1226, 0.0432, 0.1084, 0.0068, 0.0241, 0.0296, 0.1228, 0.0630], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0138, 0.0166, 0.0076, 0.0159, 0.0160, 0.0157, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 04:19:55,472 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.396e+02 3.869e+02 4.810e+02 5.847e+02 1.256e+03, threshold=9.620e+02, percent-clipped=3.0 2023-04-28 04:19:57,425 INFO [train.py:904] (7/8) Epoch 4, batch 7300, loss[loss=0.2534, simple_loss=0.3388, pruned_loss=0.08407, over 16749.00 frames. ], tot_loss[loss=0.2589, simple_loss=0.3325, pruned_loss=0.09267, over 3087832.76 frames. ], batch size: 134, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:20:24,295 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6088, 2.6813, 1.6724, 2.7244, 2.1537, 2.7366, 1.9551, 2.3558], device='cuda:7'), covar=tensor([0.0113, 0.0240, 0.1057, 0.0065, 0.0533, 0.0286, 0.0924, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0144, 0.0175, 0.0077, 0.0160, 0.0174, 0.0185, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 04:20:32,873 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0007, 3.2651, 3.2164, 1.6139, 3.4442, 3.5361, 2.5081, 2.6054], device='cuda:7'), covar=tensor([0.0895, 0.0130, 0.0215, 0.1134, 0.0064, 0.0051, 0.0402, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0083, 0.0083, 0.0140, 0.0069, 0.0072, 0.0112, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 04:20:48,210 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9733, 3.6343, 3.1382, 1.6597, 2.5012, 2.0724, 3.2568, 3.5733], device='cuda:7'), covar=tensor([0.0262, 0.0432, 0.0507, 0.1672, 0.0837, 0.0923, 0.0606, 0.0582], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0124, 0.0156, 0.0143, 0.0138, 0.0129, 0.0144, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 04:21:14,975 INFO [train.py:904] (7/8) Epoch 4, batch 7350, loss[loss=0.2215, simple_loss=0.2997, pruned_loss=0.07166, over 17168.00 frames. ], tot_loss[loss=0.2587, simple_loss=0.3322, pruned_loss=0.09259, over 3062520.87 frames. ], batch size: 46, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:21:41,374 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 04:22:27,495 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 04:22:29,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.483e+02 3.907e+02 4.910e+02 6.467e+02 1.073e+03, threshold=9.820e+02, percent-clipped=2.0 2023-04-28 04:22:31,819 INFO [train.py:904] (7/8) Epoch 4, batch 7400, loss[loss=0.2711, simple_loss=0.3649, pruned_loss=0.08864, over 16897.00 frames. ], tot_loss[loss=0.2612, simple_loss=0.334, pruned_loss=0.09417, over 3059451.08 frames. ], batch size: 96, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:22:38,180 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-28 04:23:42,298 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4185, 3.0169, 2.5017, 2.2987, 2.4665, 2.0415, 2.9669, 3.0812], device='cuda:7'), covar=tensor([0.1636, 0.0668, 0.1094, 0.1040, 0.1673, 0.1221, 0.0438, 0.0563], device='cuda:7'), in_proj_covar=tensor([0.0276, 0.0247, 0.0262, 0.0233, 0.0304, 0.0195, 0.0228, 0.0235], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 04:23:49,019 INFO [train.py:904] (7/8) Epoch 4, batch 7450, loss[loss=0.3071, simple_loss=0.3679, pruned_loss=0.1232, over 16647.00 frames. ], tot_loss[loss=0.2626, simple_loss=0.3351, pruned_loss=0.09505, over 3088988.72 frames. ], batch size: 62, lr: 1.58e-02, grad_scale: 4.0 2023-04-28 04:23:55,467 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0910, 3.7508, 3.8052, 2.6872, 3.6830, 3.7831, 3.7673, 1.5876], device='cuda:7'), covar=tensor([0.0343, 0.0035, 0.0037, 0.0209, 0.0048, 0.0101, 0.0061, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0051, 0.0056, 0.0110, 0.0058, 0.0065, 0.0060, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:24:26,770 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37924.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:25:06,402 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.829e+02 4.545e+02 5.686e+02 6.937e+02 1.455e+03, threshold=1.137e+03, percent-clipped=4.0 2023-04-28 04:25:07,728 INFO [train.py:904] (7/8) Epoch 4, batch 7500, loss[loss=0.2532, simple_loss=0.3237, pruned_loss=0.09141, over 16871.00 frames. ], tot_loss[loss=0.2629, simple_loss=0.3357, pruned_loss=0.09503, over 3085952.19 frames. ], batch size: 102, lr: 1.57e-02, grad_scale: 4.0 2023-04-28 04:25:40,562 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=37972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:25:50,763 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37979.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:26:27,338 INFO [train.py:904] (7/8) Epoch 4, batch 7550, loss[loss=0.2352, simple_loss=0.3089, pruned_loss=0.08079, over 17041.00 frames. ], tot_loss[loss=0.2649, simple_loss=0.3361, pruned_loss=0.09689, over 3050556.76 frames. ], batch size: 53, lr: 1.57e-02, grad_scale: 4.0 2023-04-28 04:26:45,152 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7137, 3.5952, 3.7790, 3.6687, 3.7531, 4.1401, 3.8940, 3.6161], device='cuda:7'), covar=tensor([0.1731, 0.1992, 0.1498, 0.2090, 0.2718, 0.1480, 0.1339, 0.2429], device='cuda:7'), in_proj_covar=tensor([0.0251, 0.0356, 0.0337, 0.0303, 0.0404, 0.0369, 0.0283, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:26:54,917 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38020.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:27:24,783 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38040.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:27:38,143 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.661e+02 3.786e+02 4.899e+02 6.266e+02 1.464e+03, threshold=9.797e+02, percent-clipped=2.0 2023-04-28 04:27:40,075 INFO [train.py:904] (7/8) Epoch 4, batch 7600, loss[loss=0.2497, simple_loss=0.3297, pruned_loss=0.08486, over 16806.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3365, pruned_loss=0.09762, over 3056642.15 frames. ], batch size: 124, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:28:05,541 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=38068.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:28:55,059 INFO [train.py:904] (7/8) Epoch 4, batch 7650, loss[loss=0.2435, simple_loss=0.3258, pruned_loss=0.08064, over 16833.00 frames. ], tot_loss[loss=0.2674, simple_loss=0.3375, pruned_loss=0.09862, over 3057089.64 frames. ], batch size: 96, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:29:17,710 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 04:29:54,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0918, 3.7174, 3.6803, 2.5667, 3.3623, 3.5247, 3.3941, 1.8698], device='cuda:7'), covar=tensor([0.0309, 0.0019, 0.0028, 0.0207, 0.0037, 0.0056, 0.0034, 0.0300], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0051, 0.0057, 0.0113, 0.0060, 0.0067, 0.0062, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:30:08,819 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.664e+02 4.200e+02 5.231e+02 7.341e+02 1.286e+03, threshold=1.046e+03, percent-clipped=4.0 2023-04-28 04:30:09,980 INFO [train.py:904] (7/8) Epoch 4, batch 7700, loss[loss=0.2645, simple_loss=0.3363, pruned_loss=0.09633, over 16392.00 frames. ], tot_loss[loss=0.2681, simple_loss=0.3378, pruned_loss=0.09922, over 3042908.93 frames. ], batch size: 68, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:30:50,596 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 04:31:22,071 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 04:31:26,748 INFO [train.py:904] (7/8) Epoch 4, batch 7750, loss[loss=0.269, simple_loss=0.3342, pruned_loss=0.1019, over 15141.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.3376, pruned_loss=0.09852, over 3052970.57 frames. ], batch size: 190, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:32:40,404 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 4.651e+02 5.629e+02 7.358e+02 1.215e+03, threshold=1.126e+03, percent-clipped=5.0 2023-04-28 04:32:42,185 INFO [train.py:904] (7/8) Epoch 4, batch 7800, loss[loss=0.2768, simple_loss=0.3537, pruned_loss=0.0999, over 16230.00 frames. ], tot_loss[loss=0.2667, simple_loss=0.3377, pruned_loss=0.09784, over 3087446.93 frames. ], batch size: 165, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:33:58,992 INFO [train.py:904] (7/8) Epoch 4, batch 7850, loss[loss=0.3011, simple_loss=0.3483, pruned_loss=0.127, over 11439.00 frames. ], tot_loss[loss=0.2673, simple_loss=0.3388, pruned_loss=0.0979, over 3088711.08 frames. ], batch size: 248, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:34:50,598 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38335.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:35:12,461 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.779e+02 3.909e+02 4.709e+02 5.998e+02 1.164e+03, threshold=9.418e+02, percent-clipped=1.0 2023-04-28 04:35:14,726 INFO [train.py:904] (7/8) Epoch 4, batch 7900, loss[loss=0.3255, simple_loss=0.3646, pruned_loss=0.1432, over 11291.00 frames. ], tot_loss[loss=0.2647, simple_loss=0.337, pruned_loss=0.09617, over 3100550.74 frames. ], batch size: 247, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:35:27,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6888, 3.6350, 3.7040, 3.6714, 3.7093, 4.1352, 3.9211, 3.6348], device='cuda:7'), covar=tensor([0.1772, 0.1758, 0.1365, 0.1923, 0.2519, 0.1305, 0.1083, 0.2112], device='cuda:7'), in_proj_covar=tensor([0.0252, 0.0352, 0.0338, 0.0308, 0.0406, 0.0371, 0.0283, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:35:28,538 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:36:21,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7409, 1.4761, 2.0688, 2.5295, 2.5344, 2.9381, 1.5830, 2.7572], device='cuda:7'), covar=tensor([0.0053, 0.0207, 0.0117, 0.0101, 0.0083, 0.0050, 0.0181, 0.0048], device='cuda:7'), in_proj_covar=tensor([0.0101, 0.0130, 0.0116, 0.0110, 0.0115, 0.0080, 0.0127, 0.0073], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 04:36:34,681 INFO [train.py:904] (7/8) Epoch 4, batch 7950, loss[loss=0.2369, simple_loss=0.3126, pruned_loss=0.08062, over 17197.00 frames. ], tot_loss[loss=0.2642, simple_loss=0.3366, pruned_loss=0.09587, over 3120252.34 frames. ], batch size: 46, lr: 1.57e-02, grad_scale: 8.0 2023-04-28 04:37:05,121 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38421.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:37:49,212 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.619e+02 4.198e+02 4.896e+02 6.041e+02 1.381e+03, threshold=9.792e+02, percent-clipped=5.0 2023-04-28 04:37:50,963 INFO [train.py:904] (7/8) Epoch 4, batch 8000, loss[loss=0.2394, simple_loss=0.3158, pruned_loss=0.08152, over 16621.00 frames. ], tot_loss[loss=0.2663, simple_loss=0.3382, pruned_loss=0.09725, over 3119511.43 frames. ], batch size: 62, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:39:04,869 INFO [train.py:904] (7/8) Epoch 4, batch 8050, loss[loss=0.2654, simple_loss=0.3408, pruned_loss=0.09497, over 16748.00 frames. ], tot_loss[loss=0.2665, simple_loss=0.3384, pruned_loss=0.09737, over 3101250.92 frames. ], batch size: 124, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:40:21,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.546e+02 4.083e+02 4.841e+02 6.395e+02 1.721e+03, threshold=9.682e+02, percent-clipped=6.0 2023-04-28 04:40:23,275 INFO [train.py:904] (7/8) Epoch 4, batch 8100, loss[loss=0.2455, simple_loss=0.3267, pruned_loss=0.08216, over 16655.00 frames. ], tot_loss[loss=0.2659, simple_loss=0.3379, pruned_loss=0.09697, over 3085420.48 frames. ], batch size: 62, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:41:41,536 INFO [train.py:904] (7/8) Epoch 4, batch 8150, loss[loss=0.2439, simple_loss=0.315, pruned_loss=0.08638, over 16447.00 frames. ], tot_loss[loss=0.2629, simple_loss=0.335, pruned_loss=0.09543, over 3093933.60 frames. ], batch size: 68, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:42:06,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5459, 4.5462, 4.4528, 3.6586, 4.4383, 1.4970, 4.2670, 4.3547], device='cuda:7'), covar=tensor([0.0064, 0.0047, 0.0075, 0.0329, 0.0053, 0.1792, 0.0080, 0.0122], device='cuda:7'), in_proj_covar=tensor([0.0083, 0.0072, 0.0110, 0.0120, 0.0082, 0.0134, 0.0098, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 04:42:34,080 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38635.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:42:57,104 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.420e+02 4.459e+02 5.761e+02 7.393e+02 1.688e+03, threshold=1.152e+03, percent-clipped=6.0 2023-04-28 04:42:59,081 INFO [train.py:904] (7/8) Epoch 4, batch 8200, loss[loss=0.2696, simple_loss=0.3414, pruned_loss=0.09894, over 16425.00 frames. ], tot_loss[loss=0.2594, simple_loss=0.3317, pruned_loss=0.09358, over 3125170.16 frames. ], batch size: 146, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:43:53,410 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=38683.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:44:23,330 INFO [train.py:904] (7/8) Epoch 4, batch 8250, loss[loss=0.2485, simple_loss=0.3167, pruned_loss=0.09013, over 12066.00 frames. ], tot_loss[loss=0.257, simple_loss=0.331, pruned_loss=0.09154, over 3103022.61 frames. ], batch size: 248, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:44:49,022 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38716.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:45:27,636 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.54 vs. limit=5.0 2023-04-28 04:45:43,968 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.522e+02 4.130e+02 5.467e+02 1.123e+03, threshold=8.260e+02, percent-clipped=0.0 2023-04-28 04:45:45,839 INFO [train.py:904] (7/8) Epoch 4, batch 8300, loss[loss=0.2306, simple_loss=0.3051, pruned_loss=0.07804, over 12312.00 frames. ], tot_loss[loss=0.251, simple_loss=0.3264, pruned_loss=0.08779, over 3070856.60 frames. ], batch size: 248, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:46:17,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6214, 3.9534, 3.3589, 3.8555, 3.4186, 3.4971, 3.6640, 3.8965], device='cuda:7'), covar=tensor([0.1583, 0.1469, 0.2828, 0.0806, 0.1379, 0.2009, 0.1148, 0.1487], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0427, 0.0373, 0.0277, 0.0273, 0.0291, 0.0348, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 04:47:07,364 INFO [train.py:904] (7/8) Epoch 4, batch 8350, loss[loss=0.2466, simple_loss=0.3283, pruned_loss=0.08245, over 15402.00 frames. ], tot_loss[loss=0.2472, simple_loss=0.3247, pruned_loss=0.08484, over 3076213.66 frames. ], batch size: 191, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:47:43,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8390, 2.4851, 2.5597, 2.0121, 2.5156, 2.5394, 2.5395, 1.6699], device='cuda:7'), covar=tensor([0.0271, 0.0035, 0.0050, 0.0194, 0.0047, 0.0070, 0.0050, 0.0326], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0049, 0.0056, 0.0110, 0.0056, 0.0065, 0.0060, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 04:47:44,536 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38823.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:48:10,370 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 04:48:20,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.00 vs. limit=5.0 2023-04-28 04:48:29,403 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.235e+02 3.176e+02 3.884e+02 4.976e+02 1.172e+03, threshold=7.768e+02, percent-clipped=5.0 2023-04-28 04:48:30,631 INFO [train.py:904] (7/8) Epoch 4, batch 8400, loss[loss=0.225, simple_loss=0.3009, pruned_loss=0.07456, over 12128.00 frames. ], tot_loss[loss=0.2424, simple_loss=0.3208, pruned_loss=0.08196, over 3054001.01 frames. ], batch size: 246, lr: 1.56e-02, grad_scale: 8.0 2023-04-28 04:48:44,164 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-28 04:49:27,023 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38884.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:49:53,150 INFO [train.py:904] (7/8) Epoch 4, batch 8450, loss[loss=0.2244, simple_loss=0.2963, pruned_loss=0.07627, over 12582.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3186, pruned_loss=0.08001, over 3040081.85 frames. ], batch size: 247, lr: 1.56e-02, grad_scale: 4.0 2023-04-28 04:51:13,826 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.235e+02 3.132e+02 4.045e+02 5.414e+02 1.265e+03, threshold=8.090e+02, percent-clipped=8.0 2023-04-28 04:51:13,846 INFO [train.py:904] (7/8) Epoch 4, batch 8500, loss[loss=0.1996, simple_loss=0.2745, pruned_loss=0.06236, over 12351.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.3131, pruned_loss=0.0762, over 3027538.50 frames. ], batch size: 247, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:51:35,372 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 04:52:39,881 INFO [train.py:904] (7/8) Epoch 4, batch 8550, loss[loss=0.2122, simple_loss=0.2798, pruned_loss=0.07228, over 11708.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.3099, pruned_loss=0.07438, over 3030492.96 frames. ], batch size: 249, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:52:45,396 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4869, 3.4136, 3.3936, 2.8369, 3.3503, 1.9798, 3.2140, 2.9291], device='cuda:7'), covar=tensor([0.0097, 0.0078, 0.0094, 0.0211, 0.0078, 0.1450, 0.0097, 0.0149], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0068, 0.0106, 0.0109, 0.0078, 0.0130, 0.0095, 0.0102], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 04:53:09,966 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39016.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:54:03,151 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-28 04:54:17,521 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 04:54:21,294 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 3.391e+02 3.909e+02 5.367e+02 1.114e+03, threshold=7.819e+02, percent-clipped=4.0 2023-04-28 04:54:21,315 INFO [train.py:904] (7/8) Epoch 4, batch 8600, loss[loss=0.242, simple_loss=0.3261, pruned_loss=0.07893, over 16374.00 frames. ], tot_loss[loss=0.2295, simple_loss=0.3111, pruned_loss=0.07395, over 3022209.76 frames. ], batch size: 146, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:54:51,179 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39064.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:55:58,616 INFO [train.py:904] (7/8) Epoch 4, batch 8650, loss[loss=0.1982, simple_loss=0.2812, pruned_loss=0.05763, over 12145.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.3078, pruned_loss=0.0712, over 3023287.02 frames. ], batch size: 247, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:57:44,934 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 3.067e+02 3.859e+02 4.894e+02 7.251e+02, threshold=7.718e+02, percent-clipped=0.0 2023-04-28 04:57:44,954 INFO [train.py:904] (7/8) Epoch 4, batch 8700, loss[loss=0.2061, simple_loss=0.2845, pruned_loss=0.06387, over 12335.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3044, pruned_loss=0.06915, over 3041135.93 frames. ], batch size: 248, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 04:58:36,857 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39179.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:58:50,426 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39185.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 04:59:20,257 INFO [train.py:904] (7/8) Epoch 4, batch 8750, loss[loss=0.2362, simple_loss=0.3274, pruned_loss=0.07244, over 16636.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.3045, pruned_loss=0.0687, over 3044743.43 frames. ], batch size: 89, lr: 1.55e-02, grad_scale: 2.0 2023-04-28 04:59:53,060 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 05:00:03,823 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4886, 4.0118, 4.0166, 2.7432, 3.7590, 4.1328, 3.9561, 1.9620], device='cuda:7'), covar=tensor([0.0286, 0.0018, 0.0026, 0.0207, 0.0033, 0.0031, 0.0033, 0.0369], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0049, 0.0056, 0.0109, 0.0055, 0.0062, 0.0059, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 05:00:43,050 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.71 vs. limit=2.0 2023-04-28 05:01:05,228 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39246.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:01:14,457 INFO [train.py:904] (7/8) Epoch 4, batch 8800, loss[loss=0.256, simple_loss=0.3332, pruned_loss=0.08933, over 12360.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3022, pruned_loss=0.06747, over 3037912.54 frames. ], batch size: 247, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 05:01:15,994 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.302e+02 3.583e+02 4.376e+02 5.208e+02 1.205e+03, threshold=8.753e+02, percent-clipped=6.0 2023-04-28 05:02:23,710 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39284.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:02:58,311 INFO [train.py:904] (7/8) Epoch 4, batch 8850, loss[loss=0.237, simple_loss=0.3262, pruned_loss=0.07389, over 16375.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.3041, pruned_loss=0.0667, over 3019566.52 frames. ], batch size: 146, lr: 1.55e-02, grad_scale: 4.0 2023-04-28 05:03:59,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6181, 3.4376, 3.0740, 1.7288, 2.6776, 2.1012, 2.9897, 3.1538], device='cuda:7'), covar=tensor([0.0289, 0.0452, 0.0474, 0.1593, 0.0669, 0.0900, 0.0717, 0.0815], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0115, 0.0154, 0.0141, 0.0133, 0.0128, 0.0139, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 05:04:32,477 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39345.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:04:44,189 INFO [train.py:904] (7/8) Epoch 4, batch 8900, loss[loss=0.2271, simple_loss=0.3132, pruned_loss=0.07049, over 16276.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3044, pruned_loss=0.06599, over 3027435.44 frames. ], batch size: 165, lr: 1.55e-02, grad_scale: 2.0 2023-04-28 05:04:49,521 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.906e+02 3.441e+02 4.038e+02 4.851e+02 9.886e+02, threshold=8.076e+02, percent-clipped=2.0 2023-04-28 05:05:20,699 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3751, 4.2559, 4.2287, 3.7262, 4.2075, 1.8477, 4.0177, 4.2698], device='cuda:7'), covar=tensor([0.0068, 0.0060, 0.0085, 0.0221, 0.0059, 0.1425, 0.0079, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0069, 0.0107, 0.0107, 0.0079, 0.0131, 0.0094, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:06:43,282 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-28 05:06:47,809 INFO [train.py:904] (7/8) Epoch 4, batch 8950, loss[loss=0.2458, simple_loss=0.3131, pruned_loss=0.0893, over 12404.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.304, pruned_loss=0.06609, over 3046356.63 frames. ], batch size: 247, lr: 1.55e-02, grad_scale: 2.0 2023-04-28 05:06:52,085 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-28 05:08:00,701 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-28 05:08:20,000 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-28 05:08:25,766 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9519, 4.1899, 3.9739, 4.0228, 3.7241, 3.6943, 3.8755, 4.1019], device='cuda:7'), covar=tensor([0.0657, 0.0740, 0.0930, 0.0407, 0.0612, 0.1132, 0.0598, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0425, 0.0364, 0.0274, 0.0272, 0.0287, 0.0342, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:08:35,746 INFO [train.py:904] (7/8) Epoch 4, batch 9000, loss[loss=0.1891, simple_loss=0.2779, pruned_loss=0.05014, over 16979.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.3005, pruned_loss=0.06427, over 3058211.00 frames. ], batch size: 116, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:08:35,747 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 05:08:45,784 INFO [train.py:938] (7/8) Epoch 4, validation: loss=0.1802, simple_loss=0.283, pruned_loss=0.0387, over 944034.00 frames. 2023-04-28 05:08:45,785 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 05:08:49,857 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.055e+02 3.044e+02 3.757e+02 4.852e+02 9.817e+02, threshold=7.514e+02, percent-clipped=1.0 2023-04-28 05:09:11,759 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39463.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:09:12,114 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-28 05:09:43,577 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39479.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:10:29,883 INFO [train.py:904] (7/8) Epoch 4, batch 9050, loss[loss=0.1851, simple_loss=0.2716, pruned_loss=0.04934, over 16887.00 frames. ], tot_loss[loss=0.217, simple_loss=0.3029, pruned_loss=0.06555, over 3079425.34 frames. ], batch size: 116, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:11:17,785 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39524.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:11:21,929 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39527.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:11:54,716 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39541.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:12:14,620 INFO [train.py:904] (7/8) Epoch 4, batch 9100, loss[loss=0.214, simple_loss=0.306, pruned_loss=0.06099, over 16451.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.3028, pruned_loss=0.06633, over 3061785.30 frames. ], batch size: 146, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:12:15,703 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7115, 4.9904, 4.7445, 4.7308, 4.3167, 4.2516, 4.4177, 4.9796], device='cuda:7'), covar=tensor([0.0561, 0.0639, 0.0809, 0.0444, 0.0669, 0.0820, 0.0551, 0.0742], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0419, 0.0359, 0.0270, 0.0272, 0.0286, 0.0337, 0.0305], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:12:18,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.564e+02 3.376e+02 4.042e+02 5.182e+02 1.418e+03, threshold=8.084e+02, percent-clipped=5.0 2023-04-28 05:14:15,390 INFO [train.py:904] (7/8) Epoch 4, batch 9150, loss[loss=0.1997, simple_loss=0.2881, pruned_loss=0.0556, over 16395.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.3027, pruned_loss=0.06552, over 3054108.17 frames. ], batch size: 146, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:15:41,163 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39639.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 05:15:43,182 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39640.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:15:58,308 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7308, 4.5482, 4.7404, 4.9714, 5.0451, 4.4144, 5.0507, 5.0163], device='cuda:7'), covar=tensor([0.0681, 0.0618, 0.0873, 0.0395, 0.0426, 0.0543, 0.0397, 0.0422], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0395, 0.0496, 0.0400, 0.0305, 0.0295, 0.0326, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:16:00,933 INFO [train.py:904] (7/8) Epoch 4, batch 9200, loss[loss=0.1767, simple_loss=0.2589, pruned_loss=0.04727, over 12378.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2979, pruned_loss=0.06418, over 3053635.88 frames. ], batch size: 248, lr: 1.54e-02, grad_scale: 4.0 2023-04-28 05:16:04,315 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.381e+02 3.414e+02 4.371e+02 6.281e+02 1.474e+03, threshold=8.741e+02, percent-clipped=12.0 2023-04-28 05:17:18,252 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7096, 4.1853, 4.3664, 3.0639, 3.9628, 4.3900, 4.1828, 2.5271], device='cuda:7'), covar=tensor([0.0273, 0.0012, 0.0020, 0.0173, 0.0025, 0.0022, 0.0020, 0.0262], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0049, 0.0055, 0.0107, 0.0055, 0.0061, 0.0058, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 05:17:21,030 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-04-28 05:17:35,064 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39700.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 05:17:35,651 INFO [train.py:904] (7/8) Epoch 4, batch 9250, loss[loss=0.2232, simple_loss=0.3075, pruned_loss=0.0695, over 15363.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2975, pruned_loss=0.06408, over 3057463.52 frames. ], batch size: 192, lr: 1.54e-02, grad_scale: 4.0 2023-04-28 05:17:52,689 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 05:19:26,138 INFO [train.py:904] (7/8) Epoch 4, batch 9300, loss[loss=0.1935, simple_loss=0.2729, pruned_loss=0.05706, over 12318.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2957, pruned_loss=0.06343, over 3034807.29 frames. ], batch size: 250, lr: 1.54e-02, grad_scale: 4.0 2023-04-28 05:19:30,023 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.165e+02 3.034e+02 3.554e+02 4.321e+02 7.893e+02, threshold=7.107e+02, percent-clipped=0.0 2023-04-28 05:20:06,151 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39767.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:21:11,789 INFO [train.py:904] (7/8) Epoch 4, batch 9350, loss[loss=0.2164, simple_loss=0.2938, pruned_loss=0.06949, over 16669.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2956, pruned_loss=0.06332, over 3053645.17 frames. ], batch size: 134, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:21:41,330 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39815.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:21:48,609 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39819.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:22:05,665 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39828.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:22:30,615 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39841.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:22:51,023 INFO [train.py:904] (7/8) Epoch 4, batch 9400, loss[loss=0.1977, simple_loss=0.2781, pruned_loss=0.05862, over 12444.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2957, pruned_loss=0.06315, over 3058289.77 frames. ], batch size: 246, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:22:57,598 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.054e+02 3.057e+02 3.854e+02 4.791e+02 1.161e+03, threshold=7.709e+02, percent-clipped=3.0 2023-04-28 05:23:22,521 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7322, 3.6012, 3.7051, 3.9260, 3.9609, 3.5074, 3.9352, 3.9733], device='cuda:7'), covar=tensor([0.0810, 0.0654, 0.1120, 0.0498, 0.0488, 0.1484, 0.0724, 0.0448], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0390, 0.0491, 0.0399, 0.0298, 0.0289, 0.0321, 0.0325], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:23:42,109 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39876.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:24:10,668 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39889.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:24:32,564 INFO [train.py:904] (7/8) Epoch 4, batch 9450, loss[loss=0.1972, simple_loss=0.2792, pruned_loss=0.05764, over 16919.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2974, pruned_loss=0.06342, over 3052594.47 frames. ], batch size: 109, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:25:54,126 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:26:13,854 INFO [train.py:904] (7/8) Epoch 4, batch 9500, loss[loss=0.1877, simple_loss=0.2842, pruned_loss=0.04559, over 16915.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2969, pruned_loss=0.06295, over 3066426.44 frames. ], batch size: 96, lr: 1.54e-02, grad_scale: 2.0 2023-04-28 05:26:21,226 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.115e+02 3.608e+02 4.489e+02 5.705e+02 9.491e+02, threshold=8.979e+02, percent-clipped=6.0 2023-04-28 05:27:31,420 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=39988.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:27:48,799 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39995.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 05:28:04,417 INFO [train.py:904] (7/8) Epoch 4, batch 9550, loss[loss=0.268, simple_loss=0.346, pruned_loss=0.09497, over 15407.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2961, pruned_loss=0.06288, over 3066855.21 frames. ], batch size: 192, lr: 1.53e-02, grad_scale: 2.0 2023-04-28 05:29:25,367 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8040, 3.0156, 3.0802, 1.9286, 2.8508, 2.9950, 2.9832, 1.7234], device='cuda:7'), covar=tensor([0.0352, 0.0027, 0.0036, 0.0257, 0.0039, 0.0044, 0.0033, 0.0312], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0050, 0.0055, 0.0108, 0.0053, 0.0063, 0.0058, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 05:29:46,549 INFO [train.py:904] (7/8) Epoch 4, batch 9600, loss[loss=0.2084, simple_loss=0.2955, pruned_loss=0.06058, over 16640.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2973, pruned_loss=0.06379, over 3060580.38 frames. ], batch size: 57, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:29:52,057 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.278e+02 3.502e+02 4.489e+02 5.494e+02 1.109e+03, threshold=8.977e+02, percent-clipped=3.0 2023-04-28 05:29:58,667 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 05:30:29,008 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3902, 3.3492, 2.8173, 2.0471, 2.3770, 2.0629, 3.3478, 3.3501], device='cuda:7'), covar=tensor([0.2268, 0.0656, 0.1132, 0.1567, 0.1887, 0.1338, 0.0414, 0.0623], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0239, 0.0255, 0.0231, 0.0241, 0.0191, 0.0222, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:31:26,237 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2845, 1.3107, 1.8235, 2.1753, 2.1352, 2.2750, 1.4169, 2.1806], device='cuda:7'), covar=tensor([0.0076, 0.0232, 0.0139, 0.0115, 0.0113, 0.0094, 0.0233, 0.0061], device='cuda:7'), in_proj_covar=tensor([0.0103, 0.0130, 0.0117, 0.0113, 0.0114, 0.0080, 0.0130, 0.0071], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 05:31:33,081 INFO [train.py:904] (7/8) Epoch 4, batch 9650, loss[loss=0.2258, simple_loss=0.2999, pruned_loss=0.07579, over 12345.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2998, pruned_loss=0.0644, over 3058219.02 frames. ], batch size: 250, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:32:12,965 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 05:32:18,223 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40119.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:32:25,203 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40123.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:33:07,409 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6118, 3.4818, 3.9993, 4.0149, 3.9764, 3.6054, 3.7179, 3.6498], device='cuda:7'), covar=tensor([0.0236, 0.0444, 0.0393, 0.0441, 0.0368, 0.0325, 0.0669, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0190, 0.0198, 0.0202, 0.0231, 0.0208, 0.0292, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-28 05:33:21,184 INFO [train.py:904] (7/8) Epoch 4, batch 9700, loss[loss=0.1975, simple_loss=0.2842, pruned_loss=0.0554, over 15275.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2989, pruned_loss=0.06414, over 3062788.02 frames. ], batch size: 190, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:33:26,547 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.154e+02 3.870e+02 5.322e+02 1.510e+03, threshold=7.740e+02, percent-clipped=2.0 2023-04-28 05:33:53,099 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40167.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:34:01,359 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40171.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:34:54,797 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40195.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:35:03,566 INFO [train.py:904] (7/8) Epoch 4, batch 9750, loss[loss=0.2036, simple_loss=0.295, pruned_loss=0.05613, over 15290.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2987, pruned_loss=0.06443, over 3059634.80 frames. ], batch size: 190, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:35:25,423 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7556, 3.9423, 2.9006, 2.2233, 2.6613, 2.0161, 3.9293, 3.8208], device='cuda:7'), covar=tensor([0.1947, 0.0459, 0.1259, 0.1465, 0.1897, 0.1520, 0.0336, 0.0337], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0239, 0.0255, 0.0231, 0.0242, 0.0191, 0.0222, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:35:34,922 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0437, 2.4151, 2.3103, 3.3015, 3.0167, 3.3414, 1.6949, 2.7190], device='cuda:7'), covar=tensor([0.1062, 0.0489, 0.0909, 0.0080, 0.0183, 0.0348, 0.1155, 0.0612], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0135, 0.0163, 0.0074, 0.0139, 0.0158, 0.0159, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 05:35:38,843 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40218.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:36:45,117 INFO [train.py:904] (7/8) Epoch 4, batch 9800, loss[loss=0.2203, simple_loss=0.327, pruned_loss=0.05682, over 16372.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2985, pruned_loss=0.06298, over 3066264.32 frames. ], batch size: 68, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:36:51,071 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.170e+02 3.844e+02 4.650e+02 7.827e+02, threshold=7.689e+02, percent-clipped=1.0 2023-04-28 05:36:56,601 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40256.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:37:25,589 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:37:37,844 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40279.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:38:18,090 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40295.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 05:38:29,694 INFO [train.py:904] (7/8) Epoch 4, batch 9850, loss[loss=0.2008, simple_loss=0.2976, pruned_loss=0.05199, over 16678.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2992, pruned_loss=0.06267, over 3061190.82 frames. ], batch size: 89, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:39:23,334 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-04-28 05:39:25,090 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-04-28 05:39:31,420 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40330.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:39:39,929 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:40:05,942 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40343.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 05:40:21,703 INFO [train.py:904] (7/8) Epoch 4, batch 9900, loss[loss=0.2163, simple_loss=0.3128, pruned_loss=0.05989, over 16690.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2996, pruned_loss=0.06285, over 3045947.97 frames. ], batch size: 134, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:40:27,912 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.084e+02 3.210e+02 3.891e+02 5.069e+02 8.589e+02, threshold=7.781e+02, percent-clipped=1.0 2023-04-28 05:41:01,923 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 05:41:57,021 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40391.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:42:18,088 INFO [train.py:904] (7/8) Epoch 4, batch 9950, loss[loss=0.2384, simple_loss=0.3123, pruned_loss=0.08225, over 12303.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.302, pruned_loss=0.06352, over 3046823.71 frames. ], batch size: 250, lr: 1.53e-02, grad_scale: 4.0 2023-04-28 05:43:13,923 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40423.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:44:20,294 INFO [train.py:904] (7/8) Epoch 4, batch 10000, loss[loss=0.205, simple_loss=0.2989, pruned_loss=0.05554, over 16778.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2994, pruned_loss=0.06195, over 3058536.18 frames. ], batch size: 76, lr: 1.53e-02, grad_scale: 8.0 2023-04-28 05:44:26,642 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.009e+02 3.138e+02 3.849e+02 4.764e+02 1.083e+03, threshold=7.697e+02, percent-clipped=5.0 2023-04-28 05:44:30,974 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4249, 3.7897, 1.6455, 3.9222, 2.3554, 3.8746, 1.7139, 2.5953], device='cuda:7'), covar=tensor([0.0130, 0.0201, 0.1696, 0.0047, 0.0970, 0.0348, 0.1743, 0.0783], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0135, 0.0170, 0.0077, 0.0153, 0.0163, 0.0181, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 05:45:01,304 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:45:01,450 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:45:42,499 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-04-28 05:45:47,314 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:46:02,940 INFO [train.py:904] (7/8) Epoch 4, batch 10050, loss[loss=0.2418, simple_loss=0.3274, pruned_loss=0.0781, over 15434.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2996, pruned_loss=0.06217, over 3046794.97 frames. ], batch size: 191, lr: 1.53e-02, grad_scale: 8.0 2023-04-28 05:46:37,303 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3754, 2.9260, 2.5895, 2.2760, 2.1377, 2.0208, 2.8908, 2.9984], device='cuda:7'), covar=tensor([0.1670, 0.0690, 0.0980, 0.1204, 0.1708, 0.1284, 0.0410, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0262, 0.0238, 0.0254, 0.0230, 0.0236, 0.0190, 0.0221, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:46:38,831 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40519.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:46:54,168 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7916, 3.6874, 3.7912, 3.7379, 3.8448, 4.1697, 3.9590, 3.6208], device='cuda:7'), covar=tensor([0.1501, 0.1641, 0.1335, 0.1811, 0.2104, 0.1251, 0.1056, 0.1989], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0332, 0.0323, 0.0290, 0.0376, 0.0350, 0.0267, 0.0381], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:47:38,804 INFO [train.py:904] (7/8) Epoch 4, batch 10100, loss[loss=0.1975, simple_loss=0.2697, pruned_loss=0.06268, over 12567.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.3001, pruned_loss=0.0628, over 3043112.39 frames. ], batch size: 248, lr: 1.52e-02, grad_scale: 8.0 2023-04-28 05:47:39,375 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40551.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:47:42,253 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40553.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:47:42,817 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.191e+02 3.527e+02 4.401e+02 5.596e+02 1.013e+03, threshold=8.802e+02, percent-clipped=4.0 2023-04-28 05:48:27,667 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40574.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:49:25,186 INFO [train.py:904] (7/8) Epoch 5, batch 0, loss[loss=0.3602, simple_loss=0.4027, pruned_loss=0.1589, over 16327.00 frames. ], tot_loss[loss=0.3602, simple_loss=0.4027, pruned_loss=0.1589, over 16327.00 frames. ], batch size: 165, lr: 1.42e-02, grad_scale: 8.0 2023-04-28 05:49:25,186 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 05:49:32,553 INFO [train.py:938] (7/8) Epoch 5, validation: loss=0.1789, simple_loss=0.2817, pruned_loss=0.03802, over 944034.00 frames. 2023-04-28 05:49:32,554 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 05:50:11,891 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40628.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:50:31,799 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7789, 4.1518, 4.3067, 1.9963, 4.5793, 4.6327, 3.4553, 3.5146], device='cuda:7'), covar=tensor([0.0615, 0.0091, 0.0169, 0.1012, 0.0035, 0.0035, 0.0231, 0.0300], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0083, 0.0076, 0.0141, 0.0068, 0.0072, 0.0109, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 05:50:42,814 INFO [train.py:904] (7/8) Epoch 5, batch 50, loss[loss=0.2711, simple_loss=0.3448, pruned_loss=0.09866, over 16641.00 frames. ], tot_loss[loss=0.2591, simple_loss=0.3266, pruned_loss=0.09583, over 757929.75 frames. ], batch size: 62, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:50:43,278 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40651.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:50:49,841 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.398e+02 3.847e+02 4.852e+02 6.011e+02 1.299e+03, threshold=9.705e+02, percent-clipped=3.0 2023-04-28 05:51:29,907 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40686.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:51:50,913 INFO [train.py:904] (7/8) Epoch 5, batch 100, loss[loss=0.2466, simple_loss=0.3166, pruned_loss=0.08832, over 16277.00 frames. ], tot_loss[loss=0.2484, simple_loss=0.3175, pruned_loss=0.08965, over 1311706.60 frames. ], batch size: 165, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:52:07,382 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40712.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:52:59,302 INFO [train.py:904] (7/8) Epoch 5, batch 150, loss[loss=0.306, simple_loss=0.3587, pruned_loss=0.1266, over 12397.00 frames. ], tot_loss[loss=0.2459, simple_loss=0.3154, pruned_loss=0.08823, over 1742277.78 frames. ], batch size: 247, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:53:08,168 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.263e+02 3.680e+02 4.214e+02 5.175e+02 1.144e+03, threshold=8.427e+02, percent-clipped=1.0 2023-04-28 05:53:28,962 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0834, 2.5055, 2.5204, 4.6060, 1.9358, 3.5363, 2.4485, 2.5088], device='cuda:7'), covar=tensor([0.0491, 0.1758, 0.0917, 0.0261, 0.2932, 0.0725, 0.1664, 0.2372], device='cuda:7'), in_proj_covar=tensor([0.0297, 0.0299, 0.0242, 0.0299, 0.0354, 0.0273, 0.0266, 0.0354], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:54:09,233 INFO [train.py:904] (7/8) Epoch 5, batch 200, loss[loss=0.2719, simple_loss=0.3265, pruned_loss=0.1086, over 16196.00 frames. ], tot_loss[loss=0.2414, simple_loss=0.3127, pruned_loss=0.08508, over 2096337.32 frames. ], batch size: 164, lr: 1.42e-02, grad_scale: 1.0 2023-04-28 05:54:17,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2101, 4.5440, 2.2825, 4.9766, 3.0877, 4.8651, 2.4745, 3.5121], device='cuda:7'), covar=tensor([0.0095, 0.0227, 0.1345, 0.0025, 0.0640, 0.0262, 0.1231, 0.0498], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0146, 0.0176, 0.0081, 0.0158, 0.0175, 0.0185, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 05:55:13,938 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:55:17,648 INFO [train.py:904] (7/8) Epoch 5, batch 250, loss[loss=0.2146, simple_loss=0.2873, pruned_loss=0.07098, over 16532.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.309, pruned_loss=0.0833, over 2366778.03 frames. ], batch size: 68, lr: 1.41e-02, grad_scale: 1.0 2023-04-28 05:55:18,144 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40851.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:55:20,612 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8649, 4.0960, 3.3634, 2.3093, 3.0949, 2.2139, 4.3928, 4.1832], device='cuda:7'), covar=tensor([0.2070, 0.0626, 0.1034, 0.1554, 0.2129, 0.1523, 0.0303, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0252, 0.0268, 0.0241, 0.0277, 0.0202, 0.0233, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:55:25,559 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.087e+02 3.707e+02 4.511e+02 5.174e+02 8.449e+02, threshold=9.022e+02, percent-clipped=1.0 2023-04-28 05:55:38,046 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40865.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:55:50,373 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40874.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:56:24,983 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40899.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:56:26,970 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40900.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:56:27,692 INFO [train.py:904] (7/8) Epoch 5, batch 300, loss[loss=0.253, simple_loss=0.3199, pruned_loss=0.09301, over 16902.00 frames. ], tot_loss[loss=0.2331, simple_loss=0.3044, pruned_loss=0.0809, over 2574559.63 frames. ], batch size: 109, lr: 1.41e-02, grad_scale: 1.0 2023-04-28 05:56:58,131 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40922.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:57:04,974 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2929, 5.6586, 5.4266, 5.5663, 4.9201, 4.7868, 5.2889, 5.8298], device='cuda:7'), covar=tensor([0.0878, 0.0804, 0.0895, 0.0415, 0.0755, 0.0665, 0.0643, 0.0732], device='cuda:7'), in_proj_covar=tensor([0.0356, 0.0478, 0.0411, 0.0310, 0.0305, 0.0317, 0.0387, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 05:57:05,109 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40926.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:57:07,813 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40928.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:57:21,185 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3983, 1.3151, 1.9221, 2.1511, 2.4188, 2.4203, 1.2939, 2.3028], device='cuda:7'), covar=tensor([0.0095, 0.0235, 0.0127, 0.0134, 0.0076, 0.0087, 0.0236, 0.0065], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0138, 0.0123, 0.0122, 0.0121, 0.0085, 0.0136, 0.0075], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 05:57:39,691 INFO [train.py:904] (7/8) Epoch 5, batch 350, loss[loss=0.2275, simple_loss=0.2915, pruned_loss=0.08179, over 15795.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.2995, pruned_loss=0.07735, over 2747404.58 frames. ], batch size: 35, lr: 1.41e-02, grad_scale: 1.0 2023-04-28 05:57:48,106 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.380e+02 3.200e+02 3.923e+02 5.275e+02 9.514e+02, threshold=7.845e+02, percent-clipped=1.0 2023-04-28 05:57:54,168 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40961.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 05:58:14,897 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=40976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:58:28,690 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40986.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:58:44,221 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40997.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:58:48,861 INFO [train.py:904] (7/8) Epoch 5, batch 400, loss[loss=0.2556, simple_loss=0.3131, pruned_loss=0.09906, over 16873.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.2985, pruned_loss=0.07704, over 2878370.32 frames. ], batch size: 116, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 05:58:55,638 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2303, 4.6003, 2.6755, 4.9961, 2.9530, 4.8702, 2.5906, 3.2838], device='cuda:7'), covar=tensor([0.0105, 0.0181, 0.1202, 0.0023, 0.0667, 0.0210, 0.1125, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0148, 0.0175, 0.0082, 0.0157, 0.0177, 0.0183, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 05:58:59,030 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41007.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:59:22,410 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-04-28 05:59:36,657 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41034.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 05:59:51,837 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41045.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:00:00,532 INFO [train.py:904] (7/8) Epoch 5, batch 450, loss[loss=0.1869, simple_loss=0.2655, pruned_loss=0.05414, over 16998.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.2959, pruned_loss=0.07544, over 2982069.40 frames. ], batch size: 41, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:00:09,413 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 3.241e+02 3.884e+02 4.938e+02 1.053e+03, threshold=7.768e+02, percent-clipped=3.0 2023-04-28 06:00:11,783 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41058.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:00:16,677 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0756, 2.3748, 2.3851, 4.6699, 1.9553, 3.8633, 2.5005, 2.5129], device='cuda:7'), covar=tensor([0.0460, 0.1811, 0.0901, 0.0241, 0.2784, 0.0595, 0.1608, 0.2302], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0304, 0.0245, 0.0304, 0.0360, 0.0282, 0.0272, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:00:42,521 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 06:01:01,583 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5694, 4.0397, 4.2817, 1.6719, 4.4640, 4.5844, 3.1760, 3.4678], device='cuda:7'), covar=tensor([0.0726, 0.0121, 0.0175, 0.1206, 0.0051, 0.0058, 0.0327, 0.0356], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0087, 0.0082, 0.0144, 0.0071, 0.0077, 0.0113, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 06:01:10,616 INFO [train.py:904] (7/8) Epoch 5, batch 500, loss[loss=0.1969, simple_loss=0.2742, pruned_loss=0.05979, over 17243.00 frames. ], tot_loss[loss=0.2211, simple_loss=0.2937, pruned_loss=0.07424, over 3053640.29 frames. ], batch size: 45, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:01:12,277 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41102.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:01:17,428 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41106.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:02:15,305 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41148.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:02:19,026 INFO [train.py:904] (7/8) Epoch 5, batch 550, loss[loss=0.214, simple_loss=0.2822, pruned_loss=0.07292, over 12309.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.2927, pruned_loss=0.07326, over 3106202.40 frames. ], batch size: 247, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:02:24,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5761, 6.0028, 5.7313, 5.8021, 5.2276, 5.0945, 5.5227, 6.1420], device='cuda:7'), covar=tensor([0.0735, 0.0720, 0.1018, 0.0387, 0.0593, 0.0543, 0.0562, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0361, 0.0495, 0.0418, 0.0317, 0.0313, 0.0316, 0.0393, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:02:27,245 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.046e+02 3.287e+02 3.927e+02 4.639e+02 9.245e+02, threshold=7.855e+02, percent-clipped=2.0 2023-04-28 06:02:36,057 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41163.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:03:22,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41196.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:03:27,700 INFO [train.py:904] (7/8) Epoch 5, batch 600, loss[loss=0.2245, simple_loss=0.3058, pruned_loss=0.07157, over 17133.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.2928, pruned_loss=0.07395, over 3155904.13 frames. ], batch size: 48, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:03:29,273 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1911, 4.8672, 5.0873, 5.4548, 5.5386, 4.7972, 5.5191, 5.4526], device='cuda:7'), covar=tensor([0.0949, 0.0741, 0.1383, 0.0408, 0.0363, 0.0500, 0.0378, 0.0420], device='cuda:7'), in_proj_covar=tensor([0.0382, 0.0472, 0.0603, 0.0476, 0.0355, 0.0348, 0.0379, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:03:39,979 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-04-28 06:03:47,573 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-04-28 06:03:54,867 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41221.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:04:23,642 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.02 vs. limit=5.0 2023-04-28 06:04:33,942 INFO [train.py:904] (7/8) Epoch 5, batch 650, loss[loss=0.2132, simple_loss=0.2837, pruned_loss=0.07135, over 16711.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.2912, pruned_loss=0.07273, over 3189142.24 frames. ], batch size: 124, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:04:41,766 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41256.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 06:04:42,118 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-28 06:04:42,450 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.330e+02 3.128e+02 3.726e+02 4.799e+02 9.270e+02, threshold=7.452e+02, percent-clipped=1.0 2023-04-28 06:05:39,911 INFO [train.py:904] (7/8) Epoch 5, batch 700, loss[loss=0.2283, simple_loss=0.2902, pruned_loss=0.08322, over 16759.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.291, pruned_loss=0.07218, over 3213523.34 frames. ], batch size: 124, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:05:49,024 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41307.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:06:49,379 INFO [train.py:904] (7/8) Epoch 5, batch 750, loss[loss=0.2364, simple_loss=0.2966, pruned_loss=0.08809, over 16873.00 frames. ], tot_loss[loss=0.218, simple_loss=0.2912, pruned_loss=0.07233, over 3245865.68 frames. ], batch size: 109, lr: 1.41e-02, grad_scale: 2.0 2023-04-28 06:06:52,781 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41353.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:06:55,099 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41355.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:06:57,958 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 3.021e+02 3.672e+02 4.220e+02 6.723e+02, threshold=7.344e+02, percent-clipped=0.0 2023-04-28 06:07:14,429 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9266, 4.5015, 4.4748, 1.9725, 4.7854, 4.7756, 3.5860, 3.9448], device='cuda:7'), covar=tensor([0.0684, 0.0106, 0.0185, 0.1111, 0.0048, 0.0075, 0.0267, 0.0283], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0087, 0.0084, 0.0144, 0.0072, 0.0079, 0.0115, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 06:07:48,728 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3565, 5.2665, 5.1390, 5.0245, 4.6416, 5.0772, 5.1848, 4.7014], device='cuda:7'), covar=tensor([0.0420, 0.0228, 0.0176, 0.0148, 0.0848, 0.0271, 0.0184, 0.0533], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0177, 0.0223, 0.0188, 0.0253, 0.0212, 0.0160, 0.0240], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:07:58,830 INFO [train.py:904] (7/8) Epoch 5, batch 800, loss[loss=0.2278, simple_loss=0.3051, pruned_loss=0.07523, over 16593.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2901, pruned_loss=0.07096, over 3270249.77 frames. ], batch size: 62, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:07:59,176 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41401.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:08:52,042 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.21 vs. limit=5.0 2023-04-28 06:08:55,199 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 06:09:08,628 INFO [train.py:904] (7/8) Epoch 5, batch 850, loss[loss=0.2202, simple_loss=0.2805, pruned_loss=0.07996, over 16826.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2896, pruned_loss=0.07049, over 3281389.30 frames. ], batch size: 116, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:09:16,405 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 3.257e+02 3.799e+02 4.682e+02 9.659e+02, threshold=7.597e+02, percent-clipped=4.0 2023-04-28 06:09:17,967 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41458.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:09:31,729 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6308, 3.6978, 1.7956, 3.8349, 2.6193, 3.8458, 2.0338, 2.9741], device='cuda:7'), covar=tensor([0.0115, 0.0275, 0.1678, 0.0117, 0.0704, 0.0385, 0.1364, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0155, 0.0177, 0.0087, 0.0161, 0.0185, 0.0189, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 06:10:01,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6021, 4.5395, 4.4947, 3.8378, 4.4227, 1.6109, 4.2744, 4.3712], device='cuda:7'), covar=tensor([0.0075, 0.0058, 0.0091, 0.0291, 0.0072, 0.1850, 0.0092, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0096, 0.0079, 0.0127, 0.0129, 0.0094, 0.0144, 0.0109, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:10:16,023 INFO [train.py:904] (7/8) Epoch 5, batch 900, loss[loss=0.191, simple_loss=0.27, pruned_loss=0.05597, over 16971.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2879, pruned_loss=0.06908, over 3294842.07 frames. ], batch size: 41, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:10:44,990 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41521.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:11:27,561 INFO [train.py:904] (7/8) Epoch 5, batch 950, loss[loss=0.1847, simple_loss=0.2682, pruned_loss=0.05062, over 17243.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2876, pruned_loss=0.06923, over 3305891.82 frames. ], batch size: 45, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:11:34,470 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41556.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:11:35,290 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.063e+02 2.970e+02 3.801e+02 5.319e+02 1.620e+03, threshold=7.602e+02, percent-clipped=9.0 2023-04-28 06:11:53,684 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41569.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:12:25,982 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8527, 3.1148, 2.5728, 4.1751, 3.7240, 3.9689, 1.6805, 2.9622], device='cuda:7'), covar=tensor([0.1247, 0.0418, 0.0954, 0.0079, 0.0294, 0.0287, 0.1270, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0136, 0.0162, 0.0080, 0.0166, 0.0165, 0.0154, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 06:12:37,216 INFO [train.py:904] (7/8) Epoch 5, batch 1000, loss[loss=0.2097, simple_loss=0.2829, pruned_loss=0.06824, over 16597.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2862, pruned_loss=0.06946, over 3313218.66 frames. ], batch size: 62, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:12:41,514 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41604.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:13:27,022 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5283, 4.2072, 3.1385, 1.9526, 2.6174, 2.1101, 3.7234, 4.0379], device='cuda:7'), covar=tensor([0.0215, 0.0339, 0.0676, 0.1569, 0.0873, 0.1087, 0.0482, 0.0565], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0132, 0.0155, 0.0142, 0.0135, 0.0126, 0.0141, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 06:13:45,707 INFO [train.py:904] (7/8) Epoch 5, batch 1050, loss[loss=0.1933, simple_loss=0.2762, pruned_loss=0.0552, over 17200.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2861, pruned_loss=0.0693, over 3313872.78 frames. ], batch size: 43, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:13:48,999 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41653.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:13:50,915 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41654.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:13:54,769 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.880e+02 3.783e+02 4.446e+02 1.011e+03, threshold=7.566e+02, percent-clipped=5.0 2023-04-28 06:14:20,679 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-28 06:14:56,090 INFO [train.py:904] (7/8) Epoch 5, batch 1100, loss[loss=0.1994, simple_loss=0.2699, pruned_loss=0.06452, over 15560.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2856, pruned_loss=0.06882, over 3308847.84 frames. ], batch size: 190, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:14:56,443 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:14:56,568 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:15:15,762 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41715.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:15:50,873 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8631, 4.4311, 4.6960, 1.7875, 4.9450, 4.9109, 3.7712, 3.7631], device='cuda:7'), covar=tensor([0.0649, 0.0083, 0.0131, 0.1134, 0.0034, 0.0038, 0.0229, 0.0320], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0086, 0.0084, 0.0141, 0.0070, 0.0079, 0.0113, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 06:16:02,694 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41749.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:16:05,300 INFO [train.py:904] (7/8) Epoch 5, batch 1150, loss[loss=0.2373, simple_loss=0.2931, pruned_loss=0.09071, over 16832.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2848, pruned_loss=0.06871, over 3313100.06 frames. ], batch size: 116, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:16:12,834 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.850e+02 3.603e+02 4.605e+02 7.939e+02, threshold=7.207e+02, percent-clipped=2.0 2023-04-28 06:16:15,808 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41758.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:17:14,377 INFO [train.py:904] (7/8) Epoch 5, batch 1200, loss[loss=0.2487, simple_loss=0.3021, pruned_loss=0.09759, over 15472.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2842, pruned_loss=0.06868, over 3309432.54 frames. ], batch size: 190, lr: 1.40e-02, grad_scale: 8.0 2023-04-28 06:17:21,147 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=41806.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:17:46,970 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9383, 4.3650, 1.7249, 4.5309, 2.6096, 4.5232, 1.9481, 3.1523], device='cuda:7'), covar=tensor([0.0131, 0.0184, 0.1695, 0.0050, 0.0895, 0.0285, 0.1647, 0.0555], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0154, 0.0175, 0.0087, 0.0162, 0.0186, 0.0187, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 06:18:23,590 INFO [train.py:904] (7/8) Epoch 5, batch 1250, loss[loss=0.1993, simple_loss=0.282, pruned_loss=0.05826, over 17190.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2844, pruned_loss=0.06867, over 3317586.83 frames. ], batch size: 46, lr: 1.40e-02, grad_scale: 8.0 2023-04-28 06:18:31,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.030e+02 3.380e+02 4.131e+02 4.913e+02 1.055e+03, threshold=8.263e+02, percent-clipped=6.0 2023-04-28 06:19:30,800 INFO [train.py:904] (7/8) Epoch 5, batch 1300, loss[loss=0.2123, simple_loss=0.2983, pruned_loss=0.06313, over 17045.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2846, pruned_loss=0.06891, over 3320779.43 frames. ], batch size: 55, lr: 1.40e-02, grad_scale: 8.0 2023-04-28 06:19:39,593 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:20:02,512 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.58 vs. limit=5.0 2023-04-28 06:20:38,675 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41948.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 06:20:42,287 INFO [train.py:904] (7/8) Epoch 5, batch 1350, loss[loss=0.2744, simple_loss=0.3241, pruned_loss=0.1123, over 16820.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2852, pruned_loss=0.06867, over 3329214.84 frames. ], batch size: 83, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:20:51,205 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.161e+02 3.407e+02 4.000e+02 4.882e+02 1.065e+03, threshold=8.000e+02, percent-clipped=1.0 2023-04-28 06:21:05,602 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41967.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:21:55,634 INFO [train.py:904] (7/8) Epoch 5, batch 1400, loss[loss=0.1891, simple_loss=0.2664, pruned_loss=0.05592, over 17194.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2853, pruned_loss=0.06847, over 3321956.21 frames. ], batch size: 45, lr: 1.40e-02, grad_scale: 4.0 2023-04-28 06:22:08,318 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42009.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 06:22:09,944 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42010.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 06:22:54,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0350, 1.8080, 2.4229, 2.8868, 2.9029, 3.5121, 1.7787, 3.2213], device='cuda:7'), covar=tensor([0.0081, 0.0216, 0.0141, 0.0126, 0.0101, 0.0064, 0.0242, 0.0071], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0139, 0.0126, 0.0125, 0.0124, 0.0089, 0.0138, 0.0078], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 06:23:05,360 INFO [train.py:904] (7/8) Epoch 5, batch 1450, loss[loss=0.2142, simple_loss=0.3028, pruned_loss=0.06282, over 16684.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2843, pruned_loss=0.06842, over 3327525.78 frames. ], batch size: 57, lr: 1.39e-02, grad_scale: 4.0 2023-04-28 06:23:15,558 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.035e+02 3.059e+02 3.831e+02 4.678e+02 9.639e+02, threshold=7.661e+02, percent-clipped=1.0 2023-04-28 06:24:14,180 INFO [train.py:904] (7/8) Epoch 5, batch 1500, loss[loss=0.2303, simple_loss=0.3022, pruned_loss=0.07923, over 15530.00 frames. ], tot_loss[loss=0.2105, simple_loss=0.2844, pruned_loss=0.06831, over 3326149.11 frames. ], batch size: 190, lr: 1.39e-02, grad_scale: 4.0 2023-04-28 06:24:14,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2521, 5.1429, 5.0963, 4.9100, 4.5579, 5.1112, 5.0821, 4.7649], device='cuda:7'), covar=tensor([0.0410, 0.0230, 0.0173, 0.0149, 0.0956, 0.0238, 0.0205, 0.0448], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0187, 0.0228, 0.0197, 0.0265, 0.0221, 0.0165, 0.0250], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:24:19,987 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3421, 4.1786, 4.7501, 4.7542, 4.7451, 4.2869, 4.3596, 4.1582], device='cuda:7'), covar=tensor([0.0243, 0.0472, 0.0356, 0.0363, 0.0387, 0.0325, 0.0773, 0.0501], device='cuda:7'), in_proj_covar=tensor([0.0241, 0.0233, 0.0239, 0.0237, 0.0281, 0.0252, 0.0358, 0.0217], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 06:25:21,155 INFO [train.py:904] (7/8) Epoch 5, batch 1550, loss[loss=0.1908, simple_loss=0.279, pruned_loss=0.05131, over 17273.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2847, pruned_loss=0.06879, over 3328730.97 frames. ], batch size: 52, lr: 1.39e-02, grad_scale: 4.0 2023-04-28 06:25:32,955 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.189e+02 3.587e+02 4.011e+02 4.553e+02 8.694e+02, threshold=8.021e+02, percent-clipped=2.0 2023-04-28 06:26:10,297 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5682, 4.9355, 4.6721, 4.7161, 4.3247, 4.2663, 4.3579, 4.9642], device='cuda:7'), covar=tensor([0.0695, 0.0679, 0.0902, 0.0437, 0.0649, 0.0930, 0.0728, 0.0682], device='cuda:7'), in_proj_covar=tensor([0.0363, 0.0502, 0.0415, 0.0315, 0.0312, 0.0317, 0.0392, 0.0355], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:26:32,638 INFO [train.py:904] (7/8) Epoch 5, batch 1600, loss[loss=0.2498, simple_loss=0.3073, pruned_loss=0.09613, over 16895.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2871, pruned_loss=0.06972, over 3323523.77 frames. ], batch size: 116, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:26:35,316 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7080, 2.7623, 2.1195, 2.4044, 3.0302, 3.0007, 3.8718, 3.4182], device='cuda:7'), covar=tensor([0.0028, 0.0154, 0.0216, 0.0196, 0.0113, 0.0144, 0.0065, 0.0086], device='cuda:7'), in_proj_covar=tensor([0.0084, 0.0152, 0.0151, 0.0149, 0.0149, 0.0153, 0.0126, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:26:42,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1918, 4.4433, 1.8521, 4.7619, 2.9415, 4.6688, 1.8585, 3.3228], device='cuda:7'), covar=tensor([0.0123, 0.0214, 0.1791, 0.0039, 0.0687, 0.0307, 0.1671, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0155, 0.0174, 0.0087, 0.0160, 0.0186, 0.0186, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 06:27:00,761 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6441, 4.2459, 4.5647, 2.2044, 4.9349, 4.8411, 3.4153, 3.8748], device='cuda:7'), covar=tensor([0.0742, 0.0132, 0.0168, 0.1044, 0.0055, 0.0049, 0.0288, 0.0303], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0086, 0.0083, 0.0141, 0.0070, 0.0078, 0.0112, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 06:27:39,758 INFO [train.py:904] (7/8) Epoch 5, batch 1650, loss[loss=0.2451, simple_loss=0.3031, pruned_loss=0.09352, over 16856.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2885, pruned_loss=0.07007, over 3329074.33 frames. ], batch size: 109, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:27:49,825 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.983e+02 3.262e+02 3.932e+02 4.837e+02 9.371e+02, threshold=7.863e+02, percent-clipped=1.0 2023-04-28 06:27:56,382 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42262.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:27:57,643 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1546, 3.7898, 3.7016, 4.3303, 4.3540, 4.0769, 4.2618, 4.3547], device='cuda:7'), covar=tensor([0.0768, 0.0921, 0.2160, 0.0687, 0.0682, 0.0835, 0.1044, 0.0731], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0498, 0.0638, 0.0506, 0.0380, 0.0369, 0.0394, 0.0419], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:28:49,303 INFO [train.py:904] (7/8) Epoch 5, batch 1700, loss[loss=0.2099, simple_loss=0.2948, pruned_loss=0.06253, over 15986.00 frames. ], tot_loss[loss=0.217, simple_loss=0.2914, pruned_loss=0.07125, over 3322652.59 frames. ], batch size: 35, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:28:54,263 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42304.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 06:29:01,335 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42310.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:29:56,510 INFO [train.py:904] (7/8) Epoch 5, batch 1750, loss[loss=0.2347, simple_loss=0.3199, pruned_loss=0.07475, over 17062.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2929, pruned_loss=0.0717, over 3320587.70 frames. ], batch size: 55, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:30:05,728 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.019e+02 3.418e+02 4.322e+02 5.636e+02 1.741e+03, threshold=8.644e+02, percent-clipped=7.0 2023-04-28 06:30:06,026 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=42358.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:30:21,997 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6383, 2.9706, 2.4734, 4.2379, 3.7341, 3.8951, 1.4908, 2.9226], device='cuda:7'), covar=tensor([0.1316, 0.0514, 0.1003, 0.0067, 0.0211, 0.0294, 0.1302, 0.0684], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0141, 0.0168, 0.0086, 0.0176, 0.0172, 0.0159, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 06:30:36,224 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42379.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:31:06,098 INFO [train.py:904] (7/8) Epoch 5, batch 1800, loss[loss=0.2553, simple_loss=0.3286, pruned_loss=0.09094, over 16890.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.294, pruned_loss=0.07217, over 3323763.07 frames. ], batch size: 96, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:32:01,401 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42440.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:32:08,265 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0525, 4.7269, 4.9449, 5.2987, 5.4401, 4.6477, 5.3720, 5.3967], device='cuda:7'), covar=tensor([0.0813, 0.0910, 0.1358, 0.0467, 0.0413, 0.0542, 0.0374, 0.0330], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0492, 0.0638, 0.0504, 0.0378, 0.0368, 0.0394, 0.0419], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:32:12,651 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6645, 2.6336, 2.2307, 3.9525, 3.4405, 3.7088, 1.5017, 2.6238], device='cuda:7'), covar=tensor([0.1320, 0.0625, 0.1237, 0.0087, 0.0286, 0.0369, 0.1319, 0.0846], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0139, 0.0166, 0.0085, 0.0175, 0.0170, 0.0157, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 06:32:16,842 INFO [train.py:904] (7/8) Epoch 5, batch 1850, loss[loss=0.1976, simple_loss=0.2928, pruned_loss=0.05117, over 17128.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.2947, pruned_loss=0.07208, over 3329817.25 frames. ], batch size: 49, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:32:26,220 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.852e+02 3.144e+02 3.803e+02 4.355e+02 7.438e+02, threshold=7.606e+02, percent-clipped=0.0 2023-04-28 06:33:24,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6518, 1.3500, 2.1069, 2.4882, 2.5476, 2.4400, 1.6063, 2.5409], device='cuda:7'), covar=tensor([0.0067, 0.0233, 0.0135, 0.0103, 0.0079, 0.0116, 0.0218, 0.0044], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0140, 0.0127, 0.0127, 0.0124, 0.0089, 0.0138, 0.0079], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 06:33:25,574 INFO [train.py:904] (7/8) Epoch 5, batch 1900, loss[loss=0.2303, simple_loss=0.2997, pruned_loss=0.08049, over 16782.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.2938, pruned_loss=0.07101, over 3315913.05 frames. ], batch size: 134, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:33:32,619 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42506.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:33:41,766 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0040, 4.3408, 2.1518, 4.6581, 2.8149, 4.5788, 2.2867, 3.2435], device='cuda:7'), covar=tensor([0.0112, 0.0218, 0.1442, 0.0053, 0.0669, 0.0283, 0.1214, 0.0483], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0154, 0.0175, 0.0086, 0.0158, 0.0186, 0.0184, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 06:34:11,574 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6755, 3.7452, 2.7634, 2.2207, 2.6197, 2.0255, 3.5692, 3.6398], device='cuda:7'), covar=tensor([0.1666, 0.0397, 0.1062, 0.1477, 0.1957, 0.1406, 0.0387, 0.0593], device='cuda:7'), in_proj_covar=tensor([0.0274, 0.0252, 0.0269, 0.0242, 0.0300, 0.0202, 0.0239, 0.0261], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:34:20,966 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2882, 1.9863, 1.4550, 1.7658, 2.3965, 2.2616, 2.5334, 2.3935], device='cuda:7'), covar=tensor([0.0058, 0.0190, 0.0237, 0.0219, 0.0101, 0.0156, 0.0110, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0084, 0.0152, 0.0151, 0.0149, 0.0148, 0.0155, 0.0127, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:34:30,484 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7909, 4.1066, 4.3695, 1.7230, 4.7172, 4.6074, 3.1787, 3.6116], device='cuda:7'), covar=tensor([0.0627, 0.0093, 0.0122, 0.1092, 0.0035, 0.0047, 0.0328, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0086, 0.0085, 0.0142, 0.0070, 0.0079, 0.0114, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 06:34:35,497 INFO [train.py:904] (7/8) Epoch 5, batch 1950, loss[loss=0.2421, simple_loss=0.3191, pruned_loss=0.08258, over 12443.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2941, pruned_loss=0.07105, over 3309636.75 frames. ], batch size: 247, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:34:47,180 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 3.199e+02 3.788e+02 4.472e+02 9.515e+02, threshold=7.576e+02, percent-clipped=2.0 2023-04-28 06:34:52,098 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42562.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:34:59,532 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:35:48,043 INFO [train.py:904] (7/8) Epoch 5, batch 2000, loss[loss=0.2399, simple_loss=0.3174, pruned_loss=0.08124, over 17041.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2938, pruned_loss=0.07043, over 3321121.97 frames. ], batch size: 53, lr: 1.39e-02, grad_scale: 8.0 2023-04-28 06:35:51,753 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42604.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 06:36:00,885 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=42610.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:36:47,631 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0441, 4.3584, 3.5750, 2.5740, 3.3558, 2.6442, 4.7115, 4.4582], device='cuda:7'), covar=tensor([0.1927, 0.0652, 0.1050, 0.1556, 0.2219, 0.1319, 0.0288, 0.0551], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0252, 0.0270, 0.0242, 0.0303, 0.0202, 0.0239, 0.0264], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:36:55,052 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42649.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:36:57,541 INFO [train.py:904] (7/8) Epoch 5, batch 2050, loss[loss=0.207, simple_loss=0.2947, pruned_loss=0.0596, over 17022.00 frames. ], tot_loss[loss=0.2172, simple_loss=0.2932, pruned_loss=0.07059, over 3321121.57 frames. ], batch size: 50, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:36:59,090 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=42652.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 06:37:03,011 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-28 06:37:06,890 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 3.076e+02 3.576e+02 4.219e+02 7.856e+02, threshold=7.152e+02, percent-clipped=1.0 2023-04-28 06:37:29,662 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7826, 4.3051, 4.3763, 1.9335, 4.7211, 4.6269, 3.3864, 3.8278], device='cuda:7'), covar=tensor([0.0686, 0.0103, 0.0143, 0.1148, 0.0047, 0.0063, 0.0277, 0.0268], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0086, 0.0083, 0.0140, 0.0069, 0.0079, 0.0112, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 06:37:45,319 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 06:37:50,969 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1358, 5.6447, 5.7870, 5.6853, 5.7969, 6.2070, 5.9281, 5.7019], device='cuda:7'), covar=tensor([0.0668, 0.1582, 0.1226, 0.1660, 0.2388, 0.0800, 0.0966, 0.1785], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0409, 0.0384, 0.0346, 0.0469, 0.0415, 0.0321, 0.0462], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:38:05,326 INFO [train.py:904] (7/8) Epoch 5, batch 2100, loss[loss=0.2503, simple_loss=0.3181, pruned_loss=0.09122, over 16841.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.2946, pruned_loss=0.0713, over 3312100.89 frames. ], batch size: 96, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:38:18,790 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42710.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:38:52,959 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42735.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:39:15,225 INFO [train.py:904] (7/8) Epoch 5, batch 2150, loss[loss=0.2357, simple_loss=0.3076, pruned_loss=0.0819, over 16543.00 frames. ], tot_loss[loss=0.2204, simple_loss=0.2962, pruned_loss=0.07225, over 3316046.94 frames. ], batch size: 75, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:39:24,100 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.183e+02 3.251e+02 3.882e+02 4.566e+02 8.930e+02, threshold=7.764e+02, percent-clipped=4.0 2023-04-28 06:39:55,148 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 06:40:23,568 INFO [train.py:904] (7/8) Epoch 5, batch 2200, loss[loss=0.2337, simple_loss=0.3036, pruned_loss=0.08189, over 16446.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.2966, pruned_loss=0.0724, over 3327020.49 frames. ], batch size: 146, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:41:34,714 INFO [train.py:904] (7/8) Epoch 5, batch 2250, loss[loss=0.258, simple_loss=0.318, pruned_loss=0.09902, over 16734.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.2969, pruned_loss=0.07244, over 3323300.90 frames. ], batch size: 89, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:41:43,711 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.043e+02 3.207e+02 3.862e+02 4.939e+02 1.226e+03, threshold=7.724e+02, percent-clipped=4.0 2023-04-28 06:41:46,075 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42859.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:41:49,535 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42862.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:42:14,408 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42879.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:42:44,589 INFO [train.py:904] (7/8) Epoch 5, batch 2300, loss[loss=0.2237, simple_loss=0.3112, pruned_loss=0.06805, over 17137.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.2976, pruned_loss=0.0725, over 3322717.44 frames. ], batch size: 48, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:43:01,606 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 06:43:11,622 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42920.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:43:31,836 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9082, 3.8062, 3.8056, 4.1433, 4.2364, 3.7744, 3.9656, 4.1798], device='cuda:7'), covar=tensor([0.0863, 0.0737, 0.1596, 0.0626, 0.0531, 0.1579, 0.1511, 0.0577], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0493, 0.0638, 0.0517, 0.0379, 0.0380, 0.0396, 0.0419], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:43:38,284 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:43:53,246 INFO [train.py:904] (7/8) Epoch 5, batch 2350, loss[loss=0.212, simple_loss=0.3046, pruned_loss=0.05973, over 17140.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.2975, pruned_loss=0.07263, over 3321679.84 frames. ], batch size: 48, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:44:03,301 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.280e+02 3.235e+02 3.889e+02 4.889e+02 8.777e+02, threshold=7.778e+02, percent-clipped=2.0 2023-04-28 06:44:16,780 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5262, 4.5629, 4.4996, 4.0066, 4.4829, 1.8217, 4.3057, 4.4145], device='cuda:7'), covar=tensor([0.0068, 0.0048, 0.0075, 0.0248, 0.0056, 0.1411, 0.0075, 0.0106], device='cuda:7'), in_proj_covar=tensor([0.0096, 0.0083, 0.0127, 0.0133, 0.0096, 0.0138, 0.0111, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:44:26,853 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0410, 3.8196, 3.7350, 4.2530, 4.3328, 3.9836, 4.0241, 4.3023], device='cuda:7'), covar=tensor([0.0945, 0.0884, 0.2252, 0.0802, 0.0706, 0.1139, 0.1622, 0.0818], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0488, 0.0632, 0.0510, 0.0372, 0.0375, 0.0392, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:45:02,234 INFO [train.py:904] (7/8) Epoch 5, batch 2400, loss[loss=0.2396, simple_loss=0.3035, pruned_loss=0.08784, over 16735.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.2984, pruned_loss=0.07346, over 3319812.05 frames. ], batch size: 134, lr: 1.38e-02, grad_scale: 8.0 2023-04-28 06:45:08,771 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43005.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:45:52,154 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43035.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:46:13,645 INFO [train.py:904] (7/8) Epoch 5, batch 2450, loss[loss=0.2055, simple_loss=0.2904, pruned_loss=0.06037, over 17238.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.2982, pruned_loss=0.07237, over 3322597.05 frames. ], batch size: 45, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:46:26,011 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 3.112e+02 3.702e+02 4.495e+02 8.852e+02, threshold=7.404e+02, percent-clipped=2.0 2023-04-28 06:46:58,057 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43083.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:47:23,976 INFO [train.py:904] (7/8) Epoch 5, batch 2500, loss[loss=0.2876, simple_loss=0.3539, pruned_loss=0.1107, over 12278.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.2976, pruned_loss=0.07191, over 3323351.85 frames. ], batch size: 246, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:48:19,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6793, 5.1041, 4.7468, 4.8712, 4.5251, 4.4249, 4.6415, 5.0296], device='cuda:7'), covar=tensor([0.0619, 0.0554, 0.0778, 0.0412, 0.0542, 0.0797, 0.0549, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0497, 0.0420, 0.0319, 0.0311, 0.0319, 0.0396, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:48:33,488 INFO [train.py:904] (7/8) Epoch 5, batch 2550, loss[loss=0.1929, simple_loss=0.269, pruned_loss=0.05844, over 16758.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.2978, pruned_loss=0.07206, over 3327525.25 frames. ], batch size: 39, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:48:45,562 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.843e+02 3.229e+02 3.908e+02 4.674e+02 8.443e+02, threshold=7.816e+02, percent-clipped=2.0 2023-04-28 06:48:48,828 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43162.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:49:43,648 INFO [train.py:904] (7/8) Epoch 5, batch 2600, loss[loss=0.2714, simple_loss=0.3419, pruned_loss=0.1004, over 12550.00 frames. ], tot_loss[loss=0.2206, simple_loss=0.2972, pruned_loss=0.072, over 3322659.00 frames. ], batch size: 246, lr: 1.38e-02, grad_scale: 4.0 2023-04-28 06:49:55,551 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43210.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:50:02,772 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43215.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:50:16,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3293, 1.9418, 1.5593, 1.7117, 2.3466, 2.2203, 2.3909, 2.4992], device='cuda:7'), covar=tensor([0.0055, 0.0178, 0.0206, 0.0214, 0.0094, 0.0149, 0.0104, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0154, 0.0154, 0.0151, 0.0151, 0.0158, 0.0132, 0.0139], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:50:31,272 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43235.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:50:53,873 INFO [train.py:904] (7/8) Epoch 5, batch 2650, loss[loss=0.1913, simple_loss=0.2726, pruned_loss=0.05505, over 16840.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.2968, pruned_loss=0.07099, over 3323805.74 frames. ], batch size: 42, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 06:50:57,689 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5099, 4.4927, 5.0567, 5.0070, 5.0503, 4.4328, 4.5669, 4.2521], device='cuda:7'), covar=tensor([0.0266, 0.0341, 0.0302, 0.0392, 0.0417, 0.0380, 0.0798, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0250, 0.0242, 0.0243, 0.0248, 0.0298, 0.0260, 0.0371, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 06:51:05,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 3.400e+02 3.860e+02 4.730e+02 9.799e+02, threshold=7.720e+02, percent-clipped=2.0 2023-04-28 06:51:28,320 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8310, 4.2997, 4.5868, 3.0888, 4.0733, 4.5538, 4.3509, 2.4897], device='cuda:7'), covar=tensor([0.0248, 0.0025, 0.0017, 0.0191, 0.0029, 0.0023, 0.0019, 0.0245], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0061, 0.0061, 0.0114, 0.0062, 0.0069, 0.0064, 0.0106], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:52:02,504 INFO [train.py:904] (7/8) Epoch 5, batch 2700, loss[loss=0.2391, simple_loss=0.3103, pruned_loss=0.08395, over 16538.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.2974, pruned_loss=0.07055, over 3323772.12 frames. ], batch size: 146, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 06:52:09,052 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43305.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:53:11,821 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3437, 5.2431, 5.1149, 4.9542, 4.6988, 5.1520, 5.1212, 4.7394], device='cuda:7'), covar=tensor([0.0336, 0.0218, 0.0165, 0.0149, 0.0815, 0.0214, 0.0210, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0196, 0.0231, 0.0201, 0.0267, 0.0225, 0.0170, 0.0257], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:53:12,585 INFO [train.py:904] (7/8) Epoch 5, batch 2750, loss[loss=0.1841, simple_loss=0.2753, pruned_loss=0.04648, over 17235.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2968, pruned_loss=0.06974, over 3327921.46 frames. ], batch size: 45, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 06:53:15,891 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43353.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:53:25,429 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.783e+02 3.325e+02 4.453e+02 8.515e+02, threshold=6.649e+02, percent-clipped=2.0 2023-04-28 06:54:22,973 INFO [train.py:904] (7/8) Epoch 5, batch 2800, loss[loss=0.1944, simple_loss=0.2836, pruned_loss=0.05258, over 17187.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.296, pruned_loss=0.06946, over 3328727.47 frames. ], batch size: 46, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:55:33,366 INFO [train.py:904] (7/8) Epoch 5, batch 2850, loss[loss=0.2226, simple_loss=0.2912, pruned_loss=0.07703, over 16457.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2952, pruned_loss=0.06974, over 3323662.05 frames. ], batch size: 75, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:55:39,988 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3625, 4.1377, 4.2809, 4.5968, 4.6805, 4.2205, 4.4850, 4.6635], device='cuda:7'), covar=tensor([0.0729, 0.0721, 0.1248, 0.0451, 0.0381, 0.0821, 0.0994, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0499, 0.0649, 0.0515, 0.0382, 0.0385, 0.0399, 0.0424], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 06:55:45,525 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.398e+02 3.090e+02 3.971e+02 4.824e+02 1.597e+03, threshold=7.942e+02, percent-clipped=16.0 2023-04-28 06:56:11,440 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 06:56:41,605 INFO [train.py:904] (7/8) Epoch 5, batch 2900, loss[loss=0.2537, simple_loss=0.332, pruned_loss=0.08774, over 16592.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.2942, pruned_loss=0.07033, over 3319493.63 frames. ], batch size: 62, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:56:49,192 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9516, 3.9770, 4.4730, 4.4052, 4.4560, 4.0489, 4.1346, 4.0648], device='cuda:7'), covar=tensor([0.0331, 0.0558, 0.0345, 0.0447, 0.0374, 0.0350, 0.0783, 0.0501], device='cuda:7'), in_proj_covar=tensor([0.0249, 0.0241, 0.0244, 0.0246, 0.0297, 0.0258, 0.0371, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 06:57:00,401 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43514.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:57:01,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43515.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:57:28,559 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43535.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:57:32,478 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.08 vs. limit=5.0 2023-04-28 06:57:49,470 INFO [train.py:904] (7/8) Epoch 5, batch 2950, loss[loss=0.2246, simple_loss=0.2834, pruned_loss=0.08287, over 16703.00 frames. ], tot_loss[loss=0.219, simple_loss=0.2946, pruned_loss=0.07173, over 3308938.98 frames. ], batch size: 89, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:58:02,033 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 3.523e+02 4.392e+02 5.474e+02 1.022e+03, threshold=8.784e+02, percent-clipped=2.0 2023-04-28 06:58:06,678 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43563.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:58:24,127 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43575.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:58:30,804 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9983, 3.1117, 3.5290, 2.3798, 3.2528, 3.5062, 3.2226, 2.0323], device='cuda:7'), covar=tensor([0.0293, 0.0081, 0.0028, 0.0214, 0.0040, 0.0044, 0.0050, 0.0252], device='cuda:7'), in_proj_covar=tensor([0.0114, 0.0060, 0.0059, 0.0112, 0.0061, 0.0068, 0.0065, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 06:58:35,030 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=43583.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 06:58:47,295 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2378, 3.8232, 3.1487, 1.9183, 2.8056, 2.2594, 3.6262, 3.5954], device='cuda:7'), covar=tensor([0.0239, 0.0500, 0.0648, 0.1554, 0.0700, 0.1000, 0.0518, 0.0707], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0132, 0.0154, 0.0140, 0.0132, 0.0125, 0.0140, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 06:58:59,871 INFO [train.py:904] (7/8) Epoch 5, batch 3000, loss[loss=0.2171, simple_loss=0.3088, pruned_loss=0.06271, over 17109.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.2957, pruned_loss=0.07291, over 3304576.79 frames. ], batch size: 49, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 06:58:59,871 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 06:59:08,831 INFO [train.py:938] (7/8) Epoch 5, validation: loss=0.1564, simple_loss=0.2632, pruned_loss=0.02477, over 944034.00 frames. 2023-04-28 06:59:08,832 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 06:59:14,023 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 06:59:16,055 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3475, 3.9725, 3.2066, 1.8408, 2.7937, 2.3000, 3.7240, 3.7229], device='cuda:7'), covar=tensor([0.0221, 0.0454, 0.0632, 0.1656, 0.0716, 0.1024, 0.0501, 0.0645], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0133, 0.0156, 0.0141, 0.0133, 0.0126, 0.0141, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 07:00:02,403 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2981, 4.5562, 2.2858, 4.8973, 2.9522, 4.7883, 2.2177, 3.4381], device='cuda:7'), covar=tensor([0.0086, 0.0146, 0.1278, 0.0032, 0.0634, 0.0248, 0.1325, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0158, 0.0175, 0.0089, 0.0158, 0.0190, 0.0185, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 07:00:18,524 INFO [train.py:904] (7/8) Epoch 5, batch 3050, loss[loss=0.2152, simple_loss=0.302, pruned_loss=0.06416, over 16742.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.2954, pruned_loss=0.07239, over 3312361.45 frames. ], batch size: 62, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 07:00:31,427 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.472e+02 3.368e+02 3.840e+02 5.233e+02 1.219e+03, threshold=7.679e+02, percent-clipped=3.0 2023-04-28 07:01:25,954 INFO [train.py:904] (7/8) Epoch 5, batch 3100, loss[loss=0.2346, simple_loss=0.2955, pruned_loss=0.08689, over 16713.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.2946, pruned_loss=0.07202, over 3316872.32 frames. ], batch size: 124, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 07:02:33,560 INFO [train.py:904] (7/8) Epoch 5, batch 3150, loss[loss=0.1804, simple_loss=0.2589, pruned_loss=0.05096, over 15929.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.2939, pruned_loss=0.07187, over 3313635.97 frames. ], batch size: 35, lr: 1.37e-02, grad_scale: 4.0 2023-04-28 07:02:46,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.292e+02 2.988e+02 3.776e+02 4.574e+02 1.068e+03, threshold=7.553e+02, percent-clipped=4.0 2023-04-28 07:02:53,138 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2030, 5.5970, 5.6661, 5.6373, 5.5169, 6.1054, 5.7324, 5.4336], device='cuda:7'), covar=tensor([0.0703, 0.1438, 0.1261, 0.1352, 0.2325, 0.0832, 0.0906, 0.1804], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0409, 0.0395, 0.0344, 0.0464, 0.0418, 0.0329, 0.0466], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:02:58,992 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 07:03:23,192 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43787.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:03:37,239 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-04-28 07:03:42,798 INFO [train.py:904] (7/8) Epoch 5, batch 3200, loss[loss=0.2411, simple_loss=0.3274, pruned_loss=0.07737, over 16735.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2929, pruned_loss=0.07165, over 3314575.11 frames. ], batch size: 57, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 07:04:48,305 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:04:52,120 INFO [train.py:904] (7/8) Epoch 5, batch 3250, loss[loss=0.245, simple_loss=0.3092, pruned_loss=0.09042, over 16835.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2918, pruned_loss=0.07052, over 3320100.96 frames. ], batch size: 116, lr: 1.37e-02, grad_scale: 8.0 2023-04-28 07:05:06,672 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.952e+02 3.819e+02 4.745e+02 7.662e+02, threshold=7.638e+02, percent-clipped=2.0 2023-04-28 07:05:19,464 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43870.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:05:45,233 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2309, 2.0855, 1.4518, 1.8680, 2.5483, 2.3420, 2.5751, 2.6654], device='cuda:7'), covar=tensor([0.0096, 0.0135, 0.0200, 0.0177, 0.0072, 0.0127, 0.0093, 0.0071], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0151, 0.0151, 0.0148, 0.0149, 0.0155, 0.0132, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:05:52,868 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0986, 1.6493, 2.4093, 2.9432, 2.9877, 3.2495, 2.2151, 3.2652], device='cuda:7'), covar=tensor([0.0067, 0.0218, 0.0112, 0.0108, 0.0079, 0.0071, 0.0155, 0.0060], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0142, 0.0127, 0.0131, 0.0126, 0.0093, 0.0137, 0.0082], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 07:05:58,776 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-04-28 07:06:01,643 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.19 vs. limit=5.0 2023-04-28 07:06:03,320 INFO [train.py:904] (7/8) Epoch 5, batch 3300, loss[loss=0.2155, simple_loss=0.2872, pruned_loss=0.0719, over 16916.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2925, pruned_loss=0.07053, over 3320684.56 frames. ], batch size: 109, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:06:03,811 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4636, 3.5605, 3.1336, 3.2034, 3.0626, 3.3734, 3.2403, 3.1044], device='cuda:7'), covar=tensor([0.0429, 0.0255, 0.0201, 0.0192, 0.0557, 0.0229, 0.0818, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0199, 0.0231, 0.0203, 0.0270, 0.0230, 0.0170, 0.0257], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:06:14,746 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:06:15,813 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1095, 5.4857, 5.1390, 5.3009, 4.8450, 4.7243, 4.8925, 5.5638], device='cuda:7'), covar=tensor([0.0614, 0.0625, 0.0855, 0.0435, 0.0577, 0.0655, 0.0618, 0.0631], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0512, 0.0434, 0.0331, 0.0321, 0.0330, 0.0409, 0.0360], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:07:12,758 INFO [train.py:904] (7/8) Epoch 5, batch 3350, loss[loss=0.2726, simple_loss=0.3352, pruned_loss=0.105, over 15441.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2928, pruned_loss=0.06995, over 3317406.18 frames. ], batch size: 190, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:07:24,283 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43959.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:07:26,940 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.942e+02 2.864e+02 3.521e+02 4.539e+02 9.461e+02, threshold=7.042e+02, percent-clipped=3.0 2023-04-28 07:07:39,209 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43970.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:08:22,254 INFO [train.py:904] (7/8) Epoch 5, batch 3400, loss[loss=0.1914, simple_loss=0.2643, pruned_loss=0.0593, over 15929.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.2933, pruned_loss=0.0699, over 3317422.39 frames. ], batch size: 35, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:08:48,427 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44020.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:09:31,566 INFO [train.py:904] (7/8) Epoch 5, batch 3450, loss[loss=0.2426, simple_loss=0.3041, pruned_loss=0.0905, over 16792.00 frames. ], tot_loss[loss=0.2156, simple_loss=0.2922, pruned_loss=0.06948, over 3314214.57 frames. ], batch size: 102, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:09:44,810 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 3.034e+02 3.683e+02 4.468e+02 1.074e+03, threshold=7.367e+02, percent-clipped=2.0 2023-04-28 07:10:01,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7059, 3.6330, 2.7035, 2.3292, 2.6664, 2.0608, 3.4239, 3.6992], device='cuda:7'), covar=tensor([0.1907, 0.0560, 0.1168, 0.1526, 0.2301, 0.1675, 0.0464, 0.0803], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0253, 0.0268, 0.0243, 0.0306, 0.0204, 0.0238, 0.0268], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:10:02,253 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4369, 4.3325, 4.4580, 4.4121, 4.2779, 4.8733, 4.6361, 4.2882], device='cuda:7'), covar=tensor([0.1269, 0.1454, 0.1327, 0.1716, 0.2774, 0.1092, 0.1069, 0.2132], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0409, 0.0393, 0.0349, 0.0467, 0.0424, 0.0328, 0.0472], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:10:18,956 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7970, 3.3635, 2.7525, 4.8425, 4.4865, 4.3011, 1.6396, 3.3695], device='cuda:7'), covar=tensor([0.1357, 0.0564, 0.1039, 0.0092, 0.0267, 0.0308, 0.1450, 0.0613], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0140, 0.0163, 0.0086, 0.0182, 0.0170, 0.0157, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 07:10:34,808 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6009, 6.0111, 5.7541, 5.8545, 5.3091, 4.9929, 5.4682, 6.1329], device='cuda:7'), covar=tensor([0.0675, 0.0767, 0.1003, 0.0418, 0.0632, 0.0690, 0.0664, 0.0609], device='cuda:7'), in_proj_covar=tensor([0.0377, 0.0513, 0.0431, 0.0329, 0.0320, 0.0331, 0.0407, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:10:39,140 INFO [train.py:904] (7/8) Epoch 5, batch 3500, loss[loss=0.24, simple_loss=0.2991, pruned_loss=0.09045, over 11607.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2907, pruned_loss=0.06869, over 3315061.18 frames. ], batch size: 248, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:11:38,574 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44143.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:11:49,624 INFO [train.py:904] (7/8) Epoch 5, batch 3550, loss[loss=0.1969, simple_loss=0.2723, pruned_loss=0.06079, over 16729.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2889, pruned_loss=0.06769, over 3322044.74 frames. ], batch size: 76, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:12:03,022 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.484e+02 3.064e+02 3.621e+02 4.319e+02 6.916e+02, threshold=7.242e+02, percent-clipped=0.0 2023-04-28 07:12:17,204 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44170.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:12:18,425 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2519, 4.2510, 4.7157, 4.7227, 4.6980, 4.3404, 4.3606, 4.1899], device='cuda:7'), covar=tensor([0.0238, 0.0322, 0.0276, 0.0343, 0.0360, 0.0265, 0.0667, 0.0422], device='cuda:7'), in_proj_covar=tensor([0.0253, 0.0246, 0.0251, 0.0253, 0.0300, 0.0264, 0.0378, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 07:12:59,203 INFO [train.py:904] (7/8) Epoch 5, batch 3600, loss[loss=0.1801, simple_loss=0.2502, pruned_loss=0.05496, over 16507.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2877, pruned_loss=0.06745, over 3319783.67 frames. ], batch size: 75, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:13:19,752 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8777, 3.9312, 4.3328, 4.2943, 4.2989, 3.9504, 3.9780, 3.9492], device='cuda:7'), covar=tensor([0.0288, 0.0383, 0.0276, 0.0380, 0.0368, 0.0300, 0.0771, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0253, 0.0247, 0.0248, 0.0254, 0.0300, 0.0264, 0.0378, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 07:13:23,276 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44218.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:14:09,674 INFO [train.py:904] (7/8) Epoch 5, batch 3650, loss[loss=0.213, simple_loss=0.2759, pruned_loss=0.07506, over 16229.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2871, pruned_loss=0.06835, over 3318425.34 frames. ], batch size: 165, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:14:25,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.873e+02 3.015e+02 3.671e+02 4.591e+02 9.976e+02, threshold=7.342e+02, percent-clipped=3.0 2023-04-28 07:14:30,556 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44265.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:15:22,470 INFO [train.py:904] (7/8) Epoch 5, batch 3700, loss[loss=0.1946, simple_loss=0.2572, pruned_loss=0.06606, over 16484.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2858, pruned_loss=0.07069, over 3294590.42 frames. ], batch size: 75, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:15:45,302 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44315.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:16:36,542 INFO [train.py:904] (7/8) Epoch 5, batch 3750, loss[loss=0.2292, simple_loss=0.3024, pruned_loss=0.07806, over 15487.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.2864, pruned_loss=0.07252, over 3276229.79 frames. ], batch size: 190, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:16:52,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.955e+02 3.644e+02 4.475e+02 7.424e+02, threshold=7.289e+02, percent-clipped=1.0 2023-04-28 07:17:17,637 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 07:17:49,980 INFO [train.py:904] (7/8) Epoch 5, batch 3800, loss[loss=0.2099, simple_loss=0.2766, pruned_loss=0.07155, over 16821.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.2874, pruned_loss=0.074, over 3279453.55 frames. ], batch size: 102, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:17:56,271 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 07:18:27,580 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 07:18:30,281 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 07:18:50,587 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44443.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:19:01,035 INFO [train.py:904] (7/8) Epoch 5, batch 3850, loss[loss=0.2246, simple_loss=0.2978, pruned_loss=0.07575, over 12387.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.2879, pruned_loss=0.07445, over 3265906.39 frames. ], batch size: 246, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:19:16,584 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.887e+02 2.978e+02 3.515e+02 4.161e+02 7.942e+02, threshold=7.030e+02, percent-clipped=2.0 2023-04-28 07:19:20,421 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.76 vs. limit=5.0 2023-04-28 07:19:57,732 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44491.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:20:11,848 INFO [train.py:904] (7/8) Epoch 5, batch 3900, loss[loss=0.2098, simple_loss=0.2784, pruned_loss=0.07055, over 16817.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.2872, pruned_loss=0.07446, over 3268855.56 frames. ], batch size: 89, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:20:49,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0129, 5.5748, 5.7158, 5.5748, 5.4805, 6.0568, 5.6919, 5.4563], device='cuda:7'), covar=tensor([0.0644, 0.1113, 0.1000, 0.1270, 0.1824, 0.0717, 0.0908, 0.1771], device='cuda:7'), in_proj_covar=tensor([0.0284, 0.0397, 0.0389, 0.0339, 0.0448, 0.0411, 0.0321, 0.0459], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:20:49,830 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44527.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:20:54,762 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9516, 5.3176, 4.9559, 5.0146, 4.6251, 4.5372, 4.7541, 5.3605], device='cuda:7'), covar=tensor([0.0673, 0.0672, 0.0932, 0.0483, 0.0657, 0.0775, 0.0694, 0.0652], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0493, 0.0418, 0.0322, 0.0313, 0.0320, 0.0399, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:21:22,437 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44550.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:21:23,257 INFO [train.py:904] (7/8) Epoch 5, batch 3950, loss[loss=0.2121, simple_loss=0.2748, pruned_loss=0.07473, over 16862.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.2871, pruned_loss=0.07506, over 3272080.85 frames. ], batch size: 96, lr: 1.36e-02, grad_scale: 8.0 2023-04-28 07:21:37,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.214e+02 3.071e+02 3.714e+02 4.451e+02 1.184e+03, threshold=7.429e+02, percent-clipped=3.0 2023-04-28 07:21:44,281 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:05,893 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2748, 2.7054, 2.5869, 4.8649, 2.2043, 3.4682, 2.6504, 2.8409], device='cuda:7'), covar=tensor([0.0452, 0.1744, 0.0920, 0.0196, 0.2605, 0.0793, 0.1681, 0.2068], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0328, 0.0262, 0.0314, 0.0368, 0.0316, 0.0293, 0.0398], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:22:16,304 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44588.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:34,894 INFO [train.py:904] (7/8) Epoch 5, batch 4000, loss[loss=0.1946, simple_loss=0.2681, pruned_loss=0.06058, over 16786.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.2869, pruned_loss=0.07481, over 3276746.48 frames. ], batch size: 102, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:22:49,519 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44611.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:51,785 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44613.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:22:54,880 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44615.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:23:01,007 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7530, 4.0328, 3.0880, 2.6020, 3.0025, 2.4588, 4.1526, 4.2907], device='cuda:7'), covar=tensor([0.2189, 0.0596, 0.1239, 0.1484, 0.2126, 0.1392, 0.0374, 0.0378], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0252, 0.0268, 0.0248, 0.0312, 0.0207, 0.0239, 0.0267], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:23:45,850 INFO [train.py:904] (7/8) Epoch 5, batch 4050, loss[loss=0.2261, simple_loss=0.3056, pruned_loss=0.0733, over 16500.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2858, pruned_loss=0.07252, over 3286870.72 frames. ], batch size: 146, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:24:02,950 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.760e+02 3.326e+02 4.025e+02 7.816e+02, threshold=6.651e+02, percent-clipped=1.0 2023-04-28 07:24:04,437 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=44663.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:25:01,130 INFO [train.py:904] (7/8) Epoch 5, batch 4100, loss[loss=0.2051, simple_loss=0.2787, pruned_loss=0.06576, over 16818.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2862, pruned_loss=0.07116, over 3275439.34 frames. ], batch size: 42, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:26:16,736 INFO [train.py:904] (7/8) Epoch 5, batch 4150, loss[loss=0.2294, simple_loss=0.3193, pruned_loss=0.06972, over 16894.00 frames. ], tot_loss[loss=0.2222, simple_loss=0.295, pruned_loss=0.07469, over 3252767.66 frames. ], batch size: 90, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:26:34,908 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.818e+02 3.479e+02 4.379e+02 9.731e+02, threshold=6.958e+02, percent-clipped=3.0 2023-04-28 07:27:18,210 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6984, 2.5109, 1.8774, 2.2648, 2.9277, 2.7791, 3.6227, 3.3363], device='cuda:7'), covar=tensor([0.0022, 0.0168, 0.0249, 0.0216, 0.0105, 0.0168, 0.0051, 0.0078], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0151, 0.0152, 0.0148, 0.0148, 0.0155, 0.0125, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:27:35,374 INFO [train.py:904] (7/8) Epoch 5, batch 4200, loss[loss=0.2391, simple_loss=0.3277, pruned_loss=0.07524, over 16800.00 frames. ], tot_loss[loss=0.2282, simple_loss=0.3027, pruned_loss=0.07688, over 3229300.21 frames. ], batch size: 83, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:27:50,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7523, 4.7577, 4.7943, 3.2293, 4.2450, 4.6622, 4.3972, 2.7208], device='cuda:7'), covar=tensor([0.0260, 0.0008, 0.0018, 0.0192, 0.0033, 0.0035, 0.0020, 0.0230], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0055, 0.0058, 0.0112, 0.0060, 0.0067, 0.0062, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:28:51,245 INFO [train.py:904] (7/8) Epoch 5, batch 4250, loss[loss=0.218, simple_loss=0.3019, pruned_loss=0.06707, over 16738.00 frames. ], tot_loss[loss=0.2308, simple_loss=0.3066, pruned_loss=0.07748, over 3214138.56 frames. ], batch size: 39, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:28:53,349 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 07:29:07,184 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 3.072e+02 3.670e+02 4.674e+02 1.163e+03, threshold=7.340e+02, percent-clipped=6.0 2023-04-28 07:29:38,865 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44883.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:29:46,086 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 07:30:06,014 INFO [train.py:904] (7/8) Epoch 5, batch 4300, loss[loss=0.2887, simple_loss=0.3741, pruned_loss=0.1016, over 16469.00 frames. ], tot_loss[loss=0.2304, simple_loss=0.3075, pruned_loss=0.07661, over 3193111.63 frames. ], batch size: 68, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:30:13,193 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:30:18,130 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8830, 5.0910, 4.8213, 4.7046, 3.9152, 4.9552, 4.9368, 4.4520], device='cuda:7'), covar=tensor([0.0549, 0.0175, 0.0286, 0.0197, 0.1392, 0.0230, 0.0218, 0.0407], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0175, 0.0207, 0.0182, 0.0240, 0.0202, 0.0151, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:31:19,636 INFO [train.py:904] (7/8) Epoch 5, batch 4350, loss[loss=0.2363, simple_loss=0.3182, pruned_loss=0.07718, over 15418.00 frames. ], tot_loss[loss=0.2341, simple_loss=0.3118, pruned_loss=0.07821, over 3189622.76 frames. ], batch size: 191, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:31:36,574 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.041e+02 3.143e+02 3.858e+02 4.477e+02 9.466e+02, threshold=7.715e+02, percent-clipped=1.0 2023-04-28 07:32:35,465 INFO [train.py:904] (7/8) Epoch 5, batch 4400, loss[loss=0.2429, simple_loss=0.3186, pruned_loss=0.08356, over 16376.00 frames. ], tot_loss[loss=0.2356, simple_loss=0.3131, pruned_loss=0.07905, over 3187183.27 frames. ], batch size: 146, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:33:49,615 INFO [train.py:904] (7/8) Epoch 5, batch 4450, loss[loss=0.2344, simple_loss=0.3066, pruned_loss=0.0811, over 11696.00 frames. ], tot_loss[loss=0.2372, simple_loss=0.3156, pruned_loss=0.07936, over 3177244.48 frames. ], batch size: 248, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:33:57,112 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45056.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:34:06,121 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.841e+02 3.317e+02 4.350e+02 8.803e+02, threshold=6.635e+02, percent-clipped=2.0 2023-04-28 07:34:18,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1463, 4.2361, 4.2881, 4.1976, 4.1283, 4.7022, 4.3667, 4.0294], device='cuda:7'), covar=tensor([0.1290, 0.1426, 0.1174, 0.1420, 0.2362, 0.0948, 0.0965, 0.2099], device='cuda:7'), in_proj_covar=tensor([0.0274, 0.0377, 0.0364, 0.0325, 0.0434, 0.0389, 0.0306, 0.0446], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:34:33,618 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45081.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:35:01,849 INFO [train.py:904] (7/8) Epoch 5, batch 4500, loss[loss=0.2436, simple_loss=0.3217, pruned_loss=0.0828, over 16814.00 frames. ], tot_loss[loss=0.2366, simple_loss=0.315, pruned_loss=0.0791, over 3180294.00 frames. ], batch size: 39, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:35:06,618 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0066, 4.9990, 4.5750, 3.9935, 4.8195, 1.8233, 4.6442, 4.6269], device='cuda:7'), covar=tensor([0.0040, 0.0030, 0.0075, 0.0283, 0.0042, 0.1540, 0.0058, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0090, 0.0080, 0.0118, 0.0128, 0.0090, 0.0135, 0.0105, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:35:26,601 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45117.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 07:36:02,317 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45142.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:36:09,009 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-28 07:36:13,672 INFO [train.py:904] (7/8) Epoch 5, batch 4550, loss[loss=0.2773, simple_loss=0.3498, pruned_loss=0.1024, over 17020.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3158, pruned_loss=0.07954, over 3193755.47 frames. ], batch size: 55, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:36:30,366 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 2.621e+02 3.061e+02 3.788e+02 6.051e+02, threshold=6.121e+02, percent-clipped=0.0 2023-04-28 07:36:36,299 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-28 07:37:00,855 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45183.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:37:27,614 INFO [train.py:904] (7/8) Epoch 5, batch 4600, loss[loss=0.2241, simple_loss=0.3109, pruned_loss=0.0686, over 16755.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.316, pruned_loss=0.07907, over 3195112.44 frames. ], batch size: 89, lr: 1.35e-02, grad_scale: 8.0 2023-04-28 07:37:35,906 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:38:12,283 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45231.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:38:31,965 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8088, 4.7805, 5.2918, 5.3432, 5.3261, 4.6917, 4.7725, 4.3571], device='cuda:7'), covar=tensor([0.0194, 0.0258, 0.0228, 0.0297, 0.0299, 0.0242, 0.0654, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0229, 0.0218, 0.0222, 0.0227, 0.0270, 0.0239, 0.0343, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 07:38:40,741 INFO [train.py:904] (7/8) Epoch 5, batch 4650, loss[loss=0.2381, simple_loss=0.3188, pruned_loss=0.07876, over 16770.00 frames. ], tot_loss[loss=0.2358, simple_loss=0.3149, pruned_loss=0.07836, over 3215639.86 frames. ], batch size: 39, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:38:45,219 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45254.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:38:47,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8411, 2.2158, 1.6853, 2.0177, 2.7123, 2.3988, 3.0647, 2.9145], device='cuda:7'), covar=tensor([0.0035, 0.0164, 0.0232, 0.0215, 0.0085, 0.0170, 0.0048, 0.0085], device='cuda:7'), in_proj_covar=tensor([0.0078, 0.0149, 0.0152, 0.0149, 0.0146, 0.0155, 0.0123, 0.0135], device='cuda:7'), out_proj_covar=tensor([9.8754e-05, 1.8317e-04, 1.8375e-04, 1.7949e-04, 1.8177e-04, 1.9293e-04, 1.4853e-04, 1.6850e-04], device='cuda:7') 2023-04-28 07:38:57,177 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.661e+02 3.124e+02 3.951e+02 6.864e+02, threshold=6.249e+02, percent-clipped=3.0 2023-04-28 07:39:16,257 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-28 07:39:27,435 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0851, 3.4078, 3.5010, 1.5473, 3.7710, 3.7339, 2.7461, 2.6662], device='cuda:7'), covar=tensor([0.0848, 0.0147, 0.0152, 0.1123, 0.0044, 0.0050, 0.0373, 0.0459], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0088, 0.0081, 0.0139, 0.0069, 0.0074, 0.0115, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 07:39:55,267 INFO [train.py:904] (7/8) Epoch 5, batch 4700, loss[loss=0.2114, simple_loss=0.2934, pruned_loss=0.06472, over 16685.00 frames. ], tot_loss[loss=0.2332, simple_loss=0.3119, pruned_loss=0.07723, over 3196288.57 frames. ], batch size: 76, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:40:06,733 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8190, 2.2861, 2.4318, 4.3237, 1.8802, 3.1181, 2.4645, 2.4407], device='cuda:7'), covar=tensor([0.0484, 0.1885, 0.0975, 0.0266, 0.2879, 0.0862, 0.1668, 0.2066], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0318, 0.0257, 0.0305, 0.0366, 0.0303, 0.0283, 0.0386], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:41:03,085 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45348.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 07:41:06,801 INFO [train.py:904] (7/8) Epoch 5, batch 4750, loss[loss=0.1937, simple_loss=0.2796, pruned_loss=0.05392, over 16904.00 frames. ], tot_loss[loss=0.2279, simple_loss=0.3068, pruned_loss=0.07449, over 3212985.26 frames. ], batch size: 90, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:41:22,808 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.785e+02 2.710e+02 3.330e+02 4.145e+02 7.902e+02, threshold=6.661e+02, percent-clipped=5.0 2023-04-28 07:41:36,561 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9805, 4.7286, 4.8856, 5.1622, 5.2731, 4.6531, 5.2835, 5.2314], device='cuda:7'), covar=tensor([0.0687, 0.0694, 0.1039, 0.0382, 0.0315, 0.0560, 0.0317, 0.0336], device='cuda:7'), in_proj_covar=tensor([0.0367, 0.0453, 0.0570, 0.0459, 0.0347, 0.0346, 0.0365, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:42:19,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3600, 4.3020, 4.7455, 4.7157, 4.6889, 4.3016, 4.3327, 3.9881], device='cuda:7'), covar=tensor([0.0198, 0.0301, 0.0235, 0.0285, 0.0343, 0.0236, 0.0650, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0228, 0.0218, 0.0224, 0.0226, 0.0271, 0.0238, 0.0338, 0.0203], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 07:42:20,543 INFO [train.py:904] (7/8) Epoch 5, batch 4800, loss[loss=0.2078, simple_loss=0.2986, pruned_loss=0.05846, over 16856.00 frames. ], tot_loss[loss=0.2254, simple_loss=0.3043, pruned_loss=0.07326, over 3195669.76 frames. ], batch size: 96, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:42:33,276 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45409.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 07:42:38,687 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45412.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 07:43:14,554 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45437.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:43:34,380 INFO [train.py:904] (7/8) Epoch 5, batch 4850, loss[loss=0.2661, simple_loss=0.3297, pruned_loss=0.1013, over 12015.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3059, pruned_loss=0.07325, over 3176890.42 frames. ], batch size: 247, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:43:42,631 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 07:43:50,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.510e+02 3.182e+02 3.913e+02 7.711e+02, threshold=6.364e+02, percent-clipped=1.0 2023-04-28 07:44:09,791 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8935, 3.8340, 3.8726, 3.2500, 3.8192, 1.4980, 3.6490, 3.7214], device='cuda:7'), covar=tensor([0.0072, 0.0072, 0.0065, 0.0308, 0.0056, 0.1726, 0.0087, 0.0125], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0076, 0.0112, 0.0121, 0.0085, 0.0130, 0.0100, 0.0112], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:44:47,403 INFO [train.py:904] (7/8) Epoch 5, batch 4900, loss[loss=0.2309, simple_loss=0.3055, pruned_loss=0.07817, over 12371.00 frames. ], tot_loss[loss=0.2255, simple_loss=0.306, pruned_loss=0.07248, over 3161122.67 frames. ], batch size: 247, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:45:23,764 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9388, 4.1112, 3.4249, 2.4556, 3.4622, 2.5669, 4.5627, 4.6877], device='cuda:7'), covar=tensor([0.1743, 0.0597, 0.0976, 0.1268, 0.1551, 0.1087, 0.0336, 0.0335], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0248, 0.0260, 0.0239, 0.0289, 0.0198, 0.0236, 0.0249], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:45:55,666 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1789, 2.3584, 1.6997, 2.0260, 2.8622, 2.5668, 3.2705, 3.1077], device='cuda:7'), covar=tensor([0.0036, 0.0169, 0.0243, 0.0206, 0.0091, 0.0160, 0.0055, 0.0088], device='cuda:7'), in_proj_covar=tensor([0.0078, 0.0151, 0.0153, 0.0149, 0.0147, 0.0154, 0.0121, 0.0135], device='cuda:7'), out_proj_covar=tensor([9.8617e-05, 1.8571e-04, 1.8441e-04, 1.7958e-04, 1.8427e-04, 1.9156e-04, 1.4567e-04, 1.6708e-04], device='cuda:7') 2023-04-28 07:46:01,014 INFO [train.py:904] (7/8) Epoch 5, batch 4950, loss[loss=0.2483, simple_loss=0.3342, pruned_loss=0.0812, over 16895.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.3067, pruned_loss=0.07251, over 3171919.98 frames. ], batch size: 116, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:46:10,671 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7895, 2.7628, 2.7146, 1.8520, 2.4317, 2.6719, 2.6478, 1.7410], device='cuda:7'), covar=tensor([0.0249, 0.0026, 0.0030, 0.0211, 0.0043, 0.0040, 0.0036, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0114, 0.0053, 0.0057, 0.0113, 0.0059, 0.0065, 0.0062, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:46:15,483 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.470e+02 3.018e+02 3.573e+02 4.428e+02 9.346e+02, threshold=7.147e+02, percent-clipped=9.0 2023-04-28 07:46:53,137 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9487, 4.8536, 4.7706, 4.6485, 4.2113, 4.7834, 4.7979, 4.5116], device='cuda:7'), covar=tensor([0.0351, 0.0242, 0.0179, 0.0151, 0.0940, 0.0281, 0.0159, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0211, 0.0184, 0.0242, 0.0209, 0.0151, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:47:04,951 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-04-28 07:47:13,005 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6393, 4.7235, 5.1621, 5.0796, 5.1440, 4.6486, 4.7052, 4.3615], device='cuda:7'), covar=tensor([0.0219, 0.0262, 0.0292, 0.0458, 0.0405, 0.0239, 0.0736, 0.0311], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0219, 0.0224, 0.0225, 0.0273, 0.0239, 0.0340, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 07:47:13,807 INFO [train.py:904] (7/8) Epoch 5, batch 5000, loss[loss=0.226, simple_loss=0.3123, pruned_loss=0.06982, over 16615.00 frames. ], tot_loss[loss=0.2266, simple_loss=0.3078, pruned_loss=0.07274, over 3177990.07 frames. ], batch size: 57, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:47:15,994 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-28 07:47:58,616 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6249, 2.7370, 2.4551, 4.3282, 3.6071, 3.8783, 1.6302, 2.8347], device='cuda:7'), covar=tensor([0.1326, 0.0605, 0.1117, 0.0073, 0.0199, 0.0300, 0.1330, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0140, 0.0165, 0.0084, 0.0169, 0.0169, 0.0160, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 07:48:27,069 INFO [train.py:904] (7/8) Epoch 5, batch 5050, loss[loss=0.2435, simple_loss=0.3176, pruned_loss=0.08468, over 16859.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.308, pruned_loss=0.07206, over 3187634.99 frames. ], batch size: 116, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:48:38,161 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-28 07:48:43,129 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.781e+02 3.302e+02 3.985e+02 6.386e+02, threshold=6.604e+02, percent-clipped=0.0 2023-04-28 07:49:39,976 INFO [train.py:904] (7/8) Epoch 5, batch 5100, loss[loss=0.2046, simple_loss=0.2912, pruned_loss=0.05897, over 16751.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3054, pruned_loss=0.07045, over 3205445.49 frames. ], batch size: 124, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:49:42,402 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45702.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:49:44,552 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45704.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 07:49:56,801 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45712.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 07:50:33,743 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45737.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:50:49,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1302, 4.0916, 4.2664, 4.2043, 4.2449, 4.7076, 4.4282, 4.0982], device='cuda:7'), covar=tensor([0.1200, 0.1742, 0.1155, 0.1525, 0.2265, 0.0912, 0.0943, 0.2212], device='cuda:7'), in_proj_covar=tensor([0.0268, 0.0364, 0.0359, 0.0317, 0.0423, 0.0384, 0.0301, 0.0442], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:50:53,051 INFO [train.py:904] (7/8) Epoch 5, batch 5150, loss[loss=0.2269, simple_loss=0.3211, pruned_loss=0.06639, over 16466.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3066, pruned_loss=0.07013, over 3201950.29 frames. ], batch size: 146, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:51:07,067 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45760.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:51:08,849 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.626e+02 3.241e+02 3.847e+02 8.905e+02, threshold=6.482e+02, percent-clipped=5.0 2023-04-28 07:51:10,645 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45763.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:51:27,012 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45774.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:51:43,499 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=45785.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:52:05,645 INFO [train.py:904] (7/8) Epoch 5, batch 5200, loss[loss=0.2525, simple_loss=0.3243, pruned_loss=0.09035, over 11853.00 frames. ], tot_loss[loss=0.2238, simple_loss=0.3061, pruned_loss=0.07075, over 3197756.53 frames. ], batch size: 246, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:52:22,497 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3040, 4.3799, 4.4919, 4.4056, 4.3658, 4.9048, 4.5270, 4.2314], device='cuda:7'), covar=tensor([0.1056, 0.1416, 0.1049, 0.1325, 0.2214, 0.0842, 0.0983, 0.2037], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0366, 0.0361, 0.0319, 0.0424, 0.0385, 0.0302, 0.0444], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:52:54,191 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45835.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:53:16,338 INFO [train.py:904] (7/8) Epoch 5, batch 5250, loss[loss=0.2043, simple_loss=0.2856, pruned_loss=0.06151, over 16640.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.3022, pruned_loss=0.06947, over 3215123.02 frames. ], batch size: 68, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:53:18,062 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4262, 2.0574, 1.5660, 1.8069, 2.4272, 2.2625, 2.5967, 2.6709], device='cuda:7'), covar=tensor([0.0043, 0.0178, 0.0234, 0.0214, 0.0098, 0.0147, 0.0062, 0.0091], device='cuda:7'), in_proj_covar=tensor([0.0078, 0.0151, 0.0154, 0.0150, 0.0147, 0.0153, 0.0121, 0.0135], device='cuda:7'), out_proj_covar=tensor([9.8086e-05, 1.8574e-04, 1.8609e-04, 1.8129e-04, 1.8388e-04, 1.8928e-04, 1.4467e-04, 1.6741e-04], device='cuda:7') 2023-04-28 07:53:27,288 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45859.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:53:31,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.734e+02 3.218e+02 3.993e+02 9.159e+02, threshold=6.436e+02, percent-clipped=4.0 2023-04-28 07:53:55,617 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3786, 4.1892, 4.1533, 2.5105, 3.6920, 4.0613, 3.8650, 2.1744], device='cuda:7'), covar=tensor([0.0326, 0.0017, 0.0030, 0.0272, 0.0048, 0.0055, 0.0034, 0.0354], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0055, 0.0060, 0.0115, 0.0060, 0.0068, 0.0064, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:54:26,350 INFO [train.py:904] (7/8) Epoch 5, batch 5300, loss[loss=0.1912, simple_loss=0.2758, pruned_loss=0.05328, over 16591.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2977, pruned_loss=0.06778, over 3219794.85 frames. ], batch size: 62, lr: 1.34e-02, grad_scale: 8.0 2023-04-28 07:54:53,156 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45920.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:54:53,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 07:55:36,423 INFO [train.py:904] (7/8) Epoch 5, batch 5350, loss[loss=0.2131, simple_loss=0.2898, pruned_loss=0.06823, over 17052.00 frames. ], tot_loss[loss=0.2148, simple_loss=0.2961, pruned_loss=0.06678, over 3232755.31 frames. ], batch size: 53, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 07:55:39,676 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 07:55:53,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.936e+02 3.263e+02 4.063e+02 1.076e+03, threshold=6.526e+02, percent-clipped=3.0 2023-04-28 07:56:52,506 INFO [train.py:904] (7/8) Epoch 5, batch 5400, loss[loss=0.2209, simple_loss=0.3045, pruned_loss=0.06864, over 16580.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2991, pruned_loss=0.06774, over 3235070.11 frames. ], batch size: 134, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 07:56:58,323 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46004.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 07:57:41,381 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9334, 3.5849, 3.4951, 2.2543, 3.1578, 3.4410, 3.4521, 1.8591], device='cuda:7'), covar=tensor([0.0331, 0.0018, 0.0031, 0.0234, 0.0045, 0.0049, 0.0028, 0.0316], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0054, 0.0061, 0.0117, 0.0061, 0.0069, 0.0064, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 07:58:09,293 INFO [train.py:904] (7/8) Epoch 5, batch 5450, loss[loss=0.2812, simple_loss=0.3556, pruned_loss=0.1034, over 16179.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.303, pruned_loss=0.06992, over 3223351.93 frames. ], batch size: 165, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 07:58:11,276 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46052.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 07:58:19,265 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46058.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 07:58:24,989 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.677e+02 3.104e+02 3.655e+02 4.572e+02 1.003e+03, threshold=7.310e+02, percent-clipped=8.0 2023-04-28 07:58:51,238 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2504, 3.2036, 3.2874, 3.4775, 3.4451, 3.2070, 3.4174, 3.5099], device='cuda:7'), covar=tensor([0.0760, 0.0676, 0.1026, 0.0489, 0.0585, 0.1593, 0.0664, 0.0514], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0461, 0.0578, 0.0468, 0.0350, 0.0349, 0.0372, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 07:59:09,264 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-28 07:59:22,164 INFO [train.py:904] (7/8) Epoch 5, batch 5500, loss[loss=0.314, simple_loss=0.3649, pruned_loss=0.1315, over 11689.00 frames. ], tot_loss[loss=0.235, simple_loss=0.3133, pruned_loss=0.07837, over 3168574.70 frames. ], batch size: 248, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:00:02,491 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 08:00:07,595 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46130.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:00:39,136 INFO [train.py:904] (7/8) Epoch 5, batch 5550, loss[loss=0.3148, simple_loss=0.3736, pruned_loss=0.128, over 16524.00 frames. ], tot_loss[loss=0.2449, simple_loss=0.3212, pruned_loss=0.0843, over 3156949.03 frames. ], batch size: 68, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:00:56,875 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.500e+02 4.429e+02 5.470e+02 6.755e+02 1.488e+03, threshold=1.094e+03, percent-clipped=17.0 2023-04-28 08:01:46,484 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-28 08:01:58,730 INFO [train.py:904] (7/8) Epoch 5, batch 5600, loss[loss=0.2309, simple_loss=0.3099, pruned_loss=0.07593, over 16555.00 frames. ], tot_loss[loss=0.2544, simple_loss=0.3277, pruned_loss=0.09051, over 3115500.01 frames. ], batch size: 62, lr: 1.33e-02, grad_scale: 16.0 2023-04-28 08:02:21,460 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3939, 3.3047, 3.2941, 2.7167, 3.3064, 2.0843, 3.1001, 2.8243], device='cuda:7'), covar=tensor([0.0093, 0.0070, 0.0107, 0.0225, 0.0072, 0.1459, 0.0093, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0078, 0.0118, 0.0127, 0.0089, 0.0138, 0.0105, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:02:23,969 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46215.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:03:23,174 INFO [train.py:904] (7/8) Epoch 5, batch 5650, loss[loss=0.234, simple_loss=0.3172, pruned_loss=0.07543, over 17031.00 frames. ], tot_loss[loss=0.263, simple_loss=0.334, pruned_loss=0.09603, over 3083961.07 frames. ], batch size: 50, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:03:42,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.653e+02 4.676e+02 6.240e+02 7.908e+02 1.367e+03, threshold=1.248e+03, percent-clipped=1.0 2023-04-28 08:04:04,261 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46276.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:04:43,835 INFO [train.py:904] (7/8) Epoch 5, batch 5700, loss[loss=0.3432, simple_loss=0.3786, pruned_loss=0.1539, over 10992.00 frames. ], tot_loss[loss=0.2662, simple_loss=0.3364, pruned_loss=0.09801, over 3080004.86 frames. ], batch size: 247, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:05:41,867 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46337.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:05:51,130 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 08:06:04,719 INFO [train.py:904] (7/8) Epoch 5, batch 5750, loss[loss=0.2839, simple_loss=0.3354, pruned_loss=0.1163, over 11484.00 frames. ], tot_loss[loss=0.269, simple_loss=0.3395, pruned_loss=0.09927, over 3079514.03 frames. ], batch size: 248, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:06:16,577 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46358.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:06:23,499 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.622e+02 4.192e+02 4.948e+02 6.377e+02 1.174e+03, threshold=9.897e+02, percent-clipped=0.0 2023-04-28 08:06:30,046 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8597, 3.6596, 3.9068, 4.0946, 4.1713, 3.7546, 4.1685, 4.1531], device='cuda:7'), covar=tensor([0.0792, 0.0807, 0.1098, 0.0503, 0.0430, 0.1020, 0.0497, 0.0436], device='cuda:7'), in_proj_covar=tensor([0.0361, 0.0444, 0.0552, 0.0453, 0.0341, 0.0340, 0.0365, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:06:52,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6015, 3.6950, 3.7704, 3.7396, 3.7379, 4.1421, 3.9550, 3.7280], device='cuda:7'), covar=tensor([0.2267, 0.1881, 0.1481, 0.2164, 0.2541, 0.1430, 0.1241, 0.2589], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0369, 0.0364, 0.0320, 0.0426, 0.0391, 0.0303, 0.0443], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 08:07:26,293 INFO [train.py:904] (7/8) Epoch 5, batch 5800, loss[loss=0.2972, simple_loss=0.3572, pruned_loss=0.1186, over 15275.00 frames. ], tot_loss[loss=0.2678, simple_loss=0.3394, pruned_loss=0.09812, over 3084876.17 frames. ], batch size: 191, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:07:35,757 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46406.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:08:14,660 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46430.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:08:46,735 INFO [train.py:904] (7/8) Epoch 5, batch 5850, loss[loss=0.2573, simple_loss=0.3341, pruned_loss=0.09025, over 16246.00 frames. ], tot_loss[loss=0.2644, simple_loss=0.337, pruned_loss=0.09593, over 3076888.63 frames. ], batch size: 165, lr: 1.33e-02, grad_scale: 4.0 2023-04-28 08:09:01,628 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8648, 3.1734, 2.2694, 4.7835, 3.9092, 4.0289, 2.0116, 2.9437], device='cuda:7'), covar=tensor([0.1377, 0.0552, 0.1380, 0.0071, 0.0349, 0.0341, 0.1279, 0.0843], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0139, 0.0165, 0.0082, 0.0172, 0.0173, 0.0161, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 08:09:06,684 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.458e+02 3.799e+02 4.852e+02 6.197e+02 1.255e+03, threshold=9.704e+02, percent-clipped=4.0 2023-04-28 08:09:29,293 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46478.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:10:09,077 INFO [train.py:904] (7/8) Epoch 5, batch 5900, loss[loss=0.3509, simple_loss=0.3819, pruned_loss=0.16, over 11507.00 frames. ], tot_loss[loss=0.264, simple_loss=0.3364, pruned_loss=0.09582, over 3067358.04 frames. ], batch size: 246, lr: 1.33e-02, grad_scale: 4.0 2023-04-28 08:10:36,168 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46515.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:10:48,553 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 08:11:06,317 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1990, 2.4697, 1.8237, 1.9755, 3.0011, 2.6597, 3.3303, 3.2079], device='cuda:7'), covar=tensor([0.0036, 0.0173, 0.0227, 0.0230, 0.0098, 0.0149, 0.0069, 0.0084], device='cuda:7'), in_proj_covar=tensor([0.0077, 0.0151, 0.0154, 0.0150, 0.0146, 0.0154, 0.0123, 0.0133], device='cuda:7'), out_proj_covar=tensor([9.6142e-05, 1.8462e-04, 1.8497e-04, 1.8031e-04, 1.8086e-04, 1.8931e-04, 1.4704e-04, 1.6495e-04], device='cuda:7') 2023-04-28 08:11:31,964 INFO [train.py:904] (7/8) Epoch 5, batch 5950, loss[loss=0.2906, simple_loss=0.3508, pruned_loss=0.1152, over 11600.00 frames. ], tot_loss[loss=0.2616, simple_loss=0.3358, pruned_loss=0.09368, over 3080893.02 frames. ], batch size: 248, lr: 1.33e-02, grad_scale: 4.0 2023-04-28 08:11:38,088 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 08:11:52,754 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46563.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:11:53,495 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.225e+02 3.478e+02 4.456e+02 5.303e+02 9.511e+02, threshold=8.911e+02, percent-clipped=0.0 2023-04-28 08:12:43,357 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46595.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:12:51,980 INFO [train.py:904] (7/8) Epoch 5, batch 6000, loss[loss=0.2159, simple_loss=0.295, pruned_loss=0.06837, over 16717.00 frames. ], tot_loss[loss=0.26, simple_loss=0.3343, pruned_loss=0.09283, over 3097888.64 frames. ], batch size: 57, lr: 1.33e-02, grad_scale: 8.0 2023-04-28 08:12:51,980 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 08:13:04,051 INFO [train.py:938] (7/8) Epoch 5, validation: loss=0.1879, simple_loss=0.2992, pruned_loss=0.03826, over 944034.00 frames. 2023-04-28 08:13:04,052 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 08:13:05,722 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46602.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:13:51,786 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46632.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:14:27,403 INFO [train.py:904] (7/8) Epoch 5, batch 6050, loss[loss=0.2308, simple_loss=0.3199, pruned_loss=0.07082, over 16917.00 frames. ], tot_loss[loss=0.258, simple_loss=0.333, pruned_loss=0.0915, over 3123279.84 frames. ], batch size: 109, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:14:37,505 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46656.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:14:48,324 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46663.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:14:49,000 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.236e+02 4.108e+02 4.980e+02 6.205e+02 1.363e+03, threshold=9.960e+02, percent-clipped=7.0 2023-04-28 08:15:45,459 INFO [train.py:904] (7/8) Epoch 5, batch 6100, loss[loss=0.2293, simple_loss=0.3164, pruned_loss=0.07111, over 16755.00 frames. ], tot_loss[loss=0.2562, simple_loss=0.332, pruned_loss=0.09019, over 3120418.83 frames. ], batch size: 83, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:16:25,016 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6800, 3.7285, 4.2073, 4.1717, 4.1698, 3.8147, 3.8183, 3.7527], device='cuda:7'), covar=tensor([0.0298, 0.0427, 0.0312, 0.0353, 0.0364, 0.0306, 0.0798, 0.0424], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0228, 0.0234, 0.0229, 0.0279, 0.0246, 0.0353, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 08:17:01,188 INFO [train.py:904] (7/8) Epoch 5, batch 6150, loss[loss=0.237, simple_loss=0.3203, pruned_loss=0.07685, over 16859.00 frames. ], tot_loss[loss=0.2537, simple_loss=0.3293, pruned_loss=0.08911, over 3110641.40 frames. ], batch size: 102, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:17:19,712 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8809, 4.0382, 3.3005, 2.5725, 3.1551, 2.4124, 4.3799, 4.1794], device='cuda:7'), covar=tensor([0.2010, 0.0610, 0.1148, 0.1327, 0.1865, 0.1329, 0.0323, 0.0556], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0250, 0.0269, 0.0243, 0.0293, 0.0202, 0.0241, 0.0251], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:17:22,723 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.183e+02 3.664e+02 4.778e+02 7.050e+02 1.596e+03, threshold=9.557e+02, percent-clipped=5.0 2023-04-28 08:17:48,282 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-28 08:18:03,005 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9969, 4.0468, 1.7820, 4.6366, 2.7549, 4.4742, 2.3249, 2.9888], device='cuda:7'), covar=tensor([0.0121, 0.0253, 0.1854, 0.0030, 0.0773, 0.0334, 0.1308, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0147, 0.0179, 0.0080, 0.0161, 0.0183, 0.0186, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:18:05,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9556, 3.9198, 3.8156, 3.2276, 3.8691, 1.7175, 3.6753, 3.5854], device='cuda:7'), covar=tensor([0.0080, 0.0063, 0.0099, 0.0296, 0.0065, 0.1797, 0.0095, 0.0146], device='cuda:7'), in_proj_covar=tensor([0.0091, 0.0078, 0.0118, 0.0126, 0.0090, 0.0141, 0.0105, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:18:21,102 INFO [train.py:904] (7/8) Epoch 5, batch 6200, loss[loss=0.2645, simple_loss=0.3375, pruned_loss=0.09573, over 16738.00 frames. ], tot_loss[loss=0.2531, simple_loss=0.3276, pruned_loss=0.0893, over 3096718.55 frames. ], batch size: 89, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:18:36,444 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46810.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:19:37,514 INFO [train.py:904] (7/8) Epoch 5, batch 6250, loss[loss=0.2226, simple_loss=0.316, pruned_loss=0.06462, over 16659.00 frames. ], tot_loss[loss=0.2519, simple_loss=0.3271, pruned_loss=0.08838, over 3111699.16 frames. ], batch size: 89, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:19:57,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.405e+02 3.628e+02 4.524e+02 5.799e+02 1.377e+03, threshold=9.049e+02, percent-clipped=5.0 2023-04-28 08:20:10,121 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46871.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:20:36,046 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1897, 1.9381, 2.1513, 3.5682, 1.7764, 2.7768, 2.1237, 2.0009], device='cuda:7'), covar=tensor([0.0671, 0.2028, 0.1112, 0.0332, 0.2942, 0.0941, 0.1887, 0.2318], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0312, 0.0259, 0.0306, 0.0367, 0.0300, 0.0281, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:20:55,078 INFO [train.py:904] (7/8) Epoch 5, batch 6300, loss[loss=0.2665, simple_loss=0.3532, pruned_loss=0.08992, over 16623.00 frames. ], tot_loss[loss=0.2507, simple_loss=0.3267, pruned_loss=0.08733, over 3132792.96 frames. ], batch size: 68, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:21:43,910 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46932.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:22:12,094 INFO [train.py:904] (7/8) Epoch 5, batch 6350, loss[loss=0.2619, simple_loss=0.3333, pruned_loss=0.09522, over 16436.00 frames. ], tot_loss[loss=0.253, simple_loss=0.3279, pruned_loss=0.08904, over 3123964.00 frames. ], batch size: 146, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:22:13,437 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46951.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 08:22:14,947 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7117, 4.4682, 3.9459, 2.2243, 3.2158, 2.7412, 3.9858, 4.0483], device='cuda:7'), covar=tensor([0.0211, 0.0363, 0.0540, 0.1537, 0.0726, 0.0878, 0.0641, 0.0750], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0127, 0.0154, 0.0142, 0.0134, 0.0126, 0.0142, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:22:22,813 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46958.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:22:31,678 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.349e+02 3.778e+02 4.701e+02 5.946e+02 1.466e+03, threshold=9.402e+02, percent-clipped=8.0 2023-04-28 08:22:57,127 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=46980.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:23:28,014 INFO [train.py:904] (7/8) Epoch 5, batch 6400, loss[loss=0.3665, simple_loss=0.4148, pruned_loss=0.1591, over 11097.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.3286, pruned_loss=0.0909, over 3097387.17 frames. ], batch size: 247, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:23:30,342 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6989, 3.0960, 2.4749, 4.6796, 3.9797, 4.1792, 1.6792, 3.0076], device='cuda:7'), covar=tensor([0.1340, 0.0517, 0.1129, 0.0077, 0.0234, 0.0333, 0.1294, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0144, 0.0168, 0.0085, 0.0179, 0.0179, 0.0163, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:24:42,569 INFO [train.py:904] (7/8) Epoch 5, batch 6450, loss[loss=0.2218, simple_loss=0.3101, pruned_loss=0.0667, over 17279.00 frames. ], tot_loss[loss=0.2535, simple_loss=0.3278, pruned_loss=0.08955, over 3090338.19 frames. ], batch size: 52, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:25:02,353 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.185e+02 3.689e+02 4.668e+02 5.826e+02 1.210e+03, threshold=9.337e+02, percent-clipped=4.0 2023-04-28 08:25:59,101 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7283, 1.5819, 2.1352, 2.7335, 2.6000, 2.9497, 1.6496, 2.8638], device='cuda:7'), covar=tensor([0.0077, 0.0256, 0.0157, 0.0103, 0.0103, 0.0077, 0.0255, 0.0068], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0141, 0.0126, 0.0123, 0.0126, 0.0090, 0.0140, 0.0079], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 08:25:59,698 INFO [train.py:904] (7/8) Epoch 5, batch 6500, loss[loss=0.229, simple_loss=0.3088, pruned_loss=0.07463, over 16749.00 frames. ], tot_loss[loss=0.2507, simple_loss=0.325, pruned_loss=0.08819, over 3105962.39 frames. ], batch size: 89, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:26:16,978 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-28 08:26:38,961 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6097, 3.8189, 1.6838, 4.0508, 2.5598, 3.9421, 1.9132, 2.7619], device='cuda:7'), covar=tensor([0.0133, 0.0228, 0.1644, 0.0045, 0.0740, 0.0361, 0.1484, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0147, 0.0179, 0.0081, 0.0162, 0.0184, 0.0187, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:27:10,666 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0088, 1.6690, 1.4534, 1.3990, 1.8067, 1.6581, 1.7558, 1.8884], device='cuda:7'), covar=tensor([0.0032, 0.0107, 0.0157, 0.0161, 0.0087, 0.0126, 0.0081, 0.0087], device='cuda:7'), in_proj_covar=tensor([0.0076, 0.0152, 0.0152, 0.0150, 0.0146, 0.0154, 0.0128, 0.0134], device='cuda:7'), out_proj_covar=tensor([9.4317e-05, 1.8561e-04, 1.8209e-04, 1.7953e-04, 1.8009e-04, 1.8941e-04, 1.5291e-04, 1.6435e-04], device='cuda:7') 2023-04-28 08:27:18,016 INFO [train.py:904] (7/8) Epoch 5, batch 6550, loss[loss=0.2843, simple_loss=0.3646, pruned_loss=0.102, over 15384.00 frames. ], tot_loss[loss=0.2529, simple_loss=0.328, pruned_loss=0.08893, over 3116989.69 frames. ], batch size: 191, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:27:37,105 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.214e+02 3.842e+02 4.732e+02 5.920e+02 1.050e+03, threshold=9.464e+02, percent-clipped=2.0 2023-04-28 08:27:41,305 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47166.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:28:33,664 INFO [train.py:904] (7/8) Epoch 5, batch 6600, loss[loss=0.3178, simple_loss=0.3741, pruned_loss=0.1307, over 15384.00 frames. ], tot_loss[loss=0.2558, simple_loss=0.3307, pruned_loss=0.0904, over 3097640.88 frames. ], batch size: 190, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:29:08,530 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 08:29:35,144 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:29:51,339 INFO [train.py:904] (7/8) Epoch 5, batch 6650, loss[loss=0.3172, simple_loss=0.3695, pruned_loss=0.1325, over 11612.00 frames. ], tot_loss[loss=0.2567, simple_loss=0.3308, pruned_loss=0.09127, over 3099967.94 frames. ], batch size: 246, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:29:52,536 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47251.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:30:02,777 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47258.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:30:11,436 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.376e+02 3.750e+02 4.777e+02 6.524e+02 9.688e+02, threshold=9.554e+02, percent-clipped=1.0 2023-04-28 08:30:25,826 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2789, 1.7897, 2.3538, 3.0959, 2.8421, 3.5930, 1.7512, 3.3004], device='cuda:7'), covar=tensor([0.0055, 0.0235, 0.0138, 0.0095, 0.0098, 0.0046, 0.0224, 0.0036], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0143, 0.0128, 0.0124, 0.0126, 0.0090, 0.0141, 0.0079], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 08:31:04,890 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47299.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:31:06,612 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-28 08:31:06,906 INFO [train.py:904] (7/8) Epoch 5, batch 6700, loss[loss=0.3285, simple_loss=0.3638, pruned_loss=0.1466, over 11759.00 frames. ], tot_loss[loss=0.2565, simple_loss=0.3297, pruned_loss=0.09163, over 3096935.66 frames. ], batch size: 247, lr: 1.32e-02, grad_scale: 8.0 2023-04-28 08:31:10,003 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47302.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:31:16,567 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47306.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:32:26,002 INFO [train.py:904] (7/8) Epoch 5, batch 6750, loss[loss=0.3173, simple_loss=0.3677, pruned_loss=0.1335, over 12027.00 frames. ], tot_loss[loss=0.2555, simple_loss=0.3284, pruned_loss=0.09131, over 3103113.95 frames. ], batch size: 247, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:32:45,844 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.669e+02 3.980e+02 4.802e+02 5.958e+02 9.394e+02, threshold=9.603e+02, percent-clipped=0.0 2023-04-28 08:32:58,055 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7124, 4.9925, 4.6538, 4.6902, 4.3813, 4.4697, 4.5167, 5.0257], device='cuda:7'), covar=tensor([0.0684, 0.0626, 0.1006, 0.0471, 0.0627, 0.0706, 0.0639, 0.0674], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0480, 0.0420, 0.0313, 0.0305, 0.0325, 0.0396, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:33:09,269 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47379.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:33:20,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9213, 4.0258, 3.7726, 3.7293, 3.4334, 3.8201, 3.5587, 3.5832], device='cuda:7'), covar=tensor([0.0441, 0.0253, 0.0220, 0.0169, 0.0715, 0.0280, 0.0771, 0.0484], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0183, 0.0207, 0.0179, 0.0235, 0.0207, 0.0152, 0.0231], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:33:40,660 INFO [train.py:904] (7/8) Epoch 5, batch 6800, loss[loss=0.2966, simple_loss=0.3467, pruned_loss=0.1233, over 11493.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.3287, pruned_loss=0.09084, over 3104297.68 frames. ], batch size: 248, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:34:21,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5869, 4.2217, 3.5351, 1.9963, 3.1332, 2.6428, 3.6474, 3.9403], device='cuda:7'), covar=tensor([0.0219, 0.0416, 0.0581, 0.1633, 0.0700, 0.0898, 0.0611, 0.0652], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0125, 0.0152, 0.0139, 0.0131, 0.0123, 0.0138, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:34:29,479 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7243, 3.5725, 3.7367, 3.6417, 3.7227, 4.0715, 3.7972, 3.5560], device='cuda:7'), covar=tensor([0.1845, 0.2080, 0.1716, 0.2185, 0.2557, 0.1554, 0.1392, 0.2436], device='cuda:7'), in_proj_covar=tensor([0.0274, 0.0372, 0.0387, 0.0332, 0.0437, 0.0399, 0.0312, 0.0456], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 08:34:41,123 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47440.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:34:57,641 INFO [train.py:904] (7/8) Epoch 5, batch 6850, loss[loss=0.2413, simple_loss=0.3328, pruned_loss=0.0749, over 16721.00 frames. ], tot_loss[loss=0.2568, simple_loss=0.3304, pruned_loss=0.09156, over 3094677.50 frames. ], batch size: 134, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:35:17,577 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.600e+02 3.467e+02 4.392e+02 5.504e+02 1.073e+03, threshold=8.784e+02, percent-clipped=2.0 2023-04-28 08:35:21,106 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:36:12,141 INFO [train.py:904] (7/8) Epoch 5, batch 6900, loss[loss=0.3272, simple_loss=0.3658, pruned_loss=0.1442, over 11185.00 frames. ], tot_loss[loss=0.2577, simple_loss=0.3327, pruned_loss=0.09136, over 3100022.10 frames. ], batch size: 250, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:36:33,891 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47514.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:36:54,661 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7401, 3.6255, 3.7349, 3.9191, 4.0191, 3.6141, 3.9473, 4.0072], device='cuda:7'), covar=tensor([0.0769, 0.0611, 0.1107, 0.0524, 0.0411, 0.1191, 0.0544, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0459, 0.0580, 0.0471, 0.0351, 0.0345, 0.0377, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:37:20,689 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1275, 4.1383, 4.0088, 3.3924, 4.0621, 1.5930, 3.8389, 3.8617], device='cuda:7'), covar=tensor([0.0068, 0.0052, 0.0085, 0.0276, 0.0053, 0.1763, 0.0081, 0.0123], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0076, 0.0118, 0.0122, 0.0087, 0.0139, 0.0103, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:37:30,509 INFO [train.py:904] (7/8) Epoch 5, batch 6950, loss[loss=0.2634, simple_loss=0.3311, pruned_loss=0.09787, over 16301.00 frames. ], tot_loss[loss=0.2619, simple_loss=0.3354, pruned_loss=0.09417, over 3083529.95 frames. ], batch size: 165, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:37:50,870 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 3.906e+02 5.173e+02 7.099e+02 1.028e+03, threshold=1.035e+03, percent-clipped=6.0 2023-04-28 08:38:32,613 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-28 08:38:42,857 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47597.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:38:49,101 INFO [train.py:904] (7/8) Epoch 5, batch 7000, loss[loss=0.2731, simple_loss=0.3479, pruned_loss=0.09912, over 15567.00 frames. ], tot_loss[loss=0.261, simple_loss=0.3355, pruned_loss=0.09326, over 3073107.81 frames. ], batch size: 191, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:38:54,308 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.54 vs. limit=5.0 2023-04-28 08:38:55,423 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4643, 4.1320, 3.5466, 1.8168, 3.1395, 2.5834, 3.6476, 3.9465], device='cuda:7'), covar=tensor([0.0203, 0.0373, 0.0536, 0.1649, 0.0670, 0.0844, 0.0516, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0124, 0.0153, 0.0140, 0.0132, 0.0122, 0.0138, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:40:06,968 INFO [train.py:904] (7/8) Epoch 5, batch 7050, loss[loss=0.2795, simple_loss=0.3462, pruned_loss=0.1063, over 15239.00 frames. ], tot_loss[loss=0.2602, simple_loss=0.3358, pruned_loss=0.09223, over 3102509.39 frames. ], batch size: 191, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:40:10,863 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-04-28 08:40:26,883 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 3.724e+02 4.498e+02 5.703e+02 1.501e+03, threshold=8.996e+02, percent-clipped=4.0 2023-04-28 08:41:10,026 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 08:41:26,199 INFO [train.py:904] (7/8) Epoch 5, batch 7100, loss[loss=0.2442, simple_loss=0.3219, pruned_loss=0.08323, over 16976.00 frames. ], tot_loss[loss=0.2583, simple_loss=0.3337, pruned_loss=0.09139, over 3100802.16 frames. ], batch size: 41, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:42:20,393 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47735.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:42:24,848 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47738.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:42:44,680 INFO [train.py:904] (7/8) Epoch 5, batch 7150, loss[loss=0.2372, simple_loss=0.3134, pruned_loss=0.08047, over 17121.00 frames. ], tot_loss[loss=0.2573, simple_loss=0.3317, pruned_loss=0.09149, over 3086615.26 frames. ], batch size: 48, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:43:03,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.709e+02 4.695e+02 5.943e+02 1.245e+03, threshold=9.389e+02, percent-clipped=5.0 2023-04-28 08:43:40,027 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47788.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:43:57,726 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47799.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:44:00,019 INFO [train.py:904] (7/8) Epoch 5, batch 7200, loss[loss=0.245, simple_loss=0.3138, pruned_loss=0.08814, over 15512.00 frames. ], tot_loss[loss=0.2542, simple_loss=0.3293, pruned_loss=0.08954, over 3079397.97 frames. ], batch size: 191, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:44:13,973 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47809.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:45:22,292 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:45:24,169 INFO [train.py:904] (7/8) Epoch 5, batch 7250, loss[loss=0.208, simple_loss=0.2816, pruned_loss=0.06718, over 16643.00 frames. ], tot_loss[loss=0.2499, simple_loss=0.3254, pruned_loss=0.08717, over 3096030.71 frames. ], batch size: 62, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:45:43,494 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.043e+02 3.368e+02 4.070e+02 5.805e+02 1.027e+03, threshold=8.141e+02, percent-clipped=1.0 2023-04-28 08:45:53,841 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47870.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 08:46:24,617 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5744, 3.4054, 3.0758, 1.9118, 2.5693, 1.9772, 3.0756, 3.2896], device='cuda:7'), covar=tensor([0.0313, 0.0426, 0.0494, 0.1508, 0.0761, 0.0957, 0.0703, 0.0632], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0124, 0.0153, 0.0140, 0.0132, 0.0123, 0.0139, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 08:46:34,998 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1509, 4.1253, 3.9593, 3.3978, 4.0306, 1.6585, 3.8292, 3.8799], device='cuda:7'), covar=tensor([0.0060, 0.0045, 0.0096, 0.0264, 0.0061, 0.1805, 0.0085, 0.0126], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0078, 0.0119, 0.0124, 0.0090, 0.0142, 0.0105, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:46:35,002 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47897.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:46:40,689 INFO [train.py:904] (7/8) Epoch 5, batch 7300, loss[loss=0.2479, simple_loss=0.3254, pruned_loss=0.08522, over 16148.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3252, pruned_loss=0.08762, over 3087116.75 frames. ], batch size: 165, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:46:48,389 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8200, 2.0563, 1.5956, 1.7692, 2.3967, 2.1559, 2.9875, 2.7462], device='cuda:7'), covar=tensor([0.0040, 0.0209, 0.0284, 0.0251, 0.0139, 0.0202, 0.0085, 0.0116], device='cuda:7'), in_proj_covar=tensor([0.0073, 0.0149, 0.0151, 0.0148, 0.0143, 0.0152, 0.0124, 0.0131], device='cuda:7'), out_proj_covar=tensor([9.0858e-05, 1.8128e-04, 1.8035e-04, 1.7671e-04, 1.7619e-04, 1.8545e-04, 1.4693e-04, 1.6110e-04], device='cuda:7') 2023-04-28 08:47:02,717 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8201, 4.1149, 3.8608, 3.9507, 3.5964, 3.7473, 3.8241, 4.0205], device='cuda:7'), covar=tensor([0.0765, 0.0678, 0.0913, 0.0465, 0.0597, 0.1143, 0.0587, 0.0868], device='cuda:7'), in_proj_covar=tensor([0.0368, 0.0478, 0.0411, 0.0307, 0.0297, 0.0323, 0.0386, 0.0345], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:47:49,458 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=47945.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:47:58,215 INFO [train.py:904] (7/8) Epoch 5, batch 7350, loss[loss=0.2369, simple_loss=0.32, pruned_loss=0.07689, over 17075.00 frames. ], tot_loss[loss=0.2512, simple_loss=0.3256, pruned_loss=0.08838, over 3057805.70 frames. ], batch size: 55, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:48:17,668 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.639e+02 4.067e+02 5.121e+02 6.309e+02 1.579e+03, threshold=1.024e+03, percent-clipped=9.0 2023-04-28 08:49:18,343 INFO [train.py:904] (7/8) Epoch 5, batch 7400, loss[loss=0.2385, simple_loss=0.3216, pruned_loss=0.0777, over 16363.00 frames. ], tot_loss[loss=0.2532, simple_loss=0.3274, pruned_loss=0.08951, over 3058751.17 frames. ], batch size: 146, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:50:13,291 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48035.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:50:38,176 INFO [train.py:904] (7/8) Epoch 5, batch 7450, loss[loss=0.3044, simple_loss=0.3511, pruned_loss=0.1288, over 11351.00 frames. ], tot_loss[loss=0.2539, simple_loss=0.3279, pruned_loss=0.09, over 3062242.65 frames. ], batch size: 246, lr: 1.31e-02, grad_scale: 8.0 2023-04-28 08:51:00,813 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.428e+02 4.070e+02 4.599e+02 6.099e+02 1.111e+03, threshold=9.198e+02, percent-clipped=2.0 2023-04-28 08:51:10,083 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 08:51:14,618 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48072.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:51:31,925 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48083.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:51:49,953 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48094.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:51:53,053 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8158, 4.7731, 5.3922, 5.3225, 5.3453, 4.9079, 4.9225, 4.5098], device='cuda:7'), covar=tensor([0.0274, 0.0330, 0.0241, 0.0330, 0.0349, 0.0247, 0.0775, 0.0357], device='cuda:7'), in_proj_covar=tensor([0.0235, 0.0222, 0.0233, 0.0230, 0.0280, 0.0246, 0.0350, 0.0206], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 08:52:00,144 INFO [train.py:904] (7/8) Epoch 5, batch 7500, loss[loss=0.2777, simple_loss=0.3335, pruned_loss=0.1109, over 11104.00 frames. ], tot_loss[loss=0.2548, simple_loss=0.329, pruned_loss=0.09027, over 3036816.89 frames. ], batch size: 248, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:52:51,019 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48133.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:53:08,589 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:53:18,040 INFO [train.py:904] (7/8) Epoch 5, batch 7550, loss[loss=0.203, simple_loss=0.2907, pruned_loss=0.05763, over 16755.00 frames. ], tot_loss[loss=0.2542, simple_loss=0.3282, pruned_loss=0.09013, over 3039439.46 frames. ], batch size: 83, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:53:38,525 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.261e+02 3.568e+02 4.415e+02 5.679e+02 1.356e+03, threshold=8.829e+02, percent-clipped=4.0 2023-04-28 08:53:40,677 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48165.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 08:54:34,588 INFO [train.py:904] (7/8) Epoch 5, batch 7600, loss[loss=0.2305, simple_loss=0.3119, pruned_loss=0.07451, over 16841.00 frames. ], tot_loss[loss=0.2542, simple_loss=0.3274, pruned_loss=0.09053, over 3041828.65 frames. ], batch size: 83, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:54:44,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6865, 4.6848, 5.2811, 5.1031, 5.2097, 4.8042, 4.7498, 4.3825], device='cuda:7'), covar=tensor([0.0258, 0.0277, 0.0268, 0.0460, 0.0341, 0.0234, 0.0757, 0.0389], device='cuda:7'), in_proj_covar=tensor([0.0239, 0.0228, 0.0240, 0.0235, 0.0286, 0.0251, 0.0357, 0.0210], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 08:54:52,559 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48212.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 08:54:52,632 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8042, 2.8821, 2.1620, 4.0597, 3.1821, 3.7728, 1.4472, 2.5940], device='cuda:7'), covar=tensor([0.1093, 0.0449, 0.1166, 0.0081, 0.0242, 0.0366, 0.1274, 0.0770], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0144, 0.0169, 0.0085, 0.0181, 0.0179, 0.0161, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 08:55:05,172 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5811, 3.6356, 4.0510, 3.9665, 3.9925, 3.7008, 3.7176, 3.7006], device='cuda:7'), covar=tensor([0.0302, 0.0394, 0.0318, 0.0422, 0.0435, 0.0324, 0.0799, 0.0422], device='cuda:7'), in_proj_covar=tensor([0.0237, 0.0226, 0.0238, 0.0233, 0.0284, 0.0249, 0.0355, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 08:55:23,912 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 08:55:51,259 INFO [train.py:904] (7/8) Epoch 5, batch 7650, loss[loss=0.2251, simple_loss=0.3093, pruned_loss=0.07049, over 16788.00 frames. ], tot_loss[loss=0.2552, simple_loss=0.3281, pruned_loss=0.09112, over 3047509.22 frames. ], batch size: 83, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 08:56:12,527 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.451e+02 3.891e+02 4.972e+02 6.319e+02 1.654e+03, threshold=9.944e+02, percent-clipped=8.0 2023-04-28 08:56:25,994 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48273.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 08:57:08,651 INFO [train.py:904] (7/8) Epoch 5, batch 7700, loss[loss=0.2736, simple_loss=0.3382, pruned_loss=0.1045, over 16750.00 frames. ], tot_loss[loss=0.2557, simple_loss=0.3281, pruned_loss=0.09165, over 3052013.40 frames. ], batch size: 124, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 08:57:44,747 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48324.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:58:26,518 INFO [train.py:904] (7/8) Epoch 5, batch 7750, loss[loss=0.2404, simple_loss=0.3269, pruned_loss=0.07696, over 16440.00 frames. ], tot_loss[loss=0.255, simple_loss=0.3281, pruned_loss=0.09098, over 3070195.24 frames. ], batch size: 75, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 08:58:35,683 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1213, 3.9042, 4.1626, 4.3946, 4.4751, 4.0022, 4.4373, 4.4623], device='cuda:7'), covar=tensor([0.0933, 0.0747, 0.1199, 0.0508, 0.0438, 0.0844, 0.0502, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0368, 0.0461, 0.0585, 0.0476, 0.0354, 0.0349, 0.0374, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 08:58:47,832 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.847e+02 4.062e+02 4.544e+02 6.017e+02 9.038e+02, threshold=9.088e+02, percent-clipped=0.0 2023-04-28 08:58:48,904 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 08:59:20,732 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48385.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:59:34,677 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48394.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:59:39,597 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48397.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 08:59:45,052 INFO [train.py:904] (7/8) Epoch 5, batch 7800, loss[loss=0.2498, simple_loss=0.3315, pruned_loss=0.08411, over 16506.00 frames. ], tot_loss[loss=0.2565, simple_loss=0.3294, pruned_loss=0.09182, over 3070570.46 frames. ], batch size: 146, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:00:21,126 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48423.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:00:28,782 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48428.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:00:49,709 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48442.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:00:51,209 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-28 09:00:52,195 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48444.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:01:02,298 INFO [train.py:904] (7/8) Epoch 5, batch 7850, loss[loss=0.2304, simple_loss=0.3162, pruned_loss=0.07236, over 16826.00 frames. ], tot_loss[loss=0.256, simple_loss=0.33, pruned_loss=0.09105, over 3086073.84 frames. ], batch size: 102, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:01:13,136 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48458.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:01:23,604 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.309e+02 3.753e+02 4.508e+02 5.540e+02 1.129e+03, threshold=9.017e+02, percent-clipped=6.0 2023-04-28 09:01:24,094 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48465.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 09:01:52,610 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48484.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:02:04,608 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48492.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:02:16,630 INFO [train.py:904] (7/8) Epoch 5, batch 7900, loss[loss=0.2503, simple_loss=0.3413, pruned_loss=0.07963, over 16810.00 frames. ], tot_loss[loss=0.253, simple_loss=0.3281, pruned_loss=0.08893, over 3120279.04 frames. ], batch size: 102, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:02:35,306 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48513.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:02:52,231 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-04-28 09:03:36,615 INFO [train.py:904] (7/8) Epoch 5, batch 7950, loss[loss=0.2977, simple_loss=0.3483, pruned_loss=0.1235, over 11532.00 frames. ], tot_loss[loss=0.2532, simple_loss=0.3279, pruned_loss=0.08925, over 3113342.77 frames. ], batch size: 247, lr: 1.30e-02, grad_scale: 4.0 2023-04-28 09:03:56,964 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 3.717e+02 4.429e+02 5.526e+02 1.268e+03, threshold=8.857e+02, percent-clipped=1.0 2023-04-28 09:04:03,017 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48568.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:04:47,725 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 09:04:50,721 INFO [train.py:904] (7/8) Epoch 5, batch 8000, loss[loss=0.3118, simple_loss=0.3516, pruned_loss=0.136, over 11060.00 frames. ], tot_loss[loss=0.2538, simple_loss=0.3284, pruned_loss=0.08964, over 3112142.49 frames. ], batch size: 248, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:06:06,087 INFO [train.py:904] (7/8) Epoch 5, batch 8050, loss[loss=0.2166, simple_loss=0.3051, pruned_loss=0.06401, over 16750.00 frames. ], tot_loss[loss=0.2529, simple_loss=0.3278, pruned_loss=0.08895, over 3112144.52 frames. ], batch size: 124, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:06:26,966 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.628e+02 3.970e+02 4.648e+02 5.523e+02 1.617e+03, threshold=9.296e+02, percent-clipped=6.0 2023-04-28 09:06:49,720 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48680.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:06:50,247 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 09:07:21,597 INFO [train.py:904] (7/8) Epoch 5, batch 8100, loss[loss=0.2969, simple_loss=0.3447, pruned_loss=0.1245, over 11599.00 frames. ], tot_loss[loss=0.2527, simple_loss=0.3275, pruned_loss=0.0889, over 3098861.25 frames. ], batch size: 247, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:08:05,995 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48728.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:08:40,611 INFO [train.py:904] (7/8) Epoch 5, batch 8150, loss[loss=0.2063, simple_loss=0.2881, pruned_loss=0.06228, over 16990.00 frames. ], tot_loss[loss=0.2504, simple_loss=0.3251, pruned_loss=0.08791, over 3104890.80 frames. ], batch size: 41, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:08:44,613 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48753.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:09:01,953 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.147e+02 3.868e+02 4.692e+02 5.916e+02 1.218e+03, threshold=9.385e+02, percent-clipped=3.0 2023-04-28 09:09:21,416 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48776.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:09:25,127 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48779.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:10:00,531 INFO [train.py:904] (7/8) Epoch 5, batch 8200, loss[loss=0.2233, simple_loss=0.3082, pruned_loss=0.06923, over 16499.00 frames. ], tot_loss[loss=0.2481, simple_loss=0.3226, pruned_loss=0.08685, over 3124113.08 frames. ], batch size: 68, lr: 1.30e-02, grad_scale: 8.0 2023-04-28 09:10:32,480 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48821.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:10:56,037 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2154, 5.5164, 5.2341, 5.2500, 4.8366, 4.7117, 5.0747, 5.5528], device='cuda:7'), covar=tensor([0.0754, 0.0687, 0.0880, 0.0433, 0.0606, 0.0617, 0.0568, 0.0678], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0488, 0.0424, 0.0313, 0.0304, 0.0330, 0.0395, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:11:01,505 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48838.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:11:22,114 INFO [train.py:904] (7/8) Epoch 5, batch 8250, loss[loss=0.243, simple_loss=0.3316, pruned_loss=0.0772, over 16374.00 frames. ], tot_loss[loss=0.246, simple_loss=0.3218, pruned_loss=0.08512, over 3097788.86 frames. ], batch size: 146, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:11:44,538 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.248e+02 3.760e+02 4.474e+02 5.646e+02 1.182e+03, threshold=8.947e+02, percent-clipped=2.0 2023-04-28 09:11:50,783 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48868.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 09:12:13,832 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48882.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:12:41,436 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48899.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:12:43,301 INFO [train.py:904] (7/8) Epoch 5, batch 8300, loss[loss=0.1974, simple_loss=0.2958, pruned_loss=0.04947, over 16465.00 frames. ], tot_loss[loss=0.241, simple_loss=0.3186, pruned_loss=0.08169, over 3074936.04 frames. ], batch size: 75, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:13:08,661 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=48916.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:14:06,486 INFO [train.py:904] (7/8) Epoch 5, batch 8350, loss[loss=0.241, simple_loss=0.3265, pruned_loss=0.07772, over 16218.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.3165, pruned_loss=0.07825, over 3073453.93 frames. ], batch size: 165, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:14:30,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.953e+02 3.108e+02 3.875e+02 4.541e+02 1.583e+03, threshold=7.750e+02, percent-clipped=2.0 2023-04-28 09:14:55,602 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48980.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:15:30,226 INFO [train.py:904] (7/8) Epoch 5, batch 8400, loss[loss=0.2113, simple_loss=0.2927, pruned_loss=0.06489, over 15185.00 frames. ], tot_loss[loss=0.2316, simple_loss=0.3128, pruned_loss=0.07525, over 3064651.98 frames. ], batch size: 191, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:15:40,918 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8843, 4.1820, 3.4747, 2.6574, 3.1330, 2.6850, 4.4083, 4.2610], device='cuda:7'), covar=tensor([0.2079, 0.0495, 0.0952, 0.1405, 0.1730, 0.1188, 0.0289, 0.0442], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0235, 0.0260, 0.0237, 0.0267, 0.0196, 0.0231, 0.0240], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:16:14,936 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49028.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:16:51,012 INFO [train.py:904] (7/8) Epoch 5, batch 8450, loss[loss=0.1812, simple_loss=0.2808, pruned_loss=0.04085, over 16773.00 frames. ], tot_loss[loss=0.2287, simple_loss=0.3106, pruned_loss=0.07343, over 3056838.79 frames. ], batch size: 76, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:16:55,268 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49053.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:17:09,898 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0320, 1.3426, 1.6916, 2.1000, 2.1072, 2.2406, 1.5971, 2.0960], device='cuda:7'), covar=tensor([0.0082, 0.0247, 0.0146, 0.0142, 0.0120, 0.0097, 0.0203, 0.0066], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0143, 0.0128, 0.0124, 0.0130, 0.0092, 0.0140, 0.0079], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 09:17:13,780 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.140e+02 2.885e+02 3.477e+02 4.145e+02 8.809e+02, threshold=6.955e+02, percent-clipped=1.0 2023-04-28 09:17:38,514 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49079.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:18:13,187 INFO [train.py:904] (7/8) Epoch 5, batch 8500, loss[loss=0.2003, simple_loss=0.2728, pruned_loss=0.06389, over 11677.00 frames. ], tot_loss[loss=0.2228, simple_loss=0.3054, pruned_loss=0.07013, over 3057860.68 frames. ], batch size: 248, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:18:13,737 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49101.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:18:57,905 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49127.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:19:07,570 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2905, 4.6437, 4.3400, 4.3959, 3.9769, 3.9927, 4.2583, 4.6175], device='cuda:7'), covar=tensor([0.0727, 0.0845, 0.0989, 0.0521, 0.0599, 0.1059, 0.0646, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0351, 0.0466, 0.0403, 0.0301, 0.0291, 0.0312, 0.0377, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:19:38,549 INFO [train.py:904] (7/8) Epoch 5, batch 8550, loss[loss=0.21, simple_loss=0.3036, pruned_loss=0.05817, over 16684.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.303, pruned_loss=0.06864, over 3039988.66 frames. ], batch size: 89, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:20:04,109 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.233e+02 3.773e+02 4.691e+02 1.040e+03, threshold=7.547e+02, percent-clipped=6.0 2023-04-28 09:20:29,798 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49177.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:20:33,024 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-04-28 09:21:04,529 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49194.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:21:17,327 INFO [train.py:904] (7/8) Epoch 5, batch 8600, loss[loss=0.1951, simple_loss=0.2871, pruned_loss=0.05157, over 16570.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.3034, pruned_loss=0.06764, over 3043590.54 frames. ], batch size: 62, lr: 1.29e-02, grad_scale: 8.0 2023-04-28 09:22:29,815 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49236.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:22:58,657 INFO [train.py:904] (7/8) Epoch 5, batch 8650, loss[loss=0.1845, simple_loss=0.2827, pruned_loss=0.04317, over 16792.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.3013, pruned_loss=0.06558, over 3065587.91 frames. ], batch size: 124, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:23:33,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.003e+02 3.105e+02 3.609e+02 4.531e+02 9.141e+02, threshold=7.218e+02, percent-clipped=3.0 2023-04-28 09:24:39,635 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49297.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:24:47,024 INFO [train.py:904] (7/8) Epoch 5, batch 8700, loss[loss=0.179, simple_loss=0.2684, pruned_loss=0.04486, over 17014.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2971, pruned_loss=0.0635, over 3065926.16 frames. ], batch size: 55, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:25:01,070 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7401, 2.7580, 1.6923, 2.8060, 2.0772, 2.8031, 1.9201, 2.4905], device='cuda:7'), covar=tensor([0.0153, 0.0305, 0.1243, 0.0082, 0.0634, 0.0384, 0.1113, 0.0516], device='cuda:7'), in_proj_covar=tensor([0.0115, 0.0145, 0.0174, 0.0075, 0.0156, 0.0170, 0.0182, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 09:25:36,254 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3356, 1.3985, 1.7667, 2.3245, 2.2818, 2.4861, 1.5968, 2.3779], device='cuda:7'), covar=tensor([0.0097, 0.0243, 0.0163, 0.0145, 0.0120, 0.0097, 0.0219, 0.0070], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0144, 0.0130, 0.0125, 0.0132, 0.0093, 0.0143, 0.0079], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 09:26:24,156 INFO [train.py:904] (7/8) Epoch 5, batch 8750, loss[loss=0.2361, simple_loss=0.3234, pruned_loss=0.07438, over 16163.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2962, pruned_loss=0.06266, over 3058956.12 frames. ], batch size: 165, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:26:37,246 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3600, 4.6549, 4.3967, 4.4186, 4.0965, 4.0724, 4.2234, 4.6283], device='cuda:7'), covar=tensor([0.0633, 0.0599, 0.0794, 0.0426, 0.0536, 0.1104, 0.0600, 0.0661], device='cuda:7'), in_proj_covar=tensor([0.0342, 0.0451, 0.0385, 0.0294, 0.0284, 0.0306, 0.0369, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:27:05,490 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.935e+02 3.673e+02 5.022e+02 7.467e+02, threshold=7.345e+02, percent-clipped=1.0 2023-04-28 09:27:31,949 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-28 09:28:17,604 INFO [train.py:904] (7/8) Epoch 5, batch 8800, loss[loss=0.2501, simple_loss=0.332, pruned_loss=0.08408, over 15376.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2942, pruned_loss=0.06147, over 3045634.83 frames. ], batch size: 190, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:28:33,923 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3480, 4.5109, 4.5121, 4.4525, 4.3983, 4.9604, 4.6218, 4.3760], device='cuda:7'), covar=tensor([0.1067, 0.1347, 0.1323, 0.1718, 0.2512, 0.0861, 0.1144, 0.2335], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0359, 0.0359, 0.0310, 0.0416, 0.0388, 0.0300, 0.0427], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:29:02,429 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:29:20,596 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1507, 3.0013, 2.6762, 1.9251, 2.5683, 2.0404, 2.6751, 2.6958], device='cuda:7'), covar=tensor([0.0289, 0.0487, 0.0449, 0.1352, 0.0612, 0.0926, 0.0603, 0.0719], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0116, 0.0148, 0.0137, 0.0129, 0.0125, 0.0134, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 09:29:24,470 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49432.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:29:24,839 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-28 09:30:04,538 INFO [train.py:904] (7/8) Epoch 5, batch 8850, loss[loss=0.2101, simple_loss=0.3053, pruned_loss=0.05743, over 16300.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2969, pruned_loss=0.06076, over 3054359.37 frames. ], batch size: 146, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:30:09,322 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6295, 3.4461, 3.0654, 1.7405, 2.5867, 2.0200, 3.0558, 3.1151], device='cuda:7'), covar=tensor([0.0309, 0.0492, 0.0521, 0.1639, 0.0768, 0.1056, 0.0746, 0.0885], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0117, 0.0150, 0.0139, 0.0131, 0.0127, 0.0136, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 09:30:38,666 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 3.290e+02 4.056e+02 5.025e+02 1.173e+03, threshold=8.111e+02, percent-clipped=7.0 2023-04-28 09:31:02,591 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49477.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:31:16,950 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49483.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:31:24,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7438, 4.0200, 3.1336, 2.3662, 2.8414, 2.3653, 4.3508, 3.8826], device='cuda:7'), covar=tensor([0.2291, 0.0569, 0.1197, 0.1595, 0.1795, 0.1439, 0.0308, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0274, 0.0240, 0.0262, 0.0237, 0.0250, 0.0198, 0.0235, 0.0237], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:31:37,619 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49493.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 09:31:39,814 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49494.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:31:54,245 INFO [train.py:904] (7/8) Epoch 5, batch 8900, loss[loss=0.1836, simple_loss=0.2764, pruned_loss=0.04535, over 16836.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2966, pruned_loss=0.05994, over 3043084.66 frames. ], batch size: 90, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:32:48,442 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49525.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:33:39,643 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49542.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:33:59,799 INFO [train.py:904] (7/8) Epoch 5, batch 8950, loss[loss=0.2036, simple_loss=0.289, pruned_loss=0.05903, over 16809.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.297, pruned_loss=0.06031, over 3078591.95 frames. ], batch size: 124, lr: 1.29e-02, grad_scale: 4.0 2023-04-28 09:34:33,545 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 09:34:35,634 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 2.796e+02 3.504e+02 4.655e+02 7.441e+02, threshold=7.007e+02, percent-clipped=0.0 2023-04-28 09:35:33,162 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49592.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:35:51,815 INFO [train.py:904] (7/8) Epoch 5, batch 9000, loss[loss=0.2224, simple_loss=0.2967, pruned_loss=0.07405, over 12136.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2935, pruned_loss=0.05851, over 3084454.13 frames. ], batch size: 250, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:35:51,816 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 09:36:02,102 INFO [train.py:938] (7/8) Epoch 5, validation: loss=0.1735, simple_loss=0.2766, pruned_loss=0.0352, over 944034.00 frames. 2023-04-28 09:36:02,103 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 09:36:03,730 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 09:36:11,126 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 09:36:33,693 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 09:37:35,937 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2157, 3.0493, 2.7646, 1.7915, 2.5039, 2.1255, 2.7620, 2.8004], device='cuda:7'), covar=tensor([0.0289, 0.0468, 0.0419, 0.1443, 0.0659, 0.0844, 0.0622, 0.0579], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0116, 0.0149, 0.0139, 0.0130, 0.0126, 0.0135, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 09:37:38,287 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 09:37:44,743 INFO [train.py:904] (7/8) Epoch 5, batch 9050, loss[loss=0.1771, simple_loss=0.2665, pruned_loss=0.04387, over 16905.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2951, pruned_loss=0.05956, over 3081775.79 frames. ], batch size: 102, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:38:18,681 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 3.141e+02 3.840e+02 5.023e+02 8.628e+02, threshold=7.679e+02, percent-clipped=5.0 2023-04-28 09:39:29,265 INFO [train.py:904] (7/8) Epoch 5, batch 9100, loss[loss=0.2089, simple_loss=0.302, pruned_loss=0.05787, over 16624.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2952, pruned_loss=0.0605, over 3088498.44 frames. ], batch size: 62, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:40:55,299 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4198, 4.3886, 4.2299, 4.0566, 3.8525, 4.3309, 4.1669, 4.0005], device='cuda:7'), covar=tensor([0.0388, 0.0335, 0.0218, 0.0182, 0.0715, 0.0295, 0.0393, 0.0522], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0174, 0.0198, 0.0169, 0.0217, 0.0199, 0.0140, 0.0224], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:40:59,831 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8780, 2.7078, 2.6178, 1.7850, 2.8269, 2.8292, 2.4659, 2.3780], device='cuda:7'), covar=tensor([0.0624, 0.0132, 0.0117, 0.0870, 0.0070, 0.0083, 0.0331, 0.0401], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0085, 0.0075, 0.0137, 0.0066, 0.0073, 0.0112, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 09:41:26,344 INFO [train.py:904] (7/8) Epoch 5, batch 9150, loss[loss=0.176, simple_loss=0.2706, pruned_loss=0.04067, over 16919.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2953, pruned_loss=0.05973, over 3090550.65 frames. ], batch size: 96, lr: 1.28e-02, grad_scale: 4.0 2023-04-28 09:42:00,770 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49766.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:42:01,430 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 3.040e+02 3.844e+02 4.947e+02 8.229e+02, threshold=7.688e+02, percent-clipped=4.0 2023-04-28 09:42:25,695 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49778.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:42:47,440 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49788.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 09:42:55,585 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8987, 3.6226, 3.5736, 4.0083, 4.1404, 3.6788, 4.1209, 4.1676], device='cuda:7'), covar=tensor([0.0873, 0.0951, 0.1981, 0.0958, 0.0747, 0.1364, 0.0913, 0.0788], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0433, 0.0547, 0.0448, 0.0340, 0.0334, 0.0358, 0.0370], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:43:09,766 INFO [train.py:904] (7/8) Epoch 5, batch 9200, loss[loss=0.1877, simple_loss=0.263, pruned_loss=0.0562, over 12232.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2913, pruned_loss=0.05904, over 3101298.52 frames. ], batch size: 247, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:43:23,493 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3201, 1.4532, 1.7088, 2.3085, 2.2562, 2.3478, 1.4692, 2.2084], device='cuda:7'), covar=tensor([0.0089, 0.0235, 0.0156, 0.0123, 0.0119, 0.0102, 0.0242, 0.0057], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0140, 0.0127, 0.0122, 0.0127, 0.0088, 0.0139, 0.0076], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 09:43:58,036 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49827.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:44:46,695 INFO [train.py:904] (7/8) Epoch 5, batch 9250, loss[loss=0.189, simple_loss=0.2689, pruned_loss=0.05456, over 12260.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2911, pruned_loss=0.0592, over 3089748.09 frames. ], batch size: 248, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:45:18,930 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.407e+02 3.305e+02 4.013e+02 4.841e+02 9.780e+02, threshold=8.027e+02, percent-clipped=1.0 2023-04-28 09:46:21,471 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49892.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:46:40,063 INFO [train.py:904] (7/8) Epoch 5, batch 9300, loss[loss=0.1894, simple_loss=0.2768, pruned_loss=0.05101, over 15298.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2893, pruned_loss=0.05841, over 3083841.00 frames. ], batch size: 190, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:47:17,342 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8262, 2.0659, 2.2969, 4.2347, 1.8249, 2.7903, 2.1946, 2.1571], device='cuda:7'), covar=tensor([0.0531, 0.2615, 0.1362, 0.0356, 0.4116, 0.1464, 0.2415, 0.3185], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0308, 0.0260, 0.0303, 0.0367, 0.0308, 0.0288, 0.0366], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:48:05,870 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=49940.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:48:05,938 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6377, 5.0123, 4.7923, 4.8210, 4.4319, 4.3921, 4.5729, 5.0705], device='cuda:7'), covar=tensor([0.0610, 0.0607, 0.0730, 0.0381, 0.0534, 0.0760, 0.0565, 0.0619], device='cuda:7'), in_proj_covar=tensor([0.0346, 0.0464, 0.0383, 0.0301, 0.0290, 0.0305, 0.0375, 0.0337], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:48:26,428 INFO [train.py:904] (7/8) Epoch 5, batch 9350, loss[loss=0.1927, simple_loss=0.2721, pruned_loss=0.05668, over 12594.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2888, pruned_loss=0.05832, over 3071802.73 frames. ], batch size: 248, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:49:00,530 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 3.133e+02 3.801e+02 4.449e+02 7.989e+02, threshold=7.602e+02, percent-clipped=0.0 2023-04-28 09:50:11,392 INFO [train.py:904] (7/8) Epoch 5, batch 9400, loss[loss=0.1994, simple_loss=0.2781, pruned_loss=0.06039, over 12240.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2882, pruned_loss=0.05799, over 3058271.06 frames. ], batch size: 248, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:51:14,941 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9473, 2.6650, 2.6385, 1.9186, 2.4838, 2.6892, 2.5702, 1.8825], device='cuda:7'), covar=tensor([0.0221, 0.0025, 0.0034, 0.0190, 0.0051, 0.0039, 0.0038, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0052, 0.0057, 0.0112, 0.0057, 0.0063, 0.0060, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 09:51:51,667 INFO [train.py:904] (7/8) Epoch 5, batch 9450, loss[loss=0.1921, simple_loss=0.2818, pruned_loss=0.05113, over 16953.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2905, pruned_loss=0.05849, over 3066682.06 frames. ], batch size: 109, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:52:01,924 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9333, 2.6298, 2.6488, 1.8927, 2.4928, 2.6250, 2.5393, 1.8250], device='cuda:7'), covar=tensor([0.0228, 0.0026, 0.0035, 0.0201, 0.0062, 0.0045, 0.0044, 0.0264], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0052, 0.0056, 0.0111, 0.0056, 0.0063, 0.0060, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 09:52:21,841 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.955e+02 3.683e+02 5.023e+02 1.227e+03, threshold=7.366e+02, percent-clipped=5.0 2023-04-28 09:52:35,548 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 09:52:45,878 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2614, 1.9510, 2.1652, 3.7192, 1.7964, 2.7574, 2.2129, 2.0620], device='cuda:7'), covar=tensor([0.0585, 0.2231, 0.1241, 0.0351, 0.3229, 0.1191, 0.2067, 0.2555], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0303, 0.0257, 0.0301, 0.0360, 0.0303, 0.0283, 0.0359], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:52:47,406 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:53:09,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50088.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 09:53:33,296 INFO [train.py:904] (7/8) Epoch 5, batch 9500, loss[loss=0.1629, simple_loss=0.2535, pruned_loss=0.03614, over 17050.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2897, pruned_loss=0.05754, over 3076797.49 frames. ], batch size: 55, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:53:59,182 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 09:54:18,376 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50122.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:54:25,651 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50126.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:54:29,417 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50128.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:54:44,418 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50136.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:55:18,512 INFO [train.py:904] (7/8) Epoch 5, batch 9550, loss[loss=0.2001, simple_loss=0.2859, pruned_loss=0.0572, over 12998.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2889, pruned_loss=0.0576, over 3058315.84 frames. ], batch size: 250, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:55:53,350 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.772e+02 3.459e+02 4.281e+02 6.687e+02, threshold=6.919e+02, percent-clipped=0.0 2023-04-28 09:56:30,771 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3347, 4.6540, 4.4641, 4.4326, 4.0783, 4.0837, 4.2251, 4.6501], device='cuda:7'), covar=tensor([0.0682, 0.0712, 0.0679, 0.0438, 0.0579, 0.1090, 0.0664, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0346, 0.0462, 0.0380, 0.0300, 0.0290, 0.0305, 0.0374, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:56:38,321 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50189.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 09:56:52,979 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50197.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 09:56:59,046 INFO [train.py:904] (7/8) Epoch 5, batch 9600, loss[loss=0.2348, simple_loss=0.321, pruned_loss=0.07428, over 15416.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2919, pruned_loss=0.05935, over 3060198.14 frames. ], batch size: 191, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:58:47,434 INFO [train.py:904] (7/8) Epoch 5, batch 9650, loss[loss=0.2011, simple_loss=0.2941, pruned_loss=0.05405, over 15449.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2938, pruned_loss=0.05995, over 3063887.72 frames. ], batch size: 191, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 09:59:05,756 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50258.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 09:59:27,461 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.258e+02 2.991e+02 3.657e+02 4.626e+02 9.582e+02, threshold=7.315e+02, percent-clipped=7.0 2023-04-28 09:59:29,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8178, 4.2081, 3.9895, 4.0457, 3.6253, 3.7383, 3.8621, 4.1338], device='cuda:7'), covar=tensor([0.0879, 0.0900, 0.0951, 0.0531, 0.0677, 0.1395, 0.0710, 0.1017], device='cuda:7'), in_proj_covar=tensor([0.0348, 0.0461, 0.0382, 0.0301, 0.0292, 0.0304, 0.0373, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 09:59:59,304 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-28 10:00:35,791 INFO [train.py:904] (7/8) Epoch 5, batch 9700, loss[loss=0.1936, simple_loss=0.2831, pruned_loss=0.05203, over 16145.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2923, pruned_loss=0.05942, over 3062061.46 frames. ], batch size: 165, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 10:02:18,561 INFO [train.py:904] (7/8) Epoch 5, batch 9750, loss[loss=0.204, simple_loss=0.2899, pruned_loss=0.05902, over 16966.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2912, pruned_loss=0.0596, over 3051849.46 frames. ], batch size: 109, lr: 1.28e-02, grad_scale: 8.0 2023-04-28 10:02:50,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.871e+02 3.120e+02 3.818e+02 4.621e+02 8.897e+02, threshold=7.636e+02, percent-clipped=1.0 2023-04-28 10:03:56,537 INFO [train.py:904] (7/8) Epoch 5, batch 9800, loss[loss=0.2307, simple_loss=0.325, pruned_loss=0.06825, over 15366.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2912, pruned_loss=0.05827, over 3072913.94 frames. ], batch size: 191, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:04:36,914 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:05:41,985 INFO [train.py:904] (7/8) Epoch 5, batch 9850, loss[loss=0.1937, simple_loss=0.2868, pruned_loss=0.0503, over 16769.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2919, pruned_loss=0.0577, over 3059182.35 frames. ], batch size: 83, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:05:53,299 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8397, 3.7657, 4.3199, 4.2896, 4.2751, 3.9943, 4.0066, 3.9178], device='cuda:7'), covar=tensor([0.0261, 0.0500, 0.0322, 0.0352, 0.0391, 0.0266, 0.0673, 0.0337], device='cuda:7'), in_proj_covar=tensor([0.0211, 0.0211, 0.0212, 0.0213, 0.0256, 0.0226, 0.0312, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-04-28 10:06:14,681 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.952e+02 3.639e+02 4.340e+02 9.232e+02, threshold=7.278e+02, percent-clipped=1.0 2023-04-28 10:06:21,891 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50470.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:06:23,607 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5282, 3.4573, 3.4453, 3.1099, 3.3820, 1.9755, 3.2107, 2.9702], device='cuda:7'), covar=tensor([0.0073, 0.0068, 0.0088, 0.0156, 0.0069, 0.1534, 0.0084, 0.0140], device='cuda:7'), in_proj_covar=tensor([0.0088, 0.0076, 0.0117, 0.0109, 0.0087, 0.0143, 0.0100, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:06:52,950 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50484.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:07:31,483 INFO [train.py:904] (7/8) Epoch 5, batch 9900, loss[loss=0.2119, simple_loss=0.3081, pruned_loss=0.05782, over 16298.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2927, pruned_loss=0.05741, over 3070223.76 frames. ], batch size: 165, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:07:40,408 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-28 10:09:17,081 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50546.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:09:27,679 INFO [train.py:904] (7/8) Epoch 5, batch 9950, loss[loss=0.1963, simple_loss=0.2923, pruned_loss=0.05012, over 15369.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2949, pruned_loss=0.05773, over 3081072.51 frames. ], batch size: 190, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:09:33,886 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50553.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:10:04,531 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.956e+02 3.001e+02 3.652e+02 4.906e+02 2.314e+03, threshold=7.303e+02, percent-clipped=6.0 2023-04-28 10:11:17,395 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 10:11:29,341 INFO [train.py:904] (7/8) Epoch 5, batch 10000, loss[loss=0.206, simple_loss=0.2943, pruned_loss=0.0588, over 15523.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2931, pruned_loss=0.05712, over 3094738.96 frames. ], batch size: 191, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:11:44,459 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50607.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:12:28,824 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2023-04-28 10:13:08,533 INFO [train.py:904] (7/8) Epoch 5, batch 10050, loss[loss=0.2289, simple_loss=0.3195, pruned_loss=0.06913, over 16338.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2927, pruned_loss=0.05669, over 3100880.99 frames. ], batch size: 146, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:13:17,220 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 10:13:38,882 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.547e+02 3.217e+02 3.917e+02 1.118e+03, threshold=6.434e+02, percent-clipped=1.0 2023-04-28 10:14:39,003 INFO [train.py:904] (7/8) Epoch 5, batch 10100, loss[loss=0.224, simple_loss=0.3101, pruned_loss=0.06896, over 15432.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.293, pruned_loss=0.0571, over 3093168.63 frames. ], batch size: 192, lr: 1.27e-02, grad_scale: 8.0 2023-04-28 10:15:38,051 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5607, 2.7091, 2.3694, 4.2091, 3.6428, 3.9799, 1.4911, 2.8886], device='cuda:7'), covar=tensor([0.1554, 0.0656, 0.1264, 0.0100, 0.0220, 0.0397, 0.1576, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0140, 0.0165, 0.0084, 0.0147, 0.0172, 0.0160, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 10:16:20,276 INFO [train.py:904] (7/8) Epoch 6, batch 0, loss[loss=0.2257, simple_loss=0.2982, pruned_loss=0.0766, over 17221.00 frames. ], tot_loss[loss=0.2257, simple_loss=0.2982, pruned_loss=0.0766, over 17221.00 frames. ], batch size: 46, lr: 1.19e-02, grad_scale: 8.0 2023-04-28 10:16:20,276 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 10:16:27,647 INFO [train.py:938] (7/8) Epoch 6, validation: loss=0.1727, simple_loss=0.2755, pruned_loss=0.03501, over 944034.00 frames. 2023-04-28 10:16:27,648 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 10:16:52,400 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 3.546e+02 4.478e+02 5.747e+02 1.222e+03, threshold=8.956e+02, percent-clipped=19.0 2023-04-28 10:17:10,736 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50782.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:17:13,842 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50784.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:17:35,098 INFO [train.py:904] (7/8) Epoch 6, batch 50, loss[loss=0.1911, simple_loss=0.2791, pruned_loss=0.05156, over 17127.00 frames. ], tot_loss[loss=0.2454, simple_loss=0.312, pruned_loss=0.08945, over 755643.26 frames. ], batch size: 47, lr: 1.19e-02, grad_scale: 2.0 2023-04-28 10:17:45,463 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3800, 3.9393, 3.0515, 1.8200, 2.6535, 2.1770, 3.8160, 3.7072], device='cuda:7'), covar=tensor([0.0229, 0.0381, 0.0662, 0.1708, 0.0801, 0.1088, 0.0449, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0116, 0.0151, 0.0139, 0.0129, 0.0127, 0.0136, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 10:17:47,202 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50808.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:18:20,194 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50832.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:18:34,787 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50843.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:18:47,055 INFO [train.py:904] (7/8) Epoch 6, batch 100, loss[loss=0.1886, simple_loss=0.2632, pruned_loss=0.05699, over 16954.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.3042, pruned_loss=0.08038, over 1323150.35 frames. ], batch size: 41, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:18:49,584 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50853.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:19:11,792 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.233e+02 3.517e+02 4.205e+02 5.493e+02 1.012e+03, threshold=8.410e+02, percent-clipped=3.0 2023-04-28 10:19:12,320 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50869.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 10:19:16,715 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-28 10:19:25,054 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1242, 4.3709, 2.0208, 4.8613, 3.1273, 4.7172, 2.3491, 3.3284], device='cuda:7'), covar=tensor([0.0113, 0.0231, 0.1469, 0.0036, 0.0607, 0.0283, 0.1267, 0.0460], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0151, 0.0176, 0.0078, 0.0158, 0.0176, 0.0185, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 10:19:54,858 INFO [train.py:904] (7/8) Epoch 6, batch 150, loss[loss=0.2306, simple_loss=0.3002, pruned_loss=0.08045, over 16858.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3021, pruned_loss=0.07906, over 1767492.50 frames. ], batch size: 96, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:19:55,167 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=50901.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:19:56,193 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50902.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:19:56,801 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-04-28 10:20:57,640 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50946.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:21:05,282 INFO [train.py:904] (7/8) Epoch 6, batch 200, loss[loss=0.2274, simple_loss=0.2931, pruned_loss=0.0809, over 16239.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.3024, pruned_loss=0.07807, over 2097883.12 frames. ], batch size: 165, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:21:28,626 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.434e+02 3.943e+02 4.796e+02 1.132e+03, threshold=7.886e+02, percent-clipped=3.0 2023-04-28 10:22:12,678 INFO [train.py:904] (7/8) Epoch 6, batch 250, loss[loss=0.2225, simple_loss=0.3064, pruned_loss=0.06934, over 16647.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.2978, pruned_loss=0.07537, over 2367159.11 frames. ], batch size: 57, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:22:21,418 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51007.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:22:24,988 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-04-28 10:22:35,892 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4701, 1.9926, 2.7234, 3.2518, 3.1504, 3.4355, 2.2032, 3.5116], device='cuda:7'), covar=tensor([0.0061, 0.0209, 0.0125, 0.0098, 0.0092, 0.0113, 0.0195, 0.0052], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0146, 0.0131, 0.0127, 0.0133, 0.0095, 0.0144, 0.0080], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 10:22:37,411 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-28 10:23:00,017 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 10:23:20,598 INFO [train.py:904] (7/8) Epoch 6, batch 300, loss[loss=0.1921, simple_loss=0.2826, pruned_loss=0.05082, over 17047.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.2925, pruned_loss=0.07254, over 2575435.89 frames. ], batch size: 50, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:23:37,419 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51062.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:23:45,704 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.797e+02 3.691e+02 4.478e+02 8.058e+02, threshold=7.381e+02, percent-clipped=1.0 2023-04-28 10:24:30,422 INFO [train.py:904] (7/8) Epoch 6, batch 350, loss[loss=0.1537, simple_loss=0.2364, pruned_loss=0.03548, over 16865.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2904, pruned_loss=0.07119, over 2737802.83 frames. ], batch size: 42, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:25:01,840 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51123.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:25:19,516 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51138.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:25:37,100 INFO [train.py:904] (7/8) Epoch 6, batch 400, loss[loss=0.2142, simple_loss=0.2802, pruned_loss=0.07405, over 12519.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.289, pruned_loss=0.07128, over 2866979.73 frames. ], batch size: 246, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:25:55,423 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51164.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:26:01,706 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.109e+02 2.940e+02 3.610e+02 4.239e+02 7.005e+02, threshold=7.220e+02, percent-clipped=1.0 2023-04-28 10:26:02,600 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-28 10:26:05,714 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51172.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:26:21,447 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9068, 4.2374, 3.2104, 2.3289, 2.9433, 2.4021, 4.3636, 4.1731], device='cuda:7'), covar=tensor([0.2161, 0.0541, 0.1245, 0.1835, 0.2317, 0.1570, 0.0364, 0.0709], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0250, 0.0271, 0.0246, 0.0271, 0.0203, 0.0244, 0.0256], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:26:45,055 INFO [train.py:904] (7/8) Epoch 6, batch 450, loss[loss=0.1843, simple_loss=0.2651, pruned_loss=0.05174, over 17186.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2876, pruned_loss=0.0696, over 2966792.84 frames. ], batch size: 44, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:26:47,135 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51202.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:27:29,628 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51233.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:27:52,970 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51250.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:27:53,840 INFO [train.py:904] (7/8) Epoch 6, batch 500, loss[loss=0.2006, simple_loss=0.2723, pruned_loss=0.06443, over 16760.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2861, pruned_loss=0.06872, over 3042751.23 frames. ], batch size: 83, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:28:16,034 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-04-28 10:28:17,363 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 3.090e+02 3.738e+02 4.449e+02 8.454e+02, threshold=7.475e+02, percent-clipped=2.0 2023-04-28 10:29:01,772 INFO [train.py:904] (7/8) Epoch 6, batch 550, loss[loss=0.2195, simple_loss=0.2919, pruned_loss=0.07356, over 15549.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2854, pruned_loss=0.06763, over 3105470.25 frames. ], batch size: 191, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:29:03,901 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51302.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:30:13,399 INFO [train.py:904] (7/8) Epoch 6, batch 600, loss[loss=0.2428, simple_loss=0.3023, pruned_loss=0.09167, over 16723.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2846, pruned_loss=0.06838, over 3154054.94 frames. ], batch size: 124, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:30:38,610 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.906e+02 3.488e+02 4.318e+02 1.008e+03, threshold=6.975e+02, percent-clipped=1.0 2023-04-28 10:30:55,727 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51382.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:31:11,016 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51392.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:31:21,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7150, 4.8840, 5.3118, 5.3525, 5.3734, 4.9619, 4.7058, 4.7039], device='cuda:7'), covar=tensor([0.0386, 0.0333, 0.0419, 0.0479, 0.0443, 0.0333, 0.1162, 0.0398], device='cuda:7'), in_proj_covar=tensor([0.0254, 0.0255, 0.0254, 0.0249, 0.0302, 0.0268, 0.0377, 0.0224], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 10:31:23,797 INFO [train.py:904] (7/8) Epoch 6, batch 650, loss[loss=0.2176, simple_loss=0.2817, pruned_loss=0.07675, over 12200.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2839, pruned_loss=0.06731, over 3193815.25 frames. ], batch size: 247, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:31:47,334 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51418.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:04,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3210, 3.2671, 3.3478, 3.4862, 3.5584, 3.2849, 3.4118, 3.5635], device='cuda:7'), covar=tensor([0.0811, 0.0688, 0.0986, 0.0475, 0.0483, 0.1720, 0.1032, 0.0549], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0526, 0.0671, 0.0533, 0.0397, 0.0391, 0.0421, 0.0454], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:32:14,628 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51438.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:21,294 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51443.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:31,063 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0188, 1.6253, 2.4877, 2.8552, 2.7337, 3.0309, 1.7269, 3.0985], device='cuda:7'), covar=tensor([0.0077, 0.0224, 0.0131, 0.0120, 0.0098, 0.0091, 0.0217, 0.0064], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0145, 0.0131, 0.0128, 0.0131, 0.0095, 0.0144, 0.0080], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 10:32:31,572 INFO [train.py:904] (7/8) Epoch 6, batch 700, loss[loss=0.2314, simple_loss=0.3074, pruned_loss=0.07765, over 16756.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2846, pruned_loss=0.06689, over 3218580.27 frames. ], batch size: 62, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:32:35,179 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51453.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:32:49,344 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51464.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 10:32:57,148 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.941e+02 3.067e+02 3.781e+02 4.816e+02 1.338e+03, threshold=7.562e+02, percent-clipped=6.0 2023-04-28 10:33:19,514 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51486.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:33:33,249 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51495.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:33:39,731 INFO [train.py:904] (7/8) Epoch 6, batch 750, loss[loss=0.2474, simple_loss=0.3065, pruned_loss=0.09412, over 16765.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2845, pruned_loss=0.06654, over 3249147.20 frames. ], batch size: 124, lr: 1.18e-02, grad_scale: 2.0 2023-04-28 10:33:56,286 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51512.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:34:19,282 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51528.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 10:34:44,232 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6650, 4.1265, 4.4469, 1.8577, 4.6614, 4.5930, 3.2934, 3.4721], device='cuda:7'), covar=tensor([0.0727, 0.0110, 0.0107, 0.1126, 0.0047, 0.0072, 0.0302, 0.0334], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0091, 0.0081, 0.0139, 0.0071, 0.0080, 0.0115, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 10:34:53,043 INFO [train.py:904] (7/8) Epoch 6, batch 800, loss[loss=0.2245, simple_loss=0.2918, pruned_loss=0.07854, over 15572.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2839, pruned_loss=0.06606, over 3254544.46 frames. ], batch size: 190, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:34:56,359 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8019, 4.2247, 4.5064, 1.9725, 4.7889, 4.7124, 3.3530, 3.5902], device='cuda:7'), covar=tensor([0.0744, 0.0130, 0.0139, 0.1088, 0.0046, 0.0075, 0.0328, 0.0335], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0092, 0.0081, 0.0139, 0.0071, 0.0080, 0.0115, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 10:35:00,632 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51556.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:35:19,940 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.998e+02 2.921e+02 3.386e+02 4.262e+02 8.289e+02, threshold=6.772e+02, percent-clipped=2.0 2023-04-28 10:36:01,915 INFO [train.py:904] (7/8) Epoch 6, batch 850, loss[loss=0.1866, simple_loss=0.2703, pruned_loss=0.05145, over 17227.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2826, pruned_loss=0.06544, over 3275219.77 frames. ], batch size: 44, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:36:04,121 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51602.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:37:10,641 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51650.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:37:11,539 INFO [train.py:904] (7/8) Epoch 6, batch 900, loss[loss=0.186, simple_loss=0.258, pruned_loss=0.05705, over 16482.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2812, pruned_loss=0.064, over 3290142.45 frames. ], batch size: 68, lr: 1.18e-02, grad_scale: 4.0 2023-04-28 10:37:39,522 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.821e+02 2.905e+02 3.526e+02 4.406e+02 8.198e+02, threshold=7.052e+02, percent-clipped=7.0 2023-04-28 10:37:54,452 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3573, 2.2414, 2.4105, 4.8381, 2.0040, 3.3339, 2.5550, 2.5249], device='cuda:7'), covar=tensor([0.0448, 0.2338, 0.1183, 0.0232, 0.2980, 0.1256, 0.1881, 0.2683], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0327, 0.0272, 0.0317, 0.0373, 0.0340, 0.0299, 0.0397], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:38:22,375 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.11 vs. limit=5.0 2023-04-28 10:38:22,600 INFO [train.py:904] (7/8) Epoch 6, batch 950, loss[loss=0.1997, simple_loss=0.2829, pruned_loss=0.05826, over 17196.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2812, pruned_loss=0.06385, over 3295908.14 frames. ], batch size: 44, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:38:46,229 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51718.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:02,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 10:39:12,997 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51738.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:25,711 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51747.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:26,802 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:30,786 INFO [train.py:904] (7/8) Epoch 6, batch 1000, loss[loss=0.2015, simple_loss=0.2785, pruned_loss=0.06229, over 17196.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2803, pruned_loss=0.06388, over 3297317.08 frames. ], batch size: 46, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:39:51,035 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51766.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:39:57,331 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 2.827e+02 3.297e+02 4.150e+02 8.828e+02, threshold=6.593e+02, percent-clipped=4.0 2023-04-28 10:40:39,947 INFO [train.py:904] (7/8) Epoch 6, batch 1050, loss[loss=0.2043, simple_loss=0.2696, pruned_loss=0.06949, over 16446.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2805, pruned_loss=0.064, over 3305869.52 frames. ], batch size: 146, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:40:49,415 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51808.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:41:17,944 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51828.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:41:45,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3134, 5.1869, 5.0927, 4.8135, 4.6086, 5.0945, 5.1295, 4.7736], device='cuda:7'), covar=tensor([0.0355, 0.0307, 0.0183, 0.0189, 0.0908, 0.0302, 0.0234, 0.0495], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0216, 0.0236, 0.0206, 0.0269, 0.0244, 0.0169, 0.0272], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 10:41:49,453 INFO [train.py:904] (7/8) Epoch 6, batch 1100, loss[loss=0.1669, simple_loss=0.2423, pruned_loss=0.04575, over 16954.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2801, pruned_loss=0.06349, over 3318425.38 frames. ], batch size: 41, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:41:49,776 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51851.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:42:16,638 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.702e+02 3.374e+02 4.258e+02 9.547e+02, threshold=6.748e+02, percent-clipped=3.0 2023-04-28 10:42:25,310 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=51876.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:42:46,347 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8480, 3.0105, 2.5400, 4.9560, 4.2225, 4.4133, 1.7783, 3.2064], device='cuda:7'), covar=tensor([0.1314, 0.0729, 0.1286, 0.0097, 0.0374, 0.0366, 0.1350, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0142, 0.0166, 0.0092, 0.0180, 0.0183, 0.0159, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 10:42:59,416 INFO [train.py:904] (7/8) Epoch 6, batch 1150, loss[loss=0.2059, simple_loss=0.2848, pruned_loss=0.06356, over 16605.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2791, pruned_loss=0.06295, over 3303365.59 frames. ], batch size: 62, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:43:51,822 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.98 vs. limit=5.0 2023-04-28 10:44:08,005 INFO [train.py:904] (7/8) Epoch 6, batch 1200, loss[loss=0.1976, simple_loss=0.2751, pruned_loss=0.0601, over 16452.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.278, pruned_loss=0.06249, over 3304289.35 frames. ], batch size: 68, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:44:33,635 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.935e+02 3.395e+02 4.046e+02 1.078e+03, threshold=6.791e+02, percent-clipped=4.0 2023-04-28 10:45:20,392 INFO [train.py:904] (7/8) Epoch 6, batch 1250, loss[loss=0.2218, simple_loss=0.2908, pruned_loss=0.07635, over 16670.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2783, pruned_loss=0.06314, over 3315204.58 frames. ], batch size: 89, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:45:52,360 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 10:45:52,488 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-04-28 10:46:13,403 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52038.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:46:16,549 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52040.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:46:26,721 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52048.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:46:30,471 INFO [train.py:904] (7/8) Epoch 6, batch 1300, loss[loss=0.2132, simple_loss=0.2813, pruned_loss=0.07258, over 16896.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2786, pruned_loss=0.06358, over 3306449.99 frames. ], batch size: 96, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:46:42,521 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.37 vs. limit=5.0 2023-04-28 10:46:52,667 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-28 10:46:58,326 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.918e+02 3.490e+02 4.202e+02 7.834e+02, threshold=6.979e+02, percent-clipped=4.0 2023-04-28 10:47:20,307 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52086.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:47:34,137 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52096.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:47:41,508 INFO [train.py:904] (7/8) Epoch 6, batch 1350, loss[loss=0.239, simple_loss=0.3164, pruned_loss=0.08079, over 16833.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2793, pruned_loss=0.06353, over 3315013.47 frames. ], batch size: 116, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:47:42,035 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52101.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:47:44,916 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52103.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:48:22,667 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-28 10:48:33,526 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9849, 4.1435, 4.3605, 1.8812, 4.6504, 4.6091, 3.3736, 3.5017], device='cuda:7'), covar=tensor([0.0590, 0.0150, 0.0146, 0.1192, 0.0042, 0.0075, 0.0300, 0.0355], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0094, 0.0083, 0.0142, 0.0072, 0.0084, 0.0117, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 10:48:51,205 INFO [train.py:904] (7/8) Epoch 6, batch 1400, loss[loss=0.2089, simple_loss=0.2731, pruned_loss=0.07237, over 11813.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2793, pruned_loss=0.06322, over 3320691.54 frames. ], batch size: 247, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:48:51,472 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52151.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:49:07,017 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3192, 3.8708, 3.2676, 2.0520, 2.8119, 2.3638, 3.6330, 3.6496], device='cuda:7'), covar=tensor([0.0230, 0.0486, 0.0579, 0.1490, 0.0716, 0.0921, 0.0508, 0.0690], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0133, 0.0152, 0.0140, 0.0130, 0.0125, 0.0140, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 10:49:07,246 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2023-04-28 10:49:19,189 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.784e+02 2.727e+02 3.349e+02 4.009e+02 8.330e+02, threshold=6.698e+02, percent-clipped=2.0 2023-04-28 10:49:49,775 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52193.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:49:58,143 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52199.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:50:00,576 INFO [train.py:904] (7/8) Epoch 6, batch 1450, loss[loss=0.1829, simple_loss=0.2771, pruned_loss=0.04435, over 17106.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2782, pruned_loss=0.06296, over 3315506.43 frames. ], batch size: 49, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:50:21,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0981, 4.3991, 3.3756, 2.4107, 3.2992, 2.6033, 4.6752, 4.3809], device='cuda:7'), covar=tensor([0.1993, 0.0606, 0.1251, 0.1818, 0.2113, 0.1449, 0.0351, 0.0641], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0256, 0.0272, 0.0248, 0.0286, 0.0205, 0.0250, 0.0272], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:50:48,750 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2790, 4.1018, 4.2653, 4.4783, 4.5846, 4.1236, 4.3838, 4.5565], device='cuda:7'), covar=tensor([0.0871, 0.0739, 0.1137, 0.0499, 0.0452, 0.0843, 0.0903, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0443, 0.0543, 0.0694, 0.0558, 0.0420, 0.0412, 0.0433, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:51:10,829 INFO [train.py:904] (7/8) Epoch 6, batch 1500, loss[loss=0.2187, simple_loss=0.2775, pruned_loss=0.07997, over 16843.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2786, pruned_loss=0.06377, over 3325094.80 frames. ], batch size: 109, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:51:15,582 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52254.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:51:38,536 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 3.021e+02 3.666e+02 4.541e+02 9.509e+02, threshold=7.332e+02, percent-clipped=3.0 2023-04-28 10:51:39,207 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3190, 2.4908, 2.5944, 5.0629, 2.3075, 3.5620, 2.6490, 2.5983], device='cuda:7'), covar=tensor([0.0487, 0.2200, 0.1242, 0.0222, 0.3109, 0.1153, 0.1859, 0.2667], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0332, 0.0276, 0.0320, 0.0376, 0.0350, 0.0303, 0.0405], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 10:52:18,652 INFO [train.py:904] (7/8) Epoch 6, batch 1550, loss[loss=0.2167, simple_loss=0.2966, pruned_loss=0.06844, over 16730.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2813, pruned_loss=0.06576, over 3316267.38 frames. ], batch size: 57, lr: 1.17e-02, grad_scale: 4.0 2023-04-28 10:52:34,692 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52312.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:53:28,032 INFO [train.py:904] (7/8) Epoch 6, batch 1600, loss[loss=0.2104, simple_loss=0.2794, pruned_loss=0.07065, over 16728.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2834, pruned_loss=0.06719, over 3311963.09 frames. ], batch size: 89, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:53:31,965 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 10:53:55,827 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.932e+02 3.332e+02 3.992e+02 7.476e+02, threshold=6.664e+02, percent-clipped=1.0 2023-04-28 10:53:59,285 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52373.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:54:07,530 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3837, 4.4530, 4.3243, 4.2215, 3.6836, 4.3554, 4.3196, 4.0508], device='cuda:7'), covar=tensor([0.0569, 0.0352, 0.0274, 0.0258, 0.1071, 0.0361, 0.0459, 0.0544], device='cuda:7'), in_proj_covar=tensor([0.0205, 0.0221, 0.0242, 0.0212, 0.0276, 0.0248, 0.0172, 0.0276], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 10:54:30,486 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52396.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:54:37,125 INFO [train.py:904] (7/8) Epoch 6, batch 1650, loss[loss=0.2084, simple_loss=0.296, pruned_loss=0.06043, over 16656.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2833, pruned_loss=0.06644, over 3320172.18 frames. ], batch size: 62, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:54:40,271 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52403.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:54:55,572 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2948, 4.2979, 4.7523, 4.7442, 4.7507, 4.3581, 4.4099, 4.2280], device='cuda:7'), covar=tensor([0.0272, 0.0465, 0.0311, 0.0393, 0.0349, 0.0263, 0.0726, 0.0421], device='cuda:7'), in_proj_covar=tensor([0.0265, 0.0267, 0.0263, 0.0262, 0.0314, 0.0276, 0.0387, 0.0229], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 10:55:45,573 INFO [train.py:904] (7/8) Epoch 6, batch 1700, loss[loss=0.2883, simple_loss=0.3481, pruned_loss=0.1142, over 12105.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.285, pruned_loss=0.06665, over 3316988.11 frames. ], batch size: 246, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:55:45,913 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52451.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:55:48,395 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 10:56:14,197 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.939e+02 3.544e+02 4.398e+02 9.866e+02, threshold=7.088e+02, percent-clipped=4.0 2023-04-28 10:56:14,604 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4905, 4.4384, 4.3692, 4.2854, 3.9965, 4.4331, 4.1972, 4.1559], device='cuda:7'), covar=tensor([0.0434, 0.0296, 0.0196, 0.0169, 0.0742, 0.0280, 0.0457, 0.0446], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0218, 0.0240, 0.0212, 0.0274, 0.0246, 0.0171, 0.0273], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 10:56:57,700 INFO [train.py:904] (7/8) Epoch 6, batch 1750, loss[loss=0.2087, simple_loss=0.2998, pruned_loss=0.05883, over 17155.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2854, pruned_loss=0.0658, over 3326398.13 frames. ], batch size: 47, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:57:15,951 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0087, 5.0195, 5.5016, 5.5260, 5.4907, 5.0928, 5.0986, 4.7950], device='cuda:7'), covar=tensor([0.0235, 0.0285, 0.0297, 0.0326, 0.0330, 0.0223, 0.0640, 0.0280], device='cuda:7'), in_proj_covar=tensor([0.0268, 0.0268, 0.0267, 0.0265, 0.0318, 0.0279, 0.0394, 0.0232], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 10:57:24,288 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6679, 3.7450, 4.0254, 3.9968, 4.0167, 3.7061, 3.7436, 3.7272], device='cuda:7'), covar=tensor([0.0307, 0.0408, 0.0343, 0.0441, 0.0401, 0.0313, 0.0673, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0268, 0.0268, 0.0267, 0.0265, 0.0319, 0.0279, 0.0394, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 10:58:05,491 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52549.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 10:58:07,617 INFO [train.py:904] (7/8) Epoch 6, batch 1800, loss[loss=0.2445, simple_loss=0.3078, pruned_loss=0.09063, over 16705.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2861, pruned_loss=0.06588, over 3331083.35 frames. ], batch size: 134, lr: 1.17e-02, grad_scale: 8.0 2023-04-28 10:58:14,157 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52555.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 10:58:36,301 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.850e+02 3.092e+02 3.773e+02 4.797e+02 9.614e+02, threshold=7.547e+02, percent-clipped=5.0 2023-04-28 10:58:56,503 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-04-28 10:59:17,732 INFO [train.py:904] (7/8) Epoch 6, batch 1850, loss[loss=0.216, simple_loss=0.3109, pruned_loss=0.06055, over 17025.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2867, pruned_loss=0.06582, over 3331286.39 frames. ], batch size: 50, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 10:59:25,912 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.06 vs. limit=5.0 2023-04-28 10:59:38,397 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52616.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 10:59:38,416 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6931, 3.1335, 2.7304, 4.3923, 3.8610, 4.2263, 1.5912, 2.9838], device='cuda:7'), covar=tensor([0.1273, 0.0481, 0.0876, 0.0079, 0.0223, 0.0307, 0.1208, 0.0703], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0144, 0.0167, 0.0095, 0.0186, 0.0187, 0.0161, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 11:00:27,100 INFO [train.py:904] (7/8) Epoch 6, batch 1900, loss[loss=0.1779, simple_loss=0.2707, pruned_loss=0.04251, over 17134.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2853, pruned_loss=0.06444, over 3336682.73 frames. ], batch size: 49, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:00:29,723 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 11:00:51,247 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52668.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:00:54,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.841e+02 2.667e+02 3.390e+02 4.181e+02 1.051e+03, threshold=6.780e+02, percent-clipped=5.0 2023-04-28 11:01:07,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1097, 5.5611, 5.6673, 5.5092, 5.5045, 6.0751, 5.6559, 5.4073], device='cuda:7'), covar=tensor([0.0606, 0.1652, 0.1583, 0.1596, 0.2642, 0.0850, 0.1081, 0.2255], device='cuda:7'), in_proj_covar=tensor([0.0303, 0.0437, 0.0429, 0.0370, 0.0498, 0.0463, 0.0352, 0.0502], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:01:30,623 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52696.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:01:36,635 INFO [train.py:904] (7/8) Epoch 6, batch 1950, loss[loss=0.2096, simple_loss=0.2944, pruned_loss=0.06245, over 17110.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2862, pruned_loss=0.06457, over 3323348.86 frames. ], batch size: 48, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:01:44,217 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7328, 4.8302, 5.2969, 5.3702, 5.3231, 4.8971, 4.8496, 4.5888], device='cuda:7'), covar=tensor([0.0285, 0.0410, 0.0353, 0.0341, 0.0341, 0.0281, 0.0760, 0.0345], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0268, 0.0266, 0.0263, 0.0322, 0.0281, 0.0397, 0.0232], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 11:01:44,336 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9371, 1.7405, 2.2345, 2.8112, 2.5130, 3.2251, 1.8403, 3.2407], device='cuda:7'), covar=tensor([0.0101, 0.0248, 0.0187, 0.0135, 0.0154, 0.0099, 0.0232, 0.0069], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0150, 0.0134, 0.0134, 0.0139, 0.0099, 0.0143, 0.0084], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 11:02:30,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0001, 4.2104, 3.2575, 2.4079, 3.1165, 2.3082, 4.4916, 4.3269], device='cuda:7'), covar=tensor([0.1945, 0.0616, 0.1127, 0.1759, 0.2025, 0.1477, 0.0342, 0.0571], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0254, 0.0269, 0.0247, 0.0284, 0.0203, 0.0245, 0.0268], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:02:36,907 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52744.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:02:47,597 INFO [train.py:904] (7/8) Epoch 6, batch 2000, loss[loss=0.2332, simple_loss=0.2959, pruned_loss=0.08527, over 16891.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.2867, pruned_loss=0.06488, over 3320960.64 frames. ], batch size: 116, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:02:49,228 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52752.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:02:49,294 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0639, 2.5393, 1.7580, 2.2111, 2.9156, 2.6933, 3.3029, 3.1013], device='cuda:7'), covar=tensor([0.0060, 0.0169, 0.0252, 0.0209, 0.0097, 0.0160, 0.0101, 0.0105], device='cuda:7'), in_proj_covar=tensor([0.0095, 0.0168, 0.0164, 0.0162, 0.0163, 0.0168, 0.0154, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:03:15,708 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.007e+02 2.923e+02 3.553e+02 4.132e+02 6.210e+02, threshold=7.105e+02, percent-clipped=0.0 2023-04-28 11:03:24,473 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8249, 4.7729, 4.6476, 4.1422, 4.7038, 1.9840, 4.5138, 4.6529], device='cuda:7'), covar=tensor([0.0071, 0.0057, 0.0111, 0.0266, 0.0069, 0.1723, 0.0095, 0.0112], device='cuda:7'), in_proj_covar=tensor([0.0103, 0.0091, 0.0140, 0.0137, 0.0106, 0.0150, 0.0121, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:03:40,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2669, 4.2096, 4.6830, 4.7204, 4.7443, 4.3516, 4.3284, 4.1337], device='cuda:7'), covar=tensor([0.0276, 0.0400, 0.0303, 0.0370, 0.0379, 0.0291, 0.0782, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0265, 0.0265, 0.0263, 0.0262, 0.0318, 0.0278, 0.0392, 0.0228], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 11:03:57,172 INFO [train.py:904] (7/8) Epoch 6, batch 2050, loss[loss=0.234, simple_loss=0.2944, pruned_loss=0.08679, over 16796.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2881, pruned_loss=0.06586, over 3316656.96 frames. ], batch size: 124, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:04:15,457 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52813.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:05:06,145 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:05:08,353 INFO [train.py:904] (7/8) Epoch 6, batch 2100, loss[loss=0.1896, simple_loss=0.276, pruned_loss=0.05154, over 16845.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2888, pruned_loss=0.06628, over 3308074.86 frames. ], batch size: 42, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:05:36,282 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.933e+02 3.490e+02 4.297e+02 7.985e+02, threshold=6.979e+02, percent-clipped=3.0 2023-04-28 11:06:12,958 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=52897.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:06:18,013 INFO [train.py:904] (7/8) Epoch 6, batch 2150, loss[loss=0.1866, simple_loss=0.2707, pruned_loss=0.05121, over 17219.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2902, pruned_loss=0.06738, over 3302897.71 frames. ], batch size: 44, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:06:33,075 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52911.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 11:07:29,318 INFO [train.py:904] (7/8) Epoch 6, batch 2200, loss[loss=0.1689, simple_loss=0.2472, pruned_loss=0.04525, over 16756.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2915, pruned_loss=0.06863, over 3282683.17 frames. ], batch size: 39, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:07:41,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9437, 4.2897, 2.1220, 4.5194, 2.9669, 4.5944, 2.2331, 3.2999], device='cuda:7'), covar=tensor([0.0143, 0.0253, 0.1419, 0.0092, 0.0762, 0.0267, 0.1400, 0.0518], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0164, 0.0179, 0.0087, 0.0161, 0.0195, 0.0187, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 11:07:53,107 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52968.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:07:56,707 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.110e+02 3.315e+02 3.926e+02 4.640e+02 8.894e+02, threshold=7.853e+02, percent-clipped=3.0 2023-04-28 11:08:38,585 INFO [train.py:904] (7/8) Epoch 6, batch 2250, loss[loss=0.1814, simple_loss=0.2622, pruned_loss=0.0503, over 16837.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2916, pruned_loss=0.06844, over 3297618.11 frames. ], batch size: 42, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:08:59,895 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53016.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:09:02,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4334, 4.2464, 3.8519, 2.0601, 3.1017, 2.4721, 3.9018, 3.8738], device='cuda:7'), covar=tensor([0.0243, 0.0443, 0.0431, 0.1460, 0.0661, 0.0881, 0.0553, 0.0917], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0135, 0.0152, 0.0139, 0.0129, 0.0124, 0.0138, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 11:09:31,800 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1788, 4.4893, 3.4416, 2.5056, 3.2750, 2.5170, 4.6581, 4.2745], device='cuda:7'), covar=tensor([0.1930, 0.0510, 0.1100, 0.1675, 0.2046, 0.1469, 0.0326, 0.0655], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0257, 0.0272, 0.0251, 0.0290, 0.0206, 0.0248, 0.0272], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:09:47,167 INFO [train.py:904] (7/8) Epoch 6, batch 2300, loss[loss=0.1966, simple_loss=0.2875, pruned_loss=0.05287, over 17143.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2919, pruned_loss=0.06826, over 3305939.40 frames. ], batch size: 48, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:09:54,619 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53056.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:10:07,103 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 11:10:15,634 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.084e+02 3.786e+02 4.874e+02 1.349e+03, threshold=7.572e+02, percent-clipped=4.0 2023-04-28 11:10:57,595 INFO [train.py:904] (7/8) Epoch 6, batch 2350, loss[loss=0.2126, simple_loss=0.2799, pruned_loss=0.07263, over 16385.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2912, pruned_loss=0.06851, over 3306688.80 frames. ], batch size: 146, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:11:08,092 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53108.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:11:21,211 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53117.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:11:46,011 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7424, 4.7614, 5.2406, 5.2665, 5.2538, 4.8331, 4.8162, 4.5619], device='cuda:7'), covar=tensor([0.0241, 0.0420, 0.0265, 0.0351, 0.0335, 0.0252, 0.0752, 0.0337], device='cuda:7'), in_proj_covar=tensor([0.0263, 0.0264, 0.0260, 0.0259, 0.0307, 0.0273, 0.0386, 0.0227], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 11:12:08,723 INFO [train.py:904] (7/8) Epoch 6, batch 2400, loss[loss=0.2438, simple_loss=0.3066, pruned_loss=0.0905, over 16943.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2918, pruned_loss=0.06842, over 3307945.14 frames. ], batch size: 109, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:12:36,944 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.873e+02 2.840e+02 3.422e+02 4.151e+02 8.672e+02, threshold=6.844e+02, percent-clipped=2.0 2023-04-28 11:12:48,877 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-28 11:13:19,230 INFO [train.py:904] (7/8) Epoch 6, batch 2450, loss[loss=0.1992, simple_loss=0.2728, pruned_loss=0.06286, over 16832.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2914, pruned_loss=0.06711, over 3316724.65 frames. ], batch size: 102, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:13:25,628 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9579, 4.2985, 4.4917, 3.1959, 3.9703, 4.2679, 3.8093, 2.2324], device='cuda:7'), covar=tensor([0.0275, 0.0027, 0.0025, 0.0201, 0.0040, 0.0045, 0.0042, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0062, 0.0062, 0.0115, 0.0063, 0.0072, 0.0065, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:13:33,788 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53211.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 11:14:16,112 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:14:29,177 INFO [train.py:904] (7/8) Epoch 6, batch 2500, loss[loss=0.2186, simple_loss=0.3043, pruned_loss=0.06645, over 17115.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2911, pruned_loss=0.06681, over 3324457.82 frames. ], batch size: 47, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:14:40,023 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53259.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:14:57,077 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.700e+02 3.263e+02 4.236e+02 1.034e+03, threshold=6.525e+02, percent-clipped=4.0 2023-04-28 11:15:23,073 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7734, 4.6720, 4.6819, 4.4520, 4.2836, 4.6198, 4.5543, 4.3643], device='cuda:7'), covar=tensor([0.0462, 0.0366, 0.0193, 0.0207, 0.0788, 0.0357, 0.0337, 0.0533], device='cuda:7'), in_proj_covar=tensor([0.0209, 0.0227, 0.0247, 0.0220, 0.0285, 0.0252, 0.0177, 0.0279], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:15:38,467 INFO [train.py:904] (7/8) Epoch 6, batch 2550, loss[loss=0.2015, simple_loss=0.2767, pruned_loss=0.06315, over 16834.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2911, pruned_loss=0.06676, over 3316338.17 frames. ], batch size: 96, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:15:40,711 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53302.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:15:43,314 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53304.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:15:58,347 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53315.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:16:48,728 INFO [train.py:904] (7/8) Epoch 6, batch 2600, loss[loss=0.2551, simple_loss=0.3186, pruned_loss=0.0958, over 12335.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2902, pruned_loss=0.0665, over 3313028.00 frames. ], batch size: 247, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:17:05,185 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3091, 4.0040, 3.2714, 1.9494, 2.8301, 2.4704, 3.5106, 3.7529], device='cuda:7'), covar=tensor([0.0266, 0.0481, 0.0612, 0.1587, 0.0792, 0.0882, 0.0680, 0.0773], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0137, 0.0154, 0.0140, 0.0131, 0.0125, 0.0140, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 11:17:09,376 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53365.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:17:11,666 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53367.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:17:16,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.796e+02 3.424e+02 4.100e+02 9.306e+02, threshold=6.847e+02, percent-clipped=3.0 2023-04-28 11:17:24,099 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53376.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:17:37,457 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 11:17:59,457 INFO [train.py:904] (7/8) Epoch 6, batch 2650, loss[loss=0.2257, simple_loss=0.2944, pruned_loss=0.07853, over 16786.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2911, pruned_loss=0.06626, over 3317906.90 frames. ], batch size: 124, lr: 1.16e-02, grad_scale: 8.0 2023-04-28 11:18:10,299 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53408.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:18:14,780 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53412.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:18:37,231 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53428.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:18:37,411 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-28 11:19:09,536 INFO [train.py:904] (7/8) Epoch 6, batch 2700, loss[loss=0.1804, simple_loss=0.2611, pruned_loss=0.04983, over 16776.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2903, pruned_loss=0.06544, over 3327493.53 frames. ], batch size: 39, lr: 1.16e-02, grad_scale: 4.0 2023-04-28 11:19:16,986 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53456.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:19:26,886 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7369, 4.7551, 5.3592, 5.3444, 5.3217, 4.9108, 4.8894, 4.6653], device='cuda:7'), covar=tensor([0.0251, 0.0386, 0.0308, 0.0376, 0.0412, 0.0282, 0.0870, 0.0313], device='cuda:7'), in_proj_covar=tensor([0.0265, 0.0265, 0.0260, 0.0261, 0.0313, 0.0277, 0.0395, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 11:19:38,366 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.946e+02 3.581e+02 4.267e+02 8.371e+02, threshold=7.163e+02, percent-clipped=3.0 2023-04-28 11:20:06,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8926, 2.7337, 1.9099, 2.3886, 3.1997, 2.7660, 3.8577, 3.3940], device='cuda:7'), covar=tensor([0.0025, 0.0186, 0.0306, 0.0239, 0.0117, 0.0197, 0.0092, 0.0099], device='cuda:7'), in_proj_covar=tensor([0.0095, 0.0167, 0.0166, 0.0164, 0.0163, 0.0169, 0.0156, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:20:19,909 INFO [train.py:904] (7/8) Epoch 6, batch 2750, loss[loss=0.1727, simple_loss=0.2558, pruned_loss=0.04477, over 16820.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2892, pruned_loss=0.06423, over 3330084.43 frames. ], batch size: 39, lr: 1.16e-02, grad_scale: 4.0 2023-04-28 11:21:14,972 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-28 11:21:31,892 INFO [train.py:904] (7/8) Epoch 6, batch 2800, loss[loss=0.1951, simple_loss=0.2795, pruned_loss=0.0554, over 17199.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2892, pruned_loss=0.06458, over 3329398.63 frames. ], batch size: 45, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:22:01,721 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 2.830e+02 3.513e+02 4.760e+02 1.218e+03, threshold=7.026e+02, percent-clipped=5.0 2023-04-28 11:22:37,902 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53597.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:22:43,373 INFO [train.py:904] (7/8) Epoch 6, batch 2850, loss[loss=0.3117, simple_loss=0.3624, pruned_loss=0.1306, over 12052.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2891, pruned_loss=0.06554, over 3310648.05 frames. ], batch size: 246, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:23:51,294 INFO [train.py:904] (7/8) Epoch 6, batch 2900, loss[loss=0.1837, simple_loss=0.2652, pruned_loss=0.05113, over 17209.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2879, pruned_loss=0.06579, over 3311093.05 frames. ], batch size: 46, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:24:04,606 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53660.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:24:16,089 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3707, 1.5769, 2.1120, 2.2788, 2.5399, 2.3854, 1.6966, 2.5933], device='cuda:7'), covar=tensor([0.0082, 0.0212, 0.0130, 0.0127, 0.0091, 0.0124, 0.0189, 0.0050], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0152, 0.0136, 0.0136, 0.0141, 0.0102, 0.0146, 0.0087], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 11:24:19,426 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53671.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:24:20,250 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.863e+02 3.412e+02 4.656e+02 8.517e+02, threshold=6.823e+02, percent-clipped=1.0 2023-04-28 11:24:59,791 INFO [train.py:904] (7/8) Epoch 6, batch 2950, loss[loss=0.2073, simple_loss=0.295, pruned_loss=0.05982, over 16701.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2879, pruned_loss=0.06688, over 3314268.43 frames. ], batch size: 57, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:25:15,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53712.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:25:30,145 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53723.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:25:30,761 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 11:26:03,591 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-04-28 11:26:08,567 INFO [train.py:904] (7/8) Epoch 6, batch 3000, loss[loss=0.2206, simple_loss=0.2893, pruned_loss=0.07594, over 16813.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2883, pruned_loss=0.06686, over 3317429.56 frames. ], batch size: 124, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:26:08,567 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 11:26:17,403 INFO [train.py:938] (7/8) Epoch 6, validation: loss=0.1514, simple_loss=0.258, pruned_loss=0.02246, over 944034.00 frames. 2023-04-28 11:26:17,403 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 11:26:29,470 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53760.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:26:45,971 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.839e+02 3.365e+02 4.010e+02 8.699e+02, threshold=6.731e+02, percent-clipped=2.0 2023-04-28 11:26:59,314 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 11:27:25,838 INFO [train.py:904] (7/8) Epoch 6, batch 3050, loss[loss=0.241, simple_loss=0.3086, pruned_loss=0.08668, over 12048.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2883, pruned_loss=0.06693, over 3297820.35 frames. ], batch size: 246, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:27:48,723 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4301, 2.3410, 2.0489, 2.2437, 2.7638, 2.6432, 3.4615, 3.0654], device='cuda:7'), covar=tensor([0.0035, 0.0183, 0.0219, 0.0224, 0.0127, 0.0174, 0.0096, 0.0105], device='cuda:7'), in_proj_covar=tensor([0.0095, 0.0164, 0.0164, 0.0161, 0.0162, 0.0168, 0.0156, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:27:53,558 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7727, 3.8755, 2.8787, 2.3956, 2.7841, 2.2042, 3.8686, 3.8146], device='cuda:7'), covar=tensor([0.1925, 0.0561, 0.1151, 0.1564, 0.2037, 0.1579, 0.0426, 0.0712], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0257, 0.0273, 0.0252, 0.0293, 0.0207, 0.0250, 0.0277], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:28:35,659 INFO [train.py:904] (7/8) Epoch 6, batch 3100, loss[loss=0.2206, simple_loss=0.2842, pruned_loss=0.07848, over 16859.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.288, pruned_loss=0.06632, over 3306092.95 frames. ], batch size: 96, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:29:05,912 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.841e+02 3.355e+02 4.130e+02 6.611e+02, threshold=6.710e+02, percent-clipped=0.0 2023-04-28 11:29:12,975 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5395, 3.2026, 2.8360, 1.8413, 2.5957, 2.1429, 3.0747, 3.1156], device='cuda:7'), covar=tensor([0.0283, 0.0639, 0.0522, 0.1550, 0.0698, 0.0900, 0.0589, 0.0823], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0136, 0.0153, 0.0138, 0.0129, 0.0123, 0.0138, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 11:29:14,059 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7165, 1.5071, 2.0799, 2.5586, 2.6200, 2.5399, 1.6511, 2.7629], device='cuda:7'), covar=tensor([0.0082, 0.0231, 0.0162, 0.0109, 0.0099, 0.0122, 0.0227, 0.0056], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0153, 0.0137, 0.0137, 0.0142, 0.0103, 0.0147, 0.0089], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 11:29:40,284 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53897.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:29:45,147 INFO [train.py:904] (7/8) Epoch 6, batch 3150, loss[loss=0.1762, simple_loss=0.2555, pruned_loss=0.04841, over 16988.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2866, pruned_loss=0.06567, over 3309985.24 frames. ], batch size: 41, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:30:49,058 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=53945.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:30:49,139 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2003, 5.5359, 5.2638, 5.3264, 4.9189, 4.7485, 5.1186, 5.6709], device='cuda:7'), covar=tensor([0.0768, 0.0723, 0.0909, 0.0538, 0.0634, 0.0710, 0.0724, 0.0729], device='cuda:7'), in_proj_covar=tensor([0.0427, 0.0570, 0.0473, 0.0362, 0.0348, 0.0353, 0.0457, 0.0398], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:30:56,671 INFO [train.py:904] (7/8) Epoch 6, batch 3200, loss[loss=0.212, simple_loss=0.2906, pruned_loss=0.06672, over 15974.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2853, pruned_loss=0.06475, over 3307688.21 frames. ], batch size: 35, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:31:09,494 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53960.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:31:25,933 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53971.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:31:26,630 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.936e+02 3.448e+02 4.117e+02 1.255e+03, threshold=6.896e+02, percent-clipped=5.0 2023-04-28 11:32:09,213 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-28 11:32:09,703 INFO [train.py:904] (7/8) Epoch 6, batch 3250, loss[loss=0.2948, simple_loss=0.3519, pruned_loss=0.1189, over 15513.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2857, pruned_loss=0.065, over 3310623.81 frames. ], batch size: 191, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:32:20,416 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=54008.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:32:35,460 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=54019.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:32:40,143 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=54023.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:33:17,051 INFO [train.py:904] (7/8) Epoch 6, batch 3300, loss[loss=0.2637, simple_loss=0.3341, pruned_loss=0.09665, over 12280.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2863, pruned_loss=0.06522, over 3311967.26 frames. ], batch size: 246, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:33:17,369 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4941, 5.9102, 5.6187, 5.7033, 5.0926, 4.9421, 5.4601, 6.0359], device='cuda:7'), covar=tensor([0.0782, 0.0681, 0.0807, 0.0462, 0.0754, 0.0545, 0.0634, 0.0609], device='cuda:7'), in_proj_covar=tensor([0.0428, 0.0570, 0.0472, 0.0361, 0.0351, 0.0354, 0.0456, 0.0398], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:33:20,464 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3374, 1.2284, 1.8907, 2.2347, 2.3663, 2.3015, 1.5341, 2.2713], device='cuda:7'), covar=tensor([0.0089, 0.0260, 0.0149, 0.0116, 0.0108, 0.0108, 0.0201, 0.0055], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0153, 0.0138, 0.0137, 0.0141, 0.0102, 0.0145, 0.0089], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 11:33:45,196 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=54071.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:33:46,071 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 3.091e+02 3.848e+02 4.464e+02 8.976e+02, threshold=7.696e+02, percent-clipped=3.0 2023-04-28 11:34:07,630 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8486, 2.4364, 1.8722, 2.0388, 2.8474, 2.7078, 3.2053, 3.0396], device='cuda:7'), covar=tensor([0.0072, 0.0186, 0.0249, 0.0269, 0.0118, 0.0149, 0.0107, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0099, 0.0170, 0.0169, 0.0166, 0.0167, 0.0171, 0.0161, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:34:26,275 INFO [train.py:904] (7/8) Epoch 6, batch 3350, loss[loss=0.2058, simple_loss=0.282, pruned_loss=0.06485, over 16763.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2879, pruned_loss=0.06593, over 3318281.57 frames. ], batch size: 102, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:35:01,047 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 11:35:34,493 INFO [train.py:904] (7/8) Epoch 6, batch 3400, loss[loss=0.2138, simple_loss=0.3075, pruned_loss=0.06009, over 17039.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2875, pruned_loss=0.06566, over 3317967.62 frames. ], batch size: 55, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:35:43,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-28 11:36:05,331 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.871e+02 3.468e+02 4.191e+02 6.688e+02, threshold=6.936e+02, percent-clipped=0.0 2023-04-28 11:36:32,858 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-04-28 11:36:46,919 INFO [train.py:904] (7/8) Epoch 6, batch 3450, loss[loss=0.1939, simple_loss=0.2567, pruned_loss=0.06557, over 16732.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2867, pruned_loss=0.06534, over 3318962.23 frames. ], batch size: 124, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:37:58,612 INFO [train.py:904] (7/8) Epoch 6, batch 3500, loss[loss=0.2223, simple_loss=0.2999, pruned_loss=0.07239, over 17095.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2848, pruned_loss=0.06448, over 3308754.12 frames. ], batch size: 53, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:38:28,905 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.000e+02 2.700e+02 3.164e+02 3.995e+02 8.205e+02, threshold=6.329e+02, percent-clipped=2.0 2023-04-28 11:39:10,635 INFO [train.py:904] (7/8) Epoch 6, batch 3550, loss[loss=0.2139, simple_loss=0.2815, pruned_loss=0.07321, over 16510.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2835, pruned_loss=0.06397, over 3308990.41 frames. ], batch size: 75, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:39:53,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8172, 4.8238, 4.6726, 4.5751, 4.2838, 4.7381, 4.6028, 4.4959], device='cuda:7'), covar=tensor([0.0540, 0.0388, 0.0249, 0.0226, 0.0908, 0.0345, 0.0366, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0212, 0.0230, 0.0253, 0.0225, 0.0287, 0.0255, 0.0177, 0.0283], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:40:21,766 INFO [train.py:904] (7/8) Epoch 6, batch 3600, loss[loss=0.1916, simple_loss=0.2795, pruned_loss=0.05187, over 17094.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2816, pruned_loss=0.06352, over 3302812.54 frames. ], batch size: 47, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:40:51,136 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.869e+02 3.018e+02 3.592e+02 4.498e+02 7.311e+02, threshold=7.184e+02, percent-clipped=7.0 2023-04-28 11:41:04,551 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0613, 4.6862, 4.9755, 5.2542, 5.3884, 4.6773, 5.3579, 5.3486], device='cuda:7'), covar=tensor([0.0919, 0.0756, 0.1226, 0.0453, 0.0353, 0.0618, 0.0408, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0443, 0.0542, 0.0700, 0.0565, 0.0420, 0.0415, 0.0433, 0.0475], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:41:33,853 INFO [train.py:904] (7/8) Epoch 6, batch 3650, loss[loss=0.2318, simple_loss=0.2947, pruned_loss=0.08443, over 16463.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2813, pruned_loss=0.06482, over 3270268.68 frames. ], batch size: 146, lr: 1.15e-02, grad_scale: 8.0 2023-04-28 11:41:37,804 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 11:42:20,201 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6819, 3.8273, 2.9346, 2.3919, 2.7396, 2.2742, 3.8712, 3.7692], device='cuda:7'), covar=tensor([0.1960, 0.0530, 0.1132, 0.1656, 0.2117, 0.1445, 0.0414, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0258, 0.0273, 0.0256, 0.0295, 0.0207, 0.0251, 0.0282], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:42:46,528 INFO [train.py:904] (7/8) Epoch 6, batch 3700, loss[loss=0.1933, simple_loss=0.2632, pruned_loss=0.06171, over 16440.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2803, pruned_loss=0.06646, over 3264107.24 frames. ], batch size: 68, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:42:53,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8489, 4.2623, 4.4767, 4.4484, 4.4090, 4.1452, 3.7583, 4.0235], device='cuda:7'), covar=tensor([0.0595, 0.0563, 0.0421, 0.0520, 0.0537, 0.0438, 0.1231, 0.0607], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0269, 0.0262, 0.0261, 0.0321, 0.0278, 0.0390, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 11:43:17,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.788e+02 2.928e+02 3.594e+02 4.333e+02 7.725e+02, threshold=7.187e+02, percent-clipped=2.0 2023-04-28 11:43:59,569 INFO [train.py:904] (7/8) Epoch 6, batch 3750, loss[loss=0.2106, simple_loss=0.2763, pruned_loss=0.07245, over 16490.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.281, pruned_loss=0.06804, over 3264181.22 frames. ], batch size: 146, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:44:10,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4908, 4.5468, 4.4157, 4.3219, 3.9797, 4.4751, 4.3090, 4.2123], device='cuda:7'), covar=tensor([0.0496, 0.0311, 0.0216, 0.0219, 0.0912, 0.0282, 0.0419, 0.0459], device='cuda:7'), in_proj_covar=tensor([0.0205, 0.0224, 0.0245, 0.0218, 0.0277, 0.0245, 0.0171, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:44:12,088 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7015, 4.0657, 3.1219, 2.4400, 3.0546, 2.4617, 4.3512, 4.1192], device='cuda:7'), covar=tensor([0.2247, 0.0605, 0.1357, 0.1621, 0.2216, 0.1383, 0.0341, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0257, 0.0272, 0.0254, 0.0294, 0.0206, 0.0249, 0.0280], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:45:13,215 INFO [train.py:904] (7/8) Epoch 6, batch 3800, loss[loss=0.1985, simple_loss=0.2653, pruned_loss=0.06592, over 16729.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2827, pruned_loss=0.0693, over 3269126.35 frames. ], batch size: 83, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:45:46,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.146e+02 2.814e+02 3.370e+02 4.517e+02 7.386e+02, threshold=6.740e+02, percent-clipped=3.0 2023-04-28 11:46:28,643 INFO [train.py:904] (7/8) Epoch 6, batch 3850, loss[loss=0.1907, simple_loss=0.2624, pruned_loss=0.05946, over 16537.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2827, pruned_loss=0.06995, over 3261301.45 frames. ], batch size: 68, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:47:02,037 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-28 11:47:06,931 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8059, 4.7797, 4.6434, 4.5384, 4.2413, 4.6801, 4.6032, 4.4380], device='cuda:7'), covar=tensor([0.0468, 0.0335, 0.0227, 0.0239, 0.0875, 0.0325, 0.0348, 0.0470], device='cuda:7'), in_proj_covar=tensor([0.0207, 0.0225, 0.0246, 0.0218, 0.0279, 0.0246, 0.0173, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:47:40,666 INFO [train.py:904] (7/8) Epoch 6, batch 3900, loss[loss=0.2044, simple_loss=0.2724, pruned_loss=0.06818, over 16497.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2821, pruned_loss=0.07029, over 3256121.32 frames. ], batch size: 146, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:47:59,643 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54664.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 11:48:10,863 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.751e+02 3.451e+02 4.155e+02 1.044e+03, threshold=6.903e+02, percent-clipped=2.0 2023-04-28 11:48:53,477 INFO [train.py:904] (7/8) Epoch 6, batch 3950, loss[loss=0.1881, simple_loss=0.2571, pruned_loss=0.05949, over 16821.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2815, pruned_loss=0.0706, over 3264549.23 frames. ], batch size: 96, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:49:27,799 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54725.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:50:04,727 INFO [train.py:904] (7/8) Epoch 6, batch 4000, loss[loss=0.2138, simple_loss=0.2814, pruned_loss=0.07312, over 16771.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2818, pruned_loss=0.07124, over 3270336.29 frames. ], batch size: 124, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 11:50:09,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9529, 2.7252, 2.7185, 1.7518, 2.9056, 2.8956, 2.4387, 2.4163], device='cuda:7'), covar=tensor([0.0820, 0.0179, 0.0156, 0.0994, 0.0083, 0.0130, 0.0384, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0095, 0.0082, 0.0139, 0.0072, 0.0083, 0.0116, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 11:50:36,657 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.045e+02 2.875e+02 3.507e+02 4.223e+02 8.153e+02, threshold=7.015e+02, percent-clipped=4.0 2023-04-28 11:51:03,997 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5846, 4.5768, 4.4504, 3.8226, 4.4789, 1.8009, 4.2100, 4.2591], device='cuda:7'), covar=tensor([0.0054, 0.0046, 0.0097, 0.0291, 0.0059, 0.1765, 0.0097, 0.0138], device='cuda:7'), in_proj_covar=tensor([0.0103, 0.0092, 0.0141, 0.0138, 0.0107, 0.0148, 0.0124, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:51:17,348 INFO [train.py:904] (7/8) Epoch 6, batch 4050, loss[loss=0.2168, simple_loss=0.2874, pruned_loss=0.0731, over 12360.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2818, pruned_loss=0.06989, over 3257023.70 frames. ], batch size: 247, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:51:52,720 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2926, 3.2301, 2.5152, 2.0650, 2.3350, 2.0852, 3.1965, 3.2181], device='cuda:7'), covar=tensor([0.2341, 0.0664, 0.1355, 0.1666, 0.2088, 0.1522, 0.0479, 0.0573], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0253, 0.0273, 0.0254, 0.0297, 0.0208, 0.0251, 0.0277], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 11:51:55,639 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5818, 2.0726, 2.1369, 4.2061, 1.8245, 2.7732, 2.2336, 2.1840], device='cuda:7'), covar=tensor([0.0630, 0.2586, 0.1418, 0.0291, 0.3325, 0.1386, 0.2381, 0.2724], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0346, 0.0282, 0.0324, 0.0383, 0.0367, 0.0314, 0.0418], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:52:28,304 INFO [train.py:904] (7/8) Epoch 6, batch 4100, loss[loss=0.1815, simple_loss=0.2638, pruned_loss=0.04961, over 16486.00 frames. ], tot_loss[loss=0.2092, simple_loss=0.2818, pruned_loss=0.06834, over 3246397.30 frames. ], batch size: 68, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:52:45,915 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54862.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:53:01,851 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.687e+02 2.498e+02 3.136e+02 3.951e+02 8.635e+02, threshold=6.272e+02, percent-clipped=2.0 2023-04-28 11:53:05,779 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2023-04-28 11:53:45,450 INFO [train.py:904] (7/8) Epoch 6, batch 4150, loss[loss=0.2212, simple_loss=0.3067, pruned_loss=0.06788, over 16583.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.2894, pruned_loss=0.07121, over 3226874.78 frames. ], batch size: 68, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:53:58,196 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 11:54:19,553 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54923.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 11:55:01,110 INFO [train.py:904] (7/8) Epoch 6, batch 4200, loss[loss=0.2234, simple_loss=0.3118, pruned_loss=0.06745, over 16765.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.2971, pruned_loss=0.0732, over 3203260.16 frames. ], batch size: 83, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:55:33,721 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.052e+02 3.214e+02 3.756e+02 4.661e+02 1.168e+03, threshold=7.512e+02, percent-clipped=6.0 2023-04-28 11:55:37,610 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8094, 4.1030, 3.8530, 3.9368, 3.5825, 3.6527, 3.7724, 4.0166], device='cuda:7'), covar=tensor([0.0777, 0.0741, 0.0832, 0.0506, 0.0658, 0.1413, 0.0731, 0.0952], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0524, 0.0438, 0.0335, 0.0324, 0.0336, 0.0428, 0.0371], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:56:15,276 INFO [train.py:904] (7/8) Epoch 6, batch 4250, loss[loss=0.2044, simple_loss=0.2787, pruned_loss=0.06504, over 12228.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3003, pruned_loss=0.07328, over 3182285.61 frames. ], batch size: 248, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:56:17,929 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 11:56:44,674 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55020.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 11:57:09,610 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-28 11:57:29,403 INFO [train.py:904] (7/8) Epoch 6, batch 4300, loss[loss=0.218, simple_loss=0.2998, pruned_loss=0.06807, over 16748.00 frames. ], tot_loss[loss=0.2226, simple_loss=0.3014, pruned_loss=0.0719, over 3196782.12 frames. ], batch size: 62, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:57:32,092 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-28 11:57:39,445 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 11:58:01,804 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.606e+02 3.247e+02 3.921e+02 6.703e+02, threshold=6.494e+02, percent-clipped=0.0 2023-04-28 11:58:43,135 INFO [train.py:904] (7/8) Epoch 6, batch 4350, loss[loss=0.2229, simple_loss=0.3011, pruned_loss=0.07237, over 16439.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3048, pruned_loss=0.0732, over 3204121.77 frames. ], batch size: 35, lr: 1.14e-02, grad_scale: 4.0 2023-04-28 11:58:45,565 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9246, 3.7034, 3.0984, 1.6163, 2.6301, 2.0923, 3.4099, 3.5336], device='cuda:7'), covar=tensor([0.0261, 0.0439, 0.0581, 0.1664, 0.0772, 0.0893, 0.0598, 0.0757], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0133, 0.0154, 0.0139, 0.0132, 0.0124, 0.0138, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 11:59:34,702 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3655, 4.3709, 4.2445, 3.6826, 4.3116, 1.5775, 4.0181, 4.1335], device='cuda:7'), covar=tensor([0.0054, 0.0044, 0.0086, 0.0258, 0.0048, 0.1863, 0.0084, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0098, 0.0088, 0.0134, 0.0133, 0.0100, 0.0145, 0.0118, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 11:59:53,025 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55148.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 11:59:56,598 INFO [train.py:904] (7/8) Epoch 6, batch 4400, loss[loss=0.2473, simple_loss=0.3278, pruned_loss=0.08347, over 16920.00 frames. ], tot_loss[loss=0.228, simple_loss=0.3072, pruned_loss=0.07437, over 3202031.51 frames. ], batch size: 109, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:00:27,150 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.770e+02 3.222e+02 4.108e+02 7.043e+02, threshold=6.445e+02, percent-clipped=2.0 2023-04-28 12:01:06,626 INFO [train.py:904] (7/8) Epoch 6, batch 4450, loss[loss=0.2594, simple_loss=0.3416, pruned_loss=0.08864, over 16391.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.3102, pruned_loss=0.0747, over 3208566.72 frames. ], batch size: 146, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:01:19,227 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55209.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:01:21,578 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55211.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:01:21,663 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7525, 1.6838, 2.2539, 2.7622, 2.6861, 3.1137, 1.7254, 2.9146], device='cuda:7'), covar=tensor([0.0100, 0.0248, 0.0145, 0.0127, 0.0105, 0.0070, 0.0240, 0.0047], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0149, 0.0132, 0.0131, 0.0137, 0.0099, 0.0144, 0.0086], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 12:01:31,362 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55218.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:01:41,370 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6361, 2.2116, 1.6655, 2.0622, 2.6182, 2.3163, 2.8476, 2.8509], device='cuda:7'), covar=tensor([0.0049, 0.0185, 0.0258, 0.0213, 0.0106, 0.0188, 0.0078, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0089, 0.0159, 0.0161, 0.0159, 0.0156, 0.0163, 0.0144, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:01:42,603 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5480, 3.9551, 4.1568, 1.9519, 4.4151, 4.5120, 3.1020, 3.3195], device='cuda:7'), covar=tensor([0.0761, 0.0147, 0.0147, 0.1086, 0.0043, 0.0034, 0.0316, 0.0323], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0095, 0.0082, 0.0141, 0.0073, 0.0080, 0.0116, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 12:02:16,415 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 12:02:16,920 INFO [train.py:904] (7/8) Epoch 6, batch 4500, loss[loss=0.2201, simple_loss=0.3092, pruned_loss=0.06552, over 16756.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.31, pruned_loss=0.07455, over 3222659.14 frames. ], batch size: 83, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:02:40,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9112, 4.0642, 2.0358, 4.6863, 2.8163, 4.5290, 2.2081, 3.0538], device='cuda:7'), covar=tensor([0.0148, 0.0220, 0.1551, 0.0019, 0.0698, 0.0246, 0.1381, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0156, 0.0177, 0.0082, 0.0160, 0.0188, 0.0183, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 12:02:41,844 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5303, 5.5159, 5.2175, 4.7748, 5.4830, 2.0047, 5.2104, 5.3011], device='cuda:7'), covar=tensor([0.0028, 0.0031, 0.0060, 0.0227, 0.0031, 0.1629, 0.0049, 0.0076], device='cuda:7'), in_proj_covar=tensor([0.0099, 0.0088, 0.0135, 0.0134, 0.0100, 0.0146, 0.0117, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:02:47,718 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:02:48,407 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.800e+02 2.322e+02 2.860e+02 3.458e+02 5.535e+02, threshold=5.719e+02, percent-clipped=0.0 2023-04-28 12:03:29,870 INFO [train.py:904] (7/8) Epoch 6, batch 4550, loss[loss=0.2308, simple_loss=0.3131, pruned_loss=0.07424, over 16621.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3102, pruned_loss=0.07461, over 3232467.12 frames. ], batch size: 62, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:03:57,526 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55320.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 12:04:41,597 INFO [train.py:904] (7/8) Epoch 6, batch 4600, loss[loss=0.229, simple_loss=0.3094, pruned_loss=0.07434, over 16681.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.3109, pruned_loss=0.07473, over 3237469.29 frames. ], batch size: 134, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:05:07,048 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55368.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:05:13,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.321e+02 2.807e+02 3.689e+02 1.502e+03, threshold=5.614e+02, percent-clipped=3.0 2023-04-28 12:05:52,757 INFO [train.py:904] (7/8) Epoch 6, batch 4650, loss[loss=0.2193, simple_loss=0.2962, pruned_loss=0.07121, over 17224.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3091, pruned_loss=0.07405, over 3242750.62 frames. ], batch size: 45, lr: 1.14e-02, grad_scale: 8.0 2023-04-28 12:07:00,149 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7589, 3.6075, 3.7691, 3.5617, 3.7213, 4.1735, 3.9776, 3.7169], device='cuda:7'), covar=tensor([0.1693, 0.1860, 0.1511, 0.2440, 0.2878, 0.1545, 0.1060, 0.2352], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0393, 0.0389, 0.0344, 0.0449, 0.0413, 0.0322, 0.0461], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 12:07:03,027 INFO [train.py:904] (7/8) Epoch 6, batch 4700, loss[loss=0.2379, simple_loss=0.3104, pruned_loss=0.08277, over 16403.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3065, pruned_loss=0.07298, over 3216911.74 frames. ], batch size: 146, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:07:30,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4510, 4.3795, 4.4345, 3.7643, 4.3898, 1.7745, 4.1437, 4.3357], device='cuda:7'), covar=tensor([0.0092, 0.0084, 0.0086, 0.0398, 0.0071, 0.1607, 0.0116, 0.0140], device='cuda:7'), in_proj_covar=tensor([0.0099, 0.0087, 0.0134, 0.0134, 0.0100, 0.0145, 0.0117, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:07:34,100 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.678e+02 2.601e+02 3.111e+02 3.691e+02 9.251e+02, threshold=6.221e+02, percent-clipped=2.0 2023-04-28 12:07:44,366 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1927, 5.5171, 5.1787, 5.3672, 4.9828, 4.7752, 4.9955, 5.5747], device='cuda:7'), covar=tensor([0.0642, 0.0665, 0.0852, 0.0433, 0.0583, 0.0548, 0.0665, 0.0712], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0527, 0.0446, 0.0338, 0.0327, 0.0344, 0.0431, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:08:04,799 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2326, 5.2917, 5.1837, 4.9562, 4.5108, 5.2032, 5.1785, 4.8890], device='cuda:7'), covar=tensor([0.0504, 0.0275, 0.0170, 0.0169, 0.0969, 0.0283, 0.0142, 0.0371], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0202, 0.0221, 0.0196, 0.0252, 0.0220, 0.0158, 0.0244], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:08:12,129 INFO [train.py:904] (7/8) Epoch 6, batch 4750, loss[loss=0.2036, simple_loss=0.2883, pruned_loss=0.05947, over 16765.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.3022, pruned_loss=0.07067, over 3219861.33 frames. ], batch size: 83, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:08:17,376 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55504.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:08:36,487 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55518.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 12:09:22,945 INFO [train.py:904] (7/8) Epoch 6, batch 4800, loss[loss=0.2017, simple_loss=0.2786, pruned_loss=0.06235, over 11624.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.2984, pruned_loss=0.06863, over 3213914.61 frames. ], batch size: 248, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:09:45,482 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55566.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:09:46,595 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:09:51,207 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 12:09:53,995 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.469e+02 2.878e+02 3.568e+02 5.612e+02, threshold=5.756e+02, percent-clipped=0.0 2023-04-28 12:10:35,259 INFO [train.py:904] (7/8) Epoch 6, batch 4850, loss[loss=0.2023, simple_loss=0.2945, pruned_loss=0.05508, over 16748.00 frames. ], tot_loss[loss=0.218, simple_loss=0.299, pruned_loss=0.06847, over 3199322.23 frames. ], batch size: 83, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:11:16,865 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8838, 1.7708, 2.2077, 3.0587, 1.8515, 2.2762, 2.0691, 1.8416], device='cuda:7'), covar=tensor([0.0746, 0.2797, 0.1174, 0.0517, 0.3347, 0.1570, 0.2413, 0.3054], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0329, 0.0272, 0.0303, 0.0373, 0.0342, 0.0297, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:11:48,127 INFO [train.py:904] (7/8) Epoch 6, batch 4900, loss[loss=0.2118, simple_loss=0.2905, pruned_loss=0.06658, over 12190.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.2991, pruned_loss=0.0681, over 3155526.91 frames. ], batch size: 246, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:11:59,529 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3283, 3.1750, 3.3541, 3.4775, 3.5464, 3.2114, 3.4749, 3.5590], device='cuda:7'), covar=tensor([0.0722, 0.0687, 0.0929, 0.0517, 0.0450, 0.1915, 0.0739, 0.0499], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0482, 0.0618, 0.0498, 0.0376, 0.0374, 0.0389, 0.0417], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:12:19,091 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.308e+02 2.929e+02 3.604e+02 9.107e+02, threshold=5.857e+02, percent-clipped=2.0 2023-04-28 12:12:59,384 INFO [train.py:904] (7/8) Epoch 6, batch 4950, loss[loss=0.2244, simple_loss=0.3097, pruned_loss=0.06954, over 16899.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.2996, pruned_loss=0.06765, over 3174168.45 frames. ], batch size: 116, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:13:11,117 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 12:13:13,995 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.50 vs. limit=5.0 2023-04-28 12:14:05,671 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55748.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:14:08,597 INFO [train.py:904] (7/8) Epoch 6, batch 5000, loss[loss=0.2312, simple_loss=0.3143, pruned_loss=0.07406, over 16905.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3013, pruned_loss=0.06789, over 3184776.32 frames. ], batch size: 109, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:14:38,598 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.723e+02 3.199e+02 3.755e+02 7.199e+02, threshold=6.398e+02, percent-clipped=5.0 2023-04-28 12:15:21,199 INFO [train.py:904] (7/8) Epoch 6, batch 5050, loss[loss=0.2074, simple_loss=0.296, pruned_loss=0.05943, over 16552.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.302, pruned_loss=0.06776, over 3201109.74 frames. ], batch size: 75, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:15:25,466 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55804.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:15:33,982 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55809.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:16:00,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1563, 1.7650, 2.0018, 3.6514, 1.7293, 2.3877, 2.0326, 1.8647], device='cuda:7'), covar=tensor([0.0756, 0.2607, 0.1374, 0.0349, 0.3317, 0.1595, 0.2325, 0.2743], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0338, 0.0278, 0.0313, 0.0379, 0.0350, 0.0304, 0.0400], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:16:28,820 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:16:31,562 INFO [train.py:904] (7/8) Epoch 6, batch 5100, loss[loss=0.2005, simple_loss=0.2893, pruned_loss=0.05589, over 16491.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.3005, pruned_loss=0.06712, over 3191399.22 frames. ], batch size: 75, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:16:32,974 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55852.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:16:54,915 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55867.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:17:04,483 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.683e+02 3.138e+02 3.930e+02 6.713e+02, threshold=6.277e+02, percent-clipped=2.0 2023-04-28 12:17:45,244 INFO [train.py:904] (7/8) Epoch 6, batch 5150, loss[loss=0.21, simple_loss=0.2933, pruned_loss=0.06335, over 16614.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.3006, pruned_loss=0.06652, over 3177028.10 frames. ], batch size: 62, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:17:57,086 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:18:06,166 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=55915.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:18:31,903 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-28 12:18:58,952 INFO [train.py:904] (7/8) Epoch 6, batch 5200, loss[loss=0.2214, simple_loss=0.295, pruned_loss=0.07387, over 12475.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2987, pruned_loss=0.06585, over 3183216.02 frames. ], batch size: 246, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:19:30,990 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.992e+02 2.558e+02 3.011e+02 3.644e+02 8.071e+02, threshold=6.021e+02, percent-clipped=1.0 2023-04-28 12:19:39,693 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7919, 1.3258, 1.5145, 1.7667, 1.7908, 1.9204, 1.4098, 1.8570], device='cuda:7'), covar=tensor([0.0099, 0.0202, 0.0095, 0.0136, 0.0124, 0.0071, 0.0194, 0.0049], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0150, 0.0130, 0.0132, 0.0136, 0.0097, 0.0144, 0.0086], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 12:20:15,830 INFO [train.py:904] (7/8) Epoch 6, batch 5250, loss[loss=0.198, simple_loss=0.2761, pruned_loss=0.0599, over 17018.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2961, pruned_loss=0.06568, over 3186133.02 frames. ], batch size: 55, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:21:28,024 INFO [train.py:904] (7/8) Epoch 6, batch 5300, loss[loss=0.1696, simple_loss=0.2542, pruned_loss=0.04249, over 16862.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2921, pruned_loss=0.06387, over 3200088.77 frames. ], batch size: 96, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:21:59,805 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.725e+02 2.658e+02 3.030e+02 3.663e+02 6.971e+02, threshold=6.060e+02, percent-clipped=3.0 2023-04-28 12:22:40,519 INFO [train.py:904] (7/8) Epoch 6, batch 5350, loss[loss=0.2087, simple_loss=0.2885, pruned_loss=0.06447, over 16732.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2903, pruned_loss=0.06285, over 3218458.50 frames. ], batch size: 62, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:22:45,586 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56104.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:22:45,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0802, 3.5936, 3.4341, 2.3573, 3.0787, 3.4505, 3.4531, 2.0818], device='cuda:7'), covar=tensor([0.0353, 0.0016, 0.0024, 0.0231, 0.0051, 0.0041, 0.0033, 0.0288], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0055, 0.0060, 0.0115, 0.0062, 0.0071, 0.0065, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 12:23:40,685 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9541, 3.4815, 3.3262, 2.3781, 2.9825, 3.3592, 3.3156, 1.9108], device='cuda:7'), covar=tensor([0.0362, 0.0021, 0.0030, 0.0230, 0.0052, 0.0052, 0.0037, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0056, 0.0060, 0.0116, 0.0062, 0.0071, 0.0065, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 12:23:44,961 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56145.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:23:52,611 INFO [train.py:904] (7/8) Epoch 6, batch 5400, loss[loss=0.2321, simple_loss=0.3114, pruned_loss=0.07641, over 15439.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2936, pruned_loss=0.06396, over 3216443.59 frames. ], batch size: 191, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:24:26,637 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.771e+02 2.610e+02 3.124e+02 3.605e+02 7.209e+02, threshold=6.248e+02, percent-clipped=3.0 2023-04-28 12:25:09,007 INFO [train.py:904] (7/8) Epoch 6, batch 5450, loss[loss=0.3072, simple_loss=0.3738, pruned_loss=0.1204, over 15321.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2977, pruned_loss=0.06629, over 3209198.18 frames. ], batch size: 190, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:25:13,312 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7623, 3.2262, 3.1098, 5.1388, 4.1744, 4.6420, 1.7560, 3.5144], device='cuda:7'), covar=tensor([0.1259, 0.0563, 0.0937, 0.0071, 0.0248, 0.0251, 0.1361, 0.0602], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0143, 0.0170, 0.0094, 0.0188, 0.0186, 0.0162, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 12:25:14,963 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56204.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:25:17,671 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:26:27,337 INFO [train.py:904] (7/8) Epoch 6, batch 5500, loss[loss=0.2802, simple_loss=0.339, pruned_loss=0.1107, over 11658.00 frames. ], tot_loss[loss=0.2254, simple_loss=0.3064, pruned_loss=0.07219, over 3188043.23 frames. ], batch size: 249, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:27:01,685 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.422e+02 4.497e+02 5.690e+02 1.317e+03, threshold=8.994e+02, percent-clipped=17.0 2023-04-28 12:27:46,061 INFO [train.py:904] (7/8) Epoch 6, batch 5550, loss[loss=0.2662, simple_loss=0.3434, pruned_loss=0.09448, over 16404.00 frames. ], tot_loss[loss=0.2365, simple_loss=0.3148, pruned_loss=0.07914, over 3150186.19 frames. ], batch size: 146, lr: 1.13e-02, grad_scale: 4.0 2023-04-28 12:28:30,586 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7181, 1.2445, 1.6772, 1.7136, 1.7544, 1.9091, 1.4493, 1.7238], device='cuda:7'), covar=tensor([0.0089, 0.0147, 0.0075, 0.0094, 0.0087, 0.0051, 0.0141, 0.0042], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0148, 0.0128, 0.0131, 0.0136, 0.0097, 0.0142, 0.0087], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 12:29:07,881 INFO [train.py:904] (7/8) Epoch 6, batch 5600, loss[loss=0.34, simple_loss=0.3766, pruned_loss=0.1517, over 10911.00 frames. ], tot_loss[loss=0.2455, simple_loss=0.3212, pruned_loss=0.08494, over 3110021.24 frames. ], batch size: 248, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:29:45,255 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.502e+02 4.330e+02 5.114e+02 6.903e+02 1.749e+03, threshold=1.023e+03, percent-clipped=9.0 2023-04-28 12:30:29,972 INFO [train.py:904] (7/8) Epoch 6, batch 5650, loss[loss=0.2672, simple_loss=0.3531, pruned_loss=0.09064, over 16882.00 frames. ], tot_loss[loss=0.2537, simple_loss=0.3275, pruned_loss=0.08993, over 3101397.00 frames. ], batch size: 96, lr: 1.13e-02, grad_scale: 8.0 2023-04-28 12:30:34,930 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56404.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:31:48,513 INFO [train.py:904] (7/8) Epoch 6, batch 5700, loss[loss=0.3897, simple_loss=0.4094, pruned_loss=0.185, over 11472.00 frames. ], tot_loss[loss=0.2579, simple_loss=0.3303, pruned_loss=0.09276, over 3082314.74 frames. ], batch size: 248, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:31:51,664 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=56452.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:31:54,249 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1164, 1.2508, 1.7873, 2.0543, 2.0366, 2.2819, 1.4933, 2.0839], device='cuda:7'), covar=tensor([0.0098, 0.0246, 0.0149, 0.0165, 0.0133, 0.0081, 0.0226, 0.0067], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0148, 0.0128, 0.0131, 0.0135, 0.0097, 0.0144, 0.0086], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 12:32:09,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5726, 2.7316, 2.4937, 4.4190, 3.4979, 4.0388, 1.4049, 2.9944], device='cuda:7'), covar=tensor([0.1411, 0.0590, 0.1128, 0.0093, 0.0292, 0.0327, 0.1466, 0.0759], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0145, 0.0171, 0.0095, 0.0194, 0.0190, 0.0164, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 12:32:16,777 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-28 12:32:25,513 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.692e+02 4.150e+02 5.062e+02 6.862e+02 1.724e+03, threshold=1.012e+03, percent-clipped=2.0 2023-04-28 12:32:36,631 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9154, 5.1802, 4.9484, 4.9823, 4.5416, 4.4240, 4.6938, 5.2746], device='cuda:7'), covar=tensor([0.0630, 0.0612, 0.0797, 0.0447, 0.0617, 0.0715, 0.0645, 0.0585], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0515, 0.0439, 0.0336, 0.0317, 0.0332, 0.0424, 0.0370], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:33:03,231 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4073, 3.5209, 1.6242, 3.7661, 2.4857, 3.7350, 1.8672, 2.7242], device='cuda:7'), covar=tensor([0.0167, 0.0270, 0.1778, 0.0045, 0.0729, 0.0393, 0.1474, 0.0571], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0153, 0.0174, 0.0080, 0.0162, 0.0185, 0.0184, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 12:33:08,909 INFO [train.py:904] (7/8) Epoch 6, batch 5750, loss[loss=0.2612, simple_loss=0.3189, pruned_loss=0.1017, over 11163.00 frames. ], tot_loss[loss=0.2609, simple_loss=0.333, pruned_loss=0.09437, over 3058621.48 frames. ], batch size: 249, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:33:09,344 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56501.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:33:09,678 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.36 vs. limit=5.0 2023-04-28 12:33:14,017 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56504.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:33:44,545 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56523.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 12:34:18,174 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-28 12:34:19,772 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5355, 3.7505, 2.8175, 2.1734, 2.8082, 2.2626, 3.8566, 3.6982], device='cuda:7'), covar=tensor([0.2494, 0.0733, 0.1401, 0.1845, 0.2021, 0.1539, 0.0484, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0277, 0.0247, 0.0265, 0.0248, 0.0280, 0.0201, 0.0245, 0.0257], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:34:30,326 INFO [train.py:904] (7/8) Epoch 6, batch 5800, loss[loss=0.2573, simple_loss=0.3386, pruned_loss=0.08799, over 16550.00 frames. ], tot_loss[loss=0.2594, simple_loss=0.3327, pruned_loss=0.09311, over 3052638.75 frames. ], batch size: 62, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:34:32,766 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=56552.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:35:05,967 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.365e+02 3.759e+02 4.783e+02 6.283e+02 1.634e+03, threshold=9.566e+02, percent-clipped=2.0 2023-04-28 12:35:23,438 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56584.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:35:49,160 INFO [train.py:904] (7/8) Epoch 6, batch 5850, loss[loss=0.2203, simple_loss=0.2997, pruned_loss=0.07048, over 17133.00 frames. ], tot_loss[loss=0.2559, simple_loss=0.33, pruned_loss=0.09093, over 3049243.14 frames. ], batch size: 47, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:36:07,780 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1104, 2.1963, 1.5444, 1.9150, 2.7189, 2.2850, 3.0472, 2.8901], device='cuda:7'), covar=tensor([0.0044, 0.0246, 0.0329, 0.0289, 0.0145, 0.0237, 0.0094, 0.0132], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0160, 0.0165, 0.0161, 0.0156, 0.0166, 0.0146, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:37:11,648 INFO [train.py:904] (7/8) Epoch 6, batch 5900, loss[loss=0.2255, simple_loss=0.3042, pruned_loss=0.07337, over 16729.00 frames. ], tot_loss[loss=0.2537, simple_loss=0.3283, pruned_loss=0.08958, over 3067599.76 frames. ], batch size: 124, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:37:52,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.523e+02 3.504e+02 4.467e+02 5.504e+02 1.096e+03, threshold=8.934e+02, percent-clipped=2.0 2023-04-28 12:38:33,951 INFO [train.py:904] (7/8) Epoch 6, batch 5950, loss[loss=0.2433, simple_loss=0.3265, pruned_loss=0.08008, over 16420.00 frames. ], tot_loss[loss=0.2526, simple_loss=0.3292, pruned_loss=0.08801, over 3082983.06 frames. ], batch size: 146, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:39:06,632 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1880, 5.6139, 5.2599, 5.3866, 4.9388, 4.7942, 5.0256, 5.6848], device='cuda:7'), covar=tensor([0.0661, 0.0590, 0.0817, 0.0430, 0.0612, 0.0620, 0.0668, 0.0639], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0511, 0.0440, 0.0337, 0.0315, 0.0333, 0.0424, 0.0371], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:39:12,889 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4745, 3.4980, 2.3540, 2.1750, 2.4688, 2.0607, 3.4249, 3.6115], device='cuda:7'), covar=tensor([0.2609, 0.0747, 0.1705, 0.1920, 0.2267, 0.1816, 0.0636, 0.0745], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0249, 0.0269, 0.0252, 0.0284, 0.0203, 0.0249, 0.0260], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:39:52,046 INFO [train.py:904] (7/8) Epoch 6, batch 6000, loss[loss=0.2505, simple_loss=0.3212, pruned_loss=0.08986, over 15311.00 frames. ], tot_loss[loss=0.2513, simple_loss=0.328, pruned_loss=0.08727, over 3110658.96 frames. ], batch size: 190, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:39:52,046 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 12:39:58,243 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6500, 3.6128, 3.6646, 2.0723, 3.8426, 3.8620, 3.1946, 3.1194], device='cuda:7'), covar=tensor([0.0610, 0.0127, 0.0162, 0.0906, 0.0053, 0.0063, 0.0250, 0.0283], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0092, 0.0080, 0.0139, 0.0072, 0.0080, 0.0114, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 12:40:01,519 INFO [train.py:938] (7/8) Epoch 6, validation: loss=0.18, simple_loss=0.2922, pruned_loss=0.03386, over 944034.00 frames. 2023-04-28 12:40:01,520 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 12:40:36,516 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.218e+02 3.558e+02 4.306e+02 5.339e+02 1.394e+03, threshold=8.611e+02, percent-clipped=5.0 2023-04-28 12:41:19,698 INFO [train.py:904] (7/8) Epoch 6, batch 6050, loss[loss=0.2404, simple_loss=0.3183, pruned_loss=0.08128, over 15325.00 frames. ], tot_loss[loss=0.2492, simple_loss=0.3261, pruned_loss=0.08614, over 3123769.00 frames. ], batch size: 190, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:41:20,634 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56801.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:41:23,969 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56803.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:42:34,241 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=56849.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:42:36,675 INFO [train.py:904] (7/8) Epoch 6, batch 6100, loss[loss=0.2207, simple_loss=0.3056, pruned_loss=0.06793, over 16538.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3254, pruned_loss=0.08515, over 3117438.14 frames. ], batch size: 75, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:42:44,446 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4602, 3.4636, 3.2589, 3.1050, 2.9982, 3.3717, 3.2281, 3.1688], device='cuda:7'), covar=tensor([0.0460, 0.0333, 0.0184, 0.0169, 0.0512, 0.0285, 0.0977, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0211, 0.0224, 0.0197, 0.0251, 0.0226, 0.0158, 0.0253], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:42:58,758 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56864.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:43:13,028 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.259e+02 3.305e+02 4.282e+02 5.206e+02 1.267e+03, threshold=8.564e+02, percent-clipped=3.0 2023-04-28 12:43:17,168 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9995, 3.2542, 3.5106, 3.4867, 3.4817, 3.2411, 3.3101, 3.3139], device='cuda:7'), covar=tensor([0.0376, 0.0509, 0.0401, 0.0444, 0.0471, 0.0415, 0.0780, 0.0473], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0252, 0.0255, 0.0253, 0.0307, 0.0274, 0.0374, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 12:43:21,510 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56879.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 12:43:56,261 INFO [train.py:904] (7/8) Epoch 6, batch 6150, loss[loss=0.1963, simple_loss=0.2751, pruned_loss=0.05878, over 16605.00 frames. ], tot_loss[loss=0.2457, simple_loss=0.3229, pruned_loss=0.08426, over 3104857.21 frames. ], batch size: 57, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:43:58,298 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2023-04-28 12:44:06,758 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0840, 5.4187, 5.0891, 5.1900, 4.7478, 4.6942, 4.8928, 5.4878], device='cuda:7'), covar=tensor([0.0694, 0.0668, 0.0927, 0.0476, 0.0645, 0.0660, 0.0642, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0409, 0.0531, 0.0453, 0.0349, 0.0327, 0.0344, 0.0436, 0.0384], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:44:08,606 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56908.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:45:17,272 INFO [train.py:904] (7/8) Epoch 6, batch 6200, loss[loss=0.2364, simple_loss=0.311, pruned_loss=0.08087, over 16500.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.3211, pruned_loss=0.08404, over 3096932.84 frames. ], batch size: 68, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:45:46,232 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56969.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:45:53,732 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.179e+02 3.479e+02 4.387e+02 5.634e+02 1.000e+03, threshold=8.774e+02, percent-clipped=2.0 2023-04-28 12:46:22,789 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56993.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:46:26,556 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56995.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:46:34,337 INFO [train.py:904] (7/8) Epoch 6, batch 6250, loss[loss=0.2457, simple_loss=0.3322, pruned_loss=0.07959, over 16687.00 frames. ], tot_loss[loss=0.2451, simple_loss=0.3214, pruned_loss=0.08444, over 3083231.70 frames. ], batch size: 57, lr: 1.12e-02, grad_scale: 4.0 2023-04-28 12:47:45,478 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0437, 2.3337, 2.4120, 2.9323, 2.4255, 3.2975, 1.8505, 2.7299], device='cuda:7'), covar=tensor([0.1032, 0.0424, 0.0834, 0.0097, 0.0176, 0.0335, 0.1094, 0.0619], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0143, 0.0170, 0.0095, 0.0198, 0.0190, 0.0164, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 12:47:50,603 INFO [train.py:904] (7/8) Epoch 6, batch 6300, loss[loss=0.2196, simple_loss=0.3009, pruned_loss=0.06909, over 16732.00 frames. ], tot_loss[loss=0.2437, simple_loss=0.3206, pruned_loss=0.08337, over 3086615.80 frames. ], batch size: 124, lr: 1.12e-02, grad_scale: 4.0 2023-04-28 12:47:56,659 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57054.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:47:59,837 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57056.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:48:29,806 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.167e+02 3.660e+02 4.420e+02 5.766e+02 1.393e+03, threshold=8.841e+02, percent-clipped=5.0 2023-04-28 12:49:09,392 INFO [train.py:904] (7/8) Epoch 6, batch 6350, loss[loss=0.2299, simple_loss=0.3051, pruned_loss=0.07737, over 17002.00 frames. ], tot_loss[loss=0.2456, simple_loss=0.3216, pruned_loss=0.0848, over 3067783.97 frames. ], batch size: 50, lr: 1.12e-02, grad_scale: 4.0 2023-04-28 12:49:43,140 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 12:50:26,705 INFO [train.py:904] (7/8) Epoch 6, batch 6400, loss[loss=0.2498, simple_loss=0.3166, pruned_loss=0.09148, over 16677.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3222, pruned_loss=0.0862, over 3059159.87 frames. ], batch size: 57, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:50:38,737 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57159.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:50:57,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9898, 1.9516, 2.2785, 3.2297, 2.0843, 2.5206, 2.2534, 2.0499], device='cuda:7'), covar=tensor([0.0653, 0.2322, 0.1071, 0.0414, 0.2759, 0.1192, 0.1920, 0.2042], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0336, 0.0277, 0.0311, 0.0383, 0.0347, 0.0303, 0.0397], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:51:01,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.642e+02 4.267e+02 5.177e+02 6.433e+02 1.516e+03, threshold=1.035e+03, percent-clipped=6.0 2023-04-28 12:51:05,166 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3133, 1.8580, 1.6302, 1.6320, 2.1880, 1.9449, 2.2142, 2.3187], device='cuda:7'), covar=tensor([0.0052, 0.0179, 0.0234, 0.0227, 0.0128, 0.0158, 0.0124, 0.0111], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0159, 0.0162, 0.0161, 0.0155, 0.0164, 0.0148, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:51:08,492 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57179.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 12:51:19,365 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57186.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:51:36,923 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57197.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:51:42,181 INFO [train.py:904] (7/8) Epoch 6, batch 6450, loss[loss=0.2316, simple_loss=0.2911, pruned_loss=0.08607, over 11512.00 frames. ], tot_loss[loss=0.246, simple_loss=0.3219, pruned_loss=0.08508, over 3059507.07 frames. ], batch size: 247, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:52:26,755 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57227.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 12:52:40,581 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7796, 3.1570, 3.1146, 1.8632, 2.8499, 3.0590, 3.0927, 1.7020], device='cuda:7'), covar=tensor([0.0372, 0.0024, 0.0034, 0.0275, 0.0051, 0.0058, 0.0036, 0.0319], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0054, 0.0059, 0.0115, 0.0062, 0.0072, 0.0065, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 12:52:52,973 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57244.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:52:57,852 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57247.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:53:03,828 INFO [train.py:904] (7/8) Epoch 6, batch 6500, loss[loss=0.2458, simple_loss=0.3198, pruned_loss=0.08589, over 16904.00 frames. ], tot_loss[loss=0.2433, simple_loss=0.3192, pruned_loss=0.08372, over 3081499.20 frames. ], batch size: 109, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:53:15,535 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57258.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 12:53:25,368 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57264.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:53:41,778 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.641e+02 4.251e+02 5.070e+02 1.044e+03, threshold=8.501e+02, percent-clipped=1.0 2023-04-28 12:54:25,252 INFO [train.py:904] (7/8) Epoch 6, batch 6550, loss[loss=0.2603, simple_loss=0.3499, pruned_loss=0.08533, over 16272.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.323, pruned_loss=0.08513, over 3088770.35 frames. ], batch size: 165, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:54:27,473 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9242, 3.8465, 3.8530, 2.9544, 3.8573, 1.6961, 3.5889, 3.5140], device='cuda:7'), covar=tensor([0.0101, 0.0094, 0.0119, 0.0391, 0.0075, 0.2166, 0.0118, 0.0254], device='cuda:7'), in_proj_covar=tensor([0.0097, 0.0085, 0.0131, 0.0129, 0.0098, 0.0148, 0.0113, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:54:32,556 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57305.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:55:40,191 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57349.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:55:42,178 INFO [train.py:904] (7/8) Epoch 6, batch 6600, loss[loss=0.2271, simple_loss=0.3113, pruned_loss=0.07143, over 16908.00 frames. ], tot_loss[loss=0.2485, simple_loss=0.3252, pruned_loss=0.08595, over 3087950.36 frames. ], batch size: 96, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:55:43,216 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57351.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:55:56,480 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5782, 4.8590, 4.5481, 4.5797, 4.2611, 4.2561, 4.3690, 4.8154], device='cuda:7'), covar=tensor([0.0835, 0.0690, 0.0958, 0.0575, 0.0670, 0.0973, 0.0752, 0.0792], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0526, 0.0447, 0.0345, 0.0328, 0.0345, 0.0431, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:56:18,771 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.503e+02 3.780e+02 4.683e+02 5.878e+02 1.016e+03, threshold=9.365e+02, percent-clipped=5.0 2023-04-28 12:56:56,523 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57398.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:57:00,530 INFO [train.py:904] (7/8) Epoch 6, batch 6650, loss[loss=0.2169, simple_loss=0.2981, pruned_loss=0.06785, over 16809.00 frames. ], tot_loss[loss=0.2485, simple_loss=0.325, pruned_loss=0.08602, over 3103463.44 frames. ], batch size: 83, lr: 1.12e-02, grad_scale: 8.0 2023-04-28 12:57:38,733 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9208, 2.6762, 2.6224, 1.8391, 2.3843, 2.5853, 2.6173, 1.8112], device='cuda:7'), covar=tensor([0.0282, 0.0029, 0.0039, 0.0217, 0.0067, 0.0065, 0.0043, 0.0268], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0055, 0.0059, 0.0115, 0.0062, 0.0073, 0.0065, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 12:58:15,922 INFO [train.py:904] (7/8) Epoch 6, batch 6700, loss[loss=0.3043, simple_loss=0.3529, pruned_loss=0.1278, over 11494.00 frames. ], tot_loss[loss=0.2478, simple_loss=0.3236, pruned_loss=0.08601, over 3082693.07 frames. ], batch size: 247, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 12:58:20,074 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6828, 2.0545, 1.6428, 1.8953, 2.5275, 2.2152, 2.8555, 2.7574], device='cuda:7'), covar=tensor([0.0057, 0.0225, 0.0286, 0.0264, 0.0130, 0.0226, 0.0089, 0.0111], device='cuda:7'), in_proj_covar=tensor([0.0087, 0.0161, 0.0165, 0.0163, 0.0157, 0.0167, 0.0150, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:58:29,514 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:58:29,677 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:58:34,680 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57462.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:58:34,758 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5332, 3.5038, 2.7320, 2.2360, 2.6107, 2.3112, 3.6148, 3.6060], device='cuda:7'), covar=tensor([0.2307, 0.0733, 0.1385, 0.1717, 0.1748, 0.1411, 0.0509, 0.0713], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0248, 0.0265, 0.0250, 0.0280, 0.0200, 0.0247, 0.0258], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 12:58:54,099 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.152e+02 3.718e+02 4.462e+02 5.305e+02 7.989e+02, threshold=8.923e+02, percent-clipped=0.0 2023-04-28 12:58:55,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1766, 4.4013, 2.0729, 4.9354, 2.8810, 4.8492, 2.3835, 2.9961], device='cuda:7'), covar=tensor([0.0139, 0.0216, 0.1626, 0.0032, 0.0787, 0.0275, 0.1453, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0156, 0.0179, 0.0084, 0.0164, 0.0190, 0.0191, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 12:59:24,887 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57495.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:59:34,130 INFO [train.py:904] (7/8) Epoch 6, batch 6750, loss[loss=0.2909, simple_loss=0.3426, pruned_loss=0.1196, over 11650.00 frames. ], tot_loss[loss=0.2476, simple_loss=0.3228, pruned_loss=0.08624, over 3080603.10 frames. ], batch size: 248, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 12:59:43,647 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57507.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 12:59:58,257 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 13:00:08,084 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57523.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:00:36,615 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57542.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:00:49,728 INFO [train.py:904] (7/8) Epoch 6, batch 6800, loss[loss=0.245, simple_loss=0.325, pruned_loss=0.08255, over 16947.00 frames. ], tot_loss[loss=0.2475, simple_loss=0.323, pruned_loss=0.08598, over 3088009.51 frames. ], batch size: 109, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:00:54,412 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57553.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 13:00:58,745 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57556.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:01:10,580 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57564.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:01:27,507 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.272e+02 3.624e+02 4.504e+02 5.898e+02 1.165e+03, threshold=9.008e+02, percent-clipped=3.0 2023-04-28 13:02:06,902 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57600.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:02:07,772 INFO [train.py:904] (7/8) Epoch 6, batch 6850, loss[loss=0.3033, simple_loss=0.3579, pruned_loss=0.1244, over 12122.00 frames. ], tot_loss[loss=0.2473, simple_loss=0.3232, pruned_loss=0.08571, over 3095794.36 frames. ], batch size: 246, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:02:24,717 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57612.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:02:49,250 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2396, 3.6403, 3.8352, 2.5397, 3.7193, 3.7794, 3.7413, 1.8970], device='cuda:7'), covar=tensor([0.0357, 0.0028, 0.0026, 0.0245, 0.0036, 0.0057, 0.0030, 0.0336], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0055, 0.0058, 0.0115, 0.0062, 0.0073, 0.0065, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 13:03:22,411 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57649.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:03:24,933 INFO [train.py:904] (7/8) Epoch 6, batch 6900, loss[loss=0.363, simple_loss=0.3899, pruned_loss=0.168, over 11127.00 frames. ], tot_loss[loss=0.2477, simple_loss=0.3254, pruned_loss=0.08503, over 3110091.71 frames. ], batch size: 246, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:03:25,325 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57651.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:04:02,714 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.992e+02 3.443e+02 4.296e+02 5.667e+02 1.150e+03, threshold=8.592e+02, percent-clipped=3.0 2023-04-28 13:04:37,931 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57697.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:04:41,445 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57699.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:04:43,543 INFO [train.py:904] (7/8) Epoch 6, batch 6950, loss[loss=0.3089, simple_loss=0.3564, pruned_loss=0.1307, over 11177.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.327, pruned_loss=0.08666, over 3105942.01 frames. ], batch size: 246, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:04:48,515 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3489, 3.3377, 2.6181, 2.1784, 2.3721, 2.1104, 3.3658, 3.4249], device='cuda:7'), covar=tensor([0.2526, 0.0803, 0.1489, 0.1854, 0.1938, 0.1621, 0.0524, 0.0771], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0248, 0.0266, 0.0250, 0.0279, 0.0200, 0.0245, 0.0257], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:05:34,171 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57733.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:05:39,676 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2158, 1.8884, 2.0394, 3.7340, 1.8594, 2.5765, 2.0584, 2.0564], device='cuda:7'), covar=tensor([0.0760, 0.2746, 0.1500, 0.0362, 0.3396, 0.1573, 0.2496, 0.2587], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0338, 0.0280, 0.0313, 0.0387, 0.0352, 0.0305, 0.0400], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:05:53,911 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9938, 3.9853, 3.9179, 3.3651, 3.9292, 1.6493, 3.7010, 3.6580], device='cuda:7'), covar=tensor([0.0080, 0.0063, 0.0107, 0.0265, 0.0068, 0.1976, 0.0097, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0098, 0.0084, 0.0130, 0.0129, 0.0098, 0.0148, 0.0113, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:06:01,332 INFO [train.py:904] (7/8) Epoch 6, batch 7000, loss[loss=0.2324, simple_loss=0.3246, pruned_loss=0.07012, over 16721.00 frames. ], tot_loss[loss=0.2504, simple_loss=0.3279, pruned_loss=0.08641, over 3103015.97 frames. ], batch size: 57, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:06:06,476 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57754.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:06:38,245 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 3.231e+02 4.685e+02 6.535e+02 1.606e+03, threshold=9.370e+02, percent-clipped=8.0 2023-04-28 13:07:07,581 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57794.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:07:09,558 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8324, 2.0001, 2.2744, 3.1439, 2.1279, 2.4242, 2.2783, 2.0684], device='cuda:7'), covar=tensor([0.0655, 0.2180, 0.1150, 0.0426, 0.2626, 0.1297, 0.1928, 0.2284], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0339, 0.0280, 0.0313, 0.0387, 0.0352, 0.0304, 0.0400], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:07:17,567 INFO [train.py:904] (7/8) Epoch 6, batch 7050, loss[loss=0.2843, simple_loss=0.3422, pruned_loss=0.1132, over 11582.00 frames. ], tot_loss[loss=0.2493, simple_loss=0.3277, pruned_loss=0.08548, over 3120122.25 frames. ], batch size: 248, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:07:44,470 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57818.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:08:20,807 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57842.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:08:27,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7070, 5.0461, 5.0987, 5.1085, 5.0378, 5.6385, 5.2131, 5.0201], device='cuda:7'), covar=tensor([0.0884, 0.1585, 0.1448, 0.1688, 0.2314, 0.0860, 0.1210, 0.2245], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0402, 0.0406, 0.0355, 0.0463, 0.0434, 0.0335, 0.0477], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 13:08:34,161 INFO [train.py:904] (7/8) Epoch 6, batch 7100, loss[loss=0.2344, simple_loss=0.31, pruned_loss=0.07943, over 15486.00 frames. ], tot_loss[loss=0.248, simple_loss=0.3259, pruned_loss=0.08504, over 3120631.21 frames. ], batch size: 191, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:08:35,327 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57851.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:08:37,977 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57853.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 13:08:41,431 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-28 13:09:03,044 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 13:09:10,528 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.825e+02 4.803e+02 6.106e+02 1.425e+03, threshold=9.606e+02, percent-clipped=4.0 2023-04-28 13:09:32,946 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57890.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:09:40,554 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4950, 4.4586, 4.3111, 4.1521, 3.8613, 4.3583, 4.2487, 4.1062], device='cuda:7'), covar=tensor([0.0483, 0.0268, 0.0214, 0.0197, 0.0862, 0.0345, 0.0365, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0210, 0.0220, 0.0190, 0.0247, 0.0226, 0.0160, 0.0254], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:09:48,848 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57900.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:09:49,599 INFO [train.py:904] (7/8) Epoch 6, batch 7150, loss[loss=0.2921, simple_loss=0.3451, pruned_loss=0.1195, over 11378.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.3236, pruned_loss=0.08458, over 3110638.60 frames. ], batch size: 247, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:09:49,972 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57901.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:10:34,825 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9008, 4.1469, 3.9427, 3.9812, 3.6554, 3.7545, 3.8893, 4.0979], device='cuda:7'), covar=tensor([0.0716, 0.0781, 0.0935, 0.0511, 0.0613, 0.1291, 0.0623, 0.0898], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0519, 0.0448, 0.0336, 0.0319, 0.0343, 0.0421, 0.0372], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:11:00,230 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=57948.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:11:04,896 INFO [train.py:904] (7/8) Epoch 6, batch 7200, loss[loss=0.2375, simple_loss=0.3101, pruned_loss=0.08249, over 11774.00 frames. ], tot_loss[loss=0.2434, simple_loss=0.3212, pruned_loss=0.08275, over 3103860.16 frames. ], batch size: 250, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:11:10,575 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 13:11:41,801 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.198e+02 3.384e+02 4.240e+02 5.535e+02 1.102e+03, threshold=8.480e+02, percent-clipped=1.0 2023-04-28 13:11:45,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0646, 3.2186, 3.5242, 3.4922, 3.4738, 3.2329, 3.2938, 3.3571], device='cuda:7'), covar=tensor([0.0340, 0.0532, 0.0326, 0.0383, 0.0457, 0.0418, 0.0727, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0256, 0.0250, 0.0257, 0.0254, 0.0301, 0.0273, 0.0373, 0.0222], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 13:12:28,080 INFO [train.py:904] (7/8) Epoch 6, batch 7250, loss[loss=0.2234, simple_loss=0.2975, pruned_loss=0.07469, over 17248.00 frames. ], tot_loss[loss=0.2409, simple_loss=0.319, pruned_loss=0.08143, over 3095599.53 frames. ], batch size: 52, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:13:22,254 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6564, 3.0758, 2.8570, 4.3812, 3.3320, 4.0795, 1.5856, 2.9639], device='cuda:7'), covar=tensor([0.1346, 0.0541, 0.0983, 0.0129, 0.0332, 0.0396, 0.1425, 0.0778], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0143, 0.0169, 0.0094, 0.0195, 0.0191, 0.0164, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 13:13:44,370 INFO [train.py:904] (7/8) Epoch 6, batch 7300, loss[loss=0.3052, simple_loss=0.3624, pruned_loss=0.124, over 11568.00 frames. ], tot_loss[loss=0.2409, simple_loss=0.3187, pruned_loss=0.0816, over 3085882.02 frames. ], batch size: 246, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:13:49,474 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58054.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:14:21,945 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.373e+02 3.804e+02 4.653e+02 5.753e+02 1.061e+03, threshold=9.306e+02, percent-clipped=1.0 2023-04-28 13:14:22,497 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1935, 4.1677, 3.9730, 3.2964, 4.0808, 1.5824, 3.8333, 3.6664], device='cuda:7'), covar=tensor([0.0059, 0.0048, 0.0121, 0.0289, 0.0054, 0.2093, 0.0090, 0.0143], device='cuda:7'), in_proj_covar=tensor([0.0098, 0.0084, 0.0130, 0.0129, 0.0098, 0.0150, 0.0113, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:14:43,462 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58089.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:15:02,133 INFO [train.py:904] (7/8) Epoch 6, batch 7350, loss[loss=0.2284, simple_loss=0.3104, pruned_loss=0.07324, over 16856.00 frames. ], tot_loss[loss=0.2413, simple_loss=0.3186, pruned_loss=0.08201, over 3064783.56 frames. ], batch size: 96, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:15:02,520 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5517, 4.8345, 4.8944, 4.8544, 4.8745, 5.3221, 4.9132, 4.7429], device='cuda:7'), covar=tensor([0.0962, 0.1420, 0.1291, 0.1560, 0.1919, 0.0893, 0.1165, 0.2035], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0405, 0.0410, 0.0355, 0.0463, 0.0441, 0.0336, 0.0479], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 13:15:03,785 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58102.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:15:28,127 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58118.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:15:56,452 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 13:15:59,411 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0402, 2.2836, 1.8667, 1.9558, 2.6501, 2.3963, 2.9602, 2.8860], device='cuda:7'), covar=tensor([0.0039, 0.0211, 0.0250, 0.0257, 0.0131, 0.0188, 0.0099, 0.0128], device='cuda:7'), in_proj_covar=tensor([0.0086, 0.0160, 0.0165, 0.0162, 0.0156, 0.0166, 0.0146, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:16:04,920 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58141.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:09,249 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:19,338 INFO [train.py:904] (7/8) Epoch 6, batch 7400, loss[loss=0.233, simple_loss=0.315, pruned_loss=0.07548, over 17238.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3198, pruned_loss=0.08206, over 3091619.27 frames. ], batch size: 45, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:16:19,755 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58151.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:19,890 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5047, 3.8518, 4.0858, 1.8502, 4.3335, 4.2966, 3.2005, 3.0528], device='cuda:7'), covar=tensor([0.0756, 0.0117, 0.0106, 0.1175, 0.0039, 0.0058, 0.0280, 0.0430], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0091, 0.0082, 0.0140, 0.0072, 0.0081, 0.0116, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 13:16:37,231 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0259, 4.1917, 1.8619, 4.6018, 2.7301, 4.5127, 2.2258, 2.8162], device='cuda:7'), covar=tensor([0.0148, 0.0259, 0.1865, 0.0047, 0.0812, 0.0346, 0.1519, 0.0696], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0152, 0.0180, 0.0084, 0.0162, 0.0188, 0.0189, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 13:16:43,940 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58166.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:16:45,171 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3171, 3.2232, 3.3103, 3.4309, 3.4390, 3.2006, 3.4164, 3.4642], device='cuda:7'), covar=tensor([0.0855, 0.0719, 0.1102, 0.0558, 0.0651, 0.1553, 0.0783, 0.0580], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0494, 0.0635, 0.0510, 0.0388, 0.0380, 0.0403, 0.0428], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:16:48,611 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6223, 2.1428, 2.2985, 4.2996, 1.9321, 2.9374, 2.4046, 2.3885], device='cuda:7'), covar=tensor([0.0601, 0.2350, 0.1384, 0.0275, 0.3181, 0.1305, 0.2085, 0.2323], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0341, 0.0282, 0.0317, 0.0388, 0.0355, 0.0305, 0.0402], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:16:57,164 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.315e+02 3.701e+02 4.363e+02 5.153e+02 1.099e+03, threshold=8.726e+02, percent-clipped=1.0 2023-04-28 13:17:34,943 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58199.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:17:37,753 INFO [train.py:904] (7/8) Epoch 6, batch 7450, loss[loss=0.3077, simple_loss=0.3523, pruned_loss=0.1316, over 11663.00 frames. ], tot_loss[loss=0.2442, simple_loss=0.321, pruned_loss=0.08368, over 3087164.91 frames. ], batch size: 248, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:17:40,431 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58202.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:17:45,735 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58205.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:18:01,686 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-28 13:18:26,957 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4986, 4.4784, 4.3543, 3.7003, 4.2942, 1.6896, 4.1256, 4.1979], device='cuda:7'), covar=tensor([0.0065, 0.0060, 0.0103, 0.0280, 0.0066, 0.1859, 0.0090, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0097, 0.0083, 0.0128, 0.0126, 0.0096, 0.0147, 0.0112, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:18:59,289 INFO [train.py:904] (7/8) Epoch 6, batch 7500, loss[loss=0.2932, simple_loss=0.3526, pruned_loss=0.1169, over 11619.00 frames. ], tot_loss[loss=0.2449, simple_loss=0.3223, pruned_loss=0.08376, over 3082354.55 frames. ], batch size: 246, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:19:07,053 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-28 13:19:39,439 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.280e+02 3.913e+02 4.700e+02 5.850e+02 1.352e+03, threshold=9.401e+02, percent-clipped=2.0 2023-04-28 13:20:17,504 INFO [train.py:904] (7/8) Epoch 6, batch 7550, loss[loss=0.2634, simple_loss=0.3323, pruned_loss=0.09729, over 16319.00 frames. ], tot_loss[loss=0.2452, simple_loss=0.3216, pruned_loss=0.0844, over 3055445.82 frames. ], batch size: 146, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:20:20,163 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-28 13:20:27,894 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7625, 3.4433, 2.7252, 5.0014, 4.2531, 4.4699, 1.6738, 3.5348], device='cuda:7'), covar=tensor([0.1319, 0.0523, 0.1146, 0.0109, 0.0318, 0.0332, 0.1345, 0.0649], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0143, 0.0170, 0.0094, 0.0195, 0.0190, 0.0164, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 13:20:47,934 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0435, 5.4263, 5.5356, 5.4093, 5.4627, 5.9423, 5.4329, 5.2460], device='cuda:7'), covar=tensor([0.0792, 0.1344, 0.1414, 0.1598, 0.2177, 0.0862, 0.1266, 0.2162], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0407, 0.0415, 0.0359, 0.0470, 0.0446, 0.0341, 0.0481], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 13:21:05,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9691, 2.3637, 2.4798, 2.9631, 2.6081, 3.2551, 1.6661, 2.8821], device='cuda:7'), covar=tensor([0.1059, 0.0427, 0.0838, 0.0104, 0.0194, 0.0353, 0.1141, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0144, 0.0171, 0.0094, 0.0196, 0.0191, 0.0165, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 13:21:33,638 INFO [train.py:904] (7/8) Epoch 6, batch 7600, loss[loss=0.3063, simple_loss=0.3441, pruned_loss=0.1343, over 11215.00 frames. ], tot_loss[loss=0.2446, simple_loss=0.3204, pruned_loss=0.08445, over 3073137.96 frames. ], batch size: 246, lr: 1.11e-02, grad_scale: 8.0 2023-04-28 13:22:14,297 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 3.777e+02 4.488e+02 5.757e+02 1.119e+03, threshold=8.975e+02, percent-clipped=2.0 2023-04-28 13:22:33,793 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58389.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:22:52,241 INFO [train.py:904] (7/8) Epoch 6, batch 7650, loss[loss=0.3056, simple_loss=0.3549, pruned_loss=0.1282, over 11712.00 frames. ], tot_loss[loss=0.2452, simple_loss=0.3208, pruned_loss=0.08485, over 3083198.90 frames. ], batch size: 248, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:23:08,020 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8371, 4.8268, 4.6635, 4.5797, 4.2383, 4.7202, 4.6274, 4.4376], device='cuda:7'), covar=tensor([0.0486, 0.0271, 0.0206, 0.0172, 0.0841, 0.0378, 0.0249, 0.0476], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0210, 0.0220, 0.0191, 0.0248, 0.0227, 0.0161, 0.0254], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:23:49,829 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58437.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:24:11,096 INFO [train.py:904] (7/8) Epoch 6, batch 7700, loss[loss=0.2612, simple_loss=0.3344, pruned_loss=0.09401, over 15371.00 frames. ], tot_loss[loss=0.2465, simple_loss=0.3216, pruned_loss=0.08572, over 3087251.79 frames. ], batch size: 190, lr: 1.11e-02, grad_scale: 4.0 2023-04-28 13:24:51,995 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 3.672e+02 4.554e+02 5.723e+02 8.873e+02, threshold=9.107e+02, percent-clipped=1.0 2023-04-28 13:25:23,661 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58497.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:25:28,826 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58500.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:25:29,685 INFO [train.py:904] (7/8) Epoch 6, batch 7750, loss[loss=0.2908, simple_loss=0.3427, pruned_loss=0.1194, over 11536.00 frames. ], tot_loss[loss=0.2464, simple_loss=0.3215, pruned_loss=0.08562, over 3086368.48 frames. ], batch size: 248, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:26:46,080 INFO [train.py:904] (7/8) Epoch 6, batch 7800, loss[loss=0.2442, simple_loss=0.3272, pruned_loss=0.08058, over 16259.00 frames. ], tot_loss[loss=0.247, simple_loss=0.3222, pruned_loss=0.08591, over 3085570.09 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:27:26,213 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.190e+02 3.955e+02 4.761e+02 5.914e+02 1.211e+03, threshold=9.523e+02, percent-clipped=2.0 2023-04-28 13:27:30,295 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8878, 3.5466, 3.1475, 1.6749, 2.7824, 2.3276, 3.3242, 3.4870], device='cuda:7'), covar=tensor([0.0282, 0.0475, 0.0549, 0.1823, 0.0742, 0.0893, 0.0630, 0.0690], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0128, 0.0152, 0.0140, 0.0133, 0.0125, 0.0137, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 13:28:01,713 INFO [train.py:904] (7/8) Epoch 6, batch 7850, loss[loss=0.2229, simple_loss=0.3042, pruned_loss=0.07079, over 16612.00 frames. ], tot_loss[loss=0.2486, simple_loss=0.324, pruned_loss=0.08659, over 3076783.43 frames. ], batch size: 62, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:29:11,662 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6725, 1.5566, 1.9760, 2.6441, 2.4169, 2.9695, 1.6467, 2.7556], device='cuda:7'), covar=tensor([0.0084, 0.0286, 0.0196, 0.0136, 0.0134, 0.0090, 0.0277, 0.0082], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0147, 0.0128, 0.0128, 0.0135, 0.0098, 0.0144, 0.0084], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 13:29:14,711 INFO [train.py:904] (7/8) Epoch 6, batch 7900, loss[loss=0.2488, simple_loss=0.3285, pruned_loss=0.08457, over 15434.00 frames. ], tot_loss[loss=0.2478, simple_loss=0.3234, pruned_loss=0.08613, over 3068525.32 frames. ], batch size: 191, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:29:53,309 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.999e+02 3.363e+02 4.237e+02 5.568e+02 1.033e+03, threshold=8.474e+02, percent-clipped=1.0 2023-04-28 13:30:02,457 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58681.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:30:31,234 INFO [train.py:904] (7/8) Epoch 6, batch 7950, loss[loss=0.2334, simple_loss=0.3117, pruned_loss=0.07755, over 16243.00 frames. ], tot_loss[loss=0.2487, simple_loss=0.3238, pruned_loss=0.08679, over 3058431.12 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:30:36,815 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3654, 4.3834, 4.1957, 4.0837, 3.7860, 4.2966, 4.1417, 3.9521], device='cuda:7'), covar=tensor([0.0530, 0.0356, 0.0240, 0.0196, 0.0866, 0.0371, 0.0415, 0.0524], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0215, 0.0224, 0.0195, 0.0253, 0.0230, 0.0164, 0.0258], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:31:33,518 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58742.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:31:46,450 INFO [train.py:904] (7/8) Epoch 6, batch 8000, loss[loss=0.321, simple_loss=0.367, pruned_loss=0.1375, over 11612.00 frames. ], tot_loss[loss=0.2502, simple_loss=0.3246, pruned_loss=0.08791, over 3031706.17 frames. ], batch size: 248, lr: 1.10e-02, grad_scale: 8.0 2023-04-28 13:32:25,085 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58775.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:32:27,145 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.294e+02 3.253e+02 3.971e+02 4.697e+02 8.095e+02, threshold=7.943e+02, percent-clipped=0.0 2023-04-28 13:32:52,242 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-28 13:32:57,932 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58797.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:33:02,629 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58800.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:33:03,024 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-04-28 13:33:03,399 INFO [train.py:904] (7/8) Epoch 6, batch 8050, loss[loss=0.2088, simple_loss=0.2972, pruned_loss=0.06017, over 16822.00 frames. ], tot_loss[loss=0.2479, simple_loss=0.3234, pruned_loss=0.0862, over 3053751.91 frames. ], batch size: 102, lr: 1.10e-02, grad_scale: 8.0 2023-04-28 13:33:26,276 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 13:33:43,375 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 13:33:58,148 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58836.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:34:13,201 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58845.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:34:16,881 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=58848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:34:20,747 INFO [train.py:904] (7/8) Epoch 6, batch 8100, loss[loss=0.2261, simple_loss=0.3112, pruned_loss=0.07053, over 16796.00 frames. ], tot_loss[loss=0.2458, simple_loss=0.3221, pruned_loss=0.08473, over 3085183.54 frames. ], batch size: 96, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:34:26,284 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6590, 2.6460, 1.7282, 2.7146, 2.1565, 2.7235, 1.9973, 2.3352], device='cuda:7'), covar=tensor([0.0204, 0.0377, 0.1184, 0.0099, 0.0577, 0.0577, 0.1061, 0.0529], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0153, 0.0177, 0.0086, 0.0161, 0.0189, 0.0189, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 13:35:03,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.302e+02 3.498e+02 4.445e+02 5.357e+02 1.535e+03, threshold=8.891e+02, percent-clipped=10.0 2023-04-28 13:35:38,306 INFO [train.py:904] (7/8) Epoch 6, batch 8150, loss[loss=0.1993, simple_loss=0.2811, pruned_loss=0.05873, over 17041.00 frames. ], tot_loss[loss=0.2436, simple_loss=0.32, pruned_loss=0.08363, over 3100425.38 frames. ], batch size: 50, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:35:44,402 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-28 13:35:46,443 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:36:54,145 INFO [train.py:904] (7/8) Epoch 6, batch 8200, loss[loss=0.2246, simple_loss=0.3158, pruned_loss=0.06675, over 16231.00 frames. ], tot_loss[loss=0.2426, simple_loss=0.3182, pruned_loss=0.08346, over 3091874.49 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:37:21,724 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58967.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:37:40,739 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.295e+02 3.597e+02 4.719e+02 5.797e+02 1.131e+03, threshold=9.438e+02, percent-clipped=6.0 2023-04-28 13:38:09,634 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.59 vs. limit=5.0 2023-04-28 13:38:17,119 INFO [train.py:904] (7/8) Epoch 6, batch 8250, loss[loss=0.2199, simple_loss=0.3081, pruned_loss=0.06585, over 16383.00 frames. ], tot_loss[loss=0.2399, simple_loss=0.3173, pruned_loss=0.08122, over 3078728.15 frames. ], batch size: 146, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:39:16,130 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59037.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:39:23,480 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6609, 1.9486, 1.5685, 1.7076, 2.3106, 2.0891, 2.3730, 2.5202], device='cuda:7'), covar=tensor([0.0039, 0.0190, 0.0253, 0.0252, 0.0124, 0.0185, 0.0093, 0.0116], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0161, 0.0164, 0.0163, 0.0159, 0.0164, 0.0147, 0.0144], device='cuda:7'), out_proj_covar=tensor([9.9366e-05, 1.8682e-04, 1.8552e-04, 1.8549e-04, 1.8612e-04, 1.9069e-04, 1.6607e-04, 1.6668e-04], device='cuda:7') 2023-04-28 13:39:37,610 INFO [train.py:904] (7/8) Epoch 6, batch 8300, loss[loss=0.2237, simple_loss=0.3152, pruned_loss=0.06609, over 16429.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.3134, pruned_loss=0.07768, over 3062424.29 frames. ], batch size: 146, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:40:14,767 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0834, 1.7262, 1.5338, 1.4301, 1.8280, 1.6259, 1.7486, 1.9347], device='cuda:7'), covar=tensor([0.0046, 0.0147, 0.0210, 0.0202, 0.0116, 0.0154, 0.0094, 0.0108], device='cuda:7'), in_proj_covar=tensor([0.0085, 0.0163, 0.0165, 0.0164, 0.0160, 0.0165, 0.0147, 0.0146], device='cuda:7'), out_proj_covar=tensor([9.9773e-05, 1.8869e-04, 1.8659e-04, 1.8646e-04, 1.8765e-04, 1.9169e-04, 1.6629e-04, 1.6836e-04], device='cuda:7') 2023-04-28 13:40:22,348 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.786e+02 3.488e+02 4.487e+02 8.597e+02, threshold=6.977e+02, percent-clipped=0.0 2023-04-28 13:40:38,433 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7040, 1.9803, 2.1332, 4.2308, 1.8485, 2.8498, 2.1457, 2.1062], device='cuda:7'), covar=tensor([0.0467, 0.2616, 0.1388, 0.0225, 0.3311, 0.1262, 0.2313, 0.2603], device='cuda:7'), in_proj_covar=tensor([0.0315, 0.0329, 0.0271, 0.0303, 0.0376, 0.0343, 0.0298, 0.0387], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:40:59,239 INFO [train.py:904] (7/8) Epoch 6, batch 8350, loss[loss=0.2143, simple_loss=0.3109, pruned_loss=0.05886, over 16918.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.3113, pruned_loss=0.07466, over 3064708.56 frames. ], batch size: 96, lr: 1.10e-02, grad_scale: 2.0 2023-04-28 13:41:49,321 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59131.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:41:51,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6112, 4.8755, 4.9750, 4.8706, 4.8932, 5.4071, 4.9321, 4.6012], device='cuda:7'), covar=tensor([0.0783, 0.1314, 0.1135, 0.1384, 0.1914, 0.0752, 0.1027, 0.2043], device='cuda:7'), in_proj_covar=tensor([0.0277, 0.0381, 0.0394, 0.0330, 0.0437, 0.0418, 0.0315, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:42:21,184 INFO [train.py:904] (7/8) Epoch 6, batch 8400, loss[loss=0.2101, simple_loss=0.2867, pruned_loss=0.06671, over 12240.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3077, pruned_loss=0.07177, over 3071134.77 frames. ], batch size: 247, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:42:42,352 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59164.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:43:05,302 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.850e+02 3.501e+02 4.655e+02 7.905e+02, threshold=7.001e+02, percent-clipped=2.0 2023-04-28 13:43:09,210 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5308, 3.5925, 3.3742, 3.1716, 3.1681, 3.4885, 3.2274, 3.2875], device='cuda:7'), covar=tensor([0.0481, 0.0364, 0.0209, 0.0168, 0.0549, 0.0328, 0.1016, 0.0385], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0208, 0.0220, 0.0188, 0.0244, 0.0223, 0.0157, 0.0249], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:43:40,731 INFO [train.py:904] (7/8) Epoch 6, batch 8450, loss[loss=0.2003, simple_loss=0.2922, pruned_loss=0.05421, over 16624.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3058, pruned_loss=0.06943, over 3080782.83 frames. ], batch size: 134, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:43:47,086 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 13:44:19,994 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59225.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:45:00,578 INFO [train.py:904] (7/8) Epoch 6, batch 8500, loss[loss=0.2048, simple_loss=0.2861, pruned_loss=0.06171, over 15327.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.3009, pruned_loss=0.06608, over 3072635.45 frames. ], batch size: 191, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:45:09,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0593, 5.3432, 5.0733, 5.1439, 4.7522, 4.7291, 4.8747, 5.3810], device='cuda:7'), covar=tensor([0.0750, 0.0767, 0.0978, 0.0533, 0.0670, 0.0700, 0.0743, 0.0736], device='cuda:7'), in_proj_covar=tensor([0.0389, 0.0510, 0.0433, 0.0331, 0.0316, 0.0338, 0.0417, 0.0372], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:45:19,664 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59262.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:45:46,052 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.033e+02 2.693e+02 3.373e+02 4.229e+02 1.183e+03, threshold=6.745e+02, percent-clipped=1.0 2023-04-28 13:46:25,538 INFO [train.py:904] (7/8) Epoch 6, batch 8550, loss[loss=0.2248, simple_loss=0.3117, pruned_loss=0.06889, over 16726.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2982, pruned_loss=0.06449, over 3051434.26 frames. ], batch size: 124, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:47:38,715 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59337.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:48:07,964 INFO [train.py:904] (7/8) Epoch 6, batch 8600, loss[loss=0.2252, simple_loss=0.297, pruned_loss=0.07668, over 12155.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2977, pruned_loss=0.06357, over 3030708.34 frames. ], batch size: 248, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:49:02,806 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5489, 2.7768, 2.5713, 4.1180, 3.1824, 4.0480, 1.3789, 2.8879], device='cuda:7'), covar=tensor([0.1467, 0.0600, 0.1019, 0.0082, 0.0164, 0.0335, 0.1572, 0.0729], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0143, 0.0167, 0.0093, 0.0181, 0.0187, 0.0165, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 13:49:03,382 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.797e+02 2.892e+02 3.852e+02 5.028e+02 1.593e+03, threshold=7.704e+02, percent-clipped=11.0 2023-04-28 13:49:16,117 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59385.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:49:46,465 INFO [train.py:904] (7/8) Epoch 6, batch 8650, loss[loss=0.1904, simple_loss=0.2795, pruned_loss=0.05061, over 16052.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2955, pruned_loss=0.06195, over 3021866.77 frames. ], batch size: 165, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:50:09,841 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.97 vs. limit=5.0 2023-04-28 13:50:55,915 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59431.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:51:04,192 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-28 13:51:31,181 INFO [train.py:904] (7/8) Epoch 6, batch 8700, loss[loss=0.2161, simple_loss=0.3042, pruned_loss=0.06401, over 15403.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2919, pruned_loss=0.06012, over 3027009.56 frames. ], batch size: 191, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:52:21,224 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.911e+02 3.776e+02 4.464e+02 8.360e+02, threshold=7.553e+02, percent-clipped=2.0 2023-04-28 13:52:22,278 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59479.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:52:57,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6444, 2.6939, 2.3623, 3.6166, 2.6716, 3.7050, 1.3059, 2.8225], device='cuda:7'), covar=tensor([0.1422, 0.0542, 0.1034, 0.0095, 0.0144, 0.0390, 0.1564, 0.0717], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0139, 0.0164, 0.0092, 0.0174, 0.0183, 0.0162, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 13:53:05,581 INFO [train.py:904] (7/8) Epoch 6, batch 8750, loss[loss=0.184, simple_loss=0.2678, pruned_loss=0.05016, over 12340.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2911, pruned_loss=0.05953, over 3010867.94 frames. ], batch size: 250, lr: 1.10e-02, grad_scale: 4.0 2023-04-28 13:53:08,835 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59502.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:53:54,480 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59520.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:54:49,292 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1326, 4.2639, 4.3187, 4.3166, 4.2733, 4.7657, 4.4448, 4.1582], device='cuda:7'), covar=tensor([0.1418, 0.1478, 0.1589, 0.1533, 0.2413, 0.0997, 0.1111, 0.2059], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0380, 0.0390, 0.0328, 0.0436, 0.0420, 0.0316, 0.0447], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:54:55,926 INFO [train.py:904] (7/8) Epoch 6, batch 8800, loss[loss=0.2093, simple_loss=0.2954, pruned_loss=0.06158, over 15252.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2896, pruned_loss=0.05781, over 3046467.55 frames. ], batch size: 191, lr: 1.10e-02, grad_scale: 8.0 2023-04-28 13:55:17,688 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59562.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:55:20,018 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59563.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:55:29,041 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0062, 1.8354, 2.2497, 3.3132, 1.9647, 2.3037, 2.1860, 1.8623], device='cuda:7'), covar=tensor([0.0715, 0.2766, 0.1257, 0.0419, 0.3560, 0.1672, 0.2263, 0.3125], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0324, 0.0272, 0.0297, 0.0376, 0.0338, 0.0298, 0.0384], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:55:52,199 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.649e+02 3.204e+02 3.980e+02 7.645e+02, threshold=6.408e+02, percent-clipped=1.0 2023-04-28 13:56:15,158 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 13:56:37,985 INFO [train.py:904] (7/8) Epoch 6, batch 8850, loss[loss=0.2014, simple_loss=0.3008, pruned_loss=0.051, over 16214.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2926, pruned_loss=0.05728, over 3063261.84 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 13:56:56,205 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59610.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 13:57:19,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5003, 3.4089, 2.7427, 2.1983, 2.3771, 2.1571, 3.5730, 3.3711], device='cuda:7'), covar=tensor([0.2333, 0.0743, 0.1344, 0.1825, 0.1826, 0.1622, 0.0433, 0.0748], device='cuda:7'), in_proj_covar=tensor([0.0276, 0.0238, 0.0261, 0.0245, 0.0250, 0.0199, 0.0238, 0.0245], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:57:42,776 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6419, 4.9133, 5.0263, 4.9278, 4.8831, 5.4474, 5.0729, 4.8381], device='cuda:7'), covar=tensor([0.0774, 0.1299, 0.1283, 0.1587, 0.2191, 0.0897, 0.1060, 0.2026], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0377, 0.0387, 0.0326, 0.0434, 0.0416, 0.0313, 0.0443], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 13:57:56,495 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 13:58:21,084 INFO [train.py:904] (7/8) Epoch 6, batch 8900, loss[loss=0.2209, simple_loss=0.3101, pruned_loss=0.06588, over 16193.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2925, pruned_loss=0.05625, over 3076503.79 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 13:59:22,392 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.981e+02 2.897e+02 3.325e+02 4.292e+02 6.901e+02, threshold=6.651e+02, percent-clipped=1.0 2023-04-28 14:00:23,144 INFO [train.py:904] (7/8) Epoch 6, batch 8950, loss[loss=0.1772, simple_loss=0.2722, pruned_loss=0.04105, over 16779.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2919, pruned_loss=0.05663, over 3089276.76 frames. ], batch size: 83, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:00:53,147 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-28 14:02:12,307 INFO [train.py:904] (7/8) Epoch 6, batch 9000, loss[loss=0.1892, simple_loss=0.2779, pruned_loss=0.05028, over 15239.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2885, pruned_loss=0.05483, over 3101276.27 frames. ], batch size: 191, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:02:12,308 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 14:02:22,190 INFO [train.py:938] (7/8) Epoch 6, validation: loss=0.1682, simple_loss=0.2716, pruned_loss=0.03235, over 944034.00 frames. 2023-04-28 14:02:22,191 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 14:02:51,130 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7299, 2.7298, 1.6705, 2.7970, 2.0981, 2.8043, 1.9660, 2.3873], device='cuda:7'), covar=tensor([0.0212, 0.0362, 0.1314, 0.0102, 0.0688, 0.0526, 0.1213, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0146, 0.0176, 0.0081, 0.0155, 0.0179, 0.0185, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-28 14:03:21,584 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.787e+02 3.475e+02 4.351e+02 1.064e+03, threshold=6.950e+02, percent-clipped=4.0 2023-04-28 14:04:06,173 INFO [train.py:904] (7/8) Epoch 6, batch 9050, loss[loss=0.2056, simple_loss=0.2858, pruned_loss=0.06271, over 12759.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2907, pruned_loss=0.05634, over 3090278.53 frames. ], batch size: 246, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:04:39,967 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8822, 1.2787, 1.6137, 1.7098, 1.7949, 1.8385, 1.4468, 1.7934], device='cuda:7'), covar=tensor([0.0100, 0.0252, 0.0114, 0.0157, 0.0158, 0.0093, 0.0239, 0.0061], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0150, 0.0131, 0.0129, 0.0139, 0.0096, 0.0147, 0.0083], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 14:04:44,616 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59820.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:05:08,085 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7568, 4.7729, 4.5934, 4.3258, 4.1959, 4.6273, 4.5801, 4.3070], device='cuda:7'), covar=tensor([0.0414, 0.0282, 0.0207, 0.0197, 0.0790, 0.0312, 0.0243, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0203, 0.0218, 0.0187, 0.0237, 0.0219, 0.0154, 0.0245], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:05:45,933 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:05:50,598 INFO [train.py:904] (7/8) Epoch 6, batch 9100, loss[loss=0.1873, simple_loss=0.2836, pruned_loss=0.0455, over 16702.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2902, pruned_loss=0.05657, over 3100435.75 frames. ], batch size: 83, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:05:58,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4701, 4.2395, 4.3656, 4.6433, 4.7632, 4.3247, 4.8014, 4.7379], device='cuda:7'), covar=tensor([0.0879, 0.0826, 0.1483, 0.0598, 0.0532, 0.0803, 0.0477, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0382, 0.0470, 0.0594, 0.0479, 0.0366, 0.0360, 0.0376, 0.0408], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:06:04,389 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59858.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:06:24,656 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=59868.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:06:55,735 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.998e+02 3.675e+02 4.654e+02 1.190e+03, threshold=7.350e+02, percent-clipped=4.0 2023-04-28 14:07:19,913 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59889.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:07:47,004 INFO [train.py:904] (7/8) Epoch 6, batch 9150, loss[loss=0.2092, simple_loss=0.2916, pruned_loss=0.06337, over 11548.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2903, pruned_loss=0.0563, over 3081100.77 frames. ], batch size: 249, lr: 1.09e-02, grad_scale: 4.0 2023-04-28 14:08:06,851 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:08:56,085 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-28 14:09:30,374 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59950.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:09:31,018 INFO [train.py:904] (7/8) Epoch 6, batch 9200, loss[loss=0.1925, simple_loss=0.2828, pruned_loss=0.05104, over 16271.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2851, pruned_loss=0.05464, over 3091811.77 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:09:51,872 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7797, 3.6767, 3.7919, 3.6758, 3.8774, 4.2120, 3.9256, 3.5621], device='cuda:7'), covar=tensor([0.1830, 0.2007, 0.1485, 0.2322, 0.2464, 0.1618, 0.1203, 0.2453], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0380, 0.0386, 0.0331, 0.0437, 0.0420, 0.0316, 0.0445], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:10:04,045 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-04-28 14:10:22,279 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 2.993e+02 3.687e+02 4.643e+02 6.569e+02, threshold=7.374e+02, percent-clipped=0.0 2023-04-28 14:11:09,075 INFO [train.py:904] (7/8) Epoch 6, batch 9250, loss[loss=0.1757, simple_loss=0.2694, pruned_loss=0.04104, over 16223.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2851, pruned_loss=0.05517, over 3084783.53 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:12:34,031 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60040.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:12:58,480 INFO [train.py:904] (7/8) Epoch 6, batch 9300, loss[loss=0.1692, simple_loss=0.2582, pruned_loss=0.04009, over 16890.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.283, pruned_loss=0.0543, over 3078668.97 frames. ], batch size: 96, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:13:01,595 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60052.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:13:13,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1864, 3.0114, 2.7850, 2.0015, 2.6108, 2.1038, 2.7575, 2.9153], device='cuda:7'), covar=tensor([0.0343, 0.0518, 0.0464, 0.1411, 0.0676, 0.0936, 0.0636, 0.0630], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0120, 0.0149, 0.0139, 0.0131, 0.0124, 0.0135, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 14:13:33,185 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60065.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:14:01,644 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:14:05,092 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.894e+02 3.605e+02 4.509e+02 1.051e+03, threshold=7.210e+02, percent-clipped=5.0 2023-04-28 14:14:44,861 INFO [train.py:904] (7/8) Epoch 6, batch 9350, loss[loss=0.2065, simple_loss=0.2812, pruned_loss=0.0659, over 12109.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2828, pruned_loss=0.05401, over 3090915.44 frames. ], batch size: 248, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:14:46,370 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60101.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 14:15:13,211 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60113.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 14:15:37,197 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60126.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:15:49,933 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2530, 3.6240, 3.6931, 1.6027, 3.9638, 3.9132, 2.9246, 2.9561], device='cuda:7'), covar=tensor([0.0782, 0.0141, 0.0166, 0.1329, 0.0042, 0.0068, 0.0357, 0.0371], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0089, 0.0079, 0.0140, 0.0067, 0.0080, 0.0113, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 14:16:03,954 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60139.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:16:25,962 INFO [train.py:904] (7/8) Epoch 6, batch 9400, loss[loss=0.1799, simple_loss=0.2603, pruned_loss=0.04975, over 12391.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2828, pruned_loss=0.05418, over 3056365.85 frames. ], batch size: 247, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:16:41,493 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60158.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:17:25,098 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 3.031e+02 3.474e+02 4.387e+02 1.018e+03, threshold=6.948e+02, percent-clipped=2.0 2023-04-28 14:18:05,601 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60198.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:18:09,621 INFO [train.py:904] (7/8) Epoch 6, batch 9450, loss[loss=0.2026, simple_loss=0.2829, pruned_loss=0.0612, over 12690.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2846, pruned_loss=0.05465, over 3035882.20 frames. ], batch size: 250, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:18:15,338 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60204.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:18:19,990 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60206.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:19:39,945 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60245.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:19:50,594 INFO [train.py:904] (7/8) Epoch 6, batch 9500, loss[loss=0.1995, simple_loss=0.2952, pruned_loss=0.05188, over 16237.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2837, pruned_loss=0.05396, over 3044489.16 frames. ], batch size: 166, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:20:11,551 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60259.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:20:32,027 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5927, 4.0483, 4.0902, 1.9747, 4.3137, 4.2619, 3.3116, 3.2811], device='cuda:7'), covar=tensor([0.0663, 0.0099, 0.0116, 0.1096, 0.0035, 0.0049, 0.0246, 0.0310], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0090, 0.0078, 0.0140, 0.0068, 0.0080, 0.0113, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 14:20:51,250 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.861e+02 2.773e+02 3.351e+02 4.137e+02 1.111e+03, threshold=6.703e+02, percent-clipped=2.0 2023-04-28 14:21:39,650 INFO [train.py:904] (7/8) Epoch 6, batch 9550, loss[loss=0.218, simple_loss=0.3138, pruned_loss=0.06112, over 15435.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2832, pruned_loss=0.05386, over 3049608.43 frames. ], batch size: 191, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:23:21,505 INFO [train.py:904] (7/8) Epoch 6, batch 9600, loss[loss=0.204, simple_loss=0.2787, pruned_loss=0.06465, over 12466.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2858, pruned_loss=0.05527, over 3038700.69 frames. ], batch size: 248, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:24:13,886 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1841, 5.2263, 5.0653, 4.7615, 4.6305, 5.0513, 4.9834, 4.6904], device='cuda:7'), covar=tensor([0.0443, 0.0308, 0.0197, 0.0158, 0.0687, 0.0319, 0.0182, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0197, 0.0212, 0.0184, 0.0234, 0.0215, 0.0149, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:24:17,217 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.768e+02 3.406e+02 4.197e+02 1.161e+03, threshold=6.812e+02, percent-clipped=3.0 2023-04-28 14:24:57,308 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60396.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 14:25:09,710 INFO [train.py:904] (7/8) Epoch 6, batch 9650, loss[loss=0.2032, simple_loss=0.2917, pruned_loss=0.05734, over 15462.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2876, pruned_loss=0.05607, over 3026131.08 frames. ], batch size: 191, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:25:28,082 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60408.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 14:25:57,596 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60421.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:26:22,171 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60434.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:26:58,015 INFO [train.py:904] (7/8) Epoch 6, batch 9700, loss[loss=0.1659, simple_loss=0.2595, pruned_loss=0.03611, over 16333.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2872, pruned_loss=0.05592, over 3044769.08 frames. ], batch size: 35, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:27:12,977 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5038, 3.5790, 3.3386, 3.2226, 3.2261, 3.4662, 3.2556, 3.2475], device='cuda:7'), covar=tensor([0.0475, 0.0332, 0.0204, 0.0172, 0.0550, 0.0296, 0.0922, 0.0412], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0198, 0.0214, 0.0186, 0.0236, 0.0217, 0.0151, 0.0243], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:27:35,119 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3519, 4.4183, 4.4834, 4.4921, 4.4465, 4.9439, 4.5402, 4.2942], device='cuda:7'), covar=tensor([0.0981, 0.1581, 0.1468, 0.1625, 0.2318, 0.0977, 0.1249, 0.2166], device='cuda:7'), in_proj_covar=tensor([0.0266, 0.0379, 0.0384, 0.0325, 0.0432, 0.0414, 0.0314, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:27:38,835 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5473, 1.9333, 1.5872, 1.7108, 2.3282, 2.0516, 2.3947, 2.4291], device='cuda:7'), covar=tensor([0.0038, 0.0245, 0.0288, 0.0292, 0.0143, 0.0212, 0.0112, 0.0140], device='cuda:7'), in_proj_covar=tensor([0.0081, 0.0161, 0.0161, 0.0161, 0.0158, 0.0161, 0.0139, 0.0142], device='cuda:7'), out_proj_covar=tensor([9.2710e-05, 1.8679e-04, 1.8177e-04, 1.8268e-04, 1.8310e-04, 1.8556e-04, 1.5381e-04, 1.6360e-04], device='cuda:7') 2023-04-28 14:27:44,417 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2023-04-28 14:27:59,431 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 2.947e+02 3.794e+02 4.727e+02 9.838e+02, threshold=7.588e+02, percent-clipped=5.0 2023-04-28 14:28:03,271 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7453, 2.7081, 1.7487, 2.7890, 2.1543, 2.8055, 1.9983, 2.4102], device='cuda:7'), covar=tensor([0.0179, 0.0340, 0.1208, 0.0098, 0.0646, 0.0430, 0.1104, 0.0580], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0148, 0.0177, 0.0083, 0.0157, 0.0178, 0.0186, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 14:28:41,233 INFO [train.py:904] (7/8) Epoch 6, batch 9750, loss[loss=0.2148, simple_loss=0.3045, pruned_loss=0.06251, over 16261.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2862, pruned_loss=0.05606, over 3023389.02 frames. ], batch size: 165, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:28:47,869 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60504.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:30:11,729 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60545.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:30:19,581 INFO [train.py:904] (7/8) Epoch 6, batch 9800, loss[loss=0.1814, simple_loss=0.284, pruned_loss=0.03944, over 16882.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2854, pruned_loss=0.05459, over 3013056.36 frames. ], batch size: 96, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:30:22,404 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60552.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:30:27,287 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60554.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:31:14,370 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 2.802e+02 3.388e+02 4.163e+02 9.547e+02, threshold=6.775e+02, percent-clipped=1.0 2023-04-28 14:31:48,143 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60593.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:32:01,288 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1951, 4.1678, 4.7219, 4.6255, 4.6620, 4.3341, 4.3113, 4.0985], device='cuda:7'), covar=tensor([0.0264, 0.0549, 0.0303, 0.0434, 0.0382, 0.0310, 0.0758, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0245, 0.0239, 0.0245, 0.0237, 0.0282, 0.0259, 0.0349, 0.0210], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-28 14:32:03,908 INFO [train.py:904] (7/8) Epoch 6, batch 9850, loss[loss=0.212, simple_loss=0.2854, pruned_loss=0.06927, over 12625.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2866, pruned_loss=0.05401, over 3035083.82 frames. ], batch size: 248, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:32:09,010 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 14:33:23,374 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1257, 3.2695, 3.6037, 3.5339, 3.5673, 3.3033, 3.3503, 3.3787], device='cuda:7'), covar=tensor([0.0346, 0.0693, 0.0436, 0.0567, 0.0369, 0.0479, 0.0706, 0.0366], device='cuda:7'), in_proj_covar=tensor([0.0244, 0.0237, 0.0245, 0.0235, 0.0280, 0.0259, 0.0347, 0.0210], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-28 14:33:56,860 INFO [train.py:904] (7/8) Epoch 6, batch 9900, loss[loss=0.1921, simple_loss=0.2751, pruned_loss=0.05458, over 12515.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2873, pruned_loss=0.05398, over 3034054.05 frames. ], batch size: 250, lr: 1.09e-02, grad_scale: 8.0 2023-04-28 14:35:04,554 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.757e+02 3.371e+02 4.013e+02 6.843e+02, threshold=6.741e+02, percent-clipped=1.0 2023-04-28 14:35:42,464 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60696.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:35:52,856 INFO [train.py:904] (7/8) Epoch 6, batch 9950, loss[loss=0.219, simple_loss=0.3087, pruned_loss=0.06461, over 15479.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2894, pruned_loss=0.05411, over 3054796.59 frames. ], batch size: 191, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:36:10,132 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60708.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:36:34,937 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60718.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:36:43,582 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60721.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:37:13,875 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60734.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:37:20,947 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6583, 4.1469, 4.1492, 2.0445, 3.5651, 2.7122, 4.0697, 3.9749], device='cuda:7'), covar=tensor([0.0150, 0.0443, 0.0441, 0.1470, 0.0488, 0.0780, 0.0511, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0119, 0.0152, 0.0139, 0.0131, 0.0123, 0.0134, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 14:37:38,498 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60744.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:37:54,651 INFO [train.py:904] (7/8) Epoch 6, batch 10000, loss[loss=0.2113, simple_loss=0.3057, pruned_loss=0.0585, over 15328.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2876, pruned_loss=0.0535, over 3079265.62 frames. ], batch size: 191, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:38:06,642 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60756.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:23,905 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-28 14:38:29,915 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60769.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:50,650 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60779.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:51,225 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.599e+02 3.122e+02 4.143e+02 8.746e+02, threshold=6.243e+02, percent-clipped=4.0 2023-04-28 14:38:57,419 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60782.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:38:57,966 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 14:39:16,250 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8302, 2.6901, 2.7181, 1.9382, 2.4839, 2.6044, 2.6328, 1.8189], device='cuda:7'), covar=tensor([0.0285, 0.0026, 0.0034, 0.0211, 0.0068, 0.0047, 0.0046, 0.0292], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0053, 0.0057, 0.0113, 0.0061, 0.0070, 0.0063, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 14:39:33,427 INFO [train.py:904] (7/8) Epoch 6, batch 10050, loss[loss=0.234, simple_loss=0.3151, pruned_loss=0.07644, over 12499.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.288, pruned_loss=0.05373, over 3071977.43 frames. ], batch size: 248, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:41:00,391 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-28 14:41:05,285 INFO [train.py:904] (7/8) Epoch 6, batch 10100, loss[loss=0.2002, simple_loss=0.285, pruned_loss=0.05767, over 16885.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2885, pruned_loss=0.05418, over 3078772.56 frames. ], batch size: 96, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:41:11,730 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60854.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:42:01,264 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.889e+02 3.366e+02 3.931e+02 7.986e+02, threshold=6.733e+02, percent-clipped=7.0 2023-04-28 14:42:22,732 INFO [train.py:904] (7/8) Epoch 6, batch 10150, loss[loss=0.214, simple_loss=0.2887, pruned_loss=0.06969, over 12396.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2873, pruned_loss=0.05455, over 3060336.19 frames. ], batch size: 250, lr: 1.08e-02, grad_scale: 8.0 2023-04-28 14:42:48,317 INFO [train.py:904] (7/8) Epoch 7, batch 0, loss[loss=0.3225, simple_loss=0.373, pruned_loss=0.136, over 15495.00 frames. ], tot_loss[loss=0.3225, simple_loss=0.373, pruned_loss=0.136, over 15495.00 frames. ], batch size: 191, lr: 1.02e-02, grad_scale: 8.0 2023-04-28 14:42:48,318 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 14:42:55,784 INFO [train.py:938] (7/8) Epoch 7, validation: loss=0.1665, simple_loss=0.2702, pruned_loss=0.03141, over 944034.00 frames. 2023-04-28 14:42:55,785 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 14:42:56,042 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=60902.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:43:13,725 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0804, 2.2518, 2.3952, 4.5728, 1.9604, 2.9411, 2.4272, 2.3990], device='cuda:7'), covar=tensor([0.0500, 0.2615, 0.1356, 0.0258, 0.3319, 0.1415, 0.2078, 0.2804], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0324, 0.0276, 0.0301, 0.0371, 0.0339, 0.0297, 0.0377], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:44:05,507 INFO [train.py:904] (7/8) Epoch 7, batch 50, loss[loss=0.2045, simple_loss=0.2965, pruned_loss=0.05626, over 17146.00 frames. ], tot_loss[loss=0.229, simple_loss=0.3029, pruned_loss=0.07753, over 760687.51 frames. ], batch size: 49, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:44:49,445 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 3.058e+02 3.620e+02 4.703e+02 1.122e+03, threshold=7.241e+02, percent-clipped=6.0 2023-04-28 14:45:15,350 INFO [train.py:904] (7/8) Epoch 7, batch 100, loss[loss=0.1834, simple_loss=0.2723, pruned_loss=0.04724, over 17236.00 frames. ], tot_loss[loss=0.2212, simple_loss=0.2971, pruned_loss=0.07264, over 1335415.30 frames. ], batch size: 45, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:46:24,665 INFO [train.py:904] (7/8) Epoch 7, batch 150, loss[loss=0.2516, simple_loss=0.3196, pruned_loss=0.0918, over 15644.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.2955, pruned_loss=0.07135, over 1771171.50 frames. ], batch size: 191, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:46:37,941 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61061.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:46:56,502 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61074.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:46:58,818 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9037, 5.2186, 4.9366, 4.9424, 4.6339, 4.5117, 4.7125, 5.2778], device='cuda:7'), covar=tensor([0.0805, 0.0760, 0.1009, 0.0563, 0.0716, 0.0814, 0.0818, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0544, 0.0453, 0.0357, 0.0337, 0.0355, 0.0448, 0.0396], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:47:07,165 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.994e+02 3.585e+02 4.192e+02 7.211e+02, threshold=7.170e+02, percent-clipped=0.0 2023-04-28 14:47:34,196 INFO [train.py:904] (7/8) Epoch 7, batch 200, loss[loss=0.2344, simple_loss=0.3037, pruned_loss=0.0826, over 16600.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.2963, pruned_loss=0.07216, over 2108944.00 frames. ], batch size: 75, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:47:44,043 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-28 14:48:01,688 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61122.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:48:44,072 INFO [train.py:904] (7/8) Epoch 7, batch 250, loss[loss=0.2199, simple_loss=0.3096, pruned_loss=0.06514, over 17063.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.2949, pruned_loss=0.07209, over 2381756.85 frames. ], batch size: 53, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:48:53,911 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6723, 4.1727, 4.3398, 1.9376, 4.5928, 4.5418, 3.2231, 3.3589], device='cuda:7'), covar=tensor([0.0747, 0.0110, 0.0135, 0.1137, 0.0046, 0.0062, 0.0361, 0.0344], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0091, 0.0080, 0.0141, 0.0069, 0.0085, 0.0116, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 14:49:24,843 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 3.071e+02 3.589e+02 4.451e+02 1.172e+03, threshold=7.179e+02, percent-clipped=5.0 2023-04-28 14:49:41,507 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-28 14:49:52,030 INFO [train.py:904] (7/8) Epoch 7, batch 300, loss[loss=0.1815, simple_loss=0.2731, pruned_loss=0.04493, over 17127.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2903, pruned_loss=0.06975, over 2581076.12 frames. ], batch size: 48, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:50:36,833 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5972, 3.9681, 4.2571, 3.2123, 3.9359, 3.9845, 4.0072, 2.3441], device='cuda:7'), covar=tensor([0.0313, 0.0032, 0.0032, 0.0196, 0.0040, 0.0073, 0.0039, 0.0315], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0060, 0.0060, 0.0115, 0.0062, 0.0075, 0.0065, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 14:50:59,331 INFO [train.py:904] (7/8) Epoch 7, batch 350, loss[loss=0.1969, simple_loss=0.2705, pruned_loss=0.06161, over 16500.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2871, pruned_loss=0.06747, over 2751801.50 frames. ], batch size: 75, lr: 1.01e-02, grad_scale: 1.0 2023-04-28 14:51:34,533 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 14:51:35,916 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8878, 1.6421, 2.1907, 2.7523, 2.5443, 3.3001, 2.0131, 3.0301], device='cuda:7'), covar=tensor([0.0126, 0.0288, 0.0200, 0.0171, 0.0162, 0.0104, 0.0260, 0.0101], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0150, 0.0134, 0.0133, 0.0137, 0.0100, 0.0146, 0.0084], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 14:51:41,730 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.825e+02 3.324e+02 4.016e+02 8.512e+02, threshold=6.649e+02, percent-clipped=2.0 2023-04-28 14:52:07,886 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9143, 4.3250, 2.2498, 4.7039, 3.0994, 4.7461, 2.3298, 3.1908], device='cuda:7'), covar=tensor([0.0161, 0.0229, 0.1432, 0.0077, 0.0658, 0.0299, 0.1381, 0.0585], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0156, 0.0179, 0.0090, 0.0158, 0.0192, 0.0188, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 14:52:08,583 INFO [train.py:904] (7/8) Epoch 7, batch 400, loss[loss=0.1735, simple_loss=0.2611, pruned_loss=0.04291, over 16851.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2852, pruned_loss=0.06674, over 2879102.10 frames. ], batch size: 42, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:53:17,212 INFO [train.py:904] (7/8) Epoch 7, batch 450, loss[loss=0.2178, simple_loss=0.2796, pruned_loss=0.07801, over 16897.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2831, pruned_loss=0.06505, over 2978718.40 frames. ], batch size: 109, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:53:48,661 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61374.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:54:00,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.768e+02 3.236e+02 4.177e+02 7.606e+02, threshold=6.472e+02, percent-clipped=3.0 2023-04-28 14:54:27,760 INFO [train.py:904] (7/8) Epoch 7, batch 500, loss[loss=0.1996, simple_loss=0.2735, pruned_loss=0.06286, over 16840.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.282, pruned_loss=0.0639, over 3062153.52 frames. ], batch size: 90, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:54:47,755 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61417.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:54:54,991 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=61422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 14:55:35,831 INFO [train.py:904] (7/8) Epoch 7, batch 550, loss[loss=0.2094, simple_loss=0.2781, pruned_loss=0.0704, over 16775.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2796, pruned_loss=0.06248, over 3120952.05 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:56:09,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4066, 4.3288, 4.3100, 4.4607, 4.3441, 4.9400, 4.5741, 4.3107], device='cuda:7'), covar=tensor([0.1221, 0.1788, 0.1866, 0.1815, 0.2697, 0.1041, 0.1319, 0.2468], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0445, 0.0442, 0.0378, 0.0507, 0.0474, 0.0360, 0.0507], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 14:56:16,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8732, 4.4037, 3.5569, 2.3492, 3.1642, 2.5665, 4.5510, 4.2335], device='cuda:7'), covar=tensor([0.2209, 0.0556, 0.1127, 0.1890, 0.2123, 0.1500, 0.0309, 0.0690], device='cuda:7'), in_proj_covar=tensor([0.0285, 0.0253, 0.0273, 0.0255, 0.0271, 0.0210, 0.0250, 0.0271], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:56:17,398 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.772e+02 2.888e+02 3.437e+02 4.263e+02 9.933e+02, threshold=6.875e+02, percent-clipped=6.0 2023-04-28 14:56:43,946 INFO [train.py:904] (7/8) Epoch 7, batch 600, loss[loss=0.1962, simple_loss=0.2833, pruned_loss=0.05459, over 17079.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2793, pruned_loss=0.06325, over 3166386.63 frames. ], batch size: 53, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:57:53,646 INFO [train.py:904] (7/8) Epoch 7, batch 650, loss[loss=0.2087, simple_loss=0.2893, pruned_loss=0.06402, over 16791.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2781, pruned_loss=0.06282, over 3201941.78 frames. ], batch size: 62, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:58:27,561 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9720, 5.3623, 5.0340, 5.1585, 4.7143, 4.6466, 4.7938, 5.3841], device='cuda:7'), covar=tensor([0.0802, 0.0751, 0.0990, 0.0476, 0.0709, 0.0788, 0.0808, 0.0756], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0569, 0.0474, 0.0376, 0.0353, 0.0369, 0.0472, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:58:35,764 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.890e+02 3.388e+02 4.314e+02 1.492e+03, threshold=6.776e+02, percent-clipped=7.0 2023-04-28 14:58:39,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7503, 4.7189, 4.5911, 3.6722, 4.6817, 1.8952, 4.3800, 4.4523], device='cuda:7'), covar=tensor([0.0090, 0.0081, 0.0139, 0.0403, 0.0076, 0.2155, 0.0120, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0089, 0.0140, 0.0131, 0.0105, 0.0158, 0.0122, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 14:59:01,951 INFO [train.py:904] (7/8) Epoch 7, batch 700, loss[loss=0.1869, simple_loss=0.2743, pruned_loss=0.04977, over 17157.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2776, pruned_loss=0.06242, over 3230182.33 frames. ], batch size: 46, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 14:59:39,181 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-28 15:00:08,132 INFO [train.py:904] (7/8) Epoch 7, batch 750, loss[loss=0.2296, simple_loss=0.2908, pruned_loss=0.08415, over 16124.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2789, pruned_loss=0.06343, over 3250102.93 frames. ], batch size: 164, lr: 1.01e-02, grad_scale: 2.0 2023-04-28 15:00:51,555 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.695e+02 2.818e+02 3.177e+02 3.782e+02 5.959e+02, threshold=6.355e+02, percent-clipped=0.0 2023-04-28 15:01:16,778 INFO [train.py:904] (7/8) Epoch 7, batch 800, loss[loss=0.2592, simple_loss=0.3166, pruned_loss=0.1009, over 12208.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2781, pruned_loss=0.06281, over 3260430.23 frames. ], batch size: 246, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:01:37,395 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61717.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:01:57,109 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61732.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:02:13,202 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61743.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:02:24,047 INFO [train.py:904] (7/8) Epoch 7, batch 850, loss[loss=0.2238, simple_loss=0.2891, pruned_loss=0.07929, over 12364.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2769, pruned_loss=0.062, over 3275493.89 frames. ], batch size: 247, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:02:42,645 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=61765.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:02:44,102 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7634, 4.3542, 4.2578, 2.4997, 3.5389, 2.8362, 4.1206, 3.9986], device='cuda:7'), covar=tensor([0.0241, 0.0562, 0.0398, 0.1353, 0.0621, 0.0789, 0.0532, 0.1015], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0131, 0.0153, 0.0139, 0.0132, 0.0124, 0.0137, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 15:02:50,063 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8431, 2.1187, 2.2920, 4.5923, 2.0059, 2.8666, 2.2129, 2.3325], device='cuda:7'), covar=tensor([0.0638, 0.2762, 0.1590, 0.0292, 0.3359, 0.1675, 0.2486, 0.2963], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0346, 0.0291, 0.0322, 0.0389, 0.0375, 0.0314, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:03:06,806 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.876e+02 3.503e+02 4.271e+02 9.516e+02, threshold=7.005e+02, percent-clipped=8.0 2023-04-28 15:03:19,338 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61793.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:03:33,712 INFO [train.py:904] (7/8) Epoch 7, batch 900, loss[loss=0.1698, simple_loss=0.2483, pruned_loss=0.04568, over 16781.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2761, pruned_loss=0.06068, over 3289469.27 frames. ], batch size: 102, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:03:36,379 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61804.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:04:40,620 INFO [train.py:904] (7/8) Epoch 7, batch 950, loss[loss=0.1731, simple_loss=0.2588, pruned_loss=0.04375, over 17233.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2763, pruned_loss=0.06068, over 3296019.32 frames. ], batch size: 45, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:05:20,760 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.637e+02 3.238e+02 3.897e+02 7.916e+02, threshold=6.477e+02, percent-clipped=2.0 2023-04-28 15:05:46,849 INFO [train.py:904] (7/8) Epoch 7, batch 1000, loss[loss=0.196, simple_loss=0.2925, pruned_loss=0.04975, over 17280.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2754, pruned_loss=0.0607, over 3298286.29 frames. ], batch size: 52, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:06:05,355 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61914.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:06:56,262 INFO [train.py:904] (7/8) Epoch 7, batch 1050, loss[loss=0.1758, simple_loss=0.2587, pruned_loss=0.04645, over 16983.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2755, pruned_loss=0.06054, over 3308988.17 frames. ], batch size: 41, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:07:14,996 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8295, 3.7032, 3.1257, 5.2396, 4.7040, 4.8663, 1.9174, 3.3930], device='cuda:7'), covar=tensor([0.1219, 0.0439, 0.0897, 0.0083, 0.0178, 0.0245, 0.1212, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0146, 0.0169, 0.0102, 0.0188, 0.0196, 0.0165, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 15:07:28,139 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61975.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:07:37,644 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.768e+02 3.346e+02 4.170e+02 9.065e+02, threshold=6.692e+02, percent-clipped=2.0 2023-04-28 15:08:07,086 INFO [train.py:904] (7/8) Epoch 7, batch 1100, loss[loss=0.2302, simple_loss=0.2864, pruned_loss=0.08698, over 16717.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.275, pruned_loss=0.06083, over 3302974.94 frames. ], batch size: 124, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:08:52,836 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0710, 3.7576, 2.7410, 5.2997, 4.7278, 4.6668, 2.0392, 3.1380], device='cuda:7'), covar=tensor([0.1024, 0.0370, 0.0946, 0.0079, 0.0201, 0.0296, 0.1030, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0145, 0.0169, 0.0102, 0.0188, 0.0196, 0.0164, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 15:09:00,258 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9338, 1.7840, 2.1298, 2.9344, 2.7678, 3.2729, 1.8756, 3.0556], device='cuda:7'), covar=tensor([0.0102, 0.0269, 0.0191, 0.0129, 0.0133, 0.0084, 0.0256, 0.0083], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0151, 0.0136, 0.0137, 0.0140, 0.0099, 0.0146, 0.0088], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 15:09:15,550 INFO [train.py:904] (7/8) Epoch 7, batch 1150, loss[loss=0.2045, simple_loss=0.2933, pruned_loss=0.0579, over 17035.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2743, pruned_loss=0.05953, over 3313547.07 frames. ], batch size: 55, lr: 1.01e-02, grad_scale: 4.0 2023-04-28 15:09:29,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6591, 4.6740, 4.8010, 4.7749, 4.6037, 5.3031, 4.8803, 4.5796], device='cuda:7'), covar=tensor([0.1204, 0.1979, 0.1987, 0.1971, 0.3395, 0.1143, 0.1443, 0.2855], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0449, 0.0450, 0.0378, 0.0514, 0.0479, 0.0357, 0.0510], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 15:09:57,385 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.657e+02 3.127e+02 3.921e+02 1.096e+03, threshold=6.253e+02, percent-clipped=5.0 2023-04-28 15:09:58,561 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4012, 1.9827, 2.1994, 4.0011, 1.9795, 2.6436, 2.0755, 2.1295], device='cuda:7'), covar=tensor([0.0713, 0.2669, 0.1585, 0.0354, 0.3098, 0.1689, 0.2663, 0.2618], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0347, 0.0290, 0.0320, 0.0386, 0.0378, 0.0315, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:10:04,188 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62088.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:10:18,797 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62099.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:10:22,595 INFO [train.py:904] (7/8) Epoch 7, batch 1200, loss[loss=0.1902, simple_loss=0.261, pruned_loss=0.05975, over 12064.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.273, pruned_loss=0.05896, over 3314191.64 frames. ], batch size: 246, lr: 1.01e-02, grad_scale: 8.0 2023-04-28 15:10:51,480 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62122.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:11:30,507 INFO [train.py:904] (7/8) Epoch 7, batch 1250, loss[loss=0.217, simple_loss=0.2964, pruned_loss=0.06883, over 17065.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2736, pruned_loss=0.06001, over 3323975.13 frames. ], batch size: 55, lr: 1.01e-02, grad_scale: 8.0 2023-04-28 15:12:13,503 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.700e+02 3.234e+02 4.045e+02 7.667e+02, threshold=6.469e+02, percent-clipped=6.0 2023-04-28 15:12:13,934 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62183.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:12:39,195 INFO [train.py:904] (7/8) Epoch 7, batch 1300, loss[loss=0.1829, simple_loss=0.2704, pruned_loss=0.04767, over 17145.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2745, pruned_loss=0.06027, over 3325226.26 frames. ], batch size: 49, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:13:49,052 INFO [train.py:904] (7/8) Epoch 7, batch 1350, loss[loss=0.2087, simple_loss=0.2803, pruned_loss=0.06855, over 16736.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2734, pruned_loss=0.05962, over 3328712.63 frames. ], batch size: 124, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:14:15,374 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62270.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:14:19,112 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:14:31,593 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.487e+02 3.145e+02 3.874e+02 9.778e+02, threshold=6.290e+02, percent-clipped=3.0 2023-04-28 15:14:36,052 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-28 15:14:56,763 INFO [train.py:904] (7/8) Epoch 7, batch 1400, loss[loss=0.1867, simple_loss=0.2764, pruned_loss=0.0485, over 17076.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.273, pruned_loss=0.05927, over 3325252.60 frames. ], batch size: 47, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:15:42,108 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62334.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:16:03,422 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-04-28 15:16:06,309 INFO [train.py:904] (7/8) Epoch 7, batch 1450, loss[loss=0.1703, simple_loss=0.2508, pruned_loss=0.04488, over 16264.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2723, pruned_loss=0.05957, over 3326402.83 frames. ], batch size: 36, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:16:47,644 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.994e+02 2.891e+02 3.326e+02 4.338e+02 8.815e+02, threshold=6.653e+02, percent-clipped=8.0 2023-04-28 15:16:53,753 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9243, 4.4815, 4.6357, 3.2760, 4.0164, 4.5682, 4.1475, 3.1060], device='cuda:7'), covar=tensor([0.0267, 0.0021, 0.0018, 0.0196, 0.0039, 0.0036, 0.0034, 0.0232], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0062, 0.0061, 0.0114, 0.0064, 0.0075, 0.0066, 0.0111], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 15:16:56,119 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62388.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:16:59,976 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62391.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:17:10,994 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62399.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:17:14,682 INFO [train.py:904] (7/8) Epoch 7, batch 1500, loss[loss=0.193, simple_loss=0.275, pruned_loss=0.0555, over 16736.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2726, pruned_loss=0.05959, over 3323905.16 frames. ], batch size: 62, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:17:41,081 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 15:18:01,115 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62436.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:18:16,545 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62447.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:18:23,649 INFO [train.py:904] (7/8) Epoch 7, batch 1550, loss[loss=0.2294, simple_loss=0.3088, pruned_loss=0.07502, over 15470.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2745, pruned_loss=0.06127, over 3325228.10 frames. ], batch size: 190, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:18:24,155 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62452.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:18:59,847 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62478.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:19:01,500 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 15:19:05,757 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 15:19:06,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.614e+02 2.773e+02 3.245e+02 3.904e+02 9.637e+02, threshold=6.490e+02, percent-clipped=1.0 2023-04-28 15:19:32,096 INFO [train.py:904] (7/8) Epoch 7, batch 1600, loss[loss=0.1908, simple_loss=0.2857, pruned_loss=0.04792, over 17120.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2763, pruned_loss=0.06164, over 3330058.08 frames. ], batch size: 48, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:19:35,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8374, 4.8667, 5.4644, 5.4135, 5.3926, 4.9928, 4.9871, 4.7595], device='cuda:7'), covar=tensor([0.0245, 0.0404, 0.0259, 0.0295, 0.0305, 0.0262, 0.0717, 0.0341], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0287, 0.0294, 0.0277, 0.0336, 0.0306, 0.0408, 0.0249], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 15:20:01,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8154, 2.1066, 2.2447, 4.5648, 1.9630, 2.8849, 2.2742, 2.4446], device='cuda:7'), covar=tensor([0.0682, 0.2855, 0.1614, 0.0276, 0.3334, 0.1727, 0.2336, 0.3036], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0347, 0.0290, 0.0319, 0.0381, 0.0377, 0.0316, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:20:31,556 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62545.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:20:40,976 INFO [train.py:904] (7/8) Epoch 7, batch 1650, loss[loss=0.218, simple_loss=0.3003, pruned_loss=0.06782, over 16664.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2784, pruned_loss=0.06295, over 3315759.34 frames. ], batch size: 62, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:20:47,942 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 15:21:05,465 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62570.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:21:22,702 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.048e+02 2.793e+02 3.410e+02 4.977e+02 9.439e+02, threshold=6.821e+02, percent-clipped=5.0 2023-04-28 15:21:25,420 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62585.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:21:49,319 INFO [train.py:904] (7/8) Epoch 7, batch 1700, loss[loss=0.1858, simple_loss=0.2672, pruned_loss=0.05218, over 16822.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2797, pruned_loss=0.06281, over 3324153.02 frames. ], batch size: 42, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:21:55,127 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62606.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:12,980 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62618.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:27,923 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62629.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:51,319 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62646.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:22:58,552 INFO [train.py:904] (7/8) Epoch 7, batch 1750, loss[loss=0.2303, simple_loss=0.2872, pruned_loss=0.08667, over 16937.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2806, pruned_loss=0.0626, over 3330117.46 frames. ], batch size: 109, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:23:18,815 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62666.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:23:21,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1281, 4.6993, 4.5620, 5.2909, 5.3614, 4.7490, 5.2408, 5.3905], device='cuda:7'), covar=tensor([0.1136, 0.0895, 0.2352, 0.0669, 0.0776, 0.0744, 0.0854, 0.0596], device='cuda:7'), in_proj_covar=tensor([0.0467, 0.0569, 0.0722, 0.0572, 0.0440, 0.0429, 0.0457, 0.0493], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:23:41,549 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.869e+02 2.853e+02 3.319e+02 4.250e+02 7.879e+02, threshold=6.638e+02, percent-clipped=1.0 2023-04-28 15:24:08,276 INFO [train.py:904] (7/8) Epoch 7, batch 1800, loss[loss=0.1901, simple_loss=0.2724, pruned_loss=0.05388, over 16433.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2815, pruned_loss=0.06208, over 3334318.88 frames. ], batch size: 146, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:24:27,497 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0383, 4.7982, 4.9754, 5.2631, 5.3926, 4.7139, 5.3833, 5.3668], device='cuda:7'), covar=tensor([0.1121, 0.0879, 0.1548, 0.0585, 0.0499, 0.0632, 0.0417, 0.0506], device='cuda:7'), in_proj_covar=tensor([0.0466, 0.0567, 0.0717, 0.0569, 0.0438, 0.0429, 0.0454, 0.0492], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:24:42,415 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62727.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:11,012 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62747.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:17,224 INFO [train.py:904] (7/8) Epoch 7, batch 1850, loss[loss=0.1841, simple_loss=0.2791, pruned_loss=0.04458, over 17218.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2817, pruned_loss=0.06195, over 3329675.03 frames. ], batch size: 44, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:25:18,967 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3036, 2.1557, 1.6303, 1.9126, 2.5652, 2.3808, 2.7027, 2.7282], device='cuda:7'), covar=tensor([0.0102, 0.0228, 0.0291, 0.0274, 0.0113, 0.0187, 0.0144, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0107, 0.0177, 0.0175, 0.0177, 0.0172, 0.0179, 0.0171, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:25:40,223 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6627, 5.0635, 4.8750, 4.7938, 4.4758, 4.4180, 4.5087, 5.1490], device='cuda:7'), covar=tensor([0.0803, 0.0681, 0.0838, 0.0613, 0.0643, 0.0908, 0.0799, 0.0650], device='cuda:7'), in_proj_covar=tensor([0.0451, 0.0579, 0.0490, 0.0389, 0.0363, 0.0375, 0.0477, 0.0420], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:25:45,811 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62773.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:53,059 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62778.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:25:59,966 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.663e+02 2.724e+02 3.242e+02 3.816e+02 7.402e+02, threshold=6.484e+02, percent-clipped=2.0 2023-04-28 15:26:08,281 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4923, 3.4186, 3.4224, 2.8363, 3.4847, 1.9832, 3.1831, 2.8200], device='cuda:7'), covar=tensor([0.0093, 0.0076, 0.0127, 0.0215, 0.0072, 0.1705, 0.0104, 0.0177], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0094, 0.0146, 0.0140, 0.0112, 0.0157, 0.0129, 0.0139], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:26:26,219 INFO [train.py:904] (7/8) Epoch 7, batch 1900, loss[loss=0.1619, simple_loss=0.2507, pruned_loss=0.03649, over 17247.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2817, pruned_loss=0.06179, over 3324873.03 frames. ], batch size: 44, lr: 1.00e-02, grad_scale: 8.0 2023-04-28 15:27:00,818 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62826.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:27:11,564 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62834.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:27:36,786 INFO [train.py:904] (7/8) Epoch 7, batch 1950, loss[loss=0.1682, simple_loss=0.2513, pruned_loss=0.04254, over 17228.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2811, pruned_loss=0.0616, over 3329625.30 frames. ], batch size: 43, lr: 9.99e-03, grad_scale: 8.0 2023-04-28 15:28:19,259 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.715e+02 3.351e+02 4.043e+02 9.306e+02, threshold=6.703e+02, percent-clipped=2.0 2023-04-28 15:28:44,330 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62901.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:28:45,198 INFO [train.py:904] (7/8) Epoch 7, batch 2000, loss[loss=0.2318, simple_loss=0.3067, pruned_loss=0.07843, over 12114.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2814, pruned_loss=0.06145, over 3320357.50 frames. ], batch size: 247, lr: 9.99e-03, grad_scale: 8.0 2023-04-28 15:28:57,703 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7445, 3.0158, 2.6647, 4.2429, 3.7417, 4.1948, 1.4380, 2.8849], device='cuda:7'), covar=tensor([0.1227, 0.0462, 0.0894, 0.0091, 0.0175, 0.0299, 0.1253, 0.0680], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0148, 0.0171, 0.0105, 0.0200, 0.0201, 0.0167, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 15:29:24,505 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62929.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:29:39,611 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62941.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:29:55,692 INFO [train.py:904] (7/8) Epoch 7, batch 2050, loss[loss=0.1985, simple_loss=0.2927, pruned_loss=0.05212, over 17132.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2813, pruned_loss=0.06191, over 3324647.30 frames. ], batch size: 48, lr: 9.99e-03, grad_scale: 16.0 2023-04-28 15:30:31,335 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=62977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:30:39,217 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.104e+02 2.968e+02 3.465e+02 4.226e+02 9.943e+02, threshold=6.931e+02, percent-clipped=3.0 2023-04-28 15:31:05,140 INFO [train.py:904] (7/8) Epoch 7, batch 2100, loss[loss=0.2223, simple_loss=0.2959, pruned_loss=0.0744, over 16672.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2834, pruned_loss=0.06308, over 3320843.02 frames. ], batch size: 68, lr: 9.98e-03, grad_scale: 8.0 2023-04-28 15:31:10,599 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1894, 3.3242, 3.3793, 1.7078, 3.5225, 3.5482, 2.8034, 2.6666], device='cuda:7'), covar=tensor([0.0738, 0.0133, 0.0154, 0.1072, 0.0074, 0.0110, 0.0373, 0.0370], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0083, 0.0140, 0.0071, 0.0090, 0.0120, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 15:31:18,049 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1872, 4.1945, 4.6301, 4.6052, 4.6260, 4.2712, 4.3533, 4.1669], device='cuda:7'), covar=tensor([0.0292, 0.0463, 0.0298, 0.0337, 0.0365, 0.0302, 0.0672, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0287, 0.0289, 0.0274, 0.0331, 0.0307, 0.0408, 0.0245], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 15:31:33,201 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63022.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:32:09,411 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63047.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:32:16,869 INFO [train.py:904] (7/8) Epoch 7, batch 2150, loss[loss=0.2013, simple_loss=0.2929, pruned_loss=0.05487, over 17254.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2846, pruned_loss=0.06332, over 3323523.13 frames. ], batch size: 52, lr: 9.98e-03, grad_scale: 8.0 2023-04-28 15:33:00,563 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.945e+02 2.751e+02 3.323e+02 4.118e+02 9.925e+02, threshold=6.647e+02, percent-clipped=2.0 2023-04-28 15:33:17,124 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63095.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:33:25,757 INFO [train.py:904] (7/8) Epoch 7, batch 2200, loss[loss=0.2467, simple_loss=0.3149, pruned_loss=0.0892, over 15634.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2858, pruned_loss=0.06371, over 3326990.50 frames. ], batch size: 190, lr: 9.98e-03, grad_scale: 8.0 2023-04-28 15:34:03,319 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63129.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:34:33,877 INFO [train.py:904] (7/8) Epoch 7, batch 2250, loss[loss=0.1952, simple_loss=0.2862, pruned_loss=0.05212, over 17221.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2858, pruned_loss=0.06418, over 3324840.98 frames. ], batch size: 45, lr: 9.97e-03, grad_scale: 8.0 2023-04-28 15:35:18,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.937e+02 3.503e+02 4.061e+02 7.762e+02, threshold=7.006e+02, percent-clipped=5.0 2023-04-28 15:35:42,446 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63201.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:35:43,093 INFO [train.py:904] (7/8) Epoch 7, batch 2300, loss[loss=0.2483, simple_loss=0.315, pruned_loss=0.09084, over 16924.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2873, pruned_loss=0.06459, over 3304568.76 frames. ], batch size: 116, lr: 9.97e-03, grad_scale: 8.0 2023-04-28 15:36:10,198 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63221.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:36:23,468 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6292, 3.0931, 2.4232, 4.2822, 3.8616, 4.1822, 1.4562, 3.0804], device='cuda:7'), covar=tensor([0.1201, 0.0473, 0.1053, 0.0103, 0.0239, 0.0360, 0.1324, 0.0586], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0147, 0.0169, 0.0104, 0.0199, 0.0201, 0.0166, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 15:36:37,756 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:36:48,221 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63249.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:36:51,523 INFO [train.py:904] (7/8) Epoch 7, batch 2350, loss[loss=0.1956, simple_loss=0.2716, pruned_loss=0.05984, over 16436.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2871, pruned_loss=0.06496, over 3314966.17 frames. ], batch size: 146, lr: 9.96e-03, grad_scale: 4.0 2023-04-28 15:37:10,369 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-28 15:37:33,944 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63282.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:37:38,008 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.053e+02 2.699e+02 3.274e+02 4.399e+02 8.377e+02, threshold=6.549e+02, percent-clipped=2.0 2023-04-28 15:37:41,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8667, 4.3148, 3.4089, 2.3595, 3.1220, 2.3290, 4.6583, 4.1455], device='cuda:7'), covar=tensor([0.2161, 0.0562, 0.1167, 0.1854, 0.2396, 0.1602, 0.0301, 0.0709], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0260, 0.0274, 0.0259, 0.0287, 0.0211, 0.0254, 0.0280], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:37:43,611 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63289.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:38:00,172 INFO [train.py:904] (7/8) Epoch 7, batch 2400, loss[loss=0.2418, simple_loss=0.3144, pruned_loss=0.08458, over 15519.00 frames. ], tot_loss[loss=0.21, simple_loss=0.288, pruned_loss=0.06595, over 3307625.81 frames. ], batch size: 191, lr: 9.96e-03, grad_scale: 8.0 2023-04-28 15:38:23,922 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.48 vs. limit=5.0 2023-04-28 15:38:27,626 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63322.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:39:10,371 INFO [train.py:904] (7/8) Epoch 7, batch 2450, loss[loss=0.2049, simple_loss=0.2803, pruned_loss=0.06475, over 16226.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2877, pruned_loss=0.06501, over 3313781.69 frames. ], batch size: 165, lr: 9.96e-03, grad_scale: 8.0 2023-04-28 15:39:34,204 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63370.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:39:55,100 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.710e+02 2.805e+02 3.565e+02 4.384e+02 1.263e+03, threshold=7.129e+02, percent-clipped=5.0 2023-04-28 15:40:12,154 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 15:40:19,249 INFO [train.py:904] (7/8) Epoch 7, batch 2500, loss[loss=0.2966, simple_loss=0.3637, pruned_loss=0.1148, over 15630.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.289, pruned_loss=0.06557, over 3315192.60 frames. ], batch size: 191, lr: 9.95e-03, grad_scale: 4.0 2023-04-28 15:40:44,226 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63420.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:40:46,546 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63422.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:40:55,630 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63429.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:41:26,924 INFO [train.py:904] (7/8) Epoch 7, batch 2550, loss[loss=0.2009, simple_loss=0.2903, pruned_loss=0.05575, over 16695.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2881, pruned_loss=0.06472, over 3328396.17 frames. ], batch size: 57, lr: 9.95e-03, grad_scale: 4.0 2023-04-28 15:42:02,624 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63477.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:42:08,333 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63481.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:42:11,081 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63483.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:42:14,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 3.238e+02 3.870e+02 4.521e+02 8.304e+02, threshold=7.740e+02, percent-clipped=2.0 2023-04-28 15:42:35,907 INFO [train.py:904] (7/8) Epoch 7, batch 2600, loss[loss=0.2249, simple_loss=0.2971, pruned_loss=0.07639, over 16709.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2872, pruned_loss=0.06417, over 3324654.16 frames. ], batch size: 89, lr: 9.94e-03, grad_scale: 4.0 2023-04-28 15:43:42,053 INFO [train.py:904] (7/8) Epoch 7, batch 2650, loss[loss=0.2019, simple_loss=0.2809, pruned_loss=0.0615, over 15644.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2875, pruned_loss=0.06367, over 3330125.89 frames. ], batch size: 190, lr: 9.94e-03, grad_scale: 4.0 2023-04-28 15:44:06,255 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63569.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:44:15,613 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63577.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:44:27,540 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.764e+02 3.348e+02 3.954e+02 9.438e+02, threshold=6.696e+02, percent-clipped=1.0 2023-04-28 15:44:27,958 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63586.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:44:48,762 INFO [train.py:904] (7/8) Epoch 7, batch 2700, loss[loss=0.2003, simple_loss=0.2769, pruned_loss=0.06185, over 16894.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2861, pruned_loss=0.06206, over 3338089.31 frames. ], batch size: 96, lr: 9.94e-03, grad_scale: 4.0 2023-04-28 15:44:55,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4156, 4.1473, 4.2560, 4.6250, 4.7061, 4.3144, 4.6304, 4.7390], device='cuda:7'), covar=tensor([0.1319, 0.1250, 0.2068, 0.0803, 0.0831, 0.0972, 0.1086, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0478, 0.0586, 0.0742, 0.0584, 0.0449, 0.0449, 0.0459, 0.0511], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:45:28,094 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63630.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:45:52,007 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63647.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:45:58,007 INFO [train.py:904] (7/8) Epoch 7, batch 2750, loss[loss=0.1822, simple_loss=0.2687, pruned_loss=0.04789, over 17169.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2852, pruned_loss=0.06098, over 3344042.70 frames. ], batch size: 46, lr: 9.93e-03, grad_scale: 4.0 2023-04-28 15:46:38,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6048, 5.9968, 5.7429, 5.7427, 5.3208, 5.1266, 5.4644, 6.0209], device='cuda:7'), covar=tensor([0.0880, 0.0655, 0.0880, 0.0521, 0.0645, 0.0569, 0.0795, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0454, 0.0587, 0.0491, 0.0387, 0.0369, 0.0379, 0.0486, 0.0433], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:46:45,846 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.888e+02 2.696e+02 3.173e+02 3.925e+02 8.532e+02, threshold=6.347e+02, percent-clipped=1.0 2023-04-28 15:47:08,539 INFO [train.py:904] (7/8) Epoch 7, batch 2800, loss[loss=0.1742, simple_loss=0.2703, pruned_loss=0.03903, over 17114.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2848, pruned_loss=0.06059, over 3345376.49 frames. ], batch size: 47, lr: 9.93e-03, grad_scale: 8.0 2023-04-28 15:48:12,851 INFO [train.py:904] (7/8) Epoch 7, batch 2850, loss[loss=0.2186, simple_loss=0.2943, pruned_loss=0.07148, over 16456.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2843, pruned_loss=0.06048, over 3339931.59 frames. ], batch size: 75, lr: 9.92e-03, grad_scale: 8.0 2023-04-28 15:48:47,929 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63776.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:48:50,257 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63778.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:49:02,037 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.714e+02 3.034e+02 3.568e+02 4.214e+02 8.361e+02, threshold=7.135e+02, percent-clipped=4.0 2023-04-28 15:49:14,346 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 15:49:23,219 INFO [train.py:904] (7/8) Epoch 7, batch 2900, loss[loss=0.235, simple_loss=0.294, pruned_loss=0.088, over 16187.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2836, pruned_loss=0.0615, over 3338297.20 frames. ], batch size: 165, lr: 9.92e-03, grad_scale: 8.0 2023-04-28 15:49:35,847 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63810.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:50:22,423 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63845.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:50:32,280 INFO [train.py:904] (7/8) Epoch 7, batch 2950, loss[loss=0.173, simple_loss=0.2516, pruned_loss=0.04718, over 15846.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2832, pruned_loss=0.06288, over 3322386.76 frames. ], batch size: 35, lr: 9.92e-03, grad_scale: 8.0 2023-04-28 15:50:58,304 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63871.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:51:05,322 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63877.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:51:17,104 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.041e+02 3.004e+02 3.573e+02 4.238e+02 9.088e+02, threshold=7.147e+02, percent-clipped=1.0 2023-04-28 15:51:38,117 INFO [train.py:904] (7/8) Epoch 7, batch 3000, loss[loss=0.1985, simple_loss=0.2876, pruned_loss=0.0547, over 17131.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2841, pruned_loss=0.06427, over 3318429.71 frames. ], batch size: 47, lr: 9.91e-03, grad_scale: 8.0 2023-04-28 15:51:38,117 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 15:51:46,841 INFO [train.py:938] (7/8) Epoch 7, validation: loss=0.1489, simple_loss=0.2553, pruned_loss=0.02124, over 944034.00 frames. 2023-04-28 15:51:46,842 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 15:51:48,002 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2849, 4.1330, 4.3474, 4.1851, 4.1675, 4.8441, 4.5447, 4.2195], device='cuda:7'), covar=tensor([0.1580, 0.2277, 0.1737, 0.2120, 0.3025, 0.1309, 0.1249, 0.2420], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0461, 0.0459, 0.0385, 0.0524, 0.0491, 0.0368, 0.0522], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 15:51:54,361 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63906.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:07,341 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63916.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:20,149 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:20,169 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:29,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5503, 4.3591, 4.0074, 2.1387, 3.0912, 2.6238, 3.8191, 4.0915], device='cuda:7'), covar=tensor([0.0290, 0.0564, 0.0500, 0.1590, 0.0771, 0.0932, 0.0630, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0140, 0.0156, 0.0140, 0.0132, 0.0124, 0.0138, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 15:52:42,754 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63942.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:52:43,458 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 15:52:56,544 INFO [train.py:904] (7/8) Epoch 7, batch 3050, loss[loss=0.235, simple_loss=0.3016, pruned_loss=0.08418, over 16653.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2832, pruned_loss=0.06359, over 3315977.19 frames. ], batch size: 134, lr: 9.91e-03, grad_scale: 4.0 2023-04-28 15:53:30,259 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:53:43,708 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.789e+02 3.428e+02 3.925e+02 6.614e+02, threshold=6.856e+02, percent-clipped=0.0 2023-04-28 15:54:06,485 INFO [train.py:904] (7/8) Epoch 7, batch 3100, loss[loss=0.2027, simple_loss=0.2892, pruned_loss=0.05808, over 17045.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2821, pruned_loss=0.06257, over 3323355.89 frames. ], batch size: 55, lr: 9.91e-03, grad_scale: 4.0 2023-04-28 15:54:10,523 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5551, 5.9724, 5.6656, 5.8205, 5.3356, 5.0920, 5.4308, 6.0418], device='cuda:7'), covar=tensor([0.0990, 0.0687, 0.0950, 0.0494, 0.0693, 0.0613, 0.0769, 0.0707], device='cuda:7'), in_proj_covar=tensor([0.0464, 0.0601, 0.0500, 0.0393, 0.0376, 0.0385, 0.0490, 0.0440], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:54:30,803 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7332, 3.6118, 2.7630, 5.0975, 4.5001, 4.8592, 1.6509, 3.4830], device='cuda:7'), covar=tensor([0.1382, 0.0542, 0.1115, 0.0147, 0.0267, 0.0292, 0.1510, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0148, 0.0168, 0.0107, 0.0201, 0.0200, 0.0166, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 15:54:36,524 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64023.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:55:16,388 INFO [train.py:904] (7/8) Epoch 7, batch 3150, loss[loss=0.2227, simple_loss=0.2958, pruned_loss=0.07479, over 16714.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2816, pruned_loss=0.06294, over 3317577.83 frames. ], batch size: 62, lr: 9.90e-03, grad_scale: 4.0 2023-04-28 15:55:49,716 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64076.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:55:52,150 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64078.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:55:52,282 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2605, 3.4652, 2.0278, 3.5025, 2.5340, 3.6209, 2.0126, 2.6858], device='cuda:7'), covar=tensor([0.0173, 0.0300, 0.1267, 0.0182, 0.0672, 0.0446, 0.1159, 0.0567], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0163, 0.0178, 0.0101, 0.0162, 0.0205, 0.0189, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 15:56:00,131 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64084.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:56:03,684 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.613e+02 3.331e+02 4.706e+02 9.304e+02, threshold=6.662e+02, percent-clipped=4.0 2023-04-28 15:56:24,283 INFO [train.py:904] (7/8) Epoch 7, batch 3200, loss[loss=0.1971, simple_loss=0.2668, pruned_loss=0.06367, over 16746.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.281, pruned_loss=0.06197, over 3314483.56 frames. ], batch size: 124, lr: 9.90e-03, grad_scale: 8.0 2023-04-28 15:56:36,010 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4207, 4.5392, 4.8824, 4.8813, 4.8897, 4.5175, 4.3448, 4.4552], device='cuda:7'), covar=tensor([0.0370, 0.0549, 0.0376, 0.0489, 0.0585, 0.0438, 0.1260, 0.0435], device='cuda:7'), in_proj_covar=tensor([0.0294, 0.0298, 0.0297, 0.0285, 0.0345, 0.0317, 0.0427, 0.0255], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 15:56:43,946 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0662, 2.3942, 1.7041, 2.2521, 2.9223, 2.6010, 3.1755, 3.0914], device='cuda:7'), covar=tensor([0.0069, 0.0222, 0.0315, 0.0227, 0.0112, 0.0202, 0.0122, 0.0116], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0177, 0.0172, 0.0174, 0.0174, 0.0177, 0.0175, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:56:54,787 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64124.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:56:57,820 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64126.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:57:32,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7364, 3.3849, 2.7553, 4.8063, 4.1156, 4.5438, 1.5338, 3.2648], device='cuda:7'), covar=tensor([0.1231, 0.0488, 0.1001, 0.0132, 0.0268, 0.0305, 0.1351, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0148, 0.0170, 0.0108, 0.0203, 0.0201, 0.0167, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 15:57:33,494 INFO [train.py:904] (7/8) Epoch 7, batch 3250, loss[loss=0.215, simple_loss=0.3028, pruned_loss=0.06357, over 16745.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2816, pruned_loss=0.06223, over 3307904.50 frames. ], batch size: 62, lr: 9.89e-03, grad_scale: 8.0 2023-04-28 15:57:53,207 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64166.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:58:21,283 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.755e+02 2.809e+02 3.257e+02 3.904e+02 7.986e+02, threshold=6.515e+02, percent-clipped=1.0 2023-04-28 15:58:42,239 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64201.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:58:43,067 INFO [train.py:904] (7/8) Epoch 7, batch 3300, loss[loss=0.1677, simple_loss=0.2509, pruned_loss=0.04226, over 17163.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.284, pruned_loss=0.06323, over 3303087.08 frames. ], batch size: 47, lr: 9.89e-03, grad_scale: 8.0 2023-04-28 15:59:16,286 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64225.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:59:25,481 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64232.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:59:34,126 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5372, 2.5254, 2.1254, 2.3646, 2.9639, 2.7135, 3.5451, 3.1514], device='cuda:7'), covar=tensor([0.0040, 0.0210, 0.0238, 0.0237, 0.0139, 0.0183, 0.0104, 0.0118], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0178, 0.0173, 0.0175, 0.0175, 0.0177, 0.0177, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 15:59:41,237 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64242.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 15:59:51,978 INFO [train.py:904] (7/8) Epoch 7, batch 3350, loss[loss=0.1738, simple_loss=0.2606, pruned_loss=0.04351, over 16409.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2842, pruned_loss=0.06252, over 3304470.62 frames. ], batch size: 75, lr: 9.89e-03, grad_scale: 8.0 2023-04-28 15:59:54,380 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5347, 2.4657, 2.1047, 2.2585, 2.8956, 2.6087, 3.4279, 3.1193], device='cuda:7'), covar=tensor([0.0051, 0.0240, 0.0256, 0.0264, 0.0139, 0.0218, 0.0127, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0178, 0.0173, 0.0176, 0.0175, 0.0178, 0.0176, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:00:20,539 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:00:22,695 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:00:42,049 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.871e+02 3.440e+02 4.131e+02 8.469e+02, threshold=6.880e+02, percent-clipped=3.0 2023-04-28 16:00:45,938 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64290.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:00:49,497 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64293.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:01:02,169 INFO [train.py:904] (7/8) Epoch 7, batch 3400, loss[loss=0.1841, simple_loss=0.2781, pruned_loss=0.04509, over 17030.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2843, pruned_loss=0.06211, over 3311070.64 frames. ], batch size: 50, lr: 9.88e-03, grad_scale: 8.0 2023-04-28 16:02:03,011 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64345.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:02:10,867 INFO [train.py:904] (7/8) Epoch 7, batch 3450, loss[loss=0.2073, simple_loss=0.2753, pruned_loss=0.06966, over 16738.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.282, pruned_loss=0.06158, over 3304350.77 frames. ], batch size: 134, lr: 9.88e-03, grad_scale: 8.0 2023-04-28 16:02:50,125 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64379.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:03:01,788 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.744e+02 2.661e+02 3.216e+02 3.896e+02 8.084e+02, threshold=6.431e+02, percent-clipped=1.0 2023-04-28 16:03:21,643 INFO [train.py:904] (7/8) Epoch 7, batch 3500, loss[loss=0.2356, simple_loss=0.297, pruned_loss=0.08714, over 16963.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2803, pruned_loss=0.06153, over 3309479.81 frames. ], batch size: 109, lr: 9.87e-03, grad_scale: 8.0 2023-04-28 16:03:28,754 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64406.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:04:30,083 INFO [train.py:904] (7/8) Epoch 7, batch 3550, loss[loss=0.1613, simple_loss=0.2441, pruned_loss=0.03926, over 16776.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2792, pruned_loss=0.06106, over 3313480.91 frames. ], batch size: 39, lr: 9.87e-03, grad_scale: 8.0 2023-04-28 16:04:48,904 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64465.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:04:50,651 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:05:00,357 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-28 16:05:18,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.678e+02 3.341e+02 4.465e+02 1.003e+03, threshold=6.682e+02, percent-clipped=6.0 2023-04-28 16:05:27,561 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 16:05:38,443 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64501.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:05:39,319 INFO [train.py:904] (7/8) Epoch 7, batch 3600, loss[loss=0.2206, simple_loss=0.2855, pruned_loss=0.07779, over 16808.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2782, pruned_loss=0.0602, over 3322035.42 frames. ], batch size: 116, lr: 9.87e-03, grad_scale: 8.0 2023-04-28 16:05:56,510 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64514.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:06:14,306 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64526.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:06:22,125 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4448, 3.8474, 3.8618, 1.8644, 4.0326, 4.0695, 3.2017, 3.1956], device='cuda:7'), covar=tensor([0.0722, 0.0105, 0.0144, 0.1125, 0.0072, 0.0103, 0.0304, 0.0329], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0085, 0.0138, 0.0073, 0.0093, 0.0120, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 16:06:24,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0410, 1.8976, 2.3021, 2.8976, 2.6700, 2.9864, 1.8947, 3.0943], device='cuda:7'), covar=tensor([0.0087, 0.0229, 0.0177, 0.0122, 0.0128, 0.0104, 0.0223, 0.0055], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0158, 0.0143, 0.0144, 0.0150, 0.0107, 0.0151, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 16:06:45,522 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64549.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:06:48,848 INFO [train.py:904] (7/8) Epoch 7, batch 3650, loss[loss=0.1784, simple_loss=0.2393, pruned_loss=0.05873, over 16723.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2767, pruned_loss=0.06109, over 3318114.34 frames. ], batch size: 83, lr: 9.86e-03, grad_scale: 8.0 2023-04-28 16:07:19,362 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64572.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:07:40,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.014e+02 2.562e+02 3.124e+02 4.018e+02 1.198e+03, threshold=6.249e+02, percent-clipped=5.0 2023-04-28 16:07:43,320 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64588.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:07:44,584 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64589.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:08:02,012 INFO [train.py:904] (7/8) Epoch 7, batch 3700, loss[loss=0.1941, simple_loss=0.266, pruned_loss=0.06112, over 16523.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2761, pruned_loss=0.06294, over 3303108.19 frames. ], batch size: 68, lr: 9.86e-03, grad_scale: 8.0 2023-04-28 16:08:25,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3964, 4.3637, 4.8119, 4.8130, 4.7913, 4.4446, 4.4613, 4.2750], device='cuda:7'), covar=tensor([0.0253, 0.0569, 0.0338, 0.0362, 0.0381, 0.0311, 0.0724, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0290, 0.0290, 0.0279, 0.0336, 0.0304, 0.0410, 0.0248], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 16:08:27,964 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64620.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:09:09,952 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64648.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:09:12,343 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64650.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:09:14,080 INFO [train.py:904] (7/8) Epoch 7, batch 3750, loss[loss=0.2331, simple_loss=0.3024, pruned_loss=0.08185, over 11479.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2772, pruned_loss=0.06458, over 3277237.73 frames. ], batch size: 246, lr: 9.86e-03, grad_scale: 8.0 2023-04-28 16:09:55,122 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64679.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:04,728 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.843e+02 3.454e+02 4.284e+02 7.405e+02, threshold=6.908e+02, percent-clipped=2.0 2023-04-28 16:10:24,974 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64701.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:25,839 INFO [train.py:904] (7/8) Epoch 7, batch 3800, loss[loss=0.1982, simple_loss=0.275, pruned_loss=0.06065, over 16518.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2778, pruned_loss=0.06576, over 3265126.58 frames. ], batch size: 68, lr: 9.85e-03, grad_scale: 4.0 2023-04-28 16:10:28,010 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64703.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:38,646 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64709.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:10:39,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5929, 3.8808, 4.1040, 2.0468, 4.2604, 4.2370, 3.2098, 3.2465], device='cuda:7'), covar=tensor([0.0718, 0.0115, 0.0105, 0.1057, 0.0056, 0.0098, 0.0285, 0.0362], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0084, 0.0138, 0.0072, 0.0091, 0.0120, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 16:11:05,639 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64727.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:31,499 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64745.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:32,816 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 16:11:39,994 INFO [train.py:904] (7/8) Epoch 7, batch 3850, loss[loss=0.2027, simple_loss=0.2702, pruned_loss=0.06763, over 16807.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2777, pruned_loss=0.06607, over 3273357.83 frames. ], batch size: 109, lr: 9.85e-03, grad_scale: 4.0 2023-04-28 16:11:41,141 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64752.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:11:47,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7605, 5.0405, 5.1478, 5.1496, 5.1081, 5.6532, 5.3358, 5.0933], device='cuda:7'), covar=tensor([0.1075, 0.1489, 0.1446, 0.1583, 0.2489, 0.0891, 0.1089, 0.2038], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0448, 0.0447, 0.0374, 0.0506, 0.0476, 0.0359, 0.0506], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 16:11:52,029 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 16:11:58,605 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64764.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:12:10,713 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8315, 4.8646, 5.3246, 5.2528, 5.3192, 4.9107, 4.7026, 4.6159], device='cuda:7'), covar=tensor([0.0302, 0.0506, 0.0302, 0.0486, 0.0433, 0.0337, 0.1109, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0290, 0.0289, 0.0280, 0.0334, 0.0305, 0.0409, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 16:12:33,102 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.797e+02 3.034e+02 3.668e+02 1.110e+03, threshold=6.069e+02, percent-clipped=1.0 2023-04-28 16:12:40,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5711, 3.1806, 2.8255, 1.8814, 2.5622, 2.0910, 2.9763, 3.2067], device='cuda:7'), covar=tensor([0.0312, 0.0633, 0.0556, 0.1531, 0.0795, 0.0880, 0.0693, 0.0743], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0137, 0.0153, 0.0139, 0.0131, 0.0123, 0.0136, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 16:12:52,611 INFO [train.py:904] (7/8) Epoch 7, batch 3900, loss[loss=0.1887, simple_loss=0.2553, pruned_loss=0.061, over 16775.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2771, pruned_loss=0.06671, over 3277144.52 frames. ], batch size: 89, lr: 9.84e-03, grad_scale: 4.0 2023-04-28 16:13:00,357 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64806.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:13:08,285 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64813.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:13:21,591 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64821.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:14:02,906 INFO [train.py:904] (7/8) Epoch 7, batch 3950, loss[loss=0.1884, simple_loss=0.2591, pruned_loss=0.05888, over 16853.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2762, pruned_loss=0.06667, over 3292642.08 frames. ], batch size: 96, lr: 9.84e-03, grad_scale: 4.0 2023-04-28 16:14:17,009 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.74 vs. limit=5.0 2023-04-28 16:14:53,036 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.973e+02 2.882e+02 3.338e+02 4.272e+02 6.667e+02, threshold=6.675e+02, percent-clipped=4.0 2023-04-28 16:14:54,194 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64888.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:15:12,241 INFO [train.py:904] (7/8) Epoch 7, batch 4000, loss[loss=0.2007, simple_loss=0.2829, pruned_loss=0.05928, over 15531.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2762, pruned_loss=0.06663, over 3299528.57 frames. ], batch size: 190, lr: 9.84e-03, grad_scale: 8.0 2023-04-28 16:16:02,049 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=64936.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:16:17,003 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64945.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:16:24,728 INFO [train.py:904] (7/8) Epoch 7, batch 4050, loss[loss=0.209, simple_loss=0.2837, pruned_loss=0.06716, over 16629.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2765, pruned_loss=0.06536, over 3290647.42 frames. ], batch size: 62, lr: 9.83e-03, grad_scale: 8.0 2023-04-28 16:16:55,136 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3005, 4.1910, 4.2422, 2.6994, 3.7205, 4.1711, 3.9445, 2.2932], device='cuda:7'), covar=tensor([0.0392, 0.0020, 0.0022, 0.0249, 0.0038, 0.0045, 0.0030, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0059, 0.0060, 0.0114, 0.0064, 0.0074, 0.0066, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 16:16:59,809 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:17:03,957 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8535, 1.8860, 2.2553, 3.2319, 1.9781, 2.2445, 2.1957, 1.9838], device='cuda:7'), covar=tensor([0.0780, 0.2623, 0.1352, 0.0421, 0.3099, 0.1571, 0.2232, 0.2609], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0363, 0.0301, 0.0326, 0.0391, 0.0396, 0.0325, 0.0430], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:17:16,149 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.141e+02 2.464e+02 3.078e+02 9.943e+02, threshold=4.928e+02, percent-clipped=3.0 2023-04-28 16:17:19,076 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0460, 3.7909, 3.7345, 2.3011, 3.3155, 3.6870, 3.5412, 2.0382], device='cuda:7'), covar=tensor([0.0363, 0.0017, 0.0020, 0.0250, 0.0039, 0.0047, 0.0032, 0.0295], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0059, 0.0060, 0.0114, 0.0064, 0.0074, 0.0066, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 16:17:29,209 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-28 16:17:35,706 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65001.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:17:36,415 INFO [train.py:904] (7/8) Epoch 7, batch 4100, loss[loss=0.2166, simple_loss=0.2973, pruned_loss=0.06795, over 16758.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2769, pruned_loss=0.06392, over 3291836.45 frames. ], batch size: 124, lr: 9.83e-03, grad_scale: 8.0 2023-04-28 16:17:39,320 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65004.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:17:58,277 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7899, 3.9716, 3.7248, 3.6096, 3.2224, 3.8454, 3.5391, 3.4786], device='cuda:7'), covar=tensor([0.0655, 0.0356, 0.0321, 0.0242, 0.1041, 0.0410, 0.0851, 0.0602], device='cuda:7'), in_proj_covar=tensor([0.0211, 0.0246, 0.0252, 0.0226, 0.0290, 0.0256, 0.0178, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 16:18:07,338 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 16:18:29,702 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65037.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:18:47,167 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65049.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:18:51,091 INFO [train.py:904] (7/8) Epoch 7, batch 4150, loss[loss=0.2268, simple_loss=0.3131, pruned_loss=0.07025, over 16672.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2859, pruned_loss=0.06802, over 3234932.43 frames. ], batch size: 134, lr: 9.83e-03, grad_scale: 8.0 2023-04-28 16:18:56,847 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65055.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:19:01,934 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65059.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:19:44,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 2.789e+02 3.341e+02 3.997e+02 8.596e+02, threshold=6.682e+02, percent-clipped=9.0 2023-04-28 16:20:03,752 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65101.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:04,513 INFO [train.py:904] (7/8) Epoch 7, batch 4200, loss[loss=0.2539, simple_loss=0.3393, pruned_loss=0.08426, over 15228.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.293, pruned_loss=0.06962, over 3221964.95 frames. ], batch size: 190, lr: 9.82e-03, grad_scale: 8.0 2023-04-28 16:20:13,292 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65108.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:24,632 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:32,595 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65121.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:20:38,954 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 16:20:57,230 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0164, 2.6964, 2.6401, 1.8077, 2.8271, 2.8176, 2.4653, 2.3703], device='cuda:7'), covar=tensor([0.0695, 0.0153, 0.0155, 0.0891, 0.0085, 0.0141, 0.0359, 0.0385], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0096, 0.0082, 0.0137, 0.0071, 0.0088, 0.0119, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 16:21:16,727 INFO [train.py:904] (7/8) Epoch 7, batch 4250, loss[loss=0.1979, simple_loss=0.2915, pruned_loss=0.05219, over 16820.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.2966, pruned_loss=0.07016, over 3188708.96 frames. ], batch size: 83, lr: 9.82e-03, grad_scale: 8.0 2023-04-28 16:21:42,347 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65169.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:22:08,031 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.682e+02 2.671e+02 3.307e+02 3.991e+02 9.211e+02, threshold=6.614e+02, percent-clipped=3.0 2023-04-28 16:22:29,060 INFO [train.py:904] (7/8) Epoch 7, batch 4300, loss[loss=0.2411, simple_loss=0.314, pruned_loss=0.08404, over 16869.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2972, pruned_loss=0.06851, over 3191106.77 frames. ], batch size: 42, lr: 9.81e-03, grad_scale: 8.0 2023-04-28 16:23:31,876 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65245.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:23:41,836 INFO [train.py:904] (7/8) Epoch 7, batch 4350, loss[loss=0.2088, simple_loss=0.2951, pruned_loss=0.06122, over 16444.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.3006, pruned_loss=0.06957, over 3181489.71 frames. ], batch size: 146, lr: 9.81e-03, grad_scale: 8.0 2023-04-28 16:24:25,435 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 16:24:30,754 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65285.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:24:33,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.885e+02 3.415e+02 4.232e+02 9.360e+02, threshold=6.829e+02, percent-clipped=2.0 2023-04-28 16:24:42,766 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65293.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:24:54,508 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 16:24:54,960 INFO [train.py:904] (7/8) Epoch 7, batch 4400, loss[loss=0.2214, simple_loss=0.3017, pruned_loss=0.07054, over 16682.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.3024, pruned_loss=0.07058, over 3174291.73 frames. ], batch size: 134, lr: 9.81e-03, grad_scale: 8.0 2023-04-28 16:24:57,767 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65304.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:25:39,699 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65332.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:25:59,830 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65346.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:26:07,822 INFO [train.py:904] (7/8) Epoch 7, batch 4450, loss[loss=0.2292, simple_loss=0.3113, pruned_loss=0.07354, over 16819.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3048, pruned_loss=0.07091, over 3182906.93 frames. ], batch size: 83, lr: 9.80e-03, grad_scale: 8.0 2023-04-28 16:26:08,160 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65352.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:26:17,519 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65359.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:26:58,768 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.495e+02 2.995e+02 3.500e+02 6.394e+02, threshold=5.989e+02, percent-clipped=0.0 2023-04-28 16:27:19,278 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65401.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:20,145 INFO [train.py:904] (7/8) Epoch 7, batch 4500, loss[loss=0.2066, simple_loss=0.2954, pruned_loss=0.05893, over 16703.00 frames. ], tot_loss[loss=0.223, simple_loss=0.3044, pruned_loss=0.07076, over 3193792.63 frames. ], batch size: 124, lr: 9.80e-03, grad_scale: 8.0 2023-04-28 16:27:26,561 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65407.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:28,764 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65408.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:32,343 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65411.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:27:45,123 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2679, 5.2632, 4.9547, 4.3467, 5.1037, 2.0184, 4.8667, 4.9489], device='cuda:7'), covar=tensor([0.0033, 0.0024, 0.0078, 0.0269, 0.0040, 0.1713, 0.0067, 0.0093], device='cuda:7'), in_proj_covar=tensor([0.0106, 0.0093, 0.0141, 0.0140, 0.0109, 0.0154, 0.0125, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:28:07,555 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 16:28:14,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.8415, 6.1191, 5.8381, 5.9515, 5.5653, 5.0668, 5.6715, 6.2450], device='cuda:7'), covar=tensor([0.0651, 0.0557, 0.0937, 0.0438, 0.0605, 0.0545, 0.0592, 0.0617], device='cuda:7'), in_proj_covar=tensor([0.0427, 0.0551, 0.0465, 0.0362, 0.0348, 0.0367, 0.0456, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:28:27,210 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65449.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:28:31,295 INFO [train.py:904] (7/8) Epoch 7, batch 4550, loss[loss=0.2307, simple_loss=0.3158, pruned_loss=0.07275, over 16702.00 frames. ], tot_loss[loss=0.2244, simple_loss=0.3055, pruned_loss=0.07164, over 3197580.61 frames. ], batch size: 89, lr: 9.80e-03, grad_scale: 8.0 2023-04-28 16:28:38,495 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65456.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:29:21,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.206e+02 2.596e+02 3.051e+02 6.081e+02, threshold=5.193e+02, percent-clipped=1.0 2023-04-28 16:29:41,305 INFO [train.py:904] (7/8) Epoch 7, batch 4600, loss[loss=0.2255, simple_loss=0.3032, pruned_loss=0.07388, over 16333.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.3061, pruned_loss=0.07143, over 3198120.62 frames. ], batch size: 35, lr: 9.79e-03, grad_scale: 8.0 2023-04-28 16:29:57,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3256, 4.0815, 4.1250, 4.4737, 4.5441, 4.1495, 4.5403, 4.6029], device='cuda:7'), covar=tensor([0.0879, 0.0851, 0.1702, 0.0607, 0.0481, 0.1040, 0.0604, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0428, 0.0526, 0.0664, 0.0531, 0.0404, 0.0417, 0.0411, 0.0453], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:30:24,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65532.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:30:50,595 INFO [train.py:904] (7/8) Epoch 7, batch 4650, loss[loss=0.2142, simple_loss=0.2907, pruned_loss=0.06885, over 17051.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3047, pruned_loss=0.07102, over 3219337.00 frames. ], batch size: 53, lr: 9.79e-03, grad_scale: 8.0 2023-04-28 16:31:00,935 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2320, 4.0235, 4.2117, 4.4504, 4.5404, 4.1204, 4.5115, 4.5740], device='cuda:7'), covar=tensor([0.0905, 0.0807, 0.1197, 0.0454, 0.0377, 0.0966, 0.0460, 0.0378], device='cuda:7'), in_proj_covar=tensor([0.0425, 0.0523, 0.0660, 0.0529, 0.0401, 0.0415, 0.0410, 0.0453], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:31:11,963 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-04-28 16:31:41,973 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.212e+02 2.707e+02 3.141e+02 8.468e+02, threshold=5.414e+02, percent-clipped=4.0 2023-04-28 16:31:49,836 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65593.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:32:03,160 INFO [train.py:904] (7/8) Epoch 7, batch 4700, loss[loss=0.1996, simple_loss=0.2859, pruned_loss=0.05662, over 16756.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3014, pruned_loss=0.06965, over 3209813.85 frames. ], batch size: 124, lr: 9.78e-03, grad_scale: 8.0 2023-04-28 16:32:24,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5070, 4.6827, 4.7311, 4.7331, 4.6748, 5.2188, 4.8380, 4.5004], device='cuda:7'), covar=tensor([0.1014, 0.1445, 0.1117, 0.1539, 0.2138, 0.0967, 0.0969, 0.2327], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0432, 0.0428, 0.0361, 0.0493, 0.0465, 0.0345, 0.0498], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 16:32:45,709 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65632.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:32:56,223 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65641.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:33:12,639 INFO [train.py:904] (7/8) Epoch 7, batch 4750, loss[loss=0.2094, simple_loss=0.2942, pruned_loss=0.06229, over 16296.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2983, pruned_loss=0.06815, over 3204570.95 frames. ], batch size: 165, lr: 9.78e-03, grad_scale: 8.0 2023-04-28 16:33:53,640 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65680.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:34:03,334 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.191e+02 2.638e+02 3.130e+02 6.213e+02, threshold=5.275e+02, percent-clipped=2.0 2023-04-28 16:34:22,803 INFO [train.py:904] (7/8) Epoch 7, batch 4800, loss[loss=0.2198, simple_loss=0.3115, pruned_loss=0.06407, over 16196.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2953, pruned_loss=0.06655, over 3190676.98 frames. ], batch size: 165, lr: 9.78e-03, grad_scale: 8.0 2023-04-28 16:34:36,975 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65711.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:35:03,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8278, 3.8352, 3.1259, 2.4199, 3.0628, 2.5616, 4.2020, 3.9387], device='cuda:7'), covar=tensor([0.2055, 0.0812, 0.1224, 0.1699, 0.1808, 0.1315, 0.0380, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0285, 0.0252, 0.0269, 0.0257, 0.0284, 0.0207, 0.0249, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:35:25,508 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 16:35:36,924 INFO [train.py:904] (7/8) Epoch 7, batch 4850, loss[loss=0.2178, simple_loss=0.2992, pruned_loss=0.06824, over 16876.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.2969, pruned_loss=0.06615, over 3184348.34 frames. ], batch size: 109, lr: 9.77e-03, grad_scale: 8.0 2023-04-28 16:35:45,502 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65759.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:36:28,554 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.411e+02 2.781e+02 3.259e+02 6.496e+02, threshold=5.561e+02, percent-clipped=1.0 2023-04-28 16:36:48,887 INFO [train.py:904] (7/8) Epoch 7, batch 4900, loss[loss=0.1889, simple_loss=0.2665, pruned_loss=0.05562, over 16486.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2963, pruned_loss=0.0652, over 3171139.97 frames. ], batch size: 35, lr: 9.77e-03, grad_scale: 8.0 2023-04-28 16:37:44,008 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65838.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:38:02,373 INFO [train.py:904] (7/8) Epoch 7, batch 4950, loss[loss=0.2143, simple_loss=0.3033, pruned_loss=0.06261, over 16886.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2958, pruned_loss=0.06446, over 3170204.17 frames. ], batch size: 116, lr: 9.77e-03, grad_scale: 8.0 2023-04-28 16:38:11,745 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 16:38:53,491 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.669e+02 3.212e+02 3.846e+02 7.226e+02, threshold=6.424e+02, percent-clipped=6.0 2023-04-28 16:38:54,457 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65888.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:39:08,356 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65899.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:39:08,429 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5091, 2.9294, 2.3232, 4.5715, 3.4405, 4.0982, 1.5568, 2.9245], device='cuda:7'), covar=tensor([0.1327, 0.0575, 0.1190, 0.0102, 0.0256, 0.0336, 0.1400, 0.0800], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0147, 0.0170, 0.0103, 0.0197, 0.0195, 0.0167, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 16:39:12,693 INFO [train.py:904] (7/8) Epoch 7, batch 5000, loss[loss=0.2196, simple_loss=0.3035, pruned_loss=0.06779, over 16202.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2975, pruned_loss=0.06476, over 3163412.28 frames. ], batch size: 35, lr: 9.76e-03, grad_scale: 8.0 2023-04-28 16:39:54,779 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-28 16:40:09,111 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65941.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:40:21,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0262, 4.9205, 4.8182, 4.6464, 4.3246, 4.8426, 4.8565, 4.5586], device='cuda:7'), covar=tensor([0.0354, 0.0299, 0.0208, 0.0179, 0.0878, 0.0295, 0.0213, 0.0459], device='cuda:7'), in_proj_covar=tensor([0.0206, 0.0242, 0.0246, 0.0220, 0.0279, 0.0249, 0.0172, 0.0280], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:40:23,706 INFO [train.py:904] (7/8) Epoch 7, batch 5050, loss[loss=0.2217, simple_loss=0.3103, pruned_loss=0.06655, over 15504.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2971, pruned_loss=0.06403, over 3181551.84 frames. ], batch size: 191, lr: 9.76e-03, grad_scale: 8.0 2023-04-28 16:41:14,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.609e+02 2.967e+02 3.441e+02 6.265e+02, threshold=5.933e+02, percent-clipped=0.0 2023-04-28 16:41:16,170 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=65989.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:41:35,940 INFO [train.py:904] (7/8) Epoch 7, batch 5100, loss[loss=0.1871, simple_loss=0.2771, pruned_loss=0.04853, over 16758.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2947, pruned_loss=0.06268, over 3199487.43 frames. ], batch size: 134, lr: 9.75e-03, grad_scale: 8.0 2023-04-28 16:42:26,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6371, 2.9582, 2.4969, 4.5549, 3.3020, 4.2169, 1.4974, 3.0894], device='cuda:7'), covar=tensor([0.1420, 0.0580, 0.1125, 0.0077, 0.0174, 0.0280, 0.1504, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0146, 0.0169, 0.0103, 0.0196, 0.0193, 0.0165, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 16:42:43,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2427, 4.1757, 4.1581, 3.4896, 4.1200, 1.7451, 3.9203, 3.9658], device='cuda:7'), covar=tensor([0.0074, 0.0063, 0.0085, 0.0321, 0.0068, 0.1802, 0.0093, 0.0138], device='cuda:7'), in_proj_covar=tensor([0.0103, 0.0091, 0.0138, 0.0137, 0.0106, 0.0151, 0.0122, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:42:46,397 INFO [train.py:904] (7/8) Epoch 7, batch 5150, loss[loss=0.2386, simple_loss=0.3259, pruned_loss=0.07567, over 16714.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2947, pruned_loss=0.06139, over 3212256.23 frames. ], batch size: 124, lr: 9.75e-03, grad_scale: 8.0 2023-04-28 16:43:37,178 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.591e+02 3.145e+02 3.855e+02 6.325e+02, threshold=6.290e+02, percent-clipped=3.0 2023-04-28 16:43:42,894 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 16:43:52,869 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 16:43:55,947 INFO [train.py:904] (7/8) Epoch 7, batch 5200, loss[loss=0.2008, simple_loss=0.2883, pruned_loss=0.05661, over 16711.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2939, pruned_loss=0.06143, over 3210995.97 frames. ], batch size: 89, lr: 9.75e-03, grad_scale: 8.0 2023-04-28 16:44:56,277 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:45:06,338 INFO [train.py:904] (7/8) Epoch 7, batch 5250, loss[loss=0.2206, simple_loss=0.3058, pruned_loss=0.06772, over 16335.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.291, pruned_loss=0.06096, over 3205596.93 frames. ], batch size: 146, lr: 9.74e-03, grad_scale: 8.0 2023-04-28 16:45:19,113 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66160.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:45:59,720 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.502e+02 2.872e+02 3.517e+02 7.118e+02, threshold=5.743e+02, percent-clipped=1.0 2023-04-28 16:46:00,096 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66188.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:46:03,723 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6615, 2.1468, 2.2422, 4.4280, 2.1092, 2.8243, 2.3135, 2.4463], device='cuda:7'), covar=tensor([0.0684, 0.2420, 0.1634, 0.0257, 0.3019, 0.1510, 0.2210, 0.2311], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0347, 0.0292, 0.0315, 0.0384, 0.0377, 0.0311, 0.0411], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:46:06,941 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66194.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:46:11,225 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5678, 2.0404, 1.7607, 1.9894, 2.4412, 2.2598, 2.5177, 2.6905], device='cuda:7'), covar=tensor([0.0064, 0.0231, 0.0289, 0.0274, 0.0131, 0.0210, 0.0112, 0.0123], device='cuda:7'), in_proj_covar=tensor([0.0102, 0.0172, 0.0172, 0.0171, 0.0167, 0.0174, 0.0162, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:46:18,402 INFO [train.py:904] (7/8) Epoch 7, batch 5300, loss[loss=0.1903, simple_loss=0.2688, pruned_loss=0.05589, over 16424.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2874, pruned_loss=0.05959, over 3210773.96 frames. ], batch size: 146, lr: 9.74e-03, grad_scale: 8.0 2023-04-28 16:46:23,352 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66205.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:46:45,731 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66221.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:47:06,139 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66236.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:47:28,968 INFO [train.py:904] (7/8) Epoch 7, batch 5350, loss[loss=0.2281, simple_loss=0.3085, pruned_loss=0.07385, over 16704.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2862, pruned_loss=0.05928, over 3206721.38 frames. ], batch size: 62, lr: 9.74e-03, grad_scale: 8.0 2023-04-28 16:47:36,938 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66258.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:48:20,491 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.423e+02 2.781e+02 3.267e+02 5.283e+02, threshold=5.561e+02, percent-clipped=0.0 2023-04-28 16:48:33,225 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5415, 4.2425, 3.8675, 2.1259, 3.1948, 2.6016, 3.8667, 4.2371], device='cuda:7'), covar=tensor([0.0216, 0.0467, 0.0518, 0.1500, 0.0707, 0.0791, 0.0573, 0.0642], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0131, 0.0155, 0.0141, 0.0133, 0.0124, 0.0136, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 16:48:40,533 INFO [train.py:904] (7/8) Epoch 7, batch 5400, loss[loss=0.2256, simple_loss=0.3081, pruned_loss=0.07153, over 17045.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2884, pruned_loss=0.05968, over 3218970.51 frames. ], batch size: 50, lr: 9.73e-03, grad_scale: 8.0 2023-04-28 16:49:05,966 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66319.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:49:40,435 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4610, 3.4685, 3.4197, 2.7787, 3.3408, 2.0435, 3.1734, 2.9019], device='cuda:7'), covar=tensor([0.0099, 0.0077, 0.0102, 0.0228, 0.0067, 0.1632, 0.0088, 0.0160], device='cuda:7'), in_proj_covar=tensor([0.0106, 0.0093, 0.0141, 0.0141, 0.0109, 0.0154, 0.0125, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:49:56,006 INFO [train.py:904] (7/8) Epoch 7, batch 5450, loss[loss=0.2426, simple_loss=0.3187, pruned_loss=0.08325, over 17105.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.2922, pruned_loss=0.06206, over 3215733.80 frames. ], batch size: 48, lr: 9.73e-03, grad_scale: 8.0 2023-04-28 16:50:49,179 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9797, 2.3190, 2.3942, 3.0461, 2.2532, 3.2872, 1.6989, 2.7262], device='cuda:7'), covar=tensor([0.1066, 0.0474, 0.0861, 0.0143, 0.0175, 0.0375, 0.1222, 0.0616], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0148, 0.0170, 0.0104, 0.0197, 0.0195, 0.0167, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 16:50:50,986 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.058e+02 3.043e+02 3.833e+02 5.336e+02 9.922e+02, threshold=7.666e+02, percent-clipped=19.0 2023-04-28 16:51:07,542 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4303, 3.3691, 2.5491, 2.2196, 2.6015, 2.1143, 3.3258, 3.4902], device='cuda:7'), covar=tensor([0.2427, 0.0780, 0.1571, 0.1693, 0.1747, 0.1590, 0.0574, 0.0640], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0250, 0.0269, 0.0255, 0.0276, 0.0205, 0.0248, 0.0267], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:51:12,550 INFO [train.py:904] (7/8) Epoch 7, batch 5500, loss[loss=0.2434, simple_loss=0.3265, pruned_loss=0.0801, over 16498.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3005, pruned_loss=0.06787, over 3195822.76 frames. ], batch size: 75, lr: 9.73e-03, grad_scale: 8.0 2023-04-28 16:52:13,818 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8692, 4.6201, 4.8271, 5.0405, 5.2185, 4.5668, 5.1930, 5.1397], device='cuda:7'), covar=tensor([0.1186, 0.0818, 0.1301, 0.0521, 0.0449, 0.0772, 0.0413, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0445, 0.0543, 0.0684, 0.0550, 0.0415, 0.0419, 0.0422, 0.0471], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:52:29,862 INFO [train.py:904] (7/8) Epoch 7, batch 5550, loss[loss=0.2223, simple_loss=0.3104, pruned_loss=0.06714, over 16717.00 frames. ], tot_loss[loss=0.2285, simple_loss=0.3087, pruned_loss=0.07415, over 3175690.65 frames. ], batch size: 89, lr: 9.72e-03, grad_scale: 8.0 2023-04-28 16:53:27,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.385e+02 3.933e+02 4.760e+02 6.147e+02 1.111e+03, threshold=9.520e+02, percent-clipped=8.0 2023-04-28 16:53:33,401 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8665, 2.6584, 2.5703, 1.8390, 2.5898, 2.6402, 2.5061, 1.8487], device='cuda:7'), covar=tensor([0.0303, 0.0036, 0.0038, 0.0247, 0.0055, 0.0070, 0.0053, 0.0265], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0057, 0.0059, 0.0116, 0.0065, 0.0075, 0.0067, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 16:53:36,448 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66494.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:53:47,081 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66500.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:53:49,155 INFO [train.py:904] (7/8) Epoch 7, batch 5600, loss[loss=0.215, simple_loss=0.2884, pruned_loss=0.07085, over 17009.00 frames. ], tot_loss[loss=0.2389, simple_loss=0.3159, pruned_loss=0.08098, over 3109982.44 frames. ], batch size: 55, lr: 9.72e-03, grad_scale: 8.0 2023-04-28 16:54:01,776 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2949, 4.0878, 4.2988, 4.5070, 4.6163, 4.1086, 4.5840, 4.5590], device='cuda:7'), covar=tensor([0.1191, 0.0868, 0.1276, 0.0505, 0.0483, 0.1067, 0.0480, 0.0491], device='cuda:7'), in_proj_covar=tensor([0.0441, 0.0538, 0.0676, 0.0543, 0.0414, 0.0416, 0.0419, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:54:11,817 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66516.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 16:54:21,748 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66521.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 16:54:54,215 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66542.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 16:55:08,335 INFO [train.py:904] (7/8) Epoch 7, batch 5650, loss[loss=0.2353, simple_loss=0.313, pruned_loss=0.07873, over 16382.00 frames. ], tot_loss[loss=0.2466, simple_loss=0.3215, pruned_loss=0.08582, over 3093276.87 frames. ], batch size: 146, lr: 9.71e-03, grad_scale: 8.0 2023-04-28 16:55:53,783 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66582.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 16:56:03,093 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.602e+02 4.256e+02 5.191e+02 6.863e+02 2.150e+03, threshold=1.038e+03, percent-clipped=9.0 2023-04-28 16:56:05,275 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-04-28 16:56:11,998 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.08 vs. limit=2.0 2023-04-28 16:56:13,702 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3508, 2.8891, 2.5749, 2.3184, 2.2270, 2.1461, 2.7348, 2.8952], device='cuda:7'), covar=tensor([0.1789, 0.0629, 0.1108, 0.1435, 0.1613, 0.1432, 0.0406, 0.0824], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0252, 0.0274, 0.0258, 0.0280, 0.0206, 0.0249, 0.0269], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:56:23,904 INFO [train.py:904] (7/8) Epoch 7, batch 5700, loss[loss=0.2201, simple_loss=0.3076, pruned_loss=0.06627, over 16619.00 frames. ], tot_loss[loss=0.2495, simple_loss=0.3233, pruned_loss=0.08784, over 3076250.82 frames. ], batch size: 57, lr: 9.71e-03, grad_scale: 8.0 2023-04-28 16:56:27,054 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-28 16:56:42,597 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66614.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 16:57:12,678 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8197, 1.9960, 2.2824, 3.1688, 2.0697, 2.3876, 2.2528, 2.0370], device='cuda:7'), covar=tensor([0.0723, 0.2363, 0.1288, 0.0434, 0.2919, 0.1366, 0.1915, 0.2519], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0351, 0.0293, 0.0319, 0.0389, 0.0381, 0.0313, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:57:41,039 INFO [train.py:904] (7/8) Epoch 7, batch 5750, loss[loss=0.2996, simple_loss=0.3427, pruned_loss=0.1282, over 11123.00 frames. ], tot_loss[loss=0.2533, simple_loss=0.3266, pruned_loss=0.09002, over 3047421.59 frames. ], batch size: 247, lr: 9.71e-03, grad_scale: 8.0 2023-04-28 16:58:30,398 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-28 16:58:39,457 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.661e+02 4.395e+02 5.448e+02 7.543e+02, threshold=8.790e+02, percent-clipped=0.0 2023-04-28 16:59:02,140 INFO [train.py:904] (7/8) Epoch 7, batch 5800, loss[loss=0.3126, simple_loss=0.3489, pruned_loss=0.1382, over 12055.00 frames. ], tot_loss[loss=0.2517, simple_loss=0.3263, pruned_loss=0.08855, over 3046748.77 frames. ], batch size: 250, lr: 9.70e-03, grad_scale: 16.0 2023-04-28 16:59:10,945 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-28 16:59:14,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7466, 4.7119, 4.5371, 3.7598, 4.6080, 1.5653, 4.3461, 4.4103], device='cuda:7'), covar=tensor([0.0072, 0.0060, 0.0121, 0.0360, 0.0063, 0.2151, 0.0100, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0103, 0.0091, 0.0137, 0.0135, 0.0105, 0.0152, 0.0120, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 16:59:48,436 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66731.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:00:19,934 INFO [train.py:904] (7/8) Epoch 7, batch 5850, loss[loss=0.2562, simple_loss=0.3177, pruned_loss=0.09731, over 11621.00 frames. ], tot_loss[loss=0.2474, simple_loss=0.3234, pruned_loss=0.08569, over 3065382.33 frames. ], batch size: 248, lr: 9.70e-03, grad_scale: 8.0 2023-04-28 17:00:21,546 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-28 17:01:20,006 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.089e+02 3.451e+02 4.238e+02 5.428e+02 1.064e+03, threshold=8.477e+02, percent-clipped=4.0 2023-04-28 17:01:25,712 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66792.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:01:39,046 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66800.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:01:41,109 INFO [train.py:904] (7/8) Epoch 7, batch 5900, loss[loss=0.2114, simple_loss=0.2966, pruned_loss=0.06313, over 17117.00 frames. ], tot_loss[loss=0.2463, simple_loss=0.3227, pruned_loss=0.08493, over 3072113.85 frames. ], batch size: 48, lr: 9.70e-03, grad_scale: 8.0 2023-04-28 17:02:08,259 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66816.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:02:51,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6375, 2.6479, 1.7637, 2.7243, 2.1410, 2.7530, 1.9864, 2.3599], device='cuda:7'), covar=tensor([0.0168, 0.0355, 0.1158, 0.0127, 0.0616, 0.0394, 0.1163, 0.0536], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0157, 0.0179, 0.0095, 0.0163, 0.0192, 0.0188, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 17:02:56,755 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66848.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:03:02,874 INFO [train.py:904] (7/8) Epoch 7, batch 5950, loss[loss=0.2433, simple_loss=0.3251, pruned_loss=0.0808, over 16242.00 frames. ], tot_loss[loss=0.2449, simple_loss=0.3232, pruned_loss=0.08331, over 3081786.53 frames. ], batch size: 165, lr: 9.69e-03, grad_scale: 8.0 2023-04-28 17:03:23,047 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66864.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:03:23,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3602, 3.6856, 3.8113, 1.7332, 3.9986, 4.0417, 3.0002, 2.9648], device='cuda:7'), covar=tensor([0.0794, 0.0143, 0.0129, 0.1195, 0.0051, 0.0082, 0.0320, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0096, 0.0081, 0.0139, 0.0071, 0.0087, 0.0118, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 17:03:41,790 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66877.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 17:03:50,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5320, 3.7159, 2.9528, 2.2690, 2.7415, 2.2952, 3.9145, 3.7460], device='cuda:7'), covar=tensor([0.2631, 0.0749, 0.1481, 0.1819, 0.2141, 0.1594, 0.0511, 0.0804], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0249, 0.0273, 0.0259, 0.0279, 0.0207, 0.0251, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:04:00,335 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 3.507e+02 4.238e+02 5.487e+02 1.167e+03, threshold=8.477e+02, percent-clipped=3.0 2023-04-28 17:04:22,274 INFO [train.py:904] (7/8) Epoch 7, batch 6000, loss[loss=0.2028, simple_loss=0.2909, pruned_loss=0.05734, over 16874.00 frames. ], tot_loss[loss=0.2441, simple_loss=0.3226, pruned_loss=0.08275, over 3090065.12 frames. ], batch size: 90, lr: 9.69e-03, grad_scale: 8.0 2023-04-28 17:04:22,274 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 17:04:32,880 INFO [train.py:938] (7/8) Epoch 7, validation: loss=0.1758, simple_loss=0.2891, pruned_loss=0.03127, over 944034.00 frames. 2023-04-28 17:04:32,881 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 17:04:51,831 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66914.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:05:53,517 INFO [train.py:904] (7/8) Epoch 7, batch 6050, loss[loss=0.2169, simple_loss=0.3156, pruned_loss=0.05906, over 16702.00 frames. ], tot_loss[loss=0.2424, simple_loss=0.3212, pruned_loss=0.08182, over 3094231.62 frames. ], batch size: 76, lr: 9.69e-03, grad_scale: 8.0 2023-04-28 17:06:02,548 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66957.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:06:03,853 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3156, 4.6216, 4.3744, 4.4461, 4.1654, 4.0820, 4.1587, 4.6349], device='cuda:7'), covar=tensor([0.0811, 0.0711, 0.0954, 0.0511, 0.0605, 0.1206, 0.0766, 0.0797], device='cuda:7'), in_proj_covar=tensor([0.0432, 0.0550, 0.0470, 0.0359, 0.0341, 0.0369, 0.0458, 0.0405], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:06:09,648 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=66962.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:06:50,896 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.435e+02 4.153e+02 5.087e+02 8.285e+02, threshold=8.306e+02, percent-clipped=0.0 2023-04-28 17:07:12,274 INFO [train.py:904] (7/8) Epoch 7, batch 6100, loss[loss=0.2197, simple_loss=0.3047, pruned_loss=0.06728, over 17130.00 frames. ], tot_loss[loss=0.2402, simple_loss=0.3197, pruned_loss=0.08038, over 3113912.49 frames. ], batch size: 48, lr: 9.68e-03, grad_scale: 8.0 2023-04-28 17:07:39,917 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67018.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:08:32,445 INFO [train.py:904] (7/8) Epoch 7, batch 6150, loss[loss=0.2073, simple_loss=0.285, pruned_loss=0.06485, over 17026.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3174, pruned_loss=0.07962, over 3107309.46 frames. ], batch size: 55, lr: 9.68e-03, grad_scale: 8.0 2023-04-28 17:09:23,265 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67083.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:09:29,265 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67087.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 17:09:31,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.004e+02 3.456e+02 4.175e+02 5.279e+02 9.590e+02, threshold=8.351e+02, percent-clipped=3.0 2023-04-28 17:09:46,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-28 17:09:51,774 INFO [train.py:904] (7/8) Epoch 7, batch 6200, loss[loss=0.2808, simple_loss=0.3347, pruned_loss=0.1135, over 11525.00 frames. ], tot_loss[loss=0.237, simple_loss=0.3157, pruned_loss=0.07912, over 3100798.11 frames. ], batch size: 247, lr: 9.67e-03, grad_scale: 4.0 2023-04-28 17:11:02,513 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67144.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:11:14,132 INFO [train.py:904] (7/8) Epoch 7, batch 6250, loss[loss=0.2377, simple_loss=0.3291, pruned_loss=0.07312, over 16696.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.3151, pruned_loss=0.07815, over 3128624.09 frames. ], batch size: 89, lr: 9.67e-03, grad_scale: 4.0 2023-04-28 17:11:52,534 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67177.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:12:11,787 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.341e+02 3.226e+02 4.153e+02 4.983e+02 8.969e+02, threshold=8.306e+02, percent-clipped=3.0 2023-04-28 17:12:29,284 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-04-28 17:12:30,907 INFO [train.py:904] (7/8) Epoch 7, batch 6300, loss[loss=0.2323, simple_loss=0.3086, pruned_loss=0.078, over 15352.00 frames. ], tot_loss[loss=0.2359, simple_loss=0.3151, pruned_loss=0.07833, over 3114579.57 frames. ], batch size: 190, lr: 9.67e-03, grad_scale: 4.0 2023-04-28 17:13:08,747 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67225.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:13:49,876 INFO [train.py:904] (7/8) Epoch 7, batch 6350, loss[loss=0.2194, simple_loss=0.2997, pruned_loss=0.06955, over 16582.00 frames. ], tot_loss[loss=0.2379, simple_loss=0.3161, pruned_loss=0.07988, over 3109143.22 frames. ], batch size: 62, lr: 9.66e-03, grad_scale: 4.0 2023-04-28 17:13:52,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5386, 2.1616, 1.7411, 1.9245, 2.4486, 2.1714, 2.5998, 2.6510], device='cuda:7'), covar=tensor([0.0075, 0.0224, 0.0309, 0.0281, 0.0133, 0.0218, 0.0113, 0.0131], device='cuda:7'), in_proj_covar=tensor([0.0101, 0.0174, 0.0173, 0.0170, 0.0167, 0.0173, 0.0166, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:14:40,164 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-28 17:14:47,875 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.015e+02 3.646e+02 4.526e+02 5.415e+02 1.244e+03, threshold=9.052e+02, percent-clipped=4.0 2023-04-28 17:15:05,417 INFO [train.py:904] (7/8) Epoch 7, batch 6400, loss[loss=0.247, simple_loss=0.3204, pruned_loss=0.08687, over 15543.00 frames. ], tot_loss[loss=0.2389, simple_loss=0.3162, pruned_loss=0.0808, over 3094859.69 frames. ], batch size: 191, lr: 9.66e-03, grad_scale: 8.0 2023-04-28 17:15:23,074 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67313.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 17:16:20,301 INFO [train.py:904] (7/8) Epoch 7, batch 6450, loss[loss=0.2258, simple_loss=0.3088, pruned_loss=0.07139, over 17202.00 frames. ], tot_loss[loss=0.2384, simple_loss=0.3161, pruned_loss=0.08036, over 3085826.33 frames. ], batch size: 44, lr: 9.66e-03, grad_scale: 4.0 2023-04-28 17:16:21,748 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-28 17:16:50,975 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6855, 3.6688, 3.7454, 3.7105, 3.8060, 4.1967, 3.9473, 3.6439], device='cuda:7'), covar=tensor([0.1816, 0.1638, 0.1615, 0.1828, 0.2411, 0.1481, 0.1256, 0.2207], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0422, 0.0431, 0.0363, 0.0488, 0.0465, 0.0344, 0.0497], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:17:16,590 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67387.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:17:22,057 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.155e+02 3.316e+02 4.009e+02 5.317e+02 1.482e+03, threshold=8.018e+02, percent-clipped=5.0 2023-04-28 17:17:37,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4314, 4.4234, 4.2322, 3.6047, 4.3006, 1.5839, 4.1044, 4.1108], device='cuda:7'), covar=tensor([0.0071, 0.0057, 0.0131, 0.0304, 0.0065, 0.2108, 0.0090, 0.0147], device='cuda:7'), in_proj_covar=tensor([0.0105, 0.0091, 0.0139, 0.0135, 0.0105, 0.0154, 0.0123, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:17:38,447 INFO [train.py:904] (7/8) Epoch 7, batch 6500, loss[loss=0.2203, simple_loss=0.303, pruned_loss=0.06881, over 16731.00 frames. ], tot_loss[loss=0.236, simple_loss=0.3136, pruned_loss=0.07926, over 3091851.82 frames. ], batch size: 124, lr: 9.65e-03, grad_scale: 4.0 2023-04-28 17:18:29,459 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67435.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:18:36,619 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67439.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:18:57,745 INFO [train.py:904] (7/8) Epoch 7, batch 6550, loss[loss=0.2232, simple_loss=0.3196, pruned_loss=0.06335, over 16434.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3165, pruned_loss=0.08, over 3100833.34 frames. ], batch size: 146, lr: 9.65e-03, grad_scale: 4.0 2023-04-28 17:19:56,002 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.411e+02 3.629e+02 4.533e+02 5.660e+02 1.346e+03, threshold=9.066e+02, percent-clipped=11.0 2023-04-28 17:20:13,107 INFO [train.py:904] (7/8) Epoch 7, batch 6600, loss[loss=0.2513, simple_loss=0.3296, pruned_loss=0.08646, over 16544.00 frames. ], tot_loss[loss=0.2403, simple_loss=0.3188, pruned_loss=0.08088, over 3102825.96 frames. ], batch size: 68, lr: 9.65e-03, grad_scale: 4.0 2023-04-28 17:20:30,829 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67513.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:21:29,933 INFO [train.py:904] (7/8) Epoch 7, batch 6650, loss[loss=0.2232, simple_loss=0.3023, pruned_loss=0.07206, over 15276.00 frames. ], tot_loss[loss=0.2412, simple_loss=0.3192, pruned_loss=0.08163, over 3103207.66 frames. ], batch size: 190, lr: 9.64e-03, grad_scale: 4.0 2023-04-28 17:21:48,353 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4923, 4.8155, 4.5288, 4.5497, 4.2354, 4.1731, 4.2814, 4.8402], device='cuda:7'), covar=tensor([0.0742, 0.0651, 0.0929, 0.0527, 0.0680, 0.1013, 0.0805, 0.0711], device='cuda:7'), in_proj_covar=tensor([0.0444, 0.0558, 0.0480, 0.0368, 0.0355, 0.0377, 0.0465, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:22:02,894 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67574.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 17:22:27,319 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.256e+02 3.378e+02 3.968e+02 4.972e+02 1.148e+03, threshold=7.935e+02, percent-clipped=1.0 2023-04-28 17:22:43,257 INFO [train.py:904] (7/8) Epoch 7, batch 6700, loss[loss=0.2656, simple_loss=0.3232, pruned_loss=0.1041, over 11575.00 frames. ], tot_loss[loss=0.2385, simple_loss=0.3166, pruned_loss=0.0802, over 3130164.21 frames. ], batch size: 247, lr: 9.64e-03, grad_scale: 4.0 2023-04-28 17:23:00,283 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67613.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:23:55,812 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0024, 3.6746, 3.7216, 2.3019, 3.3758, 3.5867, 3.4093, 2.0616], device='cuda:7'), covar=tensor([0.0377, 0.0023, 0.0024, 0.0271, 0.0055, 0.0076, 0.0044, 0.0315], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0056, 0.0060, 0.0117, 0.0065, 0.0077, 0.0067, 0.0112], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:23:56,597 INFO [train.py:904] (7/8) Epoch 7, batch 6750, loss[loss=0.2163, simple_loss=0.2968, pruned_loss=0.06791, over 16893.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.3149, pruned_loss=0.07962, over 3133701.06 frames. ], batch size: 116, lr: 9.64e-03, grad_scale: 4.0 2023-04-28 17:24:11,057 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67661.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:24:31,722 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3912, 4.4867, 4.6426, 4.5402, 4.6234, 5.1023, 4.6591, 4.3728], device='cuda:7'), covar=tensor([0.1209, 0.1633, 0.1438, 0.1716, 0.2165, 0.0888, 0.1195, 0.2305], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0430, 0.0443, 0.0368, 0.0499, 0.0473, 0.0352, 0.0508], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:24:37,895 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67680.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:24:54,248 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.424e+02 3.570e+02 4.418e+02 5.557e+02 1.177e+03, threshold=8.835e+02, percent-clipped=2.0 2023-04-28 17:24:56,114 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67692.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:25:05,897 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 17:25:10,441 INFO [train.py:904] (7/8) Epoch 7, batch 6800, loss[loss=0.2266, simple_loss=0.3107, pruned_loss=0.07123, over 17101.00 frames. ], tot_loss[loss=0.2377, simple_loss=0.3152, pruned_loss=0.0801, over 3123980.96 frames. ], batch size: 49, lr: 9.63e-03, grad_scale: 8.0 2023-04-28 17:25:13,843 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7975, 5.1691, 4.5366, 4.9592, 4.6001, 4.3822, 4.6559, 5.1937], device='cuda:7'), covar=tensor([0.1672, 0.1184, 0.2339, 0.0846, 0.1225, 0.1409, 0.1620, 0.1389], device='cuda:7'), in_proj_covar=tensor([0.0445, 0.0554, 0.0479, 0.0367, 0.0353, 0.0376, 0.0463, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:26:06,854 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67739.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:26:10,796 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67741.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:26:23,544 INFO [train.py:904] (7/8) Epoch 7, batch 6850, loss[loss=0.2183, simple_loss=0.3085, pruned_loss=0.06407, over 16847.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3166, pruned_loss=0.07999, over 3136342.93 frames. ], batch size: 42, lr: 9.63e-03, grad_scale: 8.0 2023-04-28 17:26:26,398 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67753.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:27:14,418 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=67787.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:27:20,978 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.331e+02 3.323e+02 3.993e+02 4.819e+02 7.017e+02, threshold=7.986e+02, percent-clipped=0.0 2023-04-28 17:27:35,616 INFO [train.py:904] (7/8) Epoch 7, batch 6900, loss[loss=0.2585, simple_loss=0.3346, pruned_loss=0.09117, over 16538.00 frames. ], tot_loss[loss=0.24, simple_loss=0.3197, pruned_loss=0.08012, over 3138301.88 frames. ], batch size: 68, lr: 9.63e-03, grad_scale: 2.0 2023-04-28 17:27:41,107 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4199, 4.4341, 4.3141, 3.6135, 4.3284, 1.6249, 4.0465, 4.1665], device='cuda:7'), covar=tensor([0.0084, 0.0062, 0.0113, 0.0315, 0.0067, 0.1991, 0.0118, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0090, 0.0138, 0.0135, 0.0105, 0.0154, 0.0123, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:28:21,048 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-28 17:28:46,556 INFO [train.py:904] (7/8) Epoch 7, batch 6950, loss[loss=0.3022, simple_loss=0.3524, pruned_loss=0.126, over 11342.00 frames. ], tot_loss[loss=0.2435, simple_loss=0.3221, pruned_loss=0.08243, over 3126731.32 frames. ], batch size: 247, lr: 9.62e-03, grad_scale: 2.0 2023-04-28 17:29:13,189 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67869.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:29:38,398 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1396, 5.7303, 5.8860, 5.6797, 5.6003, 6.1975, 5.7389, 5.4884], device='cuda:7'), covar=tensor([0.0604, 0.1263, 0.1188, 0.1424, 0.2132, 0.0765, 0.0911, 0.1789], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0433, 0.0447, 0.0373, 0.0500, 0.0477, 0.0353, 0.0508], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:29:46,383 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.351e+02 3.618e+02 4.486e+02 5.724e+02 1.200e+03, threshold=8.972e+02, percent-clipped=9.0 2023-04-28 17:29:59,773 INFO [train.py:904] (7/8) Epoch 7, batch 7000, loss[loss=0.212, simple_loss=0.3004, pruned_loss=0.06175, over 16814.00 frames. ], tot_loss[loss=0.2427, simple_loss=0.3217, pruned_loss=0.08188, over 3105154.26 frames. ], batch size: 39, lr: 9.62e-03, grad_scale: 2.0 2023-04-28 17:30:33,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1320, 1.8133, 1.9570, 3.6886, 1.7546, 2.4064, 2.0389, 1.9532], device='cuda:7'), covar=tensor([0.0775, 0.2787, 0.1772, 0.0350, 0.3499, 0.1679, 0.2560, 0.2943], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0354, 0.0294, 0.0319, 0.0392, 0.0381, 0.0316, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:31:13,273 INFO [train.py:904] (7/8) Epoch 7, batch 7050, loss[loss=0.2545, simple_loss=0.3289, pruned_loss=0.09006, over 16698.00 frames. ], tot_loss[loss=0.242, simple_loss=0.3221, pruned_loss=0.08092, over 3122080.57 frames. ], batch size: 134, lr: 9.61e-03, grad_scale: 2.0 2023-04-28 17:31:24,166 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0345, 3.2306, 3.5490, 3.5172, 3.4960, 3.2483, 3.3335, 3.4236], device='cuda:7'), covar=tensor([0.0355, 0.0565, 0.0363, 0.0462, 0.0516, 0.0426, 0.0789, 0.0416], device='cuda:7'), in_proj_covar=tensor([0.0269, 0.0270, 0.0271, 0.0267, 0.0320, 0.0291, 0.0386, 0.0234], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-28 17:31:35,852 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.4862, 2.6263, 2.2724, 3.8450, 2.7025, 3.8761, 1.2978, 2.6359], device='cuda:7'), covar=tensor([0.1573, 0.0723, 0.1295, 0.0134, 0.0297, 0.0374, 0.1766, 0.0875], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0148, 0.0170, 0.0104, 0.0198, 0.0196, 0.0168, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 17:31:57,450 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67981.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:32:14,209 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 3.870e+02 4.768e+02 5.699e+02 1.240e+03, threshold=9.537e+02, percent-clipped=2.0 2023-04-28 17:32:29,900 INFO [train.py:904] (7/8) Epoch 7, batch 7100, loss[loss=0.2145, simple_loss=0.2943, pruned_loss=0.06732, over 16692.00 frames. ], tot_loss[loss=0.2404, simple_loss=0.32, pruned_loss=0.08041, over 3129090.01 frames. ], batch size: 62, lr: 9.61e-03, grad_scale: 2.0 2023-04-28 17:33:21,414 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68036.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:33:30,248 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68042.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:33:37,136 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5494, 3.4944, 2.8844, 2.3368, 2.6746, 2.1924, 3.8847, 3.5811], device='cuda:7'), covar=tensor([0.2398, 0.0855, 0.1381, 0.1721, 0.1990, 0.1586, 0.0449, 0.0818], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0251, 0.0273, 0.0259, 0.0280, 0.0209, 0.0253, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:33:39,224 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68048.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:33:44,543 INFO [train.py:904] (7/8) Epoch 7, batch 7150, loss[loss=0.2514, simple_loss=0.3267, pruned_loss=0.08804, over 16870.00 frames. ], tot_loss[loss=0.2409, simple_loss=0.3191, pruned_loss=0.08137, over 3108321.24 frames. ], batch size: 116, lr: 9.61e-03, grad_scale: 2.0 2023-04-28 17:33:54,792 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0426, 3.0760, 2.8627, 5.0430, 4.2478, 4.5126, 1.9336, 3.1387], device='cuda:7'), covar=tensor([0.1171, 0.0613, 0.1011, 0.0116, 0.0385, 0.0310, 0.1224, 0.0797], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0149, 0.0171, 0.0106, 0.0200, 0.0198, 0.0169, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 17:34:20,219 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0487, 3.4991, 3.5802, 2.2323, 3.1262, 3.4795, 3.3075, 1.9747], device='cuda:7'), covar=tensor([0.0377, 0.0022, 0.0025, 0.0273, 0.0057, 0.0071, 0.0051, 0.0339], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0055, 0.0059, 0.0116, 0.0064, 0.0076, 0.0066, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:34:37,139 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68087.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:34:44,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.291e+02 3.518e+02 4.375e+02 5.869e+02 1.278e+03, threshold=8.749e+02, percent-clipped=4.0 2023-04-28 17:34:58,213 INFO [train.py:904] (7/8) Epoch 7, batch 7200, loss[loss=0.2104, simple_loss=0.299, pruned_loss=0.06087, over 16725.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.3168, pruned_loss=0.07942, over 3106786.53 frames. ], batch size: 89, lr: 9.60e-03, grad_scale: 4.0 2023-04-28 17:35:35,270 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3446, 4.6274, 4.3675, 4.4220, 4.0940, 4.0977, 4.1701, 4.6515], device='cuda:7'), covar=tensor([0.0798, 0.0699, 0.0993, 0.0566, 0.0687, 0.1111, 0.0790, 0.0771], device='cuda:7'), in_proj_covar=tensor([0.0443, 0.0555, 0.0479, 0.0367, 0.0355, 0.0377, 0.0461, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:36:11,124 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68148.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 17:36:16,031 INFO [train.py:904] (7/8) Epoch 7, batch 7250, loss[loss=0.2015, simple_loss=0.2789, pruned_loss=0.0621, over 16714.00 frames. ], tot_loss[loss=0.2352, simple_loss=0.3141, pruned_loss=0.0782, over 3084381.64 frames. ], batch size: 134, lr: 9.60e-03, grad_scale: 4.0 2023-04-28 17:36:41,762 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68169.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:36:59,899 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9425, 1.6056, 2.3024, 2.8144, 2.5918, 3.1467, 1.8856, 3.0048], device='cuda:7'), covar=tensor([0.0092, 0.0288, 0.0172, 0.0142, 0.0145, 0.0083, 0.0278, 0.0069], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0153, 0.0134, 0.0135, 0.0144, 0.0101, 0.0152, 0.0092], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 17:37:16,281 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 3.214e+02 3.991e+02 5.016e+02 1.051e+03, threshold=7.982e+02, percent-clipped=2.0 2023-04-28 17:37:29,952 INFO [train.py:904] (7/8) Epoch 7, batch 7300, loss[loss=0.235, simple_loss=0.3154, pruned_loss=0.07735, over 16155.00 frames. ], tot_loss[loss=0.2337, simple_loss=0.313, pruned_loss=0.07726, over 3110293.26 frames. ], batch size: 165, lr: 9.60e-03, grad_scale: 4.0 2023-04-28 17:37:53,279 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68217.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:38:44,626 INFO [train.py:904] (7/8) Epoch 7, batch 7350, loss[loss=0.2287, simple_loss=0.3102, pruned_loss=0.07356, over 16859.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.3135, pruned_loss=0.07777, over 3093643.45 frames. ], batch size: 102, lr: 9.59e-03, grad_scale: 4.0 2023-04-28 17:39:48,572 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.908e+02 3.085e+02 3.772e+02 4.630e+02 1.239e+03, threshold=7.544e+02, percent-clipped=2.0 2023-04-28 17:40:02,139 INFO [train.py:904] (7/8) Epoch 7, batch 7400, loss[loss=0.279, simple_loss=0.3336, pruned_loss=0.1122, over 11233.00 frames. ], tot_loss[loss=0.2354, simple_loss=0.3143, pruned_loss=0.07831, over 3100489.65 frames. ], batch size: 247, lr: 9.59e-03, grad_scale: 4.0 2023-04-28 17:40:53,709 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68336.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:40:55,676 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68337.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:41:14,111 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68348.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:41:19,096 INFO [train.py:904] (7/8) Epoch 7, batch 7450, loss[loss=0.2901, simple_loss=0.3383, pruned_loss=0.1209, over 11509.00 frames. ], tot_loss[loss=0.2388, simple_loss=0.3163, pruned_loss=0.08059, over 3063263.34 frames. ], batch size: 247, lr: 9.59e-03, grad_scale: 4.0 2023-04-28 17:41:56,252 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5957, 2.6795, 1.7995, 2.6896, 2.1173, 2.8120, 2.0134, 2.3730], device='cuda:7'), covar=tensor([0.0187, 0.0331, 0.1147, 0.0142, 0.0617, 0.0456, 0.1098, 0.0515], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0154, 0.0179, 0.0095, 0.0162, 0.0192, 0.0188, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 17:42:11,321 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68384.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:42:26,328 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.512e+02 3.665e+02 4.367e+02 5.646e+02 8.920e+02, threshold=8.734e+02, percent-clipped=5.0 2023-04-28 17:42:30,794 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68396.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:42:35,702 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68399.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:42:39,847 INFO [train.py:904] (7/8) Epoch 7, batch 7500, loss[loss=0.2671, simple_loss=0.3287, pruned_loss=0.1027, over 11422.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.3162, pruned_loss=0.07972, over 3054731.01 frames. ], batch size: 248, lr: 9.58e-03, grad_scale: 4.0 2023-04-28 17:43:06,512 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-28 17:43:14,901 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-04-28 17:43:44,352 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68443.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:43:57,935 INFO [train.py:904] (7/8) Epoch 7, batch 7550, loss[loss=0.2199, simple_loss=0.2951, pruned_loss=0.07232, over 17000.00 frames. ], tot_loss[loss=0.2384, simple_loss=0.3159, pruned_loss=0.08048, over 3040930.63 frames. ], batch size: 53, lr: 9.58e-03, grad_scale: 4.0 2023-04-28 17:44:07,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3245, 3.2398, 3.3513, 3.4552, 3.5092, 3.2336, 3.4645, 3.5543], device='cuda:7'), covar=tensor([0.0886, 0.0679, 0.0983, 0.0499, 0.0590, 0.1675, 0.0778, 0.0545], device='cuda:7'), in_proj_covar=tensor([0.0434, 0.0534, 0.0666, 0.0537, 0.0411, 0.0412, 0.0427, 0.0466], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:44:10,879 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68460.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:44:50,927 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0182, 3.2085, 3.5249, 3.5002, 3.4617, 3.2469, 3.3209, 3.3570], device='cuda:7'), covar=tensor([0.0408, 0.0621, 0.0372, 0.0401, 0.0605, 0.0495, 0.0811, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0279, 0.0279, 0.0270, 0.0326, 0.0300, 0.0396, 0.0246], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 17:44:53,023 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-28 17:44:59,735 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.309e+02 3.638e+02 4.521e+02 5.512e+02 1.190e+03, threshold=9.043e+02, percent-clipped=2.0 2023-04-28 17:45:12,771 INFO [train.py:904] (7/8) Epoch 7, batch 7600, loss[loss=0.219, simple_loss=0.3019, pruned_loss=0.06804, over 16880.00 frames. ], tot_loss[loss=0.2377, simple_loss=0.3147, pruned_loss=0.08038, over 3055326.25 frames. ], batch size: 96, lr: 9.58e-03, grad_scale: 8.0 2023-04-28 17:45:23,379 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5131, 2.5743, 2.0309, 2.3548, 2.8992, 2.7742, 3.4191, 3.2662], device='cuda:7'), covar=tensor([0.0037, 0.0223, 0.0334, 0.0261, 0.0136, 0.0214, 0.0110, 0.0139], device='cuda:7'), in_proj_covar=tensor([0.0101, 0.0175, 0.0174, 0.0173, 0.0170, 0.0177, 0.0165, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:45:53,040 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8852, 4.1580, 3.9432, 3.9790, 3.6416, 3.7680, 3.8473, 4.0986], device='cuda:7'), covar=tensor([0.0798, 0.0793, 0.0838, 0.0547, 0.0612, 0.1325, 0.0707, 0.0969], device='cuda:7'), in_proj_covar=tensor([0.0444, 0.0561, 0.0478, 0.0367, 0.0350, 0.0381, 0.0461, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:46:28,276 INFO [train.py:904] (7/8) Epoch 7, batch 7650, loss[loss=0.3063, simple_loss=0.3571, pruned_loss=0.1278, over 11156.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3155, pruned_loss=0.08056, over 3075638.58 frames. ], batch size: 246, lr: 9.57e-03, grad_scale: 8.0 2023-04-28 17:47:08,471 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68578.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:47:29,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.494e+02 3.559e+02 4.201e+02 5.687e+02 1.184e+03, threshold=8.403e+02, percent-clipped=4.0 2023-04-28 17:47:42,702 INFO [train.py:904] (7/8) Epoch 7, batch 7700, loss[loss=0.3403, simple_loss=0.3887, pruned_loss=0.1459, over 11468.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3158, pruned_loss=0.08124, over 3085226.90 frames. ], batch size: 248, lr: 9.57e-03, grad_scale: 8.0 2023-04-28 17:48:05,723 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3216, 1.4116, 1.9329, 2.3379, 2.3036, 2.5224, 1.5810, 2.5205], device='cuda:7'), covar=tensor([0.0112, 0.0286, 0.0158, 0.0171, 0.0138, 0.0101, 0.0289, 0.0068], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0154, 0.0135, 0.0136, 0.0145, 0.0102, 0.0153, 0.0092], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 17:48:35,376 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68637.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:48:38,402 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68639.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:48:57,122 INFO [train.py:904] (7/8) Epoch 7, batch 7750, loss[loss=0.2975, simple_loss=0.3534, pruned_loss=0.1208, over 11546.00 frames. ], tot_loss[loss=0.2385, simple_loss=0.3157, pruned_loss=0.08068, over 3090055.83 frames. ], batch size: 248, lr: 9.57e-03, grad_scale: 8.0 2023-04-28 17:49:43,635 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5018, 3.6410, 1.8095, 3.9412, 2.4198, 3.9349, 2.0476, 2.8236], device='cuda:7'), covar=tensor([0.0176, 0.0330, 0.1637, 0.0090, 0.0848, 0.0424, 0.1434, 0.0623], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0157, 0.0181, 0.0097, 0.0165, 0.0195, 0.0189, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 17:49:45,164 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:49:45,356 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:49:57,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.246e+02 3.570e+02 4.272e+02 5.150e+02 9.085e+02, threshold=8.545e+02, percent-clipped=2.0 2023-04-28 17:50:01,017 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1521, 5.1219, 4.9718, 4.7725, 4.4179, 4.9326, 4.9562, 4.6132], device='cuda:7'), covar=tensor([0.0694, 0.0664, 0.0302, 0.0293, 0.1159, 0.0515, 0.0332, 0.0827], device='cuda:7'), in_proj_covar=tensor([0.0205, 0.0236, 0.0237, 0.0210, 0.0267, 0.0239, 0.0167, 0.0273], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:50:06,914 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 17:50:11,064 INFO [train.py:904] (7/8) Epoch 7, batch 7800, loss[loss=0.2114, simple_loss=0.3005, pruned_loss=0.06115, over 16888.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3162, pruned_loss=0.08106, over 3098705.67 frames. ], batch size: 96, lr: 9.56e-03, grad_scale: 8.0 2023-04-28 17:51:13,409 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68743.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 17:51:17,559 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68746.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:51:25,173 INFO [train.py:904] (7/8) Epoch 7, batch 7850, loss[loss=0.2698, simple_loss=0.3282, pruned_loss=0.1057, over 11276.00 frames. ], tot_loss[loss=0.2413, simple_loss=0.3181, pruned_loss=0.08226, over 3062957.31 frames. ], batch size: 247, lr: 9.56e-03, grad_scale: 8.0 2023-04-28 17:51:30,857 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68755.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:52:24,336 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=68791.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 17:52:26,861 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.225e+02 3.517e+02 4.266e+02 5.466e+02 1.247e+03, threshold=8.532e+02, percent-clipped=3.0 2023-04-28 17:52:40,554 INFO [train.py:904] (7/8) Epoch 7, batch 7900, loss[loss=0.2629, simple_loss=0.3381, pruned_loss=0.09386, over 15439.00 frames. ], tot_loss[loss=0.2393, simple_loss=0.3168, pruned_loss=0.08096, over 3083154.83 frames. ], batch size: 190, lr: 9.56e-03, grad_scale: 8.0 2023-04-28 17:53:58,015 INFO [train.py:904] (7/8) Epoch 7, batch 7950, loss[loss=0.2826, simple_loss=0.3308, pruned_loss=0.1172, over 11671.00 frames. ], tot_loss[loss=0.2395, simple_loss=0.3168, pruned_loss=0.0811, over 3087387.25 frames. ], batch size: 248, lr: 9.55e-03, grad_scale: 2.0 2023-04-28 17:55:01,915 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-04-28 17:55:03,364 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.354e+02 3.365e+02 3.868e+02 4.958e+02 9.415e+02, threshold=7.735e+02, percent-clipped=1.0 2023-04-28 17:55:12,758 INFO [train.py:904] (7/8) Epoch 7, batch 8000, loss[loss=0.2145, simple_loss=0.2987, pruned_loss=0.06518, over 16694.00 frames. ], tot_loss[loss=0.2389, simple_loss=0.3165, pruned_loss=0.0807, over 3101061.78 frames. ], batch size: 62, lr: 9.55e-03, grad_scale: 4.0 2023-04-28 17:55:37,745 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 17:56:00,104 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68934.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:56:03,608 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4170, 3.4126, 3.4182, 2.6064, 3.3735, 1.9862, 3.0984, 2.7812], device='cuda:7'), covar=tensor([0.0156, 0.0129, 0.0166, 0.0361, 0.0095, 0.2289, 0.0140, 0.0264], device='cuda:7'), in_proj_covar=tensor([0.0104, 0.0092, 0.0141, 0.0135, 0.0106, 0.0158, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 17:56:25,023 INFO [train.py:904] (7/8) Epoch 7, batch 8050, loss[loss=0.2516, simple_loss=0.3138, pruned_loss=0.09473, over 16687.00 frames. ], tot_loss[loss=0.2378, simple_loss=0.3159, pruned_loss=0.07981, over 3119968.47 frames. ], batch size: 57, lr: 9.54e-03, grad_scale: 4.0 2023-04-28 17:56:38,679 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8384, 2.5405, 2.6179, 1.7224, 2.5005, 2.6184, 2.4805, 1.7966], device='cuda:7'), covar=tensor([0.0316, 0.0040, 0.0039, 0.0243, 0.0068, 0.0067, 0.0051, 0.0282], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0056, 0.0060, 0.0117, 0.0066, 0.0077, 0.0067, 0.0112], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:57:03,584 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9588, 3.0841, 1.6421, 3.2473, 2.2617, 3.2833, 1.8319, 2.5399], device='cuda:7'), covar=tensor([0.0221, 0.0357, 0.1489, 0.0113, 0.0743, 0.0516, 0.1462, 0.0626], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0156, 0.0181, 0.0097, 0.0164, 0.0194, 0.0190, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 17:57:30,025 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.364e+02 3.694e+02 4.392e+02 5.257e+02 9.021e+02, threshold=8.783e+02, percent-clipped=4.0 2023-04-28 17:57:41,498 INFO [train.py:904] (7/8) Epoch 7, batch 8100, loss[loss=0.227, simple_loss=0.3064, pruned_loss=0.07384, over 16698.00 frames. ], tot_loss[loss=0.2376, simple_loss=0.3159, pruned_loss=0.07967, over 3108571.55 frames. ], batch size: 134, lr: 9.54e-03, grad_scale: 4.0 2023-04-28 17:58:06,038 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9648, 3.1938, 3.4510, 2.0335, 3.1107, 3.3604, 3.1512, 1.8702], device='cuda:7'), covar=tensor([0.0394, 0.0034, 0.0032, 0.0313, 0.0064, 0.0076, 0.0053, 0.0342], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0056, 0.0059, 0.0117, 0.0066, 0.0077, 0.0067, 0.0112], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 17:58:37,688 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-28 17:58:38,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3803, 3.5600, 1.5772, 3.9823, 2.5886, 3.9501, 1.8482, 2.7706], device='cuda:7'), covar=tensor([0.0203, 0.0325, 0.1800, 0.0078, 0.0728, 0.0360, 0.1623, 0.0610], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0156, 0.0180, 0.0096, 0.0164, 0.0193, 0.0190, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 17:58:40,389 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69041.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:58:43,492 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69043.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:58:55,994 INFO [train.py:904] (7/8) Epoch 7, batch 8150, loss[loss=0.2395, simple_loss=0.3094, pruned_loss=0.08483, over 11627.00 frames. ], tot_loss[loss=0.236, simple_loss=0.3139, pruned_loss=0.07904, over 3101999.73 frames. ], batch size: 248, lr: 9.54e-03, grad_scale: 4.0 2023-04-28 17:59:00,800 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69055.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 17:59:59,700 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.706e+02 3.772e+02 4.723e+02 6.532e+02 2.181e+03, threshold=9.446e+02, percent-clipped=10.0 2023-04-28 18:00:10,172 INFO [train.py:904] (7/8) Epoch 7, batch 8200, loss[loss=0.2426, simple_loss=0.3245, pruned_loss=0.08029, over 16343.00 frames. ], tot_loss[loss=0.2344, simple_loss=0.3119, pruned_loss=0.07851, over 3112349.75 frames. ], batch size: 146, lr: 9.53e-03, grad_scale: 4.0 2023-04-28 18:00:12,049 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69103.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:00:14,176 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69104.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:01:30,742 INFO [train.py:904] (7/8) Epoch 7, batch 8250, loss[loss=0.1966, simple_loss=0.2766, pruned_loss=0.05836, over 11612.00 frames. ], tot_loss[loss=0.2323, simple_loss=0.3114, pruned_loss=0.07659, over 3085872.72 frames. ], batch size: 246, lr: 9.53e-03, grad_scale: 4.0 2023-04-28 18:01:47,922 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69162.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:01:58,245 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69168.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:02:24,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4928, 4.8155, 4.6307, 4.6329, 4.3215, 4.2717, 4.4012, 4.8803], device='cuda:7'), covar=tensor([0.0726, 0.0817, 0.0727, 0.0508, 0.0646, 0.0972, 0.0721, 0.0781], device='cuda:7'), in_proj_covar=tensor([0.0446, 0.0561, 0.0478, 0.0374, 0.0352, 0.0383, 0.0465, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:02:40,833 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.107e+02 2.942e+02 3.716e+02 4.471e+02 8.485e+02, threshold=7.433e+02, percent-clipped=0.0 2023-04-28 18:02:52,452 INFO [train.py:904] (7/8) Epoch 7, batch 8300, loss[loss=0.2104, simple_loss=0.3027, pruned_loss=0.05902, over 16462.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.3083, pruned_loss=0.07339, over 3084012.01 frames. ], batch size: 146, lr: 9.53e-03, grad_scale: 4.0 2023-04-28 18:03:01,262 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0116, 5.2937, 5.0602, 5.0710, 4.7981, 4.6812, 4.8564, 5.3941], device='cuda:7'), covar=tensor([0.0787, 0.0868, 0.0979, 0.0616, 0.0737, 0.0757, 0.0886, 0.0901], device='cuda:7'), in_proj_covar=tensor([0.0446, 0.0560, 0.0478, 0.0375, 0.0352, 0.0382, 0.0464, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:03:26,098 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69223.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:03:29,186 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69225.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:03:35,601 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69229.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:03:42,806 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69234.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:04:00,587 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69245.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:04:07,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4721, 5.7806, 5.4967, 5.5866, 5.1205, 5.0107, 5.2871, 5.8437], device='cuda:7'), covar=tensor([0.0696, 0.0731, 0.0945, 0.0467, 0.0720, 0.0541, 0.0757, 0.0726], device='cuda:7'), in_proj_covar=tensor([0.0439, 0.0552, 0.0471, 0.0369, 0.0348, 0.0376, 0.0458, 0.0409], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:04:11,886 INFO [train.py:904] (7/8) Epoch 7, batch 8350, loss[loss=0.1925, simple_loss=0.2913, pruned_loss=0.0468, over 16890.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3069, pruned_loss=0.07077, over 3073235.97 frames. ], batch size: 96, lr: 9.52e-03, grad_scale: 4.0 2023-04-28 18:04:46,423 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:05:00,053 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69282.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:05:07,155 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69286.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:05:20,697 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.640e+02 3.244e+02 4.107e+02 9.415e+02, threshold=6.488e+02, percent-clipped=2.0 2023-04-28 18:05:32,120 INFO [train.py:904] (7/8) Epoch 7, batch 8400, loss[loss=0.1897, simple_loss=0.2731, pruned_loss=0.05314, over 12430.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3032, pruned_loss=0.06801, over 3071894.03 frames. ], batch size: 246, lr: 9.52e-03, grad_scale: 8.0 2023-04-28 18:05:39,390 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69306.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 18:06:02,633 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:06:23,823 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69334.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:06:35,207 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69341.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:06:52,099 INFO [train.py:904] (7/8) Epoch 7, batch 8450, loss[loss=0.1871, simple_loss=0.2789, pruned_loss=0.04764, over 16880.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.3012, pruned_loss=0.06652, over 3052871.41 frames. ], batch size: 109, lr: 9.52e-03, grad_scale: 8.0 2023-04-28 18:07:18,774 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69368.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:28,820 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6935, 4.9115, 4.7199, 4.7325, 4.3854, 4.3460, 4.4982, 4.9793], device='cuda:7'), covar=tensor([0.0796, 0.0822, 0.1048, 0.0553, 0.0754, 0.0937, 0.0801, 0.0807], device='cuda:7'), in_proj_covar=tensor([0.0436, 0.0553, 0.0472, 0.0371, 0.0349, 0.0377, 0.0462, 0.0408], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:07:38,843 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69381.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:50,688 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69389.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:58,604 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69394.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:07:59,253 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.708e+02 3.167e+02 3.913e+02 6.077e+02, threshold=6.334e+02, percent-clipped=0.0 2023-04-28 18:08:00,196 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-28 18:08:06,589 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69399.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:08:10,415 INFO [train.py:904] (7/8) Epoch 7, batch 8500, loss[loss=0.1894, simple_loss=0.2777, pruned_loss=0.0506, over 16520.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2974, pruned_loss=0.0635, over 3068890.34 frames. ], batch size: 62, lr: 9.51e-03, grad_scale: 8.0 2023-04-28 18:08:48,587 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8199, 4.1357, 3.9515, 4.0026, 3.5671, 3.7532, 3.8729, 4.0879], device='cuda:7'), covar=tensor([0.0944, 0.0993, 0.0985, 0.0625, 0.0792, 0.1489, 0.0795, 0.1040], device='cuda:7'), in_proj_covar=tensor([0.0431, 0.0548, 0.0466, 0.0365, 0.0344, 0.0373, 0.0456, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:08:55,060 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69429.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:09:33,189 INFO [train.py:904] (7/8) Epoch 7, batch 8550, loss[loss=0.2401, simple_loss=0.3233, pruned_loss=0.07846, over 15446.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2952, pruned_loss=0.06269, over 3053543.38 frames. ], batch size: 191, lr: 9.51e-03, grad_scale: 8.0 2023-04-28 18:09:41,868 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69455.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:10:58,767 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.006e+02 3.588e+02 4.597e+02 9.217e+02, threshold=7.177e+02, percent-clipped=6.0 2023-04-28 18:11:12,041 INFO [train.py:904] (7/8) Epoch 7, batch 8600, loss[loss=0.1933, simple_loss=0.2888, pruned_loss=0.04891, over 15300.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.2946, pruned_loss=0.06095, over 3058670.99 frames. ], batch size: 190, lr: 9.51e-03, grad_scale: 8.0 2023-04-28 18:11:46,328 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69518.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:11:56,254 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69524.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:12:14,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9248, 3.1200, 3.1359, 2.0803, 2.9816, 3.1317, 2.9895, 1.6529], device='cuda:7'), covar=tensor([0.0368, 0.0028, 0.0034, 0.0261, 0.0051, 0.0044, 0.0044, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0056, 0.0058, 0.0115, 0.0064, 0.0073, 0.0066, 0.0111], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 18:12:47,749 INFO [train.py:904] (7/8) Epoch 7, batch 8650, loss[loss=0.2015, simple_loss=0.2827, pruned_loss=0.06014, over 12059.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2923, pruned_loss=0.05909, over 3059126.99 frames. ], batch size: 248, lr: 9.50e-03, grad_scale: 8.0 2023-04-28 18:12:49,835 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-28 18:12:57,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0885, 2.6408, 2.6428, 1.8567, 2.8280, 2.8659, 2.4703, 2.4701], device='cuda:7'), covar=tensor([0.0591, 0.0155, 0.0150, 0.0903, 0.0065, 0.0111, 0.0367, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0090, 0.0077, 0.0134, 0.0063, 0.0080, 0.0112, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 18:13:54,365 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69581.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:14:19,156 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.551e+02 2.997e+02 3.527e+02 6.166e+02, threshold=5.994e+02, percent-clipped=0.0 2023-04-28 18:14:31,863 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69601.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 18:14:32,629 INFO [train.py:904] (7/8) Epoch 7, batch 8700, loss[loss=0.205, simple_loss=0.2831, pruned_loss=0.06348, over 12364.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2896, pruned_loss=0.05762, over 3072581.13 frames. ], batch size: 247, lr: 9.50e-03, grad_scale: 8.0 2023-04-28 18:15:06,509 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6561, 1.9958, 2.2384, 4.2680, 1.8811, 2.7356, 2.2009, 2.2432], device='cuda:7'), covar=tensor([0.0599, 0.2978, 0.1655, 0.0265, 0.3557, 0.1605, 0.2541, 0.2740], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0341, 0.0288, 0.0304, 0.0380, 0.0363, 0.0305, 0.0397], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:15:13,983 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-28 18:15:17,963 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5841, 2.7228, 2.4405, 3.9279, 2.6428, 3.9557, 1.3743, 2.9703], device='cuda:7'), covar=tensor([0.1756, 0.0676, 0.1164, 0.0145, 0.0172, 0.0351, 0.1890, 0.0730], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0146, 0.0169, 0.0105, 0.0186, 0.0194, 0.0170, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 18:15:18,252 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 18:15:22,790 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69629.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:15:41,078 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6501, 2.7491, 1.8049, 2.8585, 2.0610, 2.8406, 2.0101, 2.4514], device='cuda:7'), covar=tensor([0.0234, 0.0348, 0.1277, 0.0134, 0.0744, 0.0486, 0.1316, 0.0654], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0148, 0.0174, 0.0090, 0.0158, 0.0181, 0.0184, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 18:16:08,643 INFO [train.py:904] (7/8) Epoch 7, batch 8750, loss[loss=0.2175, simple_loss=0.3057, pruned_loss=0.06468, over 15261.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2896, pruned_loss=0.05727, over 3079367.37 frames. ], batch size: 190, lr: 9.50e-03, grad_scale: 8.0 2023-04-28 18:17:04,338 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69674.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:17:07,499 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69676.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:17:18,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5994, 3.3884, 3.0981, 1.8159, 2.6151, 2.2970, 3.0974, 3.2875], device='cuda:7'), covar=tensor([0.0337, 0.0535, 0.0577, 0.1750, 0.0805, 0.0899, 0.0781, 0.0800], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0126, 0.0152, 0.0139, 0.0131, 0.0123, 0.0134, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 18:17:45,989 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.592e+02 3.192e+02 4.101e+02 7.533e+02, threshold=6.383e+02, percent-clipped=6.0 2023-04-28 18:17:55,801 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69699.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:18:01,178 INFO [train.py:904] (7/8) Epoch 7, batch 8800, loss[loss=0.182, simple_loss=0.2688, pruned_loss=0.04762, over 17026.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2876, pruned_loss=0.05568, over 3086312.93 frames. ], batch size: 53, lr: 9.49e-03, grad_scale: 8.0 2023-04-28 18:18:47,958 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69724.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:19:07,540 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-28 18:19:11,511 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69735.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:19:36,495 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69747.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:19:43,882 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69750.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:19:46,943 INFO [train.py:904] (7/8) Epoch 7, batch 8850, loss[loss=0.2017, simple_loss=0.3037, pruned_loss=0.04988, over 16672.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2895, pruned_loss=0.05492, over 3066555.70 frames. ], batch size: 62, lr: 9.49e-03, grad_scale: 4.0 2023-04-28 18:21:21,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.819e+02 2.840e+02 3.473e+02 4.436e+02 7.105e+02, threshold=6.946e+02, percent-clipped=4.0 2023-04-28 18:21:34,931 INFO [train.py:904] (7/8) Epoch 7, batch 8900, loss[loss=0.2011, simple_loss=0.29, pruned_loss=0.05606, over 16480.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.29, pruned_loss=0.05386, over 3084482.48 frames. ], batch size: 68, lr: 9.49e-03, grad_scale: 4.0 2023-04-28 18:21:54,580 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9048, 2.6983, 2.7064, 1.8977, 2.5423, 2.6390, 2.5810, 1.6856], device='cuda:7'), covar=tensor([0.0306, 0.0031, 0.0044, 0.0240, 0.0067, 0.0053, 0.0047, 0.0326], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0055, 0.0058, 0.0114, 0.0064, 0.0071, 0.0065, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0001, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 18:22:06,698 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69818.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:22:19,724 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 18:22:22,117 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69824.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:22:37,623 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8222, 1.2982, 1.4624, 1.7950, 1.8113, 1.8054, 1.5306, 1.8304], device='cuda:7'), covar=tensor([0.0097, 0.0217, 0.0113, 0.0151, 0.0152, 0.0112, 0.0224, 0.0081], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0149, 0.0132, 0.0132, 0.0142, 0.0096, 0.0148, 0.0089], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 18:22:50,164 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1264, 2.8837, 2.8506, 2.0171, 2.6322, 2.1556, 2.7182, 2.9486], device='cuda:7'), covar=tensor([0.0276, 0.0640, 0.0478, 0.1438, 0.0651, 0.0891, 0.0588, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0126, 0.0154, 0.0140, 0.0132, 0.0125, 0.0134, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 18:23:09,054 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8756, 5.1754, 4.9020, 4.9184, 4.6548, 4.6175, 4.6068, 5.2421], device='cuda:7'), covar=tensor([0.0771, 0.0748, 0.0989, 0.0587, 0.0716, 0.0738, 0.0847, 0.0717], device='cuda:7'), in_proj_covar=tensor([0.0429, 0.0547, 0.0460, 0.0365, 0.0344, 0.0370, 0.0455, 0.0402], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:23:38,925 INFO [train.py:904] (7/8) Epoch 7, batch 8950, loss[loss=0.1858, simple_loss=0.2742, pruned_loss=0.04871, over 15601.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2895, pruned_loss=0.05456, over 3064482.19 frames. ], batch size: 193, lr: 9.48e-03, grad_scale: 4.0 2023-04-28 18:24:09,621 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69866.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:24:22,340 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69872.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:24:43,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69881.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:24:48,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3640, 3.2773, 3.3820, 3.4905, 3.5387, 3.2272, 3.5029, 3.5646], device='cuda:7'), covar=tensor([0.0955, 0.0771, 0.0958, 0.0518, 0.0510, 0.1823, 0.0619, 0.0502], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0514, 0.0628, 0.0517, 0.0388, 0.0394, 0.0404, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:25:14,738 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.573e+02 2.911e+02 3.451e+02 7.486e+02, threshold=5.822e+02, percent-clipped=1.0 2023-04-28 18:25:27,291 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69901.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:25:27,983 INFO [train.py:904] (7/8) Epoch 7, batch 9000, loss[loss=0.1659, simple_loss=0.2561, pruned_loss=0.03788, over 16724.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2863, pruned_loss=0.05344, over 3055648.11 frames. ], batch size: 83, lr: 9.48e-03, grad_scale: 4.0 2023-04-28 18:25:27,983 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 18:25:37,200 INFO [train.py:938] (7/8) Epoch 7, validation: loss=0.1647, simple_loss=0.2682, pruned_loss=0.03062, over 944034.00 frames. 2023-04-28 18:25:37,201 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 18:26:33,159 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69929.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:26:33,224 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69929.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:26:57,586 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-28 18:27:14,854 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69949.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:27:19,727 INFO [train.py:904] (7/8) Epoch 7, batch 9050, loss[loss=0.2175, simple_loss=0.3023, pruned_loss=0.06637, over 17008.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2864, pruned_loss=0.05361, over 3056987.08 frames. ], batch size: 55, lr: 9.48e-03, grad_scale: 4.0 2023-04-28 18:27:57,451 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9743, 3.9828, 3.8394, 3.4424, 3.9250, 1.8020, 3.7357, 3.6599], device='cuda:7'), covar=tensor([0.0093, 0.0091, 0.0172, 0.0231, 0.0080, 0.1985, 0.0121, 0.0173], device='cuda:7'), in_proj_covar=tensor([0.0101, 0.0089, 0.0137, 0.0125, 0.0102, 0.0156, 0.0121, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:28:08,861 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69976.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:28:10,102 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=69977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:28:50,087 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 2.800e+02 3.566e+02 4.608e+02 8.238e+02, threshold=7.132e+02, percent-clipped=9.0 2023-04-28 18:29:06,540 INFO [train.py:904] (7/8) Epoch 7, batch 9100, loss[loss=0.2081, simple_loss=0.288, pruned_loss=0.06416, over 12545.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2862, pruned_loss=0.05394, over 3074639.38 frames. ], batch size: 248, lr: 9.47e-03, grad_scale: 4.0 2023-04-28 18:29:58,315 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:29:58,440 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:30:13,301 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70030.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:31:01,029 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70050.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:31:03,199 INFO [train.py:904] (7/8) Epoch 7, batch 9150, loss[loss=0.1912, simple_loss=0.2722, pruned_loss=0.05513, over 11966.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2862, pruned_loss=0.05367, over 3047173.41 frames. ], batch size: 250, lr: 9.47e-03, grad_scale: 4.0 2023-04-28 18:31:47,595 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70072.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:32:35,336 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.681e+02 3.453e+02 4.314e+02 7.152e+02, threshold=6.905e+02, percent-clipped=1.0 2023-04-28 18:32:36,500 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3658, 4.3797, 4.5249, 4.4503, 4.4665, 4.9956, 4.6291, 4.3418], device='cuda:7'), covar=tensor([0.1092, 0.1608, 0.1452, 0.1597, 0.2212, 0.0875, 0.1136, 0.2053], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0393, 0.0412, 0.0348, 0.0454, 0.0438, 0.0331, 0.0464], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:32:39,090 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70098.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:32:44,515 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6430, 2.7319, 1.7696, 2.8464, 2.0541, 2.7962, 1.9490, 2.3944], device='cuda:7'), covar=tensor([0.0219, 0.0317, 0.1334, 0.0140, 0.0695, 0.0422, 0.1313, 0.0587], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0149, 0.0176, 0.0090, 0.0157, 0.0180, 0.0184, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-28 18:32:45,143 INFO [train.py:904] (7/8) Epoch 7, batch 9200, loss[loss=0.207, simple_loss=0.2849, pruned_loss=0.06454, over 12287.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.282, pruned_loss=0.05243, over 3058197.44 frames. ], batch size: 247, lr: 9.47e-03, grad_scale: 8.0 2023-04-28 18:33:01,861 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70110.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:34:22,208 INFO [train.py:904] (7/8) Epoch 7, batch 9250, loss[loss=0.1737, simple_loss=0.2684, pruned_loss=0.03956, over 16778.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2815, pruned_loss=0.05218, over 3066396.64 frames. ], batch size: 83, lr: 9.46e-03, grad_scale: 8.0 2023-04-28 18:35:02,185 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70171.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:35:24,061 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9555, 3.5632, 3.4362, 1.9050, 2.9456, 2.2230, 3.3137, 3.5726], device='cuda:7'), covar=tensor([0.0306, 0.0612, 0.0482, 0.1726, 0.0705, 0.1074, 0.0682, 0.0800], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0124, 0.0151, 0.0138, 0.0132, 0.0123, 0.0133, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 18:36:01,191 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.717e+02 3.344e+02 3.929e+02 8.889e+02, threshold=6.688e+02, percent-clipped=4.0 2023-04-28 18:36:13,171 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 18:36:13,566 INFO [train.py:904] (7/8) Epoch 7, batch 9300, loss[loss=0.1691, simple_loss=0.2647, pruned_loss=0.03675, over 16875.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.28, pruned_loss=0.05166, over 3058441.82 frames. ], batch size: 90, lr: 9.46e-03, grad_scale: 8.0 2023-04-28 18:37:54,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8393, 4.0478, 4.1213, 1.9185, 4.3725, 4.3564, 3.0694, 3.4208], device='cuda:7'), covar=tensor([0.0587, 0.0117, 0.0173, 0.1099, 0.0026, 0.0050, 0.0315, 0.0301], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0091, 0.0079, 0.0135, 0.0064, 0.0081, 0.0113, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 18:37:59,035 INFO [train.py:904] (7/8) Epoch 7, batch 9350, loss[loss=0.1821, simple_loss=0.2602, pruned_loss=0.05204, over 12237.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2796, pruned_loss=0.05152, over 3060967.24 frames. ], batch size: 247, lr: 9.46e-03, grad_scale: 8.0 2023-04-28 18:38:18,386 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70260.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:39:31,297 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.755e+02 2.689e+02 3.082e+02 3.847e+02 8.568e+02, threshold=6.165e+02, percent-clipped=3.0 2023-04-28 18:39:40,132 INFO [train.py:904] (7/8) Epoch 7, batch 9400, loss[loss=0.203, simple_loss=0.2971, pruned_loss=0.05446, over 16250.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2807, pruned_loss=0.05164, over 3069935.36 frames. ], batch size: 165, lr: 9.45e-03, grad_scale: 4.0 2023-04-28 18:39:44,223 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 18:39:46,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3942, 4.5148, 4.2802, 4.1290, 3.8022, 4.3765, 4.2743, 4.0765], device='cuda:7'), covar=tensor([0.0576, 0.0352, 0.0278, 0.0214, 0.0995, 0.0416, 0.0399, 0.0529], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0223, 0.0225, 0.0199, 0.0247, 0.0227, 0.0156, 0.0262], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:40:18,694 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70321.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:40:36,614 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70330.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 18:40:43,480 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0496, 2.6147, 2.5901, 1.7786, 2.8462, 2.8394, 2.4006, 2.4453], device='cuda:7'), covar=tensor([0.0613, 0.0151, 0.0141, 0.0897, 0.0065, 0.0124, 0.0392, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0091, 0.0079, 0.0136, 0.0064, 0.0081, 0.0114, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 18:41:20,564 INFO [train.py:904] (7/8) Epoch 7, batch 9450, loss[loss=0.1919, simple_loss=0.2816, pruned_loss=0.0511, over 15465.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2827, pruned_loss=0.05239, over 3062807.55 frames. ], batch size: 191, lr: 9.45e-03, grad_scale: 4.0 2023-04-28 18:42:14,167 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70378.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 18:42:50,912 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.717e+02 3.594e+02 4.518e+02 8.204e+02, threshold=7.189e+02, percent-clipped=4.0 2023-04-28 18:43:02,106 INFO [train.py:904] (7/8) Epoch 7, batch 9500, loss[loss=0.1931, simple_loss=0.2844, pruned_loss=0.05093, over 16912.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2811, pruned_loss=0.0516, over 3059818.38 frames. ], batch size: 96, lr: 9.45e-03, grad_scale: 4.0 2023-04-28 18:44:47,149 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0996, 4.1127, 3.9436, 3.7632, 3.5721, 4.0221, 3.8357, 3.7661], device='cuda:7'), covar=tensor([0.0531, 0.0482, 0.0260, 0.0244, 0.0817, 0.0377, 0.0575, 0.0537], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0220, 0.0221, 0.0198, 0.0245, 0.0224, 0.0156, 0.0260], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:44:47,825 INFO [train.py:904] (7/8) Epoch 7, batch 9550, loss[loss=0.2098, simple_loss=0.3021, pruned_loss=0.05875, over 16285.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2812, pruned_loss=0.05207, over 3056271.26 frames. ], batch size: 166, lr: 9.44e-03, grad_scale: 4.0 2023-04-28 18:45:18,572 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:46:10,425 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-28 18:46:18,726 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.839e+02 3.500e+02 4.170e+02 7.519e+02, threshold=6.999e+02, percent-clipped=3.0 2023-04-28 18:46:26,845 INFO [train.py:904] (7/8) Epoch 7, batch 9600, loss[loss=0.2037, simple_loss=0.2954, pruned_loss=0.05597, over 16766.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.2817, pruned_loss=0.05262, over 3046938.20 frames. ], batch size: 83, lr: 9.44e-03, grad_scale: 8.0 2023-04-28 18:48:15,890 INFO [train.py:904] (7/8) Epoch 7, batch 9650, loss[loss=0.222, simple_loss=0.3133, pruned_loss=0.06537, over 15381.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.284, pruned_loss=0.0523, over 3068795.06 frames. ], batch size: 191, lr: 9.44e-03, grad_scale: 8.0 2023-04-28 18:48:43,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8072, 3.5772, 3.2452, 1.9298, 2.7957, 2.3339, 3.0953, 3.3913], device='cuda:7'), covar=tensor([0.0370, 0.0503, 0.0522, 0.1627, 0.0712, 0.0836, 0.0953, 0.0853], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0121, 0.0151, 0.0136, 0.0129, 0.0121, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 18:49:21,103 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2234, 4.2074, 4.3438, 4.2756, 4.3061, 4.7620, 4.4108, 4.1569], device='cuda:7'), covar=tensor([0.1326, 0.1694, 0.1216, 0.1704, 0.2286, 0.0932, 0.1101, 0.1949], device='cuda:7'), in_proj_covar=tensor([0.0274, 0.0394, 0.0410, 0.0344, 0.0449, 0.0433, 0.0328, 0.0455], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:49:55,425 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 2.673e+02 3.296e+02 3.931e+02 1.187e+03, threshold=6.593e+02, percent-clipped=2.0 2023-04-28 18:50:06,090 INFO [train.py:904] (7/8) Epoch 7, batch 9700, loss[loss=0.1908, simple_loss=0.2791, pruned_loss=0.05128, over 16396.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2829, pruned_loss=0.05228, over 3056678.59 frames. ], batch size: 146, lr: 9.43e-03, grad_scale: 8.0 2023-04-28 18:50:29,030 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3957, 3.0441, 2.6765, 2.2458, 2.1378, 2.1249, 2.9832, 2.8993], device='cuda:7'), covar=tensor([0.2135, 0.0668, 0.1168, 0.1801, 0.2155, 0.1716, 0.0404, 0.0840], device='cuda:7'), in_proj_covar=tensor([0.0277, 0.0238, 0.0266, 0.0251, 0.0243, 0.0204, 0.0242, 0.0252], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:50:33,470 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70616.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:50:40,075 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-28 18:51:47,807 INFO [train.py:904] (7/8) Epoch 7, batch 9750, loss[loss=0.177, simple_loss=0.2621, pruned_loss=0.04596, over 12622.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2821, pruned_loss=0.05274, over 3043221.47 frames. ], batch size: 246, lr: 9.43e-03, grad_scale: 8.0 2023-04-28 18:52:24,181 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2322, 1.8647, 2.0749, 3.7831, 1.8447, 2.3549, 2.1087, 2.0083], device='cuda:7'), covar=tensor([0.0701, 0.3240, 0.1812, 0.0321, 0.3558, 0.1885, 0.2645, 0.3231], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0340, 0.0291, 0.0304, 0.0382, 0.0365, 0.0308, 0.0399], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:52:50,889 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8253, 4.8178, 4.5588, 4.1492, 4.6365, 1.8211, 4.4615, 4.4645], device='cuda:7'), covar=tensor([0.0045, 0.0034, 0.0096, 0.0170, 0.0048, 0.1759, 0.0076, 0.0115], device='cuda:7'), in_proj_covar=tensor([0.0100, 0.0088, 0.0133, 0.0122, 0.0101, 0.0154, 0.0119, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:53:18,482 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.655e+02 3.090e+02 3.865e+02 1.050e+03, threshold=6.180e+02, percent-clipped=3.0 2023-04-28 18:53:27,264 INFO [train.py:904] (7/8) Epoch 7, batch 9800, loss[loss=0.2109, simple_loss=0.3076, pruned_loss=0.05713, over 16805.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2816, pruned_loss=0.05153, over 3048972.96 frames. ], batch size: 124, lr: 9.43e-03, grad_scale: 8.0 2023-04-28 18:55:12,750 INFO [train.py:904] (7/8) Epoch 7, batch 9850, loss[loss=0.1845, simple_loss=0.2747, pruned_loss=0.04718, over 16138.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2824, pruned_loss=0.05081, over 3057133.75 frames. ], batch size: 165, lr: 9.42e-03, grad_scale: 8.0 2023-04-28 18:55:43,748 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70766.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:56:54,664 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.569e+02 3.072e+02 3.785e+02 9.202e+02, threshold=6.144e+02, percent-clipped=3.0 2023-04-28 18:57:04,350 INFO [train.py:904] (7/8) Epoch 7, batch 9900, loss[loss=0.2001, simple_loss=0.295, pruned_loss=0.05262, over 16263.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2827, pruned_loss=0.05091, over 3040545.87 frames. ], batch size: 165, lr: 9.42e-03, grad_scale: 8.0 2023-04-28 18:57:33,476 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70814.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 18:58:34,776 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3015, 3.2688, 2.7254, 2.1161, 2.1602, 2.1633, 3.2888, 3.1307], device='cuda:7'), covar=tensor([0.2313, 0.0615, 0.1337, 0.2056, 0.1776, 0.1500, 0.0460, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0239, 0.0268, 0.0252, 0.0241, 0.0204, 0.0243, 0.0253], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 18:59:01,229 INFO [train.py:904] (7/8) Epoch 7, batch 9950, loss[loss=0.1905, simple_loss=0.2896, pruned_loss=0.04569, over 16341.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2852, pruned_loss=0.05142, over 3040392.60 frames. ], batch size: 146, lr: 9.42e-03, grad_scale: 8.0 2023-04-28 18:59:34,314 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 19:00:20,139 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3904, 1.4028, 1.8517, 2.3309, 2.2939, 2.4198, 1.6225, 2.4923], device='cuda:7'), covar=tensor([0.0095, 0.0296, 0.0183, 0.0134, 0.0151, 0.0097, 0.0260, 0.0070], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0151, 0.0135, 0.0134, 0.0141, 0.0096, 0.0149, 0.0088], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 19:00:47,806 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.726e+02 3.249e+02 3.973e+02 1.098e+03, threshold=6.498e+02, percent-clipped=6.0 2023-04-28 19:00:59,963 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4112, 1.9018, 1.6466, 1.5843, 2.1861, 1.9405, 2.1953, 2.3421], device='cuda:7'), covar=tensor([0.0061, 0.0247, 0.0272, 0.0324, 0.0132, 0.0224, 0.0116, 0.0148], device='cuda:7'), in_proj_covar=tensor([0.0097, 0.0175, 0.0171, 0.0169, 0.0167, 0.0173, 0.0155, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:01:00,574 INFO [train.py:904] (7/8) Epoch 7, batch 10000, loss[loss=0.1861, simple_loss=0.2843, pruned_loss=0.04398, over 15428.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2844, pruned_loss=0.05109, over 3067531.74 frames. ], batch size: 192, lr: 9.41e-03, grad_scale: 8.0 2023-04-28 19:01:29,811 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70916.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:02:40,837 INFO [train.py:904] (7/8) Epoch 7, batch 10050, loss[loss=0.21, simple_loss=0.2926, pruned_loss=0.06371, over 12494.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2842, pruned_loss=0.05083, over 3075052.56 frames. ], batch size: 248, lr: 9.41e-03, grad_scale: 8.0 2023-04-28 19:03:04,295 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=70964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:03:31,341 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7501, 3.6860, 3.8238, 3.6945, 3.8259, 4.2199, 3.9007, 3.6155], device='cuda:7'), covar=tensor([0.2119, 0.1891, 0.1779, 0.2132, 0.2524, 0.1552, 0.1536, 0.2611], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0401, 0.0424, 0.0353, 0.0460, 0.0440, 0.0334, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:03:56,303 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70992.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:04:04,313 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.613e+02 3.026e+02 3.875e+02 7.173e+02, threshold=6.053e+02, percent-clipped=4.0 2023-04-28 19:04:12,154 INFO [train.py:904] (7/8) Epoch 7, batch 10100, loss[loss=0.183, simple_loss=0.2601, pruned_loss=0.05298, over 12530.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2845, pruned_loss=0.05143, over 3059452.71 frames. ], batch size: 248, lr: 9.41e-03, grad_scale: 8.0 2023-04-28 19:04:20,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7246, 5.0609, 5.1535, 5.0023, 5.0322, 5.5613, 5.0405, 4.8141], device='cuda:7'), covar=tensor([0.0758, 0.1253, 0.1093, 0.1402, 0.2005, 0.0753, 0.1127, 0.1884], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0397, 0.0417, 0.0351, 0.0456, 0.0437, 0.0330, 0.0464], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:05:07,005 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71030.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:05:13,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4319, 3.1393, 3.1863, 1.9076, 2.8033, 2.0570, 3.0692, 3.0267], device='cuda:7'), covar=tensor([0.0296, 0.0569, 0.0415, 0.1630, 0.0626, 0.0947, 0.0624, 0.0815], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0120, 0.0151, 0.0137, 0.0129, 0.0121, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 19:05:25,379 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 19:05:25,999 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5131, 3.5067, 3.4898, 2.9987, 3.4473, 1.9538, 3.2771, 2.8845], device='cuda:7'), covar=tensor([0.0097, 0.0081, 0.0114, 0.0189, 0.0081, 0.1848, 0.0104, 0.0154], device='cuda:7'), in_proj_covar=tensor([0.0100, 0.0088, 0.0134, 0.0121, 0.0101, 0.0155, 0.0119, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:05:54,951 INFO [train.py:904] (7/8) Epoch 8, batch 0, loss[loss=0.244, simple_loss=0.318, pruned_loss=0.08503, over 17067.00 frames. ], tot_loss[loss=0.244, simple_loss=0.318, pruned_loss=0.08503, over 17067.00 frames. ], batch size: 53, lr: 8.86e-03, grad_scale: 8.0 2023-04-28 19:05:54,952 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 19:06:02,579 INFO [train.py:938] (7/8) Epoch 8, validation: loss=0.1627, simple_loss=0.2663, pruned_loss=0.02958, over 944034.00 frames. 2023-04-28 19:06:02,580 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 19:06:04,031 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71053.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:06:25,201 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-28 19:06:45,777 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2149, 4.1410, 4.6422, 4.6487, 4.6502, 4.2543, 4.2719, 4.1907], device='cuda:7'), covar=tensor([0.0274, 0.0696, 0.0361, 0.0373, 0.0430, 0.0360, 0.0826, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0254, 0.0262, 0.0260, 0.0248, 0.0294, 0.0278, 0.0361, 0.0227], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-28 19:06:55,602 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71091.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:07:07,690 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 3.480e+02 4.237e+02 5.161e+02 1.332e+03, threshold=8.475e+02, percent-clipped=15.0 2023-04-28 19:07:09,986 INFO [train.py:904] (7/8) Epoch 8, batch 50, loss[loss=0.2166, simple_loss=0.2863, pruned_loss=0.07341, over 16718.00 frames. ], tot_loss[loss=0.2324, simple_loss=0.3073, pruned_loss=0.07873, over 740892.49 frames. ], batch size: 124, lr: 8.86e-03, grad_scale: 1.0 2023-04-28 19:08:09,435 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-28 19:08:17,899 INFO [train.py:904] (7/8) Epoch 8, batch 100, loss[loss=0.2016, simple_loss=0.2896, pruned_loss=0.05678, over 17050.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.298, pruned_loss=0.07235, over 1312874.35 frames. ], batch size: 55, lr: 8.85e-03, grad_scale: 1.0 2023-04-28 19:08:54,550 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-28 19:09:23,829 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 2.858e+02 3.485e+02 4.000e+02 7.217e+02, threshold=6.971e+02, percent-clipped=0.0 2023-04-28 19:09:26,722 INFO [train.py:904] (7/8) Epoch 8, batch 150, loss[loss=0.2479, simple_loss=0.3085, pruned_loss=0.09365, over 16853.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.2936, pruned_loss=0.06992, over 1764139.04 frames. ], batch size: 116, lr: 8.85e-03, grad_scale: 1.0 2023-04-28 19:10:04,807 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2023-04-28 19:10:15,509 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-28 19:10:33,458 INFO [train.py:904] (7/8) Epoch 8, batch 200, loss[loss=0.173, simple_loss=0.2584, pruned_loss=0.04381, over 16990.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2938, pruned_loss=0.0697, over 2100891.26 frames. ], batch size: 41, lr: 8.85e-03, grad_scale: 1.0 2023-04-28 19:11:39,315 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6197, 4.5812, 4.4368, 3.9092, 4.5132, 1.7903, 4.3098, 4.2030], device='cuda:7'), covar=tensor([0.0086, 0.0063, 0.0120, 0.0270, 0.0065, 0.1976, 0.0091, 0.0151], device='cuda:7'), in_proj_covar=tensor([0.0107, 0.0094, 0.0144, 0.0132, 0.0109, 0.0163, 0.0127, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:11:40,018 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.597e+02 3.114e+02 3.652e+02 1.080e+03, threshold=6.228e+02, percent-clipped=1.0 2023-04-28 19:11:42,948 INFO [train.py:904] (7/8) Epoch 8, batch 250, loss[loss=0.2326, simple_loss=0.3236, pruned_loss=0.07074, over 17040.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2902, pruned_loss=0.06829, over 2367444.14 frames. ], batch size: 50, lr: 8.84e-03, grad_scale: 1.0 2023-04-28 19:11:43,388 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7957, 4.4892, 4.8052, 5.0343, 5.1934, 4.6517, 5.1725, 5.1586], device='cuda:7'), covar=tensor([0.1185, 0.0989, 0.1381, 0.0588, 0.0457, 0.0672, 0.0435, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0452, 0.0568, 0.0693, 0.0565, 0.0427, 0.0429, 0.0441, 0.0491], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:11:56,937 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3827, 2.8845, 2.6034, 2.2989, 2.1867, 2.2206, 2.8754, 2.8408], device='cuda:7'), covar=tensor([0.2007, 0.0699, 0.1123, 0.1583, 0.1762, 0.1500, 0.0394, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0245, 0.0275, 0.0257, 0.0258, 0.0209, 0.0250, 0.0269], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:12:02,801 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0511, 3.8745, 4.3574, 3.3852, 3.9256, 4.2534, 4.0217, 2.6186], device='cuda:7'), covar=tensor([0.0285, 0.0040, 0.0024, 0.0186, 0.0051, 0.0055, 0.0034, 0.0290], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0063, 0.0062, 0.0119, 0.0066, 0.0075, 0.0067, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 19:12:23,549 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-28 19:12:46,286 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3240, 3.4842, 1.7722, 3.4643, 2.5256, 3.5271, 1.9078, 2.8141], device='cuda:7'), covar=tensor([0.0216, 0.0354, 0.1646, 0.0275, 0.0788, 0.0649, 0.1404, 0.0571], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0155, 0.0181, 0.0098, 0.0163, 0.0190, 0.0191, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 19:12:47,182 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71348.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:12:52,127 INFO [train.py:904] (7/8) Epoch 8, batch 300, loss[loss=0.1776, simple_loss=0.2715, pruned_loss=0.04183, over 17110.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2854, pruned_loss=0.06576, over 2585948.26 frames. ], batch size: 47, lr: 8.84e-03, grad_scale: 1.0 2023-04-28 19:13:39,988 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71386.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:13:58,500 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.899e+02 2.511e+02 3.030e+02 3.934e+02 1.129e+03, threshold=6.060e+02, percent-clipped=6.0 2023-04-28 19:14:01,162 INFO [train.py:904] (7/8) Epoch 8, batch 350, loss[loss=0.2159, simple_loss=0.2815, pruned_loss=0.07512, over 16702.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2823, pruned_loss=0.0642, over 2742909.75 frames. ], batch size: 134, lr: 8.84e-03, grad_scale: 1.0 2023-04-28 19:14:08,856 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1683, 2.1369, 1.5835, 1.9718, 2.4913, 2.3202, 2.5289, 2.6097], device='cuda:7'), covar=tensor([0.0122, 0.0241, 0.0335, 0.0259, 0.0110, 0.0195, 0.0142, 0.0142], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0182, 0.0179, 0.0177, 0.0174, 0.0183, 0.0172, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:15:10,461 INFO [train.py:904] (7/8) Epoch 8, batch 400, loss[loss=0.1816, simple_loss=0.2548, pruned_loss=0.0542, over 17007.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2811, pruned_loss=0.06395, over 2872738.51 frames. ], batch size: 41, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:15:36,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9671, 3.5185, 2.7860, 4.5995, 3.9868, 4.4675, 1.6807, 3.2199], device='cuda:7'), covar=tensor([0.1275, 0.0412, 0.0985, 0.0096, 0.0300, 0.0316, 0.1376, 0.0680], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0149, 0.0171, 0.0109, 0.0184, 0.0200, 0.0171, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 19:16:17,820 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.889e+02 3.429e+02 4.074e+02 1.363e+03, threshold=6.859e+02, percent-clipped=4.0 2023-04-28 19:16:20,153 INFO [train.py:904] (7/8) Epoch 8, batch 450, loss[loss=0.2011, simple_loss=0.2881, pruned_loss=0.05708, over 17125.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2796, pruned_loss=0.06304, over 2975770.21 frames. ], batch size: 48, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:17:28,363 INFO [train.py:904] (7/8) Epoch 8, batch 500, loss[loss=0.2042, simple_loss=0.2821, pruned_loss=0.06318, over 16391.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2786, pruned_loss=0.06168, over 3060093.02 frames. ], batch size: 75, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:17:37,493 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8055, 2.8336, 2.2588, 2.6143, 3.2078, 2.8607, 3.7756, 3.4972], device='cuda:7'), covar=tensor([0.0038, 0.0194, 0.0270, 0.0226, 0.0117, 0.0209, 0.0119, 0.0122], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0181, 0.0177, 0.0176, 0.0173, 0.0182, 0.0172, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:18:33,696 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.499e+02 3.026e+02 3.865e+02 9.130e+02, threshold=6.053e+02, percent-clipped=1.0 2023-04-28 19:18:37,350 INFO [train.py:904] (7/8) Epoch 8, batch 550, loss[loss=0.1621, simple_loss=0.2484, pruned_loss=0.03787, over 16990.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2772, pruned_loss=0.06023, over 3112195.63 frames. ], batch size: 41, lr: 8.83e-03, grad_scale: 2.0 2023-04-28 19:18:38,191 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-28 19:18:56,802 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71616.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:19:22,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6453, 4.5884, 4.5026, 3.9254, 4.5699, 1.8596, 4.3108, 4.2874], device='cuda:7'), covar=tensor([0.0076, 0.0075, 0.0122, 0.0314, 0.0078, 0.2121, 0.0119, 0.0157], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0098, 0.0150, 0.0139, 0.0113, 0.0166, 0.0133, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:19:35,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0023, 4.0681, 4.3184, 1.8914, 4.5433, 4.6165, 3.1859, 3.4887], device='cuda:7'), covar=tensor([0.0617, 0.0145, 0.0185, 0.1099, 0.0059, 0.0093, 0.0375, 0.0347], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0080, 0.0136, 0.0066, 0.0087, 0.0115, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 19:19:41,724 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71648.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:19:45,777 INFO [train.py:904] (7/8) Epoch 8, batch 600, loss[loss=0.2006, simple_loss=0.2695, pruned_loss=0.06589, over 12358.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2755, pruned_loss=0.05939, over 3158228.24 frames. ], batch size: 246, lr: 8.82e-03, grad_scale: 2.0 2023-04-28 19:20:02,056 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 19:20:18,027 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0036, 4.2405, 2.3481, 4.7572, 3.1080, 4.6106, 2.5302, 3.3574], device='cuda:7'), covar=tensor([0.0182, 0.0226, 0.1358, 0.0078, 0.0722, 0.0357, 0.1304, 0.0547], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0157, 0.0179, 0.0099, 0.0162, 0.0194, 0.0191, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 19:20:19,064 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:20:34,038 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71686.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:20:45,955 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=71696.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:20:50,944 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.658e+02 2.476e+02 3.033e+02 3.542e+02 7.789e+02, threshold=6.066e+02, percent-clipped=2.0 2023-04-28 19:20:53,313 INFO [train.py:904] (7/8) Epoch 8, batch 650, loss[loss=0.1799, simple_loss=0.2577, pruned_loss=0.05107, over 16225.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2737, pruned_loss=0.05881, over 3200738.69 frames. ], batch size: 36, lr: 8.82e-03, grad_scale: 2.0 2023-04-28 19:21:13,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5964, 2.4971, 1.8406, 2.1565, 2.8495, 2.7119, 2.9555, 2.8877], device='cuda:7'), covar=tensor([0.0174, 0.0212, 0.0313, 0.0281, 0.0149, 0.0199, 0.0168, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0182, 0.0178, 0.0178, 0.0175, 0.0184, 0.0176, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:21:37,503 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=71734.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:22:01,904 INFO [train.py:904] (7/8) Epoch 8, batch 700, loss[loss=0.2039, simple_loss=0.2771, pruned_loss=0.06534, over 16844.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.274, pruned_loss=0.05877, over 3230498.99 frames. ], batch size: 102, lr: 8.82e-03, grad_scale: 2.0 2023-04-28 19:23:05,954 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.585e+02 3.182e+02 3.748e+02 9.363e+02, threshold=6.364e+02, percent-clipped=4.0 2023-04-28 19:23:08,650 INFO [train.py:904] (7/8) Epoch 8, batch 750, loss[loss=0.2261, simple_loss=0.2878, pruned_loss=0.08222, over 16733.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2745, pruned_loss=0.05879, over 3255538.96 frames. ], batch size: 134, lr: 8.81e-03, grad_scale: 2.0 2023-04-28 19:23:15,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2283, 4.9577, 5.0732, 5.4419, 5.5876, 4.9024, 5.4196, 5.4545], device='cuda:7'), covar=tensor([0.1212, 0.1033, 0.2080, 0.0711, 0.0622, 0.0619, 0.0790, 0.0720], device='cuda:7'), in_proj_covar=tensor([0.0483, 0.0596, 0.0738, 0.0598, 0.0454, 0.0455, 0.0471, 0.0522], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:23:43,161 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-28 19:24:17,959 INFO [train.py:904] (7/8) Epoch 8, batch 800, loss[loss=0.179, simple_loss=0.2698, pruned_loss=0.04409, over 16765.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2741, pruned_loss=0.05939, over 3266879.32 frames. ], batch size: 57, lr: 8.81e-03, grad_scale: 4.0 2023-04-28 19:24:33,389 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-28 19:24:45,255 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 19:25:05,954 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2655, 2.4261, 1.8952, 2.1793, 2.8498, 2.6397, 3.2571, 3.0963], device='cuda:7'), covar=tensor([0.0086, 0.0266, 0.0370, 0.0314, 0.0169, 0.0240, 0.0180, 0.0179], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0182, 0.0178, 0.0178, 0.0175, 0.0185, 0.0178, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:25:23,720 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.711e+02 2.663e+02 3.116e+02 3.838e+02 7.032e+02, threshold=6.231e+02, percent-clipped=1.0 2023-04-28 19:25:24,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0910, 4.4381, 2.5301, 4.7677, 3.0764, 4.6977, 2.4295, 3.3970], device='cuda:7'), covar=tensor([0.0199, 0.0240, 0.1292, 0.0088, 0.0659, 0.0325, 0.1445, 0.0550], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0156, 0.0177, 0.0099, 0.0162, 0.0194, 0.0190, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 19:25:25,980 INFO [train.py:904] (7/8) Epoch 8, batch 850, loss[loss=0.2213, simple_loss=0.2783, pruned_loss=0.08211, over 16768.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2736, pruned_loss=0.05905, over 3275893.33 frames. ], batch size: 124, lr: 8.81e-03, grad_scale: 4.0 2023-04-28 19:25:49,762 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71919.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:26:32,701 INFO [train.py:904] (7/8) Epoch 8, batch 900, loss[loss=0.1798, simple_loss=0.2625, pruned_loss=0.04854, over 17001.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2725, pruned_loss=0.05796, over 3284224.97 frames. ], batch size: 41, lr: 8.80e-03, grad_scale: 4.0 2023-04-28 19:27:00,568 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:27:13,335 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71980.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:27:38,444 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.573e+02 3.042e+02 3.857e+02 8.032e+02, threshold=6.083e+02, percent-clipped=5.0 2023-04-28 19:27:44,223 INFO [train.py:904] (7/8) Epoch 8, batch 950, loss[loss=0.1675, simple_loss=0.2503, pruned_loss=0.04236, over 17233.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2732, pruned_loss=0.05817, over 3289083.60 frames. ], batch size: 45, lr: 8.80e-03, grad_scale: 4.0 2023-04-28 19:28:21,474 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4990, 3.2030, 3.8428, 2.5314, 3.5579, 3.8279, 3.6840, 2.2813], device='cuda:7'), covar=tensor([0.0354, 0.0186, 0.0036, 0.0270, 0.0059, 0.0065, 0.0050, 0.0322], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0064, 0.0062, 0.0117, 0.0067, 0.0076, 0.0068, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 19:28:52,163 INFO [train.py:904] (7/8) Epoch 8, batch 1000, loss[loss=0.1798, simple_loss=0.2529, pruned_loss=0.05329, over 16479.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2715, pruned_loss=0.058, over 3301923.77 frames. ], batch size: 146, lr: 8.80e-03, grad_scale: 4.0 2023-04-28 19:29:58,312 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.648e+02 3.149e+02 3.724e+02 6.885e+02, threshold=6.299e+02, percent-clipped=3.0 2023-04-28 19:30:01,628 INFO [train.py:904] (7/8) Epoch 8, batch 1050, loss[loss=0.2087, simple_loss=0.2744, pruned_loss=0.07152, over 16891.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2718, pruned_loss=0.05799, over 3304061.25 frames. ], batch size: 116, lr: 8.79e-03, grad_scale: 4.0 2023-04-28 19:30:40,225 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72129.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:31:08,739 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5316, 5.9202, 5.6434, 5.7811, 5.3545, 5.0921, 5.3833, 6.0740], device='cuda:7'), covar=tensor([0.0945, 0.0913, 0.1037, 0.0600, 0.0753, 0.0631, 0.0871, 0.0817], device='cuda:7'), in_proj_covar=tensor([0.0482, 0.0621, 0.0514, 0.0416, 0.0393, 0.0408, 0.0517, 0.0457], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:31:10,580 INFO [train.py:904] (7/8) Epoch 8, batch 1100, loss[loss=0.2063, simple_loss=0.2721, pruned_loss=0.07024, over 16733.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2713, pruned_loss=0.05761, over 3313931.14 frames. ], batch size: 134, lr: 8.79e-03, grad_scale: 4.0 2023-04-28 19:32:03,566 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72190.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:32:15,694 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.462e+02 2.908e+02 3.717e+02 7.729e+02, threshold=5.817e+02, percent-clipped=1.0 2023-04-28 19:32:18,239 INFO [train.py:904] (7/8) Epoch 8, batch 1150, loss[loss=0.2083, simple_loss=0.2911, pruned_loss=0.06272, over 16688.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2707, pruned_loss=0.05645, over 3324641.53 frames. ], batch size: 62, lr: 8.79e-03, grad_scale: 4.0 2023-04-28 19:33:15,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7386, 4.0974, 4.3970, 1.6534, 4.6209, 4.7031, 3.0837, 3.4109], device='cuda:7'), covar=tensor([0.0772, 0.0171, 0.0150, 0.1289, 0.0054, 0.0080, 0.0396, 0.0386], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0095, 0.0084, 0.0140, 0.0069, 0.0091, 0.0120, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 19:33:26,648 INFO [train.py:904] (7/8) Epoch 8, batch 1200, loss[loss=0.1951, simple_loss=0.2633, pruned_loss=0.06348, over 16912.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2695, pruned_loss=0.05617, over 3318971.57 frames. ], batch size: 109, lr: 8.79e-03, grad_scale: 8.0 2023-04-28 19:33:53,190 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:33:57,022 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72275.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:34:05,224 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 19:34:06,451 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2023-04-28 19:34:27,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1515, 4.4559, 4.6195, 1.9815, 4.8725, 4.9283, 3.2819, 3.6503], device='cuda:7'), covar=tensor([0.0631, 0.0147, 0.0198, 0.1134, 0.0058, 0.0084, 0.0355, 0.0371], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0095, 0.0085, 0.0140, 0.0070, 0.0092, 0.0121, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 19:34:30,642 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.822e+02 3.117e+02 3.864e+02 9.153e+02, threshold=6.234e+02, percent-clipped=4.0 2023-04-28 19:34:32,978 INFO [train.py:904] (7/8) Epoch 8, batch 1250, loss[loss=0.2247, simple_loss=0.2877, pruned_loss=0.08089, over 16555.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2698, pruned_loss=0.05633, over 3319234.79 frames. ], batch size: 146, lr: 8.78e-03, grad_scale: 8.0 2023-04-28 19:34:58,529 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=72320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:35:06,229 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72326.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:35:41,553 INFO [train.py:904] (7/8) Epoch 8, batch 1300, loss[loss=0.2294, simple_loss=0.301, pruned_loss=0.07896, over 16726.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2698, pruned_loss=0.05602, over 3326887.28 frames. ], batch size: 124, lr: 8.78e-03, grad_scale: 8.0 2023-04-28 19:36:31,370 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72387.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:36:49,364 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.738e+02 2.612e+02 3.072e+02 3.829e+02 7.494e+02, threshold=6.144e+02, percent-clipped=3.0 2023-04-28 19:36:52,126 INFO [train.py:904] (7/8) Epoch 8, batch 1350, loss[loss=0.1799, simple_loss=0.2585, pruned_loss=0.05067, over 16240.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2701, pruned_loss=0.05589, over 3320582.38 frames. ], batch size: 165, lr: 8.78e-03, grad_scale: 8.0 2023-04-28 19:37:08,984 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6029, 2.2480, 1.5570, 1.9875, 2.7504, 2.6273, 2.9964, 2.7607], device='cuda:7'), covar=tensor([0.0130, 0.0251, 0.0370, 0.0319, 0.0154, 0.0204, 0.0143, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0187, 0.0182, 0.0182, 0.0181, 0.0188, 0.0185, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:38:01,933 INFO [train.py:904] (7/8) Epoch 8, batch 1400, loss[loss=0.2082, simple_loss=0.2771, pruned_loss=0.06967, over 16792.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2708, pruned_loss=0.05638, over 3324663.73 frames. ], batch size: 124, lr: 8.77e-03, grad_scale: 8.0 2023-04-28 19:38:11,384 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7921, 1.4937, 2.1230, 2.6497, 2.6582, 2.5627, 1.6841, 2.8770], device='cuda:7'), covar=tensor([0.0101, 0.0317, 0.0214, 0.0156, 0.0148, 0.0154, 0.0318, 0.0063], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0157, 0.0143, 0.0143, 0.0150, 0.0104, 0.0157, 0.0097], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 19:38:17,203 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4792, 4.3010, 4.5315, 4.7386, 4.8157, 4.3673, 4.6090, 4.7826], device='cuda:7'), covar=tensor([0.1163, 0.0900, 0.1132, 0.0470, 0.0520, 0.0903, 0.1165, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0486, 0.0607, 0.0753, 0.0610, 0.0462, 0.0463, 0.0472, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:38:47,856 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72485.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:39:07,468 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.583e+02 2.954e+02 3.513e+02 6.005e+02, threshold=5.909e+02, percent-clipped=0.0 2023-04-28 19:39:11,115 INFO [train.py:904] (7/8) Epoch 8, batch 1450, loss[loss=0.2041, simple_loss=0.275, pruned_loss=0.06658, over 16727.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2697, pruned_loss=0.05625, over 3324453.30 frames. ], batch size: 83, lr: 8.77e-03, grad_scale: 8.0 2023-04-28 19:40:20,583 INFO [train.py:904] (7/8) Epoch 8, batch 1500, loss[loss=0.1622, simple_loss=0.2355, pruned_loss=0.04442, over 15878.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2693, pruned_loss=0.05639, over 3324654.78 frames. ], batch size: 35, lr: 8.77e-03, grad_scale: 8.0 2023-04-28 19:40:52,136 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72575.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:41:25,450 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.563e+02 3.021e+02 3.528e+02 6.404e+02, threshold=6.041e+02, percent-clipped=2.0 2023-04-28 19:41:28,324 INFO [train.py:904] (7/8) Epoch 8, batch 1550, loss[loss=0.2191, simple_loss=0.2807, pruned_loss=0.07878, over 16676.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2703, pruned_loss=0.05735, over 3330348.64 frames. ], batch size: 124, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:41:58,135 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=72623.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:42:25,345 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0839, 4.6428, 4.7032, 5.2479, 5.3926, 4.7902, 5.3238, 5.3177], device='cuda:7'), covar=tensor([0.1073, 0.1084, 0.2159, 0.0754, 0.0644, 0.0640, 0.0638, 0.0657], device='cuda:7'), in_proj_covar=tensor([0.0492, 0.0612, 0.0761, 0.0618, 0.0463, 0.0471, 0.0479, 0.0525], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:42:31,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8117, 3.2543, 2.6334, 4.6442, 3.9214, 4.4364, 1.5908, 3.2648], device='cuda:7'), covar=tensor([0.1380, 0.0557, 0.1105, 0.0126, 0.0310, 0.0367, 0.1525, 0.0737], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0154, 0.0174, 0.0117, 0.0196, 0.0206, 0.0172, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 19:42:34,503 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 19:42:39,327 INFO [train.py:904] (7/8) Epoch 8, batch 1600, loss[loss=0.2112, simple_loss=0.2806, pruned_loss=0.07096, over 16855.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2721, pruned_loss=0.05781, over 3335879.54 frames. ], batch size: 96, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:42:49,555 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-28 19:43:00,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4127, 4.3041, 4.8457, 4.8184, 4.8293, 4.5088, 4.5126, 4.2279], device='cuda:7'), covar=tensor([0.0300, 0.0493, 0.0306, 0.0376, 0.0359, 0.0300, 0.0683, 0.0560], device='cuda:7'), in_proj_covar=tensor([0.0302, 0.0311, 0.0306, 0.0293, 0.0349, 0.0327, 0.0427, 0.0266], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 19:43:20,458 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72682.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:43:44,413 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.814e+02 3.483e+02 4.056e+02 7.586e+02, threshold=6.967e+02, percent-clipped=5.0 2023-04-28 19:43:47,315 INFO [train.py:904] (7/8) Epoch 8, batch 1650, loss[loss=0.1757, simple_loss=0.2653, pruned_loss=0.04302, over 17130.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2741, pruned_loss=0.05972, over 3319329.83 frames. ], batch size: 48, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:44:58,252 INFO [train.py:904] (7/8) Epoch 8, batch 1700, loss[loss=0.1988, simple_loss=0.2942, pruned_loss=0.05172, over 15905.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2762, pruned_loss=0.06012, over 3314218.56 frames. ], batch size: 35, lr: 8.76e-03, grad_scale: 8.0 2023-04-28 19:45:33,808 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8510, 4.0936, 1.8698, 4.5828, 2.7823, 4.5473, 1.9842, 3.0236], device='cuda:7'), covar=tensor([0.0215, 0.0342, 0.1798, 0.0139, 0.0780, 0.0343, 0.1770, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0162, 0.0179, 0.0106, 0.0164, 0.0200, 0.0189, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 19:45:43,063 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6533, 3.5833, 3.9502, 2.7392, 3.4820, 3.8797, 3.7490, 2.2793], device='cuda:7'), covar=tensor([0.0320, 0.0150, 0.0029, 0.0236, 0.0069, 0.0053, 0.0043, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0065, 0.0062, 0.0117, 0.0068, 0.0077, 0.0069, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 19:45:44,244 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72785.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:46:05,113 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.772e+02 2.648e+02 3.116e+02 3.582e+02 6.867e+02, threshold=6.231e+02, percent-clipped=0.0 2023-04-28 19:46:07,556 INFO [train.py:904] (7/8) Epoch 8, batch 1750, loss[loss=0.2119, simple_loss=0.2817, pruned_loss=0.07105, over 16817.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2768, pruned_loss=0.06041, over 3306719.45 frames. ], batch size: 102, lr: 8.75e-03, grad_scale: 8.0 2023-04-28 19:46:50,163 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=72833.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:47:16,122 INFO [train.py:904] (7/8) Epoch 8, batch 1800, loss[loss=0.1699, simple_loss=0.2563, pruned_loss=0.04175, over 17242.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2784, pruned_loss=0.05983, over 3312690.56 frames. ], batch size: 45, lr: 8.75e-03, grad_scale: 4.0 2023-04-28 19:48:15,311 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72894.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:48:24,992 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 2.735e+02 3.300e+02 4.209e+02 8.740e+02, threshold=6.600e+02, percent-clipped=4.0 2023-04-28 19:48:26,879 INFO [train.py:904] (7/8) Epoch 8, batch 1850, loss[loss=0.1955, simple_loss=0.2952, pruned_loss=0.04794, over 17133.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2794, pruned_loss=0.06, over 3307125.31 frames. ], batch size: 48, lr: 8.75e-03, grad_scale: 4.0 2023-04-28 19:49:35,232 INFO [train.py:904] (7/8) Epoch 8, batch 1900, loss[loss=0.1921, simple_loss=0.2659, pruned_loss=0.05918, over 16896.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2785, pruned_loss=0.05889, over 3311667.45 frames. ], batch size: 109, lr: 8.74e-03, grad_scale: 4.0 2023-04-28 19:49:39,777 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72955.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:49:52,098 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72964.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 19:50:12,194 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1571, 5.0592, 4.9291, 4.3540, 4.9252, 1.7865, 4.7705, 4.8631], device='cuda:7'), covar=tensor([0.0058, 0.0045, 0.0106, 0.0299, 0.0059, 0.2025, 0.0092, 0.0133], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0104, 0.0157, 0.0150, 0.0121, 0.0169, 0.0142, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:50:16,643 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72982.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:50:41,098 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.549e+02 2.878e+02 3.342e+02 6.835e+02, threshold=5.756e+02, percent-clipped=2.0 2023-04-28 19:50:42,505 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-28 19:50:43,039 INFO [train.py:904] (7/8) Epoch 8, batch 1950, loss[loss=0.2017, simple_loss=0.279, pruned_loss=0.06213, over 16796.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2777, pruned_loss=0.0578, over 3316741.59 frames. ], batch size: 83, lr: 8.74e-03, grad_scale: 4.0 2023-04-28 19:51:14,263 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73025.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 19:51:21,295 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73030.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:51:50,205 INFO [train.py:904] (7/8) Epoch 8, batch 2000, loss[loss=0.2272, simple_loss=0.2865, pruned_loss=0.084, over 16910.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2768, pruned_loss=0.05703, over 3319354.24 frames. ], batch size: 116, lr: 8.74e-03, grad_scale: 8.0 2023-04-28 19:52:27,241 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3545, 4.3748, 4.4588, 4.3783, 4.3385, 4.9514, 4.6310, 4.3246], device='cuda:7'), covar=tensor([0.1425, 0.1685, 0.1580, 0.1864, 0.2699, 0.1025, 0.1115, 0.2126], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0460, 0.0478, 0.0407, 0.0529, 0.0507, 0.0380, 0.0539], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 19:52:58,785 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.718e+02 3.255e+02 3.787e+02 1.366e+03, threshold=6.510e+02, percent-clipped=4.0 2023-04-28 19:53:00,016 INFO [train.py:904] (7/8) Epoch 8, batch 2050, loss[loss=0.1577, simple_loss=0.2412, pruned_loss=0.03706, over 17014.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2775, pruned_loss=0.05807, over 3311852.54 frames. ], batch size: 41, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:53:34,993 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1255, 5.0973, 4.9749, 4.4066, 4.9233, 1.8656, 4.7645, 5.0413], device='cuda:7'), covar=tensor([0.0061, 0.0052, 0.0116, 0.0346, 0.0080, 0.1952, 0.0109, 0.0133], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0105, 0.0157, 0.0151, 0.0121, 0.0169, 0.0142, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:53:45,941 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1427, 3.3323, 3.5599, 3.5262, 3.5085, 3.3487, 3.3676, 3.3653], device='cuda:7'), covar=tensor([0.0355, 0.0504, 0.0406, 0.0451, 0.0455, 0.0369, 0.0707, 0.0439], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0310, 0.0311, 0.0296, 0.0350, 0.0330, 0.0432, 0.0264], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 19:53:54,676 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1765, 4.3462, 4.4857, 3.6263, 3.9278, 4.5291, 4.0828, 3.0102], device='cuda:7'), covar=tensor([0.0285, 0.0029, 0.0021, 0.0171, 0.0066, 0.0041, 0.0043, 0.0253], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0067, 0.0064, 0.0119, 0.0069, 0.0080, 0.0072, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 19:54:09,348 INFO [train.py:904] (7/8) Epoch 8, batch 2100, loss[loss=0.21, simple_loss=0.2816, pruned_loss=0.0692, over 16416.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2784, pruned_loss=0.05917, over 3302936.72 frames. ], batch size: 146, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:55:14,764 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.680e+02 3.272e+02 3.798e+02 7.634e+02, threshold=6.545e+02, percent-clipped=2.0 2023-04-28 19:55:16,489 INFO [train.py:904] (7/8) Epoch 8, batch 2150, loss[loss=0.2156, simple_loss=0.3045, pruned_loss=0.06339, over 17091.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.278, pruned_loss=0.05853, over 3307149.11 frames. ], batch size: 53, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:55:31,761 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9843, 4.3465, 3.1499, 2.3922, 3.0015, 2.4401, 4.6263, 3.9589], device='cuda:7'), covar=tensor([0.2202, 0.0606, 0.1367, 0.2019, 0.2406, 0.1639, 0.0307, 0.0907], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0255, 0.0275, 0.0262, 0.0279, 0.0213, 0.0254, 0.0286], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 19:56:08,706 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-28 19:56:21,758 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73250.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:56:23,751 INFO [train.py:904] (7/8) Epoch 8, batch 2200, loss[loss=0.2027, simple_loss=0.2927, pruned_loss=0.05632, over 16738.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.28, pruned_loss=0.06015, over 3300437.09 frames. ], batch size: 57, lr: 8.73e-03, grad_scale: 8.0 2023-04-28 19:56:47,681 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-28 19:57:20,306 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73293.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:57:29,785 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.707e+02 2.689e+02 3.130e+02 3.780e+02 7.244e+02, threshold=6.259e+02, percent-clipped=1.0 2023-04-28 19:57:31,984 INFO [train.py:904] (7/8) Epoch 8, batch 2250, loss[loss=0.1596, simple_loss=0.248, pruned_loss=0.03563, over 17195.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2798, pruned_loss=0.05951, over 3308838.57 frames. ], batch size: 44, lr: 8.72e-03, grad_scale: 8.0 2023-04-28 19:57:56,992 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73320.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 19:58:40,766 INFO [train.py:904] (7/8) Epoch 8, batch 2300, loss[loss=0.1898, simple_loss=0.2651, pruned_loss=0.05728, over 16833.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2802, pruned_loss=0.05998, over 3313258.36 frames. ], batch size: 102, lr: 8.72e-03, grad_scale: 8.0 2023-04-28 19:58:44,929 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73354.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 19:58:54,229 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-28 19:59:46,141 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-28 19:59:48,863 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.910e+02 3.420e+02 4.127e+02 1.105e+03, threshold=6.841e+02, percent-clipped=4.0 2023-04-28 19:59:49,986 INFO [train.py:904] (7/8) Epoch 8, batch 2350, loss[loss=0.2058, simple_loss=0.2767, pruned_loss=0.06743, over 16410.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2809, pruned_loss=0.06069, over 3316388.23 frames. ], batch size: 68, lr: 8.72e-03, grad_scale: 8.0 2023-04-28 20:00:58,123 INFO [train.py:904] (7/8) Epoch 8, batch 2400, loss[loss=0.2402, simple_loss=0.3025, pruned_loss=0.08896, over 16435.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2816, pruned_loss=0.06074, over 3316163.20 frames. ], batch size: 146, lr: 8.71e-03, grad_scale: 8.0 2023-04-28 20:01:02,971 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7493, 1.4510, 2.1304, 2.6053, 2.5953, 2.6317, 1.7259, 2.8330], device='cuda:7'), covar=tensor([0.0079, 0.0299, 0.0181, 0.0144, 0.0130, 0.0117, 0.0262, 0.0057], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0160, 0.0146, 0.0148, 0.0154, 0.0107, 0.0160, 0.0099], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 20:01:24,302 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-04-28 20:02:04,600 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.977e+02 2.637e+02 3.145e+02 3.743e+02 9.455e+02, threshold=6.290e+02, percent-clipped=3.0 2023-04-28 20:02:05,790 INFO [train.py:904] (7/8) Epoch 8, batch 2450, loss[loss=0.2901, simple_loss=0.3475, pruned_loss=0.1163, over 11866.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2822, pruned_loss=0.0603, over 3309134.76 frames. ], batch size: 247, lr: 8.71e-03, grad_scale: 8.0 2023-04-28 20:03:12,707 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73550.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:03:14,442 INFO [train.py:904] (7/8) Epoch 8, batch 2500, loss[loss=0.1757, simple_loss=0.2555, pruned_loss=0.04797, over 16807.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2819, pruned_loss=0.06012, over 3314039.02 frames. ], batch size: 39, lr: 8.71e-03, grad_scale: 8.0 2023-04-28 20:04:07,514 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73590.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:04:18,853 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73598.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:04:22,387 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.774e+02 3.514e+02 4.177e+02 6.912e+02, threshold=7.028e+02, percent-clipped=4.0 2023-04-28 20:04:23,442 INFO [train.py:904] (7/8) Epoch 8, batch 2550, loss[loss=0.2745, simple_loss=0.3338, pruned_loss=0.1076, over 12158.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2815, pruned_loss=0.0604, over 3316283.23 frames. ], batch size: 248, lr: 8.70e-03, grad_scale: 8.0 2023-04-28 20:04:49,570 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73620.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 20:05:29,136 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73649.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:05:32,280 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73651.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:05:32,980 INFO [train.py:904] (7/8) Epoch 8, batch 2600, loss[loss=0.1959, simple_loss=0.2742, pruned_loss=0.05883, over 16914.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2816, pruned_loss=0.05963, over 3317671.28 frames. ], batch size: 96, lr: 8.70e-03, grad_scale: 8.0 2023-04-28 20:05:55,318 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73668.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 20:06:00,372 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-28 20:06:32,415 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6249, 3.9294, 4.0895, 2.9300, 3.6487, 4.0399, 3.7151, 2.4569], device='cuda:7'), covar=tensor([0.0320, 0.0083, 0.0027, 0.0230, 0.0062, 0.0052, 0.0046, 0.0290], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0067, 0.0064, 0.0120, 0.0070, 0.0080, 0.0072, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 20:06:43,002 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.639e+02 3.290e+02 4.275e+02 7.096e+02, threshold=6.579e+02, percent-clipped=1.0 2023-04-28 20:06:43,017 INFO [train.py:904] (7/8) Epoch 8, batch 2650, loss[loss=0.1664, simple_loss=0.2597, pruned_loss=0.03656, over 16946.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2823, pruned_loss=0.0596, over 3319407.39 frames. ], batch size: 41, lr: 8.70e-03, grad_scale: 4.0 2023-04-28 20:07:23,275 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73732.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:07:50,970 INFO [train.py:904] (7/8) Epoch 8, batch 2700, loss[loss=0.185, simple_loss=0.27, pruned_loss=0.04996, over 16746.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2828, pruned_loss=0.05965, over 3306692.81 frames. ], batch size: 83, lr: 8.70e-03, grad_scale: 4.0 2023-04-28 20:07:54,293 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 20:08:11,261 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1776, 4.3332, 3.4685, 2.7080, 3.3378, 2.7584, 4.7402, 4.2808], device='cuda:7'), covar=tensor([0.2167, 0.0706, 0.1307, 0.1786, 0.2093, 0.1464, 0.0404, 0.0741], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0258, 0.0278, 0.0265, 0.0283, 0.0213, 0.0258, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 20:08:46,276 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73793.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:08:57,116 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.547e+02 3.040e+02 3.898e+02 5.528e+02, threshold=6.080e+02, percent-clipped=0.0 2023-04-28 20:08:57,137 INFO [train.py:904] (7/8) Epoch 8, batch 2750, loss[loss=0.1941, simple_loss=0.2802, pruned_loss=0.05394, over 17239.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2828, pruned_loss=0.05884, over 3312623.05 frames. ], batch size: 45, lr: 8.69e-03, grad_scale: 4.0 2023-04-28 20:09:13,373 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.59 vs. limit=5.0 2023-04-28 20:09:23,127 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-28 20:10:05,135 INFO [train.py:904] (7/8) Epoch 8, batch 2800, loss[loss=0.1741, simple_loss=0.269, pruned_loss=0.03956, over 17123.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2819, pruned_loss=0.05878, over 3310926.77 frames. ], batch size: 48, lr: 8.69e-03, grad_scale: 8.0 2023-04-28 20:11:14,614 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.850e+02 2.765e+02 3.378e+02 4.166e+02 1.030e+03, threshold=6.755e+02, percent-clipped=2.0 2023-04-28 20:11:14,636 INFO [train.py:904] (7/8) Epoch 8, batch 2850, loss[loss=0.1921, simple_loss=0.2836, pruned_loss=0.05033, over 17035.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2813, pruned_loss=0.05863, over 3305815.31 frames. ], batch size: 50, lr: 8.69e-03, grad_scale: 8.0 2023-04-28 20:11:17,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0338, 4.4419, 1.8556, 4.6894, 2.9690, 4.6468, 2.0498, 3.1978], device='cuda:7'), covar=tensor([0.0186, 0.0206, 0.1745, 0.0123, 0.0732, 0.0305, 0.1815, 0.0552], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0166, 0.0182, 0.0110, 0.0166, 0.0205, 0.0192, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 20:12:16,879 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73946.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:12:20,843 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73949.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:12:23,988 INFO [train.py:904] (7/8) Epoch 8, batch 2900, loss[loss=0.1925, simple_loss=0.2655, pruned_loss=0.0598, over 16915.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2808, pruned_loss=0.05869, over 3311477.10 frames. ], batch size: 96, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:13:13,943 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7509, 3.1927, 2.6144, 4.8041, 4.0038, 4.5144, 1.6321, 3.1573], device='cuda:7'), covar=tensor([0.1376, 0.0593, 0.1080, 0.0140, 0.0239, 0.0317, 0.1410, 0.0712], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0152, 0.0173, 0.0121, 0.0200, 0.0207, 0.0171, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 20:13:28,888 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=73997.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:13:38,232 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.787e+02 3.423e+02 3.955e+02 9.086e+02, threshold=6.847e+02, percent-clipped=3.0 2023-04-28 20:13:38,254 INFO [train.py:904] (7/8) Epoch 8, batch 2950, loss[loss=0.1743, simple_loss=0.2672, pruned_loss=0.04066, over 17097.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2802, pruned_loss=0.05965, over 3309265.54 frames. ], batch size: 53, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:14:17,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4949, 4.8258, 5.1934, 5.1751, 5.2193, 4.8656, 4.4294, 4.5274], device='cuda:7'), covar=tensor([0.0537, 0.0557, 0.0505, 0.0705, 0.0581, 0.0546, 0.1271, 0.0500], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0306, 0.0308, 0.0292, 0.0345, 0.0322, 0.0429, 0.0262], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 20:14:46,264 INFO [train.py:904] (7/8) Epoch 8, batch 3000, loss[loss=0.213, simple_loss=0.2993, pruned_loss=0.06329, over 17069.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2805, pruned_loss=0.05973, over 3308588.61 frames. ], batch size: 53, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:14:46,264 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 20:14:55,854 INFO [train.py:938] (7/8) Epoch 8, validation: loss=0.1462, simple_loss=0.2525, pruned_loss=0.01995, over 944034.00 frames. 2023-04-28 20:14:55,855 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 20:15:13,423 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74064.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:15:23,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3663, 4.6041, 4.3864, 4.4198, 4.1596, 4.1732, 4.1807, 4.6345], device='cuda:7'), covar=tensor([0.0841, 0.0794, 0.0906, 0.0601, 0.0789, 0.1144, 0.0800, 0.0819], device='cuda:7'), in_proj_covar=tensor([0.0490, 0.0628, 0.0514, 0.0419, 0.0392, 0.0402, 0.0517, 0.0463], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:15:45,926 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74088.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:16:06,423 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-28 20:16:06,764 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.715e+02 3.274e+02 3.889e+02 8.736e+02, threshold=6.548e+02, percent-clipped=1.0 2023-04-28 20:16:06,780 INFO [train.py:904] (7/8) Epoch 8, batch 3050, loss[loss=0.2127, simple_loss=0.2983, pruned_loss=0.06352, over 16625.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2799, pruned_loss=0.05889, over 3314044.15 frames. ], batch size: 62, lr: 8.68e-03, grad_scale: 8.0 2023-04-28 20:16:38,677 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74125.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:17:13,378 INFO [train.py:904] (7/8) Epoch 8, batch 3100, loss[loss=0.239, simple_loss=0.2952, pruned_loss=0.09139, over 16854.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2793, pruned_loss=0.05913, over 3313050.81 frames. ], batch size: 96, lr: 8.67e-03, grad_scale: 8.0 2023-04-28 20:18:16,377 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8822, 4.8764, 5.3931, 5.4116, 5.4098, 5.0562, 4.9795, 4.8385], device='cuda:7'), covar=tensor([0.0269, 0.0360, 0.0368, 0.0446, 0.0390, 0.0288, 0.0814, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0310, 0.0311, 0.0299, 0.0350, 0.0328, 0.0438, 0.0267], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 20:18:21,290 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 2.546e+02 2.917e+02 3.424e+02 8.256e+02, threshold=5.834e+02, percent-clipped=1.0 2023-04-28 20:18:21,306 INFO [train.py:904] (7/8) Epoch 8, batch 3150, loss[loss=0.2161, simple_loss=0.2808, pruned_loss=0.07567, over 16789.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2789, pruned_loss=0.05853, over 3315584.00 frames. ], batch size: 124, lr: 8.67e-03, grad_scale: 8.0 2023-04-28 20:18:45,883 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.93 vs. limit=5.0 2023-04-28 20:19:15,046 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74240.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:19:23,327 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74246.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:19:28,077 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-28 20:19:32,374 INFO [train.py:904] (7/8) Epoch 8, batch 3200, loss[loss=0.1783, simple_loss=0.2767, pruned_loss=0.0399, over 17154.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.279, pruned_loss=0.05882, over 3296806.57 frames. ], batch size: 48, lr: 8.67e-03, grad_scale: 8.0 2023-04-28 20:19:54,243 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3560, 5.0706, 5.2880, 5.5009, 5.6785, 5.0510, 5.5830, 5.5704], device='cuda:7'), covar=tensor([0.1210, 0.0875, 0.1296, 0.0536, 0.0394, 0.0593, 0.0429, 0.0452], device='cuda:7'), in_proj_covar=tensor([0.0503, 0.0618, 0.0770, 0.0630, 0.0470, 0.0475, 0.0489, 0.0537], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:20:31,326 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74294.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:20:40,615 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74301.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:20:41,303 INFO [train.py:904] (7/8) Epoch 8, batch 3250, loss[loss=0.2183, simple_loss=0.2936, pruned_loss=0.07145, over 16659.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2782, pruned_loss=0.05819, over 3305328.67 frames. ], batch size: 76, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:20:42,360 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.591e+02 3.147e+02 3.802e+02 9.622e+02, threshold=6.293e+02, percent-clipped=5.0 2023-04-28 20:21:30,262 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-04-28 20:21:36,452 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5272, 2.4085, 1.9336, 2.0979, 2.8900, 2.6795, 3.3571, 3.1192], device='cuda:7'), covar=tensor([0.0058, 0.0239, 0.0321, 0.0287, 0.0145, 0.0219, 0.0136, 0.0131], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0185, 0.0181, 0.0183, 0.0183, 0.0187, 0.0189, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:21:41,137 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5817, 4.5141, 4.4273, 3.9289, 4.5046, 1.8306, 4.2454, 4.2765], device='cuda:7'), covar=tensor([0.0074, 0.0063, 0.0112, 0.0275, 0.0067, 0.1831, 0.0102, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0108, 0.0160, 0.0154, 0.0124, 0.0169, 0.0145, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:21:52,450 INFO [train.py:904] (7/8) Epoch 8, batch 3300, loss[loss=0.2368, simple_loss=0.2996, pruned_loss=0.08703, over 16898.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2786, pruned_loss=0.0586, over 3306648.25 frames. ], batch size: 109, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:22:24,917 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.39 vs. limit=5.0 2023-04-28 20:22:35,055 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2451, 5.2055, 5.1250, 4.4612, 5.0923, 1.8145, 4.8294, 5.1473], device='cuda:7'), covar=tensor([0.0061, 0.0058, 0.0110, 0.0342, 0.0068, 0.2010, 0.0111, 0.0118], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0108, 0.0160, 0.0155, 0.0125, 0.0169, 0.0145, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:22:41,338 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74388.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:23:01,501 INFO [train.py:904] (7/8) Epoch 8, batch 3350, loss[loss=0.2158, simple_loss=0.2797, pruned_loss=0.07597, over 16743.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2795, pruned_loss=0.05869, over 3317217.98 frames. ], batch size: 134, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:23:02,763 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.835e+02 2.621e+02 3.247e+02 4.157e+02 8.305e+02, threshold=6.494e+02, percent-clipped=1.0 2023-04-28 20:23:27,944 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74420.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:23:49,804 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74436.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:24:11,182 INFO [train.py:904] (7/8) Epoch 8, batch 3400, loss[loss=0.2301, simple_loss=0.3044, pruned_loss=0.07788, over 15522.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.279, pruned_loss=0.05847, over 3314089.97 frames. ], batch size: 191, lr: 8.66e-03, grad_scale: 4.0 2023-04-28 20:24:30,029 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7770, 6.1680, 5.8790, 5.9978, 5.4353, 5.2556, 5.7778, 6.2031], device='cuda:7'), covar=tensor([0.0955, 0.0733, 0.0971, 0.0561, 0.0725, 0.0598, 0.0603, 0.0768], device='cuda:7'), in_proj_covar=tensor([0.0489, 0.0623, 0.0514, 0.0421, 0.0391, 0.0407, 0.0517, 0.0462], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:25:21,720 INFO [train.py:904] (7/8) Epoch 8, batch 3450, loss[loss=0.1978, simple_loss=0.2905, pruned_loss=0.05255, over 16730.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2785, pruned_loss=0.0584, over 3306455.52 frames. ], batch size: 57, lr: 8.65e-03, grad_scale: 4.0 2023-04-28 20:25:22,839 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.611e+02 3.016e+02 3.656e+02 6.729e+02, threshold=6.032e+02, percent-clipped=3.0 2023-04-28 20:26:30,770 INFO [train.py:904] (7/8) Epoch 8, batch 3500, loss[loss=0.2104, simple_loss=0.2864, pruned_loss=0.06718, over 16738.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2766, pruned_loss=0.0572, over 3310307.58 frames. ], batch size: 124, lr: 8.65e-03, grad_scale: 4.0 2023-04-28 20:27:20,489 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-28 20:27:32,541 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74595.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:27:34,455 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74596.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:27:42,866 INFO [train.py:904] (7/8) Epoch 8, batch 3550, loss[loss=0.1958, simple_loss=0.2711, pruned_loss=0.06022, over 16704.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2748, pruned_loss=0.05621, over 3318374.04 frames. ], batch size: 89, lr: 8.65e-03, grad_scale: 4.0 2023-04-28 20:27:43,961 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.810e+02 2.461e+02 3.024e+02 3.861e+02 7.667e+02, threshold=6.049e+02, percent-clipped=4.0 2023-04-28 20:27:54,363 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-28 20:28:07,947 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7471, 3.9471, 3.0272, 2.3957, 2.7022, 2.3547, 3.9447, 3.6518], device='cuda:7'), covar=tensor([0.2157, 0.0525, 0.1281, 0.1985, 0.2212, 0.1598, 0.0428, 0.0908], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0257, 0.0278, 0.0265, 0.0285, 0.0212, 0.0258, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:28:51,896 INFO [train.py:904] (7/8) Epoch 8, batch 3600, loss[loss=0.1998, simple_loss=0.2798, pruned_loss=0.05991, over 16563.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2732, pruned_loss=0.05566, over 3320526.62 frames. ], batch size: 68, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:28:56,992 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74656.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:29:22,004 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-28 20:30:00,914 INFO [train.py:904] (7/8) Epoch 8, batch 3650, loss[loss=0.2198, simple_loss=0.2882, pruned_loss=0.07568, over 11346.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.273, pruned_loss=0.057, over 3290600.51 frames. ], batch size: 247, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:30:02,113 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.393e+02 2.913e+02 3.939e+02 7.282e+02, threshold=5.826e+02, percent-clipped=2.0 2023-04-28 20:30:27,893 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74720.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:30:40,548 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7080, 2.4018, 1.6967, 2.1017, 2.8488, 2.6236, 3.0075, 2.8739], device='cuda:7'), covar=tensor([0.0086, 0.0219, 0.0317, 0.0302, 0.0108, 0.0197, 0.0125, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0185, 0.0180, 0.0182, 0.0182, 0.0186, 0.0189, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:31:13,895 INFO [train.py:904] (7/8) Epoch 8, batch 3700, loss[loss=0.1822, simple_loss=0.262, pruned_loss=0.05117, over 16516.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2716, pruned_loss=0.05859, over 3266183.18 frames. ], batch size: 62, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:31:38,766 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74768.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:32:29,734 INFO [train.py:904] (7/8) Epoch 8, batch 3750, loss[loss=0.1943, simple_loss=0.2572, pruned_loss=0.06573, over 16772.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2735, pruned_loss=0.06098, over 3256808.11 frames. ], batch size: 102, lr: 8.64e-03, grad_scale: 8.0 2023-04-28 20:32:30,689 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.639e+02 3.130e+02 3.703e+02 1.006e+03, threshold=6.260e+02, percent-clipped=3.0 2023-04-28 20:33:41,200 INFO [train.py:904] (7/8) Epoch 8, batch 3800, loss[loss=0.2111, simple_loss=0.286, pruned_loss=0.0681, over 16986.00 frames. ], tot_loss[loss=0.2, simple_loss=0.275, pruned_loss=0.06248, over 3263784.72 frames. ], batch size: 53, lr: 8.63e-03, grad_scale: 8.0 2023-04-28 20:34:20,762 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9652, 1.5991, 2.3585, 2.7621, 2.7238, 2.6794, 1.7782, 2.9011], device='cuda:7'), covar=tensor([0.0091, 0.0284, 0.0183, 0.0152, 0.0137, 0.0123, 0.0293, 0.0061], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0160, 0.0145, 0.0148, 0.0154, 0.0110, 0.0159, 0.0101], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 20:34:45,383 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74896.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:34:49,503 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-28 20:34:51,951 INFO [train.py:904] (7/8) Epoch 8, batch 3850, loss[loss=0.2058, simple_loss=0.2657, pruned_loss=0.07291, over 16877.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2746, pruned_loss=0.06285, over 3266117.61 frames. ], batch size: 90, lr: 8.63e-03, grad_scale: 8.0 2023-04-28 20:34:53,141 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.503e+02 3.025e+02 3.649e+02 5.657e+02, threshold=6.049e+02, percent-clipped=0.0 2023-04-28 20:35:03,515 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74909.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:35:52,801 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=74944.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:36:01,480 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74951.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:36:02,295 INFO [train.py:904] (7/8) Epoch 8, batch 3900, loss[loss=0.189, simple_loss=0.266, pruned_loss=0.05594, over 16663.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2737, pruned_loss=0.06293, over 3271032.72 frames. ], batch size: 62, lr: 8.63e-03, grad_scale: 8.0 2023-04-28 20:36:13,756 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8175, 1.5104, 2.2044, 2.6058, 2.6616, 2.4573, 1.6852, 2.8070], device='cuda:7'), covar=tensor([0.0090, 0.0293, 0.0193, 0.0169, 0.0137, 0.0174, 0.0305, 0.0069], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0160, 0.0144, 0.0148, 0.0154, 0.0110, 0.0159, 0.0100], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 20:36:20,395 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74964.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:36:21,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6693, 3.7939, 3.9790, 1.9766, 4.1652, 4.0993, 3.2511, 3.0597], device='cuda:7'), covar=tensor([0.0705, 0.0131, 0.0115, 0.1064, 0.0045, 0.0097, 0.0298, 0.0401], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0098, 0.0085, 0.0138, 0.0070, 0.0093, 0.0120, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 20:36:28,141 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74970.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:37:12,349 INFO [train.py:904] (7/8) Epoch 8, batch 3950, loss[loss=0.2567, simple_loss=0.3203, pruned_loss=0.09651, over 12441.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2728, pruned_loss=0.06343, over 3267587.69 frames. ], batch size: 246, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:37:14,095 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.648e+02 3.131e+02 3.735e+02 8.073e+02, threshold=6.262e+02, percent-clipped=3.0 2023-04-28 20:37:46,173 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75025.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:38:23,633 INFO [train.py:904] (7/8) Epoch 8, batch 4000, loss[loss=0.2064, simple_loss=0.287, pruned_loss=0.0629, over 15613.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2727, pruned_loss=0.06342, over 3276141.09 frames. ], batch size: 190, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:39:10,255 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.01 vs. limit=5.0 2023-04-28 20:39:35,123 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0753, 4.9658, 5.1053, 5.3123, 5.4753, 4.8655, 5.4330, 5.4518], device='cuda:7'), covar=tensor([0.1123, 0.0828, 0.1287, 0.0444, 0.0362, 0.0523, 0.0372, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0483, 0.0591, 0.0737, 0.0602, 0.0454, 0.0453, 0.0463, 0.0516], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:39:36,991 INFO [train.py:904] (7/8) Epoch 8, batch 4050, loss[loss=0.1822, simple_loss=0.2625, pruned_loss=0.05093, over 16690.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2723, pruned_loss=0.06159, over 3277454.96 frames. ], batch size: 76, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:39:38,165 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.447e+02 2.780e+02 3.424e+02 6.417e+02, threshold=5.561e+02, percent-clipped=2.0 2023-04-28 20:40:48,400 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0468, 4.9657, 4.7867, 4.1500, 4.9701, 1.7972, 4.6789, 4.6671], device='cuda:7'), covar=tensor([0.0057, 0.0054, 0.0098, 0.0305, 0.0049, 0.2159, 0.0092, 0.0134], device='cuda:7'), in_proj_covar=tensor([0.0118, 0.0106, 0.0156, 0.0152, 0.0122, 0.0165, 0.0142, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:40:49,065 INFO [train.py:904] (7/8) Epoch 8, batch 4100, loss[loss=0.1898, simple_loss=0.2696, pruned_loss=0.055, over 16597.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2736, pruned_loss=0.0609, over 3264009.18 frames. ], batch size: 68, lr: 8.62e-03, grad_scale: 8.0 2023-04-28 20:40:55,401 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-28 20:42:02,417 INFO [train.py:904] (7/8) Epoch 8, batch 4150, loss[loss=0.2172, simple_loss=0.306, pruned_loss=0.06423, over 16286.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2808, pruned_loss=0.06363, over 3224606.38 frames. ], batch size: 165, lr: 8.61e-03, grad_scale: 8.0 2023-04-28 20:42:04,253 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.780e+02 2.372e+02 3.014e+02 3.829e+02 9.608e+02, threshold=6.029e+02, percent-clipped=8.0 2023-04-28 20:42:43,272 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75228.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:19,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75251.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:20,235 INFO [train.py:904] (7/8) Epoch 8, batch 4200, loss[loss=0.2353, simple_loss=0.3204, pruned_loss=0.07512, over 16706.00 frames. ], tot_loss[loss=0.209, simple_loss=0.288, pruned_loss=0.06502, over 3212953.68 frames. ], batch size: 124, lr: 8.61e-03, grad_scale: 8.0 2023-04-28 20:43:21,321 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8566, 4.8026, 4.7081, 4.5247, 4.3177, 4.7769, 4.7106, 4.4576], device='cuda:7'), covar=tensor([0.0416, 0.0256, 0.0194, 0.0180, 0.0762, 0.0239, 0.0215, 0.0478], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0259, 0.0258, 0.0231, 0.0288, 0.0262, 0.0176, 0.0293], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:43:41,462 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75265.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:43:48,076 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75270.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:44:16,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75289.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:44:27,275 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-28 20:44:29,252 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6201, 4.4572, 4.6055, 4.8130, 4.9153, 4.4128, 4.8973, 4.9089], device='cuda:7'), covar=tensor([0.1063, 0.0887, 0.1185, 0.0506, 0.0381, 0.0841, 0.0503, 0.0417], device='cuda:7'), in_proj_covar=tensor([0.0466, 0.0569, 0.0710, 0.0580, 0.0438, 0.0442, 0.0447, 0.0498], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:44:30,800 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75299.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:44:34,943 INFO [train.py:904] (7/8) Epoch 8, batch 4250, loss[loss=0.1732, simple_loss=0.2648, pruned_loss=0.04082, over 15551.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2918, pruned_loss=0.06562, over 3195125.68 frames. ], batch size: 191, lr: 8.61e-03, grad_scale: 8.0 2023-04-28 20:44:36,197 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.663e+02 3.246e+02 3.785e+02 9.237e+02, threshold=6.492e+02, percent-clipped=4.0 2023-04-28 20:45:01,380 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:45:18,510 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75331.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 20:45:48,481 INFO [train.py:904] (7/8) Epoch 8, batch 4300, loss[loss=0.2163, simple_loss=0.3055, pruned_loss=0.06353, over 16478.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2935, pruned_loss=0.06515, over 3176649.62 frames. ], batch size: 75, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:47:02,658 INFO [train.py:904] (7/8) Epoch 8, batch 4350, loss[loss=0.2561, simple_loss=0.3205, pruned_loss=0.09587, over 11734.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2973, pruned_loss=0.06641, over 3178889.55 frames. ], batch size: 248, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:47:03,853 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.615e+02 3.065e+02 3.856e+02 8.729e+02, threshold=6.129e+02, percent-clipped=2.0 2023-04-28 20:48:16,622 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1633, 1.5212, 1.7793, 2.0647, 2.1155, 2.3672, 1.5776, 2.2046], device='cuda:7'), covar=tensor([0.0121, 0.0231, 0.0135, 0.0159, 0.0136, 0.0080, 0.0238, 0.0055], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0159, 0.0143, 0.0148, 0.0153, 0.0108, 0.0160, 0.0100], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 20:48:17,344 INFO [train.py:904] (7/8) Epoch 8, batch 4400, loss[loss=0.234, simple_loss=0.3122, pruned_loss=0.07786, over 15411.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.2998, pruned_loss=0.0675, over 3184011.13 frames. ], batch size: 191, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:48:38,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7423, 5.0661, 4.7988, 4.8893, 4.5062, 4.4507, 4.5266, 5.1364], device='cuda:7'), covar=tensor([0.0864, 0.0693, 0.0933, 0.0489, 0.0741, 0.0815, 0.0839, 0.0724], device='cuda:7'), in_proj_covar=tensor([0.0476, 0.0590, 0.0500, 0.0404, 0.0377, 0.0392, 0.0496, 0.0439], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:48:38,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1017, 3.6934, 3.6622, 2.3615, 3.4277, 3.6904, 3.5479, 1.8290], device='cuda:7'), covar=tensor([0.0383, 0.0023, 0.0025, 0.0284, 0.0045, 0.0045, 0.0033, 0.0340], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0064, 0.0063, 0.0119, 0.0069, 0.0077, 0.0070, 0.0112], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 20:48:38,952 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1789, 1.9740, 2.0343, 3.8438, 1.7786, 2.5257, 2.1317, 2.1477], device='cuda:7'), covar=tensor([0.0774, 0.2759, 0.1817, 0.0332, 0.3445, 0.1724, 0.2327, 0.2766], device='cuda:7'), in_proj_covar=tensor([0.0348, 0.0371, 0.0309, 0.0326, 0.0397, 0.0416, 0.0331, 0.0440], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:49:26,874 INFO [train.py:904] (7/8) Epoch 8, batch 4450, loss[loss=0.2086, simple_loss=0.2996, pruned_loss=0.05878, over 16688.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3021, pruned_loss=0.06768, over 3203923.13 frames. ], batch size: 57, lr: 8.60e-03, grad_scale: 8.0 2023-04-28 20:49:28,913 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.422e+02 2.913e+02 3.508e+02 6.103e+02, threshold=5.826e+02, percent-clipped=0.0 2023-04-28 20:49:54,305 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7374, 4.5432, 4.5781, 3.1677, 3.9853, 4.5301, 4.0800, 2.5631], device='cuda:7'), covar=tensor([0.0325, 0.0014, 0.0021, 0.0215, 0.0046, 0.0040, 0.0029, 0.0265], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0064, 0.0063, 0.0120, 0.0069, 0.0078, 0.0070, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 20:50:34,406 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6625, 1.9624, 1.4596, 1.8690, 2.4771, 2.1081, 2.6627, 2.7281], device='cuda:7'), covar=tensor([0.0075, 0.0258, 0.0413, 0.0290, 0.0144, 0.0254, 0.0132, 0.0133], device='cuda:7'), in_proj_covar=tensor([0.0114, 0.0180, 0.0179, 0.0180, 0.0179, 0.0183, 0.0179, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:50:37,433 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1075, 4.8069, 5.0516, 5.2325, 5.3628, 4.7182, 5.3521, 5.3542], device='cuda:7'), covar=tensor([0.1013, 0.0863, 0.1038, 0.0404, 0.0335, 0.0597, 0.0356, 0.0341], device='cuda:7'), in_proj_covar=tensor([0.0460, 0.0570, 0.0707, 0.0578, 0.0435, 0.0440, 0.0445, 0.0490], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:50:38,184 INFO [train.py:904] (7/8) Epoch 8, batch 4500, loss[loss=0.2019, simple_loss=0.2899, pruned_loss=0.05701, over 17254.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.3018, pruned_loss=0.06786, over 3221805.09 frames. ], batch size: 52, lr: 8.59e-03, grad_scale: 8.0 2023-04-28 20:50:57,714 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75565.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:51:14,213 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8309, 1.6588, 2.1783, 2.7570, 2.6709, 3.0839, 1.8983, 2.9695], device='cuda:7'), covar=tensor([0.0091, 0.0286, 0.0168, 0.0153, 0.0138, 0.0075, 0.0273, 0.0053], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0160, 0.0143, 0.0148, 0.0154, 0.0108, 0.0159, 0.0099], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 20:51:24,897 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75584.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:51:51,103 INFO [train.py:904] (7/8) Epoch 8, batch 4550, loss[loss=0.2411, simple_loss=0.3199, pruned_loss=0.08118, over 16210.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.3023, pruned_loss=0.06834, over 3215990.00 frames. ], batch size: 165, lr: 8.59e-03, grad_scale: 8.0 2023-04-28 20:51:52,275 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.189e+02 2.672e+02 3.111e+02 5.807e+02, threshold=5.345e+02, percent-clipped=0.0 2023-04-28 20:52:06,457 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75613.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:52:17,209 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75620.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:52:25,172 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75626.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 20:53:02,634 INFO [train.py:904] (7/8) Epoch 8, batch 4600, loss[loss=0.2148, simple_loss=0.3007, pruned_loss=0.06442, over 17134.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.3029, pruned_loss=0.0685, over 3213299.77 frames. ], batch size: 48, lr: 8.59e-03, grad_scale: 8.0 2023-04-28 20:53:26,300 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75668.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:54:12,111 INFO [train.py:904] (7/8) Epoch 8, batch 4650, loss[loss=0.2008, simple_loss=0.2873, pruned_loss=0.05709, over 16714.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.3024, pruned_loss=0.06891, over 3222314.41 frames. ], batch size: 62, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:54:13,263 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.294e+02 2.810e+02 3.370e+02 9.861e+02, threshold=5.619e+02, percent-clipped=3.0 2023-04-28 20:55:23,492 INFO [train.py:904] (7/8) Epoch 8, batch 4700, loss[loss=0.2114, simple_loss=0.3087, pruned_loss=0.05708, over 16849.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3001, pruned_loss=0.06767, over 3220883.25 frames. ], batch size: 102, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:55:57,978 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7586, 3.5548, 3.7942, 3.5998, 3.7274, 4.1646, 3.9160, 3.5448], device='cuda:7'), covar=tensor([0.1980, 0.2184, 0.1625, 0.2271, 0.2645, 0.1660, 0.1282, 0.2539], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0429, 0.0444, 0.0377, 0.0502, 0.0474, 0.0363, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 20:56:10,381 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9076, 2.6781, 2.0516, 2.4079, 3.1276, 2.7398, 3.5966, 3.4884], device='cuda:7'), covar=tensor([0.0026, 0.0216, 0.0320, 0.0258, 0.0122, 0.0216, 0.0089, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0178, 0.0177, 0.0178, 0.0177, 0.0181, 0.0177, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:56:31,946 INFO [train.py:904] (7/8) Epoch 8, batch 4750, loss[loss=0.197, simple_loss=0.281, pruned_loss=0.05652, over 16700.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2964, pruned_loss=0.06614, over 3206718.70 frames. ], batch size: 89, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:56:33,074 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 2.114e+02 2.531e+02 3.127e+02 7.196e+02, threshold=5.061e+02, percent-clipped=1.0 2023-04-28 20:56:58,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-28 20:57:32,608 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7245, 2.2067, 2.3498, 4.4294, 2.0428, 2.8346, 2.2866, 2.4725], device='cuda:7'), covar=tensor([0.0686, 0.2839, 0.1705, 0.0297, 0.3402, 0.1736, 0.2487, 0.2625], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0369, 0.0307, 0.0322, 0.0399, 0.0414, 0.0328, 0.0433], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 20:57:44,155 INFO [train.py:904] (7/8) Epoch 8, batch 4800, loss[loss=0.2153, simple_loss=0.3035, pruned_loss=0.06356, over 16366.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2928, pruned_loss=0.06402, over 3210199.14 frames. ], batch size: 165, lr: 8.58e-03, grad_scale: 8.0 2023-04-28 20:58:32,107 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75884.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 20:58:58,555 INFO [train.py:904] (7/8) Epoch 8, batch 4850, loss[loss=0.2159, simple_loss=0.3048, pruned_loss=0.06352, over 15459.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2929, pruned_loss=0.06333, over 3181234.27 frames. ], batch size: 190, lr: 8.57e-03, grad_scale: 8.0 2023-04-28 20:59:01,504 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.327e+02 2.698e+02 3.138e+02 6.949e+02, threshold=5.395e+02, percent-clipped=1.0 2023-04-28 20:59:22,727 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6358, 2.4783, 2.2886, 3.3762, 2.3739, 3.7038, 1.4795, 2.7711], device='cuda:7'), covar=tensor([0.1311, 0.0603, 0.1085, 0.0102, 0.0102, 0.0298, 0.1454, 0.0707], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0151, 0.0173, 0.0117, 0.0200, 0.0202, 0.0171, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 20:59:36,924 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75926.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 20:59:46,581 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75932.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:00:17,707 INFO [train.py:904] (7/8) Epoch 8, batch 4900, loss[loss=0.1812, simple_loss=0.264, pruned_loss=0.04916, over 17214.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2924, pruned_loss=0.06235, over 3171955.86 frames. ], batch size: 45, lr: 8.57e-03, grad_scale: 8.0 2023-04-28 21:00:49,947 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=75974.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:01:08,314 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2389, 4.2910, 4.0551, 3.9090, 3.7376, 4.1617, 3.9134, 3.8776], device='cuda:7'), covar=tensor([0.0474, 0.0317, 0.0237, 0.0207, 0.0887, 0.0340, 0.0488, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0213, 0.0250, 0.0250, 0.0222, 0.0279, 0.0251, 0.0169, 0.0282], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:01:34,539 INFO [train.py:904] (7/8) Epoch 8, batch 4950, loss[loss=0.2295, simple_loss=0.3143, pruned_loss=0.0724, over 16308.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.292, pruned_loss=0.06175, over 3183845.60 frames. ], batch size: 165, lr: 8.57e-03, grad_scale: 8.0 2023-04-28 21:01:36,823 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.340e+02 2.828e+02 3.481e+02 8.052e+02, threshold=5.656e+02, percent-clipped=2.0 2023-04-28 21:01:48,855 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7115, 2.0780, 1.6373, 1.8453, 2.4921, 2.2604, 2.6064, 2.6990], device='cuda:7'), covar=tensor([0.0071, 0.0293, 0.0400, 0.0356, 0.0162, 0.0272, 0.0130, 0.0160], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0178, 0.0176, 0.0177, 0.0177, 0.0181, 0.0175, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:02:18,124 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-04-28 21:02:45,336 INFO [train.py:904] (7/8) Epoch 8, batch 5000, loss[loss=0.2004, simple_loss=0.2943, pruned_loss=0.05324, over 16508.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2929, pruned_loss=0.06133, over 3205479.60 frames. ], batch size: 75, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:03:46,438 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5229, 2.6889, 2.4947, 4.2534, 3.2780, 4.0248, 1.4746, 3.0267], device='cuda:7'), covar=tensor([0.1384, 0.0691, 0.1103, 0.0119, 0.0214, 0.0335, 0.1477, 0.0716], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0154, 0.0177, 0.0120, 0.0205, 0.0206, 0.0175, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 21:03:50,914 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4949, 4.0751, 3.8461, 2.2052, 3.1220, 2.6761, 3.9449, 3.8610], device='cuda:7'), covar=tensor([0.0221, 0.0439, 0.0490, 0.1514, 0.0665, 0.0761, 0.0558, 0.0766], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0135, 0.0155, 0.0141, 0.0133, 0.0123, 0.0136, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 21:03:55,758 INFO [train.py:904] (7/8) Epoch 8, batch 5050, loss[loss=0.2104, simple_loss=0.2971, pruned_loss=0.06184, over 16563.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2931, pruned_loss=0.06079, over 3221998.37 frames. ], batch size: 75, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:03:57,928 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.576e+02 2.969e+02 3.606e+02 8.836e+02, threshold=5.938e+02, percent-clipped=5.0 2023-04-28 21:04:58,397 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-28 21:05:07,318 INFO [train.py:904] (7/8) Epoch 8, batch 5100, loss[loss=0.1989, simple_loss=0.2838, pruned_loss=0.05701, over 16819.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2912, pruned_loss=0.05991, over 3224459.52 frames. ], batch size: 39, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:05:13,715 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8751, 2.1610, 1.7967, 2.0759, 2.5917, 2.3107, 2.7457, 2.8098], device='cuda:7'), covar=tensor([0.0056, 0.0276, 0.0330, 0.0290, 0.0147, 0.0253, 0.0089, 0.0150], device='cuda:7'), in_proj_covar=tensor([0.0110, 0.0178, 0.0177, 0.0176, 0.0176, 0.0180, 0.0174, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:06:06,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6229, 3.1894, 3.0144, 1.7553, 2.6999, 2.2044, 3.2269, 3.2232], device='cuda:7'), covar=tensor([0.0248, 0.0489, 0.0569, 0.1623, 0.0688, 0.0890, 0.0559, 0.0603], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0134, 0.0154, 0.0140, 0.0132, 0.0122, 0.0134, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 21:06:15,385 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2345, 4.4226, 4.6867, 2.3280, 5.0326, 5.0023, 3.6287, 3.8420], device='cuda:7'), covar=tensor([0.0616, 0.0140, 0.0101, 0.1031, 0.0026, 0.0040, 0.0241, 0.0339], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0082, 0.0138, 0.0068, 0.0089, 0.0118, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 21:06:20,881 INFO [train.py:904] (7/8) Epoch 8, batch 5150, loss[loss=0.2336, simple_loss=0.3281, pruned_loss=0.06954, over 16867.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2918, pruned_loss=0.05915, over 3224756.78 frames. ], batch size: 116, lr: 8.56e-03, grad_scale: 8.0 2023-04-28 21:06:24,099 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.361e+02 2.714e+02 3.177e+02 5.443e+02, threshold=5.429e+02, percent-clipped=0.0 2023-04-28 21:06:33,904 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6728, 3.6367, 4.0070, 4.0217, 3.9635, 3.7342, 3.7471, 3.7426], device='cuda:7'), covar=tensor([0.0266, 0.0532, 0.0371, 0.0377, 0.0457, 0.0334, 0.0781, 0.0413], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0289, 0.0289, 0.0277, 0.0331, 0.0305, 0.0407, 0.0251], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 21:06:58,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3057, 4.1194, 4.3327, 4.5129, 4.6149, 4.2364, 4.5631, 4.6046], device='cuda:7'), covar=tensor([0.1073, 0.0809, 0.1184, 0.0486, 0.0392, 0.0929, 0.0486, 0.0412], device='cuda:7'), in_proj_covar=tensor([0.0462, 0.0562, 0.0705, 0.0577, 0.0435, 0.0434, 0.0442, 0.0495], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:07:18,245 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1950, 4.0508, 4.1979, 4.3890, 4.5036, 4.1032, 4.4358, 4.5000], device='cuda:7'), covar=tensor([0.1162, 0.0797, 0.1283, 0.0531, 0.0424, 0.0949, 0.0553, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0464, 0.0563, 0.0708, 0.0579, 0.0436, 0.0434, 0.0443, 0.0496], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:07:33,713 INFO [train.py:904] (7/8) Epoch 8, batch 5200, loss[loss=0.1816, simple_loss=0.2629, pruned_loss=0.05011, over 16609.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2908, pruned_loss=0.05913, over 3225886.70 frames. ], batch size: 75, lr: 8.55e-03, grad_scale: 8.0 2023-04-28 21:07:47,123 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-28 21:07:59,835 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76269.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:08:36,430 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76294.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:08:48,739 INFO [train.py:904] (7/8) Epoch 8, batch 5250, loss[loss=0.2108, simple_loss=0.2941, pruned_loss=0.06381, over 16595.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2883, pruned_loss=0.05862, over 3226784.06 frames. ], batch size: 62, lr: 8.55e-03, grad_scale: 8.0 2023-04-28 21:08:51,149 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.354e+02 2.757e+02 3.436e+02 6.643e+02, threshold=5.515e+02, percent-clipped=1.0 2023-04-28 21:09:19,337 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76323.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 21:09:30,983 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76330.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:09:36,561 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-04-28 21:09:49,726 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6448, 2.8696, 2.6045, 4.1264, 3.0827, 4.0705, 1.4966, 3.0777], device='cuda:7'), covar=tensor([0.1321, 0.0598, 0.0974, 0.0094, 0.0253, 0.0279, 0.1492, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0153, 0.0175, 0.0118, 0.0202, 0.0205, 0.0173, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 21:10:01,834 INFO [train.py:904] (7/8) Epoch 8, batch 5300, loss[loss=0.183, simple_loss=0.266, pruned_loss=0.05004, over 16904.00 frames. ], tot_loss[loss=0.199, simple_loss=0.284, pruned_loss=0.05701, over 3226558.85 frames. ], batch size: 116, lr: 8.55e-03, grad_scale: 8.0 2023-04-28 21:10:06,502 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76355.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:10:34,039 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0331, 5.0028, 4.8353, 4.1873, 4.8966, 1.8660, 4.6626, 4.7290], device='cuda:7'), covar=tensor([0.0058, 0.0052, 0.0098, 0.0339, 0.0065, 0.1950, 0.0095, 0.0123], device='cuda:7'), in_proj_covar=tensor([0.0111, 0.0098, 0.0147, 0.0144, 0.0114, 0.0160, 0.0131, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:10:48,008 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76384.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:11:13,485 INFO [train.py:904] (7/8) Epoch 8, batch 5350, loss[loss=0.1985, simple_loss=0.2888, pruned_loss=0.05409, over 16493.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2824, pruned_loss=0.05594, over 3227593.59 frames. ], batch size: 68, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:11:15,925 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.303e+02 2.727e+02 3.252e+02 5.747e+02, threshold=5.455e+02, percent-clipped=1.0 2023-04-28 21:12:26,520 INFO [train.py:904] (7/8) Epoch 8, batch 5400, loss[loss=0.224, simple_loss=0.3149, pruned_loss=0.06656, over 16620.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2854, pruned_loss=0.0571, over 3215467.65 frames. ], batch size: 134, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:13:43,644 INFO [train.py:904] (7/8) Epoch 8, batch 5450, loss[loss=0.2333, simple_loss=0.3073, pruned_loss=0.0796, over 12112.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2894, pruned_loss=0.05951, over 3198914.30 frames. ], batch size: 246, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:13:46,710 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.643e+02 3.277e+02 3.879e+02 8.643e+02, threshold=6.553e+02, percent-clipped=9.0 2023-04-28 21:15:01,748 INFO [train.py:904] (7/8) Epoch 8, batch 5500, loss[loss=0.2723, simple_loss=0.3394, pruned_loss=0.1026, over 15319.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2974, pruned_loss=0.06509, over 3175580.95 frames. ], batch size: 191, lr: 8.54e-03, grad_scale: 8.0 2023-04-28 21:16:22,250 INFO [train.py:904] (7/8) Epoch 8, batch 5550, loss[loss=0.2659, simple_loss=0.3409, pruned_loss=0.09545, over 15287.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.3048, pruned_loss=0.07054, over 3159961.68 frames. ], batch size: 191, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:16:26,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.250e+02 3.650e+02 4.259e+02 5.295e+02 1.196e+03, threshold=8.517e+02, percent-clipped=11.0 2023-04-28 21:17:00,709 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76625.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 21:17:41,011 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76650.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:17:43,811 INFO [train.py:904] (7/8) Epoch 8, batch 5600, loss[loss=0.2128, simple_loss=0.2952, pruned_loss=0.06516, over 16561.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3104, pruned_loss=0.07583, over 3121128.46 frames. ], batch size: 68, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:18:28,912 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76679.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:19:06,832 INFO [train.py:904] (7/8) Epoch 8, batch 5650, loss[loss=0.2654, simple_loss=0.3358, pruned_loss=0.09746, over 16677.00 frames. ], tot_loss[loss=0.2379, simple_loss=0.3158, pruned_loss=0.07999, over 3113807.92 frames. ], batch size: 134, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:19:10,212 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.391e+02 3.706e+02 4.487e+02 5.532e+02 1.181e+03, threshold=8.975e+02, percent-clipped=2.0 2023-04-28 21:19:12,700 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0424, 5.3595, 4.8594, 5.2641, 4.8608, 4.6788, 5.1445, 5.4012], device='cuda:7'), covar=tensor([0.1596, 0.1274, 0.2015, 0.0949, 0.1236, 0.1197, 0.1347, 0.1583], device='cuda:7'), in_proj_covar=tensor([0.0461, 0.0578, 0.0492, 0.0397, 0.0370, 0.0380, 0.0483, 0.0427], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:20:01,395 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9656, 4.7398, 4.9307, 5.1881, 5.2958, 4.6957, 5.2776, 5.2703], device='cuda:7'), covar=tensor([0.1247, 0.0920, 0.1334, 0.0487, 0.0454, 0.0614, 0.0533, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0460, 0.0560, 0.0693, 0.0568, 0.0431, 0.0429, 0.0444, 0.0492], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:20:04,703 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-04-28 21:20:27,963 INFO [train.py:904] (7/8) Epoch 8, batch 5700, loss[loss=0.3461, simple_loss=0.3761, pruned_loss=0.158, over 11120.00 frames. ], tot_loss[loss=0.2417, simple_loss=0.3179, pruned_loss=0.08271, over 3077195.71 frames. ], batch size: 248, lr: 8.53e-03, grad_scale: 8.0 2023-04-28 21:21:49,370 INFO [train.py:904] (7/8) Epoch 8, batch 5750, loss[loss=0.2324, simple_loss=0.3132, pruned_loss=0.07581, over 16432.00 frames. ], tot_loss[loss=0.2452, simple_loss=0.3208, pruned_loss=0.08476, over 3047878.59 frames. ], batch size: 146, lr: 8.52e-03, grad_scale: 4.0 2023-04-28 21:21:54,078 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.644e+02 3.742e+02 4.943e+02 6.173e+02 1.099e+03, threshold=9.887e+02, percent-clipped=1.0 2023-04-28 21:22:52,094 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76839.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:23:12,546 INFO [train.py:904] (7/8) Epoch 8, batch 5800, loss[loss=0.2314, simple_loss=0.2991, pruned_loss=0.08184, over 12051.00 frames. ], tot_loss[loss=0.2443, simple_loss=0.3206, pruned_loss=0.08399, over 3026202.46 frames. ], batch size: 247, lr: 8.52e-03, grad_scale: 4.0 2023-04-28 21:23:22,179 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5059, 3.6526, 3.7357, 1.7158, 3.9459, 4.0235, 3.1214, 2.9642], device='cuda:7'), covar=tensor([0.0788, 0.0154, 0.0165, 0.1260, 0.0066, 0.0094, 0.0327, 0.0426], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0097, 0.0083, 0.0140, 0.0070, 0.0091, 0.0118, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 21:23:28,987 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6964, 5.0052, 4.7363, 4.7087, 4.4679, 4.4352, 4.4583, 5.0802], device='cuda:7'), covar=tensor([0.0758, 0.0699, 0.0868, 0.0669, 0.0654, 0.0920, 0.0806, 0.0703], device='cuda:7'), in_proj_covar=tensor([0.0465, 0.0581, 0.0498, 0.0399, 0.0371, 0.0385, 0.0487, 0.0428], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:23:32,448 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7258, 3.5844, 3.8019, 3.6901, 3.7483, 4.1828, 3.8881, 3.5872], device='cuda:7'), covar=tensor([0.1938, 0.2217, 0.1869, 0.2241, 0.2777, 0.1409, 0.1468, 0.2828], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0432, 0.0452, 0.0378, 0.0505, 0.0474, 0.0368, 0.0517], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 21:24:30,625 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76900.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:24:33,004 INFO [train.py:904] (7/8) Epoch 8, batch 5850, loss[loss=0.2226, simple_loss=0.3071, pruned_loss=0.069, over 16338.00 frames. ], tot_loss[loss=0.2413, simple_loss=0.3186, pruned_loss=0.08199, over 3034547.51 frames. ], batch size: 146, lr: 8.52e-03, grad_scale: 4.0 2023-04-28 21:24:37,995 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 3.434e+02 4.196e+02 5.124e+02 1.086e+03, threshold=8.393e+02, percent-clipped=1.0 2023-04-28 21:25:09,383 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76925.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 21:25:21,001 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-28 21:25:53,590 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76950.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:25:56,483 INFO [train.py:904] (7/8) Epoch 8, batch 5900, loss[loss=0.2264, simple_loss=0.3063, pruned_loss=0.07321, over 16172.00 frames. ], tot_loss[loss=0.2406, simple_loss=0.3181, pruned_loss=0.08159, over 3040741.33 frames. ], batch size: 165, lr: 8.51e-03, grad_scale: 4.0 2023-04-28 21:26:34,317 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=76973.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:26:42,725 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76979.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:27:11,067 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=76998.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:27:17,813 INFO [train.py:904] (7/8) Epoch 8, batch 5950, loss[loss=0.2255, simple_loss=0.3081, pruned_loss=0.07139, over 16446.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3187, pruned_loss=0.07984, over 3059140.32 frames. ], batch size: 146, lr: 8.51e-03, grad_scale: 4.0 2023-04-28 21:27:21,561 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.299e+02 3.417e+02 4.127e+02 4.793e+02 1.224e+03, threshold=8.253e+02, percent-clipped=3.0 2023-04-28 21:27:25,371 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6527, 2.4998, 2.1255, 3.4490, 2.5772, 3.6364, 1.3713, 2.6275], device='cuda:7'), covar=tensor([0.1297, 0.0633, 0.1252, 0.0143, 0.0207, 0.0398, 0.1559, 0.0861], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0152, 0.0174, 0.0117, 0.0202, 0.0205, 0.0173, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 21:27:55,399 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77027.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:28:33,566 INFO [train.py:904] (7/8) Epoch 8, batch 6000, loss[loss=0.2219, simple_loss=0.303, pruned_loss=0.07038, over 16853.00 frames. ], tot_loss[loss=0.2387, simple_loss=0.3178, pruned_loss=0.0798, over 3056500.62 frames. ], batch size: 96, lr: 8.51e-03, grad_scale: 8.0 2023-04-28 21:28:33,566 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 21:28:41,747 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8225, 3.0185, 2.5910, 2.8065, 3.2148, 2.9050, 3.7354, 3.5666], device='cuda:7'), covar=tensor([0.0042, 0.0190, 0.0259, 0.0211, 0.0135, 0.0211, 0.0094, 0.0102], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0177, 0.0176, 0.0177, 0.0174, 0.0178, 0.0175, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:28:44,114 INFO [train.py:938] (7/8) Epoch 8, validation: loss=0.1707, simple_loss=0.284, pruned_loss=0.02871, over 944034.00 frames. 2023-04-28 21:28:44,115 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 21:29:48,579 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77095.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:30:00,324 INFO [train.py:904] (7/8) Epoch 8, batch 6050, loss[loss=0.2428, simple_loss=0.3093, pruned_loss=0.08821, over 12009.00 frames. ], tot_loss[loss=0.2371, simple_loss=0.3162, pruned_loss=0.07894, over 3064373.90 frames. ], batch size: 246, lr: 8.51e-03, grad_scale: 8.0 2023-04-28 21:30:04,231 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.264e+02 3.621e+02 4.460e+02 5.601e+02 1.783e+03, threshold=8.919e+02, percent-clipped=7.0 2023-04-28 21:30:41,608 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 21:30:45,874 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-28 21:31:19,265 INFO [train.py:904] (7/8) Epoch 8, batch 6100, loss[loss=0.2417, simple_loss=0.3092, pruned_loss=0.08709, over 11761.00 frames. ], tot_loss[loss=0.2363, simple_loss=0.3163, pruned_loss=0.07821, over 3069734.87 frames. ], batch size: 248, lr: 8.50e-03, grad_scale: 8.0 2023-04-28 21:31:26,067 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77156.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 21:32:15,288 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6361, 5.0870, 5.3132, 5.1247, 5.1095, 5.7255, 5.2227, 5.0352], device='cuda:7'), covar=tensor([0.0993, 0.1622, 0.1313, 0.1510, 0.2353, 0.0710, 0.1171, 0.2211], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0434, 0.0458, 0.0385, 0.0515, 0.0481, 0.0373, 0.0522], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 21:32:26,666 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77195.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:32:37,672 INFO [train.py:904] (7/8) Epoch 8, batch 6150, loss[loss=0.2883, simple_loss=0.3417, pruned_loss=0.1175, over 11746.00 frames. ], tot_loss[loss=0.2351, simple_loss=0.3143, pruned_loss=0.07792, over 3049894.34 frames. ], batch size: 248, lr: 8.50e-03, grad_scale: 8.0 2023-04-28 21:32:42,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.089e+02 3.253e+02 3.802e+02 4.768e+02 1.072e+03, threshold=7.605e+02, percent-clipped=4.0 2023-04-28 21:32:52,977 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77211.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:33:59,292 INFO [train.py:904] (7/8) Epoch 8, batch 6200, loss[loss=0.2174, simple_loss=0.2954, pruned_loss=0.06963, over 16467.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3117, pruned_loss=0.07677, over 3066330.16 frames. ], batch size: 62, lr: 8.50e-03, grad_scale: 4.0 2023-04-28 21:34:31,319 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:34:33,235 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77273.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:34:39,304 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77277.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:35:16,106 INFO [train.py:904] (7/8) Epoch 8, batch 6250, loss[loss=0.2455, simple_loss=0.3272, pruned_loss=0.0819, over 16180.00 frames. ], tot_loss[loss=0.2327, simple_loss=0.3118, pruned_loss=0.07677, over 3070998.64 frames. ], batch size: 165, lr: 8.50e-03, grad_scale: 4.0 2023-04-28 21:35:22,797 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 3.064e+02 3.768e+02 4.806e+02 9.942e+02, threshold=7.536e+02, percent-clipped=2.0 2023-04-28 21:35:29,841 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7502, 3.5857, 3.7733, 3.6183, 3.7255, 4.1427, 3.8875, 3.6261], device='cuda:7'), covar=tensor([0.2030, 0.2402, 0.2166, 0.2453, 0.3109, 0.1640, 0.1398, 0.2697], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0434, 0.0459, 0.0385, 0.0511, 0.0483, 0.0372, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 21:35:39,906 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.25 vs. limit=5.0 2023-04-28 21:36:07,309 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77334.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:36:13,011 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77338.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:36:14,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2856, 5.5957, 5.2836, 5.3482, 4.9607, 4.8698, 5.0717, 5.6198], device='cuda:7'), covar=tensor([0.0790, 0.0727, 0.0847, 0.0586, 0.0777, 0.0611, 0.0782, 0.0814], device='cuda:7'), in_proj_covar=tensor([0.0468, 0.0588, 0.0497, 0.0400, 0.0371, 0.0384, 0.0487, 0.0429], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:36:35,487 INFO [train.py:904] (7/8) Epoch 8, batch 6300, loss[loss=0.1965, simple_loss=0.2928, pruned_loss=0.05008, over 16542.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.3106, pruned_loss=0.07506, over 3089232.47 frames. ], batch size: 68, lr: 8.49e-03, grad_scale: 4.0 2023-04-28 21:37:27,782 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4448, 3.4608, 3.2685, 3.0191, 3.0591, 3.3646, 3.2184, 3.1584], device='cuda:7'), covar=tensor([0.0514, 0.0377, 0.0215, 0.0215, 0.0529, 0.0331, 0.1109, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0215, 0.0255, 0.0249, 0.0220, 0.0278, 0.0257, 0.0172, 0.0290], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:37:40,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2427, 1.5157, 1.9726, 2.2124, 2.2156, 2.3873, 1.7296, 2.3482], device='cuda:7'), covar=tensor([0.0130, 0.0303, 0.0161, 0.0183, 0.0168, 0.0120, 0.0268, 0.0070], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0160, 0.0144, 0.0144, 0.0153, 0.0109, 0.0161, 0.0100], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 21:37:54,087 INFO [train.py:904] (7/8) Epoch 8, batch 6350, loss[loss=0.237, simple_loss=0.314, pruned_loss=0.08, over 16161.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.3118, pruned_loss=0.07688, over 3075572.62 frames. ], batch size: 165, lr: 8.49e-03, grad_scale: 4.0 2023-04-28 21:38:00,464 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.093e+02 3.385e+02 4.021e+02 5.215e+02 8.857e+02, threshold=8.042e+02, percent-clipped=2.0 2023-04-28 21:38:02,985 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5631, 2.1535, 1.7342, 1.9736, 2.4610, 2.2194, 2.5878, 2.7041], device='cuda:7'), covar=tensor([0.0069, 0.0235, 0.0295, 0.0280, 0.0119, 0.0209, 0.0128, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0109, 0.0180, 0.0178, 0.0179, 0.0176, 0.0180, 0.0177, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:38:41,650 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3111, 4.1249, 4.3587, 4.5556, 4.6760, 4.2124, 4.6181, 4.6443], device='cuda:7'), covar=tensor([0.1346, 0.1066, 0.1349, 0.0542, 0.0455, 0.0990, 0.0543, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0469, 0.0574, 0.0708, 0.0586, 0.0443, 0.0439, 0.0457, 0.0508], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:39:10,458 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77451.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:39:11,216 INFO [train.py:904] (7/8) Epoch 8, batch 6400, loss[loss=0.2152, simple_loss=0.2952, pruned_loss=0.06758, over 16808.00 frames. ], tot_loss[loss=0.2341, simple_loss=0.312, pruned_loss=0.0781, over 3078527.93 frames. ], batch size: 83, lr: 8.49e-03, grad_scale: 8.0 2023-04-28 21:40:10,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3103, 1.9359, 2.0462, 3.7852, 1.8445, 2.4990, 2.0855, 2.0925], device='cuda:7'), covar=tensor([0.0750, 0.3062, 0.1897, 0.0360, 0.3501, 0.1847, 0.2652, 0.2796], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0361, 0.0301, 0.0315, 0.0395, 0.0402, 0.0322, 0.0426], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:40:17,241 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77495.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:40:26,244 INFO [train.py:904] (7/8) Epoch 8, batch 6450, loss[loss=0.2194, simple_loss=0.2916, pruned_loss=0.07364, over 15279.00 frames. ], tot_loss[loss=0.2335, simple_loss=0.3119, pruned_loss=0.0775, over 3075629.34 frames. ], batch size: 190, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:40:33,079 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 3.422e+02 4.688e+02 6.041e+02 9.477e+02, threshold=9.377e+02, percent-clipped=7.0 2023-04-28 21:41:31,693 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77543.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:41:45,641 INFO [train.py:904] (7/8) Epoch 8, batch 6500, loss[loss=0.2694, simple_loss=0.3178, pruned_loss=0.1105, over 11243.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.3089, pruned_loss=0.07613, over 3064298.28 frames. ], batch size: 246, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:42:09,111 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77567.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:42:24,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3269, 4.3492, 4.4956, 4.4062, 4.5133, 4.9822, 4.5717, 4.3458], device='cuda:7'), covar=tensor([0.1301, 0.1552, 0.1557, 0.1728, 0.2092, 0.0891, 0.1186, 0.2044], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0434, 0.0463, 0.0390, 0.0515, 0.0485, 0.0377, 0.0524], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 21:43:05,177 INFO [train.py:904] (7/8) Epoch 8, batch 6550, loss[loss=0.3043, simple_loss=0.356, pruned_loss=0.1263, over 11276.00 frames. ], tot_loss[loss=0.2332, simple_loss=0.3122, pruned_loss=0.07708, over 3075573.37 frames. ], batch size: 248, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:43:11,242 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.922e+02 3.133e+02 3.759e+02 4.463e+02 1.371e+03, threshold=7.519e+02, percent-clipped=1.0 2023-04-28 21:43:28,379 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77616.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:43:47,937 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77629.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:43:54,006 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77633.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:44:22,377 INFO [train.py:904] (7/8) Epoch 8, batch 6600, loss[loss=0.2821, simple_loss=0.3386, pruned_loss=0.1128, over 11582.00 frames. ], tot_loss[loss=0.2348, simple_loss=0.3146, pruned_loss=0.07746, over 3089758.02 frames. ], batch size: 248, lr: 8.48e-03, grad_scale: 8.0 2023-04-28 21:44:41,557 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.16 vs. limit=5.0 2023-04-28 21:45:00,855 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:45:00,869 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:45:37,087 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4791, 3.5139, 3.2587, 3.1151, 3.0948, 3.3929, 3.2728, 3.1998], device='cuda:7'), covar=tensor([0.0467, 0.0352, 0.0203, 0.0190, 0.0546, 0.0291, 0.0906, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0257, 0.0251, 0.0221, 0.0279, 0.0257, 0.0173, 0.0290], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:45:38,852 INFO [train.py:904] (7/8) Epoch 8, batch 6650, loss[loss=0.241, simple_loss=0.3126, pruned_loss=0.08471, over 16805.00 frames. ], tot_loss[loss=0.2355, simple_loss=0.3149, pruned_loss=0.07807, over 3097671.14 frames. ], batch size: 116, lr: 8.47e-03, grad_scale: 8.0 2023-04-28 21:45:45,542 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.195e+02 3.528e+02 4.149e+02 5.029e+02 1.289e+03, threshold=8.299e+02, percent-clipped=5.0 2023-04-28 21:46:17,762 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9946, 4.0210, 3.8139, 3.6707, 3.5659, 3.9329, 3.6228, 3.7024], device='cuda:7'), covar=tensor([0.0488, 0.0363, 0.0223, 0.0192, 0.0709, 0.0310, 0.0729, 0.0530], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0257, 0.0251, 0.0222, 0.0279, 0.0257, 0.0173, 0.0291], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:46:35,370 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77738.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:46:53,783 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77751.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 21:46:54,548 INFO [train.py:904] (7/8) Epoch 8, batch 6700, loss[loss=0.2239, simple_loss=0.3093, pruned_loss=0.06927, over 16792.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.3146, pruned_loss=0.07906, over 3084520.14 frames. ], batch size: 102, lr: 8.47e-03, grad_scale: 4.0 2023-04-28 21:47:35,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8480, 2.9888, 2.5817, 4.2158, 3.3563, 4.0906, 1.6061, 2.9700], device='cuda:7'), covar=tensor([0.1156, 0.0552, 0.1040, 0.0123, 0.0270, 0.0348, 0.1361, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0150, 0.0172, 0.0118, 0.0199, 0.0203, 0.0172, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 21:47:52,556 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8354, 5.1624, 4.8624, 4.8848, 4.5690, 4.5762, 4.6326, 5.2333], device='cuda:7'), covar=tensor([0.0861, 0.0716, 0.0948, 0.0589, 0.0780, 0.0808, 0.0867, 0.0703], device='cuda:7'), in_proj_covar=tensor([0.0473, 0.0587, 0.0497, 0.0400, 0.0371, 0.0386, 0.0488, 0.0431], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:48:06,760 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77799.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:48:10,960 INFO [train.py:904] (7/8) Epoch 8, batch 6750, loss[loss=0.2967, simple_loss=0.3489, pruned_loss=0.1223, over 11538.00 frames. ], tot_loss[loss=0.2356, simple_loss=0.3134, pruned_loss=0.07891, over 3084208.43 frames. ], batch size: 247, lr: 8.47e-03, grad_scale: 4.0 2023-04-28 21:48:18,412 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.361e+02 3.344e+02 4.001e+02 5.088e+02 1.053e+03, threshold=8.003e+02, percent-clipped=2.0 2023-04-28 21:48:32,362 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0919, 3.0522, 3.1133, 1.7249, 3.3310, 3.3460, 2.6054, 2.5739], device='cuda:7'), covar=tensor([0.0817, 0.0175, 0.0187, 0.1175, 0.0063, 0.0129, 0.0440, 0.0444], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0084, 0.0138, 0.0068, 0.0090, 0.0118, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 21:49:25,323 INFO [train.py:904] (7/8) Epoch 8, batch 6800, loss[loss=0.2206, simple_loss=0.307, pruned_loss=0.06706, over 17025.00 frames. ], tot_loss[loss=0.235, simple_loss=0.3133, pruned_loss=0.07831, over 3097728.51 frames. ], batch size: 55, lr: 8.47e-03, grad_scale: 8.0 2023-04-28 21:49:49,014 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77867.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:50:44,164 INFO [train.py:904] (7/8) Epoch 8, batch 6850, loss[loss=0.2099, simple_loss=0.3094, pruned_loss=0.05518, over 16489.00 frames. ], tot_loss[loss=0.2359, simple_loss=0.3146, pruned_loss=0.07866, over 3084150.09 frames. ], batch size: 75, lr: 8.46e-03, grad_scale: 4.0 2023-04-28 21:50:53,214 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.176e+02 3.270e+02 3.895e+02 4.577e+02 9.421e+02, threshold=7.790e+02, percent-clipped=1.0 2023-04-28 21:51:03,726 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77915.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:51:24,317 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77929.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:51:26,098 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-28 21:51:30,460 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77933.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:51:59,518 INFO [train.py:904] (7/8) Epoch 8, batch 6900, loss[loss=0.238, simple_loss=0.3202, pruned_loss=0.07788, over 16368.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.317, pruned_loss=0.0779, over 3114811.62 frames. ], batch size: 146, lr: 8.46e-03, grad_scale: 4.0 2023-04-28 21:52:31,664 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77972.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:52:36,478 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-28 21:52:39,054 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77977.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:52:45,340 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=77981.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:53:20,824 INFO [train.py:904] (7/8) Epoch 8, batch 6950, loss[loss=0.229, simple_loss=0.3178, pruned_loss=0.07008, over 16826.00 frames. ], tot_loss[loss=0.2386, simple_loss=0.3185, pruned_loss=0.0793, over 3118249.62 frames. ], batch size: 102, lr: 8.46e-03, grad_scale: 4.0 2023-04-28 21:53:29,770 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.004e+02 3.295e+02 4.352e+02 5.795e+02 9.816e+02, threshold=8.703e+02, percent-clipped=9.0 2023-04-28 21:53:52,848 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1528, 3.0516, 3.1195, 1.7013, 3.3347, 3.3792, 2.7009, 2.5171], device='cuda:7'), covar=tensor([0.0712, 0.0192, 0.0171, 0.1101, 0.0058, 0.0117, 0.0363, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0096, 0.0082, 0.0137, 0.0067, 0.0089, 0.0117, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 21:54:10,667 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78033.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:54:35,357 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4244, 4.0778, 3.8338, 1.9495, 3.1779, 2.8525, 3.8111, 4.0461], device='cuda:7'), covar=tensor([0.0217, 0.0419, 0.0467, 0.1611, 0.0595, 0.0707, 0.0481, 0.0600], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0132, 0.0154, 0.0140, 0.0133, 0.0123, 0.0135, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 21:54:35,458 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-28 21:54:38,879 INFO [train.py:904] (7/8) Epoch 8, batch 7000, loss[loss=0.1906, simple_loss=0.2951, pruned_loss=0.0431, over 16820.00 frames. ], tot_loss[loss=0.2379, simple_loss=0.3181, pruned_loss=0.07878, over 3102416.27 frames. ], batch size: 39, lr: 8.45e-03, grad_scale: 4.0 2023-04-28 21:54:47,335 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-28 21:55:53,786 INFO [train.py:904] (7/8) Epoch 8, batch 7050, loss[loss=0.2887, simple_loss=0.3447, pruned_loss=0.1163, over 11569.00 frames. ], tot_loss[loss=0.2375, simple_loss=0.3181, pruned_loss=0.0785, over 3094072.62 frames. ], batch size: 248, lr: 8.45e-03, grad_scale: 4.0 2023-04-28 21:56:03,906 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.852e+02 3.078e+02 3.817e+02 4.566e+02 8.440e+02, threshold=7.634e+02, percent-clipped=0.0 2023-04-28 21:56:21,228 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2372, 2.9524, 2.4597, 2.2485, 2.3162, 2.0049, 2.9090, 2.8517], device='cuda:7'), covar=tensor([0.2125, 0.0831, 0.1539, 0.1653, 0.1944, 0.1887, 0.0628, 0.0900], device='cuda:7'), in_proj_covar=tensor([0.0294, 0.0255, 0.0276, 0.0264, 0.0280, 0.0212, 0.0260, 0.0277], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 21:57:11,219 INFO [train.py:904] (7/8) Epoch 8, batch 7100, loss[loss=0.2118, simple_loss=0.3023, pruned_loss=0.06066, over 16854.00 frames. ], tot_loss[loss=0.2385, simple_loss=0.3177, pruned_loss=0.07965, over 3050300.96 frames. ], batch size: 83, lr: 8.45e-03, grad_scale: 2.0 2023-04-28 21:58:26,616 INFO [train.py:904] (7/8) Epoch 8, batch 7150, loss[loss=0.233, simple_loss=0.3163, pruned_loss=0.07487, over 16387.00 frames. ], tot_loss[loss=0.2355, simple_loss=0.3147, pruned_loss=0.07817, over 3073735.16 frames. ], batch size: 146, lr: 8.45e-03, grad_scale: 2.0 2023-04-28 21:58:36,176 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.673e+02 3.732e+02 4.325e+02 5.377e+02 1.184e+03, threshold=8.651e+02, percent-clipped=7.0 2023-04-28 21:59:09,631 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78232.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 21:59:39,856 INFO [train.py:904] (7/8) Epoch 8, batch 7200, loss[loss=0.2105, simple_loss=0.286, pruned_loss=0.06748, over 11765.00 frames. ], tot_loss[loss=0.2328, simple_loss=0.3125, pruned_loss=0.07648, over 3077982.01 frames. ], batch size: 248, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:00:09,976 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78272.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:00:44,987 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78293.0, num_to_drop=1, layers_to_drop={3} 2023-04-28 22:00:46,993 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8189, 3.3618, 3.2259, 1.6860, 2.7506, 2.2613, 3.3707, 3.4015], device='cuda:7'), covar=tensor([0.0324, 0.0553, 0.0565, 0.1870, 0.0795, 0.0913, 0.0601, 0.0817], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0133, 0.0155, 0.0141, 0.0133, 0.0124, 0.0136, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 22:00:59,553 INFO [train.py:904] (7/8) Epoch 8, batch 7250, loss[loss=0.204, simple_loss=0.2849, pruned_loss=0.06148, over 16877.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.3099, pruned_loss=0.0748, over 3073016.10 frames. ], batch size: 109, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:01:10,052 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.959e+02 3.722e+02 4.557e+02 8.668e+02, threshold=7.444e+02, percent-clipped=1.0 2023-04-28 22:01:27,081 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=78320.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:01:30,623 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-28 22:01:47,750 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:02:14,474 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2741, 1.3898, 1.9148, 2.2042, 2.1765, 2.4871, 1.5330, 2.3002], device='cuda:7'), covar=tensor([0.0121, 0.0315, 0.0179, 0.0182, 0.0164, 0.0089, 0.0289, 0.0074], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0159, 0.0142, 0.0142, 0.0149, 0.0106, 0.0157, 0.0098], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 22:02:16,551 INFO [train.py:904] (7/8) Epoch 8, batch 7300, loss[loss=0.2149, simple_loss=0.2971, pruned_loss=0.06635, over 16613.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.3087, pruned_loss=0.07395, over 3074089.38 frames. ], batch size: 57, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:02:44,864 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-28 22:03:02,794 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=78381.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:03:34,481 INFO [train.py:904] (7/8) Epoch 8, batch 7350, loss[loss=0.2585, simple_loss=0.3148, pruned_loss=0.1011, over 10967.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3089, pruned_loss=0.07448, over 3066242.96 frames. ], batch size: 250, lr: 8.44e-03, grad_scale: 4.0 2023-04-28 22:03:45,282 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.445e+02 3.549e+02 4.373e+02 5.467e+02 2.414e+03, threshold=8.746e+02, percent-clipped=10.0 2023-04-28 22:04:55,420 INFO [train.py:904] (7/8) Epoch 8, batch 7400, loss[loss=0.2113, simple_loss=0.3007, pruned_loss=0.06099, over 16976.00 frames. ], tot_loss[loss=0.2304, simple_loss=0.3104, pruned_loss=0.07523, over 3069985.29 frames. ], batch size: 41, lr: 8.43e-03, grad_scale: 4.0 2023-04-28 22:05:08,386 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78460.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:06:08,223 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3023, 3.3135, 1.4877, 3.5515, 2.2538, 3.5339, 1.7264, 2.5473], device='cuda:7'), covar=tensor([0.0178, 0.0318, 0.1827, 0.0091, 0.0853, 0.0445, 0.1629, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0157, 0.0181, 0.0101, 0.0165, 0.0196, 0.0190, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 22:06:13,356 INFO [train.py:904] (7/8) Epoch 8, batch 7450, loss[loss=0.2426, simple_loss=0.3312, pruned_loss=0.07696, over 15313.00 frames. ], tot_loss[loss=0.2325, simple_loss=0.3116, pruned_loss=0.07674, over 3046626.43 frames. ], batch size: 190, lr: 8.43e-03, grad_scale: 4.0 2023-04-28 22:06:19,879 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6252, 4.4604, 4.5399, 4.8084, 5.0251, 4.5406, 5.0141, 4.9759], device='cuda:7'), covar=tensor([0.1578, 0.1125, 0.1882, 0.0813, 0.0607, 0.0684, 0.0562, 0.0606], device='cuda:7'), in_proj_covar=tensor([0.0463, 0.0569, 0.0702, 0.0576, 0.0441, 0.0430, 0.0459, 0.0500], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:06:26,559 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 3.338e+02 4.187e+02 5.103e+02 1.080e+03, threshold=8.375e+02, percent-clipped=2.0 2023-04-28 22:06:47,424 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78521.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:06:49,017 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 22:07:34,693 INFO [train.py:904] (7/8) Epoch 8, batch 7500, loss[loss=0.2137, simple_loss=0.2985, pruned_loss=0.06445, over 16449.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3123, pruned_loss=0.07641, over 3053973.98 frames. ], batch size: 146, lr: 8.43e-03, grad_scale: 4.0 2023-04-28 22:07:42,054 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78556.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:07:53,028 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-28 22:08:33,369 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78588.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:08:55,094 INFO [train.py:904] (7/8) Epoch 8, batch 7550, loss[loss=0.2128, simple_loss=0.2907, pruned_loss=0.06745, over 17250.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3107, pruned_loss=0.07637, over 3050423.74 frames. ], batch size: 52, lr: 8.42e-03, grad_scale: 4.0 2023-04-28 22:09:05,666 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.406e+02 3.301e+02 4.162e+02 5.431e+02 1.258e+03, threshold=8.325e+02, percent-clipped=7.0 2023-04-28 22:09:20,013 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78617.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:10:05,648 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6966, 3.9217, 4.3423, 1.9716, 4.6583, 4.6396, 3.0714, 3.2399], device='cuda:7'), covar=tensor([0.0715, 0.0186, 0.0127, 0.1139, 0.0040, 0.0074, 0.0365, 0.0404], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0094, 0.0080, 0.0134, 0.0065, 0.0087, 0.0115, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 22:10:12,507 INFO [train.py:904] (7/8) Epoch 8, batch 7600, loss[loss=0.2037, simple_loss=0.286, pruned_loss=0.06077, over 16625.00 frames. ], tot_loss[loss=0.2319, simple_loss=0.3104, pruned_loss=0.07663, over 3062224.84 frames. ], batch size: 62, lr: 8.42e-03, grad_scale: 8.0 2023-04-28 22:10:20,572 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1286, 3.3249, 3.5425, 3.5074, 3.5013, 3.3035, 3.3443, 3.3914], device='cuda:7'), covar=tensor([0.0351, 0.0562, 0.0426, 0.0438, 0.0464, 0.0449, 0.0774, 0.0506], device='cuda:7'), in_proj_covar=tensor([0.0295, 0.0298, 0.0301, 0.0290, 0.0343, 0.0318, 0.0419, 0.0262], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 22:10:46,394 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9076, 3.3194, 3.3411, 2.1320, 3.1588, 3.3619, 3.2057, 1.7069], device='cuda:7'), covar=tensor([0.0400, 0.0048, 0.0047, 0.0315, 0.0079, 0.0102, 0.0083, 0.0420], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0062, 0.0064, 0.0121, 0.0069, 0.0079, 0.0071, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 22:11:30,208 INFO [train.py:904] (7/8) Epoch 8, batch 7650, loss[loss=0.309, simple_loss=0.354, pruned_loss=0.132, over 11320.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3122, pruned_loss=0.07768, over 3066342.12 frames. ], batch size: 246, lr: 8.42e-03, grad_scale: 8.0 2023-04-28 22:11:40,449 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 3.310e+02 4.230e+02 5.150e+02 8.626e+02, threshold=8.460e+02, percent-clipped=2.0 2023-04-28 22:12:45,897 INFO [train.py:904] (7/8) Epoch 8, batch 7700, loss[loss=0.2236, simple_loss=0.2983, pruned_loss=0.07446, over 15363.00 frames. ], tot_loss[loss=0.2346, simple_loss=0.3126, pruned_loss=0.07831, over 3067936.50 frames. ], batch size: 190, lr: 8.42e-03, grad_scale: 4.0 2023-04-28 22:13:26,829 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4394, 3.3611, 2.7852, 2.0847, 2.3237, 2.0531, 3.5666, 3.2690], device='cuda:7'), covar=tensor([0.2612, 0.0753, 0.1433, 0.2020, 0.2097, 0.1860, 0.0477, 0.0932], device='cuda:7'), in_proj_covar=tensor([0.0294, 0.0252, 0.0275, 0.0262, 0.0277, 0.0211, 0.0259, 0.0276], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:13:29,039 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-28 22:14:03,989 INFO [train.py:904] (7/8) Epoch 8, batch 7750, loss[loss=0.2264, simple_loss=0.3141, pruned_loss=0.06932, over 16430.00 frames. ], tot_loss[loss=0.2338, simple_loss=0.3123, pruned_loss=0.07765, over 3084570.09 frames. ], batch size: 146, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:14:11,787 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.08 vs. limit=5.0 2023-04-28 22:14:17,810 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.129e+02 3.510e+02 4.111e+02 5.372e+02 9.340e+02, threshold=8.221e+02, percent-clipped=3.0 2023-04-28 22:14:26,065 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78816.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:15:19,817 INFO [train.py:904] (7/8) Epoch 8, batch 7800, loss[loss=0.2122, simple_loss=0.2927, pruned_loss=0.0659, over 16397.00 frames. ], tot_loss[loss=0.2347, simple_loss=0.313, pruned_loss=0.0782, over 3084174.43 frames. ], batch size: 68, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:16:16,753 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78888.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:16:38,239 INFO [train.py:904] (7/8) Epoch 8, batch 7850, loss[loss=0.2383, simple_loss=0.3214, pruned_loss=0.07763, over 16939.00 frames. ], tot_loss[loss=0.2353, simple_loss=0.314, pruned_loss=0.07832, over 3066986.59 frames. ], batch size: 109, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:16:50,961 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.229e+02 3.386e+02 3.845e+02 4.858e+02 8.286e+02, threshold=7.691e+02, percent-clipped=1.0 2023-04-28 22:16:54,019 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78912.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:17:29,763 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=78936.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:17:54,236 INFO [train.py:904] (7/8) Epoch 8, batch 7900, loss[loss=0.2271, simple_loss=0.3151, pruned_loss=0.06956, over 16718.00 frames. ], tot_loss[loss=0.233, simple_loss=0.3124, pruned_loss=0.07682, over 3100219.20 frames. ], batch size: 83, lr: 8.41e-03, grad_scale: 2.0 2023-04-28 22:18:15,561 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78965.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:19:13,458 INFO [train.py:904] (7/8) Epoch 8, batch 7950, loss[loss=0.2313, simple_loss=0.3111, pruned_loss=0.07577, over 16410.00 frames. ], tot_loss[loss=0.2332, simple_loss=0.3125, pruned_loss=0.07696, over 3116073.66 frames. ], batch size: 146, lr: 8.40e-03, grad_scale: 2.0 2023-04-28 22:19:14,053 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6713, 2.7993, 2.3245, 4.2354, 3.1520, 4.0907, 1.2740, 2.9264], device='cuda:7'), covar=tensor([0.1297, 0.0618, 0.1201, 0.0150, 0.0270, 0.0350, 0.1590, 0.0767], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0152, 0.0173, 0.0119, 0.0201, 0.0202, 0.0174, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 22:19:28,063 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.154e+02 3.667e+02 4.128e+02 4.915e+02 9.776e+02, threshold=8.255e+02, percent-clipped=2.0 2023-04-28 22:19:35,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3730, 3.9498, 3.9355, 2.6789, 3.5572, 3.9405, 3.6605, 2.2836], device='cuda:7'), covar=tensor([0.0347, 0.0027, 0.0029, 0.0255, 0.0056, 0.0061, 0.0043, 0.0320], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0061, 0.0063, 0.0119, 0.0067, 0.0079, 0.0069, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 22:19:52,376 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79026.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:20:32,290 INFO [train.py:904] (7/8) Epoch 8, batch 8000, loss[loss=0.207, simple_loss=0.2991, pruned_loss=0.05744, over 16803.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.3133, pruned_loss=0.07757, over 3118604.38 frames. ], batch size: 102, lr: 8.40e-03, grad_scale: 4.0 2023-04-28 22:21:12,821 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8024, 3.8777, 4.1896, 4.1319, 4.1700, 3.8551, 3.8072, 3.8700], device='cuda:7'), covar=tensor([0.0400, 0.0614, 0.0487, 0.0596, 0.0648, 0.0535, 0.1328, 0.0512], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0295, 0.0298, 0.0287, 0.0336, 0.0313, 0.0416, 0.0257], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 22:21:48,619 INFO [train.py:904] (7/8) Epoch 8, batch 8050, loss[loss=0.2049, simple_loss=0.2969, pruned_loss=0.05652, over 16802.00 frames. ], tot_loss[loss=0.2339, simple_loss=0.3132, pruned_loss=0.07731, over 3113158.56 frames. ], batch size: 102, lr: 8.40e-03, grad_scale: 4.0 2023-04-28 22:21:53,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5053, 3.5230, 3.2124, 3.0982, 3.0851, 3.4413, 3.2818, 3.2334], device='cuda:7'), covar=tensor([0.0479, 0.0387, 0.0230, 0.0207, 0.0542, 0.0328, 0.0996, 0.0423], device='cuda:7'), in_proj_covar=tensor([0.0213, 0.0256, 0.0250, 0.0222, 0.0277, 0.0257, 0.0174, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:22:02,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.318e+02 3.198e+02 3.767e+02 4.640e+02 1.099e+03, threshold=7.534e+02, percent-clipped=1.0 2023-04-28 22:22:11,214 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:22:28,353 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9232, 1.9456, 2.2451, 3.1946, 2.1690, 2.3381, 2.2048, 2.1119], device='cuda:7'), covar=tensor([0.0838, 0.2871, 0.1657, 0.0529, 0.3264, 0.1844, 0.2476, 0.2721], device='cuda:7'), in_proj_covar=tensor([0.0344, 0.0370, 0.0310, 0.0320, 0.0407, 0.0409, 0.0329, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:23:05,324 INFO [train.py:904] (7/8) Epoch 8, batch 8100, loss[loss=0.2471, simple_loss=0.3124, pruned_loss=0.0909, over 11611.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3122, pruned_loss=0.0765, over 3105089.88 frames. ], batch size: 247, lr: 8.40e-03, grad_scale: 4.0 2023-04-28 22:23:18,244 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6806, 3.8460, 4.1049, 1.9480, 4.4186, 4.4376, 3.0867, 3.1123], device='cuda:7'), covar=tensor([0.0729, 0.0153, 0.0143, 0.1183, 0.0040, 0.0063, 0.0325, 0.0434], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0097, 0.0083, 0.0139, 0.0067, 0.0089, 0.0118, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 22:23:24,437 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79164.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:24:08,462 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-28 22:24:22,974 INFO [train.py:904] (7/8) Epoch 8, batch 8150, loss[loss=0.2014, simple_loss=0.2771, pruned_loss=0.06292, over 16912.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3104, pruned_loss=0.0758, over 3105400.25 frames. ], batch size: 109, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:24:36,889 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.139e+02 3.692e+02 4.407e+02 5.311e+02 8.589e+02, threshold=8.814e+02, percent-clipped=3.0 2023-04-28 22:24:38,598 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79212.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:24:50,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9221, 3.3496, 3.3998, 2.2066, 3.1904, 3.4279, 3.2856, 1.8823], device='cuda:7'), covar=tensor([0.0412, 0.0034, 0.0040, 0.0307, 0.0064, 0.0069, 0.0047, 0.0336], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0061, 0.0064, 0.0120, 0.0069, 0.0080, 0.0069, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 22:25:39,755 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8737, 4.6361, 4.5297, 3.4191, 4.2227, 4.4579, 4.1431, 2.3014], device='cuda:7'), covar=tensor([0.0378, 0.0033, 0.0042, 0.0240, 0.0061, 0.0122, 0.0053, 0.0414], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0061, 0.0064, 0.0120, 0.0069, 0.0080, 0.0069, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 22:25:42,520 INFO [train.py:904] (7/8) Epoch 8, batch 8200, loss[loss=0.2361, simple_loss=0.3184, pruned_loss=0.0769, over 16507.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.3081, pruned_loss=0.07529, over 3115315.07 frames. ], batch size: 68, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:25:55,508 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79260.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:26:36,834 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79285.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:27:04,320 INFO [train.py:904] (7/8) Epoch 8, batch 8250, loss[loss=0.2001, simple_loss=0.2802, pruned_loss=0.05999, over 12108.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3069, pruned_loss=0.07253, over 3115635.80 frames. ], batch size: 246, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:27:19,443 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 3.091e+02 3.915e+02 5.485e+02 9.423e+02, threshold=7.829e+02, percent-clipped=2.0 2023-04-28 22:27:37,074 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79321.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:27:56,943 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79333.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:28:02,794 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1114, 2.6943, 2.7036, 1.8699, 2.8561, 2.8651, 2.5102, 2.4738], device='cuda:7'), covar=tensor([0.0619, 0.0161, 0.0176, 0.0930, 0.0068, 0.0124, 0.0334, 0.0337], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0096, 0.0081, 0.0138, 0.0065, 0.0087, 0.0117, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 22:28:17,700 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79346.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:28:26,474 INFO [train.py:904] (7/8) Epoch 8, batch 8300, loss[loss=0.2201, simple_loss=0.3084, pruned_loss=0.06589, over 16885.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.3038, pruned_loss=0.06936, over 3109832.14 frames. ], batch size: 116, lr: 8.39e-03, grad_scale: 4.0 2023-04-28 22:28:47,037 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79364.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:29:04,967 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79375.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:29:19,421 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6813, 4.6346, 5.0360, 4.9647, 4.9694, 4.7143, 4.6399, 4.4116], device='cuda:7'), covar=tensor([0.0234, 0.0438, 0.0306, 0.0372, 0.0402, 0.0282, 0.0887, 0.0386], device='cuda:7'), in_proj_covar=tensor([0.0284, 0.0291, 0.0293, 0.0283, 0.0331, 0.0307, 0.0409, 0.0253], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 22:29:36,159 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79394.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:29:48,239 INFO [train.py:904] (7/8) Epoch 8, batch 8350, loss[loss=0.2048, simple_loss=0.3013, pruned_loss=0.05419, over 15364.00 frames. ], tot_loss[loss=0.2179, simple_loss=0.3024, pruned_loss=0.06668, over 3103655.87 frames. ], batch size: 191, lr: 8.38e-03, grad_scale: 4.0 2023-04-28 22:30:02,873 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 2.645e+02 3.311e+02 4.098e+02 6.833e+02, threshold=6.622e+02, percent-clipped=0.0 2023-04-28 22:30:15,528 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 22:30:26,278 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79425.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:30:28,236 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2023-04-28 22:30:43,756 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79436.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:30:59,991 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-04-28 22:31:09,026 INFO [train.py:904] (7/8) Epoch 8, batch 8400, loss[loss=0.1905, simple_loss=0.278, pruned_loss=0.05148, over 15453.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2993, pruned_loss=0.06449, over 3083518.83 frames. ], batch size: 191, lr: 8.38e-03, grad_scale: 8.0 2023-04-28 22:32:27,073 INFO [train.py:904] (7/8) Epoch 8, batch 8450, loss[loss=0.2012, simple_loss=0.285, pruned_loss=0.05875, over 12319.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2974, pruned_loss=0.06283, over 3071957.44 frames. ], batch size: 246, lr: 8.38e-03, grad_scale: 8.0 2023-04-28 22:32:42,130 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.576e+02 3.232e+02 4.028e+02 7.324e+02, threshold=6.464e+02, percent-clipped=2.0 2023-04-28 22:32:46,949 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5559, 3.5635, 3.5200, 3.0277, 3.4782, 2.0461, 3.3027, 2.9953], device='cuda:7'), covar=tensor([0.0085, 0.0070, 0.0116, 0.0175, 0.0066, 0.1819, 0.0092, 0.0139], device='cuda:7'), in_proj_covar=tensor([0.0108, 0.0094, 0.0141, 0.0136, 0.0112, 0.0160, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:32:54,624 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8614, 5.1409, 5.3068, 5.1716, 5.1199, 5.6643, 5.1337, 4.9657], device='cuda:7'), covar=tensor([0.0780, 0.1557, 0.1350, 0.1651, 0.2246, 0.0844, 0.1207, 0.2004], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0418, 0.0449, 0.0376, 0.0494, 0.0473, 0.0366, 0.0500], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 22:33:39,597 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8906, 3.4539, 3.4110, 1.9684, 2.7617, 2.3493, 3.3375, 3.5909], device='cuda:7'), covar=tensor([0.0279, 0.0598, 0.0459, 0.1651, 0.0744, 0.0908, 0.0671, 0.0699], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0128, 0.0149, 0.0136, 0.0128, 0.0121, 0.0132, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-28 22:33:47,278 INFO [train.py:904] (7/8) Epoch 8, batch 8500, loss[loss=0.1839, simple_loss=0.2796, pruned_loss=0.04409, over 16703.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2928, pruned_loss=0.05984, over 3073122.48 frames. ], batch size: 124, lr: 8.37e-03, grad_scale: 8.0 2023-04-28 22:34:40,506 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79585.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:35:09,525 INFO [train.py:904] (7/8) Epoch 8, batch 8550, loss[loss=0.2, simple_loss=0.2804, pruned_loss=0.05978, over 11941.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2906, pruned_loss=0.0592, over 3039452.13 frames. ], batch size: 246, lr: 8.37e-03, grad_scale: 8.0 2023-04-28 22:35:15,195 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-28 22:35:26,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.652e+02 3.316e+02 4.199e+02 1.038e+03, threshold=6.632e+02, percent-clipped=3.0 2023-04-28 22:35:47,573 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79621.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:36:28,528 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79641.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:36:37,437 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79646.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:36:48,482 INFO [train.py:904] (7/8) Epoch 8, batch 8600, loss[loss=0.174, simple_loss=0.2727, pruned_loss=0.03768, over 16645.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2906, pruned_loss=0.05803, over 3045658.68 frames. ], batch size: 89, lr: 8.37e-03, grad_scale: 8.0 2023-04-28 22:37:00,178 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79657.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:37:25,669 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79669.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:37:27,103 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9015, 1.9909, 2.3045, 3.2186, 2.0333, 2.3114, 2.1951, 2.0545], device='cuda:7'), covar=tensor([0.0843, 0.2897, 0.1705, 0.0461, 0.3568, 0.1948, 0.2614, 0.2912], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0355, 0.0299, 0.0305, 0.0388, 0.0390, 0.0317, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:38:02,804 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79689.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:38:26,215 INFO [train.py:904] (7/8) Epoch 8, batch 8650, loss[loss=0.1937, simple_loss=0.2802, pruned_loss=0.05357, over 16847.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2884, pruned_loss=0.05682, over 3006718.76 frames. ], batch size: 124, lr: 8.37e-03, grad_scale: 4.0 2023-04-28 22:38:50,265 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.014e+02 2.698e+02 3.186e+02 3.919e+02 1.176e+03, threshold=6.372e+02, percent-clipped=4.0 2023-04-28 22:39:04,705 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79718.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:39:09,491 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79720.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:39:32,982 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79731.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:39:33,371 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 22:39:47,559 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.65 vs. limit=5.0 2023-04-28 22:40:12,075 INFO [train.py:904] (7/8) Epoch 8, batch 8700, loss[loss=0.1769, simple_loss=0.2644, pruned_loss=0.0447, over 16602.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2853, pruned_loss=0.05494, over 3027229.86 frames. ], batch size: 57, lr: 8.36e-03, grad_scale: 4.0 2023-04-28 22:40:33,056 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79762.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:40:47,917 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-28 22:41:50,481 INFO [train.py:904] (7/8) Epoch 8, batch 8750, loss[loss=0.2001, simple_loss=0.295, pruned_loss=0.05263, over 15181.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2849, pruned_loss=0.05452, over 3024902.07 frames. ], batch size: 191, lr: 8.36e-03, grad_scale: 4.0 2023-04-28 22:42:15,365 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.669e+02 3.166e+02 4.113e+02 7.290e+02, threshold=6.332e+02, percent-clipped=2.0 2023-04-28 22:42:44,175 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79823.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:43:44,539 INFO [train.py:904] (7/8) Epoch 8, batch 8800, loss[loss=0.1825, simple_loss=0.2753, pruned_loss=0.0449, over 16252.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2828, pruned_loss=0.05325, over 3025175.14 frames. ], batch size: 165, lr: 8.36e-03, grad_scale: 8.0 2023-04-28 22:45:24,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2279, 4.2950, 4.4553, 4.3915, 4.2958, 4.7913, 4.4808, 4.2570], device='cuda:7'), covar=tensor([0.1368, 0.1460, 0.1061, 0.1358, 0.1997, 0.0859, 0.1022, 0.1925], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0407, 0.0439, 0.0368, 0.0482, 0.0460, 0.0356, 0.0484], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:45:31,369 INFO [train.py:904] (7/8) Epoch 8, batch 8850, loss[loss=0.1978, simple_loss=0.3014, pruned_loss=0.04715, over 16667.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2852, pruned_loss=0.05236, over 3025705.91 frames. ], batch size: 57, lr: 8.36e-03, grad_scale: 8.0 2023-04-28 22:45:51,486 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.661e+02 3.236e+02 3.873e+02 8.211e+02, threshold=6.471e+02, percent-clipped=3.0 2023-04-28 22:46:58,249 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79941.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:46:58,289 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79941.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 22:47:20,941 INFO [train.py:904] (7/8) Epoch 8, batch 8900, loss[loss=0.2046, simple_loss=0.2903, pruned_loss=0.05943, over 16563.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2847, pruned_loss=0.05126, over 3033362.79 frames. ], batch size: 62, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:48:55,759 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=79989.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:48:55,794 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79989.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:49:29,484 INFO [train.py:904] (7/8) Epoch 8, batch 8950, loss[loss=0.1816, simple_loss=0.2716, pruned_loss=0.04575, over 16684.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2846, pruned_loss=0.05177, over 3050763.60 frames. ], batch size: 134, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:49:31,091 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2219, 4.1641, 4.1051, 3.6319, 4.0820, 1.5370, 3.8460, 3.8261], device='cuda:7'), covar=tensor([0.0065, 0.0053, 0.0103, 0.0225, 0.0069, 0.2178, 0.0113, 0.0167], device='cuda:7'), in_proj_covar=tensor([0.0107, 0.0093, 0.0139, 0.0132, 0.0111, 0.0162, 0.0127, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:49:50,118 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-28 22:49:50,493 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.495e+02 3.051e+02 3.831e+02 8.360e+02, threshold=6.103e+02, percent-clipped=2.0 2023-04-28 22:49:54,999 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80013.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:50:03,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6023, 2.9186, 2.4439, 4.2078, 3.0157, 4.1447, 1.3484, 3.0791], device='cuda:7'), covar=tensor([0.1471, 0.0618, 0.1145, 0.0075, 0.0134, 0.0308, 0.1571, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0148, 0.0169, 0.0113, 0.0183, 0.0195, 0.0170, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 22:50:08,399 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80020.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:50:30,266 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8055, 3.6731, 3.8479, 3.9610, 4.0413, 3.6255, 4.0345, 4.0589], device='cuda:7'), covar=tensor([0.1012, 0.0846, 0.1140, 0.0542, 0.0499, 0.1542, 0.0505, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0541, 0.0661, 0.0551, 0.0420, 0.0413, 0.0436, 0.0482], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:50:32,311 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80031.0, num_to_drop=1, layers_to_drop={0} 2023-04-28 22:50:39,725 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-28 22:50:46,302 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80037.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:50:46,418 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2046, 3.9969, 4.2483, 4.3812, 4.5295, 4.0933, 4.5141, 4.5080], device='cuda:7'), covar=tensor([0.1135, 0.0936, 0.1262, 0.0586, 0.0474, 0.0897, 0.0448, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0541, 0.0661, 0.0551, 0.0420, 0.0412, 0.0436, 0.0482], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 22:51:17,204 INFO [train.py:904] (7/8) Epoch 8, batch 9000, loss[loss=0.2079, simple_loss=0.2828, pruned_loss=0.06653, over 12029.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2812, pruned_loss=0.05029, over 3059613.70 frames. ], batch size: 248, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:51:17,204 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 22:51:27,534 INFO [train.py:938] (7/8) Epoch 8, validation: loss=0.1608, simple_loss=0.265, pruned_loss=0.02828, over 944034.00 frames. 2023-04-28 22:51:27,535 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 22:51:53,097 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3021, 1.5002, 1.7381, 2.1604, 2.1551, 2.2326, 1.5800, 2.3399], device='cuda:7'), covar=tensor([0.0127, 0.0303, 0.0214, 0.0190, 0.0203, 0.0149, 0.0329, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0156, 0.0142, 0.0138, 0.0146, 0.0103, 0.0155, 0.0094], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 22:52:04,336 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80068.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:52:26,511 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80079.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 22:52:47,028 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-28 22:52:56,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4591, 4.5342, 4.9088, 4.8495, 4.8990, 4.6518, 4.5068, 4.4057], device='cuda:7'), covar=tensor([0.0366, 0.0697, 0.0418, 0.0497, 0.0443, 0.0384, 0.0853, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0276, 0.0278, 0.0279, 0.0273, 0.0320, 0.0299, 0.0386, 0.0244], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-28 22:53:14,312 INFO [train.py:904] (7/8) Epoch 8, batch 9050, loss[loss=0.1748, simple_loss=0.2588, pruned_loss=0.04541, over 16861.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2826, pruned_loss=0.05095, over 3083594.07 frames. ], batch size: 116, lr: 8.35e-03, grad_scale: 8.0 2023-04-28 22:53:35,359 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 2.777e+02 3.475e+02 4.225e+02 7.695e+02, threshold=6.949e+02, percent-clipped=5.0 2023-04-28 22:53:48,945 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80118.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:54:59,320 INFO [train.py:904] (7/8) Epoch 8, batch 9100, loss[loss=0.2044, simple_loss=0.2975, pruned_loss=0.05567, over 16671.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2822, pruned_loss=0.05125, over 3087466.45 frames. ], batch size: 62, lr: 8.34e-03, grad_scale: 8.0 2023-04-28 22:56:59,476 INFO [train.py:904] (7/8) Epoch 8, batch 9150, loss[loss=0.1893, simple_loss=0.275, pruned_loss=0.05176, over 16516.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2833, pruned_loss=0.05108, over 3089486.48 frames. ], batch size: 68, lr: 8.34e-03, grad_scale: 8.0 2023-04-28 22:57:20,189 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.002e+02 2.760e+02 3.111e+02 3.901e+02 6.426e+02, threshold=6.222e+02, percent-clipped=0.0 2023-04-28 22:57:36,665 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-28 22:57:54,226 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3360, 4.0405, 4.0251, 2.6136, 3.6500, 4.0331, 3.7785, 2.0093], device='cuda:7'), covar=tensor([0.0367, 0.0018, 0.0027, 0.0273, 0.0055, 0.0049, 0.0034, 0.0369], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0060, 0.0061, 0.0117, 0.0067, 0.0076, 0.0067, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 22:58:26,806 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80241.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:58:44,961 INFO [train.py:904] (7/8) Epoch 8, batch 9200, loss[loss=0.1685, simple_loss=0.2508, pruned_loss=0.04306, over 12222.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2791, pruned_loss=0.05022, over 3080719.05 frames. ], batch size: 248, lr: 8.34e-03, grad_scale: 8.0 2023-04-28 22:59:41,140 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-28 22:59:43,163 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80283.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 22:59:56,007 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80289.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:00:22,119 INFO [train.py:904] (7/8) Epoch 8, batch 9250, loss[loss=0.1837, simple_loss=0.2759, pruned_loss=0.04573, over 15243.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2795, pruned_loss=0.05057, over 3063161.91 frames. ], batch size: 190, lr: 8.34e-03, grad_scale: 4.0 2023-04-28 23:00:42,872 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 2.839e+02 3.478e+02 4.215e+02 8.620e+02, threshold=6.955e+02, percent-clipped=3.0 2023-04-28 23:00:44,594 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80313.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:01:56,412 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80344.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:02:12,075 INFO [train.py:904] (7/8) Epoch 8, batch 9300, loss[loss=0.171, simple_loss=0.2639, pruned_loss=0.03904, over 15575.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2772, pruned_loss=0.04946, over 3056697.49 frames. ], batch size: 191, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:02:32,767 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80361.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:03:40,830 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6849, 4.3261, 4.3326, 2.9252, 3.9371, 4.2514, 3.8565, 2.1934], device='cuda:7'), covar=tensor([0.0339, 0.0022, 0.0024, 0.0269, 0.0054, 0.0062, 0.0038, 0.0380], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0060, 0.0062, 0.0118, 0.0068, 0.0077, 0.0068, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-28 23:03:42,307 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80393.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:03:55,746 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2023-04-28 23:03:56,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4118, 3.6171, 1.7930, 3.8119, 2.6275, 3.7421, 2.0290, 2.8319], device='cuda:7'), covar=tensor([0.0205, 0.0253, 0.1575, 0.0116, 0.0724, 0.0495, 0.1478, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0149, 0.0177, 0.0099, 0.0159, 0.0185, 0.0187, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-28 23:03:57,486 INFO [train.py:904] (7/8) Epoch 8, batch 9350, loss[loss=0.1929, simple_loss=0.2864, pruned_loss=0.04973, over 16772.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.277, pruned_loss=0.04943, over 3064682.91 frames. ], batch size: 124, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:04:22,277 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.575e+02 3.007e+02 3.554e+02 5.975e+02, threshold=6.013e+02, percent-clipped=0.0 2023-04-28 23:04:34,127 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80418.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:05:40,200 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9761, 4.1938, 4.0257, 4.0717, 3.7434, 3.8010, 3.8623, 4.1936], device='cuda:7'), covar=tensor([0.0833, 0.0878, 0.0917, 0.0562, 0.0651, 0.1421, 0.0786, 0.0890], device='cuda:7'), in_proj_covar=tensor([0.0439, 0.0555, 0.0464, 0.0380, 0.0345, 0.0370, 0.0463, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:05:40,912 INFO [train.py:904] (7/8) Epoch 8, batch 9400, loss[loss=0.17, simple_loss=0.2521, pruned_loss=0.04399, over 12555.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2771, pruned_loss=0.04937, over 3051388.42 frames. ], batch size: 250, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:05:46,309 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80454.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:06:05,129 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80463.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:06:10,060 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80466.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:06:49,879 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1574, 3.2805, 3.3234, 1.7455, 3.5747, 3.5962, 2.8245, 2.7210], device='cuda:7'), covar=tensor([0.0782, 0.0151, 0.0141, 0.1132, 0.0054, 0.0105, 0.0372, 0.0385], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0094, 0.0080, 0.0137, 0.0064, 0.0086, 0.0115, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-28 23:07:07,307 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80494.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:07:09,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3616, 3.4750, 1.8466, 3.7655, 2.4549, 3.6255, 2.0211, 2.7778], device='cuda:7'), covar=tensor([0.0208, 0.0296, 0.1675, 0.0118, 0.0911, 0.0580, 0.1450, 0.0634], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0151, 0.0178, 0.0099, 0.0161, 0.0186, 0.0189, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-28 23:07:19,877 INFO [train.py:904] (7/8) Epoch 8, batch 9450, loss[loss=0.1835, simple_loss=0.2778, pruned_loss=0.04455, over 15356.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2796, pruned_loss=0.04999, over 3047589.45 frames. ], batch size: 191, lr: 8.33e-03, grad_scale: 4.0 2023-04-28 23:07:38,815 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.597e+02 3.133e+02 4.086e+02 1.022e+03, threshold=6.266e+02, percent-clipped=6.0 2023-04-28 23:08:04,822 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80524.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:08:58,874 INFO [train.py:904] (7/8) Epoch 8, batch 9500, loss[loss=0.1934, simple_loss=0.2857, pruned_loss=0.05057, over 16120.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2786, pruned_loss=0.04943, over 3065159.21 frames. ], batch size: 165, lr: 8.32e-03, grad_scale: 4.0 2023-04-28 23:09:08,391 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80555.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:09:47,446 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-28 23:10:20,927 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4234, 1.9870, 2.1289, 4.0387, 1.9306, 2.5446, 2.0774, 2.1485], device='cuda:7'), covar=tensor([0.0721, 0.3027, 0.1899, 0.0308, 0.3510, 0.1820, 0.2864, 0.2962], device='cuda:7'), in_proj_covar=tensor([0.0330, 0.0351, 0.0301, 0.0308, 0.0387, 0.0386, 0.0317, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:10:46,570 INFO [train.py:904] (7/8) Epoch 8, batch 9550, loss[loss=0.1939, simple_loss=0.2769, pruned_loss=0.05547, over 12435.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2777, pruned_loss=0.04915, over 3079233.41 frames. ], batch size: 250, lr: 8.32e-03, grad_scale: 4.0 2023-04-28 23:11:10,121 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.584e+02 3.110e+02 3.705e+02 8.746e+02, threshold=6.220e+02, percent-clipped=3.0 2023-04-28 23:11:22,135 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-28 23:12:04,488 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80639.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:12:25,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7811, 1.4630, 2.2185, 2.6740, 2.3753, 2.8044, 1.5093, 2.9280], device='cuda:7'), covar=tensor([0.0090, 0.0336, 0.0178, 0.0138, 0.0165, 0.0110, 0.0390, 0.0057], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0157, 0.0142, 0.0139, 0.0146, 0.0103, 0.0155, 0.0094], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 23:12:27,102 INFO [train.py:904] (7/8) Epoch 8, batch 9600, loss[loss=0.2053, simple_loss=0.2752, pruned_loss=0.06763, over 11871.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2785, pruned_loss=0.04983, over 3056787.36 frames. ], batch size: 248, lr: 8.32e-03, grad_scale: 8.0 2023-04-28 23:13:00,206 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5769, 4.5681, 4.3715, 3.9386, 4.3838, 1.6451, 4.2015, 4.3101], device='cuda:7'), covar=tensor([0.0065, 0.0052, 0.0116, 0.0227, 0.0080, 0.2171, 0.0101, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0105, 0.0093, 0.0135, 0.0127, 0.0108, 0.0161, 0.0123, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-28 23:14:15,062 INFO [train.py:904] (7/8) Epoch 8, batch 9650, loss[loss=0.1951, simple_loss=0.2761, pruned_loss=0.05707, over 12591.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2801, pruned_loss=0.04978, over 3050961.65 frames. ], batch size: 250, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:14:42,864 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.581e+02 3.070e+02 3.886e+02 7.619e+02, threshold=6.139e+02, percent-clipped=2.0 2023-04-28 23:15:30,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2121, 3.2912, 3.6379, 3.6086, 3.6112, 3.3875, 3.4566, 3.4816], device='cuda:7'), covar=tensor([0.0377, 0.0758, 0.0457, 0.0467, 0.0496, 0.0459, 0.0724, 0.0388], device='cuda:7'), in_proj_covar=tensor([0.0274, 0.0275, 0.0279, 0.0274, 0.0314, 0.0296, 0.0383, 0.0241], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-28 23:15:30,924 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80736.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:15:59,710 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80749.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:16:03,281 INFO [train.py:904] (7/8) Epoch 8, batch 9700, loss[loss=0.1941, simple_loss=0.2853, pruned_loss=0.05146, over 15296.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2788, pruned_loss=0.04943, over 3049760.17 frames. ], batch size: 190, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:16:46,536 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80773.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:17:38,603 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80797.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:17:46,358 INFO [train.py:904] (7/8) Epoch 8, batch 9750, loss[loss=0.1903, simple_loss=0.27, pruned_loss=0.05528, over 12010.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2778, pruned_loss=0.04953, over 3053119.58 frames. ], batch size: 246, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:17:49,614 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-04-28 23:18:01,195 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4010, 3.3364, 3.4181, 3.5356, 3.5571, 3.2476, 3.5334, 3.6005], device='cuda:7'), covar=tensor([0.0894, 0.0738, 0.0953, 0.0553, 0.0553, 0.1996, 0.0734, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0533, 0.0653, 0.0545, 0.0409, 0.0404, 0.0428, 0.0470], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:18:08,283 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.771e+02 2.800e+02 3.435e+02 4.029e+02 7.858e+02, threshold=6.871e+02, percent-clipped=2.0 2023-04-28 23:18:21,095 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80819.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:18:55,756 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80834.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:19:18,781 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2908, 4.0890, 4.3585, 4.4799, 4.6018, 4.1972, 4.5934, 4.6082], device='cuda:7'), covar=tensor([0.1217, 0.0962, 0.1152, 0.0541, 0.0439, 0.0795, 0.0425, 0.0401], device='cuda:7'), in_proj_covar=tensor([0.0434, 0.0531, 0.0649, 0.0542, 0.0407, 0.0401, 0.0425, 0.0466], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:19:22,175 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3668, 2.9539, 2.6188, 2.2383, 2.1469, 2.1094, 2.8581, 2.8621], device='cuda:7'), covar=tensor([0.2150, 0.0661, 0.1219, 0.1773, 0.2078, 0.1708, 0.0420, 0.0923], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0240, 0.0265, 0.0252, 0.0243, 0.0202, 0.0243, 0.0255], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:19:24,290 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80850.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:19:26,331 INFO [train.py:904] (7/8) Epoch 8, batch 9800, loss[loss=0.1823, simple_loss=0.2826, pruned_loss=0.04097, over 16833.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2773, pruned_loss=0.04854, over 3058702.63 frames. ], batch size: 90, lr: 8.31e-03, grad_scale: 8.0 2023-04-28 23:19:38,197 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4086, 4.4600, 4.6011, 4.4553, 4.5016, 4.9953, 4.6444, 4.2420], device='cuda:7'), covar=tensor([0.1133, 0.1709, 0.1410, 0.1781, 0.2318, 0.0864, 0.1261, 0.2594], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0414, 0.0441, 0.0367, 0.0480, 0.0459, 0.0357, 0.0482], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:19:39,569 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2120, 3.9098, 4.1009, 4.3760, 4.5075, 4.1904, 4.5789, 4.5268], device='cuda:7'), covar=tensor([0.1227, 0.1124, 0.1586, 0.0781, 0.0720, 0.0877, 0.0577, 0.0702], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0533, 0.0651, 0.0545, 0.0408, 0.0402, 0.0427, 0.0468], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:21:11,985 INFO [train.py:904] (7/8) Epoch 8, batch 9850, loss[loss=0.1741, simple_loss=0.2692, pruned_loss=0.03954, over 17279.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2785, pruned_loss=0.04837, over 3055732.56 frames. ], batch size: 52, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:21:23,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9071, 2.3670, 2.3196, 3.0861, 2.2158, 3.3676, 1.6049, 2.7846], device='cuda:7'), covar=tensor([0.1160, 0.0547, 0.0951, 0.0130, 0.0090, 0.0366, 0.1319, 0.0615], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0150, 0.0170, 0.0113, 0.0175, 0.0197, 0.0169, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 23:21:33,320 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.396e+02 3.091e+02 3.765e+02 1.204e+03, threshold=6.182e+02, percent-clipped=3.0 2023-04-28 23:22:37,462 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80939.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:23:04,046 INFO [train.py:904] (7/8) Epoch 8, batch 9900, loss[loss=0.2085, simple_loss=0.3014, pruned_loss=0.05783, over 16914.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.279, pruned_loss=0.04868, over 3050702.08 frames. ], batch size: 116, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:24:29,640 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=80987.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:24:42,451 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3193, 4.4042, 4.5170, 4.4730, 4.4473, 4.9642, 4.5534, 4.2574], device='cuda:7'), covar=tensor([0.1173, 0.1534, 0.1603, 0.1712, 0.2538, 0.0950, 0.1275, 0.2303], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0416, 0.0443, 0.0370, 0.0480, 0.0461, 0.0354, 0.0482], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:25:03,340 INFO [train.py:904] (7/8) Epoch 8, batch 9950, loss[loss=0.2011, simple_loss=0.2894, pruned_loss=0.05642, over 12370.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2821, pruned_loss=0.04935, over 3060807.11 frames. ], batch size: 248, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:25:29,451 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.486e+02 3.063e+02 3.674e+02 7.431e+02, threshold=6.127e+02, percent-clipped=1.0 2023-04-28 23:27:02,307 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81049.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:27:07,185 INFO [train.py:904] (7/8) Epoch 8, batch 10000, loss[loss=0.1918, simple_loss=0.2871, pruned_loss=0.04823, over 16688.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2801, pruned_loss=0.04825, over 3095499.36 frames. ], batch size: 134, lr: 8.30e-03, grad_scale: 8.0 2023-04-28 23:27:19,245 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5997, 1.4800, 2.0567, 2.6194, 2.4238, 2.6380, 1.9072, 2.7088], device='cuda:7'), covar=tensor([0.0108, 0.0355, 0.0198, 0.0156, 0.0178, 0.0132, 0.0292, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0158, 0.0143, 0.0139, 0.0147, 0.0104, 0.0156, 0.0094], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-28 23:28:30,847 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81092.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:28:40,811 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81097.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:28:47,990 INFO [train.py:904] (7/8) Epoch 8, batch 10050, loss[loss=0.1706, simple_loss=0.2696, pruned_loss=0.0358, over 16503.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.28, pruned_loss=0.04781, over 3094860.02 frames. ], batch size: 68, lr: 8.29e-03, grad_scale: 8.0 2023-04-28 23:29:08,282 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.518e+02 2.989e+02 3.596e+02 8.754e+02, threshold=5.978e+02, percent-clipped=2.0 2023-04-28 23:29:15,230 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:29:21,147 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81119.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:29:40,305 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81129.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:30:18,858 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81150.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:30:21,540 INFO [train.py:904] (7/8) Epoch 8, batch 10100, loss[loss=0.1657, simple_loss=0.258, pruned_loss=0.03671, over 16849.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2808, pruned_loss=0.04826, over 3105269.20 frames. ], batch size: 96, lr: 8.29e-03, grad_scale: 8.0 2023-04-28 23:30:44,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1708, 4.0468, 4.2667, 4.3888, 4.5326, 4.1515, 4.5380, 4.5019], device='cuda:7'), covar=tensor([0.1245, 0.0841, 0.1239, 0.0622, 0.0497, 0.0857, 0.0485, 0.0591], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0534, 0.0653, 0.0543, 0.0408, 0.0403, 0.0428, 0.0471], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:30:51,850 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81167.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:31:14,358 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81177.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:31:37,454 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81198.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:32:08,879 INFO [train.py:904] (7/8) Epoch 9, batch 0, loss[loss=0.2866, simple_loss=0.3411, pruned_loss=0.116, over 16288.00 frames. ], tot_loss[loss=0.2866, simple_loss=0.3411, pruned_loss=0.116, over 16288.00 frames. ], batch size: 165, lr: 7.85e-03, grad_scale: 8.0 2023-04-28 23:32:08,880 INFO [train.py:929] (7/8) Computing validation loss 2023-04-28 23:32:16,264 INFO [train.py:938] (7/8) Epoch 9, validation: loss=0.1602, simple_loss=0.2637, pruned_loss=0.02837, over 944034.00 frames. 2023-04-28 23:32:16,265 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-28 23:32:36,788 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.828e+02 3.571e+02 4.681e+02 1.164e+03, threshold=7.142e+02, percent-clipped=9.0 2023-04-28 23:33:25,088 INFO [train.py:904] (7/8) Epoch 9, batch 50, loss[loss=0.2076, simple_loss=0.2765, pruned_loss=0.06937, over 16859.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.2939, pruned_loss=0.06888, over 747157.91 frames. ], batch size: 96, lr: 7.85e-03, grad_scale: 1.0 2023-04-28 23:34:31,161 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81299.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:34:34,774 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81301.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:34:35,497 INFO [train.py:904] (7/8) Epoch 9, batch 100, loss[loss=0.2159, simple_loss=0.3035, pruned_loss=0.06421, over 17067.00 frames. ], tot_loss[loss=0.2133, simple_loss=0.2915, pruned_loss=0.06759, over 1311657.99 frames. ], batch size: 55, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:34:54,539 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.676e+02 3.088e+02 4.301e+02 1.054e+03, threshold=6.177e+02, percent-clipped=2.0 2023-04-28 23:35:40,982 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-28 23:35:42,951 INFO [train.py:904] (7/8) Epoch 9, batch 150, loss[loss=0.1812, simple_loss=0.2641, pruned_loss=0.04915, over 17238.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2861, pruned_loss=0.06381, over 1756966.68 frames. ], batch size: 45, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:35:55,902 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81360.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:35:58,309 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81362.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:36:40,144 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81392.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:36:54,435 INFO [train.py:904] (7/8) Epoch 9, batch 200, loss[loss=0.2131, simple_loss=0.3103, pruned_loss=0.05802, over 16683.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2865, pruned_loss=0.0638, over 2106066.32 frames. ], batch size: 57, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:37:13,045 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.694e+02 3.180e+02 3.951e+02 1.154e+03, threshold=6.361e+02, percent-clipped=2.0 2023-04-28 23:37:32,076 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81429.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:37:46,219 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81440.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:38:01,970 INFO [train.py:904] (7/8) Epoch 9, batch 250, loss[loss=0.1948, simple_loss=0.2699, pruned_loss=0.05984, over 16832.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2838, pruned_loss=0.06292, over 2375483.12 frames. ], batch size: 96, lr: 7.84e-03, grad_scale: 1.0 2023-04-28 23:38:02,740 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.07 vs. limit=5.0 2023-04-28 23:38:17,739 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-28 23:38:29,720 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81472.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:38:36,873 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81477.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:39:10,407 INFO [train.py:904] (7/8) Epoch 9, batch 300, loss[loss=0.1625, simple_loss=0.2466, pruned_loss=0.03919, over 15941.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2808, pruned_loss=0.06151, over 2587437.99 frames. ], batch size: 35, lr: 7.83e-03, grad_scale: 1.0 2023-04-28 23:39:29,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.475e+02 2.919e+02 3.755e+02 7.155e+02, threshold=5.837e+02, percent-clipped=3.0 2023-04-28 23:40:17,834 INFO [train.py:904] (7/8) Epoch 9, batch 350, loss[loss=0.2062, simple_loss=0.2937, pruned_loss=0.05933, over 17103.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2794, pruned_loss=0.06059, over 2742608.08 frames. ], batch size: 48, lr: 7.83e-03, grad_scale: 1.0 2023-04-28 23:41:07,944 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-28 23:41:08,706 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7190, 4.7441, 4.5757, 4.4236, 3.9705, 4.7095, 4.6878, 4.2671], device='cuda:7'), covar=tensor([0.0709, 0.0889, 0.0383, 0.0324, 0.1187, 0.0878, 0.0425, 0.0866], device='cuda:7'), in_proj_covar=tensor([0.0226, 0.0264, 0.0260, 0.0233, 0.0286, 0.0267, 0.0179, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:41:23,826 INFO [train.py:904] (7/8) Epoch 9, batch 400, loss[loss=0.2264, simple_loss=0.303, pruned_loss=0.07489, over 16871.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2773, pruned_loss=0.05934, over 2873094.99 frames. ], batch size: 96, lr: 7.83e-03, grad_scale: 2.0 2023-04-28 23:41:43,594 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.386e+02 2.843e+02 3.463e+02 6.249e+02, threshold=5.687e+02, percent-clipped=1.0 2023-04-28 23:42:22,408 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-28 23:42:28,137 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-28 23:42:33,330 INFO [train.py:904] (7/8) Epoch 9, batch 450, loss[loss=0.1946, simple_loss=0.2856, pruned_loss=0.05178, over 17115.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2749, pruned_loss=0.05765, over 2973764.36 frames. ], batch size: 49, lr: 7.83e-03, grad_scale: 2.0 2023-04-28 23:42:37,371 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81655.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:42:40,764 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81657.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:42:58,812 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81670.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:43:42,348 INFO [train.py:904] (7/8) Epoch 9, batch 500, loss[loss=0.1852, simple_loss=0.2766, pruned_loss=0.04692, over 16670.00 frames. ], tot_loss[loss=0.193, simple_loss=0.273, pruned_loss=0.05654, over 3051267.00 frames. ], batch size: 57, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:44:01,903 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.446e+02 2.865e+02 3.583e+02 8.926e+02, threshold=5.729e+02, percent-clipped=4.0 2023-04-28 23:44:17,632 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-28 23:44:22,692 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81731.0, num_to_drop=1, layers_to_drop={2} 2023-04-28 23:44:50,839 INFO [train.py:904] (7/8) Epoch 9, batch 550, loss[loss=0.1661, simple_loss=0.2512, pruned_loss=0.04052, over 16835.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.272, pruned_loss=0.05591, over 3117913.52 frames. ], batch size: 42, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:45:20,615 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81772.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:45:40,389 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-28 23:46:02,617 INFO [train.py:904] (7/8) Epoch 9, batch 600, loss[loss=0.2065, simple_loss=0.2965, pruned_loss=0.05824, over 17062.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2723, pruned_loss=0.05662, over 3163111.11 frames. ], batch size: 53, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:46:15,413 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2053, 3.3890, 3.5859, 3.5564, 3.5847, 3.3864, 3.3880, 3.4250], device='cuda:7'), covar=tensor([0.0364, 0.0571, 0.0470, 0.0460, 0.0417, 0.0413, 0.0764, 0.0471], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0310, 0.0312, 0.0303, 0.0348, 0.0330, 0.0425, 0.0266], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-28 23:46:21,006 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8889, 2.0622, 2.1611, 4.6519, 2.0085, 2.6595, 2.2118, 2.3214], device='cuda:7'), covar=tensor([0.0741, 0.3251, 0.2141, 0.0330, 0.3731, 0.2114, 0.2745, 0.3419], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0369, 0.0311, 0.0323, 0.0399, 0.0408, 0.0330, 0.0434], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:46:21,576 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.753e+02 2.670e+02 3.296e+02 4.052e+02 8.860e+02, threshold=6.592e+02, percent-clipped=6.0 2023-04-28 23:46:26,577 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=81820.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:46:28,717 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2738, 3.5245, 3.3196, 2.1590, 2.9575, 2.5077, 3.8000, 3.5702], device='cuda:7'), covar=tensor([0.0224, 0.0621, 0.0596, 0.1533, 0.0710, 0.0852, 0.0461, 0.0845], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0135, 0.0157, 0.0143, 0.0135, 0.0125, 0.0133, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-28 23:46:46,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4903, 4.4448, 4.3809, 3.2275, 4.3819, 1.6249, 4.0503, 4.0646], device='cuda:7'), covar=tensor([0.0189, 0.0149, 0.0238, 0.0733, 0.0151, 0.3300, 0.0262, 0.0377], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0104, 0.0152, 0.0144, 0.0121, 0.0173, 0.0137, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:47:09,565 INFO [train.py:904] (7/8) Epoch 9, batch 650, loss[loss=0.1868, simple_loss=0.2791, pruned_loss=0.04726, over 17135.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2706, pruned_loss=0.05661, over 3189936.30 frames. ], batch size: 49, lr: 7.82e-03, grad_scale: 2.0 2023-04-28 23:47:39,039 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8061, 4.5017, 4.5700, 5.0786, 5.2321, 4.6932, 5.2385, 5.1427], device='cuda:7'), covar=tensor([0.1501, 0.1478, 0.2341, 0.0914, 0.0846, 0.0898, 0.0762, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0503, 0.0616, 0.0764, 0.0622, 0.0469, 0.0462, 0.0494, 0.0543], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:48:18,171 INFO [train.py:904] (7/8) Epoch 9, batch 700, loss[loss=0.221, simple_loss=0.2856, pruned_loss=0.07818, over 16911.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2705, pruned_loss=0.05582, over 3230052.02 frames. ], batch size: 109, lr: 7.81e-03, grad_scale: 2.0 2023-04-28 23:48:37,201 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.553e+02 3.037e+02 4.738e+02 2.852e+03, threshold=6.073e+02, percent-clipped=12.0 2023-04-28 23:49:25,164 INFO [train.py:904] (7/8) Epoch 9, batch 750, loss[loss=0.2079, simple_loss=0.2775, pruned_loss=0.06914, over 16696.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2709, pruned_loss=0.05617, over 3246752.61 frames. ], batch size: 134, lr: 7.81e-03, grad_scale: 2.0 2023-04-28 23:49:29,115 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81955.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:49:31,610 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81957.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:06,028 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8218, 3.6886, 3.8770, 3.6947, 3.7534, 4.2308, 3.9506, 3.5971], device='cuda:7'), covar=tensor([0.1891, 0.2299, 0.1963, 0.2515, 0.3293, 0.1762, 0.1346, 0.2470], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0465, 0.0497, 0.0412, 0.0550, 0.0522, 0.0395, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 23:50:14,429 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81987.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:31,368 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7218, 3.2842, 2.6569, 5.0463, 4.1370, 4.6146, 1.6110, 3.2360], device='cuda:7'), covar=tensor([0.1261, 0.0539, 0.1108, 0.0093, 0.0255, 0.0302, 0.1362, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0153, 0.0174, 0.0124, 0.0190, 0.0208, 0.0173, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-28 23:50:37,833 INFO [train.py:904] (7/8) Epoch 9, batch 800, loss[loss=0.1948, simple_loss=0.2633, pruned_loss=0.06316, over 16837.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2713, pruned_loss=0.05649, over 3264598.25 frames. ], batch size: 116, lr: 7.81e-03, grad_scale: 4.0 2023-04-28 23:50:39,979 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82003.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:42,265 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82005.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:50:56,931 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.619e+02 2.968e+02 3.371e+02 5.495e+02, threshold=5.935e+02, percent-clipped=0.0 2023-04-28 23:51:10,880 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82026.0, num_to_drop=1, layers_to_drop={1} 2023-04-28 23:51:40,939 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82048.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:51:43,245 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7773, 2.3261, 1.5903, 1.9913, 2.8327, 2.5684, 3.0658, 2.8921], device='cuda:7'), covar=tensor([0.0131, 0.0320, 0.0473, 0.0384, 0.0161, 0.0247, 0.0181, 0.0184], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0195, 0.0190, 0.0191, 0.0190, 0.0194, 0.0194, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:51:45,047 INFO [train.py:904] (7/8) Epoch 9, batch 850, loss[loss=0.211, simple_loss=0.2992, pruned_loss=0.06137, over 17093.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2704, pruned_loss=0.05597, over 3277209.46 frames. ], batch size: 55, lr: 7.81e-03, grad_scale: 4.0 2023-04-28 23:51:49,665 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82055.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:51:56,947 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-04-28 23:52:54,663 INFO [train.py:904] (7/8) Epoch 9, batch 900, loss[loss=0.1845, simple_loss=0.2763, pruned_loss=0.0464, over 17119.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2699, pruned_loss=0.05512, over 3289532.80 frames. ], batch size: 49, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:53:13,857 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.604e+02 3.157e+02 3.589e+02 5.638e+02, threshold=6.315e+02, percent-clipped=0.0 2023-04-28 23:53:15,023 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82116.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:53:32,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9674, 5.1957, 4.8812, 4.6270, 3.9849, 5.0498, 5.2813, 4.5796], device='cuda:7'), covar=tensor([0.0907, 0.0374, 0.0444, 0.0350, 0.2213, 0.0436, 0.0276, 0.0687], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0277, 0.0273, 0.0245, 0.0301, 0.0281, 0.0189, 0.0317], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-28 23:54:03,815 INFO [train.py:904] (7/8) Epoch 9, batch 950, loss[loss=0.1796, simple_loss=0.2661, pruned_loss=0.04649, over 17130.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2694, pruned_loss=0.05481, over 3282572.88 frames. ], batch size: 49, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:55:11,105 INFO [train.py:904] (7/8) Epoch 9, batch 1000, loss[loss=0.1645, simple_loss=0.249, pruned_loss=0.03997, over 17230.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2676, pruned_loss=0.05418, over 3286063.87 frames. ], batch size: 45, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:55:31,955 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.342e+02 2.920e+02 3.464e+02 5.943e+02, threshold=5.839e+02, percent-clipped=0.0 2023-04-28 23:56:20,447 INFO [train.py:904] (7/8) Epoch 9, batch 1050, loss[loss=0.1778, simple_loss=0.2696, pruned_loss=0.04299, over 17119.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2668, pruned_loss=0.0538, over 3295747.87 frames. ], batch size: 48, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:57:18,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0674, 4.8067, 5.0640, 5.3149, 5.5061, 4.8453, 5.4208, 5.4248], device='cuda:7'), covar=tensor([0.1402, 0.1004, 0.1506, 0.0513, 0.0394, 0.0717, 0.0422, 0.0422], device='cuda:7'), in_proj_covar=tensor([0.0512, 0.0625, 0.0785, 0.0638, 0.0477, 0.0473, 0.0504, 0.0554], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-28 23:57:28,539 INFO [train.py:904] (7/8) Epoch 9, batch 1100, loss[loss=0.1925, simple_loss=0.2618, pruned_loss=0.06158, over 16910.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2663, pruned_loss=0.05393, over 3305088.95 frames. ], batch size: 116, lr: 7.80e-03, grad_scale: 4.0 2023-04-28 23:57:47,217 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.514e+02 3.056e+02 3.616e+02 1.290e+03, threshold=6.113e+02, percent-clipped=7.0 2023-04-28 23:58:01,828 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82326.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:58:24,293 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82343.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:58:28,024 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82346.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:58:35,320 INFO [train.py:904] (7/8) Epoch 9, batch 1150, loss[loss=0.2003, simple_loss=0.2703, pruned_loss=0.06518, over 16735.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2665, pruned_loss=0.0537, over 3313212.60 frames. ], batch size: 124, lr: 7.79e-03, grad_scale: 4.0 2023-04-28 23:59:04,944 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82374.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:59:44,531 INFO [train.py:904] (7/8) Epoch 9, batch 1200, loss[loss=0.1598, simple_loss=0.2533, pruned_loss=0.03318, over 17099.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2656, pruned_loss=0.05302, over 3312695.32 frames. ], batch size: 47, lr: 7.79e-03, grad_scale: 8.0 2023-04-28 23:59:50,739 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82407.0, num_to_drop=0, layers_to_drop=set() 2023-04-28 23:59:55,316 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82411.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:00:02,694 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.375e+02 3.047e+02 3.937e+02 1.504e+03, threshold=6.095e+02, percent-clipped=2.0 2023-04-29 00:00:17,302 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 00:00:37,533 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6412, 2.9782, 2.5854, 4.7754, 3.7455, 4.4926, 1.7035, 3.0766], device='cuda:7'), covar=tensor([0.1454, 0.0696, 0.1206, 0.0175, 0.0335, 0.0367, 0.1481, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0154, 0.0174, 0.0126, 0.0195, 0.0211, 0.0173, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 00:00:50,247 INFO [train.py:904] (7/8) Epoch 9, batch 1250, loss[loss=0.1988, simple_loss=0.2729, pruned_loss=0.06236, over 16533.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.266, pruned_loss=0.05343, over 3311616.04 frames. ], batch size: 68, lr: 7.79e-03, grad_scale: 4.0 2023-04-29 00:01:47,128 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82493.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 00:01:53,840 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-04-29 00:01:58,480 INFO [train.py:904] (7/8) Epoch 9, batch 1300, loss[loss=0.2071, simple_loss=0.2716, pruned_loss=0.07129, over 16747.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2661, pruned_loss=0.05409, over 3307449.45 frames. ], batch size: 134, lr: 7.79e-03, grad_scale: 4.0 2023-04-29 00:02:18,041 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.404e+02 2.965e+02 3.987e+02 6.881e+02, threshold=5.930e+02, percent-clipped=4.0 2023-04-29 00:02:32,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6185, 2.6685, 2.3428, 3.6729, 3.1158, 3.8377, 1.4853, 2.7494], device='cuda:7'), covar=tensor([0.1335, 0.0577, 0.1173, 0.0166, 0.0254, 0.0398, 0.1420, 0.0805], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0154, 0.0174, 0.0125, 0.0195, 0.0210, 0.0173, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 00:02:39,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8669, 4.6294, 4.8778, 5.0801, 5.2308, 4.5886, 5.1989, 5.1957], device='cuda:7'), covar=tensor([0.1311, 0.0944, 0.1366, 0.0524, 0.0479, 0.0775, 0.0523, 0.0408], device='cuda:7'), in_proj_covar=tensor([0.0523, 0.0638, 0.0801, 0.0654, 0.0490, 0.0485, 0.0514, 0.0566], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:03:05,227 INFO [train.py:904] (7/8) Epoch 9, batch 1350, loss[loss=0.1733, simple_loss=0.2628, pruned_loss=0.04188, over 15521.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2663, pruned_loss=0.0541, over 3306867.55 frames. ], batch size: 191, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:03:08,030 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82554.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 00:03:29,841 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82569.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:04:12,578 INFO [train.py:904] (7/8) Epoch 9, batch 1400, loss[loss=0.1702, simple_loss=0.2584, pruned_loss=0.04107, over 17091.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2656, pruned_loss=0.05372, over 3302724.33 frames. ], batch size: 55, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:04:33,570 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.260e+02 2.970e+02 4.048e+02 1.598e+03, threshold=5.940e+02, percent-clipped=4.0 2023-04-29 00:04:53,143 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82630.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:05:00,354 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 00:05:09,610 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82643.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:05:22,666 INFO [train.py:904] (7/8) Epoch 9, batch 1450, loss[loss=0.1693, simple_loss=0.2445, pruned_loss=0.04705, over 15592.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2655, pruned_loss=0.054, over 3308519.32 frames. ], batch size: 191, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:06:01,098 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 00:06:03,764 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 00:06:04,981 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 00:06:16,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:06:26,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0388, 5.4915, 5.6400, 5.4761, 5.4511, 6.0213, 5.5976, 5.2823], device='cuda:7'), covar=tensor([0.0786, 0.1790, 0.1625, 0.1915, 0.2845, 0.0945, 0.1203, 0.2247], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0471, 0.0503, 0.0417, 0.0555, 0.0527, 0.0398, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 00:06:29,722 INFO [train.py:904] (7/8) Epoch 9, batch 1500, loss[loss=0.1479, simple_loss=0.2286, pruned_loss=0.03362, over 16756.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2655, pruned_loss=0.05406, over 3316543.90 frames. ], batch size: 39, lr: 7.78e-03, grad_scale: 4.0 2023-04-29 00:06:29,982 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82702.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:06:42,904 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82711.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:06:51,616 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.540e+02 3.048e+02 3.562e+02 8.559e+02, threshold=6.096e+02, percent-clipped=4.0 2023-04-29 00:06:52,038 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2170, 5.8376, 5.9754, 5.8435, 5.8298, 6.2774, 5.9099, 5.5660], device='cuda:7'), covar=tensor([0.0711, 0.1507, 0.1509, 0.1580, 0.2056, 0.0846, 0.1155, 0.2122], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0466, 0.0499, 0.0412, 0.0550, 0.0524, 0.0395, 0.0550], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 00:07:39,186 INFO [train.py:904] (7/8) Epoch 9, batch 1550, loss[loss=0.1648, simple_loss=0.2451, pruned_loss=0.04227, over 16793.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2667, pruned_loss=0.0548, over 3316645.43 frames. ], batch size: 39, lr: 7.77e-03, grad_scale: 4.0 2023-04-29 00:07:49,786 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=82759.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:08:02,815 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 00:08:04,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9413, 4.6696, 4.9234, 5.1476, 5.3352, 4.6885, 5.3365, 5.2741], device='cuda:7'), covar=tensor([0.1298, 0.1098, 0.1625, 0.0668, 0.0477, 0.0723, 0.0442, 0.0467], device='cuda:7'), in_proj_covar=tensor([0.0525, 0.0640, 0.0807, 0.0658, 0.0492, 0.0485, 0.0512, 0.0569], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:08:24,262 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8557, 4.3395, 4.4149, 3.2343, 3.8689, 4.3649, 3.8788, 2.6916], device='cuda:7'), covar=tensor([0.0334, 0.0031, 0.0028, 0.0223, 0.0064, 0.0063, 0.0063, 0.0287], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0067, 0.0066, 0.0121, 0.0071, 0.0081, 0.0072, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 00:08:48,754 INFO [train.py:904] (7/8) Epoch 9, batch 1600, loss[loss=0.1905, simple_loss=0.2768, pruned_loss=0.05207, over 16016.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2684, pruned_loss=0.05531, over 3316751.82 frames. ], batch size: 35, lr: 7.77e-03, grad_scale: 8.0 2023-04-29 00:09:09,716 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.591e+02 3.276e+02 4.037e+02 8.145e+02, threshold=6.551e+02, percent-clipped=5.0 2023-04-29 00:09:52,326 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82849.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 00:09:56,034 INFO [train.py:904] (7/8) Epoch 9, batch 1650, loss[loss=0.2159, simple_loss=0.3048, pruned_loss=0.06352, over 16688.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2711, pruned_loss=0.05662, over 3312852.62 frames. ], batch size: 62, lr: 7.77e-03, grad_scale: 8.0 2023-04-29 00:10:01,473 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5450, 5.5273, 5.3272, 4.7133, 5.3769, 2.5392, 5.1426, 5.4136], device='cuda:7'), covar=tensor([0.0051, 0.0045, 0.0106, 0.0304, 0.0059, 0.1727, 0.0088, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0111, 0.0161, 0.0155, 0.0130, 0.0176, 0.0145, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:10:54,406 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 00:11:05,719 INFO [train.py:904] (7/8) Epoch 9, batch 1700, loss[loss=0.1647, simple_loss=0.2568, pruned_loss=0.03632, over 17175.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2735, pruned_loss=0.05713, over 3310580.09 frames. ], batch size: 46, lr: 7.77e-03, grad_scale: 8.0 2023-04-29 00:11:24,537 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 2.790e+02 3.377e+02 4.170e+02 9.157e+02, threshold=6.754e+02, percent-clipped=3.0 2023-04-29 00:11:37,186 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82925.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:12:13,191 INFO [train.py:904] (7/8) Epoch 9, batch 1750, loss[loss=0.2152, simple_loss=0.2974, pruned_loss=0.06653, over 16207.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2747, pruned_loss=0.05703, over 3313378.53 frames. ], batch size: 164, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:12:28,851 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82963.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:12:34,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1817, 3.0653, 3.4481, 2.3567, 3.1743, 3.5409, 3.2638, 1.9059], device='cuda:7'), covar=tensor([0.0339, 0.0105, 0.0036, 0.0252, 0.0064, 0.0049, 0.0058, 0.0320], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0067, 0.0066, 0.0121, 0.0071, 0.0080, 0.0071, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 00:13:19,623 INFO [train.py:904] (7/8) Epoch 9, batch 1800, loss[loss=0.1975, simple_loss=0.2928, pruned_loss=0.05108, over 17237.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2762, pruned_loss=0.05731, over 3313627.17 frames. ], batch size: 52, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:13:20,690 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:13:39,565 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3893, 3.4669, 3.5942, 1.9923, 3.7756, 3.7661, 3.1105, 2.8409], device='cuda:7'), covar=tensor([0.0723, 0.0149, 0.0143, 0.0992, 0.0057, 0.0116, 0.0313, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0097, 0.0086, 0.0141, 0.0069, 0.0097, 0.0119, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 00:13:40,267 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.771e+02 2.500e+02 2.970e+02 3.668e+02 6.624e+02, threshold=5.940e+02, percent-clipped=0.0 2023-04-29 00:13:50,976 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83024.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:14:26,662 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83050.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:14:28,696 INFO [train.py:904] (7/8) Epoch 9, batch 1850, loss[loss=0.2111, simple_loss=0.3042, pruned_loss=0.05897, over 16748.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2764, pruned_loss=0.05703, over 3324529.68 frames. ], batch size: 57, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:15:37,693 INFO [train.py:904] (7/8) Epoch 9, batch 1900, loss[loss=0.1804, simple_loss=0.2737, pruned_loss=0.04359, over 17104.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2748, pruned_loss=0.0563, over 3321385.00 frames. ], batch size: 47, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:15:59,153 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.589e+02 2.476e+02 2.876e+02 3.479e+02 6.930e+02, threshold=5.751e+02, percent-clipped=2.0 2023-04-29 00:16:43,375 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83149.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 00:16:46,840 INFO [train.py:904] (7/8) Epoch 9, batch 1950, loss[loss=0.2311, simple_loss=0.3117, pruned_loss=0.0752, over 12088.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2749, pruned_loss=0.05594, over 3319055.45 frames. ], batch size: 246, lr: 7.76e-03, grad_scale: 8.0 2023-04-29 00:17:48,389 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83196.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:17:49,340 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83197.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 00:17:55,092 INFO [train.py:904] (7/8) Epoch 9, batch 2000, loss[loss=0.2232, simple_loss=0.2896, pruned_loss=0.07842, over 16777.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.274, pruned_loss=0.05548, over 3323332.84 frames. ], batch size: 83, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:18:17,513 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.480e+02 2.857e+02 3.443e+02 6.900e+02, threshold=5.715e+02, percent-clipped=1.0 2023-04-29 00:18:26,864 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83225.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:19:04,153 INFO [train.py:904] (7/8) Epoch 9, batch 2050, loss[loss=0.173, simple_loss=0.258, pruned_loss=0.04397, over 17229.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2737, pruned_loss=0.0555, over 3316334.00 frames. ], batch size: 45, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:19:12,821 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83257.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:19:33,984 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83273.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:19:48,950 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83284.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:20:13,209 INFO [train.py:904] (7/8) Epoch 9, batch 2100, loss[loss=0.1543, simple_loss=0.2455, pruned_loss=0.03156, over 17241.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2749, pruned_loss=0.05673, over 3310706.83 frames. ], batch size: 46, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:20:35,028 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.645e+02 3.321e+02 4.089e+02 1.048e+03, threshold=6.642e+02, percent-clipped=7.0 2023-04-29 00:20:37,134 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83319.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:21:11,912 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83345.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:21:20,745 INFO [train.py:904] (7/8) Epoch 9, batch 2150, loss[loss=0.2217, simple_loss=0.2875, pruned_loss=0.07799, over 16836.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2755, pruned_loss=0.05719, over 3308066.62 frames. ], batch size: 116, lr: 7.75e-03, grad_scale: 8.0 2023-04-29 00:22:31,371 INFO [train.py:904] (7/8) Epoch 9, batch 2200, loss[loss=0.1681, simple_loss=0.2498, pruned_loss=0.04317, over 17015.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2762, pruned_loss=0.05694, over 3313600.59 frames. ], batch size: 41, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:22:52,223 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5555, 3.6349, 3.7115, 1.9191, 3.9008, 3.8868, 3.0384, 2.9742], device='cuda:7'), covar=tensor([0.0675, 0.0126, 0.0185, 0.1032, 0.0057, 0.0117, 0.0400, 0.0358], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0098, 0.0087, 0.0142, 0.0070, 0.0098, 0.0121, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 00:22:54,055 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.797e+02 3.398e+02 4.283e+02 9.169e+02, threshold=6.797e+02, percent-clipped=3.0 2023-04-29 00:23:41,050 INFO [train.py:904] (7/8) Epoch 9, batch 2250, loss[loss=0.1757, simple_loss=0.2753, pruned_loss=0.03805, over 17130.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2766, pruned_loss=0.05741, over 3323836.92 frames. ], batch size: 48, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:24:49,197 INFO [train.py:904] (7/8) Epoch 9, batch 2300, loss[loss=0.1945, simple_loss=0.2683, pruned_loss=0.06035, over 16827.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2772, pruned_loss=0.0578, over 3306100.41 frames. ], batch size: 96, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:25:12,016 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.674e+02 3.137e+02 4.026e+02 1.366e+03, threshold=6.274e+02, percent-clipped=5.0 2023-04-29 00:25:59,045 INFO [train.py:904] (7/8) Epoch 9, batch 2350, loss[loss=0.2341, simple_loss=0.3065, pruned_loss=0.08081, over 12146.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2782, pruned_loss=0.05893, over 3303692.51 frames. ], batch size: 246, lr: 7.74e-03, grad_scale: 8.0 2023-04-29 00:25:59,363 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83552.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:26:10,092 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6856, 2.8275, 2.3137, 4.0995, 3.2786, 4.0232, 1.5448, 2.7517], device='cuda:7'), covar=tensor([0.1286, 0.0596, 0.1154, 0.0136, 0.0182, 0.0381, 0.1345, 0.0803], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0154, 0.0175, 0.0128, 0.0199, 0.0210, 0.0172, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 00:26:44,569 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3570, 3.8058, 3.7084, 2.1308, 3.0462, 2.5509, 3.6101, 3.7637], device='cuda:7'), covar=tensor([0.0257, 0.0657, 0.0478, 0.1583, 0.0741, 0.0811, 0.0647, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0141, 0.0156, 0.0141, 0.0134, 0.0123, 0.0136, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 00:27:06,421 INFO [train.py:904] (7/8) Epoch 9, batch 2400, loss[loss=0.1873, simple_loss=0.2858, pruned_loss=0.04438, over 17050.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2795, pruned_loss=0.05943, over 3315171.44 frames. ], batch size: 55, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:27:29,674 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 2.500e+02 2.959e+02 3.579e+02 6.970e+02, threshold=5.919e+02, percent-clipped=2.0 2023-04-29 00:27:31,437 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83619.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:27:59,213 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83640.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:28:14,971 INFO [train.py:904] (7/8) Epoch 9, batch 2450, loss[loss=0.2432, simple_loss=0.3064, pruned_loss=0.09004, over 16860.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.28, pruned_loss=0.05868, over 3313228.75 frames. ], batch size: 96, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:28:35,098 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83667.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:29:22,521 INFO [train.py:904] (7/8) Epoch 9, batch 2500, loss[loss=0.2224, simple_loss=0.2981, pruned_loss=0.07335, over 16471.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2792, pruned_loss=0.05847, over 3314806.07 frames. ], batch size: 146, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:29:44,597 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.505e+02 2.999e+02 3.639e+02 7.354e+02, threshold=5.999e+02, percent-clipped=3.0 2023-04-29 00:30:28,785 INFO [train.py:904] (7/8) Epoch 9, batch 2550, loss[loss=0.1947, simple_loss=0.2751, pruned_loss=0.05712, over 15560.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.28, pruned_loss=0.05877, over 3315194.53 frames. ], batch size: 191, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:30:34,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7016, 2.6856, 2.2655, 2.4636, 2.9554, 2.7403, 3.5338, 3.2357], device='cuda:7'), covar=tensor([0.0055, 0.0241, 0.0271, 0.0281, 0.0145, 0.0236, 0.0122, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0192, 0.0187, 0.0192, 0.0190, 0.0194, 0.0198, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:30:41,819 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83760.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:31:13,079 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 00:31:38,617 INFO [train.py:904] (7/8) Epoch 9, batch 2600, loss[loss=0.2167, simple_loss=0.2852, pruned_loss=0.07415, over 16880.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2801, pruned_loss=0.05866, over 3310181.68 frames. ], batch size: 109, lr: 7.73e-03, grad_scale: 8.0 2023-04-29 00:31:59,380 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.494e+02 3.064e+02 3.833e+02 6.863e+02, threshold=6.129e+02, percent-clipped=2.0 2023-04-29 00:32:04,244 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83821.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:32:45,549 INFO [train.py:904] (7/8) Epoch 9, batch 2650, loss[loss=0.2012, simple_loss=0.2933, pruned_loss=0.05453, over 16500.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2801, pruned_loss=0.05757, over 3313537.25 frames. ], batch size: 75, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:32:45,888 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:33:51,932 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83900.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:33:53,992 INFO [train.py:904] (7/8) Epoch 9, batch 2700, loss[loss=0.2222, simple_loss=0.2922, pruned_loss=0.07615, over 16715.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2798, pruned_loss=0.05655, over 3310613.35 frames. ], batch size: 134, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:34:17,452 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.501e+02 2.856e+02 3.411e+02 7.585e+02, threshold=5.711e+02, percent-clipped=1.0 2023-04-29 00:34:47,165 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83940.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:34:58,193 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83949.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:35:02,470 INFO [train.py:904] (7/8) Epoch 9, batch 2750, loss[loss=0.184, simple_loss=0.2804, pruned_loss=0.04383, over 17061.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.279, pruned_loss=0.05567, over 3314110.21 frames. ], batch size: 50, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:35:46,143 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4424, 2.0857, 2.2007, 4.3035, 2.0266, 2.6430, 2.2410, 2.3612], device='cuda:7'), covar=tensor([0.0893, 0.3095, 0.1953, 0.0329, 0.3406, 0.1925, 0.2835, 0.2692], device='cuda:7'), in_proj_covar=tensor([0.0355, 0.0376, 0.0316, 0.0325, 0.0400, 0.0427, 0.0335, 0.0445], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:35:48,352 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83985.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:35:52,468 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=83988.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:36:15,271 INFO [train.py:904] (7/8) Epoch 9, batch 2800, loss[loss=0.1625, simple_loss=0.2478, pruned_loss=0.03862, over 16956.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2782, pruned_loss=0.05572, over 3316911.39 frames. ], batch size: 41, lr: 7.72e-03, grad_scale: 8.0 2023-04-29 00:36:26,411 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84010.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 00:36:26,616 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 00:36:36,298 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.751e+02 3.290e+02 3.928e+02 7.223e+02, threshold=6.581e+02, percent-clipped=1.0 2023-04-29 00:37:16,100 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84046.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:37:16,150 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6897, 4.2328, 3.9868, 2.3344, 3.1981, 3.0827, 3.8871, 4.2066], device='cuda:7'), covar=tensor([0.0238, 0.0578, 0.0478, 0.1526, 0.0734, 0.0724, 0.0654, 0.0895], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0143, 0.0157, 0.0142, 0.0134, 0.0124, 0.0135, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 00:37:23,283 INFO [train.py:904] (7/8) Epoch 9, batch 2850, loss[loss=0.3327, simple_loss=0.3652, pruned_loss=0.1501, over 11839.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2775, pruned_loss=0.05597, over 3308777.06 frames. ], batch size: 247, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:38:25,999 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84097.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:38:32,601 INFO [train.py:904] (7/8) Epoch 9, batch 2900, loss[loss=0.211, simple_loss=0.2709, pruned_loss=0.07552, over 16845.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2768, pruned_loss=0.05663, over 3308281.07 frames. ], batch size: 116, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:38:52,456 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84116.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:38:54,444 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.515e+02 3.057e+02 3.611e+02 6.139e+02, threshold=6.114e+02, percent-clipped=0.0 2023-04-29 00:39:20,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4265, 2.4071, 1.7312, 2.0514, 2.8349, 2.5876, 3.2992, 3.0903], device='cuda:7'), covar=tensor([0.0077, 0.0320, 0.0417, 0.0367, 0.0183, 0.0266, 0.0170, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0193, 0.0187, 0.0191, 0.0191, 0.0194, 0.0200, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:39:21,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3100, 2.5994, 1.9974, 2.2742, 2.9585, 2.7065, 3.2414, 3.1222], device='cuda:7'), covar=tensor([0.0078, 0.0247, 0.0307, 0.0274, 0.0136, 0.0202, 0.0154, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0194, 0.0187, 0.0191, 0.0191, 0.0194, 0.0200, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:39:37,788 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2008, 3.2711, 1.9434, 3.3786, 2.4930, 3.4494, 1.9280, 2.6114], device='cuda:7'), covar=tensor([0.0215, 0.0339, 0.1267, 0.0224, 0.0665, 0.0573, 0.1185, 0.0581], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0165, 0.0184, 0.0121, 0.0164, 0.0208, 0.0191, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 00:39:43,189 INFO [train.py:904] (7/8) Epoch 9, batch 2950, loss[loss=0.19, simple_loss=0.2761, pruned_loss=0.05195, over 17121.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.277, pruned_loss=0.05728, over 3313185.27 frames. ], batch size: 47, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:39:51,248 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:40:48,809 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84199.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:40:52,821 INFO [train.py:904] (7/8) Epoch 9, batch 3000, loss[loss=0.1985, simple_loss=0.2679, pruned_loss=0.06456, over 16863.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2774, pruned_loss=0.05803, over 3314845.40 frames. ], batch size: 109, lr: 7.71e-03, grad_scale: 8.0 2023-04-29 00:40:52,821 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 00:41:02,065 INFO [train.py:938] (7/8) Epoch 9, validation: loss=0.1444, simple_loss=0.2507, pruned_loss=0.019, over 944034.00 frames. 2023-04-29 00:41:02,065 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 00:41:21,229 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6788, 3.9301, 4.2056, 2.0495, 4.4219, 4.4553, 3.1127, 3.5514], device='cuda:7'), covar=tensor([0.0694, 0.0167, 0.0164, 0.1059, 0.0062, 0.0141, 0.0403, 0.0328], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0097, 0.0088, 0.0141, 0.0071, 0.0101, 0.0122, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 00:41:23,147 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.086e+02 2.665e+02 3.010e+02 4.048e+02 8.692e+02, threshold=6.019e+02, percent-clipped=1.0 2023-04-29 00:41:47,925 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84236.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:42:09,947 INFO [train.py:904] (7/8) Epoch 9, batch 3050, loss[loss=0.1779, simple_loss=0.2716, pruned_loss=0.04207, over 17139.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2778, pruned_loss=0.05759, over 3316834.75 frames. ], batch size: 49, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:42:11,955 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 00:42:20,915 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84260.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:43:11,826 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84297.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:43:14,521 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-29 00:43:17,166 INFO [train.py:904] (7/8) Epoch 9, batch 3100, loss[loss=0.1991, simple_loss=0.2785, pruned_loss=0.05981, over 16496.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2768, pruned_loss=0.05715, over 3325558.16 frames. ], batch size: 75, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:43:22,151 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84305.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 00:43:39,996 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.846e+02 2.519e+02 3.109e+02 3.888e+02 1.112e+03, threshold=6.217e+02, percent-clipped=5.0 2023-04-29 00:44:06,194 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:44:12,270 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84341.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:44:28,370 INFO [train.py:904] (7/8) Epoch 9, batch 3150, loss[loss=0.2067, simple_loss=0.2831, pruned_loss=0.06516, over 12276.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2759, pruned_loss=0.05677, over 3321825.83 frames. ], batch size: 247, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:45:30,335 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84397.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:45:36,495 INFO [train.py:904] (7/8) Epoch 9, batch 3200, loss[loss=0.1962, simple_loss=0.2679, pruned_loss=0.06223, over 16763.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.276, pruned_loss=0.05693, over 3319032.89 frames. ], batch size: 124, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:45:56,425 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84416.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:45:59,095 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 2.601e+02 3.159e+02 4.189e+02 1.012e+03, threshold=6.318e+02, percent-clipped=3.0 2023-04-29 00:46:34,013 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-29 00:46:45,891 INFO [train.py:904] (7/8) Epoch 9, batch 3250, loss[loss=0.2455, simple_loss=0.3113, pruned_loss=0.08989, over 12060.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2752, pruned_loss=0.0567, over 3308759.07 frames. ], batch size: 246, lr: 7.70e-03, grad_scale: 8.0 2023-04-29 00:46:47,426 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84453.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:47:03,266 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84464.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:47:14,813 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9535, 3.3836, 3.0121, 1.9116, 2.6071, 2.2797, 3.3717, 3.4477], device='cuda:7'), covar=tensor([0.0253, 0.0638, 0.0647, 0.1624, 0.0782, 0.0893, 0.0513, 0.0721], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0146, 0.0158, 0.0142, 0.0136, 0.0126, 0.0137, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 00:47:55,760 INFO [train.py:904] (7/8) Epoch 9, batch 3300, loss[loss=0.2036, simple_loss=0.2838, pruned_loss=0.06175, over 16782.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.276, pruned_loss=0.05693, over 3308164.09 frames. ], batch size: 102, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:48:18,092 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.660e+02 3.024e+02 3.740e+02 8.260e+02, threshold=6.049e+02, percent-clipped=1.0 2023-04-29 00:48:27,029 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-29 00:48:27,929 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84525.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:49:04,558 INFO [train.py:904] (7/8) Epoch 9, batch 3350, loss[loss=0.1693, simple_loss=0.2626, pruned_loss=0.03802, over 17147.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2752, pruned_loss=0.0563, over 3313640.08 frames. ], batch size: 49, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:49:10,310 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:49:18,477 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 00:49:53,245 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84586.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:50:01,513 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84592.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:50:16,103 INFO [train.py:904] (7/8) Epoch 9, batch 3400, loss[loss=0.2023, simple_loss=0.2972, pruned_loss=0.05367, over 17033.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.275, pruned_loss=0.05609, over 3312993.47 frames. ], batch size: 55, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:50:21,026 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84605.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:50:38,680 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.352e+02 2.993e+02 3.807e+02 8.102e+02, threshold=5.985e+02, percent-clipped=2.0 2023-04-29 00:50:46,957 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8233, 5.2782, 5.3921, 5.2889, 5.2061, 5.9054, 5.4023, 5.0975], device='cuda:7'), covar=tensor([0.0918, 0.1708, 0.1755, 0.1767, 0.2913, 0.0877, 0.1190, 0.2252], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0475, 0.0502, 0.0419, 0.0554, 0.0525, 0.0399, 0.0565], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 00:50:57,236 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7953, 4.1290, 3.0797, 2.4437, 2.8817, 2.3642, 4.2328, 3.7629], device='cuda:7'), covar=tensor([0.2232, 0.0520, 0.1356, 0.1928, 0.2216, 0.1615, 0.0422, 0.1009], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0254, 0.0276, 0.0266, 0.0279, 0.0214, 0.0261, 0.0289], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:50:59,423 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84633.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:51:10,742 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84641.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:51:12,038 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2403, 3.3080, 1.7490, 3.3813, 2.4857, 3.4451, 1.7977, 2.5933], device='cuda:7'), covar=tensor([0.0217, 0.0310, 0.1665, 0.0246, 0.0705, 0.0425, 0.1500, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0167, 0.0188, 0.0124, 0.0168, 0.0210, 0.0194, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 00:51:24,589 INFO [train.py:904] (7/8) Epoch 9, batch 3450, loss[loss=0.1748, simple_loss=0.253, pruned_loss=0.04828, over 16757.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2732, pruned_loss=0.05551, over 3323976.48 frames. ], batch size: 39, lr: 7.69e-03, grad_scale: 8.0 2023-04-29 00:51:26,679 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84653.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:51:46,297 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-29 00:52:16,087 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84689.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:20,956 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:24,207 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84694.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:52:35,104 INFO [train.py:904] (7/8) Epoch 9, batch 3500, loss[loss=0.1832, simple_loss=0.2698, pruned_loss=0.04831, over 17224.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2724, pruned_loss=0.05484, over 3311261.33 frames. ], batch size: 45, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:52:56,980 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.287e+02 2.792e+02 3.603e+02 8.093e+02, threshold=5.583e+02, percent-clipped=2.0 2023-04-29 00:53:21,377 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9385, 4.3996, 3.1717, 2.2301, 3.0446, 2.2451, 4.6347, 3.9843], device='cuda:7'), covar=tensor([0.2197, 0.0441, 0.1414, 0.2166, 0.2283, 0.1693, 0.0340, 0.0753], device='cuda:7'), in_proj_covar=tensor([0.0293, 0.0258, 0.0279, 0.0270, 0.0283, 0.0217, 0.0264, 0.0292], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 00:53:44,902 INFO [train.py:904] (7/8) Epoch 9, batch 3550, loss[loss=0.1568, simple_loss=0.236, pruned_loss=0.03879, over 16991.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2711, pruned_loss=0.05415, over 3318231.48 frames. ], batch size: 41, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:53:47,052 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84753.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:54:53,011 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84800.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:54:54,906 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84801.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:54:55,838 INFO [train.py:904] (7/8) Epoch 9, batch 3600, loss[loss=0.1978, simple_loss=0.2738, pruned_loss=0.06086, over 16233.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2701, pruned_loss=0.05453, over 3312074.86 frames. ], batch size: 164, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:55:17,924 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.544e+02 2.999e+02 3.772e+02 8.043e+02, threshold=5.998e+02, percent-clipped=5.0 2023-04-29 00:56:07,142 INFO [train.py:904] (7/8) Epoch 9, batch 3650, loss[loss=0.2003, simple_loss=0.2738, pruned_loss=0.06336, over 15579.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2695, pruned_loss=0.05533, over 3305570.99 frames. ], batch size: 191, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:56:12,843 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84855.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:56:21,117 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84861.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:56:45,910 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 00:56:50,583 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84881.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:57:07,433 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84892.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:57:21,445 INFO [train.py:904] (7/8) Epoch 9, batch 3700, loss[loss=0.1946, simple_loss=0.2616, pruned_loss=0.06384, over 16725.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.268, pruned_loss=0.0567, over 3298307.50 frames. ], batch size: 134, lr: 7.68e-03, grad_scale: 8.0 2023-04-29 00:57:23,532 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84903.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:57:46,407 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.577e+02 2.579e+02 2.947e+02 3.773e+02 6.988e+02, threshold=5.894e+02, percent-clipped=1.0 2023-04-29 00:58:19,042 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=84940.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:58:35,033 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84951.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:58:35,667 INFO [train.py:904] (7/8) Epoch 9, batch 3750, loss[loss=0.2181, simple_loss=0.284, pruned_loss=0.07604, over 16918.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2684, pruned_loss=0.05785, over 3292353.35 frames. ], batch size: 109, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 00:59:29,108 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84989.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:59:34,083 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84992.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 00:59:47,564 INFO [train.py:904] (7/8) Epoch 9, batch 3800, loss[loss=0.2325, simple_loss=0.3007, pruned_loss=0.08213, over 15505.00 frames. ], tot_loss[loss=0.194, simple_loss=0.27, pruned_loss=0.05906, over 3283653.88 frames. ], batch size: 190, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 01:00:04,186 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85012.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:00:11,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.825e+02 2.585e+02 2.931e+02 3.484e+02 5.921e+02, threshold=5.862e+02, percent-clipped=1.0 2023-04-29 01:00:43,935 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:01:00,362 INFO [train.py:904] (7/8) Epoch 9, batch 3850, loss[loss=0.1887, simple_loss=0.2596, pruned_loss=0.05888, over 16715.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2701, pruned_loss=0.0596, over 3286410.30 frames. ], batch size: 83, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 01:01:06,206 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 01:02:13,003 INFO [train.py:904] (7/8) Epoch 9, batch 3900, loss[loss=0.1889, simple_loss=0.2596, pruned_loss=0.05907, over 16897.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2701, pruned_loss=0.05963, over 3276107.34 frames. ], batch size: 109, lr: 7.67e-03, grad_scale: 8.0 2023-04-29 01:02:21,792 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85107.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:02:32,613 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:02:36,234 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.478e+02 2.844e+02 3.462e+02 6.399e+02, threshold=5.687e+02, percent-clipped=3.0 2023-04-29 01:03:24,370 INFO [train.py:904] (7/8) Epoch 9, batch 3950, loss[loss=0.1974, simple_loss=0.2668, pruned_loss=0.06398, over 16362.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2697, pruned_loss=0.06015, over 3276393.86 frames. ], batch size: 146, lr: 7.66e-03, grad_scale: 8.0 2023-04-29 01:03:32,224 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85156.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:03:50,066 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85168.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:03:51,164 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6580, 3.8402, 2.2468, 4.0282, 2.8412, 3.9723, 2.0516, 3.0320], device='cuda:7'), covar=tensor([0.0177, 0.0310, 0.1356, 0.0162, 0.0611, 0.0485, 0.1434, 0.0565], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0164, 0.0183, 0.0119, 0.0164, 0.0205, 0.0190, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:04:00,754 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:04:07,294 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85181.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:04:20,053 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-29 01:04:37,103 INFO [train.py:904] (7/8) Epoch 9, batch 4000, loss[loss=0.1826, simple_loss=0.2597, pruned_loss=0.05272, over 16635.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2696, pruned_loss=0.0605, over 3275732.85 frames. ], batch size: 57, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:05:00,772 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.490e+02 3.071e+02 3.606e+02 5.108e+02, threshold=6.141e+02, percent-clipped=0.0 2023-04-29 01:05:17,727 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85229.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:05:51,196 INFO [train.py:904] (7/8) Epoch 9, batch 4050, loss[loss=0.1798, simple_loss=0.2589, pruned_loss=0.05039, over 16997.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2695, pruned_loss=0.05939, over 3280553.78 frames. ], batch size: 55, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:06:23,066 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1284, 4.1326, 4.5011, 4.4570, 4.4771, 4.1727, 4.1604, 4.0670], device='cuda:7'), covar=tensor([0.0256, 0.0546, 0.0331, 0.0383, 0.0361, 0.0289, 0.0770, 0.0413], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0320, 0.0322, 0.0311, 0.0366, 0.0338, 0.0443, 0.0270], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 01:06:45,612 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85289.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:07:04,626 INFO [train.py:904] (7/8) Epoch 9, batch 4100, loss[loss=0.2042, simple_loss=0.2951, pruned_loss=0.05665, over 16624.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2703, pruned_loss=0.05802, over 3268614.01 frames. ], batch size: 62, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:07:12,447 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85307.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:07:28,047 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.051e+02 2.369e+02 2.790e+02 6.834e+02, threshold=4.737e+02, percent-clipped=1.0 2023-04-29 01:07:57,981 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85337.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:08:20,253 INFO [train.py:904] (7/8) Epoch 9, batch 4150, loss[loss=0.1949, simple_loss=0.2842, pruned_loss=0.05284, over 16892.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2787, pruned_loss=0.06161, over 3224104.59 frames. ], batch size: 96, lr: 7.66e-03, grad_scale: 16.0 2023-04-29 01:08:27,932 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0131, 2.6032, 2.6166, 1.8977, 2.7538, 2.7593, 2.3983, 2.3675], device='cuda:7'), covar=tensor([0.0666, 0.0184, 0.0206, 0.0806, 0.0095, 0.0169, 0.0420, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0085, 0.0138, 0.0069, 0.0098, 0.0120, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 01:08:54,613 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-29 01:08:58,145 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0352, 1.9004, 2.3759, 2.9754, 2.9061, 3.4070, 1.9492, 3.2416], device='cuda:7'), covar=tensor([0.0125, 0.0301, 0.0206, 0.0153, 0.0152, 0.0091, 0.0304, 0.0077], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0162, 0.0150, 0.0152, 0.0156, 0.0116, 0.0162, 0.0105], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 01:09:37,260 INFO [train.py:904] (7/8) Epoch 9, batch 4200, loss[loss=0.2413, simple_loss=0.3267, pruned_loss=0.07792, over 16584.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2871, pruned_loss=0.06439, over 3201034.67 frames. ], batch size: 62, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:10:02,496 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.639e+02 3.175e+02 3.973e+02 9.081e+02, threshold=6.349e+02, percent-clipped=14.0 2023-04-29 01:10:05,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2982, 1.9097, 2.0842, 3.8811, 1.9511, 2.4534, 2.0668, 2.1515], device='cuda:7'), covar=tensor([0.0887, 0.3675, 0.2057, 0.0369, 0.3903, 0.2154, 0.2880, 0.3141], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0382, 0.0317, 0.0324, 0.0400, 0.0432, 0.0340, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:10:17,810 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85428.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:10:51,796 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7408, 3.2798, 3.3549, 1.8654, 2.8746, 2.3075, 3.3035, 3.2297], device='cuda:7'), covar=tensor([0.0215, 0.0550, 0.0470, 0.1676, 0.0654, 0.0793, 0.0588, 0.0721], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0141, 0.0156, 0.0141, 0.0135, 0.0124, 0.0135, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:10:52,284 INFO [train.py:904] (7/8) Epoch 9, batch 4250, loss[loss=0.1875, simple_loss=0.2799, pruned_loss=0.04753, over 16468.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2908, pruned_loss=0.0643, over 3199352.86 frames. ], batch size: 146, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:10:59,129 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85456.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:11:08,336 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85463.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:11:10,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8316, 2.1576, 1.5360, 1.7959, 2.5483, 2.3048, 2.9001, 2.8338], device='cuda:7'), covar=tensor([0.0096, 0.0359, 0.0443, 0.0368, 0.0181, 0.0276, 0.0140, 0.0166], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0191, 0.0187, 0.0186, 0.0187, 0.0190, 0.0192, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:11:20,276 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:11:39,014 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85483.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:11:48,332 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85489.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:12:07,386 INFO [train.py:904] (7/8) Epoch 9, batch 4300, loss[loss=0.2216, simple_loss=0.3014, pruned_loss=0.07093, over 11426.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2915, pruned_loss=0.06318, over 3176999.47 frames. ], batch size: 246, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:12:08,450 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8334, 3.5967, 3.8126, 3.5680, 3.7754, 4.1832, 3.9525, 3.5905], device='cuda:7'), covar=tensor([0.1790, 0.2228, 0.1777, 0.2498, 0.2692, 0.1654, 0.1170, 0.2474], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0462, 0.0489, 0.0406, 0.0534, 0.0516, 0.0392, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 01:12:11,407 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85504.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:12:30,792 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.412e+02 2.949e+02 3.511e+02 9.601e+02, threshold=5.898e+02, percent-clipped=2.0 2023-04-29 01:12:49,836 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85530.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:13:10,293 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85544.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:13:21,371 INFO [train.py:904] (7/8) Epoch 9, batch 4350, loss[loss=0.1974, simple_loss=0.289, pruned_loss=0.05287, over 16781.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.295, pruned_loss=0.06414, over 3188264.23 frames. ], batch size: 83, lr: 7.65e-03, grad_scale: 16.0 2023-04-29 01:13:32,849 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 01:14:01,391 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5089, 2.2957, 2.2025, 4.3145, 2.1082, 2.7637, 2.2824, 2.4457], device='cuda:7'), covar=tensor([0.0818, 0.2639, 0.1992, 0.0329, 0.3342, 0.1830, 0.2452, 0.2911], device='cuda:7'), in_proj_covar=tensor([0.0361, 0.0386, 0.0321, 0.0325, 0.0407, 0.0435, 0.0342, 0.0455], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:14:18,989 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85591.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:14:19,399 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 01:14:22,978 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 01:14:34,737 INFO [train.py:904] (7/8) Epoch 9, batch 4400, loss[loss=0.2127, simple_loss=0.2985, pruned_loss=0.06348, over 16524.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2965, pruned_loss=0.06482, over 3183956.80 frames. ], batch size: 75, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:14:42,370 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85607.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:14:56,973 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.712e+02 3.095e+02 3.599e+02 6.298e+02, threshold=6.190e+02, percent-clipped=2.0 2023-04-29 01:15:07,607 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:15:46,027 INFO [train.py:904] (7/8) Epoch 9, batch 4450, loss[loss=0.2449, simple_loss=0.3089, pruned_loss=0.09047, over 11817.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2994, pruned_loss=0.06576, over 3171993.54 frames. ], batch size: 248, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:15:50,470 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85655.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:16:34,309 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85686.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 01:16:56,348 INFO [train.py:904] (7/8) Epoch 9, batch 4500, loss[loss=0.2065, simple_loss=0.2935, pruned_loss=0.05971, over 17125.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2988, pruned_loss=0.06551, over 3195736.50 frames. ], batch size: 47, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:17:10,907 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 01:17:12,495 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9146, 3.3608, 3.1313, 1.7489, 2.6888, 2.0945, 3.0728, 3.3139], device='cuda:7'), covar=tensor([0.0339, 0.0719, 0.0631, 0.1971, 0.0876, 0.1025, 0.0898, 0.0975], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0141, 0.0155, 0.0140, 0.0133, 0.0123, 0.0135, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:17:20,319 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.131e+02 2.570e+02 3.015e+02 5.229e+02, threshold=5.140e+02, percent-clipped=0.0 2023-04-29 01:17:33,529 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3935, 3.5819, 3.8623, 1.6567, 4.2397, 4.2050, 3.0148, 3.0442], device='cuda:7'), covar=tensor([0.0756, 0.0173, 0.0135, 0.1141, 0.0034, 0.0056, 0.0331, 0.0378], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0095, 0.0084, 0.0136, 0.0067, 0.0095, 0.0117, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-29 01:18:07,074 INFO [train.py:904] (7/8) Epoch 9, batch 4550, loss[loss=0.2114, simple_loss=0.2933, pruned_loss=0.06476, over 17106.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2987, pruned_loss=0.06584, over 3211372.59 frames. ], batch size: 49, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:18:23,752 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85763.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:18:35,922 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85771.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:18:54,474 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:19:20,633 INFO [train.py:904] (7/8) Epoch 9, batch 4600, loss[loss=0.2131, simple_loss=0.2934, pruned_loss=0.06643, over 16794.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2993, pruned_loss=0.06556, over 3211526.87 frames. ], batch size: 39, lr: 7.64e-03, grad_scale: 16.0 2023-04-29 01:19:32,661 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85811.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:19:42,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.222e+02 2.575e+02 3.036e+02 5.036e+02, threshold=5.150e+02, percent-clipped=0.0 2023-04-29 01:19:44,979 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=85819.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:19:52,044 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-04-29 01:20:13,019 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85839.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:20:17,291 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0443, 3.4739, 3.4811, 1.7316, 2.9172, 2.3227, 3.3506, 3.5237], device='cuda:7'), covar=tensor([0.0262, 0.0596, 0.0490, 0.1845, 0.0731, 0.0876, 0.0635, 0.0753], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0140, 0.0155, 0.0140, 0.0133, 0.0123, 0.0134, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:20:30,862 INFO [train.py:904] (7/8) Epoch 9, batch 4650, loss[loss=0.2132, simple_loss=0.2871, pruned_loss=0.06961, over 16260.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2988, pruned_loss=0.066, over 3202914.14 frames. ], batch size: 35, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:20:43,174 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 01:21:04,863 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85875.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:21:20,432 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85886.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:21:23,431 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0959, 3.9756, 4.1610, 4.3465, 4.4204, 4.0723, 4.3582, 4.4560], device='cuda:7'), covar=tensor([0.1129, 0.0888, 0.1082, 0.0469, 0.0405, 0.1024, 0.0565, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0482, 0.0592, 0.0736, 0.0599, 0.0457, 0.0461, 0.0471, 0.0526], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:21:32,729 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 01:21:42,734 INFO [train.py:904] (7/8) Epoch 9, batch 4700, loss[loss=0.186, simple_loss=0.2672, pruned_loss=0.05237, over 16594.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2956, pruned_loss=0.06449, over 3212850.26 frames. ], batch size: 57, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:22:06,318 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.169e+02 2.505e+02 2.903e+02 6.033e+02, threshold=5.010e+02, percent-clipped=1.0 2023-04-29 01:22:25,271 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85930.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:22:33,527 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85936.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:22:42,599 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2498, 5.5349, 5.2277, 5.2986, 5.0014, 4.7561, 4.9530, 5.5673], device='cuda:7'), covar=tensor([0.0750, 0.0656, 0.0844, 0.0542, 0.0611, 0.0659, 0.0789, 0.0754], device='cuda:7'), in_proj_covar=tensor([0.0486, 0.0608, 0.0510, 0.0416, 0.0381, 0.0395, 0.0507, 0.0460], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:22:55,510 INFO [train.py:904] (7/8) Epoch 9, batch 4750, loss[loss=0.1824, simple_loss=0.2677, pruned_loss=0.04857, over 16713.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2915, pruned_loss=0.06264, over 3214212.52 frames. ], batch size: 76, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:23:04,320 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85957.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:23:39,898 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85981.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:23:49,334 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 01:23:50,338 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6846, 4.1341, 3.9632, 2.2861, 3.1622, 2.7036, 3.8830, 4.2032], device='cuda:7'), covar=tensor([0.0225, 0.0504, 0.0487, 0.1505, 0.0748, 0.0743, 0.0560, 0.0697], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0142, 0.0157, 0.0142, 0.0135, 0.0125, 0.0136, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:23:55,278 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85991.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:24:13,774 INFO [train.py:904] (7/8) Epoch 9, batch 4800, loss[loss=0.2283, simple_loss=0.3194, pruned_loss=0.06857, over 16370.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2886, pruned_loss=0.0609, over 3218915.46 frames. ], batch size: 165, lr: 7.63e-03, grad_scale: 16.0 2023-04-29 01:24:14,297 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:24:37,230 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.079e+02 2.389e+02 2.964e+02 6.421e+02, threshold=4.777e+02, percent-clipped=3.0 2023-04-29 01:24:38,391 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:25:20,096 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5627, 2.6601, 2.3623, 3.8714, 2.5284, 3.8852, 1.2073, 2.8607], device='cuda:7'), covar=tensor([0.1335, 0.0618, 0.1126, 0.0115, 0.0156, 0.0344, 0.1626, 0.0708], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0151, 0.0173, 0.0125, 0.0197, 0.0203, 0.0172, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 01:25:28,115 INFO [train.py:904] (7/8) Epoch 9, batch 4850, loss[loss=0.214, simple_loss=0.3047, pruned_loss=0.06171, over 16693.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2905, pruned_loss=0.06094, over 3194349.77 frames. ], batch size: 134, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:25:29,143 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86052.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:25:36,421 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7808, 1.3332, 1.6326, 1.6898, 1.8810, 1.9179, 1.4974, 1.8224], device='cuda:7'), covar=tensor([0.0144, 0.0256, 0.0128, 0.0176, 0.0142, 0.0108, 0.0238, 0.0062], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0163, 0.0148, 0.0152, 0.0156, 0.0114, 0.0162, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 01:25:45,873 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86063.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:25:54,723 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86069.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:26:16,542 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86084.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:26:42,270 INFO [train.py:904] (7/8) Epoch 9, batch 4900, loss[loss=0.2053, simple_loss=0.2921, pruned_loss=0.05926, over 16910.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2899, pruned_loss=0.0601, over 3180246.34 frames. ], batch size: 109, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:26:46,073 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2121, 3.3021, 1.7237, 3.5367, 2.3265, 3.4858, 2.0035, 2.5928], device='cuda:7'), covar=tensor([0.0190, 0.0320, 0.1659, 0.0088, 0.0845, 0.0402, 0.1404, 0.0672], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0160, 0.0182, 0.0109, 0.0162, 0.0198, 0.0189, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:26:59,866 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86113.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:05,787 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.491e+02 2.278e+02 2.647e+02 3.027e+02 5.480e+02, threshold=5.294e+02, percent-clipped=1.0 2023-04-29 01:27:23,802 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86130.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:26,528 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:37,366 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86139.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:27:55,683 INFO [train.py:904] (7/8) Epoch 9, batch 4950, loss[loss=0.208, simple_loss=0.3024, pruned_loss=0.05684, over 15407.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2894, pruned_loss=0.05928, over 3194351.93 frames. ], batch size: 191, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:28:20,575 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 01:28:45,984 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86186.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:28:47,115 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86187.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:29:08,684 INFO [train.py:904] (7/8) Epoch 9, batch 5000, loss[loss=0.2135, simple_loss=0.295, pruned_loss=0.06607, over 16474.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2905, pruned_loss=0.0591, over 3203883.54 frames. ], batch size: 35, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:29:28,541 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.64 vs. limit=5.0 2023-04-29 01:29:32,026 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 2.365e+02 2.931e+02 3.414e+02 6.733e+02, threshold=5.862e+02, percent-clipped=2.0 2023-04-29 01:29:50,446 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86231.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:29:54,587 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86234.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:30:11,791 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86246.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:30:19,725 INFO [train.py:904] (7/8) Epoch 9, batch 5050, loss[loss=0.2033, simple_loss=0.288, pruned_loss=0.05927, over 16466.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2907, pruned_loss=0.05874, over 3215674.85 frames. ], batch size: 68, lr: 7.62e-03, grad_scale: 16.0 2023-04-29 01:31:01,314 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:31:07,988 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86286.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:31:29,534 INFO [train.py:904] (7/8) Epoch 9, batch 5100, loss[loss=0.1843, simple_loss=0.2661, pruned_loss=0.05121, over 17025.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2887, pruned_loss=0.05788, over 3211518.17 frames. ], batch size: 41, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:31:37,513 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86307.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:31:45,610 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86313.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:31:52,764 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.422e+02 2.762e+02 3.223e+02 5.276e+02, threshold=5.525e+02, percent-clipped=0.0 2023-04-29 01:32:08,054 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:32:41,548 INFO [train.py:904] (7/8) Epoch 9, batch 5150, loss[loss=0.2085, simple_loss=0.2887, pruned_loss=0.0642, over 16644.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2893, pruned_loss=0.05727, over 3204288.00 frames. ], batch size: 62, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:32:51,006 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86358.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:33:19,779 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 01:33:54,434 INFO [train.py:904] (7/8) Epoch 9, batch 5200, loss[loss=0.1961, simple_loss=0.2863, pruned_loss=0.05294, over 16895.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2869, pruned_loss=0.05648, over 3215374.00 frames. ], batch size: 96, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:34:02,835 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86408.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:34:04,444 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2023-04-29 01:34:16,788 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.07 vs. limit=5.0 2023-04-29 01:34:17,300 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.533e+02 2.298e+02 2.670e+02 3.115e+02 5.719e+02, threshold=5.340e+02, percent-clipped=1.0 2023-04-29 01:34:27,690 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86425.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:34:50,056 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86440.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:34:57,851 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86445.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:35:06,331 INFO [train.py:904] (7/8) Epoch 9, batch 5250, loss[loss=0.1901, simple_loss=0.278, pruned_loss=0.05116, over 16422.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2847, pruned_loss=0.05608, over 3218411.06 frames. ], batch size: 68, lr: 7.61e-03, grad_scale: 16.0 2023-04-29 01:35:41,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6893, 4.0526, 3.2277, 2.3783, 2.9932, 2.6049, 4.3633, 3.8851], device='cuda:7'), covar=tensor([0.2361, 0.0636, 0.1276, 0.1807, 0.1982, 0.1426, 0.0381, 0.0742], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0252, 0.0274, 0.0268, 0.0274, 0.0212, 0.0261, 0.0278], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:36:16,641 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 01:36:18,409 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86501.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:36:19,015 INFO [train.py:904] (7/8) Epoch 9, batch 5300, loss[loss=0.1792, simple_loss=0.2623, pruned_loss=0.0481, over 16862.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2812, pruned_loss=0.05494, over 3229436.19 frames. ], batch size: 90, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:36:25,663 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86506.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:36:42,028 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.448e+02 2.767e+02 3.292e+02 5.111e+02, threshold=5.534e+02, percent-clipped=0.0 2023-04-29 01:36:47,618 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4780, 3.8207, 4.1071, 1.9952, 4.3834, 4.3526, 3.1304, 3.1740], device='cuda:7'), covar=tensor([0.0793, 0.0153, 0.0133, 0.1091, 0.0033, 0.0060, 0.0328, 0.0406], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0096, 0.0084, 0.0137, 0.0067, 0.0094, 0.0118, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-29 01:36:58,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0154, 3.5253, 3.1754, 1.8450, 2.8212, 2.3406, 3.4326, 3.5039], device='cuda:7'), covar=tensor([0.0224, 0.0524, 0.0615, 0.1760, 0.0755, 0.0860, 0.0618, 0.0792], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0138, 0.0156, 0.0140, 0.0134, 0.0124, 0.0135, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:37:01,886 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86531.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:37:32,906 INFO [train.py:904] (7/8) Epoch 9, batch 5350, loss[loss=0.2207, simple_loss=0.2948, pruned_loss=0.07332, over 12276.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2798, pruned_loss=0.05463, over 3221641.16 frames. ], batch size: 246, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:38:12,587 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86579.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:38:13,774 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86580.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:38:23,806 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86586.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:38:45,874 INFO [train.py:904] (7/8) Epoch 9, batch 5400, loss[loss=0.1951, simple_loss=0.2893, pruned_loss=0.05048, over 16735.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2822, pruned_loss=0.05494, over 3220546.02 frames. ], batch size: 89, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:38:46,255 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86602.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:39:01,736 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86613.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:39:09,035 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.152e+02 2.423e+02 2.907e+02 4.710e+02, threshold=4.847e+02, percent-clipped=0.0 2023-04-29 01:39:12,914 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9413, 3.8135, 3.8514, 4.1941, 4.2527, 3.9335, 4.2149, 4.3099], device='cuda:7'), covar=tensor([0.1473, 0.1125, 0.1944, 0.0745, 0.0675, 0.1228, 0.0769, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0499, 0.0606, 0.0752, 0.0614, 0.0465, 0.0466, 0.0483, 0.0536], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:39:32,145 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86634.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:39:41,804 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86641.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:39:54,465 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-29 01:40:00,497 INFO [train.py:904] (7/8) Epoch 9, batch 5450, loss[loss=0.2234, simple_loss=0.3071, pruned_loss=0.06989, over 16564.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2859, pruned_loss=0.05712, over 3218827.58 frames. ], batch size: 75, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:40:04,665 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 01:40:11,018 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86658.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:40:15,979 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86661.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:41:18,962 INFO [train.py:904] (7/8) Epoch 9, batch 5500, loss[loss=0.2815, simple_loss=0.3416, pruned_loss=0.1106, over 11603.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2936, pruned_loss=0.06219, over 3190996.23 frames. ], batch size: 247, lr: 7.60e-03, grad_scale: 16.0 2023-04-29 01:41:24,419 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86706.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:41:28,634 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86708.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:41:43,595 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.927e+02 3.483e+02 4.212e+02 5.366e+02 1.166e+03, threshold=8.424e+02, percent-clipped=36.0 2023-04-29 01:41:54,538 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86725.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:42:36,884 INFO [train.py:904] (7/8) Epoch 9, batch 5550, loss[loss=0.3321, simple_loss=0.3724, pruned_loss=0.1458, over 10869.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.3026, pruned_loss=0.06948, over 3127339.35 frames. ], batch size: 247, lr: 7.59e-03, grad_scale: 16.0 2023-04-29 01:42:43,144 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86756.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:43:09,699 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86773.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:43:45,645 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86796.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:43:53,983 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86801.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:43:54,671 INFO [train.py:904] (7/8) Epoch 9, batch 5600, loss[loss=0.3537, simple_loss=0.3876, pruned_loss=0.1599, over 11246.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3087, pruned_loss=0.07423, over 3116991.88 frames. ], batch size: 247, lr: 7.59e-03, grad_scale: 8.0 2023-04-29 01:44:14,623 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86814.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:44:21,511 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 3.990e+02 4.962e+02 6.279e+02 1.585e+03, threshold=9.923e+02, percent-clipped=6.0 2023-04-29 01:45:17,296 INFO [train.py:904] (7/8) Epoch 9, batch 5650, loss[loss=0.224, simple_loss=0.306, pruned_loss=0.07099, over 16534.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.3144, pruned_loss=0.07919, over 3083387.92 frames. ], batch size: 68, lr: 7.59e-03, grad_scale: 4.0 2023-04-29 01:45:19,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7156, 3.9753, 3.1543, 2.3610, 3.0733, 2.4476, 4.3276, 3.8140], device='cuda:7'), covar=tensor([0.2587, 0.0712, 0.1407, 0.2035, 0.2204, 0.1592, 0.0416, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0293, 0.0254, 0.0278, 0.0270, 0.0280, 0.0212, 0.0262, 0.0278], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:45:31,923 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-29 01:45:53,437 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86875.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:45:58,543 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86878.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:46:23,165 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7193, 1.6374, 1.4404, 1.5346, 1.8773, 1.6409, 1.6909, 1.9038], device='cuda:7'), covar=tensor([0.0119, 0.0201, 0.0287, 0.0253, 0.0131, 0.0183, 0.0133, 0.0132], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0189, 0.0187, 0.0186, 0.0186, 0.0189, 0.0188, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:46:33,112 INFO [train.py:904] (7/8) Epoch 9, batch 5700, loss[loss=0.2925, simple_loss=0.3444, pruned_loss=0.1203, over 10993.00 frames. ], tot_loss[loss=0.2395, simple_loss=0.3163, pruned_loss=0.08136, over 3064744.28 frames. ], batch size: 247, lr: 7.59e-03, grad_scale: 4.0 2023-04-29 01:46:33,478 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86902.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:46:59,885 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.388e+02 3.612e+02 4.435e+02 5.746e+02 9.391e+02, threshold=8.870e+02, percent-clipped=0.0 2023-04-29 01:47:25,202 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86936.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:47:31,000 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86939.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:47:48,290 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=86950.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:47:50,857 INFO [train.py:904] (7/8) Epoch 9, batch 5750, loss[loss=0.2239, simple_loss=0.3121, pruned_loss=0.06782, over 16484.00 frames. ], tot_loss[loss=0.2413, simple_loss=0.3183, pruned_loss=0.08209, over 3066976.87 frames. ], batch size: 68, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:48:12,845 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.8352, 6.2116, 5.8072, 5.9607, 5.5023, 5.3107, 5.6405, 6.2073], device='cuda:7'), covar=tensor([0.0915, 0.0648, 0.1054, 0.0643, 0.0750, 0.0588, 0.0923, 0.0713], device='cuda:7'), in_proj_covar=tensor([0.0482, 0.0609, 0.0508, 0.0412, 0.0379, 0.0392, 0.0503, 0.0457], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:49:12,520 INFO [train.py:904] (7/8) Epoch 9, batch 5800, loss[loss=0.1891, simple_loss=0.2868, pruned_loss=0.04566, over 16839.00 frames. ], tot_loss[loss=0.2391, simple_loss=0.3177, pruned_loss=0.08024, over 3063554.86 frames. ], batch size: 102, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:49:13,790 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87002.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 01:49:40,723 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 3.181e+02 3.841e+02 4.855e+02 7.782e+02, threshold=7.681e+02, percent-clipped=0.0 2023-04-29 01:50:30,453 INFO [train.py:904] (7/8) Epoch 9, batch 5850, loss[loss=0.2071, simple_loss=0.2909, pruned_loss=0.06161, over 17124.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.3152, pruned_loss=0.07814, over 3077138.55 frames. ], batch size: 48, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:50:41,892 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4878, 3.5776, 2.7669, 2.0971, 2.6737, 2.1681, 3.7767, 3.6451], device='cuda:7'), covar=tensor([0.2776, 0.0741, 0.1584, 0.2252, 0.1984, 0.1712, 0.0483, 0.0781], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0254, 0.0279, 0.0272, 0.0282, 0.0214, 0.0264, 0.0279], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 01:50:46,914 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 01:50:47,461 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87063.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:51:25,286 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6702, 4.9368, 5.0968, 4.8950, 4.8443, 5.5036, 4.9634, 4.7449], device='cuda:7'), covar=tensor([0.0987, 0.1613, 0.1489, 0.1672, 0.2357, 0.0850, 0.1384, 0.2251], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0445, 0.0473, 0.0391, 0.0520, 0.0500, 0.0383, 0.0528], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 01:51:36,432 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-29 01:51:42,196 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87096.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:51:50,079 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87101.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 01:51:50,877 INFO [train.py:904] (7/8) Epoch 9, batch 5900, loss[loss=0.2184, simple_loss=0.3005, pruned_loss=0.06811, over 16316.00 frames. ], tot_loss[loss=0.2359, simple_loss=0.3154, pruned_loss=0.07827, over 3080019.47 frames. ], batch size: 165, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:52:22,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.063e+02 3.002e+02 3.675e+02 4.500e+02 8.851e+02, threshold=7.350e+02, percent-clipped=1.0 2023-04-29 01:53:01,051 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87144.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:53:09,129 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87149.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:53:13,007 INFO [train.py:904] (7/8) Epoch 9, batch 5950, loss[loss=0.2417, simple_loss=0.3327, pruned_loss=0.07537, over 17216.00 frames. ], tot_loss[loss=0.235, simple_loss=0.3163, pruned_loss=0.07683, over 3085596.52 frames. ], batch size: 52, lr: 7.58e-03, grad_scale: 4.0 2023-04-29 01:53:42,084 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87170.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:53:48,778 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9529, 2.8141, 2.6384, 1.9773, 2.5969, 2.0922, 2.7471, 2.9193], device='cuda:7'), covar=tensor([0.0249, 0.0505, 0.0515, 0.1449, 0.0678, 0.0888, 0.0510, 0.0612], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0139, 0.0158, 0.0142, 0.0134, 0.0125, 0.0136, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 01:54:02,311 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 01:54:33,111 INFO [train.py:904] (7/8) Epoch 9, batch 6000, loss[loss=0.2068, simple_loss=0.2868, pruned_loss=0.06342, over 16557.00 frames. ], tot_loss[loss=0.2345, simple_loss=0.3151, pruned_loss=0.07691, over 3076068.84 frames. ], batch size: 68, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:54:33,112 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 01:54:44,301 INFO [train.py:938] (7/8) Epoch 9, validation: loss=0.1674, simple_loss=0.2809, pruned_loss=0.02692, over 944034.00 frames. 2023-04-29 01:54:44,302 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 01:55:11,802 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.247e+02 3.795e+02 4.870e+02 1.523e+03, threshold=7.589e+02, percent-clipped=3.0 2023-04-29 01:55:28,619 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87230.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:55:34,844 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87234.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:55:37,553 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87236.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:56:03,562 INFO [train.py:904] (7/8) Epoch 9, batch 6050, loss[loss=0.2653, simple_loss=0.3167, pruned_loss=0.107, over 12069.00 frames. ], tot_loss[loss=0.2322, simple_loss=0.313, pruned_loss=0.07567, over 3080628.36 frames. ], batch size: 248, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:56:56,178 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87284.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:57:06,956 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87291.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 01:57:23,341 INFO [train.py:904] (7/8) Epoch 9, batch 6100, loss[loss=0.2195, simple_loss=0.2997, pruned_loss=0.06968, over 16308.00 frames. ], tot_loss[loss=0.2309, simple_loss=0.3123, pruned_loss=0.07473, over 3087921.15 frames. ], batch size: 165, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:57:52,502 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.938e+02 3.700e+02 4.608e+02 9.243e+02, threshold=7.400e+02, percent-clipped=2.0 2023-04-29 01:58:42,528 INFO [train.py:904] (7/8) Epoch 9, batch 6150, loss[loss=0.2597, simple_loss=0.3198, pruned_loss=0.09977, over 11359.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3108, pruned_loss=0.07465, over 3074888.80 frames. ], batch size: 248, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 01:58:52,854 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87358.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 01:58:59,191 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87362.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:00:00,641 INFO [train.py:904] (7/8) Epoch 9, batch 6200, loss[loss=0.2726, simple_loss=0.3242, pruned_loss=0.1105, over 11457.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.3088, pruned_loss=0.07389, over 3073436.96 frames. ], batch size: 246, lr: 7.57e-03, grad_scale: 8.0 2023-04-29 02:00:28,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.203e+02 3.670e+02 4.311e+02 5.768e+02 9.493e+02, threshold=8.622e+02, percent-clipped=7.0 2023-04-29 02:00:34,963 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87423.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:01:18,845 INFO [train.py:904] (7/8) Epoch 9, batch 6250, loss[loss=0.1904, simple_loss=0.2862, pruned_loss=0.04731, over 16882.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3072, pruned_loss=0.07261, over 3097262.17 frames. ], batch size: 96, lr: 7.56e-03, grad_scale: 4.0 2023-04-29 02:01:32,680 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87461.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:01:38,894 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8122, 3.3277, 3.1434, 1.9192, 2.7237, 2.2149, 3.3181, 3.3435], device='cuda:7'), covar=tensor([0.0242, 0.0586, 0.0597, 0.1698, 0.0775, 0.0896, 0.0577, 0.0757], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0140, 0.0158, 0.0142, 0.0135, 0.0125, 0.0136, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:01:46,085 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87470.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:02:03,858 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1937, 4.1949, 4.6369, 4.6001, 4.5910, 4.3043, 4.3122, 4.1677], device='cuda:7'), covar=tensor([0.0287, 0.0558, 0.0347, 0.0387, 0.0384, 0.0314, 0.0804, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0307, 0.0319, 0.0304, 0.0358, 0.0328, 0.0440, 0.0268], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 02:02:14,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5347, 1.5698, 2.0263, 2.5473, 2.5449, 2.7888, 1.6726, 2.7898], device='cuda:7'), covar=tensor([0.0136, 0.0364, 0.0226, 0.0211, 0.0187, 0.0123, 0.0370, 0.0087], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0162, 0.0145, 0.0149, 0.0156, 0.0112, 0.0163, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 02:02:33,542 INFO [train.py:904] (7/8) Epoch 9, batch 6300, loss[loss=0.236, simple_loss=0.3235, pruned_loss=0.07425, over 16826.00 frames. ], tot_loss[loss=0.2252, simple_loss=0.3068, pruned_loss=0.07182, over 3108278.32 frames. ], batch size: 102, lr: 7.56e-03, grad_scale: 4.0 2023-04-29 02:02:38,949 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87505.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:02:59,938 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87518.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:03:04,119 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 3.209e+02 3.966e+02 4.812e+02 1.231e+03, threshold=7.932e+02, percent-clipped=2.0 2023-04-29 02:03:06,807 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87522.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:03:13,147 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4631, 2.9434, 2.6658, 2.2208, 2.3353, 2.1497, 2.8657, 2.9078], device='cuda:7'), covar=tensor([0.1896, 0.0767, 0.1300, 0.1931, 0.1741, 0.1700, 0.0513, 0.0796], device='cuda:7'), in_proj_covar=tensor([0.0299, 0.0254, 0.0282, 0.0273, 0.0283, 0.0214, 0.0264, 0.0281], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:03:24,787 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87534.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:03:51,243 INFO [train.py:904] (7/8) Epoch 9, batch 6350, loss[loss=0.2144, simple_loss=0.2904, pruned_loss=0.06922, over 17032.00 frames. ], tot_loss[loss=0.2263, simple_loss=0.3072, pruned_loss=0.07267, over 3118537.64 frames. ], batch size: 53, lr: 7.56e-03, grad_scale: 4.0 2023-04-29 02:03:57,873 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6453, 4.0326, 4.2005, 1.6928, 4.5229, 4.6696, 3.2130, 3.0963], device='cuda:7'), covar=tensor([0.0951, 0.0129, 0.0152, 0.1346, 0.0054, 0.0071, 0.0364, 0.0540], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0100, 0.0085, 0.0139, 0.0067, 0.0095, 0.0119, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 02:04:12,993 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87566.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:04:37,491 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87582.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:04:40,833 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87584.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:04:44,426 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87586.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:04:50,169 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87589.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:05:08,858 INFO [train.py:904] (7/8) Epoch 9, batch 6400, loss[loss=0.286, simple_loss=0.346, pruned_loss=0.113, over 11669.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.3088, pruned_loss=0.07493, over 3076170.76 frames. ], batch size: 246, lr: 7.56e-03, grad_scale: 8.0 2023-04-29 02:05:14,040 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87605.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:05:15,401 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7125, 2.3396, 2.3954, 4.5347, 2.1556, 2.8236, 2.3817, 2.6115], device='cuda:7'), covar=tensor([0.0821, 0.3090, 0.2004, 0.0312, 0.3663, 0.2057, 0.2666, 0.2517], device='cuda:7'), in_proj_covar=tensor([0.0351, 0.0373, 0.0313, 0.0318, 0.0403, 0.0423, 0.0335, 0.0439], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:05:37,573 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.591e+02 3.429e+02 4.257e+02 5.158e+02 9.236e+02, threshold=8.515e+02, percent-clipped=3.0 2023-04-29 02:05:51,902 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9830, 5.3122, 5.0363, 5.0946, 4.8066, 4.6082, 4.6950, 5.3905], device='cuda:7'), covar=tensor([0.0949, 0.0797, 0.0972, 0.0667, 0.0716, 0.0850, 0.0955, 0.0766], device='cuda:7'), in_proj_covar=tensor([0.0497, 0.0626, 0.0525, 0.0426, 0.0388, 0.0405, 0.0523, 0.0466], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:06:02,419 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7497, 3.4000, 3.1709, 1.8021, 2.6816, 2.0236, 3.3917, 3.4632], device='cuda:7'), covar=tensor([0.0247, 0.0561, 0.0530, 0.1864, 0.0877, 0.1013, 0.0491, 0.0631], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0138, 0.0157, 0.0141, 0.0134, 0.0124, 0.0135, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:06:05,313 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1531, 5.0621, 4.9794, 4.3056, 4.9836, 1.8200, 4.7402, 4.8574], device='cuda:7'), covar=tensor([0.0082, 0.0084, 0.0131, 0.0380, 0.0094, 0.2160, 0.0130, 0.0142], device='cuda:7'), in_proj_covar=tensor([0.0116, 0.0104, 0.0152, 0.0145, 0.0121, 0.0166, 0.0136, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:06:14,998 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87645.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:06:22,248 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:06:24,601 INFO [train.py:904] (7/8) Epoch 9, batch 6450, loss[loss=0.2224, simple_loss=0.3043, pruned_loss=0.07021, over 16684.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.3083, pruned_loss=0.07352, over 3089242.16 frames. ], batch size: 134, lr: 7.55e-03, grad_scale: 4.0 2023-04-29 02:06:34,933 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87658.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:06:46,454 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87666.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:07:41,891 INFO [train.py:904] (7/8) Epoch 9, batch 6500, loss[loss=0.2492, simple_loss=0.3263, pruned_loss=0.0861, over 15418.00 frames. ], tot_loss[loss=0.2255, simple_loss=0.3063, pruned_loss=0.07239, over 3113549.32 frames. ], batch size: 191, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:07:48,430 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87706.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:08:06,568 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87718.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:08:13,256 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.348e+02 3.236e+02 4.021e+02 5.179e+02 1.078e+03, threshold=8.043e+02, percent-clipped=2.0 2023-04-29 02:08:56,393 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-29 02:08:58,276 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9464, 4.9095, 5.3862, 5.4426, 5.3758, 5.0507, 5.0277, 4.7245], device='cuda:7'), covar=tensor([0.0256, 0.0483, 0.0460, 0.0323, 0.0377, 0.0388, 0.0725, 0.0371], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0306, 0.0316, 0.0301, 0.0355, 0.0327, 0.0435, 0.0267], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 02:09:00,851 INFO [train.py:904] (7/8) Epoch 9, batch 6550, loss[loss=0.2048, simple_loss=0.3108, pruned_loss=0.04943, over 16786.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.309, pruned_loss=0.07357, over 3091656.06 frames. ], batch size: 102, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:09:35,393 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7620, 4.5661, 4.5658, 2.8735, 3.9473, 4.4709, 4.1594, 2.3143], device='cuda:7'), covar=tensor([0.0340, 0.0026, 0.0022, 0.0273, 0.0052, 0.0061, 0.0038, 0.0339], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0063, 0.0063, 0.0119, 0.0070, 0.0082, 0.0070, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 02:09:49,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5408, 4.0340, 3.6079, 1.9145, 2.9673, 2.8362, 3.6769, 3.9813], device='cuda:7'), covar=tensor([0.0250, 0.0616, 0.0641, 0.1954, 0.0863, 0.0788, 0.0696, 0.0772], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0137, 0.0155, 0.0139, 0.0132, 0.0123, 0.0134, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:10:18,528 INFO [train.py:904] (7/8) Epoch 9, batch 6600, loss[loss=0.2877, simple_loss=0.3442, pruned_loss=0.1156, over 11768.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3106, pruned_loss=0.07361, over 3107942.59 frames. ], batch size: 248, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:10:20,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5082, 2.7195, 2.3630, 3.9699, 3.0512, 4.0156, 1.2435, 2.8059], device='cuda:7'), covar=tensor([0.1381, 0.0596, 0.1159, 0.0137, 0.0272, 0.0352, 0.1569, 0.0788], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0152, 0.0176, 0.0127, 0.0201, 0.0204, 0.0173, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 02:10:27,401 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3047, 4.6389, 4.3682, 4.3477, 4.1098, 4.1487, 4.1536, 4.6636], device='cuda:7'), covar=tensor([0.0961, 0.0715, 0.0978, 0.0650, 0.0757, 0.1185, 0.0882, 0.0805], device='cuda:7'), in_proj_covar=tensor([0.0498, 0.0622, 0.0523, 0.0428, 0.0388, 0.0406, 0.0521, 0.0467], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:10:42,563 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87817.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:10:51,648 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.077e+02 3.184e+02 4.033e+02 5.145e+02 1.254e+03, threshold=8.065e+02, percent-clipped=5.0 2023-04-29 02:11:36,980 INFO [train.py:904] (7/8) Epoch 9, batch 6650, loss[loss=0.2015, simple_loss=0.2929, pruned_loss=0.05501, over 16840.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3107, pruned_loss=0.07435, over 3103426.97 frames. ], batch size: 102, lr: 7.55e-03, grad_scale: 2.0 2023-04-29 02:11:51,662 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87861.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:12:29,428 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87886.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:12:39,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3923, 3.0655, 2.8057, 1.7811, 2.5737, 2.1593, 2.9705, 3.0671], device='cuda:7'), covar=tensor([0.0326, 0.0593, 0.0674, 0.1827, 0.0856, 0.0913, 0.0697, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0139, 0.0157, 0.0140, 0.0133, 0.0124, 0.0135, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:12:53,912 INFO [train.py:904] (7/8) Epoch 9, batch 6700, loss[loss=0.2332, simple_loss=0.3132, pruned_loss=0.07657, over 16433.00 frames. ], tot_loss[loss=0.2298, simple_loss=0.3101, pruned_loss=0.07471, over 3089172.53 frames. ], batch size: 146, lr: 7.54e-03, grad_scale: 2.0 2023-04-29 02:13:13,445 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8802, 4.8806, 4.6779, 4.4676, 4.2642, 4.7684, 4.6934, 4.3727], device='cuda:7'), covar=tensor([0.0414, 0.0229, 0.0217, 0.0209, 0.0869, 0.0266, 0.0280, 0.0597], device='cuda:7'), in_proj_covar=tensor([0.0224, 0.0271, 0.0261, 0.0237, 0.0285, 0.0274, 0.0180, 0.0307], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:13:26,705 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.105e+02 3.383e+02 4.194e+02 5.361e+02 9.838e+02, threshold=8.388e+02, percent-clipped=4.0 2023-04-29 02:13:44,326 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=87934.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:13:52,494 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87940.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 02:14:00,295 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87945.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:14:10,417 INFO [train.py:904] (7/8) Epoch 9, batch 6750, loss[loss=0.2217, simple_loss=0.3017, pruned_loss=0.07087, over 16560.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3091, pruned_loss=0.07461, over 3097977.53 frames. ], batch size: 75, lr: 7.54e-03, grad_scale: 2.0 2023-04-29 02:14:24,859 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87961.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:15:11,819 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-29 02:15:29,295 INFO [train.py:904] (7/8) Epoch 9, batch 6800, loss[loss=0.2436, simple_loss=0.3149, pruned_loss=0.08622, over 11818.00 frames. ], tot_loss[loss=0.2296, simple_loss=0.3093, pruned_loss=0.07499, over 3077746.70 frames. ], batch size: 246, lr: 7.54e-03, grad_scale: 4.0 2023-04-29 02:15:35,140 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5624, 3.9669, 3.4116, 3.8122, 3.4783, 3.6105, 3.5868, 3.9161], device='cuda:7'), covar=tensor([0.2993, 0.1727, 0.3595, 0.1391, 0.1812, 0.2738, 0.2550, 0.1845], device='cuda:7'), in_proj_covar=tensor([0.0497, 0.0627, 0.0526, 0.0426, 0.0390, 0.0408, 0.0521, 0.0468], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:15:39,247 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-04-29 02:15:54,176 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:16:02,235 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 3.136e+02 3.707e+02 4.804e+02 8.151e+02, threshold=7.414e+02, percent-clipped=0.0 2023-04-29 02:16:42,917 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7188, 2.9428, 2.8119, 4.8720, 3.6478, 4.4774, 1.6624, 3.1391], device='cuda:7'), covar=tensor([0.1310, 0.0590, 0.0943, 0.0078, 0.0273, 0.0286, 0.1361, 0.0688], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0152, 0.0176, 0.0126, 0.0200, 0.0203, 0.0173, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 02:16:45,314 INFO [train.py:904] (7/8) Epoch 9, batch 6850, loss[loss=0.204, simple_loss=0.3092, pruned_loss=0.0494, over 16688.00 frames. ], tot_loss[loss=0.2318, simple_loss=0.3109, pruned_loss=0.07635, over 3073609.33 frames. ], batch size: 89, lr: 7.54e-03, grad_scale: 4.0 2023-04-29 02:17:06,663 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88066.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:17:11,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8246, 3.4322, 3.2749, 1.8511, 2.8403, 2.4131, 3.3822, 3.4916], device='cuda:7'), covar=tensor([0.0254, 0.0558, 0.0571, 0.1804, 0.0764, 0.0876, 0.0580, 0.0713], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0140, 0.0158, 0.0143, 0.0135, 0.0126, 0.0137, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:17:54,134 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4529, 2.0507, 1.5774, 1.7854, 2.3747, 2.1248, 2.5007, 2.6079], device='cuda:7'), covar=tensor([0.0108, 0.0248, 0.0361, 0.0357, 0.0152, 0.0269, 0.0130, 0.0175], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0188, 0.0187, 0.0185, 0.0185, 0.0189, 0.0188, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:17:59,066 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88100.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:18:01,007 INFO [train.py:904] (7/8) Epoch 9, batch 6900, loss[loss=0.249, simple_loss=0.3252, pruned_loss=0.08638, over 15390.00 frames. ], tot_loss[loss=0.2326, simple_loss=0.3135, pruned_loss=0.0759, over 3092146.14 frames. ], batch size: 191, lr: 7.54e-03, grad_scale: 4.0 2023-04-29 02:18:24,695 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88117.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:18:33,102 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.937e+02 3.211e+02 3.844e+02 4.463e+02 9.504e+02, threshold=7.687e+02, percent-clipped=4.0 2023-04-29 02:19:18,083 INFO [train.py:904] (7/8) Epoch 9, batch 6950, loss[loss=0.2845, simple_loss=0.3376, pruned_loss=0.1157, over 11161.00 frames. ], tot_loss[loss=0.2357, simple_loss=0.3157, pruned_loss=0.07785, over 3086130.44 frames. ], batch size: 246, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:19:31,395 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 02:19:32,176 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88161.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:19:32,327 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88161.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:19:37,258 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9991, 3.6292, 3.4853, 1.7521, 2.9259, 2.3993, 3.4783, 3.5984], device='cuda:7'), covar=tensor([0.0243, 0.0550, 0.0499, 0.1916, 0.0748, 0.0875, 0.0641, 0.0835], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0138, 0.0158, 0.0142, 0.0134, 0.0125, 0.0136, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:19:38,906 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88165.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:20:07,231 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 02:20:31,304 INFO [train.py:904] (7/8) Epoch 9, batch 7000, loss[loss=0.2179, simple_loss=0.305, pruned_loss=0.06544, over 17070.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.3143, pruned_loss=0.07615, over 3100929.58 frames. ], batch size: 53, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:20:43,254 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88209.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:21:01,194 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8416, 1.6051, 2.2723, 2.8259, 2.5447, 2.9656, 1.8669, 2.9434], device='cuda:7'), covar=tensor([0.0118, 0.0354, 0.0232, 0.0171, 0.0204, 0.0134, 0.0362, 0.0093], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0162, 0.0144, 0.0146, 0.0155, 0.0113, 0.0163, 0.0103], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 02:21:03,412 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 3.320e+02 4.209e+02 4.909e+02 8.638e+02, threshold=8.417e+02, percent-clipped=2.0 2023-04-29 02:21:25,621 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.43 vs. limit=5.0 2023-04-29 02:21:29,128 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88240.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:21:36,654 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88245.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:21:47,201 INFO [train.py:904] (7/8) Epoch 9, batch 7050, loss[loss=0.2446, simple_loss=0.3121, pruned_loss=0.08856, over 11548.00 frames. ], tot_loss[loss=0.233, simple_loss=0.315, pruned_loss=0.07549, over 3098756.13 frames. ], batch size: 247, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:22:00,414 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88261.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:22:41,151 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88288.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:22:44,390 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4522, 3.5201, 1.8339, 3.9572, 2.4613, 3.8628, 2.0106, 2.6695], device='cuda:7'), covar=tensor([0.0214, 0.0359, 0.1705, 0.0104, 0.0837, 0.0467, 0.1552, 0.0761], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0160, 0.0184, 0.0111, 0.0164, 0.0200, 0.0192, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:22:45,497 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88291.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:22:47,890 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88293.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:23:02,093 INFO [train.py:904] (7/8) Epoch 9, batch 7100, loss[loss=0.2087, simple_loss=0.2979, pruned_loss=0.05972, over 16896.00 frames. ], tot_loss[loss=0.2333, simple_loss=0.3144, pruned_loss=0.07613, over 3069368.39 frames. ], batch size: 96, lr: 7.53e-03, grad_scale: 4.0 2023-04-29 02:23:12,647 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88309.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:23:33,105 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.153e+02 3.325e+02 4.085e+02 5.101e+02 9.859e+02, threshold=8.169e+02, percent-clipped=1.0 2023-04-29 02:24:13,971 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-04-29 02:24:15,985 INFO [train.py:904] (7/8) Epoch 9, batch 7150, loss[loss=0.2029, simple_loss=0.2871, pruned_loss=0.05936, over 16832.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3116, pruned_loss=0.07517, over 3086703.13 frames. ], batch size: 116, lr: 7.52e-03, grad_scale: 4.0 2023-04-29 02:24:17,120 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88352.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:25:28,112 INFO [train.py:904] (7/8) Epoch 9, batch 7200, loss[loss=0.2265, simple_loss=0.2974, pruned_loss=0.07778, over 11584.00 frames. ], tot_loss[loss=0.2275, simple_loss=0.3088, pruned_loss=0.07303, over 3086963.70 frames. ], batch size: 247, lr: 7.52e-03, grad_scale: 8.0 2023-04-29 02:26:00,149 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.048e+02 3.005e+02 3.439e+02 4.228e+02 8.504e+02, threshold=6.879e+02, percent-clipped=1.0 2023-04-29 02:26:47,217 INFO [train.py:904] (7/8) Epoch 9, batch 7250, loss[loss=0.1991, simple_loss=0.2797, pruned_loss=0.05928, over 16839.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.306, pruned_loss=0.0711, over 3103339.61 frames. ], batch size: 116, lr: 7.52e-03, grad_scale: 8.0 2023-04-29 02:26:53,324 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88456.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:28:00,990 INFO [train.py:904] (7/8) Epoch 9, batch 7300, loss[loss=0.2069, simple_loss=0.306, pruned_loss=0.05392, over 16746.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.3056, pruned_loss=0.07135, over 3092187.47 frames. ], batch size: 89, lr: 7.52e-03, grad_scale: 8.0 2023-04-29 02:28:19,682 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2808, 4.3808, 4.5267, 4.4325, 4.4682, 4.9851, 4.5039, 4.2653], device='cuda:7'), covar=tensor([0.1481, 0.1819, 0.1523, 0.1700, 0.2415, 0.0987, 0.1701, 0.2690], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0453, 0.0486, 0.0403, 0.0530, 0.0514, 0.0393, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 02:28:33,455 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.255e+02 4.092e+02 5.788e+02 1.345e+03, threshold=8.184e+02, percent-clipped=12.0 2023-04-29 02:28:36,372 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3035, 2.8081, 2.5885, 2.3368, 2.3237, 2.1338, 2.8070, 2.8079], device='cuda:7'), covar=tensor([0.2005, 0.0739, 0.1313, 0.1596, 0.1799, 0.1659, 0.0476, 0.0883], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0255, 0.0280, 0.0269, 0.0280, 0.0214, 0.0263, 0.0281], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:29:14,118 INFO [train.py:904] (7/8) Epoch 9, batch 7350, loss[loss=0.2119, simple_loss=0.2964, pruned_loss=0.06375, over 16221.00 frames. ], tot_loss[loss=0.226, simple_loss=0.3068, pruned_loss=0.07259, over 3067275.53 frames. ], batch size: 165, lr: 7.52e-03, grad_scale: 4.0 2023-04-29 02:30:27,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9984, 3.4178, 3.2211, 1.7456, 2.8395, 2.3107, 3.4511, 3.4699], device='cuda:7'), covar=tensor([0.0263, 0.0607, 0.0607, 0.1999, 0.0798, 0.0919, 0.0672, 0.0866], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0140, 0.0158, 0.0143, 0.0135, 0.0127, 0.0138, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:30:27,931 INFO [train.py:904] (7/8) Epoch 9, batch 7400, loss[loss=0.212, simple_loss=0.2986, pruned_loss=0.0627, over 16900.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.3078, pruned_loss=0.07374, over 3046256.14 frames. ], batch size: 109, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:30:41,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 02:30:56,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4847, 4.7752, 4.5051, 4.5367, 4.2736, 4.2351, 4.3230, 4.8572], device='cuda:7'), covar=tensor([0.0977, 0.0931, 0.1127, 0.0691, 0.0788, 0.1115, 0.0895, 0.0865], device='cuda:7'), in_proj_covar=tensor([0.0489, 0.0617, 0.0520, 0.0418, 0.0381, 0.0400, 0.0511, 0.0465], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:31:01,753 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.901e+02 3.181e+02 3.584e+02 4.494e+02 7.659e+02, threshold=7.169e+02, percent-clipped=0.0 2023-04-29 02:31:38,446 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88647.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:31:44,622 INFO [train.py:904] (7/8) Epoch 9, batch 7450, loss[loss=0.3094, simple_loss=0.3594, pruned_loss=0.1297, over 11352.00 frames. ], tot_loss[loss=0.2304, simple_loss=0.3096, pruned_loss=0.07555, over 3041767.07 frames. ], batch size: 247, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:32:51,457 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-04-29 02:33:03,523 INFO [train.py:904] (7/8) Epoch 9, batch 7500, loss[loss=0.2003, simple_loss=0.2898, pruned_loss=0.05545, over 16697.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3096, pruned_loss=0.07522, over 3055423.05 frames. ], batch size: 89, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:33:36,931 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 3.551e+02 4.142e+02 5.326e+02 1.060e+03, threshold=8.283e+02, percent-clipped=5.0 2023-04-29 02:34:09,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3198, 1.9812, 1.6659, 1.8139, 2.3329, 2.0760, 2.3606, 2.5471], device='cuda:7'), covar=tensor([0.0081, 0.0264, 0.0322, 0.0288, 0.0149, 0.0234, 0.0141, 0.0144], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0187, 0.0185, 0.0184, 0.0185, 0.0187, 0.0186, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:34:18,168 INFO [train.py:904] (7/8) Epoch 9, batch 7550, loss[loss=0.2159, simple_loss=0.2988, pruned_loss=0.06649, over 17050.00 frames. ], tot_loss[loss=0.2283, simple_loss=0.3076, pruned_loss=0.07447, over 3073699.05 frames. ], batch size: 53, lr: 7.51e-03, grad_scale: 4.0 2023-04-29 02:34:23,607 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88756.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:34:45,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 02:35:25,774 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7692, 3.3849, 2.9116, 1.8132, 2.6564, 2.4433, 3.1753, 3.3992], device='cuda:7'), covar=tensor([0.0352, 0.0581, 0.0764, 0.1897, 0.0890, 0.0864, 0.0754, 0.0873], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0140, 0.0157, 0.0142, 0.0134, 0.0126, 0.0137, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:35:33,603 INFO [train.py:904] (7/8) Epoch 9, batch 7600, loss[loss=0.2275, simple_loss=0.3093, pruned_loss=0.07289, over 16244.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.3066, pruned_loss=0.07383, over 3082984.77 frames. ], batch size: 165, lr: 7.51e-03, grad_scale: 8.0 2023-04-29 02:35:37,865 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88804.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:36:05,557 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.836e+02 3.313e+02 3.789e+02 4.639e+02 1.052e+03, threshold=7.577e+02, percent-clipped=2.0 2023-04-29 02:36:45,926 INFO [train.py:904] (7/8) Epoch 9, batch 7650, loss[loss=0.2117, simple_loss=0.3061, pruned_loss=0.0587, over 16924.00 frames. ], tot_loss[loss=0.2277, simple_loss=0.307, pruned_loss=0.07416, over 3081539.34 frames. ], batch size: 96, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:37:40,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9818, 4.0382, 3.8437, 3.6980, 3.5705, 3.9563, 3.5911, 3.7309], device='cuda:7'), covar=tensor([0.0544, 0.0471, 0.0244, 0.0223, 0.0705, 0.0380, 0.0908, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0268, 0.0255, 0.0235, 0.0277, 0.0267, 0.0179, 0.0299], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:38:01,886 INFO [train.py:904] (7/8) Epoch 9, batch 7700, loss[loss=0.2159, simple_loss=0.3, pruned_loss=0.06588, over 16871.00 frames. ], tot_loss[loss=0.2293, simple_loss=0.3081, pruned_loss=0.07523, over 3079369.78 frames. ], batch size: 116, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:38:34,568 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 3.502e+02 4.245e+02 5.386e+02 9.706e+02, threshold=8.489e+02, percent-clipped=2.0 2023-04-29 02:39:08,834 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88947.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 02:39:16,836 INFO [train.py:904] (7/8) Epoch 9, batch 7750, loss[loss=0.1978, simple_loss=0.2912, pruned_loss=0.05218, over 16875.00 frames. ], tot_loss[loss=0.2291, simple_loss=0.3086, pruned_loss=0.07484, over 3098395.16 frames. ], batch size: 116, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:39:34,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2278, 4.4037, 4.1078, 3.9884, 3.3238, 4.3444, 4.1760, 3.9063], device='cuda:7'), covar=tensor([0.0863, 0.0617, 0.0461, 0.0384, 0.1873, 0.0497, 0.0655, 0.0733], device='cuda:7'), in_proj_covar=tensor([0.0223, 0.0269, 0.0257, 0.0236, 0.0278, 0.0270, 0.0180, 0.0301], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:40:20,795 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=88995.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:40:29,349 INFO [train.py:904] (7/8) Epoch 9, batch 7800, loss[loss=0.2178, simple_loss=0.298, pruned_loss=0.06879, over 16660.00 frames. ], tot_loss[loss=0.2301, simple_loss=0.3095, pruned_loss=0.07531, over 3104960.55 frames. ], batch size: 134, lr: 7.50e-03, grad_scale: 8.0 2023-04-29 02:41:02,677 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.440e+02 3.396e+02 4.306e+02 5.337e+02 1.534e+03, threshold=8.611e+02, percent-clipped=4.0 2023-04-29 02:41:07,705 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89027.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:41:42,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1166, 5.8204, 6.0650, 5.7579, 5.8168, 6.3026, 5.7995, 5.5182], device='cuda:7'), covar=tensor([0.0818, 0.1550, 0.1594, 0.1551, 0.1899, 0.0930, 0.1367, 0.2459], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0451, 0.0485, 0.0402, 0.0524, 0.0515, 0.0391, 0.0541], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 02:41:44,464 INFO [train.py:904] (7/8) Epoch 9, batch 7850, loss[loss=0.2155, simple_loss=0.3018, pruned_loss=0.06458, over 16850.00 frames. ], tot_loss[loss=0.2305, simple_loss=0.3103, pruned_loss=0.0753, over 3091778.64 frames. ], batch size: 116, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:42:39,023 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:42:57,238 INFO [train.py:904] (7/8) Epoch 9, batch 7900, loss[loss=0.2469, simple_loss=0.3254, pruned_loss=0.08418, over 16381.00 frames. ], tot_loss[loss=0.2292, simple_loss=0.3091, pruned_loss=0.07463, over 3084621.48 frames. ], batch size: 146, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:43:25,681 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-04-29 02:43:28,447 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.998e+02 3.084e+02 3.809e+02 4.466e+02 8.463e+02, threshold=7.617e+02, percent-clipped=0.0 2023-04-29 02:43:34,434 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1779, 4.2309, 4.4115, 4.2133, 4.2910, 4.7880, 4.3570, 4.0578], device='cuda:7'), covar=tensor([0.1519, 0.1745, 0.1743, 0.1989, 0.2456, 0.1049, 0.1457, 0.2733], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0454, 0.0486, 0.0404, 0.0527, 0.0516, 0.0393, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 02:44:12,933 INFO [train.py:904] (7/8) Epoch 9, batch 7950, loss[loss=0.2158, simple_loss=0.2987, pruned_loss=0.06643, over 16725.00 frames. ], tot_loss[loss=0.2306, simple_loss=0.3101, pruned_loss=0.07553, over 3075965.22 frames. ], batch size: 89, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:44:39,177 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89169.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:44:43,023 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0920, 3.6697, 3.4238, 1.9795, 2.9896, 2.3421, 3.5641, 3.7139], device='cuda:7'), covar=tensor([0.0263, 0.0543, 0.0545, 0.1720, 0.0775, 0.0894, 0.0657, 0.0814], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0137, 0.0155, 0.0140, 0.0132, 0.0123, 0.0135, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 02:45:26,763 INFO [train.py:904] (7/8) Epoch 9, batch 8000, loss[loss=0.2227, simple_loss=0.3116, pruned_loss=0.06687, over 16450.00 frames. ], tot_loss[loss=0.2311, simple_loss=0.3107, pruned_loss=0.07577, over 3091993.78 frames. ], batch size: 68, lr: 7.49e-03, grad_scale: 8.0 2023-04-29 02:45:59,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.399e+02 3.668e+02 4.137e+02 4.605e+02 8.803e+02, threshold=8.275e+02, percent-clipped=3.0 2023-04-29 02:46:09,204 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89230.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:46:39,460 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3260, 1.4727, 1.9846, 2.2224, 2.3268, 2.5675, 1.5141, 2.3645], device='cuda:7'), covar=tensor([0.0128, 0.0329, 0.0195, 0.0179, 0.0181, 0.0106, 0.0332, 0.0062], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0162, 0.0146, 0.0147, 0.0158, 0.0115, 0.0165, 0.0104], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 02:46:40,085 INFO [train.py:904] (7/8) Epoch 9, batch 8050, loss[loss=0.2326, simple_loss=0.3137, pruned_loss=0.07575, over 16365.00 frames. ], tot_loss[loss=0.231, simple_loss=0.3105, pruned_loss=0.07577, over 3075600.58 frames. ], batch size: 146, lr: 7.49e-03, grad_scale: 4.0 2023-04-29 02:47:55,032 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9917, 3.4258, 3.3812, 2.2367, 3.1212, 3.3172, 3.1942, 1.8049], device='cuda:7'), covar=tensor([0.0390, 0.0025, 0.0033, 0.0273, 0.0061, 0.0083, 0.0052, 0.0352], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0063, 0.0065, 0.0123, 0.0072, 0.0083, 0.0072, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 02:47:55,803 INFO [train.py:904] (7/8) Epoch 9, batch 8100, loss[loss=0.2389, simple_loss=0.3177, pruned_loss=0.0801, over 16247.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3101, pruned_loss=0.07491, over 3087339.05 frames. ], batch size: 165, lr: 7.48e-03, grad_scale: 4.0 2023-04-29 02:48:29,433 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.308e+02 2.948e+02 3.621e+02 4.670e+02 8.254e+02, threshold=7.243e+02, percent-clipped=0.0 2023-04-29 02:49:10,975 INFO [train.py:904] (7/8) Epoch 9, batch 8150, loss[loss=0.2294, simple_loss=0.2991, pruned_loss=0.07983, over 11489.00 frames. ], tot_loss[loss=0.2269, simple_loss=0.3073, pruned_loss=0.07323, over 3111075.67 frames. ], batch size: 247, lr: 7.48e-03, grad_scale: 4.0 2023-04-29 02:49:34,657 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8609, 4.0526, 4.4177, 2.3077, 4.6892, 4.6610, 3.2596, 3.2923], device='cuda:7'), covar=tensor([0.0699, 0.0141, 0.0103, 0.0977, 0.0033, 0.0066, 0.0323, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0099, 0.0087, 0.0142, 0.0068, 0.0096, 0.0120, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 02:49:45,633 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89374.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:49:58,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1510, 5.1261, 4.9178, 4.2236, 4.9796, 1.7139, 4.6655, 4.7503], device='cuda:7'), covar=tensor([0.0065, 0.0049, 0.0132, 0.0367, 0.0063, 0.2266, 0.0098, 0.0163], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0103, 0.0153, 0.0146, 0.0120, 0.0168, 0.0136, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:49:59,925 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89383.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:50:07,714 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-29 02:50:28,014 INFO [train.py:904] (7/8) Epoch 9, batch 8200, loss[loss=0.2292, simple_loss=0.3109, pruned_loss=0.07374, over 16783.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.3044, pruned_loss=0.07226, over 3106262.98 frames. ], batch size: 124, lr: 7.48e-03, grad_scale: 4.0 2023-04-29 02:51:05,348 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89425.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:51:06,153 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.992e+02 3.313e+02 4.005e+02 4.576e+02 8.683e+02, threshold=8.011e+02, percent-clipped=3.0 2023-04-29 02:51:20,837 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89435.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:51:27,474 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89439.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 02:51:47,440 INFO [train.py:904] (7/8) Epoch 9, batch 8250, loss[loss=0.2342, simple_loss=0.3295, pruned_loss=0.06942, over 16365.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.3039, pruned_loss=0.0705, over 3081544.47 frames. ], batch size: 146, lr: 7.48e-03, grad_scale: 2.0 2023-04-29 02:51:55,456 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2893, 4.2468, 4.1314, 3.9028, 3.7200, 4.1944, 3.9553, 3.9101], device='cuda:7'), covar=tensor([0.0471, 0.0476, 0.0262, 0.0258, 0.0963, 0.0424, 0.0558, 0.0615], device='cuda:7'), in_proj_covar=tensor([0.0224, 0.0272, 0.0260, 0.0240, 0.0281, 0.0272, 0.0180, 0.0305], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:52:22,333 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-04-29 02:52:43,699 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89486.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:53:06,385 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89500.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 02:53:08,817 INFO [train.py:904] (7/8) Epoch 9, batch 8300, loss[loss=0.1962, simple_loss=0.2933, pruned_loss=0.04957, over 16196.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.3014, pruned_loss=0.06765, over 3069300.68 frames. ], batch size: 165, lr: 7.48e-03, grad_scale: 2.0 2023-04-29 02:53:21,169 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89509.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:53:47,778 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89525.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:53:48,528 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.763e+02 2.745e+02 3.164e+02 3.682e+02 6.265e+02, threshold=6.329e+02, percent-clipped=0.0 2023-04-29 02:54:31,546 INFO [train.py:904] (7/8) Epoch 9, batch 8350, loss[loss=0.1954, simple_loss=0.2926, pruned_loss=0.04909, over 16852.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.3001, pruned_loss=0.06566, over 3055677.02 frames. ], batch size: 116, lr: 7.47e-03, grad_scale: 2.0 2023-04-29 02:55:01,207 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89570.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:55:08,565 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 02:55:20,941 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 02:55:49,015 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.52 vs. limit=5.0 2023-04-29 02:55:52,203 INFO [train.py:904] (7/8) Epoch 9, batch 8400, loss[loss=0.1911, simple_loss=0.276, pruned_loss=0.05309, over 12244.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2971, pruned_loss=0.06327, over 3046939.90 frames. ], batch size: 248, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 02:56:31,317 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.203e+02 2.904e+02 3.369e+02 3.930e+02 8.032e+02, threshold=6.737e+02, percent-clipped=2.0 2023-04-29 02:57:13,776 INFO [train.py:904] (7/8) Epoch 9, batch 8450, loss[loss=0.1709, simple_loss=0.2703, pruned_loss=0.03581, over 17263.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.295, pruned_loss=0.06129, over 3060802.51 frames. ], batch size: 52, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 02:57:26,062 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3356, 4.1420, 4.3986, 4.5489, 4.6816, 4.1623, 4.6281, 4.6467], device='cuda:7'), covar=tensor([0.1380, 0.1109, 0.1181, 0.0587, 0.0463, 0.1056, 0.0490, 0.0574], device='cuda:7'), in_proj_covar=tensor([0.0471, 0.0582, 0.0707, 0.0587, 0.0454, 0.0454, 0.0471, 0.0531], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:57:44,477 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.38 vs. limit=5.0 2023-04-29 02:58:00,189 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:58:04,757 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89683.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:58:34,973 INFO [train.py:904] (7/8) Epoch 9, batch 8500, loss[loss=0.1817, simple_loss=0.27, pruned_loss=0.04668, over 15239.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2911, pruned_loss=0.05855, over 3066889.55 frames. ], batch size: 190, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 02:59:08,298 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1781, 4.1960, 4.0174, 3.8213, 3.7003, 4.1658, 3.9072, 3.8583], device='cuda:7'), covar=tensor([0.0514, 0.0453, 0.0265, 0.0240, 0.0785, 0.0379, 0.0484, 0.0578], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0265, 0.0254, 0.0235, 0.0275, 0.0265, 0.0176, 0.0297], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 02:59:14,100 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.444e+02 3.113e+02 3.731e+02 7.658e+02, threshold=6.225e+02, percent-clipped=1.0 2023-04-29 02:59:20,993 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89730.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:59:22,204 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=89731.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:59:39,437 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89741.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 02:59:47,158 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-29 02:59:58,605 INFO [train.py:904] (7/8) Epoch 9, batch 8550, loss[loss=0.1814, simple_loss=0.2793, pruned_loss=0.04181, over 16835.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2887, pruned_loss=0.05749, over 3043055.52 frames. ], batch size: 96, lr: 7.47e-03, grad_scale: 4.0 2023-04-29 03:00:00,481 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-29 03:00:55,447 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:01:12,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5853, 2.8498, 2.5058, 4.1183, 2.7799, 4.1227, 1.3783, 3.0426], device='cuda:7'), covar=tensor([0.1353, 0.0579, 0.1005, 0.0139, 0.0144, 0.0317, 0.1482, 0.0612], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0149, 0.0173, 0.0124, 0.0193, 0.0201, 0.0173, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 03:01:24,774 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89795.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 03:01:37,952 INFO [train.py:904] (7/8) Epoch 9, batch 8600, loss[loss=0.1828, simple_loss=0.2642, pruned_loss=0.05072, over 12224.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2887, pruned_loss=0.0567, over 3036344.72 frames. ], batch size: 246, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:02:25,781 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89825.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:02:26,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.775e+02 2.695e+02 3.313e+02 4.176e+02 8.025e+02, threshold=6.625e+02, percent-clipped=2.0 2023-04-29 03:03:15,787 INFO [train.py:904] (7/8) Epoch 9, batch 8650, loss[loss=0.1713, simple_loss=0.2598, pruned_loss=0.04136, over 12050.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2863, pruned_loss=0.05407, over 3066221.62 frames. ], batch size: 247, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:03:47,703 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:03:47,824 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:03:55,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6047, 2.7195, 2.4427, 3.9614, 2.5599, 4.0507, 1.3047, 2.9367], device='cuda:7'), covar=tensor([0.1578, 0.0676, 0.1169, 0.0173, 0.0144, 0.0311, 0.1799, 0.0683], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0150, 0.0174, 0.0124, 0.0192, 0.0202, 0.0173, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 03:04:05,356 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=89873.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:04:35,237 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 03:04:45,303 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3883, 3.4639, 2.0560, 3.7455, 2.4842, 3.7278, 1.9376, 2.7621], device='cuda:7'), covar=tensor([0.0200, 0.0334, 0.1496, 0.0127, 0.0840, 0.0462, 0.1613, 0.0644], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0153, 0.0179, 0.0108, 0.0160, 0.0190, 0.0187, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 03:04:55,403 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1178, 1.3475, 1.7681, 2.0499, 2.0618, 2.3199, 1.6161, 2.1871], device='cuda:7'), covar=tensor([0.0130, 0.0324, 0.0187, 0.0192, 0.0201, 0.0118, 0.0282, 0.0095], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0161, 0.0145, 0.0144, 0.0156, 0.0111, 0.0162, 0.0102], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 03:04:57,405 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-04-29 03:05:02,355 INFO [train.py:904] (7/8) Epoch 9, batch 8700, loss[loss=0.2007, simple_loss=0.285, pruned_loss=0.0582, over 16289.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2832, pruned_loss=0.05256, over 3070549.42 frames. ], batch size: 165, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:05:45,062 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.547e+02 3.087e+02 3.551e+02 5.785e+02, threshold=6.175e+02, percent-clipped=0.0 2023-04-29 03:05:45,920 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89926.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 03:06:18,824 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 03:06:36,272 INFO [train.py:904] (7/8) Epoch 9, batch 8750, loss[loss=0.2069, simple_loss=0.2986, pruned_loss=0.05762, over 16188.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2823, pruned_loss=0.0517, over 3067018.74 frames. ], batch size: 165, lr: 7.46e-03, grad_scale: 4.0 2023-04-29 03:07:14,212 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 03:07:23,052 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2023-04-29 03:08:32,647 INFO [train.py:904] (7/8) Epoch 9, batch 8800, loss[loss=0.1745, simple_loss=0.2641, pruned_loss=0.04248, over 16723.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2801, pruned_loss=0.05024, over 3077716.71 frames. ], batch size: 76, lr: 7.46e-03, grad_scale: 8.0 2023-04-29 03:09:09,955 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 03:09:21,971 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.546e+02 3.019e+02 3.613e+02 7.330e+02, threshold=6.037e+02, percent-clipped=3.0 2023-04-29 03:09:31,882 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90030.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:09:44,820 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:10:17,869 INFO [train.py:904] (7/8) Epoch 9, batch 8850, loss[loss=0.1655, simple_loss=0.2516, pruned_loss=0.03971, over 12144.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2822, pruned_loss=0.04999, over 3068296.64 frames. ], batch size: 246, lr: 7.45e-03, grad_scale: 8.0 2023-04-29 03:10:22,191 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3611, 2.9696, 2.5793, 2.1776, 2.1545, 2.0626, 2.9404, 2.7791], device='cuda:7'), covar=tensor([0.2233, 0.0674, 0.1289, 0.1834, 0.2066, 0.1615, 0.0431, 0.1019], device='cuda:7'), in_proj_covar=tensor([0.0284, 0.0240, 0.0267, 0.0257, 0.0253, 0.0204, 0.0249, 0.0263], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:11:12,094 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90078.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:11:18,704 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90081.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:11:37,085 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 03:11:38,904 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0663, 5.4076, 5.2046, 5.1627, 4.8013, 4.7644, 4.8294, 5.4356], device='cuda:7'), covar=tensor([0.0900, 0.0707, 0.0828, 0.0643, 0.0712, 0.0719, 0.0895, 0.0793], device='cuda:7'), in_proj_covar=tensor([0.0470, 0.0598, 0.0493, 0.0407, 0.0367, 0.0392, 0.0498, 0.0451], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:11:41,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3232, 1.9311, 2.0119, 3.8500, 1.9146, 2.3910, 2.0842, 2.0697], device='cuda:7'), covar=tensor([0.0820, 0.3103, 0.2188, 0.0385, 0.3925, 0.2110, 0.2954, 0.3157], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0367, 0.0311, 0.0309, 0.0400, 0.0413, 0.0329, 0.0427], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:11:50,489 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90095.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 03:12:02,865 INFO [train.py:904] (7/8) Epoch 9, batch 8900, loss[loss=0.1758, simple_loss=0.276, pruned_loss=0.0378, over 16898.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2822, pruned_loss=0.04919, over 3055551.13 frames. ], batch size: 96, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:12:57,515 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.859e+02 2.641e+02 3.095e+02 3.699e+02 6.742e+02, threshold=6.190e+02, percent-clipped=4.0 2023-04-29 03:13:07,419 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90129.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:13:10,508 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90130.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:13:10,679 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9817, 2.0167, 2.3281, 3.2785, 2.0606, 2.3026, 2.1967, 2.1008], device='cuda:7'), covar=tensor([0.0807, 0.2909, 0.1749, 0.0456, 0.3564, 0.1944, 0.2718, 0.2988], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0366, 0.0311, 0.0308, 0.0400, 0.0411, 0.0329, 0.0426], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:13:48,682 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90143.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 03:14:08,715 INFO [train.py:904] (7/8) Epoch 9, batch 8950, loss[loss=0.1918, simple_loss=0.2741, pruned_loss=0.0547, over 12792.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2821, pruned_loss=0.04967, over 3063011.62 frames. ], batch size: 248, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:14:25,994 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-29 03:14:39,029 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90165.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:14:45,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9176, 3.3336, 3.3618, 1.8590, 2.8118, 2.2928, 3.4346, 3.3901], device='cuda:7'), covar=tensor([0.0212, 0.0593, 0.0500, 0.1684, 0.0708, 0.0860, 0.0582, 0.0709], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0131, 0.0153, 0.0139, 0.0131, 0.0123, 0.0133, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 03:15:35,826 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90191.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:15:38,471 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 03:15:57,362 INFO [train.py:904] (7/8) Epoch 9, batch 9000, loss[loss=0.1575, simple_loss=0.2509, pruned_loss=0.032, over 16928.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2786, pruned_loss=0.048, over 3058250.62 frames. ], batch size: 102, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:15:57,363 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 03:16:07,532 INFO [train.py:938] (7/8) Epoch 9, validation: loss=0.1581, simple_loss=0.2623, pruned_loss=0.02697, over 944034.00 frames. 2023-04-29 03:16:07,533 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 03:16:31,273 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90213.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:16:48,426 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90221.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 03:16:53,706 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-29 03:16:58,711 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.410e+02 2.881e+02 3.616e+02 7.746e+02, threshold=5.761e+02, percent-clipped=2.0 2023-04-29 03:17:09,244 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90232.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:17:51,879 INFO [train.py:904] (7/8) Epoch 9, batch 9050, loss[loss=0.2055, simple_loss=0.2831, pruned_loss=0.06402, over 12709.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2804, pruned_loss=0.04906, over 3075994.76 frames. ], batch size: 246, lr: 7.45e-03, grad_scale: 4.0 2023-04-29 03:18:03,794 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1228, 2.8991, 2.8138, 2.0170, 2.6128, 2.1819, 2.7329, 2.9590], device='cuda:7'), covar=tensor([0.0296, 0.0556, 0.0445, 0.1549, 0.0667, 0.0841, 0.0677, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0130, 0.0152, 0.0138, 0.0130, 0.0122, 0.0133, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 03:19:19,963 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90293.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:19:36,855 INFO [train.py:904] (7/8) Epoch 9, batch 9100, loss[loss=0.1861, simple_loss=0.2697, pruned_loss=0.05127, over 12242.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2797, pruned_loss=0.0495, over 3052296.67 frames. ], batch size: 248, lr: 7.44e-03, grad_scale: 4.0 2023-04-29 03:20:34,112 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.733e+02 2.617e+02 3.274e+02 4.462e+02 7.495e+02, threshold=6.548e+02, percent-clipped=8.0 2023-04-29 03:20:58,667 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:21:34,629 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90351.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:21:35,380 INFO [train.py:904] (7/8) Epoch 9, batch 9150, loss[loss=0.1799, simple_loss=0.2687, pruned_loss=0.04553, over 16725.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2806, pruned_loss=0.04905, over 3079548.96 frames. ], batch size: 134, lr: 7.44e-03, grad_scale: 4.0 2023-04-29 03:21:47,332 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 03:22:14,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6449, 4.9285, 4.3962, 4.8610, 4.5240, 4.3840, 4.5177, 4.9695], device='cuda:7'), covar=tensor([0.1550, 0.1225, 0.2205, 0.0894, 0.1146, 0.1382, 0.1593, 0.1390], device='cuda:7'), in_proj_covar=tensor([0.0467, 0.0593, 0.0488, 0.0404, 0.0366, 0.0387, 0.0494, 0.0444], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:22:45,426 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:23:22,759 INFO [train.py:904] (7/8) Epoch 9, batch 9200, loss[loss=0.1787, simple_loss=0.2738, pruned_loss=0.04176, over 16299.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2761, pruned_loss=0.04811, over 3058681.66 frames. ], batch size: 165, lr: 7.44e-03, grad_scale: 8.0 2023-04-29 03:23:42,457 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90412.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:24:07,629 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.608e+02 2.952e+02 3.728e+02 7.219e+02, threshold=5.904e+02, percent-clipped=3.0 2023-04-29 03:24:58,585 INFO [train.py:904] (7/8) Epoch 9, batch 9250, loss[loss=0.1717, simple_loss=0.2687, pruned_loss=0.03735, over 15403.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2758, pruned_loss=0.0482, over 3039725.85 frames. ], batch size: 190, lr: 7.44e-03, grad_scale: 8.0 2023-04-29 03:25:14,868 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90459.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:26:13,027 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90486.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:26:34,801 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-29 03:26:38,541 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90496.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:26:49,743 INFO [train.py:904] (7/8) Epoch 9, batch 9300, loss[loss=0.1579, simple_loss=0.2471, pruned_loss=0.0344, over 16535.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2741, pruned_loss=0.04754, over 3034294.27 frames. ], batch size: 68, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:27:35,128 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90520.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:27:36,970 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90521.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 03:27:46,916 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.664e+02 2.420e+02 2.895e+02 3.675e+02 7.033e+02, threshold=5.790e+02, percent-clipped=1.0 2023-04-29 03:28:13,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0475, 3.0615, 1.7681, 3.3216, 2.3336, 3.2764, 1.9415, 2.5578], device='cuda:7'), covar=tensor([0.0213, 0.0385, 0.1509, 0.0169, 0.0803, 0.0524, 0.1422, 0.0601], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0154, 0.0180, 0.0108, 0.0159, 0.0188, 0.0187, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 03:28:36,081 INFO [train.py:904] (7/8) Epoch 9, batch 9350, loss[loss=0.1979, simple_loss=0.2759, pruned_loss=0.05993, over 12299.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2746, pruned_loss=0.04735, over 3050102.72 frames. ], batch size: 246, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:28:48,558 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90557.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:29:14,441 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90569.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:29:50,799 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90588.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:30:16,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6204, 3.2542, 3.1621, 1.8452, 2.7252, 2.2439, 3.1144, 3.2907], device='cuda:7'), covar=tensor([0.0300, 0.0601, 0.0505, 0.1855, 0.0759, 0.0947, 0.0714, 0.0786], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0130, 0.0151, 0.0140, 0.0130, 0.0123, 0.0132, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 03:30:18,691 INFO [train.py:904] (7/8) Epoch 9, batch 9400, loss[loss=0.1891, simple_loss=0.2795, pruned_loss=0.04937, over 12468.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2745, pruned_loss=0.047, over 3038968.32 frames. ], batch size: 248, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:30:29,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9709, 1.9633, 2.2823, 3.2136, 2.0652, 2.2047, 2.2193, 2.0534], device='cuda:7'), covar=tensor([0.0875, 0.3212, 0.1860, 0.0473, 0.3677, 0.2235, 0.2609, 0.3205], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0362, 0.0308, 0.0305, 0.0394, 0.0405, 0.0326, 0.0422], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:30:58,429 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 03:31:09,567 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.474e+02 2.959e+02 3.714e+02 8.908e+02, threshold=5.918e+02, percent-clipped=5.0 2023-04-29 03:32:00,907 INFO [train.py:904] (7/8) Epoch 9, batch 9450, loss[loss=0.1921, simple_loss=0.2982, pruned_loss=0.04299, over 15512.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.276, pruned_loss=0.04733, over 3032461.13 frames. ], batch size: 191, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:32:38,782 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6055, 4.0533, 4.1994, 2.7898, 3.7609, 4.1219, 3.8754, 2.5793], device='cuda:7'), covar=tensor([0.0340, 0.0020, 0.0021, 0.0263, 0.0059, 0.0055, 0.0040, 0.0298], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0062, 0.0064, 0.0121, 0.0072, 0.0080, 0.0071, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 03:32:53,635 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0317, 4.0280, 3.8986, 3.4333, 3.9602, 1.6030, 3.7496, 3.6541], device='cuda:7'), covar=tensor([0.0071, 0.0067, 0.0126, 0.0209, 0.0069, 0.2364, 0.0101, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0112, 0.0099, 0.0144, 0.0133, 0.0115, 0.0165, 0.0130, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-29 03:33:41,135 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6405, 3.7228, 3.0818, 2.1991, 2.5133, 2.3110, 4.0208, 3.4811], device='cuda:7'), covar=tensor([0.2479, 0.0582, 0.1412, 0.2201, 0.2258, 0.1718, 0.0356, 0.0887], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0244, 0.0270, 0.0259, 0.0248, 0.0207, 0.0251, 0.0265], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:33:43,982 INFO [train.py:904] (7/8) Epoch 9, batch 9500, loss[loss=0.1597, simple_loss=0.2446, pruned_loss=0.03741, over 12953.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2748, pruned_loss=0.04681, over 3032932.16 frames. ], batch size: 248, lr: 7.43e-03, grad_scale: 8.0 2023-04-29 03:33:58,720 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:34:05,429 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-04-29 03:34:35,318 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 2.298e+02 2.954e+02 3.660e+02 6.291e+02, threshold=5.908e+02, percent-clipped=3.0 2023-04-29 03:34:40,639 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-29 03:35:30,047 INFO [train.py:904] (7/8) Epoch 9, batch 9550, loss[loss=0.2245, simple_loss=0.3151, pruned_loss=0.06691, over 15325.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2741, pruned_loss=0.04695, over 3036346.66 frames. ], batch size: 190, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:36:36,602 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1227, 2.8317, 2.7728, 2.0197, 2.5944, 2.1999, 2.6968, 2.9208], device='cuda:7'), covar=tensor([0.0333, 0.0702, 0.0511, 0.1684, 0.0700, 0.0887, 0.0714, 0.0686], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0130, 0.0152, 0.0140, 0.0130, 0.0123, 0.0132, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 03:36:43,126 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90786.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:37:11,178 INFO [train.py:904] (7/8) Epoch 9, batch 9600, loss[loss=0.2191, simple_loss=0.2888, pruned_loss=0.07474, over 12352.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2757, pruned_loss=0.04786, over 3025147.51 frames. ], batch size: 248, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:37:35,923 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 03:37:37,097 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90815.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:37:59,122 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.542e+02 3.036e+02 4.023e+02 8.440e+02, threshold=6.073e+02, percent-clipped=4.0 2023-04-29 03:38:17,859 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90834.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:38:55,264 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90849.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:39:01,500 INFO [train.py:904] (7/8) Epoch 9, batch 9650, loss[loss=0.1866, simple_loss=0.276, pruned_loss=0.04856, over 16864.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2777, pruned_loss=0.04809, over 3038266.88 frames. ], batch size: 116, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:39:03,593 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:39:12,401 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4892, 3.1026, 3.1277, 2.0220, 2.8046, 2.2056, 2.9650, 3.0423], device='cuda:7'), covar=tensor([0.0359, 0.0609, 0.0450, 0.1621, 0.0692, 0.0883, 0.0802, 0.0860], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0129, 0.0152, 0.0140, 0.0130, 0.0122, 0.0132, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 03:40:19,175 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90888.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:40:42,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0816, 4.0829, 4.4989, 4.4699, 4.4641, 4.1358, 4.2145, 4.1136], device='cuda:7'), covar=tensor([0.0300, 0.0612, 0.0414, 0.0417, 0.0425, 0.0404, 0.0730, 0.0388], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0294, 0.0300, 0.0292, 0.0339, 0.0317, 0.0410, 0.0256], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-29 03:40:48,113 INFO [train.py:904] (7/8) Epoch 9, batch 9700, loss[loss=0.1806, simple_loss=0.2732, pruned_loss=0.04399, over 15175.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2768, pruned_loss=0.0479, over 3043920.89 frames. ], batch size: 190, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:41:04,671 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90910.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:41:40,522 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.339e+02 3.062e+02 3.716e+02 7.920e+02, threshold=6.123e+02, percent-clipped=1.0 2023-04-29 03:42:00,641 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=90936.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:42:31,495 INFO [train.py:904] (7/8) Epoch 9, batch 9750, loss[loss=0.1678, simple_loss=0.2646, pruned_loss=0.03547, over 16803.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2754, pruned_loss=0.04766, over 3052642.94 frames. ], batch size: 90, lr: 7.42e-03, grad_scale: 8.0 2023-04-29 03:43:57,965 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-29 03:44:10,872 INFO [train.py:904] (7/8) Epoch 9, batch 9800, loss[loss=0.1645, simple_loss=0.2655, pruned_loss=0.03169, over 16863.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2754, pruned_loss=0.04635, over 3068788.07 frames. ], batch size: 96, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:44:21,914 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91007.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:44:41,391 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9815, 4.0159, 4.3652, 4.3459, 4.3653, 4.0931, 4.0728, 3.9918], device='cuda:7'), covar=tensor([0.0281, 0.0474, 0.0406, 0.0451, 0.0408, 0.0312, 0.0786, 0.0379], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0292, 0.0297, 0.0286, 0.0333, 0.0312, 0.0402, 0.0251], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 03:44:57,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.313e+02 2.751e+02 3.431e+02 5.847e+02, threshold=5.502e+02, percent-clipped=0.0 2023-04-29 03:45:57,963 INFO [train.py:904] (7/8) Epoch 9, batch 9850, loss[loss=0.1844, simple_loss=0.276, pruned_loss=0.04638, over 16188.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.276, pruned_loss=0.04614, over 3063137.10 frames. ], batch size: 165, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:46:05,840 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91055.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:46:47,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3445, 3.0294, 2.5572, 2.1314, 2.1665, 1.9295, 2.9958, 2.8350], device='cuda:7'), covar=tensor([0.2250, 0.0735, 0.1526, 0.2202, 0.2427, 0.2036, 0.0511, 0.1041], device='cuda:7'), in_proj_covar=tensor([0.0288, 0.0244, 0.0268, 0.0260, 0.0245, 0.0207, 0.0251, 0.0264], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:47:48,706 INFO [train.py:904] (7/8) Epoch 9, batch 9900, loss[loss=0.1961, simple_loss=0.2806, pruned_loss=0.05583, over 12425.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2761, pruned_loss=0.04629, over 3036680.80 frames. ], batch size: 246, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:48:20,588 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:48:47,940 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.434e+02 3.232e+02 4.008e+02 9.044e+02, threshold=6.464e+02, percent-clipped=5.0 2023-04-29 03:49:47,999 INFO [train.py:904] (7/8) Epoch 9, batch 9950, loss[loss=0.1835, simple_loss=0.2754, pruned_loss=0.04581, over 16424.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2788, pruned_loss=0.04684, over 3055682.35 frames. ], batch size: 68, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:49:49,187 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91152.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:50:00,149 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3069, 1.9488, 1.7034, 1.6948, 2.2005, 1.9902, 2.1140, 2.3338], device='cuda:7'), covar=tensor([0.0066, 0.0237, 0.0292, 0.0299, 0.0125, 0.0204, 0.0139, 0.0139], device='cuda:7'), in_proj_covar=tensor([0.0117, 0.0190, 0.0184, 0.0184, 0.0184, 0.0187, 0.0178, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:50:14,540 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91163.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:50:50,881 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 03:51:19,955 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9898, 5.3140, 5.0552, 5.0676, 4.7172, 4.7516, 4.7183, 5.3738], device='cuda:7'), covar=tensor([0.0798, 0.0759, 0.0834, 0.0574, 0.0683, 0.0703, 0.0824, 0.0773], device='cuda:7'), in_proj_covar=tensor([0.0468, 0.0593, 0.0486, 0.0406, 0.0371, 0.0385, 0.0493, 0.0442], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:51:20,043 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7024, 4.7020, 4.4778, 4.0836, 4.5748, 1.5818, 4.3037, 4.4051], device='cuda:7'), covar=tensor([0.0059, 0.0062, 0.0109, 0.0216, 0.0059, 0.2327, 0.0096, 0.0155], device='cuda:7'), in_proj_covar=tensor([0.0113, 0.0100, 0.0145, 0.0132, 0.0116, 0.0167, 0.0132, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-29 03:51:44,352 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91200.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:51:47,249 INFO [train.py:904] (7/8) Epoch 9, batch 10000, loss[loss=0.1947, simple_loss=0.2926, pruned_loss=0.04833, over 15317.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2776, pruned_loss=0.04646, over 3077900.68 frames. ], batch size: 191, lr: 7.41e-03, grad_scale: 8.0 2023-04-29 03:51:53,884 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91205.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:52:35,747 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.574e+02 2.985e+02 3.622e+02 7.238e+02, threshold=5.969e+02, percent-clipped=2.0 2023-04-29 03:52:37,462 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2023-04-29 03:53:27,231 INFO [train.py:904] (7/8) Epoch 9, batch 10050, loss[loss=0.1837, simple_loss=0.2808, pruned_loss=0.04332, over 16916.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2771, pruned_loss=0.04574, over 3082670.00 frames. ], batch size: 116, lr: 7.40e-03, grad_scale: 8.0 2023-04-29 03:53:33,748 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4624, 3.8706, 4.0209, 2.7314, 3.6622, 4.0442, 3.7943, 2.4901], device='cuda:7'), covar=tensor([0.0349, 0.0026, 0.0025, 0.0243, 0.0056, 0.0041, 0.0038, 0.0294], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0062, 0.0064, 0.0120, 0.0072, 0.0078, 0.0071, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 03:54:00,794 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5843, 2.0532, 2.1803, 4.2663, 1.9541, 2.5903, 2.2582, 2.2789], device='cuda:7'), covar=tensor([0.0743, 0.3107, 0.2035, 0.0309, 0.3764, 0.1990, 0.2691, 0.3005], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0360, 0.0306, 0.0304, 0.0390, 0.0400, 0.0322, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:54:30,776 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91285.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:55:01,276 INFO [train.py:904] (7/8) Epoch 9, batch 10100, loss[loss=0.1803, simple_loss=0.2714, pruned_loss=0.04461, over 16326.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2773, pruned_loss=0.04593, over 3072500.07 frames. ], batch size: 166, lr: 7.40e-03, grad_scale: 8.0 2023-04-29 03:55:52,064 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.558e+02 3.237e+02 3.853e+02 8.628e+02, threshold=6.474e+02, percent-clipped=2.0 2023-04-29 03:56:14,374 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91346.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 03:56:45,268 INFO [train.py:904] (7/8) Epoch 10, batch 0, loss[loss=0.1945, simple_loss=0.2798, pruned_loss=0.05461, over 17122.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2798, pruned_loss=0.05461, over 17122.00 frames. ], batch size: 48, lr: 7.04e-03, grad_scale: 8.0 2023-04-29 03:56:45,268 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 03:56:50,434 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7734, 6.1488, 5.7853, 6.0217, 5.6449, 5.5997, 5.5513, 6.1550], device='cuda:7'), covar=tensor([0.0866, 0.0666, 0.0743, 0.0567, 0.0769, 0.0287, 0.0855, 0.0648], device='cuda:7'), in_proj_covar=tensor([0.0467, 0.0596, 0.0485, 0.0407, 0.0373, 0.0388, 0.0496, 0.0443], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:56:52,894 INFO [train.py:938] (7/8) Epoch 10, validation: loss=0.158, simple_loss=0.2614, pruned_loss=0.02732, over 944034.00 frames. 2023-04-29 03:56:52,894 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 03:58:02,504 INFO [train.py:904] (7/8) Epoch 10, batch 50, loss[loss=0.2237, simple_loss=0.3011, pruned_loss=0.07316, over 16873.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.2935, pruned_loss=0.071, over 749303.10 frames. ], batch size: 109, lr: 7.04e-03, grad_scale: 2.0 2023-04-29 03:58:07,599 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 03:58:09,085 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1750, 1.8986, 2.1445, 3.6162, 1.9562, 2.2153, 2.0678, 2.0464], device='cuda:7'), covar=tensor([0.0864, 0.3107, 0.2131, 0.0503, 0.3484, 0.2188, 0.2891, 0.2775], device='cuda:7'), in_proj_covar=tensor([0.0337, 0.0362, 0.0308, 0.0305, 0.0392, 0.0403, 0.0326, 0.0419], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 03:58:17,043 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9604, 3.8917, 4.3349, 4.3339, 4.3509, 4.0285, 4.0600, 4.0025], device='cuda:7'), covar=tensor([0.0307, 0.0556, 0.0392, 0.0404, 0.0403, 0.0365, 0.0799, 0.0437], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0293, 0.0297, 0.0283, 0.0331, 0.0313, 0.0404, 0.0252], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 03:58:39,942 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.984e+02 3.603e+02 4.569e+02 8.591e+02, threshold=7.207e+02, percent-clipped=1.0 2023-04-29 03:59:08,808 INFO [train.py:904] (7/8) Epoch 10, batch 100, loss[loss=0.2129, simple_loss=0.3004, pruned_loss=0.06272, over 17104.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2889, pruned_loss=0.06648, over 1309089.56 frames. ], batch size: 53, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 03:59:38,050 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-04-29 03:59:58,522 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4607, 4.3654, 4.9345, 4.8835, 4.9446, 4.4943, 4.5810, 4.3939], device='cuda:7'), covar=tensor([0.0362, 0.0617, 0.0358, 0.0475, 0.0446, 0.0420, 0.0875, 0.0495], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0303, 0.0307, 0.0293, 0.0342, 0.0325, 0.0417, 0.0261], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-29 04:00:16,898 INFO [train.py:904] (7/8) Epoch 10, batch 150, loss[loss=0.1938, simple_loss=0.2809, pruned_loss=0.05332, over 17190.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2853, pruned_loss=0.06417, over 1754473.99 frames. ], batch size: 46, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:00:22,900 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91505.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:00:56,397 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.683e+02 3.349e+02 4.064e+02 6.042e+02, threshold=6.698e+02, percent-clipped=0.0 2023-04-29 04:01:12,437 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91542.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 04:01:17,742 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 04:01:26,562 INFO [train.py:904] (7/8) Epoch 10, batch 200, loss[loss=0.2302, simple_loss=0.2898, pruned_loss=0.0853, over 16716.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2848, pruned_loss=0.06335, over 2106946.00 frames. ], batch size: 134, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:01:28,066 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91553.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:01:51,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8268, 3.6747, 2.9202, 2.3551, 2.7131, 2.2156, 3.8038, 3.4781], device='cuda:7'), covar=tensor([0.2219, 0.0742, 0.1372, 0.2162, 0.1939, 0.1647, 0.0527, 0.1185], device='cuda:7'), in_proj_covar=tensor([0.0293, 0.0250, 0.0277, 0.0267, 0.0255, 0.0213, 0.0261, 0.0278], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:02:34,677 INFO [train.py:904] (7/8) Epoch 10, batch 250, loss[loss=0.2056, simple_loss=0.2726, pruned_loss=0.06934, over 16879.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2814, pruned_loss=0.06205, over 2382898.92 frames. ], batch size: 109, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:02:36,459 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2065, 5.2513, 5.0086, 4.5266, 5.0243, 1.9071, 4.7985, 5.0384], device='cuda:7'), covar=tensor([0.0070, 0.0059, 0.0133, 0.0306, 0.0075, 0.2120, 0.0102, 0.0152], device='cuda:7'), in_proj_covar=tensor([0.0119, 0.0106, 0.0153, 0.0141, 0.0123, 0.0174, 0.0140, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:02:36,497 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91603.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 04:02:52,863 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8567, 3.0599, 2.7419, 4.4370, 3.7973, 4.3256, 1.6179, 3.1120], device='cuda:7'), covar=tensor([0.1225, 0.0503, 0.0878, 0.0123, 0.0197, 0.0335, 0.1284, 0.0647], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0153, 0.0176, 0.0128, 0.0185, 0.0207, 0.0176, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:03:11,344 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.504e+02 3.085e+02 3.691e+02 6.376e+02, threshold=6.169e+02, percent-clipped=0.0 2023-04-29 04:03:28,156 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91641.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:03:42,034 INFO [train.py:904] (7/8) Epoch 10, batch 300, loss[loss=0.1708, simple_loss=0.2626, pruned_loss=0.03954, over 16846.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2781, pruned_loss=0.06028, over 2592417.05 frames. ], batch size: 42, lr: 7.03e-03, grad_scale: 1.0 2023-04-29 04:04:46,229 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2023-04-29 04:04:51,285 INFO [train.py:904] (7/8) Epoch 10, batch 350, loss[loss=0.1878, simple_loss=0.2696, pruned_loss=0.05303, over 16783.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2749, pruned_loss=0.05793, over 2752736.59 frames. ], batch size: 102, lr: 7.02e-03, grad_scale: 1.0 2023-04-29 04:05:02,219 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7283, 2.5022, 2.3453, 3.6180, 3.0047, 3.8043, 1.4271, 2.8060], device='cuda:7'), covar=tensor([0.1334, 0.0644, 0.1087, 0.0150, 0.0178, 0.0375, 0.1500, 0.0723], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0152, 0.0176, 0.0128, 0.0186, 0.0206, 0.0175, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:05:28,622 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.384e+02 2.940e+02 3.585e+02 5.710e+02, threshold=5.881e+02, percent-clipped=0.0 2023-04-29 04:05:59,428 INFO [train.py:904] (7/8) Epoch 10, batch 400, loss[loss=0.2135, simple_loss=0.2727, pruned_loss=0.07714, over 16893.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2737, pruned_loss=0.05771, over 2877041.41 frames. ], batch size: 109, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:07:11,422 INFO [train.py:904] (7/8) Epoch 10, batch 450, loss[loss=0.1892, simple_loss=0.2849, pruned_loss=0.04678, over 17110.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2727, pruned_loss=0.05663, over 2975149.77 frames. ], batch size: 47, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:07:46,581 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9065, 2.1445, 2.3100, 4.6848, 2.2245, 2.6885, 2.3136, 2.4534], device='cuda:7'), covar=tensor([0.0733, 0.3336, 0.2133, 0.0312, 0.3574, 0.2163, 0.2874, 0.3066], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0377, 0.0318, 0.0319, 0.0404, 0.0423, 0.0338, 0.0439], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:07:50,762 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.266e+02 2.904e+02 3.599e+02 6.333e+02, threshold=5.808e+02, percent-clipped=1.0 2023-04-29 04:08:20,258 INFO [train.py:904] (7/8) Epoch 10, batch 500, loss[loss=0.1908, simple_loss=0.2697, pruned_loss=0.056, over 15437.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2711, pruned_loss=0.05552, over 3041630.41 frames. ], batch size: 190, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:08:26,608 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-04-29 04:09:04,393 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1705, 5.1213, 4.8198, 4.3937, 4.8757, 1.8993, 4.6645, 4.8608], device='cuda:7'), covar=tensor([0.0064, 0.0058, 0.0149, 0.0293, 0.0076, 0.2300, 0.0095, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0108, 0.0158, 0.0145, 0.0126, 0.0176, 0.0143, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:09:13,574 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 04:09:15,179 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91891.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:09:23,903 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91898.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 04:09:26,854 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1696, 4.1159, 4.0996, 3.6076, 4.1006, 1.6789, 3.8847, 3.7455], device='cuda:7'), covar=tensor([0.0094, 0.0089, 0.0120, 0.0219, 0.0079, 0.2283, 0.0119, 0.0164], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0108, 0.0157, 0.0145, 0.0126, 0.0176, 0.0143, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:09:29,473 INFO [train.py:904] (7/8) Epoch 10, batch 550, loss[loss=0.1777, simple_loss=0.2704, pruned_loss=0.04251, over 17114.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2701, pruned_loss=0.05498, over 3104170.23 frames. ], batch size: 47, lr: 7.02e-03, grad_scale: 2.0 2023-04-29 04:10:07,829 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.517e+02 3.166e+02 3.975e+02 1.037e+03, threshold=6.333e+02, percent-clipped=5.0 2023-04-29 04:10:22,761 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91941.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:10:38,343 INFO [train.py:904] (7/8) Epoch 10, batch 600, loss[loss=0.1698, simple_loss=0.2401, pruned_loss=0.04973, over 16682.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2691, pruned_loss=0.05484, over 3152598.91 frames. ], batch size: 89, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:10:38,855 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91952.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:10:42,875 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91955.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:11:00,514 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8706, 3.9970, 3.0096, 2.3247, 2.6422, 2.3290, 4.0105, 3.5442], device='cuda:7'), covar=tensor([0.2114, 0.0500, 0.1331, 0.2212, 0.2208, 0.1740, 0.0405, 0.1111], device='cuda:7'), in_proj_covar=tensor([0.0297, 0.0253, 0.0278, 0.0271, 0.0267, 0.0216, 0.0263, 0.0287], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:11:19,514 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91981.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:11:30,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=91989.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:11:51,297 INFO [train.py:904] (7/8) Epoch 10, batch 650, loss[loss=0.1902, simple_loss=0.2604, pruned_loss=0.06006, over 12536.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.268, pruned_loss=0.05525, over 3185334.85 frames. ], batch size: 246, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:12:12,126 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92016.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:12:22,141 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 04:12:31,802 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.417e+02 3.017e+02 3.569e+02 8.138e+02, threshold=6.033e+02, percent-clipped=1.0 2023-04-29 04:12:49,553 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92042.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 04:12:55,657 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1963, 5.2188, 5.0144, 4.5886, 4.4771, 5.1259, 5.1039, 4.6782], device='cuda:7'), covar=tensor([0.0647, 0.0340, 0.0288, 0.0311, 0.1316, 0.0356, 0.0294, 0.0681], device='cuda:7'), in_proj_covar=tensor([0.0244, 0.0298, 0.0284, 0.0262, 0.0309, 0.0298, 0.0193, 0.0334], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:13:03,072 INFO [train.py:904] (7/8) Epoch 10, batch 700, loss[loss=0.1899, simple_loss=0.2733, pruned_loss=0.05322, over 16620.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2666, pruned_loss=0.05397, over 3215767.42 frames. ], batch size: 62, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:13:53,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9090, 4.6723, 4.9474, 5.2005, 5.3450, 4.6658, 5.3385, 5.3071], device='cuda:7'), covar=tensor([0.1524, 0.1151, 0.1705, 0.0662, 0.0509, 0.0736, 0.0522, 0.0514], device='cuda:7'), in_proj_covar=tensor([0.0522, 0.0637, 0.0777, 0.0643, 0.0490, 0.0492, 0.0514, 0.0577], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:14:09,243 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0030, 1.7361, 2.3794, 2.7920, 2.7445, 2.9594, 1.7793, 3.0015], device='cuda:7'), covar=tensor([0.0116, 0.0328, 0.0210, 0.0185, 0.0170, 0.0146, 0.0359, 0.0076], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0167, 0.0152, 0.0155, 0.0163, 0.0119, 0.0168, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 04:14:12,209 INFO [train.py:904] (7/8) Epoch 10, batch 750, loss[loss=0.2024, simple_loss=0.288, pruned_loss=0.05839, over 16707.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2667, pruned_loss=0.05365, over 3240071.52 frames. ], batch size: 57, lr: 7.01e-03, grad_scale: 2.0 2023-04-29 04:14:50,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5250, 3.5543, 2.0220, 3.7794, 2.5593, 3.7397, 1.9561, 2.7224], device='cuda:7'), covar=tensor([0.0190, 0.0317, 0.1492, 0.0176, 0.0783, 0.0493, 0.1401, 0.0645], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0164, 0.0188, 0.0124, 0.0168, 0.0204, 0.0194, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 04:14:52,017 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.724e+02 2.681e+02 3.065e+02 4.087e+02 6.685e+02, threshold=6.130e+02, percent-clipped=5.0 2023-04-29 04:15:22,696 INFO [train.py:904] (7/8) Epoch 10, batch 800, loss[loss=0.1879, simple_loss=0.2777, pruned_loss=0.0491, over 17080.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2665, pruned_loss=0.05357, over 3263151.87 frames. ], batch size: 53, lr: 7.01e-03, grad_scale: 4.0 2023-04-29 04:16:27,576 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92198.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:16:32,818 INFO [train.py:904] (7/8) Epoch 10, batch 850, loss[loss=0.1978, simple_loss=0.2726, pruned_loss=0.0615, over 12065.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2654, pruned_loss=0.05241, over 3277966.44 frames. ], batch size: 246, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:17:10,137 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.453e+02 2.911e+02 3.766e+02 9.676e+02, threshold=5.821e+02, percent-clipped=2.0 2023-04-29 04:17:33,409 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92246.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:17:34,389 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92247.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:17:41,172 INFO [train.py:904] (7/8) Epoch 10, batch 900, loss[loss=0.1723, simple_loss=0.2487, pruned_loss=0.04791, over 16524.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2646, pruned_loss=0.05166, over 3289246.19 frames. ], batch size: 146, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:18:50,844 INFO [train.py:904] (7/8) Epoch 10, batch 950, loss[loss=0.1741, simple_loss=0.267, pruned_loss=0.04062, over 17115.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2643, pruned_loss=0.05118, over 3293444.70 frames. ], batch size: 49, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:19:04,231 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92311.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:19:27,900 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4533, 3.4918, 1.9367, 3.6729, 2.6046, 3.6218, 1.9666, 2.7484], device='cuda:7'), covar=tensor([0.0206, 0.0359, 0.1464, 0.0232, 0.0739, 0.0769, 0.1341, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0164, 0.0188, 0.0124, 0.0166, 0.0203, 0.0192, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 04:19:29,771 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.335e+02 2.688e+02 3.318e+02 6.424e+02, threshold=5.375e+02, percent-clipped=3.0 2023-04-29 04:19:39,563 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92337.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 04:19:58,804 INFO [train.py:904] (7/8) Epoch 10, batch 1000, loss[loss=0.2006, simple_loss=0.2868, pruned_loss=0.05719, over 17136.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2634, pruned_loss=0.05145, over 3290992.82 frames. ], batch size: 48, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:20:47,764 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 04:20:59,983 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1481, 5.1075, 4.9223, 4.3618, 4.9730, 1.8894, 4.6838, 4.8683], device='cuda:7'), covar=tensor([0.0069, 0.0058, 0.0138, 0.0353, 0.0085, 0.2323, 0.0120, 0.0171], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0111, 0.0162, 0.0151, 0.0131, 0.0179, 0.0149, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:21:03,428 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5495, 3.5770, 2.0395, 3.7466, 2.6910, 3.7270, 1.9867, 2.7270], device='cuda:7'), covar=tensor([0.0201, 0.0392, 0.1500, 0.0238, 0.0704, 0.0631, 0.1514, 0.0654], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0163, 0.0186, 0.0123, 0.0164, 0.0202, 0.0191, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 04:21:09,130 INFO [train.py:904] (7/8) Epoch 10, batch 1050, loss[loss=0.1767, simple_loss=0.2542, pruned_loss=0.04964, over 16454.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2633, pruned_loss=0.05122, over 3300743.04 frames. ], batch size: 75, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:21:48,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.612e+02 3.041e+02 3.664e+02 7.677e+02, threshold=6.083e+02, percent-clipped=4.0 2023-04-29 04:22:18,523 INFO [train.py:904] (7/8) Epoch 10, batch 1100, loss[loss=0.1748, simple_loss=0.2658, pruned_loss=0.04191, over 17249.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2626, pruned_loss=0.05049, over 3306428.31 frames. ], batch size: 45, lr: 7.00e-03, grad_scale: 4.0 2023-04-29 04:23:28,219 INFO [train.py:904] (7/8) Epoch 10, batch 1150, loss[loss=0.1895, simple_loss=0.2766, pruned_loss=0.05124, over 17122.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2627, pruned_loss=0.05075, over 3307660.42 frames. ], batch size: 48, lr: 6.99e-03, grad_scale: 4.0 2023-04-29 04:23:59,452 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9542, 1.8457, 2.2705, 2.7979, 2.7205, 3.3118, 2.0014, 3.1218], device='cuda:7'), covar=tensor([0.0151, 0.0343, 0.0235, 0.0213, 0.0201, 0.0119, 0.0352, 0.0112], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0156, 0.0158, 0.0166, 0.0121, 0.0171, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 04:24:08,394 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.186e+02 2.590e+02 3.516e+02 6.121e+02, threshold=5.179e+02, percent-clipped=1.0 2023-04-29 04:24:32,784 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92547.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:24:38,826 INFO [train.py:904] (7/8) Epoch 10, batch 1200, loss[loss=0.2002, simple_loss=0.2838, pruned_loss=0.0583, over 16680.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2619, pruned_loss=0.05031, over 3313141.88 frames. ], batch size: 62, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:25:06,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1006, 5.1096, 5.6919, 5.6647, 5.6483, 5.2652, 5.2178, 4.9406], device='cuda:7'), covar=tensor([0.0256, 0.0398, 0.0304, 0.0342, 0.0350, 0.0277, 0.0745, 0.0373], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0339, 0.0338, 0.0326, 0.0380, 0.0356, 0.0463, 0.0286], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 04:25:11,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1702, 4.5274, 4.6937, 3.4301, 4.0899, 4.5019, 4.2098, 3.0139], device='cuda:7'), covar=tensor([0.0308, 0.0052, 0.0023, 0.0223, 0.0060, 0.0062, 0.0049, 0.0286], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0070, 0.0068, 0.0125, 0.0077, 0.0084, 0.0075, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:25:39,102 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92595.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:25:47,779 INFO [train.py:904] (7/8) Epoch 10, batch 1250, loss[loss=0.185, simple_loss=0.2555, pruned_loss=0.0573, over 12461.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2625, pruned_loss=0.051, over 3312561.79 frames. ], batch size: 246, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:26:01,791 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92611.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:26:21,257 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:26:27,275 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.554e+02 2.910e+02 3.549e+02 5.849e+02, threshold=5.821e+02, percent-clipped=4.0 2023-04-29 04:26:37,464 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92637.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 04:26:58,809 INFO [train.py:904] (7/8) Epoch 10, batch 1300, loss[loss=0.2041, simple_loss=0.2699, pruned_loss=0.06911, over 12066.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2621, pruned_loss=0.05114, over 3303724.76 frames. ], batch size: 246, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:27:07,466 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:27:44,065 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=92685.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 04:27:45,927 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92686.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:28:07,567 INFO [train.py:904] (7/8) Epoch 10, batch 1350, loss[loss=0.2089, simple_loss=0.2768, pruned_loss=0.07053, over 16906.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2631, pruned_loss=0.05089, over 3307681.81 frames. ], batch size: 109, lr: 6.99e-03, grad_scale: 8.0 2023-04-29 04:28:34,079 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92720.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:28:47,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.580e+02 3.033e+02 3.471e+02 5.594e+02, threshold=6.066e+02, percent-clipped=0.0 2023-04-29 04:28:50,758 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6745, 3.0355, 2.7556, 4.9354, 4.1275, 4.5685, 1.4950, 3.2007], device='cuda:7'), covar=tensor([0.1364, 0.0625, 0.1104, 0.0143, 0.0238, 0.0348, 0.1517, 0.0717], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0154, 0.0177, 0.0135, 0.0195, 0.0211, 0.0177, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:29:01,929 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-04-29 04:29:18,981 INFO [train.py:904] (7/8) Epoch 10, batch 1400, loss[loss=0.1864, simple_loss=0.2526, pruned_loss=0.06013, over 15525.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2627, pruned_loss=0.05073, over 3309316.23 frames. ], batch size: 191, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:29:43,210 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 04:30:00,614 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9182, 2.7059, 2.6710, 1.9241, 2.6457, 2.6889, 2.6702, 1.8712], device='cuda:7'), covar=tensor([0.0341, 0.0066, 0.0046, 0.0289, 0.0072, 0.0066, 0.0071, 0.0310], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0070, 0.0068, 0.0126, 0.0077, 0.0085, 0.0076, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:30:00,653 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:30:12,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4596, 2.9098, 2.6855, 2.2809, 2.2702, 2.2279, 2.9278, 2.8620], device='cuda:7'), covar=tensor([0.2221, 0.0800, 0.1364, 0.1910, 0.2166, 0.1763, 0.0523, 0.0973], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0256, 0.0280, 0.0271, 0.0274, 0.0218, 0.0264, 0.0292], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:30:28,855 INFO [train.py:904] (7/8) Epoch 10, batch 1450, loss[loss=0.1786, simple_loss=0.2536, pruned_loss=0.05179, over 12198.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.262, pruned_loss=0.05034, over 3297196.72 frames. ], batch size: 246, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:31:07,995 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.441e+02 3.023e+02 3.567e+02 6.383e+02, threshold=6.046e+02, percent-clipped=1.0 2023-04-29 04:31:09,671 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92831.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:31:38,229 INFO [train.py:904] (7/8) Epoch 10, batch 1500, loss[loss=0.1845, simple_loss=0.2698, pruned_loss=0.04956, over 17230.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2614, pruned_loss=0.04997, over 3307833.02 frames. ], batch size: 46, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:32:34,117 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92892.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:32:43,935 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9191, 4.0969, 3.1952, 2.4509, 2.8995, 2.4241, 4.0834, 3.7880], device='cuda:7'), covar=tensor([0.2107, 0.0510, 0.1278, 0.2215, 0.2080, 0.1589, 0.0454, 0.1124], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0256, 0.0280, 0.0272, 0.0274, 0.0218, 0.0265, 0.0293], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:32:45,537 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92899.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:32:48,543 INFO [train.py:904] (7/8) Epoch 10, batch 1550, loss[loss=0.1815, simple_loss=0.2741, pruned_loss=0.04445, over 17127.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.264, pruned_loss=0.05169, over 3298202.07 frames. ], batch size: 49, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:33:26,145 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.944e+02 3.490e+02 4.097e+02 8.318e+02, threshold=6.980e+02, percent-clipped=5.0 2023-04-29 04:33:28,363 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92931.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:33:54,806 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 04:33:56,442 INFO [train.py:904] (7/8) Epoch 10, batch 1600, loss[loss=0.1826, simple_loss=0.2721, pruned_loss=0.04654, over 17064.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2662, pruned_loss=0.05301, over 3309026.39 frames. ], batch size: 55, lr: 6.98e-03, grad_scale: 8.0 2023-04-29 04:33:59,060 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7741, 4.7284, 4.6240, 4.3255, 4.2487, 4.7152, 4.6125, 4.3675], device='cuda:7'), covar=tensor([0.0636, 0.0515, 0.0276, 0.0309, 0.0995, 0.0411, 0.0384, 0.0678], device='cuda:7'), in_proj_covar=tensor([0.0249, 0.0305, 0.0289, 0.0268, 0.0318, 0.0305, 0.0199, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:34:08,481 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92960.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:29,032 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92975.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:37,950 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92981.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:50,555 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 04:34:52,601 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92992.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:34:55,587 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0790, 2.3329, 2.5538, 4.8998, 2.2506, 2.9852, 2.4787, 2.6994], device='cuda:7'), covar=tensor([0.0749, 0.3337, 0.2011, 0.0260, 0.3657, 0.2037, 0.2742, 0.2981], device='cuda:7'), in_proj_covar=tensor([0.0360, 0.0384, 0.0326, 0.0324, 0.0409, 0.0437, 0.0346, 0.0453], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:35:06,236 INFO [train.py:904] (7/8) Epoch 10, batch 1650, loss[loss=0.2031, simple_loss=0.2708, pruned_loss=0.0677, over 16852.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2685, pruned_loss=0.0538, over 3315270.17 frames. ], batch size: 42, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:35:18,338 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93010.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:35:45,200 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.314e+02 2.824e+02 3.373e+02 5.658e+02, threshold=5.648e+02, percent-clipped=0.0 2023-04-29 04:35:53,121 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:35:53,147 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:36:09,640 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6678, 2.6908, 2.5500, 4.0803, 3.6240, 4.1540, 1.3356, 3.1513], device='cuda:7'), covar=tensor([0.1283, 0.0552, 0.0991, 0.0108, 0.0155, 0.0271, 0.1384, 0.0585], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0153, 0.0176, 0.0135, 0.0194, 0.0209, 0.0175, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:36:15,639 INFO [train.py:904] (7/8) Epoch 10, batch 1700, loss[loss=0.1805, simple_loss=0.2774, pruned_loss=0.04184, over 17012.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2707, pruned_loss=0.05436, over 3314096.36 frames. ], batch size: 50, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:36:42,621 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93071.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:36:48,269 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93076.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:37:17,825 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93097.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 04:37:23,068 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7827, 3.1820, 3.0471, 5.0171, 4.2790, 4.6185, 1.4841, 3.1871], device='cuda:7'), covar=tensor([0.1333, 0.0613, 0.0979, 0.0129, 0.0239, 0.0351, 0.1548, 0.0744], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0154, 0.0177, 0.0136, 0.0195, 0.0211, 0.0176, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:37:23,700 INFO [train.py:904] (7/8) Epoch 10, batch 1750, loss[loss=0.1771, simple_loss=0.2706, pruned_loss=0.0418, over 17110.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2706, pruned_loss=0.05373, over 3323426.67 frames. ], batch size: 49, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:37:27,640 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4515, 2.9520, 2.9150, 1.9320, 2.6466, 2.1789, 3.0031, 3.0957], device='cuda:7'), covar=tensor([0.0242, 0.0675, 0.0537, 0.1659, 0.0725, 0.0958, 0.0590, 0.0757], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0142, 0.0157, 0.0141, 0.0135, 0.0125, 0.0136, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 04:37:42,848 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:38:01,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.469e+02 2.889e+02 3.646e+02 7.131e+02, threshold=5.778e+02, percent-clipped=4.0 2023-04-29 04:38:19,673 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 04:38:32,573 INFO [train.py:904] (7/8) Epoch 10, batch 1800, loss[loss=0.1616, simple_loss=0.2482, pruned_loss=0.03745, over 16807.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2721, pruned_loss=0.0544, over 3318659.49 frames. ], batch size: 39, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:39:06,726 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:39:20,464 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93187.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:39:42,546 INFO [train.py:904] (7/8) Epoch 10, batch 1850, loss[loss=0.2007, simple_loss=0.2897, pruned_loss=0.05581, over 16666.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2735, pruned_loss=0.05441, over 3302369.84 frames. ], batch size: 57, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:39:54,385 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 04:40:04,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8877, 3.9609, 3.0747, 2.3547, 2.8443, 2.4881, 4.1929, 3.7677], device='cuda:7'), covar=tensor([0.2315, 0.0758, 0.1511, 0.2076, 0.2104, 0.1668, 0.0496, 0.0974], device='cuda:7'), in_proj_covar=tensor([0.0295, 0.0255, 0.0279, 0.0270, 0.0274, 0.0218, 0.0263, 0.0293], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:40:21,085 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.500e+02 2.909e+02 3.446e+02 8.007e+02, threshold=5.817e+02, percent-clipped=2.0 2023-04-29 04:40:31,126 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2654, 4.5839, 4.3853, 4.4081, 4.0884, 4.1416, 4.1810, 4.6590], device='cuda:7'), covar=tensor([0.1017, 0.0868, 0.0942, 0.0710, 0.0810, 0.1245, 0.0954, 0.0840], device='cuda:7'), in_proj_covar=tensor([0.0524, 0.0669, 0.0549, 0.0460, 0.0420, 0.0428, 0.0559, 0.0503], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:40:52,072 INFO [train.py:904] (7/8) Epoch 10, batch 1900, loss[loss=0.1954, simple_loss=0.2734, pruned_loss=0.0587, over 16499.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2724, pruned_loss=0.05395, over 3307061.93 frames. ], batch size: 68, lr: 6.97e-03, grad_scale: 8.0 2023-04-29 04:40:52,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8493, 5.1963, 5.3087, 5.1802, 5.1673, 5.7684, 5.3410, 5.0234], device='cuda:7'), covar=tensor([0.1033, 0.1679, 0.1783, 0.1712, 0.2405, 0.0915, 0.1305, 0.2166], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0496, 0.0520, 0.0428, 0.0570, 0.0549, 0.0421, 0.0572], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:40:57,482 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93255.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:41:15,159 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7133, 2.7061, 2.3065, 4.1532, 3.3339, 4.0370, 1.6839, 2.6979], device='cuda:7'), covar=tensor([0.1383, 0.0679, 0.1263, 0.0171, 0.0211, 0.0398, 0.1395, 0.0927], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0156, 0.0178, 0.0137, 0.0196, 0.0212, 0.0177, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:41:27,622 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0335, 4.9546, 5.5539, 5.5479, 5.4836, 5.1396, 5.1226, 4.7823], device='cuda:7'), covar=tensor([0.0298, 0.0353, 0.0272, 0.0333, 0.0443, 0.0298, 0.0804, 0.0405], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0338, 0.0342, 0.0326, 0.0386, 0.0359, 0.0467, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 04:41:33,469 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:41:41,660 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93287.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:41:51,283 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2906, 3.3522, 3.7577, 2.6761, 3.4381, 3.6913, 3.6069, 2.3629], device='cuda:7'), covar=tensor([0.0362, 0.0163, 0.0035, 0.0253, 0.0079, 0.0081, 0.0062, 0.0330], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0069, 0.0067, 0.0123, 0.0076, 0.0083, 0.0074, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:42:02,737 INFO [train.py:904] (7/8) Epoch 10, batch 1950, loss[loss=0.1844, simple_loss=0.277, pruned_loss=0.04587, over 17142.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2713, pruned_loss=0.05324, over 3308732.85 frames. ], batch size: 49, lr: 6.96e-03, grad_scale: 8.0 2023-04-29 04:42:40,591 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:42:41,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.458e+02 3.015e+02 3.561e+02 8.313e+02, threshold=6.031e+02, percent-clipped=4.0 2023-04-29 04:42:43,595 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93331.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:43:12,565 INFO [train.py:904] (7/8) Epoch 10, batch 2000, loss[loss=0.1588, simple_loss=0.2506, pruned_loss=0.03353, over 17002.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2711, pruned_loss=0.05375, over 3307405.04 frames. ], batch size: 41, lr: 6.96e-03, grad_scale: 8.0 2023-04-29 04:43:28,505 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-29 04:43:31,526 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93366.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:43:45,294 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93376.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:44:08,430 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93392.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 04:44:21,504 INFO [train.py:904] (7/8) Epoch 10, batch 2050, loss[loss=0.1984, simple_loss=0.2912, pruned_loss=0.05279, over 17111.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2716, pruned_loss=0.05402, over 3306162.18 frames. ], batch size: 49, lr: 6.96e-03, grad_scale: 4.0 2023-04-29 04:44:51,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93424.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:45:00,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.531e+02 2.855e+02 3.295e+02 5.942e+02, threshold=5.709e+02, percent-clipped=0.0 2023-04-29 04:45:12,458 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 04:45:29,956 INFO [train.py:904] (7/8) Epoch 10, batch 2100, loss[loss=0.2032, simple_loss=0.2829, pruned_loss=0.06174, over 16367.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2715, pruned_loss=0.05403, over 3311299.07 frames. ], batch size: 165, lr: 6.96e-03, grad_scale: 4.0 2023-04-29 04:45:47,697 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.71 vs. limit=5.0 2023-04-29 04:45:56,825 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:46:15,505 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2903, 3.6735, 3.4233, 2.1116, 3.0633, 2.4616, 3.6839, 3.7821], device='cuda:7'), covar=tensor([0.0253, 0.0641, 0.0546, 0.1650, 0.0709, 0.0853, 0.0551, 0.0729], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0143, 0.0156, 0.0142, 0.0135, 0.0125, 0.0136, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 04:46:18,716 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93487.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:46:40,126 INFO [train.py:904] (7/8) Epoch 10, batch 2150, loss[loss=0.1928, simple_loss=0.2711, pruned_loss=0.05727, over 16875.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2734, pruned_loss=0.05573, over 3307677.93 frames. ], batch size: 83, lr: 6.96e-03, grad_scale: 4.0 2023-04-29 04:47:18,312 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.473e+02 3.061e+02 3.473e+02 5.653e+02, threshold=6.122e+02, percent-clipped=0.0 2023-04-29 04:47:24,634 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93535.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:47:30,906 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1806, 5.8492, 5.9726, 5.7614, 5.7356, 6.2883, 5.8488, 5.4851], device='cuda:7'), covar=tensor([0.0918, 0.1598, 0.1323, 0.1723, 0.2364, 0.0812, 0.1262, 0.2125], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0498, 0.0522, 0.0431, 0.0577, 0.0549, 0.0422, 0.0579], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:47:46,708 INFO [train.py:904] (7/8) Epoch 10, batch 2200, loss[loss=0.2203, simple_loss=0.3017, pruned_loss=0.06944, over 16639.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2737, pruned_loss=0.05576, over 3303070.98 frames. ], batch size: 62, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:47:51,992 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:48:35,502 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93587.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:48:54,037 INFO [train.py:904] (7/8) Epoch 10, batch 2250, loss[loss=0.1947, simple_loss=0.2831, pruned_loss=0.05322, over 17031.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2737, pruned_loss=0.0555, over 3316656.47 frames. ], batch size: 55, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:48:56,180 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93603.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:49:20,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9721, 2.8863, 2.5557, 2.7342, 3.2324, 3.0153, 3.8511, 3.4602], device='cuda:7'), covar=tensor([0.0060, 0.0227, 0.0285, 0.0268, 0.0153, 0.0214, 0.0143, 0.0152], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0202, 0.0198, 0.0199, 0.0200, 0.0201, 0.0210, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:49:33,830 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.560e+02 3.200e+02 3.958e+02 7.230e+02, threshold=6.400e+02, percent-clipped=2.0 2023-04-29 04:49:34,882 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93631.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:49:39,264 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93635.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:50:04,001 INFO [train.py:904] (7/8) Epoch 10, batch 2300, loss[loss=0.2, simple_loss=0.2923, pruned_loss=0.05384, over 16732.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2738, pruned_loss=0.05511, over 3324849.63 frames. ], batch size: 57, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:50:10,681 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 04:50:22,499 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93666.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:50:32,627 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0097, 4.3453, 4.6178, 4.5773, 4.6001, 4.2774, 3.9816, 4.1623], device='cuda:7'), covar=tensor([0.0690, 0.0873, 0.0591, 0.0734, 0.0680, 0.0666, 0.1657, 0.0821], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0344, 0.0348, 0.0331, 0.0390, 0.0362, 0.0471, 0.0291], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 04:50:39,794 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93679.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:50:57,985 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:51:05,812 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2023-04-29 04:51:09,775 INFO [train.py:904] (7/8) Epoch 10, batch 2350, loss[loss=0.1744, simple_loss=0.2637, pruned_loss=0.04257, over 17243.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2738, pruned_loss=0.05582, over 3327270.87 frames. ], batch size: 44, lr: 6.95e-03, grad_scale: 4.0 2023-04-29 04:51:27,738 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93714.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:51:49,912 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.363e+02 2.805e+02 3.305e+02 9.718e+02, threshold=5.610e+02, percent-clipped=1.0 2023-04-29 04:51:56,166 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 04:52:03,193 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:52:03,820 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-29 04:52:17,497 INFO [train.py:904] (7/8) Epoch 10, batch 2400, loss[loss=0.2114, simple_loss=0.2898, pruned_loss=0.06653, over 16406.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2752, pruned_loss=0.05604, over 3329613.44 frames. ], batch size: 75, lr: 6.95e-03, grad_scale: 8.0 2023-04-29 04:52:27,525 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6239, 2.7324, 2.4908, 2.6573, 2.9422, 2.8188, 3.5543, 3.2085], device='cuda:7'), covar=tensor([0.0079, 0.0234, 0.0290, 0.0273, 0.0178, 0.0242, 0.0149, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0199, 0.0196, 0.0198, 0.0200, 0.0200, 0.0210, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:52:41,485 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93770.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:52:43,197 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93771.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:53:02,146 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7857, 4.0748, 2.3820, 4.6043, 2.9678, 4.5493, 2.3744, 3.3582], device='cuda:7'), covar=tensor([0.0231, 0.0322, 0.1423, 0.0149, 0.0859, 0.0362, 0.1470, 0.0598], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0169, 0.0188, 0.0129, 0.0169, 0.0210, 0.0196, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 04:53:15,411 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8362, 5.1922, 4.9186, 4.9695, 4.6412, 4.5584, 4.7153, 5.2762], device='cuda:7'), covar=tensor([0.1094, 0.0776, 0.0981, 0.0648, 0.0730, 0.0957, 0.0946, 0.0812], device='cuda:7'), in_proj_covar=tensor([0.0531, 0.0671, 0.0557, 0.0461, 0.0423, 0.0427, 0.0559, 0.0508], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:53:26,813 INFO [train.py:904] (7/8) Epoch 10, batch 2450, loss[loss=0.1859, simple_loss=0.2617, pruned_loss=0.05501, over 16885.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2749, pruned_loss=0.05536, over 3333045.12 frames. ], batch size: 96, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:53:50,748 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=93819.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:54:05,073 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.505e+02 3.029e+02 3.825e+02 7.236e+02, threshold=6.058e+02, percent-clipped=4.0 2023-04-29 04:54:05,565 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93831.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:54:12,339 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2277, 4.0871, 4.2290, 4.4410, 4.5235, 4.0999, 4.2612, 4.5046], device='cuda:7'), covar=tensor([0.1294, 0.0889, 0.1330, 0.0556, 0.0540, 0.1120, 0.1646, 0.0546], device='cuda:7'), in_proj_covar=tensor([0.0538, 0.0656, 0.0809, 0.0672, 0.0502, 0.0516, 0.0531, 0.0598], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:54:34,536 INFO [train.py:904] (7/8) Epoch 10, batch 2500, loss[loss=0.1787, simple_loss=0.2514, pruned_loss=0.05297, over 16786.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2749, pruned_loss=0.05483, over 3326575.01 frames. ], batch size: 83, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:55:06,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1354, 3.3261, 3.5632, 2.4036, 3.2302, 3.6303, 3.3933, 2.0799], device='cuda:7'), covar=tensor([0.0372, 0.0088, 0.0035, 0.0274, 0.0071, 0.0054, 0.0050, 0.0323], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0069, 0.0069, 0.0124, 0.0077, 0.0084, 0.0075, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 04:55:15,120 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0105, 5.3902, 5.1012, 5.1789, 4.8296, 4.6888, 4.9084, 5.4908], device='cuda:7'), covar=tensor([0.1107, 0.0865, 0.0954, 0.0637, 0.0748, 0.0889, 0.0962, 0.0854], device='cuda:7'), in_proj_covar=tensor([0.0528, 0.0667, 0.0556, 0.0458, 0.0421, 0.0426, 0.0555, 0.0507], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:55:19,256 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4837, 3.2431, 2.5960, 2.0849, 2.2611, 2.1017, 3.3510, 3.0907], device='cuda:7'), covar=tensor([0.2380, 0.0782, 0.1512, 0.2278, 0.2301, 0.1815, 0.0564, 0.1251], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0259, 0.0282, 0.0274, 0.0281, 0.0220, 0.0266, 0.0297], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:55:43,668 INFO [train.py:904] (7/8) Epoch 10, batch 2550, loss[loss=0.1704, simple_loss=0.2514, pruned_loss=0.0447, over 16789.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.275, pruned_loss=0.05461, over 3326136.98 frames. ], batch size: 83, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:56:23,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.408e+02 2.927e+02 3.582e+02 7.180e+02, threshold=5.854e+02, percent-clipped=2.0 2023-04-29 04:56:38,020 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9089, 2.2824, 2.4650, 4.7375, 2.2628, 2.8560, 2.4727, 2.5748], device='cuda:7'), covar=tensor([0.0868, 0.3310, 0.2093, 0.0319, 0.3656, 0.2178, 0.2799, 0.3080], device='cuda:7'), in_proj_covar=tensor([0.0360, 0.0384, 0.0323, 0.0324, 0.0406, 0.0438, 0.0344, 0.0454], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 04:56:52,859 INFO [train.py:904] (7/8) Epoch 10, batch 2600, loss[loss=0.2213, simple_loss=0.2912, pruned_loss=0.07565, over 16820.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2751, pruned_loss=0.05428, over 3324605.27 frames. ], batch size: 124, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:57:32,320 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93981.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:57:49,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6493, 2.7168, 2.4227, 4.3549, 3.3759, 4.1074, 1.4415, 2.8298], device='cuda:7'), covar=tensor([0.1480, 0.0773, 0.1278, 0.0188, 0.0350, 0.0474, 0.1588, 0.0912], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0156, 0.0179, 0.0139, 0.0199, 0.0214, 0.0177, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 04:58:03,851 INFO [train.py:904] (7/8) Epoch 10, batch 2650, loss[loss=0.179, simple_loss=0.2728, pruned_loss=0.04264, over 17128.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2756, pruned_loss=0.05386, over 3315982.86 frames. ], batch size: 49, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:58:43,805 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.229e+02 2.744e+02 3.271e+02 8.724e+02, threshold=5.488e+02, percent-clipped=1.0 2023-04-29 04:59:00,274 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94042.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 04:59:11,206 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 04:59:13,558 INFO [train.py:904] (7/8) Epoch 10, batch 2700, loss[loss=0.1819, simple_loss=0.268, pruned_loss=0.0479, over 17242.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2757, pruned_loss=0.05342, over 3321388.99 frames. ], batch size: 45, lr: 6.94e-03, grad_scale: 8.0 2023-04-29 04:59:57,903 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 04:59:58,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9853, 4.9602, 5.4698, 5.4466, 5.4524, 5.0431, 5.0448, 4.7661], device='cuda:7'), covar=tensor([0.0298, 0.0495, 0.0326, 0.0410, 0.0418, 0.0307, 0.0878, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0337, 0.0346, 0.0350, 0.0332, 0.0390, 0.0363, 0.0474, 0.0292], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 05:00:13,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7988, 2.9807, 2.6052, 4.8784, 4.0791, 4.5426, 1.5978, 3.3771], device='cuda:7'), covar=tensor([0.1187, 0.0623, 0.1140, 0.0169, 0.0259, 0.0361, 0.1316, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0156, 0.0179, 0.0139, 0.0199, 0.0214, 0.0176, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 05:00:19,705 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 05:00:23,320 INFO [train.py:904] (7/8) Epoch 10, batch 2750, loss[loss=0.2101, simple_loss=0.2831, pruned_loss=0.06848, over 16663.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2753, pruned_loss=0.05274, over 3322411.16 frames. ], batch size: 89, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:00:50,356 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4505, 5.8403, 5.5350, 5.6160, 5.1883, 5.0581, 5.3089, 5.9602], device='cuda:7'), covar=tensor([0.1122, 0.0836, 0.1057, 0.0708, 0.0818, 0.0697, 0.0986, 0.0922], device='cuda:7'), in_proj_covar=tensor([0.0536, 0.0683, 0.0567, 0.0466, 0.0427, 0.0436, 0.0565, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:00:55,376 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94126.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:01:01,089 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.410e+02 2.958e+02 3.419e+02 5.641e+02, threshold=5.917e+02, percent-clipped=1.0 2023-04-29 05:01:29,689 INFO [train.py:904] (7/8) Epoch 10, batch 2800, loss[loss=0.1894, simple_loss=0.279, pruned_loss=0.04986, over 16772.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2747, pruned_loss=0.0525, over 3329354.75 frames. ], batch size: 57, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:02:24,398 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.46 vs. limit=5.0 2023-04-29 05:02:26,318 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7890, 4.9532, 5.0810, 4.9833, 4.8719, 5.6042, 5.1208, 4.8359], device='cuda:7'), covar=tensor([0.1288, 0.1695, 0.1685, 0.1777, 0.2644, 0.0932, 0.1448, 0.2438], device='cuda:7'), in_proj_covar=tensor([0.0346, 0.0492, 0.0517, 0.0428, 0.0567, 0.0546, 0.0418, 0.0573], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:02:39,412 INFO [train.py:904] (7/8) Epoch 10, batch 2850, loss[loss=0.1924, simple_loss=0.2832, pruned_loss=0.05076, over 17271.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2747, pruned_loss=0.05273, over 3333831.68 frames. ], batch size: 52, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:03:09,857 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94223.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:03:20,117 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.429e+02 2.851e+02 3.333e+02 6.061e+02, threshold=5.703e+02, percent-clipped=1.0 2023-04-29 05:03:21,889 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0642, 3.6895, 3.0007, 5.1636, 4.3844, 4.7200, 1.8130, 3.3913], device='cuda:7'), covar=tensor([0.1142, 0.0469, 0.0961, 0.0127, 0.0224, 0.0335, 0.1326, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0156, 0.0178, 0.0138, 0.0199, 0.0212, 0.0176, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 05:03:49,049 INFO [train.py:904] (7/8) Epoch 10, batch 2900, loss[loss=0.2825, simple_loss=0.3356, pruned_loss=0.1147, over 11856.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2734, pruned_loss=0.05297, over 3331255.60 frames. ], batch size: 246, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:03:56,234 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-29 05:04:33,882 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94284.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:04:36,172 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3513, 5.3206, 5.1965, 4.7780, 4.6679, 5.2869, 5.2562, 4.8689], device='cuda:7'), covar=tensor([0.0584, 0.0317, 0.0262, 0.0291, 0.1180, 0.0381, 0.0216, 0.0683], device='cuda:7'), in_proj_covar=tensor([0.0255, 0.0315, 0.0297, 0.0276, 0.0327, 0.0312, 0.0205, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:04:58,069 INFO [train.py:904] (7/8) Epoch 10, batch 2950, loss[loss=0.2009, simple_loss=0.2719, pruned_loss=0.06499, over 16875.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2725, pruned_loss=0.05294, over 3330288.20 frames. ], batch size: 96, lr: 6.93e-03, grad_scale: 8.0 2023-04-29 05:05:39,550 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.713e+02 3.214e+02 3.938e+02 7.856e+02, threshold=6.427e+02, percent-clipped=3.0 2023-04-29 05:05:49,257 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94337.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:06:03,150 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94348.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:06:08,015 INFO [train.py:904] (7/8) Epoch 10, batch 3000, loss[loss=0.1802, simple_loss=0.2648, pruned_loss=0.04787, over 17039.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2728, pruned_loss=0.05355, over 3336204.18 frames. ], batch size: 53, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:06:08,015 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 05:06:17,142 INFO [train.py:938] (7/8) Epoch 10, validation: loss=0.1426, simple_loss=0.2488, pruned_loss=0.01818, over 944034.00 frames. 2023-04-29 05:06:17,143 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 05:06:21,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9753, 4.0845, 2.5736, 4.6830, 3.0184, 4.5983, 2.6483, 3.3593], device='cuda:7'), covar=tensor([0.0186, 0.0287, 0.1265, 0.0169, 0.0721, 0.0412, 0.1143, 0.0556], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0167, 0.0187, 0.0131, 0.0167, 0.0211, 0.0195, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 05:07:26,698 INFO [train.py:904] (7/8) Epoch 10, batch 3050, loss[loss=0.2156, simple_loss=0.2791, pruned_loss=0.07611, over 16847.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2727, pruned_loss=0.05408, over 3335585.63 frames. ], batch size: 116, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:07:36,849 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94409.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:07:57,753 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94426.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:08:05,350 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.835e+02 2.667e+02 3.252e+02 4.083e+02 7.974e+02, threshold=6.505e+02, percent-clipped=3.0 2023-04-29 05:08:12,090 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.4825, 2.9707, 2.6129, 4.9000, 3.9314, 4.4500, 1.6910, 3.1349], device='cuda:7'), covar=tensor([0.1595, 0.0732, 0.1271, 0.0158, 0.0331, 0.0381, 0.1606, 0.0787], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0157, 0.0179, 0.0139, 0.0200, 0.0214, 0.0176, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 05:08:33,250 INFO [train.py:904] (7/8) Epoch 10, batch 3100, loss[loss=0.1654, simple_loss=0.2423, pruned_loss=0.04425, over 16823.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2723, pruned_loss=0.05405, over 3336391.02 frames. ], batch size: 83, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:08:56,914 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3789, 3.3455, 3.3535, 3.5243, 3.5585, 3.2620, 3.4864, 3.5984], device='cuda:7'), covar=tensor([0.1044, 0.0809, 0.1152, 0.0567, 0.0629, 0.2857, 0.1038, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0547, 0.0672, 0.0832, 0.0683, 0.0511, 0.0530, 0.0543, 0.0602], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:09:04,482 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=94474.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:09:43,506 INFO [train.py:904] (7/8) Epoch 10, batch 3150, loss[loss=0.1984, simple_loss=0.2729, pruned_loss=0.06192, over 16839.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2717, pruned_loss=0.05394, over 3330189.28 frames. ], batch size: 90, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:09:47,736 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 05:10:18,027 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-29 05:10:23,686 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.494e+02 3.047e+02 3.554e+02 8.509e+02, threshold=6.093e+02, percent-clipped=1.0 2023-04-29 05:10:38,390 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4945, 3.5921, 2.0788, 3.8205, 2.5746, 3.7214, 2.0218, 2.7974], device='cuda:7'), covar=tensor([0.0181, 0.0372, 0.1342, 0.0175, 0.0755, 0.0591, 0.1350, 0.0603], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0167, 0.0187, 0.0131, 0.0168, 0.0210, 0.0195, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 05:10:45,610 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8572, 4.1677, 3.9062, 4.0146, 3.6743, 3.7681, 3.7816, 4.1586], device='cuda:7'), covar=tensor([0.1038, 0.0972, 0.1153, 0.0774, 0.0880, 0.1657, 0.1087, 0.1042], device='cuda:7'), in_proj_covar=tensor([0.0540, 0.0687, 0.0568, 0.0470, 0.0430, 0.0435, 0.0570, 0.0521], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:10:52,180 INFO [train.py:904] (7/8) Epoch 10, batch 3200, loss[loss=0.1874, simple_loss=0.2625, pruned_loss=0.0562, over 16943.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2705, pruned_loss=0.05331, over 3327530.45 frames. ], batch size: 116, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:11:32,218 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94579.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:12:04,565 INFO [train.py:904] (7/8) Epoch 10, batch 3250, loss[loss=0.1656, simple_loss=0.2463, pruned_loss=0.0424, over 16215.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2702, pruned_loss=0.05308, over 3319175.63 frames. ], batch size: 36, lr: 6.92e-03, grad_scale: 8.0 2023-04-29 05:12:44,904 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.346e+02 2.940e+02 3.507e+02 9.203e+02, threshold=5.881e+02, percent-clipped=1.0 2023-04-29 05:12:51,253 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 05:12:53,020 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94637.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:13:14,618 INFO [train.py:904] (7/8) Epoch 10, batch 3300, loss[loss=0.1709, simple_loss=0.255, pruned_loss=0.04343, over 16781.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2707, pruned_loss=0.05322, over 3320499.88 frames. ], batch size: 102, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:14:02,322 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=94685.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:14:24,568 INFO [train.py:904] (7/8) Epoch 10, batch 3350, loss[loss=0.1799, simple_loss=0.257, pruned_loss=0.05144, over 16761.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2724, pruned_loss=0.05349, over 3323107.08 frames. ], batch size: 102, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:14:28,567 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94704.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:15:05,006 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.488e+02 2.932e+02 3.912e+02 8.438e+02, threshold=5.863e+02, percent-clipped=4.0 2023-04-29 05:15:10,681 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 05:15:35,793 INFO [train.py:904] (7/8) Epoch 10, batch 3400, loss[loss=0.1877, simple_loss=0.2659, pruned_loss=0.05475, over 16866.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2723, pruned_loss=0.05385, over 3323400.79 frames. ], batch size: 116, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:16:10,962 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3683, 5.3494, 5.1745, 4.6483, 5.2139, 2.0519, 4.9890, 5.1923], device='cuda:7'), covar=tensor([0.0067, 0.0061, 0.0136, 0.0344, 0.0069, 0.2079, 0.0106, 0.0142], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0120, 0.0169, 0.0162, 0.0139, 0.0180, 0.0158, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:16:26,722 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7651, 4.7666, 4.6636, 4.1631, 4.7302, 1.9193, 4.4707, 4.5244], device='cuda:7'), covar=tensor([0.0114, 0.0075, 0.0134, 0.0317, 0.0074, 0.2257, 0.0135, 0.0163], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0120, 0.0169, 0.0162, 0.0138, 0.0179, 0.0158, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:16:43,125 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9370, 1.8025, 2.2926, 2.7356, 2.7702, 2.5875, 1.7875, 2.9029], device='cuda:7'), covar=tensor([0.0116, 0.0315, 0.0253, 0.0181, 0.0177, 0.0200, 0.0351, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0172, 0.0158, 0.0163, 0.0167, 0.0124, 0.0171, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 05:16:44,921 INFO [train.py:904] (7/8) Epoch 10, batch 3450, loss[loss=0.218, simple_loss=0.2917, pruned_loss=0.07216, over 16299.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2702, pruned_loss=0.05324, over 3329947.98 frames. ], batch size: 165, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:17:15,347 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2817, 5.6880, 5.3868, 5.4622, 5.0687, 4.9638, 5.2159, 5.7593], device='cuda:7'), covar=tensor([0.1038, 0.0886, 0.1082, 0.0632, 0.0844, 0.0707, 0.0952, 0.0950], device='cuda:7'), in_proj_covar=tensor([0.0539, 0.0689, 0.0563, 0.0469, 0.0432, 0.0437, 0.0568, 0.0523], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:17:26,309 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.690e+02 2.564e+02 2.957e+02 3.532e+02 6.981e+02, threshold=5.915e+02, percent-clipped=3.0 2023-04-29 05:17:47,256 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0305, 3.9293, 4.4873, 2.0644, 4.7412, 4.6730, 3.2803, 3.6389], device='cuda:7'), covar=tensor([0.0630, 0.0191, 0.0127, 0.1072, 0.0036, 0.0095, 0.0342, 0.0323], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0100, 0.0089, 0.0138, 0.0070, 0.0103, 0.0122, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 05:17:56,543 INFO [train.py:904] (7/8) Epoch 10, batch 3500, loss[loss=0.195, simple_loss=0.2684, pruned_loss=0.06074, over 16481.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2692, pruned_loss=0.05308, over 3327902.74 frames. ], batch size: 146, lr: 6.91e-03, grad_scale: 8.0 2023-04-29 05:17:58,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9465, 4.0429, 4.3524, 4.3396, 4.3439, 4.0791, 4.1414, 3.9920], device='cuda:7'), covar=tensor([0.0376, 0.0484, 0.0404, 0.0418, 0.0467, 0.0364, 0.0693, 0.0574], device='cuda:7'), in_proj_covar=tensor([0.0344, 0.0350, 0.0356, 0.0337, 0.0399, 0.0371, 0.0478, 0.0297], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 05:18:12,675 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1915, 4.1366, 4.2921, 4.1393, 4.1625, 4.7335, 4.3808, 4.0503], device='cuda:7'), covar=tensor([0.1863, 0.2152, 0.1666, 0.2389, 0.3411, 0.1265, 0.1407, 0.2750], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0497, 0.0524, 0.0432, 0.0571, 0.0550, 0.0423, 0.0583], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:18:35,816 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94879.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:19:06,956 INFO [train.py:904] (7/8) Epoch 10, batch 3550, loss[loss=0.1578, simple_loss=0.2419, pruned_loss=0.0368, over 17216.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2675, pruned_loss=0.05188, over 3329269.80 frames. ], batch size: 44, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:19:07,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9964, 4.9585, 4.8653, 4.5481, 4.4117, 4.9647, 4.8524, 4.5025], device='cuda:7'), covar=tensor([0.0622, 0.0472, 0.0280, 0.0287, 0.1056, 0.0354, 0.0318, 0.0628], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0321, 0.0303, 0.0282, 0.0332, 0.0320, 0.0210, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:19:42,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=94927.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:19:47,559 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.230e+02 2.675e+02 3.251e+02 5.912e+02, threshold=5.350e+02, percent-clipped=0.0 2023-04-29 05:20:10,579 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 05:20:17,541 INFO [train.py:904] (7/8) Epoch 10, batch 3600, loss[loss=0.2041, simple_loss=0.2672, pruned_loss=0.07053, over 16890.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2668, pruned_loss=0.05154, over 3333445.63 frames. ], batch size: 116, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:21:02,154 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2023-04-29 05:21:28,703 INFO [train.py:904] (7/8) Epoch 10, batch 3650, loss[loss=0.1799, simple_loss=0.2546, pruned_loss=0.05256, over 16283.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2659, pruned_loss=0.05231, over 3317613.54 frames. ], batch size: 165, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:21:33,035 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95004.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:22:10,218 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.260e+02 2.711e+02 3.366e+02 9.321e+02, threshold=5.423e+02, percent-clipped=5.0 2023-04-29 05:22:42,268 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3386, 4.2218, 4.2083, 4.0336, 3.9761, 4.2770, 4.0096, 4.0296], device='cuda:7'), covar=tensor([0.0503, 0.0528, 0.0231, 0.0242, 0.0686, 0.0380, 0.0585, 0.0510], device='cuda:7'), in_proj_covar=tensor([0.0252, 0.0314, 0.0298, 0.0277, 0.0325, 0.0313, 0.0205, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:22:43,025 INFO [train.py:904] (7/8) Epoch 10, batch 3700, loss[loss=0.2007, simple_loss=0.2676, pruned_loss=0.06688, over 16433.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2653, pruned_loss=0.05421, over 3302001.88 frames. ], batch size: 146, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:22:43,414 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95052.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:23:56,097 INFO [train.py:904] (7/8) Epoch 10, batch 3750, loss[loss=0.236, simple_loss=0.3179, pruned_loss=0.07703, over 17219.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.266, pruned_loss=0.05583, over 3295918.08 frames. ], batch size: 52, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:24:00,778 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95105.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:24:37,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4487, 3.0130, 2.6616, 2.1718, 2.2249, 2.1237, 2.9985, 2.8739], device='cuda:7'), covar=tensor([0.2245, 0.0653, 0.1313, 0.2032, 0.2124, 0.1748, 0.0481, 0.0912], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0260, 0.0282, 0.0277, 0.0287, 0.0221, 0.0270, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:24:38,276 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.465e+02 2.786e+02 3.363e+02 5.307e+02, threshold=5.572e+02, percent-clipped=0.0 2023-04-29 05:25:07,908 INFO [train.py:904] (7/8) Epoch 10, batch 3800, loss[loss=0.181, simple_loss=0.2611, pruned_loss=0.05038, over 16187.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.2681, pruned_loss=0.05747, over 3301673.25 frames. ], batch size: 165, lr: 6.90e-03, grad_scale: 8.0 2023-04-29 05:25:28,687 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95166.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:25:52,518 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8255, 5.0782, 5.3855, 5.3412, 5.3477, 5.0900, 4.8254, 4.6567], device='cuda:7'), covar=tensor([0.0382, 0.0380, 0.0284, 0.0376, 0.0435, 0.0325, 0.1048, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0341, 0.0344, 0.0326, 0.0385, 0.0361, 0.0466, 0.0285], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 05:26:20,913 INFO [train.py:904] (7/8) Epoch 10, batch 3850, loss[loss=0.1788, simple_loss=0.2506, pruned_loss=0.05346, over 16339.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2679, pruned_loss=0.05771, over 3298473.92 frames. ], batch size: 165, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:27:00,986 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.608e+02 2.452e+02 2.920e+02 3.414e+02 5.310e+02, threshold=5.839e+02, percent-clipped=0.0 2023-04-29 05:27:31,953 INFO [train.py:904] (7/8) Epoch 10, batch 3900, loss[loss=0.1801, simple_loss=0.2571, pruned_loss=0.05153, over 17042.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.267, pruned_loss=0.05798, over 3291464.96 frames. ], batch size: 55, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:27:43,919 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95260.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:27:56,176 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:28:07,355 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3311, 3.6629, 3.8636, 2.5359, 3.5466, 3.9055, 3.7395, 2.3122], device='cuda:7'), covar=tensor([0.0370, 0.0066, 0.0027, 0.0280, 0.0051, 0.0054, 0.0043, 0.0302], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0067, 0.0067, 0.0121, 0.0075, 0.0083, 0.0073, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:28:45,303 INFO [train.py:904] (7/8) Epoch 10, batch 3950, loss[loss=0.1858, simple_loss=0.2517, pruned_loss=0.05999, over 16856.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2655, pruned_loss=0.05793, over 3293990.43 frames. ], batch size: 116, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:28:51,203 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 05:29:13,237 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:29:23,117 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:29:25,660 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.393e+02 2.876e+02 3.494e+02 7.568e+02, threshold=5.751e+02, percent-clipped=4.0 2023-04-29 05:29:38,098 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95339.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:29:56,316 INFO [train.py:904] (7/8) Epoch 10, batch 4000, loss[loss=0.1867, simple_loss=0.2647, pruned_loss=0.05431, over 16598.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2654, pruned_loss=0.05833, over 3294943.84 frames. ], batch size: 62, lr: 6.89e-03, grad_scale: 8.0 2023-04-29 05:30:29,062 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3274, 3.3798, 1.9930, 3.5181, 2.5205, 3.5982, 1.9739, 2.7464], device='cuda:7'), covar=tensor([0.0187, 0.0333, 0.1322, 0.0170, 0.0679, 0.0410, 0.1252, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0164, 0.0186, 0.0127, 0.0166, 0.0207, 0.0190, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 05:30:29,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8470, 3.2006, 3.0526, 1.9993, 2.7257, 2.2514, 3.2922, 3.3318], device='cuda:7'), covar=tensor([0.0231, 0.0660, 0.0572, 0.1587, 0.0771, 0.0920, 0.0510, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0145, 0.0156, 0.0142, 0.0135, 0.0124, 0.0136, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 05:31:05,668 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95400.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:31:07,592 INFO [train.py:904] (7/8) Epoch 10, batch 4050, loss[loss=0.1911, simple_loss=0.2708, pruned_loss=0.05574, over 16492.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2659, pruned_loss=0.05731, over 3294265.21 frames. ], batch size: 68, lr: 6.89e-03, grad_scale: 16.0 2023-04-29 05:31:49,147 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 1.964e+02 2.335e+02 2.702e+02 4.319e+02, threshold=4.671e+02, percent-clipped=0.0 2023-04-29 05:32:20,038 INFO [train.py:904] (7/8) Epoch 10, batch 4100, loss[loss=0.2274, simple_loss=0.3021, pruned_loss=0.0763, over 15314.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2678, pruned_loss=0.05689, over 3292285.15 frames. ], batch size: 190, lr: 6.88e-03, grad_scale: 16.0 2023-04-29 05:32:34,869 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95461.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:32:37,050 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1391, 2.3796, 1.7636, 2.0226, 2.6917, 2.4117, 2.9975, 3.0334], device='cuda:7'), covar=tensor([0.0071, 0.0238, 0.0362, 0.0325, 0.0158, 0.0246, 0.0126, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0198, 0.0196, 0.0195, 0.0200, 0.0200, 0.0208, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:33:33,944 INFO [train.py:904] (7/8) Epoch 10, batch 4150, loss[loss=0.2133, simple_loss=0.3032, pruned_loss=0.06167, over 16732.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2759, pruned_loss=0.06015, over 3233751.67 frames. ], batch size: 89, lr: 6.88e-03, grad_scale: 16.0 2023-04-29 05:34:17,118 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.709e+02 3.150e+02 3.936e+02 7.135e+02, threshold=6.300e+02, percent-clipped=10.0 2023-04-29 05:34:32,374 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9993, 1.8911, 2.0348, 3.5282, 1.8324, 2.2897, 2.0571, 2.0725], device='cuda:7'), covar=tensor([0.1103, 0.3450, 0.2064, 0.0495, 0.3860, 0.2247, 0.3017, 0.3157], device='cuda:7'), in_proj_covar=tensor([0.0362, 0.0391, 0.0324, 0.0324, 0.0406, 0.0447, 0.0350, 0.0461], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:34:49,635 INFO [train.py:904] (7/8) Epoch 10, batch 4200, loss[loss=0.21, simple_loss=0.31, pruned_loss=0.05501, over 16449.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2837, pruned_loss=0.06178, over 3223972.39 frames. ], batch size: 68, lr: 6.88e-03, grad_scale: 16.0 2023-04-29 05:35:30,548 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6570, 3.7711, 4.0687, 4.0356, 4.0730, 3.7799, 3.7569, 3.8094], device='cuda:7'), covar=tensor([0.0331, 0.0453, 0.0382, 0.0388, 0.0360, 0.0378, 0.0916, 0.0461], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0329, 0.0332, 0.0316, 0.0376, 0.0353, 0.0454, 0.0280], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 05:36:04,065 INFO [train.py:904] (7/8) Epoch 10, batch 4250, loss[loss=0.184, simple_loss=0.2853, pruned_loss=0.04137, over 16762.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2872, pruned_loss=0.06214, over 3194118.74 frames. ], batch size: 124, lr: 6.88e-03, grad_scale: 8.0 2023-04-29 05:36:24,787 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95616.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:36:26,057 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95617.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:36:38,766 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95624.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:36:49,134 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.325e+02 2.809e+02 3.314e+02 5.860e+02, threshold=5.619e+02, percent-clipped=0.0 2023-04-29 05:37:19,471 INFO [train.py:904] (7/8) Epoch 10, batch 4300, loss[loss=0.1956, simple_loss=0.2907, pruned_loss=0.05027, over 17208.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2883, pruned_loss=0.06134, over 3176121.11 frames. ], batch size: 44, lr: 6.88e-03, grad_scale: 8.0 2023-04-29 05:37:36,437 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5257, 4.4967, 4.9154, 4.8281, 4.8705, 4.5214, 4.5104, 4.3498], device='cuda:7'), covar=tensor([0.0259, 0.0379, 0.0277, 0.0385, 0.0438, 0.0311, 0.0799, 0.0399], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0330, 0.0332, 0.0317, 0.0377, 0.0352, 0.0456, 0.0280], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 05:37:59,687 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95678.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:38:24,136 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95695.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:38:33,669 INFO [train.py:904] (7/8) Epoch 10, batch 4350, loss[loss=0.1926, simple_loss=0.2871, pruned_loss=0.04903, over 16732.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2913, pruned_loss=0.06194, over 3188040.88 frames. ], batch size: 83, lr: 6.88e-03, grad_scale: 8.0 2023-04-29 05:39:18,686 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.523e+02 3.025e+02 3.679e+02 8.417e+02, threshold=6.050e+02, percent-clipped=3.0 2023-04-29 05:39:49,471 INFO [train.py:904] (7/8) Epoch 10, batch 4400, loss[loss=0.2106, simple_loss=0.2954, pruned_loss=0.06294, over 16501.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2932, pruned_loss=0.06327, over 3201708.83 frames. ], batch size: 146, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:40:02,631 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95761.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:41:01,598 INFO [train.py:904] (7/8) Epoch 10, batch 4450, loss[loss=0.2248, simple_loss=0.3085, pruned_loss=0.07051, over 15387.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.296, pruned_loss=0.06393, over 3210440.92 frames. ], batch size: 190, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:41:13,573 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95809.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:41:17,847 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95812.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:41:24,422 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0075, 4.8487, 4.7746, 3.2651, 4.1771, 4.6278, 4.2164, 2.7951], device='cuda:7'), covar=tensor([0.0310, 0.0011, 0.0020, 0.0236, 0.0049, 0.0049, 0.0039, 0.0277], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0067, 0.0067, 0.0122, 0.0076, 0.0083, 0.0073, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 05:41:46,154 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.254e+02 2.672e+02 3.297e+02 5.015e+02, threshold=5.344e+02, percent-clipped=0.0 2023-04-29 05:42:08,844 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95847.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:42:15,009 INFO [train.py:904] (7/8) Epoch 10, batch 4500, loss[loss=0.2075, simple_loss=0.2915, pruned_loss=0.06178, over 16829.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2959, pruned_loss=0.06382, over 3223679.32 frames. ], batch size: 83, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:42:37,438 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8541, 1.3077, 1.6917, 1.6910, 1.8684, 1.9380, 1.5007, 1.8114], device='cuda:7'), covar=tensor([0.0144, 0.0253, 0.0129, 0.0187, 0.0148, 0.0103, 0.0238, 0.0077], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0169, 0.0152, 0.0157, 0.0164, 0.0121, 0.0168, 0.0111], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 05:42:46,011 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95873.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:43:27,210 INFO [train.py:904] (7/8) Epoch 10, batch 4550, loss[loss=0.2255, simple_loss=0.3113, pruned_loss=0.0699, over 16251.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2961, pruned_loss=0.06429, over 3221789.94 frames. ], batch size: 165, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:43:35,577 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95908.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 05:43:47,808 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95916.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:43:59,566 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:44:10,340 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.676e+02 2.125e+02 2.355e+02 2.832e+02 4.769e+02, threshold=4.710e+02, percent-clipped=0.0 2023-04-29 05:44:39,221 INFO [train.py:904] (7/8) Epoch 10, batch 4600, loss[loss=0.2105, simple_loss=0.2938, pruned_loss=0.06357, over 16906.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2968, pruned_loss=0.06429, over 3244060.56 frames. ], batch size: 109, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:44:57,866 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95964.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:10,196 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=95972.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:12,085 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95973.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:20,971 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95980.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:43,328 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95995.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:45:55,191 INFO [train.py:904] (7/8) Epoch 10, batch 4650, loss[loss=0.2448, simple_loss=0.3104, pruned_loss=0.0896, over 11394.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2958, pruned_loss=0.06397, over 3245565.90 frames. ], batch size: 248, lr: 6.87e-03, grad_scale: 8.0 2023-04-29 05:46:40,628 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.110e+02 2.403e+02 2.925e+02 6.538e+02, threshold=4.805e+02, percent-clipped=1.0 2023-04-29 05:46:52,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2242, 2.3945, 1.8437, 2.2062, 2.7215, 2.4053, 2.9884, 2.9461], device='cuda:7'), covar=tensor([0.0059, 0.0279, 0.0407, 0.0318, 0.0164, 0.0277, 0.0131, 0.0166], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0195, 0.0193, 0.0190, 0.0194, 0.0197, 0.0198, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:46:55,334 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96041.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:46:58,682 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96043.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:46:58,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6357, 2.1766, 2.2582, 4.3887, 2.1849, 2.6248, 2.2196, 2.4397], device='cuda:7'), covar=tensor([0.0852, 0.2904, 0.2065, 0.0352, 0.3514, 0.2132, 0.2674, 0.2927], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0385, 0.0318, 0.0318, 0.0407, 0.0441, 0.0346, 0.0452], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:47:10,309 INFO [train.py:904] (7/8) Epoch 10, batch 4700, loss[loss=0.2148, simple_loss=0.2872, pruned_loss=0.07116, over 11756.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2932, pruned_loss=0.06301, over 3233767.66 frames. ], batch size: 247, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:47:13,894 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0557, 2.6920, 2.6621, 2.0246, 2.6082, 2.1221, 2.7270, 2.8106], device='cuda:7'), covar=tensor([0.0298, 0.0754, 0.0479, 0.1602, 0.0744, 0.0869, 0.0583, 0.0708], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0143, 0.0155, 0.0142, 0.0134, 0.0123, 0.0134, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 05:47:52,307 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96080.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:48:19,071 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4298, 4.4393, 4.3235, 4.0152, 3.8892, 4.3399, 4.1811, 4.0478], device='cuda:7'), covar=tensor([0.0513, 0.0412, 0.0228, 0.0243, 0.0888, 0.0409, 0.0453, 0.0532], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0287, 0.0274, 0.0252, 0.0298, 0.0285, 0.0187, 0.0319], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:48:24,007 INFO [train.py:904] (7/8) Epoch 10, batch 4750, loss[loss=0.1708, simple_loss=0.2556, pruned_loss=0.04303, over 17109.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.289, pruned_loss=0.0609, over 3224437.68 frames. ], batch size: 49, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:49:08,954 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.162e+02 2.549e+02 3.399e+02 6.347e+02, threshold=5.097e+02, percent-clipped=5.0 2023-04-29 05:49:23,245 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96141.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:49:38,027 INFO [train.py:904] (7/8) Epoch 10, batch 4800, loss[loss=0.2183, simple_loss=0.299, pruned_loss=0.06883, over 16892.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2853, pruned_loss=0.05868, over 3236203.62 frames. ], batch size: 109, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:49:58,731 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96165.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:49:59,969 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2557, 1.9779, 1.6225, 1.8492, 2.3196, 2.0718, 2.1484, 2.4821], device='cuda:7'), covar=tensor([0.0096, 0.0291, 0.0369, 0.0331, 0.0153, 0.0264, 0.0110, 0.0161], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0197, 0.0195, 0.0193, 0.0196, 0.0199, 0.0200, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:50:03,018 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96168.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 05:50:54,660 INFO [train.py:904] (7/8) Epoch 10, batch 4850, loss[loss=0.1863, simple_loss=0.2678, pruned_loss=0.05236, over 16656.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2871, pruned_loss=0.05868, over 3206339.70 frames. ], batch size: 57, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:50:56,285 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96203.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 05:51:21,224 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5980, 4.5833, 4.4306, 4.1512, 4.0615, 4.4675, 4.3215, 4.1851], device='cuda:7'), covar=tensor([0.0465, 0.0260, 0.0245, 0.0226, 0.0909, 0.0368, 0.0366, 0.0563], device='cuda:7'), in_proj_covar=tensor([0.0227, 0.0283, 0.0272, 0.0248, 0.0294, 0.0282, 0.0184, 0.0314], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:51:32,234 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96226.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:51:40,159 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.214e+02 2.693e+02 3.150e+02 7.967e+02, threshold=5.386e+02, percent-clipped=5.0 2023-04-29 05:52:10,086 INFO [train.py:904] (7/8) Epoch 10, batch 4900, loss[loss=0.2039, simple_loss=0.2852, pruned_loss=0.06124, over 12213.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2863, pruned_loss=0.05733, over 3188245.95 frames. ], batch size: 250, lr: 6.86e-03, grad_scale: 8.0 2023-04-29 05:52:42,917 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96273.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:53:24,553 INFO [train.py:904] (7/8) Epoch 10, batch 4950, loss[loss=0.2131, simple_loss=0.2974, pruned_loss=0.06438, over 16698.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2857, pruned_loss=0.05668, over 3184117.49 frames. ], batch size: 124, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:53:52,126 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:53:56,318 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96324.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:54:05,873 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.230e+02 2.596e+02 3.100e+02 4.863e+02, threshold=5.193e+02, percent-clipped=0.0 2023-04-29 05:54:12,444 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:54:33,271 INFO [train.py:904] (7/8) Epoch 10, batch 5000, loss[loss=0.2059, simple_loss=0.3001, pruned_loss=0.05588, over 16779.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2874, pruned_loss=0.05696, over 3190515.34 frames. ], batch size: 89, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:55:21,106 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96385.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:55:28,753 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96390.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:55:44,808 INFO [train.py:904] (7/8) Epoch 10, batch 5050, loss[loss=0.1875, simple_loss=0.275, pruned_loss=0.04996, over 17277.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2882, pruned_loss=0.05703, over 3190825.22 frames. ], batch size: 52, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:56:02,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7542, 4.6643, 4.6691, 3.8735, 4.6629, 1.7643, 4.3416, 4.5051], device='cuda:7'), covar=tensor([0.0079, 0.0074, 0.0098, 0.0436, 0.0085, 0.2149, 0.0116, 0.0167], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0110, 0.0154, 0.0151, 0.0126, 0.0169, 0.0144, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 05:56:13,867 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0207, 4.0154, 4.5590, 2.2396, 4.9030, 4.8539, 3.1813, 3.4988], device='cuda:7'), covar=tensor([0.0617, 0.0190, 0.0105, 0.1016, 0.0030, 0.0037, 0.0338, 0.0377], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0099, 0.0085, 0.0139, 0.0070, 0.0096, 0.0120, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 05:56:27,894 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.386e+02 2.799e+02 3.376e+02 6.474e+02, threshold=5.598e+02, percent-clipped=3.0 2023-04-29 05:56:33,460 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96436.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:56:55,159 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96451.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:56:55,806 INFO [train.py:904] (7/8) Epoch 10, batch 5100, loss[loss=0.1987, simple_loss=0.2797, pruned_loss=0.0588, over 16631.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2859, pruned_loss=0.05609, over 3210955.12 frames. ], batch size: 68, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:57:20,224 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96468.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:58:08,614 INFO [train.py:904] (7/8) Epoch 10, batch 5150, loss[loss=0.1931, simple_loss=0.2891, pruned_loss=0.04856, over 16941.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2855, pruned_loss=0.05486, over 3218505.81 frames. ], batch size: 90, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:58:11,209 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96503.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:58:30,252 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96516.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 05:58:31,603 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2062, 1.5007, 1.8800, 2.1024, 2.1882, 2.3300, 1.5908, 2.2572], device='cuda:7'), covar=tensor([0.0129, 0.0288, 0.0169, 0.0185, 0.0179, 0.0109, 0.0309, 0.0059], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0151, 0.0156, 0.0163, 0.0119, 0.0170, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 05:58:37,350 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96521.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:58:50,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2023-04-29 05:58:52,024 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.088e+02 2.479e+02 2.937e+02 7.130e+02, threshold=4.958e+02, percent-clipped=1.0 2023-04-29 05:59:16,596 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-29 05:59:19,749 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1382, 1.9526, 2.5133, 3.0545, 2.8796, 3.4213, 1.9906, 3.4066], device='cuda:7'), covar=tensor([0.0113, 0.0335, 0.0213, 0.0155, 0.0184, 0.0089, 0.0370, 0.0069], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0167, 0.0152, 0.0157, 0.0164, 0.0119, 0.0171, 0.0110], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 05:59:21,946 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96551.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 05:59:22,748 INFO [train.py:904] (7/8) Epoch 10, batch 5200, loss[loss=0.2003, simple_loss=0.289, pruned_loss=0.05582, over 16419.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2847, pruned_loss=0.05455, over 3215715.50 frames. ], batch size: 146, lr: 6.85e-03, grad_scale: 8.0 2023-04-29 05:59:32,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1988, 3.3342, 1.7024, 3.4523, 2.3757, 3.5024, 1.8820, 2.6102], device='cuda:7'), covar=tensor([0.0193, 0.0245, 0.1526, 0.0106, 0.0775, 0.0365, 0.1505, 0.0633], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0162, 0.0186, 0.0119, 0.0166, 0.0200, 0.0192, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 05:59:33,403 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5611, 3.4979, 3.4703, 2.8582, 3.4147, 1.9599, 3.1412, 2.9727], device='cuda:7'), covar=tensor([0.0113, 0.0096, 0.0118, 0.0267, 0.0081, 0.2008, 0.0121, 0.0180], device='cuda:7'), in_proj_covar=tensor([0.0121, 0.0110, 0.0155, 0.0153, 0.0127, 0.0170, 0.0144, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:00:16,139 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 06:00:35,349 INFO [train.py:904] (7/8) Epoch 10, batch 5250, loss[loss=0.1889, simple_loss=0.281, pruned_loss=0.04835, over 15245.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2826, pruned_loss=0.05431, over 3212861.55 frames. ], batch size: 190, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:01:21,023 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.301e+02 2.656e+02 3.144e+02 5.435e+02, threshold=5.311e+02, percent-clipped=2.0 2023-04-29 06:01:26,355 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96636.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:01:31,307 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 06:01:37,023 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.34 vs. limit=5.0 2023-04-29 06:01:48,985 INFO [train.py:904] (7/8) Epoch 10, batch 5300, loss[loss=0.156, simple_loss=0.2314, pruned_loss=0.04032, over 17030.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2793, pruned_loss=0.05334, over 3212466.89 frames. ], batch size: 41, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:02:11,608 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 06:02:30,167 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:02:36,967 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96684.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:03:03,546 INFO [train.py:904] (7/8) Epoch 10, batch 5350, loss[loss=0.2311, simple_loss=0.317, pruned_loss=0.07262, over 16371.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.2779, pruned_loss=0.05337, over 3185029.86 frames. ], batch size: 146, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:03:05,538 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 06:03:48,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.305e+02 2.858e+02 3.473e+02 6.088e+02, threshold=5.716e+02, percent-clipped=3.0 2023-04-29 06:03:53,436 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96736.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:04:08,550 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96746.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:04:15,627 INFO [train.py:904] (7/8) Epoch 10, batch 5400, loss[loss=0.2254, simple_loss=0.3055, pruned_loss=0.07264, over 16437.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.281, pruned_loss=0.0548, over 3165360.68 frames. ], batch size: 146, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:04:52,069 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96776.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:05:03,163 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:05:31,408 INFO [train.py:904] (7/8) Epoch 10, batch 5450, loss[loss=0.2735, simple_loss=0.3327, pruned_loss=0.1072, over 11825.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2843, pruned_loss=0.0567, over 3168863.49 frames. ], batch size: 247, lr: 6.84e-03, grad_scale: 4.0 2023-04-29 06:06:02,278 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96821.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:06:20,034 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.545e+02 3.377e+02 4.195e+02 1.255e+03, threshold=6.754e+02, percent-clipped=10.0 2023-04-29 06:06:27,780 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96837.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:06:49,560 INFO [train.py:904] (7/8) Epoch 10, batch 5500, loss[loss=0.2964, simple_loss=0.3501, pruned_loss=0.1213, over 11853.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2924, pruned_loss=0.06213, over 3145276.49 frames. ], batch size: 246, lr: 6.83e-03, grad_scale: 4.0 2023-04-29 06:07:04,531 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 06:07:17,927 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=96869.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:08:08,536 INFO [train.py:904] (7/8) Epoch 10, batch 5550, loss[loss=0.2615, simple_loss=0.335, pruned_loss=0.09402, over 16672.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3003, pruned_loss=0.06792, over 3121985.13 frames. ], batch size: 134, lr: 6.83e-03, grad_scale: 4.0 2023-04-29 06:08:37,654 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3848, 4.6319, 4.3607, 4.4027, 4.1067, 4.0619, 4.2048, 4.6476], device='cuda:7'), covar=tensor([0.0936, 0.0821, 0.1004, 0.0697, 0.0800, 0.1373, 0.0965, 0.0894], device='cuda:7'), in_proj_covar=tensor([0.0504, 0.0639, 0.0532, 0.0441, 0.0403, 0.0412, 0.0530, 0.0491], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:09:01,355 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.332e+02 3.697e+02 4.404e+02 5.267e+02 9.227e+02, threshold=8.809e+02, percent-clipped=8.0 2023-04-29 06:09:16,522 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-04-29 06:09:20,826 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4699, 4.6594, 4.8148, 4.6995, 4.6792, 5.2207, 4.8081, 4.5792], device='cuda:7'), covar=tensor([0.1157, 0.1765, 0.1598, 0.1743, 0.2371, 0.0978, 0.1362, 0.2375], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0470, 0.0497, 0.0411, 0.0547, 0.0535, 0.0404, 0.0559], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 06:09:28,006 INFO [train.py:904] (7/8) Epoch 10, batch 5600, loss[loss=0.2498, simple_loss=0.3247, pruned_loss=0.08749, over 16230.00 frames. ], tot_loss[loss=0.2253, simple_loss=0.3057, pruned_loss=0.07242, over 3098147.95 frames. ], batch size: 165, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:09:58,355 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-04-29 06:10:15,998 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96980.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:10:51,245 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97000.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:10:53,972 INFO [train.py:904] (7/8) Epoch 10, batch 5650, loss[loss=0.2283, simple_loss=0.3006, pruned_loss=0.078, over 16706.00 frames. ], tot_loss[loss=0.2331, simple_loss=0.3114, pruned_loss=0.07741, over 3076459.24 frames. ], batch size: 134, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:11:07,600 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 06:11:14,639 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1026, 3.3353, 3.5300, 3.4831, 3.5096, 3.2918, 3.3128, 3.4083], device='cuda:7'), covar=tensor([0.0424, 0.0617, 0.0416, 0.0495, 0.0511, 0.0520, 0.0891, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0322, 0.0325, 0.0310, 0.0371, 0.0345, 0.0443, 0.0277], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 06:11:17,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7810, 1.2415, 1.6233, 1.5609, 1.8035, 1.8534, 1.4821, 1.6873], device='cuda:7'), covar=tensor([0.0174, 0.0243, 0.0125, 0.0184, 0.0162, 0.0098, 0.0255, 0.0066], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0165, 0.0150, 0.0155, 0.0161, 0.0118, 0.0167, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 06:11:34,647 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97028.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:11:36,383 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2023-04-29 06:11:43,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.521e+02 3.909e+02 4.849e+02 5.778e+02 1.289e+03, threshold=9.698e+02, percent-clipped=3.0 2023-04-29 06:12:02,869 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97046.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:12:11,114 INFO [train.py:904] (7/8) Epoch 10, batch 5700, loss[loss=0.2135, simple_loss=0.3039, pruned_loss=0.06156, over 16592.00 frames. ], tot_loss[loss=0.2364, simple_loss=0.3137, pruned_loss=0.07952, over 3063427.49 frames. ], batch size: 57, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:12:25,387 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97061.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:13:10,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6117, 4.7136, 4.5813, 2.9934, 4.0474, 4.6069, 4.0991, 2.7817], device='cuda:7'), covar=tensor([0.0401, 0.0015, 0.0028, 0.0285, 0.0054, 0.0058, 0.0045, 0.0295], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0065, 0.0067, 0.0123, 0.0076, 0.0086, 0.0074, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 06:13:16,962 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97094.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:13:29,504 INFO [train.py:904] (7/8) Epoch 10, batch 5750, loss[loss=0.2363, simple_loss=0.3105, pruned_loss=0.0811, over 15326.00 frames. ], tot_loss[loss=0.2383, simple_loss=0.3158, pruned_loss=0.08039, over 3059006.87 frames. ], batch size: 191, lr: 6.83e-03, grad_scale: 2.0 2023-04-29 06:13:29,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4145, 3.5142, 3.2548, 3.0485, 3.1226, 3.3946, 3.2303, 3.1751], device='cuda:7'), covar=tensor([0.0570, 0.0382, 0.0251, 0.0241, 0.0591, 0.0371, 0.1176, 0.0511], device='cuda:7'), in_proj_covar=tensor([0.0235, 0.0292, 0.0277, 0.0253, 0.0300, 0.0292, 0.0188, 0.0323], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:14:17,792 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:14:22,109 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.407e+02 3.370e+02 4.173e+02 4.986e+02 1.197e+03, threshold=8.346e+02, percent-clipped=2.0 2023-04-29 06:14:49,733 INFO [train.py:904] (7/8) Epoch 10, batch 5800, loss[loss=0.2247, simple_loss=0.2983, pruned_loss=0.07552, over 11892.00 frames. ], tot_loss[loss=0.2374, simple_loss=0.3155, pruned_loss=0.07967, over 3043531.54 frames. ], batch size: 246, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:16:07,869 INFO [train.py:904] (7/8) Epoch 10, batch 5850, loss[loss=0.2237, simple_loss=0.3067, pruned_loss=0.07029, over 16943.00 frames. ], tot_loss[loss=0.2346, simple_loss=0.3134, pruned_loss=0.07795, over 3033132.44 frames. ], batch size: 109, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:16:55,027 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 06:16:58,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5837, 4.3039, 4.1982, 2.7650, 3.8831, 4.2550, 3.9134, 2.2239], device='cuda:7'), covar=tensor([0.0421, 0.0028, 0.0037, 0.0361, 0.0052, 0.0086, 0.0047, 0.0403], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0066, 0.0068, 0.0125, 0.0076, 0.0086, 0.0074, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 06:17:00,835 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.062e+02 2.965e+02 3.707e+02 4.622e+02 9.015e+02, threshold=7.415e+02, percent-clipped=1.0 2023-04-29 06:17:28,953 INFO [train.py:904] (7/8) Epoch 10, batch 5900, loss[loss=0.2579, simple_loss=0.3222, pruned_loss=0.0968, over 11764.00 frames. ], tot_loss[loss=0.2342, simple_loss=0.3131, pruned_loss=0.07763, over 3051149.03 frames. ], batch size: 247, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:17:45,186 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97260.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:18:49,540 INFO [train.py:904] (7/8) Epoch 10, batch 5950, loss[loss=0.2388, simple_loss=0.3158, pruned_loss=0.08085, over 16215.00 frames. ], tot_loss[loss=0.2323, simple_loss=0.3134, pruned_loss=0.07559, over 3067936.18 frames. ], batch size: 165, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:18:53,220 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4890, 2.5677, 2.1404, 2.3272, 3.0688, 2.6625, 3.3559, 3.2146], device='cuda:7'), covar=tensor([0.0051, 0.0257, 0.0325, 0.0299, 0.0133, 0.0257, 0.0111, 0.0143], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0191, 0.0191, 0.0190, 0.0191, 0.0194, 0.0195, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:19:04,429 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-04-29 06:19:20,346 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:19:41,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.897e+02 3.264e+02 4.239e+02 7.603e+02, threshold=6.529e+02, percent-clipped=1.0 2023-04-29 06:19:52,890 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97342.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:20:01,462 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97347.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:20:09,130 INFO [train.py:904] (7/8) Epoch 10, batch 6000, loss[loss=0.199, simple_loss=0.2804, pruned_loss=0.05878, over 16398.00 frames. ], tot_loss[loss=0.2302, simple_loss=0.3119, pruned_loss=0.07424, over 3090470.68 frames. ], batch size: 68, lr: 6.82e-03, grad_scale: 4.0 2023-04-29 06:20:09,130 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 06:20:23,727 INFO [train.py:938] (7/8) Epoch 10, validation: loss=0.165, simple_loss=0.2783, pruned_loss=0.02583, over 944034.00 frames. 2023-04-29 06:20:23,728 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 06:20:30,456 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:20:37,253 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-04-29 06:20:58,260 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97374.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:21:10,980 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 06:21:42,487 INFO [train.py:904] (7/8) Epoch 10, batch 6050, loss[loss=0.2219, simple_loss=0.3084, pruned_loss=0.06766, over 16523.00 frames. ], tot_loss[loss=0.2273, simple_loss=0.3095, pruned_loss=0.0726, over 3106647.44 frames. ], batch size: 75, lr: 6.82e-03, grad_scale: 2.0 2023-04-29 06:21:45,490 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97403.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 06:21:52,866 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97408.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:22:30,440 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:22:35,017 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97435.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:22:35,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 3.012e+02 3.614e+02 4.338e+02 6.849e+02, threshold=7.229e+02, percent-clipped=1.0 2023-04-29 06:22:47,505 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2496, 1.5487, 1.9029, 2.1071, 2.3635, 2.4334, 1.5987, 2.3513], device='cuda:7'), covar=tensor([0.0150, 0.0333, 0.0194, 0.0212, 0.0182, 0.0117, 0.0321, 0.0079], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0164, 0.0149, 0.0153, 0.0160, 0.0117, 0.0167, 0.0108], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 06:23:02,108 INFO [train.py:904] (7/8) Epoch 10, batch 6100, loss[loss=0.2093, simple_loss=0.2943, pruned_loss=0.0622, over 16632.00 frames. ], tot_loss[loss=0.2253, simple_loss=0.3084, pruned_loss=0.07112, over 3123074.57 frames. ], batch size: 62, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:23:26,032 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2305, 4.9245, 4.7783, 5.3079, 5.5417, 4.8266, 5.4558, 5.5081], device='cuda:7'), covar=tensor([0.1391, 0.1124, 0.2370, 0.0897, 0.0803, 0.0897, 0.0940, 0.0835], device='cuda:7'), in_proj_covar=tensor([0.0506, 0.0626, 0.0760, 0.0637, 0.0485, 0.0486, 0.0501, 0.0567], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:23:49,433 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97480.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:23:55,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2818, 3.1979, 3.2297, 3.4183, 3.4466, 3.1944, 3.4178, 3.5047], device='cuda:7'), covar=tensor([0.1132, 0.0978, 0.1211, 0.0648, 0.0722, 0.2137, 0.0885, 0.0744], device='cuda:7'), in_proj_covar=tensor([0.0503, 0.0623, 0.0757, 0.0634, 0.0483, 0.0483, 0.0499, 0.0565], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:24:23,713 INFO [train.py:904] (7/8) Epoch 10, batch 6150, loss[loss=0.2074, simple_loss=0.2914, pruned_loss=0.06172, over 16745.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3066, pruned_loss=0.07085, over 3109964.05 frames. ], batch size: 124, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:25:17,522 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.161e+02 3.229e+02 3.939e+02 5.020e+02 8.476e+02, threshold=7.879e+02, percent-clipped=2.0 2023-04-29 06:25:41,185 INFO [train.py:904] (7/8) Epoch 10, batch 6200, loss[loss=0.2492, simple_loss=0.3099, pruned_loss=0.09422, over 11410.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.3054, pruned_loss=0.0714, over 3081062.97 frames. ], batch size: 246, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:25:46,642 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:26:09,858 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-29 06:26:23,509 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1289, 3.3645, 3.5684, 3.5415, 3.5353, 3.3458, 3.3858, 3.4301], device='cuda:7'), covar=tensor([0.0408, 0.0613, 0.0435, 0.0452, 0.0470, 0.0506, 0.0795, 0.0446], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0328, 0.0329, 0.0316, 0.0377, 0.0351, 0.0452, 0.0281], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 06:26:57,946 INFO [train.py:904] (7/8) Epoch 10, batch 6250, loss[loss=0.2397, simple_loss=0.3226, pruned_loss=0.07835, over 16248.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.3042, pruned_loss=0.07034, over 3092911.05 frames. ], batch size: 165, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:27:14,683 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9760, 1.7622, 2.3387, 2.8039, 2.7344, 3.1194, 1.7714, 3.2117], device='cuda:7'), covar=tensor([0.0105, 0.0337, 0.0210, 0.0178, 0.0179, 0.0113, 0.0401, 0.0075], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0164, 0.0149, 0.0153, 0.0160, 0.0117, 0.0168, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 06:27:18,730 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:27:18,871 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:27:47,826 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 3.287e+02 3.846e+02 4.993e+02 1.247e+03, threshold=7.692e+02, percent-clipped=7.0 2023-04-29 06:28:11,802 INFO [train.py:904] (7/8) Epoch 10, batch 6300, loss[loss=0.2029, simple_loss=0.2915, pruned_loss=0.05715, over 16663.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.3045, pruned_loss=0.07026, over 3085686.63 frames. ], batch size: 57, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:28:18,615 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97656.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:28:27,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0121, 1.9369, 2.4579, 2.9665, 2.8693, 3.3312, 1.9246, 3.2308], device='cuda:7'), covar=tensor([0.0123, 0.0326, 0.0209, 0.0163, 0.0179, 0.0092, 0.0379, 0.0100], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0164, 0.0148, 0.0153, 0.0160, 0.0117, 0.0168, 0.0109], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 06:28:48,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2885, 4.0498, 4.0030, 2.7047, 3.5293, 3.9461, 3.8333, 2.1323], device='cuda:7'), covar=tensor([0.0384, 0.0024, 0.0030, 0.0268, 0.0064, 0.0084, 0.0041, 0.0329], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0066, 0.0068, 0.0126, 0.0077, 0.0087, 0.0075, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 06:28:49,006 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97674.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 06:29:24,504 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97698.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 06:29:30,251 INFO [train.py:904] (7/8) Epoch 10, batch 6350, loss[loss=0.2569, simple_loss=0.329, pruned_loss=0.09235, over 16209.00 frames. ], tot_loss[loss=0.2254, simple_loss=0.3062, pruned_loss=0.07225, over 3086525.26 frames. ], batch size: 165, lr: 6.81e-03, grad_scale: 2.0 2023-04-29 06:29:31,971 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97703.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:29:33,675 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:30:13,654 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97730.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:30:22,204 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0013, 4.9978, 4.8017, 4.1480, 4.8538, 1.7591, 4.6088, 4.6902], device='cuda:7'), covar=tensor([0.0054, 0.0044, 0.0104, 0.0300, 0.0057, 0.2298, 0.0083, 0.0120], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0110, 0.0157, 0.0153, 0.0128, 0.0172, 0.0144, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:30:22,238 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97735.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 06:30:22,848 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.422e+02 4.156e+02 5.083e+02 8.318e+02, threshold=8.312e+02, percent-clipped=1.0 2023-04-29 06:30:46,256 INFO [train.py:904] (7/8) Epoch 10, batch 6400, loss[loss=0.2079, simple_loss=0.2981, pruned_loss=0.05887, over 16603.00 frames. ], tot_loss[loss=0.2266, simple_loss=0.3065, pruned_loss=0.07339, over 3080318.66 frames. ], batch size: 75, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:32:00,790 INFO [train.py:904] (7/8) Epoch 10, batch 6450, loss[loss=0.2173, simple_loss=0.2953, pruned_loss=0.06965, over 15300.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.3067, pruned_loss=0.07307, over 3068662.13 frames. ], batch size: 191, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:32:57,258 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.110e+02 2.867e+02 3.528e+02 4.342e+02 1.041e+03, threshold=7.056e+02, percent-clipped=1.0 2023-04-29 06:33:21,809 INFO [train.py:904] (7/8) Epoch 10, batch 6500, loss[loss=0.2198, simple_loss=0.3054, pruned_loss=0.06709, over 16817.00 frames. ], tot_loss[loss=0.224, simple_loss=0.304, pruned_loss=0.07204, over 3075040.60 frames. ], batch size: 124, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:34:17,491 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97888.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:34:41,153 INFO [train.py:904] (7/8) Epoch 10, batch 6550, loss[loss=0.2435, simple_loss=0.3368, pruned_loss=0.07509, over 17198.00 frames. ], tot_loss[loss=0.2266, simple_loss=0.3069, pruned_loss=0.0731, over 3073156.71 frames. ], batch size: 44, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:34:56,262 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97911.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:35:04,955 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97916.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:35:35,379 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.197e+02 3.236e+02 3.952e+02 5.112e+02 9.277e+02, threshold=7.905e+02, percent-clipped=2.0 2023-04-29 06:35:56,924 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97949.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:36:00,724 INFO [train.py:904] (7/8) Epoch 10, batch 6600, loss[loss=0.2845, simple_loss=0.3406, pruned_loss=0.1142, over 11574.00 frames. ], tot_loss[loss=0.2289, simple_loss=0.3099, pruned_loss=0.07396, over 3065103.25 frames. ], batch size: 246, lr: 6.80e-03, grad_scale: 4.0 2023-04-29 06:36:19,495 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=97964.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:37:14,856 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97998.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:37:16,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 06:37:22,489 INFO [train.py:904] (7/8) Epoch 10, batch 6650, loss[loss=0.1823, simple_loss=0.2664, pruned_loss=0.04908, over 16758.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3092, pruned_loss=0.07404, over 3076873.47 frames. ], batch size: 39, lr: 6.79e-03, grad_scale: 4.0 2023-04-29 06:37:24,771 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98003.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:05,458 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98030.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 06:38:05,481 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98030.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:14,887 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 3.281e+02 3.856e+02 4.909e+02 8.809e+02, threshold=7.713e+02, percent-clipped=1.0 2023-04-29 06:38:31,005 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98046.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:39,001 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98051.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:38:39,802 INFO [train.py:904] (7/8) Epoch 10, batch 6700, loss[loss=0.2003, simple_loss=0.2824, pruned_loss=0.05905, over 16618.00 frames. ], tot_loss[loss=0.2282, simple_loss=0.3081, pruned_loss=0.07416, over 3066202.41 frames. ], batch size: 62, lr: 6.79e-03, grad_scale: 4.0 2023-04-29 06:38:45,376 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2254, 3.1453, 3.3322, 1.6756, 3.5385, 3.6102, 2.7176, 2.6283], device='cuda:7'), covar=tensor([0.0766, 0.0201, 0.0160, 0.1148, 0.0054, 0.0101, 0.0411, 0.0456], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0098, 0.0087, 0.0139, 0.0068, 0.0096, 0.0121, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 06:39:20,941 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98078.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:39:57,936 INFO [train.py:904] (7/8) Epoch 10, batch 6750, loss[loss=0.1925, simple_loss=0.2738, pruned_loss=0.05554, over 17142.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.3068, pruned_loss=0.07395, over 3067720.53 frames. ], batch size: 49, lr: 6.79e-03, grad_scale: 4.0 2023-04-29 06:40:04,400 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98106.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:40:49,803 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.169e+02 3.584e+02 4.478e+02 5.468e+02 7.454e+02, threshold=8.956e+02, percent-clipped=0.0 2023-04-29 06:41:15,040 INFO [train.py:904] (7/8) Epoch 10, batch 6800, loss[loss=0.2259, simple_loss=0.308, pruned_loss=0.07192, over 16400.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.3067, pruned_loss=0.07386, over 3064099.06 frames. ], batch size: 146, lr: 6.79e-03, grad_scale: 8.0 2023-04-29 06:41:39,116 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98167.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:42:33,973 INFO [train.py:904] (7/8) Epoch 10, batch 6850, loss[loss=0.2008, simple_loss=0.3022, pruned_loss=0.04968, over 17090.00 frames. ], tot_loss[loss=0.2274, simple_loss=0.3077, pruned_loss=0.07351, over 3080726.60 frames. ], batch size: 49, lr: 6.79e-03, grad_scale: 8.0 2023-04-29 06:42:46,942 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-29 06:42:47,690 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98211.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:43:24,608 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.168e+02 2.932e+02 3.586e+02 4.673e+02 8.284e+02, threshold=7.173e+02, percent-clipped=0.0 2023-04-29 06:43:37,323 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98244.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:43:49,046 INFO [train.py:904] (7/8) Epoch 10, batch 6900, loss[loss=0.2916, simple_loss=0.3474, pruned_loss=0.1179, over 11436.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.3104, pruned_loss=0.0742, over 3059985.96 frames. ], batch size: 248, lr: 6.79e-03, grad_scale: 8.0 2023-04-29 06:44:01,761 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98259.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:45:09,163 INFO [train.py:904] (7/8) Epoch 10, batch 6950, loss[loss=0.227, simple_loss=0.3077, pruned_loss=0.0732, over 16891.00 frames. ], tot_loss[loss=0.2312, simple_loss=0.3118, pruned_loss=0.07527, over 3067029.61 frames. ], batch size: 109, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:45:54,288 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98330.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 06:46:01,762 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 3.204e+02 3.949e+02 4.845e+02 8.099e+02, threshold=7.898e+02, percent-clipped=3.0 2023-04-29 06:46:27,406 INFO [train.py:904] (7/8) Epoch 10, batch 7000, loss[loss=0.2257, simple_loss=0.317, pruned_loss=0.06723, over 16429.00 frames. ], tot_loss[loss=0.2297, simple_loss=0.3112, pruned_loss=0.07404, over 3073576.23 frames. ], batch size: 146, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:46:40,053 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 06:46:57,872 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-04-29 06:47:08,045 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98378.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 06:47:32,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6159, 2.1631, 2.3203, 4.2414, 2.1965, 2.6150, 2.2760, 2.3389], device='cuda:7'), covar=tensor([0.0872, 0.3181, 0.2021, 0.0378, 0.3634, 0.2068, 0.2845, 0.3031], device='cuda:7'), in_proj_covar=tensor([0.0356, 0.0382, 0.0320, 0.0315, 0.0406, 0.0432, 0.0345, 0.0448], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:47:43,139 INFO [train.py:904] (7/8) Epoch 10, batch 7050, loss[loss=0.2377, simple_loss=0.3193, pruned_loss=0.07804, over 16676.00 frames. ], tot_loss[loss=0.2294, simple_loss=0.3117, pruned_loss=0.07354, over 3080798.45 frames. ], batch size: 76, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:47:57,207 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98410.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:48:34,004 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-04-29 06:48:34,412 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 3.125e+02 3.922e+02 4.883e+02 1.063e+03, threshold=7.844e+02, percent-clipped=4.0 2023-04-29 06:48:55,986 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 06:48:59,560 INFO [train.py:904] (7/8) Epoch 10, batch 7100, loss[loss=0.2351, simple_loss=0.3158, pruned_loss=0.07719, over 15294.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.3104, pruned_loss=0.07321, over 3072189.15 frames. ], batch size: 190, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:49:14,096 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98462.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:49:27,978 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:49:37,784 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3393, 2.5468, 1.9525, 2.1684, 2.9046, 2.4859, 3.1908, 3.0946], device='cuda:7'), covar=tensor([0.0062, 0.0259, 0.0390, 0.0340, 0.0162, 0.0289, 0.0156, 0.0155], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0193, 0.0192, 0.0192, 0.0193, 0.0197, 0.0196, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:50:12,814 INFO [train.py:904] (7/8) Epoch 10, batch 7150, loss[loss=0.2553, simple_loss=0.3404, pruned_loss=0.08504, over 16276.00 frames. ], tot_loss[loss=0.2273, simple_loss=0.3085, pruned_loss=0.07302, over 3062900.33 frames. ], batch size: 165, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:50:47,380 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6157, 2.1617, 1.7719, 1.9876, 2.5119, 2.2087, 2.6527, 2.7611], device='cuda:7'), covar=tensor([0.0104, 0.0277, 0.0378, 0.0341, 0.0184, 0.0293, 0.0129, 0.0167], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0195, 0.0194, 0.0193, 0.0194, 0.0197, 0.0198, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:51:03,788 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.197e+02 3.117e+02 3.720e+02 4.761e+02 8.896e+02, threshold=7.439e+02, percent-clipped=3.0 2023-04-29 06:51:15,661 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98543.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:51:16,770 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98544.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:51:29,379 INFO [train.py:904] (7/8) Epoch 10, batch 7200, loss[loss=0.2156, simple_loss=0.2889, pruned_loss=0.07115, over 11536.00 frames. ], tot_loss[loss=0.2248, simple_loss=0.3067, pruned_loss=0.0714, over 3064484.19 frames. ], batch size: 248, lr: 6.78e-03, grad_scale: 8.0 2023-04-29 06:51:48,144 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 06:52:32,713 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98592.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:52:49,213 INFO [train.py:904] (7/8) Epoch 10, batch 7250, loss[loss=0.2393, simple_loss=0.3063, pruned_loss=0.08619, over 11241.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.3039, pruned_loss=0.07001, over 3059841.99 frames. ], batch size: 248, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:52:50,548 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9960, 2.2600, 2.3222, 2.6940, 2.0551, 3.1908, 1.7253, 2.6374], device='cuda:7'), covar=tensor([0.1074, 0.0567, 0.0887, 0.0126, 0.0125, 0.0346, 0.1216, 0.0642], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0156, 0.0179, 0.0136, 0.0202, 0.0205, 0.0178, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 06:52:53,547 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98604.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:52:57,510 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4919, 3.3265, 2.7320, 2.0712, 2.3351, 2.1616, 3.4953, 3.2021], device='cuda:7'), covar=tensor([0.2455, 0.0675, 0.1475, 0.2127, 0.2170, 0.1820, 0.0473, 0.0999], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0255, 0.0281, 0.0275, 0.0279, 0.0217, 0.0264, 0.0287], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:53:38,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2983, 3.0052, 3.0462, 1.8683, 2.7935, 2.0925, 3.0489, 3.2001], device='cuda:7'), covar=tensor([0.0310, 0.0609, 0.0549, 0.1802, 0.0738, 0.0993, 0.0647, 0.0737], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0143, 0.0159, 0.0144, 0.0136, 0.0127, 0.0138, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 06:53:45,151 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 3.052e+02 3.513e+02 4.472e+02 1.147e+03, threshold=7.026e+02, percent-clipped=3.0 2023-04-29 06:54:05,881 INFO [train.py:904] (7/8) Epoch 10, batch 7300, loss[loss=0.2239, simple_loss=0.3099, pruned_loss=0.06899, over 16671.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3034, pruned_loss=0.07029, over 3046655.97 frames. ], batch size: 134, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:55:01,035 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.91 vs. limit=5.0 2023-04-29 06:55:12,541 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 06:55:23,874 INFO [train.py:904] (7/8) Epoch 10, batch 7350, loss[loss=0.2124, simple_loss=0.2924, pruned_loss=0.06616, over 17256.00 frames. ], tot_loss[loss=0.2231, simple_loss=0.3037, pruned_loss=0.07124, over 3021254.07 frames. ], batch size: 52, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:55:39,570 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-29 06:56:19,126 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.271e+02 3.294e+02 3.832e+02 4.622e+02 6.263e+02, threshold=7.664e+02, percent-clipped=0.0 2023-04-29 06:56:41,016 INFO [train.py:904] (7/8) Epoch 10, batch 7400, loss[loss=0.2393, simple_loss=0.3184, pruned_loss=0.08009, over 16875.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.306, pruned_loss=0.0726, over 3024634.05 frames. ], batch size: 42, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:56:57,662 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98762.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:57:03,074 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98766.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:57:25,559 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7438, 4.0610, 3.2729, 2.3436, 2.9438, 2.5754, 4.2899, 3.7133], device='cuda:7'), covar=tensor([0.2693, 0.0615, 0.1330, 0.2116, 0.2232, 0.1619, 0.0404, 0.0931], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0256, 0.0283, 0.0276, 0.0280, 0.0217, 0.0265, 0.0289], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:57:29,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5734, 3.6546, 2.9488, 2.1089, 2.5025, 2.2996, 3.8310, 3.4207], device='cuda:7'), covar=tensor([0.2642, 0.0602, 0.1400, 0.2247, 0.2289, 0.1750, 0.0427, 0.1006], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0256, 0.0282, 0.0276, 0.0280, 0.0217, 0.0265, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 06:57:59,092 INFO [train.py:904] (7/8) Epoch 10, batch 7450, loss[loss=0.212, simple_loss=0.2982, pruned_loss=0.06292, over 16851.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.3068, pruned_loss=0.07299, over 3048417.22 frames. ], batch size: 42, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:58:14,963 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=98810.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 06:58:57,292 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.965e+02 3.303e+02 3.942e+02 4.934e+02 1.157e+03, threshold=7.883e+02, percent-clipped=3.0 2023-04-29 06:59:19,909 INFO [train.py:904] (7/8) Epoch 10, batch 7500, loss[loss=0.23, simple_loss=0.3137, pruned_loss=0.07312, over 16735.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.3072, pruned_loss=0.07219, over 3056294.30 frames. ], batch size: 124, lr: 6.77e-03, grad_scale: 2.0 2023-04-29 06:59:53,002 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5700, 2.5738, 2.2993, 3.7862, 2.8652, 3.8663, 1.3255, 2.6110], device='cuda:7'), covar=tensor([0.1410, 0.0696, 0.1236, 0.0144, 0.0264, 0.0369, 0.1625, 0.0867], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0156, 0.0179, 0.0135, 0.0200, 0.0205, 0.0178, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 07:00:20,508 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98890.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:00:23,056 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1212, 1.9096, 1.4608, 1.5199, 2.1103, 1.8523, 2.0841, 2.2997], device='cuda:7'), covar=tensor([0.0111, 0.0329, 0.0475, 0.0421, 0.0199, 0.0300, 0.0168, 0.0203], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0193, 0.0191, 0.0191, 0.0192, 0.0195, 0.0195, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:00:27,937 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98895.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:00:34,240 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98899.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:00:38,289 INFO [train.py:904] (7/8) Epoch 10, batch 7550, loss[loss=0.2473, simple_loss=0.3093, pruned_loss=0.09261, over 11526.00 frames. ], tot_loss[loss=0.2258, simple_loss=0.3061, pruned_loss=0.0727, over 3051414.20 frames. ], batch size: 248, lr: 6.76e-03, grad_scale: 2.0 2023-04-29 07:01:32,285 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.425e+02 3.280e+02 3.927e+02 5.001e+02 7.830e+02, threshold=7.854e+02, percent-clipped=0.0 2023-04-29 07:01:53,306 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98951.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:01:53,982 INFO [train.py:904] (7/8) Epoch 10, batch 7600, loss[loss=0.2394, simple_loss=0.3168, pruned_loss=0.08094, over 16918.00 frames. ], tot_loss[loss=0.2261, simple_loss=0.3058, pruned_loss=0.07318, over 3054020.25 frames. ], batch size: 109, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:02:01,639 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98956.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:02:24,347 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98971.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:03:11,989 INFO [train.py:904] (7/8) Epoch 10, batch 7650, loss[loss=0.2413, simple_loss=0.3222, pruned_loss=0.08022, over 15288.00 frames. ], tot_loss[loss=0.2268, simple_loss=0.3065, pruned_loss=0.07353, over 3059369.13 frames. ], batch size: 191, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:03:12,435 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0415, 5.3307, 5.0684, 5.1043, 4.7878, 4.7263, 4.8620, 5.4159], device='cuda:7'), covar=tensor([0.0999, 0.0783, 0.0966, 0.0708, 0.0732, 0.0779, 0.0935, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0513, 0.0638, 0.0535, 0.0443, 0.0399, 0.0423, 0.0534, 0.0485], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:03:26,731 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0420, 2.5556, 2.6171, 1.8803, 2.7652, 2.8481, 2.3518, 2.4228], device='cuda:7'), covar=tensor([0.0654, 0.0201, 0.0208, 0.0866, 0.0087, 0.0184, 0.0408, 0.0386], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0099, 0.0085, 0.0137, 0.0068, 0.0096, 0.0120, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 07:03:44,996 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6082, 2.6261, 2.4230, 3.9394, 2.9403, 3.9774, 1.3547, 2.7612], device='cuda:7'), covar=tensor([0.1348, 0.0660, 0.1117, 0.0136, 0.0247, 0.0381, 0.1567, 0.0851], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0157, 0.0179, 0.0136, 0.0202, 0.0207, 0.0179, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 07:03:59,235 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:04:08,574 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.981e+02 3.374e+02 4.235e+02 5.349e+02 9.496e+02, threshold=8.471e+02, percent-clipped=6.0 2023-04-29 07:04:28,273 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99051.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:04:29,017 INFO [train.py:904] (7/8) Epoch 10, batch 7700, loss[loss=0.2207, simple_loss=0.3127, pruned_loss=0.06433, over 16469.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.3065, pruned_loss=0.07381, over 3061293.48 frames. ], batch size: 68, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:04:50,187 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99066.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:05:32,311 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0851, 5.0798, 4.8801, 4.2036, 4.9873, 1.6664, 4.6749, 4.7051], device='cuda:7'), covar=tensor([0.0065, 0.0050, 0.0129, 0.0329, 0.0059, 0.2496, 0.0100, 0.0150], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0107, 0.0155, 0.0150, 0.0126, 0.0173, 0.0142, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:05:43,704 INFO [train.py:904] (7/8) Epoch 10, batch 7750, loss[loss=0.2282, simple_loss=0.3096, pruned_loss=0.07342, over 16225.00 frames. ], tot_loss[loss=0.2266, simple_loss=0.3066, pruned_loss=0.0733, over 3074166.55 frames. ], batch size: 165, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:05:51,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3913, 2.0241, 2.0889, 3.9763, 2.0125, 2.4605, 2.1178, 2.2288], device='cuda:7'), covar=tensor([0.0917, 0.3156, 0.2245, 0.0417, 0.3678, 0.2175, 0.2914, 0.2944], device='cuda:7'), in_proj_covar=tensor([0.0355, 0.0385, 0.0321, 0.0319, 0.0413, 0.0436, 0.0346, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:05:52,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2983, 1.9762, 2.0881, 3.8340, 1.9873, 2.4329, 2.1067, 2.1738], device='cuda:7'), covar=tensor([0.0905, 0.3143, 0.2138, 0.0405, 0.3596, 0.2113, 0.2884, 0.2831], device='cuda:7'), in_proj_covar=tensor([0.0355, 0.0385, 0.0321, 0.0319, 0.0413, 0.0436, 0.0346, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:06:00,265 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99112.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:06:02,526 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99114.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:06:06,044 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 07:06:38,060 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.181e+02 3.321e+02 3.885e+02 5.264e+02 1.269e+03, threshold=7.770e+02, percent-clipped=1.0 2023-04-29 07:06:59,343 INFO [train.py:904] (7/8) Epoch 10, batch 7800, loss[loss=0.2748, simple_loss=0.3266, pruned_loss=0.1115, over 11336.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.3078, pruned_loss=0.07419, over 3074141.69 frames. ], batch size: 247, lr: 6.76e-03, grad_scale: 4.0 2023-04-29 07:07:05,296 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4921, 3.5199, 1.8627, 3.9095, 2.5158, 3.9042, 2.0340, 2.7445], device='cuda:7'), covar=tensor([0.0228, 0.0394, 0.1755, 0.0130, 0.0810, 0.0596, 0.1563, 0.0705], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0163, 0.0187, 0.0120, 0.0165, 0.0204, 0.0192, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 07:08:12,881 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99199.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:08:16,580 INFO [train.py:904] (7/8) Epoch 10, batch 7850, loss[loss=0.2254, simple_loss=0.304, pruned_loss=0.07333, over 15276.00 frames. ], tot_loss[loss=0.2286, simple_loss=0.3087, pruned_loss=0.07429, over 3066131.14 frames. ], batch size: 190, lr: 6.75e-03, grad_scale: 4.0 2023-04-29 07:08:57,403 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99229.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:10,228 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.079e+02 2.953e+02 3.751e+02 4.670e+02 9.934e+02, threshold=7.502e+02, percent-clipped=3.0 2023-04-29 07:09:22,420 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99246.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:24,275 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99247.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:29,708 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99251.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:09:30,446 INFO [train.py:904] (7/8) Epoch 10, batch 7900, loss[loss=0.2107, simple_loss=0.2954, pruned_loss=0.06299, over 16926.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.3079, pruned_loss=0.07329, over 3076168.33 frames. ], batch size: 109, lr: 6.75e-03, grad_scale: 4.0 2023-04-29 07:09:31,893 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3038, 5.5844, 5.2651, 5.4258, 4.9942, 4.9203, 5.0302, 5.6860], device='cuda:7'), covar=tensor([0.1042, 0.0877, 0.1157, 0.0730, 0.0800, 0.0751, 0.1100, 0.0876], device='cuda:7'), in_proj_covar=tensor([0.0511, 0.0637, 0.0536, 0.0445, 0.0400, 0.0424, 0.0538, 0.0487], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:09:55,590 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:10:31,561 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99290.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:10:49,033 INFO [train.py:904] (7/8) Epoch 10, batch 7950, loss[loss=0.2038, simple_loss=0.2863, pruned_loss=0.0606, over 16150.00 frames. ], tot_loss[loss=0.2271, simple_loss=0.3074, pruned_loss=0.07338, over 3072577.98 frames. ], batch size: 35, lr: 6.75e-03, grad_scale: 4.0 2023-04-29 07:11:23,009 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99324.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:11:27,255 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99327.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:11:30,575 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:11:43,786 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.817e+02 3.138e+02 3.909e+02 4.608e+02 7.916e+02, threshold=7.818e+02, percent-clipped=1.0 2023-04-29 07:12:06,027 INFO [train.py:904] (7/8) Epoch 10, batch 8000, loss[loss=0.2307, simple_loss=0.3128, pruned_loss=0.07428, over 15282.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.3077, pruned_loss=0.07418, over 3067099.60 frames. ], batch size: 190, lr: 6.75e-03, grad_scale: 8.0 2023-04-29 07:12:35,349 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 07:12:57,279 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99385.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:13:21,644 INFO [train.py:904] (7/8) Epoch 10, batch 8050, loss[loss=0.2308, simple_loss=0.3135, pruned_loss=0.07409, over 16893.00 frames. ], tot_loss[loss=0.2282, simple_loss=0.3079, pruned_loss=0.07423, over 3058464.53 frames. ], batch size: 109, lr: 6.75e-03, grad_scale: 8.0 2023-04-29 07:13:29,878 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99407.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:13:50,408 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99420.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:14:01,661 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7066, 2.6330, 2.1448, 2.3035, 3.0452, 2.7084, 3.4693, 3.3309], device='cuda:7'), covar=tensor([0.0053, 0.0264, 0.0362, 0.0335, 0.0163, 0.0264, 0.0145, 0.0144], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0194, 0.0193, 0.0192, 0.0193, 0.0196, 0.0197, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:14:18,159 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.046e+02 3.114e+02 3.682e+02 4.495e+02 9.204e+02, threshold=7.364e+02, percent-clipped=1.0 2023-04-29 07:14:39,577 INFO [train.py:904] (7/8) Epoch 10, batch 8100, loss[loss=0.2272, simple_loss=0.2965, pruned_loss=0.07895, over 11833.00 frames. ], tot_loss[loss=0.2268, simple_loss=0.3073, pruned_loss=0.07316, over 3068403.72 frames. ], batch size: 249, lr: 6.75e-03, grad_scale: 8.0 2023-04-29 07:14:46,212 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6341, 4.6271, 4.4658, 4.2668, 4.0756, 4.5863, 4.4260, 4.2636], device='cuda:7'), covar=tensor([0.0551, 0.0420, 0.0262, 0.0245, 0.1015, 0.0403, 0.0414, 0.0604], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0297, 0.0274, 0.0254, 0.0297, 0.0290, 0.0190, 0.0319], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:14:59,165 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6415, 4.5853, 4.4575, 3.5992, 4.4842, 1.5324, 4.2695, 4.2556], device='cuda:7'), covar=tensor([0.0085, 0.0068, 0.0157, 0.0415, 0.0090, 0.2581, 0.0113, 0.0214], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0108, 0.0157, 0.0151, 0.0127, 0.0174, 0.0143, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:15:02,145 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9969, 3.9794, 3.9314, 3.2838, 3.9559, 1.7296, 3.7701, 3.6456], device='cuda:7'), covar=tensor([0.0105, 0.0077, 0.0135, 0.0287, 0.0079, 0.2327, 0.0107, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0108, 0.0157, 0.0151, 0.0127, 0.0174, 0.0143, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:15:03,979 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99468.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:15:23,113 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99481.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:15:54,079 INFO [train.py:904] (7/8) Epoch 10, batch 8150, loss[loss=0.227, simple_loss=0.2988, pruned_loss=0.07761, over 11509.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3045, pruned_loss=0.07152, over 3083531.57 frames. ], batch size: 248, lr: 6.74e-03, grad_scale: 8.0 2023-04-29 07:16:21,968 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-04-29 07:16:35,876 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99529.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:16:45,358 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6630, 3.7333, 4.0803, 4.0662, 4.0642, 3.7720, 3.7936, 3.7476], device='cuda:7'), covar=tensor([0.0336, 0.0601, 0.0406, 0.0393, 0.0438, 0.0436, 0.0888, 0.0548], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0326, 0.0329, 0.0314, 0.0378, 0.0348, 0.0450, 0.0284], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 07:16:49,693 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.193e+02 2.958e+02 3.724e+02 4.411e+02 7.985e+02, threshold=7.447e+02, percent-clipped=3.0 2023-04-29 07:16:55,296 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-29 07:17:01,462 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99546.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:17:09,015 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99551.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:17:09,695 INFO [train.py:904] (7/8) Epoch 10, batch 8200, loss[loss=0.2589, simple_loss=0.3138, pruned_loss=0.102, over 11330.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3027, pruned_loss=0.0713, over 3082785.07 frames. ], batch size: 246, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:17:36,173 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99568.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:04,719 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99585.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:11,887 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99589.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:20,877 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99594.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:28,366 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99599.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:18:32,984 INFO [train.py:904] (7/8) Epoch 10, batch 8250, loss[loss=0.2187, simple_loss=0.3081, pruned_loss=0.06469, over 16236.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.3015, pruned_loss=0.0683, over 3074084.42 frames. ], batch size: 165, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:19:10,808 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99624.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:16,588 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99627.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:19,944 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99629.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:36,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 2.844e+02 3.459e+02 4.111e+02 7.773e+02, threshold=6.919e+02, percent-clipped=2.0 2023-04-29 07:19:55,496 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:19:57,356 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7621, 1.2202, 1.6290, 1.6644, 1.8481, 1.8859, 1.6048, 1.8086], device='cuda:7'), covar=tensor([0.0181, 0.0287, 0.0134, 0.0161, 0.0168, 0.0133, 0.0273, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0167, 0.0148, 0.0152, 0.0163, 0.0117, 0.0168, 0.0107], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 07:19:57,963 INFO [train.py:904] (7/8) Epoch 10, batch 8300, loss[loss=0.1937, simple_loss=0.2704, pruned_loss=0.05851, over 12278.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.298, pruned_loss=0.06513, over 3061433.57 frames. ], batch size: 246, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:20:37,237 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99675.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:20:44,718 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:20:44,801 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:21:21,641 INFO [train.py:904] (7/8) Epoch 10, batch 8350, loss[loss=0.2058, simple_loss=0.2997, pruned_loss=0.0559, over 15348.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2977, pruned_loss=0.06359, over 3051243.81 frames. ], batch size: 191, lr: 6.74e-03, grad_scale: 4.0 2023-04-29 07:21:22,317 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5611, 3.5443, 2.8802, 2.0573, 2.3243, 2.1745, 3.8251, 3.3452], device='cuda:7'), covar=tensor([0.2485, 0.0596, 0.1425, 0.2496, 0.2341, 0.1846, 0.0366, 0.0983], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0251, 0.0278, 0.0273, 0.0277, 0.0217, 0.0261, 0.0283], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:21:30,533 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:22:05,487 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-29 07:22:21,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.332e+02 2.884e+02 3.614e+02 8.033e+02, threshold=5.769e+02, percent-clipped=2.0 2023-04-29 07:22:26,872 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99741.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:22:43,915 INFO [train.py:904] (7/8) Epoch 10, batch 8400, loss[loss=0.2056, simple_loss=0.2954, pruned_loss=0.05787, over 16339.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2942, pruned_loss=0.06019, over 3083207.36 frames. ], batch size: 146, lr: 6.74e-03, grad_scale: 8.0 2023-04-29 07:22:49,992 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99755.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:23:25,437 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99776.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:23:46,725 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3201, 2.5383, 2.0390, 2.0417, 2.8645, 2.4985, 3.1252, 3.1013], device='cuda:7'), covar=tensor([0.0075, 0.0252, 0.0369, 0.0362, 0.0192, 0.0259, 0.0122, 0.0150], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0190, 0.0189, 0.0187, 0.0189, 0.0192, 0.0189, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:24:06,262 INFO [train.py:904] (7/8) Epoch 10, batch 8450, loss[loss=0.1866, simple_loss=0.2804, pruned_loss=0.04641, over 16226.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2921, pruned_loss=0.05853, over 3082181.89 frames. ], batch size: 165, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:24:42,818 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99824.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:25:06,073 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.551e+02 3.005e+02 3.693e+02 6.066e+02, threshold=6.011e+02, percent-clipped=2.0 2023-04-29 07:25:17,302 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 07:25:25,941 INFO [train.py:904] (7/8) Epoch 10, batch 8500, loss[loss=0.1843, simple_loss=0.2619, pruned_loss=0.05339, over 12034.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2878, pruned_loss=0.05575, over 3085260.89 frames. ], batch size: 247, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:26:03,677 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7899, 2.6023, 2.2586, 3.5008, 2.4839, 3.7366, 1.3466, 2.8791], device='cuda:7'), covar=tensor([0.1225, 0.0567, 0.1055, 0.0116, 0.0127, 0.0329, 0.1543, 0.0600], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0156, 0.0177, 0.0133, 0.0196, 0.0203, 0.0178, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 07:26:21,612 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99885.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:26:50,590 INFO [train.py:904] (7/8) Epoch 10, batch 8550, loss[loss=0.2242, simple_loss=0.3149, pruned_loss=0.06674, over 16759.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2854, pruned_loss=0.05497, over 3061275.04 frames. ], batch size: 124, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:27:34,916 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:27:35,009 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:27:50,571 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:28:02,245 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.540e+02 3.329e+02 3.854e+02 7.309e+02, threshold=6.657e+02, percent-clipped=4.0 2023-04-29 07:28:16,663 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99945.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:28:29,561 INFO [train.py:904] (7/8) Epoch 10, batch 8600, loss[loss=0.1763, simple_loss=0.2633, pruned_loss=0.04459, over 12335.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2859, pruned_loss=0.05463, over 3040681.82 frames. ], batch size: 246, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:29:11,033 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=99972.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:29:26,476 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99980.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:30:11,026 INFO [train.py:904] (7/8) Epoch 10, batch 8650, loss[loss=0.1725, simple_loss=0.2763, pruned_loss=0.03436, over 16871.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2834, pruned_loss=0.05262, over 3041392.13 frames. ], batch size: 83, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:30:24,875 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100007.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:31:10,607 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100028.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:31:23,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0528, 4.1675, 3.4080, 2.4798, 2.9335, 2.5937, 4.5603, 3.7945], device='cuda:7'), covar=tensor([0.2229, 0.0707, 0.1316, 0.2176, 0.2251, 0.1665, 0.0364, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0245, 0.0271, 0.0265, 0.0265, 0.0211, 0.0254, 0.0274], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:31:27,928 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:31:31,867 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.482e+02 2.900e+02 4.130e+02 1.018e+03, threshold=5.800e+02, percent-clipped=4.0 2023-04-29 07:31:56,454 INFO [train.py:904] (7/8) Epoch 10, batch 8700, loss[loss=0.1783, simple_loss=0.2706, pruned_loss=0.04301, over 15235.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2803, pruned_loss=0.05096, over 3043103.93 frames. ], batch size: 190, lr: 6.73e-03, grad_scale: 8.0 2023-04-29 07:32:00,778 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.81 vs. limit=5.0 2023-04-29 07:32:06,101 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 07:32:19,592 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 07:32:28,614 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100068.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:32:43,284 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100076.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:33:35,366 INFO [train.py:904] (7/8) Epoch 10, batch 8750, loss[loss=0.1943, simple_loss=0.2739, pruned_loss=0.05741, over 12175.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2796, pruned_loss=0.05001, over 3046830.42 frames. ], batch size: 248, lr: 6.72e-03, grad_scale: 8.0 2023-04-29 07:34:32,195 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100124.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:34:32,250 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100124.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:35:02,705 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.530e+02 3.072e+02 4.161e+02 6.852e+02, threshold=6.144e+02, percent-clipped=8.0 2023-04-29 07:35:29,752 INFO [train.py:904] (7/8) Epoch 10, batch 8800, loss[loss=0.1814, simple_loss=0.278, pruned_loss=0.04236, over 16421.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2777, pruned_loss=0.04875, over 3064683.39 frames. ], batch size: 68, lr: 6.72e-03, grad_scale: 8.0 2023-04-29 07:36:00,654 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 07:36:12,283 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100172.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:37:15,891 INFO [train.py:904] (7/8) Epoch 10, batch 8850, loss[loss=0.2049, simple_loss=0.3054, pruned_loss=0.0522, over 16773.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2807, pruned_loss=0.04824, over 3073933.84 frames. ], batch size: 124, lr: 6.72e-03, grad_scale: 8.0 2023-04-29 07:37:34,193 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5383, 3.5792, 2.7017, 2.1529, 2.3543, 2.2057, 3.7576, 3.3099], device='cuda:7'), covar=tensor([0.2546, 0.0654, 0.1583, 0.2276, 0.2279, 0.1808, 0.0445, 0.0990], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0243, 0.0270, 0.0262, 0.0261, 0.0210, 0.0252, 0.0272], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:38:04,284 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100224.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:38:34,771 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100238.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:38:36,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 2.435e+02 3.067e+02 3.598e+02 6.365e+02, threshold=6.134e+02, percent-clipped=1.0 2023-04-29 07:38:48,308 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100245.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:38:50,625 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8965, 1.8786, 2.1580, 3.2383, 2.0177, 2.0618, 2.0993, 1.9319], device='cuda:7'), covar=tensor([0.0917, 0.3407, 0.2068, 0.0493, 0.4046, 0.2564, 0.3028, 0.3694], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0373, 0.0314, 0.0308, 0.0401, 0.0419, 0.0336, 0.0435], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:39:01,674 INFO [train.py:904] (7/8) Epoch 10, batch 8900, loss[loss=0.2053, simple_loss=0.2981, pruned_loss=0.05625, over 15286.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2804, pruned_loss=0.04746, over 3057391.98 frames. ], batch size: 190, lr: 6.72e-03, grad_scale: 4.0 2023-04-29 07:39:14,669 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5595, 3.5969, 3.3961, 3.2048, 3.2147, 3.5227, 3.2703, 3.3083], device='cuda:7'), covar=tensor([0.0491, 0.0411, 0.0230, 0.0215, 0.0502, 0.0394, 0.1263, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0225, 0.0281, 0.0263, 0.0243, 0.0283, 0.0278, 0.0183, 0.0307], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:39:43,457 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100272.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:40:45,778 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100293.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:41:01,124 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100299.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:41:05,921 INFO [train.py:904] (7/8) Epoch 10, batch 8950, loss[loss=0.1902, simple_loss=0.2759, pruned_loss=0.05227, over 16815.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2803, pruned_loss=0.0477, over 3083237.47 frames. ], batch size: 124, lr: 6.72e-03, grad_scale: 4.0 2023-04-29 07:41:24,950 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 07:42:22,253 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:42:29,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.659e+02 2.569e+02 2.987e+02 3.833e+02 6.749e+02, threshold=5.974e+02, percent-clipped=2.0 2023-04-29 07:42:55,809 INFO [train.py:904] (7/8) Epoch 10, batch 9000, loss[loss=0.2029, simple_loss=0.2814, pruned_loss=0.06222, over 12080.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2772, pruned_loss=0.0465, over 3068738.85 frames. ], batch size: 248, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:42:55,810 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 07:43:05,470 INFO [train.py:938] (7/8) Epoch 10, validation: loss=0.1565, simple_loss=0.2604, pruned_loss=0.02634, over 944034.00 frames. 2023-04-29 07:43:05,471 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 07:43:30,282 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100363.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:44:12,989 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:44:23,624 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-29 07:44:50,493 INFO [train.py:904] (7/8) Epoch 10, batch 9050, loss[loss=0.1846, simple_loss=0.2685, pruned_loss=0.05033, over 16825.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2778, pruned_loss=0.04688, over 3076374.82 frames. ], batch size: 116, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:44:52,388 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100402.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:45:35,874 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1535, 4.2195, 2.4362, 4.8224, 3.0577, 4.6943, 2.5202, 3.3494], device='cuda:7'), covar=tensor([0.0137, 0.0186, 0.1342, 0.0089, 0.0653, 0.0286, 0.1287, 0.0510], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0151, 0.0178, 0.0112, 0.0157, 0.0187, 0.0186, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 07:45:52,444 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7959, 3.7807, 3.8866, 3.8073, 3.8961, 4.2700, 4.0115, 3.7496], device='cuda:7'), covar=tensor([0.2159, 0.2226, 0.2014, 0.2455, 0.2904, 0.1996, 0.1413, 0.2664], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0447, 0.0478, 0.0391, 0.0511, 0.0509, 0.0393, 0.0526], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:46:08,566 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.813e+02 2.533e+02 2.923e+02 3.957e+02 1.423e+03, threshold=5.846e+02, percent-clipped=5.0 2023-04-29 07:46:37,397 INFO [train.py:904] (7/8) Epoch 10, batch 9100, loss[loss=0.1851, simple_loss=0.2837, pruned_loss=0.04323, over 16618.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2775, pruned_loss=0.04732, over 3084304.06 frames. ], batch size: 62, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:46:58,853 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100463.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 07:48:34,514 INFO [train.py:904] (7/8) Epoch 10, batch 9150, loss[loss=0.199, simple_loss=0.2797, pruned_loss=0.05914, over 12136.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2776, pruned_loss=0.04667, over 3069895.09 frames. ], batch size: 250, lr: 6.71e-03, grad_scale: 4.0 2023-04-29 07:49:54,596 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.281e+02 2.663e+02 3.521e+02 5.779e+02, threshold=5.326e+02, percent-clipped=0.0 2023-04-29 07:50:13,983 INFO [train.py:904] (7/8) Epoch 10, batch 9200, loss[loss=0.1716, simple_loss=0.2553, pruned_loss=0.04396, over 11930.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2735, pruned_loss=0.04604, over 3061698.77 frames. ], batch size: 250, lr: 6.71e-03, grad_scale: 8.0 2023-04-29 07:51:04,427 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5019, 2.6492, 2.6952, 4.0586, 3.1561, 4.0493, 1.1211, 3.3194], device='cuda:7'), covar=tensor([0.1479, 0.0663, 0.1012, 0.0130, 0.0171, 0.0298, 0.1665, 0.0565], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0155, 0.0178, 0.0133, 0.0188, 0.0202, 0.0179, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 07:51:35,732 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100594.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 07:51:50,966 INFO [train.py:904] (7/8) Epoch 10, batch 9250, loss[loss=0.1865, simple_loss=0.2784, pruned_loss=0.04734, over 16331.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2733, pruned_loss=0.04615, over 3040464.71 frames. ], batch size: 146, lr: 6.71e-03, grad_scale: 8.0 2023-04-29 07:53:14,010 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.424e+02 2.858e+02 3.550e+02 5.723e+02, threshold=5.716e+02, percent-clipped=3.0 2023-04-29 07:53:42,539 INFO [train.py:904] (7/8) Epoch 10, batch 9300, loss[loss=0.1799, simple_loss=0.256, pruned_loss=0.05185, over 12779.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2722, pruned_loss=0.04552, over 3066041.10 frames. ], batch size: 250, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:53:46,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5201, 3.5320, 3.7959, 1.8541, 4.0058, 4.1123, 3.1115, 3.0539], device='cuda:7'), covar=tensor([0.0768, 0.0236, 0.0197, 0.1224, 0.0044, 0.0079, 0.0337, 0.0417], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0096, 0.0082, 0.0135, 0.0065, 0.0093, 0.0116, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-29 07:54:09,482 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100663.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 07:55:29,740 INFO [train.py:904] (7/8) Epoch 10, batch 9350, loss[loss=0.201, simple_loss=0.2885, pruned_loss=0.05676, over 16906.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2725, pruned_loss=0.04569, over 3083984.90 frames. ], batch size: 116, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:55:50,309 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100711.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:56:29,136 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100730.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 07:56:48,378 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.395e+02 2.806e+02 3.199e+02 7.223e+02, threshold=5.612e+02, percent-clipped=2.0 2023-04-29 07:57:05,895 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9775, 4.0343, 3.8843, 3.6147, 3.5503, 3.9398, 3.6432, 3.7437], device='cuda:7'), covar=tensor([0.0580, 0.0513, 0.0277, 0.0299, 0.0774, 0.0556, 0.0846, 0.0573], device='cuda:7'), in_proj_covar=tensor([0.0222, 0.0277, 0.0261, 0.0241, 0.0281, 0.0278, 0.0182, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 07:57:09,977 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1326, 3.2291, 3.1488, 2.2189, 2.9043, 3.2326, 3.0388, 1.8638], device='cuda:7'), covar=tensor([0.0355, 0.0031, 0.0035, 0.0289, 0.0087, 0.0060, 0.0070, 0.0397], device='cuda:7'), in_proj_covar=tensor([0.0122, 0.0062, 0.0066, 0.0121, 0.0075, 0.0082, 0.0073, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 07:57:12,455 INFO [train.py:904] (7/8) Epoch 10, batch 9400, loss[loss=0.1603, simple_loss=0.2464, pruned_loss=0.03708, over 12207.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2723, pruned_loss=0.04529, over 3084343.90 frames. ], batch size: 248, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 07:57:25,514 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100758.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:57:27,918 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3108, 3.9348, 3.9303, 2.1363, 3.1176, 2.4930, 3.6623, 3.9444], device='cuda:7'), covar=tensor([0.0237, 0.0557, 0.0404, 0.1658, 0.0723, 0.0836, 0.0646, 0.0719], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0133, 0.0153, 0.0139, 0.0132, 0.0122, 0.0133, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 07:58:33,037 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100791.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 07:58:52,912 INFO [train.py:904] (7/8) Epoch 10, batch 9450, loss[loss=0.1837, simple_loss=0.2727, pruned_loss=0.04733, over 16917.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2739, pruned_loss=0.04551, over 3071020.73 frames. ], batch size: 116, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:00:10,875 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.411e+02 3.092e+02 3.645e+02 7.379e+02, threshold=6.183e+02, percent-clipped=4.0 2023-04-29 08:00:34,006 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0548, 2.6252, 2.7185, 1.8446, 2.8574, 2.9794, 2.5377, 2.4909], device='cuda:7'), covar=tensor([0.0627, 0.0204, 0.0173, 0.0975, 0.0074, 0.0151, 0.0371, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0096, 0.0081, 0.0134, 0.0065, 0.0093, 0.0115, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:7') 2023-04-29 08:00:34,625 INFO [train.py:904] (7/8) Epoch 10, batch 9500, loss[loss=0.1618, simple_loss=0.2506, pruned_loss=0.03645, over 12951.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2728, pruned_loss=0.04528, over 3053520.05 frames. ], batch size: 248, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:01:04,318 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9413, 3.4773, 3.3175, 1.8718, 2.9148, 2.1171, 3.3825, 3.4068], device='cuda:7'), covar=tensor([0.0236, 0.0634, 0.0539, 0.1811, 0.0749, 0.1003, 0.0594, 0.0849], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0134, 0.0155, 0.0141, 0.0134, 0.0125, 0.0133, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:01:09,235 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3316, 2.0359, 2.1972, 3.8577, 2.0476, 2.3637, 2.2015, 2.1838], device='cuda:7'), covar=tensor([0.0795, 0.3259, 0.2181, 0.0360, 0.3713, 0.2291, 0.2856, 0.3228], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0370, 0.0313, 0.0305, 0.0399, 0.0413, 0.0334, 0.0430], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:01:15,698 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9508, 2.0730, 2.3590, 3.2455, 2.1076, 2.2058, 2.2311, 2.0992], device='cuda:7'), covar=tensor([0.0817, 0.3161, 0.1881, 0.0492, 0.3670, 0.2257, 0.2770, 0.3241], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0370, 0.0313, 0.0305, 0.0400, 0.0413, 0.0334, 0.0430], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:01:31,823 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9416, 5.2161, 4.9579, 5.0259, 4.7049, 4.7834, 4.6516, 5.2790], device='cuda:7'), covar=tensor([0.0918, 0.0895, 0.0967, 0.0621, 0.0765, 0.0758, 0.1014, 0.0853], device='cuda:7'), in_proj_covar=tensor([0.0486, 0.0611, 0.0502, 0.0425, 0.0386, 0.0404, 0.0512, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:02:03,500 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100894.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:02:09,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0019, 4.2389, 4.0262, 4.0848, 3.7297, 3.8683, 3.8281, 4.2079], device='cuda:7'), covar=tensor([0.0929, 0.0887, 0.0970, 0.0673, 0.0727, 0.1458, 0.0932, 0.1018], device='cuda:7'), in_proj_covar=tensor([0.0487, 0.0612, 0.0503, 0.0426, 0.0386, 0.0405, 0.0513, 0.0470], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:02:20,740 INFO [train.py:904] (7/8) Epoch 10, batch 9550, loss[loss=0.1793, simple_loss=0.2767, pruned_loss=0.04096, over 16926.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2727, pruned_loss=0.04527, over 3045817.15 frames. ], batch size: 96, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:03:23,192 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 08:03:40,250 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.820e+02 2.447e+02 2.728e+02 3.779e+02 7.191e+02, threshold=5.455e+02, percent-clipped=3.0 2023-04-29 08:03:45,458 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=100942.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:04:02,631 INFO [train.py:904] (7/8) Epoch 10, batch 9600, loss[loss=0.1997, simple_loss=0.297, pruned_loss=0.0512, over 16266.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2743, pruned_loss=0.04616, over 3041834.74 frames. ], batch size: 165, lr: 6.70e-03, grad_scale: 8.0 2023-04-29 08:04:13,555 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5157, 4.6110, 4.8009, 4.7148, 4.6460, 5.2306, 4.8061, 4.5242], device='cuda:7'), covar=tensor([0.1126, 0.1726, 0.1521, 0.1759, 0.2425, 0.0860, 0.1238, 0.2241], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0448, 0.0479, 0.0388, 0.0511, 0.0513, 0.0390, 0.0525], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:05:06,944 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0060, 4.0388, 3.8700, 3.6692, 3.5478, 3.9528, 3.6273, 3.7190], device='cuda:7'), covar=tensor([0.0506, 0.0508, 0.0270, 0.0239, 0.0754, 0.0462, 0.0886, 0.0558], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0274, 0.0260, 0.0239, 0.0279, 0.0274, 0.0181, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:05:52,889 INFO [train.py:904] (7/8) Epoch 10, batch 9650, loss[loss=0.1882, simple_loss=0.2797, pruned_loss=0.04833, over 15356.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.276, pruned_loss=0.04635, over 3042728.39 frames. ], batch size: 192, lr: 6.69e-03, grad_scale: 8.0 2023-04-29 08:07:15,510 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.568e+02 3.331e+02 4.260e+02 1.013e+03, threshold=6.663e+02, percent-clipped=7.0 2023-04-29 08:07:41,465 INFO [train.py:904] (7/8) Epoch 10, batch 9700, loss[loss=0.1685, simple_loss=0.2532, pruned_loss=0.04186, over 12211.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2753, pruned_loss=0.04613, over 3047791.68 frames. ], batch size: 248, lr: 6.69e-03, grad_scale: 8.0 2023-04-29 08:07:52,605 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101058.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:07:57,877 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7422, 3.3910, 3.4751, 1.9224, 2.8451, 2.2826, 3.2982, 3.3300], device='cuda:7'), covar=tensor([0.0280, 0.0618, 0.0427, 0.1771, 0.0760, 0.0940, 0.0713, 0.0882], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0132, 0.0152, 0.0140, 0.0132, 0.0123, 0.0132, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:08:54,372 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101086.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:09:25,211 INFO [train.py:904] (7/8) Epoch 10, batch 9750, loss[loss=0.1721, simple_loss=0.2698, pruned_loss=0.03718, over 16850.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2731, pruned_loss=0.04583, over 3053637.71 frames. ], batch size: 96, lr: 6.69e-03, grad_scale: 8.0 2023-04-29 08:09:32,963 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=101106.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:09:35,568 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101107.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:10:45,039 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.458e+02 3.043e+02 3.868e+02 6.520e+02, threshold=6.087e+02, percent-clipped=0.0 2023-04-29 08:11:02,906 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9459, 3.8572, 4.0298, 4.1486, 4.2401, 3.7732, 4.2416, 4.2624], device='cuda:7'), covar=tensor([0.1336, 0.1024, 0.1170, 0.0594, 0.0510, 0.1395, 0.0526, 0.0512], device='cuda:7'), in_proj_covar=tensor([0.0470, 0.0589, 0.0708, 0.0605, 0.0458, 0.0459, 0.0472, 0.0537], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:11:05,050 INFO [train.py:904] (7/8) Epoch 10, batch 9800, loss[loss=0.1665, simple_loss=0.2683, pruned_loss=0.03236, over 16513.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2738, pruned_loss=0.04494, over 3072640.05 frames. ], batch size: 62, lr: 6.69e-03, grad_scale: 4.0 2023-04-29 08:11:37,132 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101168.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:11:41,466 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3447, 1.8854, 1.5836, 1.5488, 2.1879, 1.8265, 2.0693, 2.2975], device='cuda:7'), covar=tensor([0.0084, 0.0327, 0.0399, 0.0411, 0.0208, 0.0321, 0.0155, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0193, 0.0193, 0.0191, 0.0192, 0.0195, 0.0187, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:12:49,201 INFO [train.py:904] (7/8) Epoch 10, batch 9850, loss[loss=0.1985, simple_loss=0.2884, pruned_loss=0.05426, over 16650.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2752, pruned_loss=0.04511, over 3061507.12 frames. ], batch size: 134, lr: 6.69e-03, grad_scale: 4.0 2023-04-29 08:14:04,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7826, 5.0256, 5.2229, 5.0350, 5.0826, 5.5802, 5.0781, 4.7998], device='cuda:7'), covar=tensor([0.0777, 0.1497, 0.1430, 0.1630, 0.2000, 0.0766, 0.1257, 0.2207], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0440, 0.0470, 0.0381, 0.0499, 0.0502, 0.0382, 0.0514], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:14:17,930 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.461e+02 2.965e+02 3.704e+02 9.099e+02, threshold=5.931e+02, percent-clipped=1.0 2023-04-29 08:14:41,634 INFO [train.py:904] (7/8) Epoch 10, batch 9900, loss[loss=0.1723, simple_loss=0.2786, pruned_loss=0.03296, over 16748.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2752, pruned_loss=0.04488, over 3063501.82 frames. ], batch size: 83, lr: 6.69e-03, grad_scale: 4.0 2023-04-29 08:15:07,742 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 08:16:23,613 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2310, 4.1650, 4.6258, 4.6019, 4.5758, 4.3168, 4.3044, 4.1828], device='cuda:7'), covar=tensor([0.0250, 0.0514, 0.0365, 0.0370, 0.0426, 0.0290, 0.0760, 0.0368], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0300, 0.0304, 0.0289, 0.0348, 0.0320, 0.0406, 0.0261], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 08:16:26,320 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5945, 4.4127, 4.5734, 4.7548, 4.9197, 4.4444, 4.9441, 4.9202], device='cuda:7'), covar=tensor([0.1313, 0.0904, 0.1328, 0.0607, 0.0490, 0.0716, 0.0358, 0.0499], device='cuda:7'), in_proj_covar=tensor([0.0466, 0.0583, 0.0700, 0.0597, 0.0455, 0.0452, 0.0468, 0.0532], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:16:40,044 INFO [train.py:904] (7/8) Epoch 10, batch 9950, loss[loss=0.1761, simple_loss=0.2714, pruned_loss=0.04042, over 16615.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2772, pruned_loss=0.04534, over 3068674.75 frames. ], batch size: 57, lr: 6.68e-03, grad_scale: 4.0 2023-04-29 08:17:30,109 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2438, 4.2543, 4.1178, 3.4582, 4.1402, 1.3991, 3.9179, 3.8514], device='cuda:7'), covar=tensor([0.0099, 0.0097, 0.0139, 0.0321, 0.0111, 0.2780, 0.0123, 0.0286], device='cuda:7'), in_proj_covar=tensor([0.0120, 0.0106, 0.0149, 0.0137, 0.0123, 0.0171, 0.0136, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-29 08:18:13,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.784e+02 2.620e+02 2.929e+02 3.358e+02 7.407e+02, threshold=5.858e+02, percent-clipped=2.0 2023-04-29 08:18:42,240 INFO [train.py:904] (7/8) Epoch 10, batch 10000, loss[loss=0.1766, simple_loss=0.2716, pruned_loss=0.04083, over 16974.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2756, pruned_loss=0.04496, over 3074204.86 frames. ], batch size: 109, lr: 6.68e-03, grad_scale: 8.0 2023-04-29 08:19:17,686 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7400, 3.1147, 3.0463, 1.6823, 2.5965, 2.0256, 3.1210, 3.3156], device='cuda:7'), covar=tensor([0.0241, 0.0686, 0.0649, 0.2102, 0.0979, 0.1135, 0.0747, 0.0731], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0131, 0.0152, 0.0139, 0.0132, 0.0122, 0.0131, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:19:54,034 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101386.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 08:20:00,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1826, 4.1994, 4.0322, 3.7865, 3.7012, 4.1039, 3.8488, 3.8510], device='cuda:7'), covar=tensor([0.0480, 0.0445, 0.0264, 0.0259, 0.0766, 0.0435, 0.0745, 0.0590], device='cuda:7'), in_proj_covar=tensor([0.0220, 0.0272, 0.0258, 0.0237, 0.0273, 0.0270, 0.0177, 0.0299], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-29 08:20:23,603 INFO [train.py:904] (7/8) Epoch 10, batch 10050, loss[loss=0.2042, simple_loss=0.2985, pruned_loss=0.05489, over 16277.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2758, pruned_loss=0.04473, over 3076955.99 frames. ], batch size: 146, lr: 6.68e-03, grad_scale: 8.0 2023-04-29 08:21:00,602 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4157, 3.2041, 2.7611, 2.0799, 2.2176, 2.2051, 3.3899, 2.9813], device='cuda:7'), covar=tensor([0.2552, 0.0711, 0.1385, 0.2398, 0.2354, 0.1712, 0.0457, 0.1136], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0241, 0.0269, 0.0261, 0.0250, 0.0208, 0.0251, 0.0271], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:21:25,330 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=101434.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 08:21:36,390 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.448e+02 2.853e+02 3.607e+02 8.482e+02, threshold=5.706e+02, percent-clipped=4.0 2023-04-29 08:21:56,149 INFO [train.py:904] (7/8) Epoch 10, batch 10100, loss[loss=0.1714, simple_loss=0.2646, pruned_loss=0.03907, over 16746.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2759, pruned_loss=0.04478, over 3084573.44 frames. ], batch size: 76, lr: 6.68e-03, grad_scale: 8.0 2023-04-29 08:22:16,007 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101463.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:23:07,602 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7483, 4.5171, 4.7669, 4.9575, 5.0934, 4.5579, 5.0987, 5.0686], device='cuda:7'), covar=tensor([0.1429, 0.1001, 0.1357, 0.0550, 0.0431, 0.0703, 0.0375, 0.0516], device='cuda:7'), in_proj_covar=tensor([0.0468, 0.0583, 0.0704, 0.0600, 0.0453, 0.0454, 0.0469, 0.0532], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:23:38,435 INFO [train.py:904] (7/8) Epoch 11, batch 0, loss[loss=0.2392, simple_loss=0.3156, pruned_loss=0.08136, over 16754.00 frames. ], tot_loss[loss=0.2392, simple_loss=0.3156, pruned_loss=0.08136, over 16754.00 frames. ], batch size: 62, lr: 6.37e-03, grad_scale: 8.0 2023-04-29 08:23:38,435 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 08:23:45,829 INFO [train.py:938] (7/8) Epoch 11, validation: loss=0.1557, simple_loss=0.2595, pruned_loss=0.02591, over 944034.00 frames. 2023-04-29 08:23:45,829 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 08:24:32,851 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1660, 4.9710, 5.5550, 5.5463, 5.6086, 5.1628, 5.1191, 4.7896], device='cuda:7'), covar=tensor([0.0237, 0.0426, 0.0305, 0.0425, 0.0461, 0.0358, 0.0958, 0.0420], device='cuda:7'), in_proj_covar=tensor([0.0293, 0.0302, 0.0306, 0.0290, 0.0347, 0.0322, 0.0408, 0.0263], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 08:24:43,129 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.573e+02 2.991e+02 3.932e+02 8.479e+02, threshold=5.981e+02, percent-clipped=4.0 2023-04-29 08:24:43,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9503, 4.2643, 3.1587, 2.2835, 2.8960, 2.4059, 4.6311, 3.7435], device='cuda:7'), covar=tensor([0.2380, 0.0637, 0.1419, 0.2169, 0.2154, 0.1745, 0.0336, 0.0985], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0244, 0.0274, 0.0264, 0.0254, 0.0211, 0.0255, 0.0276], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:24:54,829 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-29 08:24:55,127 INFO [train.py:904] (7/8) Epoch 11, batch 50, loss[loss=0.207, simple_loss=0.2935, pruned_loss=0.06024, over 17084.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2911, pruned_loss=0.06588, over 752331.02 frames. ], batch size: 53, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:25:29,726 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5652, 3.2955, 3.6301, 2.6895, 3.2148, 3.5802, 3.4330, 2.1190], device='cuda:7'), covar=tensor([0.0361, 0.0162, 0.0040, 0.0259, 0.0087, 0.0086, 0.0067, 0.0379], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0065, 0.0067, 0.0123, 0.0076, 0.0084, 0.0074, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:26:05,602 INFO [train.py:904] (7/8) Epoch 11, batch 100, loss[loss=0.195, simple_loss=0.2895, pruned_loss=0.05027, over 16730.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2848, pruned_loss=0.06252, over 1318728.04 frames. ], batch size: 62, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:26:35,203 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0200, 3.3183, 2.9564, 5.0912, 4.3643, 4.6405, 1.7635, 3.5438], device='cuda:7'), covar=tensor([0.1158, 0.0550, 0.0989, 0.0177, 0.0232, 0.0348, 0.1359, 0.0620], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0155, 0.0179, 0.0134, 0.0184, 0.0203, 0.0180, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:27:03,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.523e+02 3.127e+02 3.796e+02 9.987e+02, threshold=6.254e+02, percent-clipped=2.0 2023-04-29 08:27:12,722 INFO [train.py:904] (7/8) Epoch 11, batch 150, loss[loss=0.1608, simple_loss=0.2495, pruned_loss=0.03605, over 17216.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2818, pruned_loss=0.05983, over 1755443.25 frames. ], batch size: 44, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:27:15,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6970, 4.6413, 4.6559, 4.1690, 4.5799, 1.9393, 4.3946, 4.4707], device='cuda:7'), covar=tensor([0.0121, 0.0100, 0.0140, 0.0290, 0.0107, 0.2149, 0.0126, 0.0182], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0111, 0.0155, 0.0143, 0.0127, 0.0177, 0.0142, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:27:52,904 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1404, 3.4143, 3.2646, 2.0739, 2.9227, 2.3649, 3.6221, 3.5508], device='cuda:7'), covar=tensor([0.0243, 0.0720, 0.0654, 0.1652, 0.0760, 0.0928, 0.0562, 0.0725], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0134, 0.0154, 0.0141, 0.0133, 0.0123, 0.0133, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:27:59,445 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101684.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:28:09,636 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 08:28:23,287 INFO [train.py:904] (7/8) Epoch 11, batch 200, loss[loss=0.2099, simple_loss=0.2963, pruned_loss=0.06179, over 16474.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2807, pruned_loss=0.05957, over 2098989.70 frames. ], batch size: 68, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:29:21,761 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.429e+02 2.923e+02 3.406e+02 5.474e+02, threshold=5.846e+02, percent-clipped=1.0 2023-04-29 08:29:23,312 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101745.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:29:31,806 INFO [train.py:904] (7/8) Epoch 11, batch 250, loss[loss=0.151, simple_loss=0.2319, pruned_loss=0.03501, over 15866.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2785, pruned_loss=0.05856, over 2368532.01 frames. ], batch size: 35, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:29:46,176 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101763.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:29:58,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7673, 3.8801, 3.0420, 2.4234, 2.7637, 2.5279, 4.0804, 3.5586], device='cuda:7'), covar=tensor([0.2425, 0.0672, 0.1428, 0.2116, 0.2182, 0.1639, 0.0514, 0.1109], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0250, 0.0279, 0.0269, 0.0265, 0.0217, 0.0262, 0.0286], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:30:37,965 INFO [train.py:904] (7/8) Epoch 11, batch 300, loss[loss=0.1755, simple_loss=0.2752, pruned_loss=0.03786, over 17063.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2756, pruned_loss=0.05667, over 2581360.82 frames. ], batch size: 50, lr: 6.37e-03, grad_scale: 1.0 2023-04-29 08:30:51,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=101811.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:30:57,292 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 08:31:35,547 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.476e+02 3.003e+02 3.668e+02 8.355e+02, threshold=6.006e+02, percent-clipped=2.0 2023-04-29 08:31:47,613 INFO [train.py:904] (7/8) Epoch 11, batch 350, loss[loss=0.1673, simple_loss=0.2465, pruned_loss=0.04407, over 15948.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2719, pruned_loss=0.05458, over 2753462.43 frames. ], batch size: 35, lr: 6.36e-03, grad_scale: 1.0 2023-04-29 08:32:07,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5173, 3.5818, 2.0843, 3.7538, 2.7071, 3.8085, 2.1773, 2.8864], device='cuda:7'), covar=tensor([0.0221, 0.0363, 0.1430, 0.0254, 0.0702, 0.0514, 0.1306, 0.0603], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0161, 0.0187, 0.0125, 0.0167, 0.0203, 0.0196, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:32:56,339 INFO [train.py:904] (7/8) Epoch 11, batch 400, loss[loss=0.151, simple_loss=0.2386, pruned_loss=0.03171, over 17047.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2701, pruned_loss=0.0539, over 2882350.33 frames. ], batch size: 41, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:33:22,135 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101920.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:33:29,702 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1532, 4.1555, 4.5574, 4.5490, 4.6790, 4.3172, 4.1756, 4.1591], device='cuda:7'), covar=tensor([0.0498, 0.0846, 0.0602, 0.0726, 0.0616, 0.0540, 0.1437, 0.0623], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0330, 0.0335, 0.0313, 0.0375, 0.0349, 0.0447, 0.0283], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 08:33:54,607 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.298e+02 2.716e+02 3.213e+02 6.833e+02, threshold=5.433e+02, percent-clipped=1.0 2023-04-29 08:34:06,168 INFO [train.py:904] (7/8) Epoch 11, batch 450, loss[loss=0.2006, simple_loss=0.2879, pruned_loss=0.05659, over 17117.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2682, pruned_loss=0.05315, over 2985865.86 frames. ], batch size: 49, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:34:46,983 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101981.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 08:34:54,572 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101987.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:35:19,215 INFO [train.py:904] (7/8) Epoch 11, batch 500, loss[loss=0.1675, simple_loss=0.2487, pruned_loss=0.04319, over 15606.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2676, pruned_loss=0.05256, over 3063776.59 frames. ], batch size: 190, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:36:13,501 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:36:15,848 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 08:36:19,096 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.707e+02 2.413e+02 2.842e+02 3.807e+02 1.018e+03, threshold=5.684e+02, percent-clipped=5.0 2023-04-29 08:36:24,848 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102048.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:36:29,132 INFO [train.py:904] (7/8) Epoch 11, batch 550, loss[loss=0.1619, simple_loss=0.2584, pruned_loss=0.03265, over 17281.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2669, pruned_loss=0.05182, over 3117580.42 frames. ], batch size: 52, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:36:50,097 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 08:36:57,870 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-29 08:37:07,484 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8376, 4.1775, 4.3134, 3.3629, 3.7039, 4.2241, 3.9625, 2.5007], device='cuda:7'), covar=tensor([0.0330, 0.0053, 0.0035, 0.0224, 0.0076, 0.0073, 0.0054, 0.0335], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0069, 0.0069, 0.0125, 0.0078, 0.0087, 0.0077, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:37:12,998 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7791, 3.0978, 3.0280, 1.9744, 2.6298, 2.1909, 3.3360, 3.2891], device='cuda:7'), covar=tensor([0.0254, 0.0804, 0.0590, 0.1727, 0.0783, 0.0933, 0.0554, 0.0852], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0139, 0.0156, 0.0142, 0.0135, 0.0125, 0.0135, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:37:40,178 INFO [train.py:904] (7/8) Epoch 11, batch 600, loss[loss=0.2099, simple_loss=0.2735, pruned_loss=0.07315, over 16776.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2672, pruned_loss=0.05323, over 3163654.93 frames. ], batch size: 124, lr: 6.36e-03, grad_scale: 2.0 2023-04-29 08:38:38,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.665e+02 3.250e+02 4.151e+02 1.142e+03, threshold=6.501e+02, percent-clipped=8.0 2023-04-29 08:38:49,713 INFO [train.py:904] (7/8) Epoch 11, batch 650, loss[loss=0.1832, simple_loss=0.2774, pruned_loss=0.04445, over 17112.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2654, pruned_loss=0.05279, over 3194970.56 frames. ], batch size: 47, lr: 6.35e-03, grad_scale: 2.0 2023-04-29 08:38:54,699 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8135, 3.9947, 2.3896, 4.5534, 2.9982, 4.5821, 2.5032, 3.1588], device='cuda:7'), covar=tensor([0.0253, 0.0359, 0.1479, 0.0223, 0.0770, 0.0389, 0.1442, 0.0689], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0165, 0.0189, 0.0129, 0.0169, 0.0207, 0.0198, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 08:38:55,178 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-29 08:39:58,899 INFO [train.py:904] (7/8) Epoch 11, batch 700, loss[loss=0.1717, simple_loss=0.2504, pruned_loss=0.04649, over 16769.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.265, pruned_loss=0.05265, over 3211072.34 frames. ], batch size: 83, lr: 6.35e-03, grad_scale: 2.0 2023-04-29 08:40:23,193 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9278, 4.9555, 5.4221, 5.4541, 5.4388, 5.0374, 5.0039, 4.7465], device='cuda:7'), covar=tensor([0.0300, 0.0417, 0.0452, 0.0442, 0.0452, 0.0344, 0.0925, 0.0424], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0339, 0.0346, 0.0320, 0.0388, 0.0359, 0.0459, 0.0290], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 08:40:57,198 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.406e+02 3.042e+02 3.688e+02 8.109e+02, threshold=6.084e+02, percent-clipped=2.0 2023-04-29 08:41:09,232 INFO [train.py:904] (7/8) Epoch 11, batch 750, loss[loss=0.2158, simple_loss=0.2792, pruned_loss=0.07614, over 16821.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2651, pruned_loss=0.05249, over 3242840.22 frames. ], batch size: 109, lr: 6.35e-03, grad_scale: 2.0 2023-04-29 08:41:42,369 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102276.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 08:42:18,054 INFO [train.py:904] (7/8) Epoch 11, batch 800, loss[loss=0.1785, simple_loss=0.2731, pruned_loss=0.04194, over 17027.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2647, pruned_loss=0.052, over 3261187.23 frames. ], batch size: 50, lr: 6.35e-03, grad_scale: 4.0 2023-04-29 08:42:40,082 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2159, 5.5626, 5.2931, 5.3606, 5.0501, 4.9297, 5.0035, 5.6112], device='cuda:7'), covar=tensor([0.0999, 0.0769, 0.0920, 0.0670, 0.0693, 0.0817, 0.0943, 0.0868], device='cuda:7'), in_proj_covar=tensor([0.0544, 0.0681, 0.0564, 0.0476, 0.0431, 0.0445, 0.0574, 0.0523], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 08:43:11,969 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102340.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:43:15,335 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102343.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:43:16,114 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.524e+02 3.019e+02 3.517e+02 6.392e+02, threshold=6.037e+02, percent-clipped=1.0 2023-04-29 08:43:27,552 INFO [train.py:904] (7/8) Epoch 11, batch 850, loss[loss=0.1701, simple_loss=0.2556, pruned_loss=0.04233, over 16796.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2638, pruned_loss=0.0515, over 3274493.05 frames. ], batch size: 102, lr: 6.35e-03, grad_scale: 4.0 2023-04-29 08:44:17,573 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=102388.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:44:31,812 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8848, 4.1444, 4.1825, 3.1108, 3.8280, 4.1989, 3.9139, 2.0296], device='cuda:7'), covar=tensor([0.0395, 0.0071, 0.0062, 0.0324, 0.0111, 0.0114, 0.0104, 0.0544], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0071, 0.0071, 0.0128, 0.0079, 0.0089, 0.0078, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:44:37,484 INFO [train.py:904] (7/8) Epoch 11, batch 900, loss[loss=0.2, simple_loss=0.2787, pruned_loss=0.06063, over 16713.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2627, pruned_loss=0.05087, over 3278637.02 frames. ], batch size: 76, lr: 6.35e-03, grad_scale: 4.0 2023-04-29 08:45:35,222 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.314e+02 2.859e+02 3.467e+02 6.346e+02, threshold=5.718e+02, percent-clipped=1.0 2023-04-29 08:45:45,394 INFO [train.py:904] (7/8) Epoch 11, batch 950, loss[loss=0.1751, simple_loss=0.2519, pruned_loss=0.04919, over 16202.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2627, pruned_loss=0.05108, over 3284588.51 frames. ], batch size: 164, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:46:54,283 INFO [train.py:904] (7/8) Epoch 11, batch 1000, loss[loss=0.1707, simple_loss=0.242, pruned_loss=0.04966, over 16847.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2615, pruned_loss=0.0505, over 3289725.06 frames. ], batch size: 96, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:47:52,028 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.460e+02 2.841e+02 3.268e+02 6.263e+02, threshold=5.683e+02, percent-clipped=1.0 2023-04-29 08:48:02,262 INFO [train.py:904] (7/8) Epoch 11, batch 1050, loss[loss=0.1927, simple_loss=0.266, pruned_loss=0.05972, over 16626.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.262, pruned_loss=0.05071, over 3291770.67 frames. ], batch size: 134, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:48:36,348 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102576.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:49:12,614 INFO [train.py:904] (7/8) Epoch 11, batch 1100, loss[loss=0.1614, simple_loss=0.2404, pruned_loss=0.04116, over 16827.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2612, pruned_loss=0.05037, over 3298521.43 frames. ], batch size: 102, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:49:43,776 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=102624.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 08:50:08,332 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102643.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:50:09,059 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.199e+02 2.657e+02 3.494e+02 5.570e+02, threshold=5.313e+02, percent-clipped=0.0 2023-04-29 08:50:20,437 INFO [train.py:904] (7/8) Epoch 11, batch 1150, loss[loss=0.1623, simple_loss=0.2487, pruned_loss=0.03798, over 17216.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2609, pruned_loss=0.05002, over 3305277.76 frames. ], batch size: 44, lr: 6.34e-03, grad_scale: 4.0 2023-04-29 08:50:27,381 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-29 08:51:14,263 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=102691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:51:14,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9345, 4.4807, 4.5604, 3.3623, 3.8961, 4.4960, 4.0987, 2.9251], device='cuda:7'), covar=tensor([0.0335, 0.0037, 0.0028, 0.0224, 0.0076, 0.0065, 0.0048, 0.0297], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0070, 0.0070, 0.0125, 0.0078, 0.0088, 0.0076, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:51:27,908 INFO [train.py:904] (7/8) Epoch 11, batch 1200, loss[loss=0.1871, simple_loss=0.2572, pruned_loss=0.05844, over 16754.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2601, pruned_loss=0.0491, over 3312696.31 frames. ], batch size: 83, lr: 6.34e-03, grad_scale: 8.0 2023-04-29 08:52:27,637 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.248e+02 2.693e+02 3.149e+02 5.178e+02, threshold=5.386e+02, percent-clipped=0.0 2023-04-29 08:52:39,082 INFO [train.py:904] (7/8) Epoch 11, batch 1250, loss[loss=0.1617, simple_loss=0.2613, pruned_loss=0.03107, over 17126.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2609, pruned_loss=0.04971, over 3309348.23 frames. ], batch size: 48, lr: 6.34e-03, grad_scale: 8.0 2023-04-29 08:52:49,867 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-04-29 08:53:18,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6569, 3.7172, 3.9402, 1.9932, 4.0558, 4.0706, 3.1300, 2.8933], device='cuda:7'), covar=tensor([0.0724, 0.0165, 0.0180, 0.1132, 0.0075, 0.0147, 0.0398, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0097, 0.0088, 0.0136, 0.0069, 0.0101, 0.0121, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 08:53:31,119 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2293, 4.6322, 4.7065, 3.5208, 4.0507, 4.6736, 4.1517, 3.0782], device='cuda:7'), covar=tensor([0.0321, 0.0039, 0.0025, 0.0223, 0.0065, 0.0060, 0.0051, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0070, 0.0070, 0.0125, 0.0078, 0.0088, 0.0076, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:53:49,409 INFO [train.py:904] (7/8) Epoch 11, batch 1300, loss[loss=0.1731, simple_loss=0.2692, pruned_loss=0.03844, over 17132.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2608, pruned_loss=0.04957, over 3311187.57 frames. ], batch size: 47, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:54:17,578 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 08:54:46,487 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.481e+02 2.988e+02 3.715e+02 8.832e+02, threshold=5.975e+02, percent-clipped=5.0 2023-04-29 08:54:58,065 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102851.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:54:58,773 INFO [train.py:904] (7/8) Epoch 11, batch 1350, loss[loss=0.1771, simple_loss=0.2591, pruned_loss=0.04759, over 17211.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2605, pruned_loss=0.04924, over 3313164.25 frames. ], batch size: 44, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:55:45,239 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102887.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:56:05,630 INFO [train.py:904] (7/8) Epoch 11, batch 1400, loss[loss=0.1472, simple_loss=0.2331, pruned_loss=0.03066, over 16773.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.261, pruned_loss=0.04924, over 3319169.68 frames. ], batch size: 39, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:56:20,779 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102912.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:56:53,961 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 08:57:05,104 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.762e+02 2.362e+02 2.982e+02 3.517e+02 5.508e+02, threshold=5.963e+02, percent-clipped=1.0 2023-04-29 08:57:10,230 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102948.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 08:57:15,719 INFO [train.py:904] (7/8) Epoch 11, batch 1450, loss[loss=0.1758, simple_loss=0.2654, pruned_loss=0.04309, over 17108.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2598, pruned_loss=0.04903, over 3310788.16 frames. ], batch size: 49, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:57:20,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8404, 4.7422, 4.7474, 4.4653, 4.3798, 4.8077, 4.6537, 4.5102], device='cuda:7'), covar=tensor([0.0656, 0.0667, 0.0273, 0.0261, 0.0864, 0.0438, 0.0378, 0.0634], device='cuda:7'), in_proj_covar=tensor([0.0258, 0.0323, 0.0301, 0.0280, 0.0325, 0.0319, 0.0207, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:57:23,042 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3119, 4.3009, 4.4734, 4.3848, 4.3511, 4.9603, 4.5480, 4.2463], device='cuda:7'), covar=tensor([0.1766, 0.1974, 0.1842, 0.2203, 0.3069, 0.1228, 0.1553, 0.2756], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0491, 0.0533, 0.0425, 0.0565, 0.0559, 0.0422, 0.0575], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 08:57:35,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1272, 1.9972, 2.5978, 3.0672, 2.8244, 3.3578, 2.3766, 3.4310], device='cuda:7'), covar=tensor([0.0137, 0.0333, 0.0221, 0.0189, 0.0206, 0.0129, 0.0304, 0.0087], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0173, 0.0157, 0.0157, 0.0168, 0.0123, 0.0170, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 08:58:25,009 INFO [train.py:904] (7/8) Epoch 11, batch 1500, loss[loss=0.2222, simple_loss=0.3026, pruned_loss=0.0709, over 16787.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2607, pruned_loss=0.04968, over 3316464.17 frames. ], batch size: 76, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:59:24,565 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.416e+02 2.850e+02 3.321e+02 6.158e+02, threshold=5.699e+02, percent-clipped=1.0 2023-04-29 08:59:35,065 INFO [train.py:904] (7/8) Epoch 11, batch 1550, loss[loss=0.2024, simple_loss=0.278, pruned_loss=0.06338, over 11912.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.262, pruned_loss=0.0506, over 3314456.59 frames. ], batch size: 246, lr: 6.33e-03, grad_scale: 8.0 2023-04-29 08:59:44,354 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 09:00:40,055 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103098.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:00:45,013 INFO [train.py:904] (7/8) Epoch 11, batch 1600, loss[loss=0.1933, simple_loss=0.2651, pruned_loss=0.0607, over 16771.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.264, pruned_loss=0.0514, over 3308171.06 frames. ], batch size: 83, lr: 6.32e-03, grad_scale: 8.0 2023-04-29 09:00:57,401 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6114, 2.6459, 2.0127, 2.1685, 2.9614, 2.7004, 3.5263, 3.2965], device='cuda:7'), covar=tensor([0.0084, 0.0348, 0.0494, 0.0446, 0.0233, 0.0320, 0.0174, 0.0201], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0208, 0.0205, 0.0205, 0.0208, 0.0208, 0.0213, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:01:25,966 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:01:42,629 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 2.417e+02 3.013e+02 3.629e+02 9.117e+02, threshold=6.027e+02, percent-clipped=4.0 2023-04-29 09:01:53,532 INFO [train.py:904] (7/8) Epoch 11, batch 1650, loss[loss=0.2472, simple_loss=0.3233, pruned_loss=0.08559, over 12235.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2655, pruned_loss=0.05177, over 3304210.27 frames. ], batch size: 247, lr: 6.32e-03, grad_scale: 8.0 2023-04-29 09:02:00,686 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 09:02:03,376 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103159.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:02:13,722 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-29 09:02:24,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1296, 1.9108, 2.6445, 3.1065, 2.7669, 3.5408, 2.2475, 3.5172], device='cuda:7'), covar=tensor([0.0158, 0.0356, 0.0213, 0.0190, 0.0210, 0.0109, 0.0327, 0.0094], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0173, 0.0159, 0.0159, 0.0170, 0.0123, 0.0171, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 09:02:50,039 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103193.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:02:57,206 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9826, 3.5296, 2.9910, 5.3125, 4.5469, 4.6715, 1.7971, 3.6911], device='cuda:7'), covar=tensor([0.1221, 0.0575, 0.1020, 0.0139, 0.0254, 0.0346, 0.1418, 0.0593], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0159, 0.0181, 0.0144, 0.0196, 0.0211, 0.0181, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 09:03:02,608 INFO [train.py:904] (7/8) Epoch 11, batch 1700, loss[loss=0.2004, simple_loss=0.2778, pruned_loss=0.06146, over 16784.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2678, pruned_loss=0.0529, over 3311408.22 frames. ], batch size: 83, lr: 6.32e-03, grad_scale: 8.0 2023-04-29 09:03:10,665 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103207.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:04:01,771 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103243.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:04:02,561 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 2.478e+02 3.058e+02 3.640e+02 7.869e+02, threshold=6.116e+02, percent-clipped=2.0 2023-04-29 09:04:13,826 INFO [train.py:904] (7/8) Epoch 11, batch 1750, loss[loss=0.2375, simple_loss=0.2995, pruned_loss=0.08777, over 16469.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2694, pruned_loss=0.05334, over 3305063.27 frames. ], batch size: 146, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:04:19,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 09:05:22,490 INFO [train.py:904] (7/8) Epoch 11, batch 1800, loss[loss=0.1922, simple_loss=0.2755, pruned_loss=0.05448, over 16474.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2705, pruned_loss=0.05331, over 3311120.38 frames. ], batch size: 75, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:05:33,660 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4368, 4.3200, 4.7300, 2.5145, 5.0293, 4.9837, 3.5675, 3.9009], device='cuda:7'), covar=tensor([0.0539, 0.0146, 0.0189, 0.0996, 0.0042, 0.0113, 0.0276, 0.0307], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0099, 0.0090, 0.0139, 0.0070, 0.0104, 0.0121, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 09:05:37,566 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-29 09:06:21,371 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.500e+02 2.964e+02 3.680e+02 1.127e+03, threshold=5.929e+02, percent-clipped=5.0 2023-04-29 09:06:31,969 INFO [train.py:904] (7/8) Epoch 11, batch 1850, loss[loss=0.168, simple_loss=0.2537, pruned_loss=0.04118, over 17245.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2715, pruned_loss=0.05338, over 3311838.26 frames. ], batch size: 45, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:06:51,484 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103366.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 09:07:24,721 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-04-29 09:07:39,139 INFO [train.py:904] (7/8) Epoch 11, batch 1900, loss[loss=0.186, simple_loss=0.2615, pruned_loss=0.05525, over 16806.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2709, pruned_loss=0.05288, over 3317869.16 frames. ], batch size: 102, lr: 6.32e-03, grad_scale: 4.0 2023-04-29 09:08:16,846 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103427.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 09:08:40,925 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.233e+02 2.712e+02 3.149e+02 7.257e+02, threshold=5.425e+02, percent-clipped=1.0 2023-04-29 09:08:51,896 INFO [train.py:904] (7/8) Epoch 11, batch 1950, loss[loss=0.1861, simple_loss=0.2796, pruned_loss=0.04633, over 17035.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2698, pruned_loss=0.05154, over 3319708.53 frames. ], batch size: 53, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:08:55,136 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103454.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:09:41,438 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103488.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:10:00,151 INFO [train.py:904] (7/8) Epoch 11, batch 2000, loss[loss=0.212, simple_loss=0.2865, pruned_loss=0.06881, over 16466.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2706, pruned_loss=0.05132, over 3322785.99 frames. ], batch size: 146, lr: 6.31e-03, grad_scale: 8.0 2023-04-29 09:10:07,698 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103507.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:10:27,081 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-04-29 09:10:58,966 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103543.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:11:00,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.374e+02 2.796e+02 3.607e+02 7.066e+02, threshold=5.592e+02, percent-clipped=4.0 2023-04-29 09:11:10,943 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 09:11:11,330 INFO [train.py:904] (7/8) Epoch 11, batch 2050, loss[loss=0.1519, simple_loss=0.237, pruned_loss=0.03343, over 16842.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2701, pruned_loss=0.05144, over 3318983.82 frames. ], batch size: 39, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:11:16,491 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103555.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:12:05,794 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103591.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:12:21,606 INFO [train.py:904] (7/8) Epoch 11, batch 2100, loss[loss=0.2066, simple_loss=0.2823, pruned_loss=0.06547, over 16544.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2719, pruned_loss=0.05332, over 3311148.65 frames. ], batch size: 75, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:13:15,930 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 09:13:22,848 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.732e+02 2.751e+02 3.128e+02 3.657e+02 9.004e+02, threshold=6.256e+02, percent-clipped=1.0 2023-04-29 09:13:31,363 INFO [train.py:904] (7/8) Epoch 11, batch 2150, loss[loss=0.2128, simple_loss=0.2786, pruned_loss=0.07348, over 16347.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2724, pruned_loss=0.05351, over 3305458.45 frames. ], batch size: 165, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:13:40,596 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 09:13:52,966 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-29 09:14:42,114 INFO [train.py:904] (7/8) Epoch 11, batch 2200, loss[loss=0.2046, simple_loss=0.2757, pruned_loss=0.06675, over 16758.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2727, pruned_loss=0.05339, over 3310625.34 frames. ], batch size: 124, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:15:11,430 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103722.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:15:35,676 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:15:44,095 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.339e+02 2.759e+02 3.363e+02 7.654e+02, threshold=5.518e+02, percent-clipped=1.0 2023-04-29 09:15:51,105 INFO [train.py:904] (7/8) Epoch 11, batch 2250, loss[loss=0.1998, simple_loss=0.2882, pruned_loss=0.05569, over 17041.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2724, pruned_loss=0.05371, over 3307379.37 frames. ], batch size: 50, lr: 6.31e-03, grad_scale: 4.0 2023-04-29 09:15:54,348 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103754.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:16:12,474 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103765.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:16:44,277 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103788.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:01,647 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103801.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:02,364 INFO [train.py:904] (7/8) Epoch 11, batch 2300, loss[loss=0.1488, simple_loss=0.2345, pruned_loss=0.03159, over 16869.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2712, pruned_loss=0.05277, over 3314970.51 frames. ], batch size: 42, lr: 6.30e-03, grad_scale: 4.0 2023-04-29 09:17:02,657 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103802.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:08,406 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3082, 5.6360, 5.3765, 5.4578, 5.0551, 4.9316, 5.0899, 5.7330], device='cuda:7'), covar=tensor([0.1153, 0.0843, 0.0967, 0.0641, 0.0823, 0.0705, 0.1056, 0.0931], device='cuda:7'), in_proj_covar=tensor([0.0544, 0.0687, 0.0565, 0.0479, 0.0429, 0.0442, 0.0572, 0.0521], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:17:35,801 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103826.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:17:39,923 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-29 09:17:45,722 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 09:17:48,571 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=103836.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:18:02,534 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.457e+02 2.837e+02 3.610e+02 7.865e+02, threshold=5.674e+02, percent-clipped=5.0 2023-04-29 09:18:11,660 INFO [train.py:904] (7/8) Epoch 11, batch 2350, loss[loss=0.1877, simple_loss=0.2778, pruned_loss=0.04882, over 17054.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2718, pruned_loss=0.05269, over 3326705.01 frames. ], batch size: 55, lr: 6.30e-03, grad_scale: 4.0 2023-04-29 09:18:45,314 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103876.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:19:19,661 INFO [train.py:904] (7/8) Epoch 11, batch 2400, loss[loss=0.1709, simple_loss=0.2679, pruned_loss=0.0369, over 16660.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2721, pruned_loss=0.0524, over 3329575.07 frames. ], batch size: 57, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:20:09,877 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103937.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:20:20,789 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.318e+02 2.704e+02 3.589e+02 9.063e+02, threshold=5.409e+02, percent-clipped=3.0 2023-04-29 09:20:28,635 INFO [train.py:904] (7/8) Epoch 11, batch 2450, loss[loss=0.1907, simple_loss=0.2614, pruned_loss=0.05999, over 16832.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2719, pruned_loss=0.0519, over 3336889.97 frames. ], batch size: 83, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:20:33,720 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8028, 5.2081, 5.3933, 5.1723, 5.1538, 5.8065, 5.3900, 5.1298], device='cuda:7'), covar=tensor([0.1145, 0.1873, 0.1690, 0.2030, 0.2652, 0.0919, 0.1197, 0.2198], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0501, 0.0545, 0.0439, 0.0576, 0.0571, 0.0428, 0.0588], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:21:22,382 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103990.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:21:32,660 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103997.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:21:42,817 INFO [train.py:904] (7/8) Epoch 11, batch 2500, loss[loss=0.2125, simple_loss=0.2971, pruned_loss=0.06397, over 16447.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.272, pruned_loss=0.0517, over 3338220.28 frames. ], batch size: 68, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:22:11,694 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104022.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:22:43,747 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.442e+02 2.893e+02 3.416e+02 6.492e+02, threshold=5.787e+02, percent-clipped=3.0 2023-04-29 09:22:47,626 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=104049.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:22:49,940 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104051.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:22:50,609 INFO [train.py:904] (7/8) Epoch 11, batch 2550, loss[loss=0.1675, simple_loss=0.2604, pruned_loss=0.0373, over 17119.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2722, pruned_loss=0.05188, over 3331608.04 frames. ], batch size: 47, lr: 6.30e-03, grad_scale: 8.0 2023-04-29 09:23:00,518 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104058.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:23:15,997 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104070.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 09:23:27,453 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2023-04-29 09:23:51,817 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104096.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:23:59,574 INFO [train.py:904] (7/8) Epoch 11, batch 2600, loss[loss=0.2242, simple_loss=0.2912, pruned_loss=0.07859, over 16910.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2729, pruned_loss=0.05246, over 3322714.85 frames. ], batch size: 116, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:24:12,043 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104110.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:24:26,718 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104121.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:25:01,162 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.655e+02 3.049e+02 3.595e+02 6.269e+02, threshold=6.098e+02, percent-clipped=2.0 2023-04-29 09:25:10,402 INFO [train.py:904] (7/8) Epoch 11, batch 2650, loss[loss=0.1959, simple_loss=0.2857, pruned_loss=0.05301, over 16687.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.274, pruned_loss=0.05233, over 3321363.42 frames. ], batch size: 57, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:26:18,844 INFO [train.py:904] (7/8) Epoch 11, batch 2700, loss[loss=0.1776, simple_loss=0.268, pruned_loss=0.04365, over 17253.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2742, pruned_loss=0.05176, over 3324339.60 frames. ], batch size: 45, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:26:41,393 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9577, 1.7466, 2.3836, 2.8257, 2.7956, 2.9622, 1.9323, 3.0246], device='cuda:7'), covar=tensor([0.0122, 0.0345, 0.0235, 0.0164, 0.0171, 0.0137, 0.0356, 0.0087], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0173, 0.0158, 0.0161, 0.0170, 0.0126, 0.0172, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 09:27:00,800 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104232.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 09:27:19,009 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.677e+02 2.334e+02 2.747e+02 3.753e+02 9.915e+02, threshold=5.495e+02, percent-clipped=4.0 2023-04-29 09:27:27,257 INFO [train.py:904] (7/8) Epoch 11, batch 2750, loss[loss=0.1839, simple_loss=0.2693, pruned_loss=0.04924, over 16801.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2745, pruned_loss=0.05128, over 3334242.65 frames. ], batch size: 83, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:28:01,853 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-29 09:28:09,400 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3740, 2.2788, 1.8160, 2.0435, 2.6397, 2.4915, 2.7637, 2.7961], device='cuda:7'), covar=tensor([0.0130, 0.0285, 0.0360, 0.0325, 0.0155, 0.0220, 0.0166, 0.0179], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0210, 0.0202, 0.0204, 0.0209, 0.0208, 0.0217, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:28:36,581 INFO [train.py:904] (7/8) Epoch 11, batch 2800, loss[loss=0.1797, simple_loss=0.2727, pruned_loss=0.04332, over 16039.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2746, pruned_loss=0.05196, over 3330060.92 frames. ], batch size: 35, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:29:37,341 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.406e+02 2.849e+02 3.526e+02 7.229e+02, threshold=5.698e+02, percent-clipped=5.0 2023-04-29 09:29:37,666 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104346.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:29:44,359 INFO [train.py:904] (7/8) Epoch 11, batch 2850, loss[loss=0.1856, simple_loss=0.2726, pruned_loss=0.04929, over 16139.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2738, pruned_loss=0.05215, over 3322622.03 frames. ], batch size: 35, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:29:45,784 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104353.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:30:44,765 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104396.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:30:52,883 INFO [train.py:904] (7/8) Epoch 11, batch 2900, loss[loss=0.2016, simple_loss=0.2716, pruned_loss=0.06579, over 16887.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2727, pruned_loss=0.05258, over 3320415.43 frames. ], batch size: 116, lr: 6.29e-03, grad_scale: 8.0 2023-04-29 09:30:58,432 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104405.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:31:05,865 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8540, 4.2021, 4.2913, 2.8767, 3.6413, 4.1651, 3.8977, 2.7238], device='cuda:7'), covar=tensor([0.0345, 0.0046, 0.0031, 0.0286, 0.0090, 0.0076, 0.0054, 0.0312], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0071, 0.0071, 0.0126, 0.0079, 0.0090, 0.0077, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:31:20,041 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104421.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:31:49,218 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-29 09:31:53,048 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104444.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:31:55,137 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.403e+02 2.933e+02 3.429e+02 6.056e+02, threshold=5.866e+02, percent-clipped=1.0 2023-04-29 09:32:04,095 INFO [train.py:904] (7/8) Epoch 11, batch 2950, loss[loss=0.3102, simple_loss=0.3547, pruned_loss=0.1329, over 11988.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2725, pruned_loss=0.05295, over 3320191.38 frames. ], batch size: 248, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:32:28,559 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104469.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:33:12,391 INFO [train.py:904] (7/8) Epoch 11, batch 3000, loss[loss=0.2052, simple_loss=0.283, pruned_loss=0.06374, over 15613.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2722, pruned_loss=0.05324, over 3321687.19 frames. ], batch size: 191, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:33:12,391 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 09:33:22,063 INFO [train.py:938] (7/8) Epoch 11, validation: loss=0.1413, simple_loss=0.2475, pruned_loss=0.01754, over 944034.00 frames. 2023-04-29 09:33:22,063 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 09:33:36,255 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-29 09:34:04,316 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104532.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 09:34:20,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.544e+02 3.030e+02 3.628e+02 6.112e+02, threshold=6.060e+02, percent-clipped=1.0 2023-04-29 09:34:22,222 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.78 vs. limit=5.0 2023-04-29 09:34:30,338 INFO [train.py:904] (7/8) Epoch 11, batch 3050, loss[loss=0.1737, simple_loss=0.2485, pruned_loss=0.04947, over 16899.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2711, pruned_loss=0.05267, over 3332086.23 frames. ], batch size: 90, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:34:39,667 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2143, 4.1864, 4.1025, 3.9248, 3.8458, 4.1931, 3.8155, 3.9487], device='cuda:7'), covar=tensor([0.0584, 0.0490, 0.0285, 0.0243, 0.0713, 0.0370, 0.0852, 0.0547], device='cuda:7'), in_proj_covar=tensor([0.0264, 0.0332, 0.0312, 0.0290, 0.0335, 0.0330, 0.0213, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:35:07,353 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104580.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:35:37,120 INFO [train.py:904] (7/8) Epoch 11, batch 3100, loss[loss=0.2219, simple_loss=0.2826, pruned_loss=0.0806, over 16837.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2705, pruned_loss=0.05238, over 3327841.76 frames. ], batch size: 96, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:36:39,260 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.673e+02 2.465e+02 3.037e+02 3.411e+02 8.178e+02, threshold=6.075e+02, percent-clipped=4.0 2023-04-29 09:36:39,610 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104646.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:36:47,520 INFO [train.py:904] (7/8) Epoch 11, batch 3150, loss[loss=0.1598, simple_loss=0.2519, pruned_loss=0.03387, over 16809.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2708, pruned_loss=0.05269, over 3322108.61 frames. ], batch size: 42, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:36:49,892 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104653.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:37:39,187 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9512, 4.9849, 5.4979, 5.4835, 5.4814, 5.0725, 5.0329, 4.7808], device='cuda:7'), covar=tensor([0.0316, 0.0452, 0.0329, 0.0418, 0.0469, 0.0346, 0.0967, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0343, 0.0358, 0.0356, 0.0336, 0.0403, 0.0373, 0.0484, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 09:37:46,971 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104694.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:37:51,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5427, 2.4670, 2.0214, 2.3195, 2.8233, 2.6090, 3.4131, 3.1549], device='cuda:7'), covar=tensor([0.0074, 0.0314, 0.0390, 0.0367, 0.0211, 0.0291, 0.0174, 0.0175], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0209, 0.0203, 0.0204, 0.0209, 0.0206, 0.0218, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:37:56,591 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104701.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:37:57,455 INFO [train.py:904] (7/8) Epoch 11, batch 3200, loss[loss=0.2009, simple_loss=0.2703, pruned_loss=0.06576, over 16747.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2694, pruned_loss=0.05175, over 3323844.49 frames. ], batch size: 124, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:38:01,367 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104705.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:38:42,316 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2669, 5.7193, 5.8651, 5.5395, 5.5345, 6.1402, 5.8099, 5.5385], device='cuda:7'), covar=tensor([0.0751, 0.1703, 0.1708, 0.1986, 0.3003, 0.1167, 0.1269, 0.2446], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0508, 0.0548, 0.0438, 0.0585, 0.0576, 0.0433, 0.0594], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:38:57,881 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.472e+02 2.965e+02 3.891e+02 6.936e+02, threshold=5.930e+02, percent-clipped=3.0 2023-04-29 09:39:06,606 INFO [train.py:904] (7/8) Epoch 11, batch 3250, loss[loss=0.1817, simple_loss=0.2709, pruned_loss=0.04626, over 16708.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2695, pruned_loss=0.05259, over 3320874.12 frames. ], batch size: 62, lr: 6.28e-03, grad_scale: 8.0 2023-04-29 09:39:08,059 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=104753.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:39:09,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1622, 5.6596, 5.9093, 5.6363, 5.6466, 6.2078, 5.7943, 5.4735], device='cuda:7'), covar=tensor([0.0815, 0.1619, 0.1741, 0.1817, 0.2900, 0.0950, 0.1360, 0.2334], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0510, 0.0550, 0.0440, 0.0588, 0.0578, 0.0434, 0.0597], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:39:17,705 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 09:39:43,494 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2023-04-29 09:39:56,469 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3048, 5.3330, 5.1544, 4.7646, 4.7551, 5.2924, 5.2628, 4.8329], device='cuda:7'), covar=tensor([0.0698, 0.0401, 0.0281, 0.0288, 0.1161, 0.0339, 0.0267, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0266, 0.0336, 0.0315, 0.0293, 0.0336, 0.0333, 0.0214, 0.0364], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:40:15,732 INFO [train.py:904] (7/8) Epoch 11, batch 3300, loss[loss=0.1878, simple_loss=0.2676, pruned_loss=0.054, over 16773.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2701, pruned_loss=0.0525, over 3313913.74 frames. ], batch size: 102, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:41:16,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.335e+02 2.858e+02 3.500e+02 6.085e+02, threshold=5.716e+02, percent-clipped=1.0 2023-04-29 09:41:16,718 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8194, 4.9152, 5.3861, 5.4200, 5.3662, 4.9668, 4.9718, 4.6802], device='cuda:7'), covar=tensor([0.0361, 0.0447, 0.0409, 0.0374, 0.0422, 0.0361, 0.0818, 0.0407], device='cuda:7'), in_proj_covar=tensor([0.0342, 0.0357, 0.0357, 0.0335, 0.0404, 0.0373, 0.0478, 0.0301], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 09:41:24,650 INFO [train.py:904] (7/8) Epoch 11, batch 3350, loss[loss=0.1835, simple_loss=0.2716, pruned_loss=0.04765, over 17065.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2704, pruned_loss=0.05193, over 3309357.77 frames. ], batch size: 53, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:42:23,835 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 09:42:33,971 INFO [train.py:904] (7/8) Epoch 11, batch 3400, loss[loss=0.21, simple_loss=0.2813, pruned_loss=0.06931, over 16877.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.27, pruned_loss=0.05178, over 3315427.97 frames. ], batch size: 116, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:43:18,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0606, 5.5653, 5.8199, 5.5258, 5.5329, 6.1232, 5.7122, 5.4463], device='cuda:7'), covar=tensor([0.0814, 0.1733, 0.1796, 0.1657, 0.2628, 0.0939, 0.1243, 0.2194], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0506, 0.0546, 0.0436, 0.0581, 0.0576, 0.0432, 0.0593], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:43:33,849 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.318e+02 2.811e+02 3.324e+02 7.497e+02, threshold=5.622e+02, percent-clipped=1.0 2023-04-29 09:43:41,666 INFO [train.py:904] (7/8) Epoch 11, batch 3450, loss[loss=0.1889, simple_loss=0.2711, pruned_loss=0.05342, over 16529.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2682, pruned_loss=0.05144, over 3320913.09 frames. ], batch size: 68, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:44:06,365 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7316, 2.2777, 2.3801, 4.4342, 2.1425, 2.8679, 2.3669, 2.5538], device='cuda:7'), covar=tensor([0.0865, 0.3295, 0.2136, 0.0354, 0.3673, 0.1935, 0.2858, 0.2897], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0393, 0.0331, 0.0327, 0.0414, 0.0455, 0.0357, 0.0465], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:44:50,185 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2704, 5.3333, 5.1318, 4.7322, 4.5048, 5.2567, 5.2187, 4.7117], device='cuda:7'), covar=tensor([0.0680, 0.0465, 0.0351, 0.0346, 0.1352, 0.0395, 0.0279, 0.0762], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0339, 0.0315, 0.0294, 0.0340, 0.0335, 0.0216, 0.0366], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:44:52,762 INFO [train.py:904] (7/8) Epoch 11, batch 3500, loss[loss=0.1789, simple_loss=0.273, pruned_loss=0.04234, over 17014.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2674, pruned_loss=0.05115, over 3307577.30 frames. ], batch size: 50, lr: 6.27e-03, grad_scale: 8.0 2023-04-29 09:45:55,146 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.297e+02 2.659e+02 3.238e+02 6.959e+02, threshold=5.317e+02, percent-clipped=2.0 2023-04-29 09:46:03,261 INFO [train.py:904] (7/8) Epoch 11, batch 3550, loss[loss=0.2083, simple_loss=0.2798, pruned_loss=0.06843, over 11869.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2656, pruned_loss=0.05039, over 3306264.46 frames. ], batch size: 247, lr: 6.27e-03, grad_scale: 4.0 2023-04-29 09:46:15,244 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4386, 5.9275, 5.6867, 5.7391, 5.2754, 5.2475, 5.3799, 6.0750], device='cuda:7'), covar=tensor([0.1241, 0.0938, 0.0883, 0.0682, 0.0827, 0.0645, 0.0945, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0568, 0.0708, 0.0584, 0.0494, 0.0445, 0.0454, 0.0590, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:46:29,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1100, 5.5492, 5.7260, 5.4552, 5.5233, 6.0945, 5.6762, 5.3692], device='cuda:7'), covar=tensor([0.0826, 0.1664, 0.1939, 0.1910, 0.2688, 0.1068, 0.1380, 0.2525], device='cuda:7'), in_proj_covar=tensor([0.0357, 0.0504, 0.0545, 0.0437, 0.0581, 0.0575, 0.0432, 0.0593], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:47:12,604 INFO [train.py:904] (7/8) Epoch 11, batch 3600, loss[loss=0.2185, simple_loss=0.2842, pruned_loss=0.07645, over 11421.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2651, pruned_loss=0.05064, over 3288171.38 frames. ], batch size: 246, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:48:03,735 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6585, 4.6023, 4.5377, 4.2541, 4.2403, 4.6283, 4.3425, 4.3347], device='cuda:7'), covar=tensor([0.0647, 0.0658, 0.0298, 0.0321, 0.0835, 0.0495, 0.0544, 0.0617], device='cuda:7'), in_proj_covar=tensor([0.0266, 0.0336, 0.0313, 0.0292, 0.0337, 0.0333, 0.0215, 0.0365], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:48:17,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.381e+02 2.999e+02 3.559e+02 5.389e+02, threshold=5.997e+02, percent-clipped=2.0 2023-04-29 09:48:24,362 INFO [train.py:904] (7/8) Epoch 11, batch 3650, loss[loss=0.1994, simple_loss=0.2659, pruned_loss=0.06649, over 16213.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2633, pruned_loss=0.05038, over 3285936.23 frames. ], batch size: 165, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:49:37,408 INFO [train.py:904] (7/8) Epoch 11, batch 3700, loss[loss=0.2099, simple_loss=0.2773, pruned_loss=0.07128, over 16773.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2621, pruned_loss=0.05186, over 3292537.59 frames. ], batch size: 102, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:50:41,017 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.627e+02 2.380e+02 2.864e+02 3.354e+02 5.031e+02, threshold=5.728e+02, percent-clipped=0.0 2023-04-29 09:50:48,715 INFO [train.py:904] (7/8) Epoch 11, batch 3750, loss[loss=0.2269, simple_loss=0.301, pruned_loss=0.07643, over 11822.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2641, pruned_loss=0.05363, over 3288877.35 frames. ], batch size: 247, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:51:57,090 INFO [train.py:904] (7/8) Epoch 11, batch 3800, loss[loss=0.1846, simple_loss=0.2629, pruned_loss=0.05313, over 16469.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2651, pruned_loss=0.05452, over 3283421.29 frames. ], batch size: 68, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:51:59,982 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-29 09:53:02,334 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.313e+02 2.608e+02 3.308e+02 8.159e+02, threshold=5.217e+02, percent-clipped=3.0 2023-04-29 09:53:08,979 INFO [train.py:904] (7/8) Epoch 11, batch 3850, loss[loss=0.1636, simple_loss=0.2544, pruned_loss=0.03641, over 16605.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2647, pruned_loss=0.0546, over 3279213.64 frames. ], batch size: 35, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:53:09,343 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0148, 3.8464, 3.9045, 4.2098, 4.2557, 3.9151, 4.0879, 4.2697], device='cuda:7'), covar=tensor([0.1260, 0.1065, 0.1675, 0.0740, 0.0711, 0.1404, 0.1564, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0542, 0.0678, 0.0833, 0.0694, 0.0526, 0.0534, 0.0534, 0.0616], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:53:20,611 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4183, 3.6146, 3.8478, 2.6425, 3.4104, 3.8702, 3.5650, 2.2838], device='cuda:7'), covar=tensor([0.0397, 0.0083, 0.0034, 0.0292, 0.0073, 0.0069, 0.0062, 0.0335], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0070, 0.0070, 0.0124, 0.0078, 0.0089, 0.0077, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:53:34,444 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5957, 1.7352, 2.2159, 2.5373, 2.5612, 2.4121, 1.7237, 2.7679], device='cuda:7'), covar=tensor([0.0122, 0.0346, 0.0226, 0.0198, 0.0187, 0.0227, 0.0353, 0.0078], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0172, 0.0158, 0.0160, 0.0168, 0.0125, 0.0169, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 09:54:00,703 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105388.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:54:19,884 INFO [train.py:904] (7/8) Epoch 11, batch 3900, loss[loss=0.1713, simple_loss=0.2494, pruned_loss=0.04661, over 16336.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2643, pruned_loss=0.05509, over 3276313.40 frames. ], batch size: 68, lr: 6.26e-03, grad_scale: 8.0 2023-04-29 09:55:00,196 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-29 09:55:06,110 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 09:55:25,283 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.279e+02 2.712e+02 3.395e+02 5.874e+02, threshold=5.424e+02, percent-clipped=4.0 2023-04-29 09:55:28,939 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105449.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 09:55:31,849 INFO [train.py:904] (7/8) Epoch 11, batch 3950, loss[loss=0.1739, simple_loss=0.2483, pruned_loss=0.0498, over 16456.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2642, pruned_loss=0.056, over 3274346.24 frames. ], batch size: 146, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:55:50,459 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5170, 3.6031, 1.9524, 3.7332, 2.6856, 3.7131, 2.1258, 2.8248], device='cuda:7'), covar=tensor([0.0164, 0.0287, 0.1379, 0.0157, 0.0635, 0.0526, 0.1277, 0.0538], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0167, 0.0186, 0.0133, 0.0166, 0.0211, 0.0193, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 09:56:21,781 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105486.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:56:44,044 INFO [train.py:904] (7/8) Epoch 11, batch 4000, loss[loss=0.2259, simple_loss=0.2918, pruned_loss=0.08001, over 16920.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2647, pruned_loss=0.05676, over 3277794.93 frames. ], batch size: 109, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:56:44,509 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105502.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:57:04,757 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2114, 2.5033, 1.9129, 2.2889, 2.9627, 2.5233, 3.0820, 3.0937], device='cuda:7'), covar=tensor([0.0087, 0.0275, 0.0418, 0.0337, 0.0151, 0.0290, 0.0142, 0.0144], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0205, 0.0200, 0.0200, 0.0204, 0.0202, 0.0212, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 09:57:38,507 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4438, 4.3623, 4.4404, 2.6042, 3.7851, 4.1499, 3.8537, 1.9265], device='cuda:7'), covar=tensor([0.0468, 0.0030, 0.0045, 0.0387, 0.0081, 0.0111, 0.0065, 0.0492], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0070, 0.0070, 0.0124, 0.0078, 0.0089, 0.0077, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 09:57:48,231 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.126e+02 2.510e+02 3.113e+02 5.391e+02, threshold=5.020e+02, percent-clipped=0.0 2023-04-29 09:57:48,858 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105547.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:57:55,347 INFO [train.py:904] (7/8) Epoch 11, batch 4050, loss[loss=0.1789, simple_loss=0.2562, pruned_loss=0.05081, over 16741.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2647, pruned_loss=0.05549, over 3284025.47 frames. ], batch size: 62, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 09:58:11,881 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105563.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 09:59:08,734 INFO [train.py:904] (7/8) Epoch 11, batch 4100, loss[loss=0.1899, simple_loss=0.2766, pruned_loss=0.05156, over 16765.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2658, pruned_loss=0.05479, over 3272674.65 frames. ], batch size: 83, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:00:18,929 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 2.197e+02 2.575e+02 3.230e+02 7.189e+02, threshold=5.150e+02, percent-clipped=4.0 2023-04-29 10:00:26,727 INFO [train.py:904] (7/8) Epoch 11, batch 4150, loss[loss=0.269, simple_loss=0.3363, pruned_loss=0.1009, over 11328.00 frames. ], tot_loss[loss=0.194, simple_loss=0.273, pruned_loss=0.05747, over 3236115.88 frames. ], batch size: 248, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:01:03,998 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2395, 1.9614, 2.7877, 3.2506, 3.0498, 3.7733, 2.1220, 3.5300], device='cuda:7'), covar=tensor([0.0127, 0.0331, 0.0198, 0.0149, 0.0165, 0.0075, 0.0331, 0.0075], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0171, 0.0155, 0.0159, 0.0167, 0.0123, 0.0167, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 10:01:28,595 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-29 10:01:44,703 INFO [train.py:904] (7/8) Epoch 11, batch 4200, loss[loss=0.2143, simple_loss=0.3036, pruned_loss=0.0625, over 16687.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.28, pruned_loss=0.0592, over 3200444.20 frames. ], batch size: 57, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:01:50,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7858, 5.0470, 4.8380, 4.8606, 4.5703, 4.4691, 4.5250, 5.1519], device='cuda:7'), covar=tensor([0.1004, 0.0825, 0.0835, 0.0691, 0.0780, 0.0958, 0.0943, 0.0820], device='cuda:7'), in_proj_covar=tensor([0.0545, 0.0675, 0.0561, 0.0472, 0.0428, 0.0439, 0.0565, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:02:47,017 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6873, 3.6091, 3.5708, 3.9196, 3.9408, 3.5800, 3.9263, 4.0088], device='cuda:7'), covar=tensor([0.1448, 0.1056, 0.1694, 0.0735, 0.0706, 0.2067, 0.0911, 0.0685], device='cuda:7'), in_proj_covar=tensor([0.0516, 0.0649, 0.0788, 0.0663, 0.0503, 0.0509, 0.0511, 0.0589], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:02:49,770 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105744.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 10:02:50,182 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-29 10:02:53,625 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.308e+02 2.787e+02 3.444e+02 5.705e+02, threshold=5.573e+02, percent-clipped=4.0 2023-04-29 10:02:55,440 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0666, 4.3296, 2.7763, 5.0160, 3.3212, 4.9159, 2.7567, 3.5010], device='cuda:7'), covar=tensor([0.0175, 0.0221, 0.1273, 0.0144, 0.0627, 0.0254, 0.1337, 0.0509], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0164, 0.0185, 0.0129, 0.0165, 0.0207, 0.0192, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 10:02:59,809 INFO [train.py:904] (7/8) Epoch 11, batch 4250, loss[loss=0.198, simple_loss=0.286, pruned_loss=0.05502, over 15320.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2831, pruned_loss=0.05841, over 3207592.21 frames. ], batch size: 190, lr: 6.25e-03, grad_scale: 8.0 2023-04-29 10:03:13,422 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9394, 4.1363, 4.4307, 1.8895, 4.7829, 4.9112, 3.3350, 3.5077], device='cuda:7'), covar=tensor([0.0714, 0.0197, 0.0188, 0.1268, 0.0051, 0.0056, 0.0362, 0.0396], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0101, 0.0090, 0.0139, 0.0071, 0.0104, 0.0122, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 10:04:00,218 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105792.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:04:08,594 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105799.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:04:12,705 INFO [train.py:904] (7/8) Epoch 11, batch 4300, loss[loss=0.2053, simple_loss=0.296, pruned_loss=0.05734, over 16652.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2844, pruned_loss=0.05754, over 3200150.78 frames. ], batch size: 57, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:04:21,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-29 10:05:11,676 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105842.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:05:17,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.245e+02 2.727e+02 3.248e+02 5.954e+02, threshold=5.453e+02, percent-clipped=2.0 2023-04-29 10:05:25,483 INFO [train.py:904] (7/8) Epoch 11, batch 4350, loss[loss=0.2195, simple_loss=0.2972, pruned_loss=0.0709, over 17051.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2881, pruned_loss=0.05907, over 3202983.43 frames. ], batch size: 53, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:05:27,957 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105853.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:05:34,705 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105858.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:05:37,358 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105860.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:06:38,345 INFO [train.py:904] (7/8) Epoch 11, batch 4400, loss[loss=0.2095, simple_loss=0.3, pruned_loss=0.05946, over 16464.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2903, pruned_loss=0.06058, over 3199147.61 frames. ], batch size: 35, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:06:57,903 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5251, 4.3123, 4.5524, 4.7195, 4.8449, 4.3335, 4.8036, 4.8431], device='cuda:7'), covar=tensor([0.1209, 0.0895, 0.1254, 0.0528, 0.0392, 0.0858, 0.0443, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0515, 0.0649, 0.0793, 0.0663, 0.0503, 0.0509, 0.0510, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:07:29,843 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-04-29 10:07:40,709 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.548e+02 2.924e+02 3.370e+02 5.572e+02, threshold=5.847e+02, percent-clipped=1.0 2023-04-29 10:07:48,899 INFO [train.py:904] (7/8) Epoch 11, batch 4450, loss[loss=0.2113, simple_loss=0.3015, pruned_loss=0.06055, over 15363.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2934, pruned_loss=0.06127, over 3216909.62 frames. ], batch size: 190, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:07:55,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5578, 3.5367, 3.4739, 2.8936, 3.3515, 1.9667, 3.0723, 2.7678], device='cuda:7'), covar=tensor([0.0086, 0.0079, 0.0130, 0.0203, 0.0069, 0.2056, 0.0102, 0.0190], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0117, 0.0163, 0.0156, 0.0134, 0.0175, 0.0151, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:09:04,903 INFO [train.py:904] (7/8) Epoch 11, batch 4500, loss[loss=0.2115, simple_loss=0.2873, pruned_loss=0.0679, over 11988.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2937, pruned_loss=0.06181, over 3220449.63 frames. ], batch size: 246, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:10:05,566 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106044.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:10:09,394 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 1.945e+02 2.292e+02 2.586e+02 4.524e+02, threshold=4.584e+02, percent-clipped=0.0 2023-04-29 10:10:17,385 INFO [train.py:904] (7/8) Epoch 11, batch 4550, loss[loss=0.2153, simple_loss=0.3035, pruned_loss=0.06353, over 16720.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2941, pruned_loss=0.0625, over 3221112.12 frames. ], batch size: 134, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:10:35,974 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8075, 3.2239, 3.3312, 1.7958, 2.8322, 1.9620, 3.3396, 3.2377], device='cuda:7'), covar=tensor([0.0199, 0.0610, 0.0529, 0.1986, 0.0767, 0.1113, 0.0575, 0.0850], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0145, 0.0157, 0.0144, 0.0135, 0.0124, 0.0136, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 10:10:53,641 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106077.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:11:15,421 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106092.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:11:29,149 INFO [train.py:904] (7/8) Epoch 11, batch 4600, loss[loss=0.1933, simple_loss=0.2772, pruned_loss=0.0547, over 17007.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2946, pruned_loss=0.06201, over 3241175.39 frames. ], batch size: 41, lr: 6.24e-03, grad_scale: 8.0 2023-04-29 10:11:35,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7155, 4.3195, 4.3145, 2.8468, 3.7921, 4.1757, 3.8927, 2.0774], device='cuda:7'), covar=tensor([0.0372, 0.0026, 0.0036, 0.0310, 0.0068, 0.0089, 0.0070, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0124, 0.0068, 0.0068, 0.0123, 0.0077, 0.0088, 0.0075, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 10:12:22,913 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106138.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:28,052 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106142.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:35,554 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.926e+02 2.251e+02 2.695e+02 4.536e+02, threshold=4.502e+02, percent-clipped=0.0 2023-04-29 10:12:37,707 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106148.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:41,948 INFO [train.py:904] (7/8) Epoch 11, batch 4650, loss[loss=0.1933, simple_loss=0.2786, pruned_loss=0.05396, over 16439.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2933, pruned_loss=0.06181, over 3238338.75 frames. ], batch size: 68, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:12:47,399 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106155.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:12:51,853 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:13:35,797 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 10:13:37,646 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106190.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:13:55,749 INFO [train.py:904] (7/8) Epoch 11, batch 4700, loss[loss=0.2006, simple_loss=0.2828, pruned_loss=0.05916, over 16725.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2904, pruned_loss=0.06054, over 3237012.33 frames. ], batch size: 124, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:14:01,902 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106206.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:14:13,231 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 10:14:28,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4950, 4.5151, 4.3754, 4.0948, 3.9317, 4.4415, 4.2587, 4.0922], device='cuda:7'), covar=tensor([0.0557, 0.0474, 0.0247, 0.0246, 0.0928, 0.0494, 0.0402, 0.0628], device='cuda:7'), in_proj_covar=tensor([0.0235, 0.0296, 0.0278, 0.0259, 0.0299, 0.0295, 0.0192, 0.0322], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:15:01,683 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.142e+02 2.451e+02 2.986e+02 5.737e+02, threshold=4.902e+02, percent-clipped=3.0 2023-04-29 10:15:09,058 INFO [train.py:904] (7/8) Epoch 11, batch 4750, loss[loss=0.1672, simple_loss=0.2525, pruned_loss=0.04094, over 16504.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2863, pruned_loss=0.05857, over 3237331.13 frames. ], batch size: 75, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:15:32,920 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:16:22,065 INFO [train.py:904] (7/8) Epoch 11, batch 4800, loss[loss=0.2157, simple_loss=0.3003, pruned_loss=0.06551, over 16318.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2836, pruned_loss=0.05705, over 3228728.12 frames. ], batch size: 165, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:16:45,151 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2008, 3.4397, 3.6728, 1.8032, 3.7645, 3.8719, 2.8557, 2.8757], device='cuda:7'), covar=tensor([0.0802, 0.0189, 0.0129, 0.1097, 0.0067, 0.0086, 0.0389, 0.0420], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0100, 0.0090, 0.0137, 0.0070, 0.0102, 0.0120, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 10:16:50,592 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7627, 3.0838, 2.6921, 5.1116, 3.9022, 4.3927, 1.7073, 3.2250], device='cuda:7'), covar=tensor([0.1240, 0.0630, 0.1102, 0.0107, 0.0317, 0.0336, 0.1389, 0.0705], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0156, 0.0178, 0.0144, 0.0197, 0.0205, 0.0178, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 10:17:02,224 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:17:12,325 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106336.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:17:27,380 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.167e+02 2.529e+02 2.965e+02 6.623e+02, threshold=5.058e+02, percent-clipped=3.0 2023-04-29 10:17:35,193 INFO [train.py:904] (7/8) Epoch 11, batch 4850, loss[loss=0.2249, simple_loss=0.3087, pruned_loss=0.07051, over 16747.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2846, pruned_loss=0.05674, over 3215930.53 frames. ], batch size: 134, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:17:46,965 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.04 vs. limit=5.0 2023-04-29 10:17:49,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5750, 4.5713, 4.9730, 4.9869, 4.9587, 4.5961, 4.5943, 4.2726], device='cuda:7'), covar=tensor([0.0258, 0.0412, 0.0335, 0.0341, 0.0417, 0.0311, 0.0857, 0.0452], device='cuda:7'), in_proj_covar=tensor([0.0317, 0.0328, 0.0331, 0.0312, 0.0374, 0.0349, 0.0450, 0.0281], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 10:18:38,707 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106397.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:18:45,468 INFO [train.py:904] (7/8) Epoch 11, batch 4900, loss[loss=0.2152, simple_loss=0.285, pruned_loss=0.07275, over 11961.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.284, pruned_loss=0.05557, over 3183741.31 frames. ], batch size: 246, lr: 6.23e-03, grad_scale: 8.0 2023-04-29 10:19:29,955 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106433.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:19:50,195 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.157e+02 2.578e+02 2.904e+02 4.407e+02, threshold=5.156e+02, percent-clipped=0.0 2023-04-29 10:19:52,369 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106448.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:19:56,929 INFO [train.py:904] (7/8) Epoch 11, batch 4950, loss[loss=0.201, simple_loss=0.2974, pruned_loss=0.05227, over 16864.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2833, pruned_loss=0.05498, over 3187786.57 frames. ], batch size: 96, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:20:01,969 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106454.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:20:03,090 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106455.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:20:55,986 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-29 10:20:59,708 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.4925, 2.6736, 2.2732, 3.9369, 2.7931, 3.8238, 1.3475, 2.7008], device='cuda:7'), covar=tensor([0.1345, 0.0659, 0.1211, 0.0134, 0.0218, 0.0365, 0.1544, 0.0834], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0157, 0.0180, 0.0144, 0.0198, 0.0207, 0.0179, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 10:21:00,655 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106496.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:21:08,424 INFO [train.py:904] (7/8) Epoch 11, batch 5000, loss[loss=0.2262, simple_loss=0.3066, pruned_loss=0.07288, over 11795.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2856, pruned_loss=0.05519, over 3192414.90 frames. ], batch size: 246, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:21:10,944 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106503.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:21:28,699 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106515.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:22:14,241 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.415e+02 2.751e+02 3.160e+02 5.367e+02, threshold=5.502e+02, percent-clipped=2.0 2023-04-29 10:22:21,433 INFO [train.py:904] (7/8) Epoch 11, batch 5050, loss[loss=0.1883, simple_loss=0.2818, pruned_loss=0.0474, over 16420.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2859, pruned_loss=0.05557, over 3201700.97 frames. ], batch size: 146, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:23:32,385 INFO [train.py:904] (7/8) Epoch 11, batch 5100, loss[loss=0.1718, simple_loss=0.2637, pruned_loss=0.03993, over 16700.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2844, pruned_loss=0.0546, over 3205458.47 frames. ], batch size: 89, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:23:56,172 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.76 vs. limit=5.0 2023-04-29 10:24:03,652 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106624.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:24:23,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3642, 5.6735, 5.3898, 5.4749, 5.1032, 4.9823, 5.0684, 5.7687], device='cuda:7'), covar=tensor([0.0936, 0.0676, 0.0865, 0.0590, 0.0706, 0.0594, 0.0834, 0.0686], device='cuda:7'), in_proj_covar=tensor([0.0525, 0.0650, 0.0539, 0.0449, 0.0413, 0.0423, 0.0538, 0.0500], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:24:38,777 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.185e+02 2.500e+02 2.940e+02 7.724e+02, threshold=5.000e+02, percent-clipped=1.0 2023-04-29 10:24:44,362 INFO [train.py:904] (7/8) Epoch 11, batch 5150, loss[loss=0.2028, simple_loss=0.2994, pruned_loss=0.05305, over 15306.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2841, pruned_loss=0.0534, over 3206231.71 frames. ], batch size: 190, lr: 6.22e-03, grad_scale: 4.0 2023-04-29 10:24:47,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9291, 4.9118, 4.8234, 4.1697, 4.8106, 1.8064, 4.5655, 4.7767], device='cuda:7'), covar=tensor([0.0073, 0.0059, 0.0111, 0.0398, 0.0075, 0.2265, 0.0100, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0117, 0.0164, 0.0159, 0.0135, 0.0179, 0.0151, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:25:12,947 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3283, 3.0023, 2.6432, 2.1699, 2.2521, 2.2200, 2.9473, 2.9137], device='cuda:7'), covar=tensor([0.2179, 0.0638, 0.1306, 0.2054, 0.1814, 0.1650, 0.0467, 0.0849], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0255, 0.0281, 0.0275, 0.0281, 0.0220, 0.0265, 0.0294], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:25:29,612 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9669, 2.3304, 1.9606, 2.1645, 2.7394, 2.4236, 2.8434, 2.9026], device='cuda:7'), covar=tensor([0.0085, 0.0311, 0.0361, 0.0342, 0.0156, 0.0300, 0.0135, 0.0198], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0199, 0.0195, 0.0194, 0.0198, 0.0200, 0.0201, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:25:43,753 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:25:56,056 INFO [train.py:904] (7/8) Epoch 11, batch 5200, loss[loss=0.1864, simple_loss=0.2775, pruned_loss=0.04762, over 15420.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2821, pruned_loss=0.0525, over 3206603.88 frames. ], batch size: 190, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:26:42,787 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106733.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:27:02,572 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8402, 3.9699, 2.2335, 4.4700, 2.8375, 4.3951, 2.3438, 2.9709], device='cuda:7'), covar=tensor([0.0177, 0.0265, 0.1550, 0.0109, 0.0817, 0.0300, 0.1392, 0.0731], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0162, 0.0186, 0.0124, 0.0166, 0.0203, 0.0192, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 10:27:04,339 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.187e+02 2.709e+02 3.054e+02 5.980e+02, threshold=5.418e+02, percent-clipped=1.0 2023-04-29 10:27:11,357 INFO [train.py:904] (7/8) Epoch 11, batch 5250, loss[loss=0.1725, simple_loss=0.2597, pruned_loss=0.04271, over 16485.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.28, pruned_loss=0.05274, over 3193145.36 frames. ], batch size: 68, lr: 6.22e-03, grad_scale: 8.0 2023-04-29 10:27:52,602 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:28:22,269 INFO [train.py:904] (7/8) Epoch 11, batch 5300, loss[loss=0.1534, simple_loss=0.2357, pruned_loss=0.03557, over 17209.00 frames. ], tot_loss[loss=0.1899, simple_loss=0.2765, pruned_loss=0.05168, over 3190848.07 frames. ], batch size: 45, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:28:32,968 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106810.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:29:09,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6218, 2.1963, 2.2964, 4.4961, 2.1994, 2.7661, 2.3077, 2.5181], device='cuda:7'), covar=tensor([0.0877, 0.3140, 0.2186, 0.0308, 0.3416, 0.2083, 0.2886, 0.2688], device='cuda:7'), in_proj_covar=tensor([0.0361, 0.0390, 0.0325, 0.0318, 0.0408, 0.0448, 0.0352, 0.0456], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:29:23,962 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 10:29:27,202 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.160e+02 2.565e+02 2.956e+02 5.223e+02, threshold=5.130e+02, percent-clipped=0.0 2023-04-29 10:29:33,914 INFO [train.py:904] (7/8) Epoch 11, batch 5350, loss[loss=0.1808, simple_loss=0.2685, pruned_loss=0.04652, over 17112.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2746, pruned_loss=0.05096, over 3205748.72 frames. ], batch size: 47, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:29:50,846 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-04-29 10:29:59,832 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9251, 3.7858, 3.8455, 4.1189, 4.2369, 3.8958, 4.1542, 4.2531], device='cuda:7'), covar=tensor([0.1392, 0.1124, 0.1988, 0.0852, 0.0697, 0.1366, 0.0930, 0.0701], device='cuda:7'), in_proj_covar=tensor([0.0519, 0.0652, 0.0796, 0.0669, 0.0508, 0.0514, 0.0515, 0.0591], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:30:26,728 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-04-29 10:30:45,869 INFO [train.py:904] (7/8) Epoch 11, batch 5400, loss[loss=0.2702, simple_loss=0.3452, pruned_loss=0.0976, over 12190.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2775, pruned_loss=0.05168, over 3207638.85 frames. ], batch size: 246, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:31:18,899 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106924.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:31:20,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0249, 3.6246, 3.5441, 2.3240, 3.3040, 3.5399, 3.3197, 2.0330], device='cuda:7'), covar=tensor([0.0472, 0.0029, 0.0036, 0.0332, 0.0062, 0.0081, 0.0067, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0069, 0.0070, 0.0127, 0.0078, 0.0091, 0.0078, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 10:31:54,547 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.177e+02 2.658e+02 3.216e+02 5.881e+02, threshold=5.316e+02, percent-clipped=3.0 2023-04-29 10:32:02,036 INFO [train.py:904] (7/8) Epoch 11, batch 5450, loss[loss=0.2135, simple_loss=0.2896, pruned_loss=0.06872, over 11728.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2809, pruned_loss=0.05364, over 3191098.07 frames. ], batch size: 248, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:32:34,452 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=106972.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:33:04,756 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106992.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:33:19,274 INFO [train.py:904] (7/8) Epoch 11, batch 5500, loss[loss=0.222, simple_loss=0.3088, pruned_loss=0.06763, over 16652.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2885, pruned_loss=0.05825, over 3180746.32 frames. ], batch size: 89, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:34:18,945 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:34:31,621 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 3.119e+02 3.959e+02 4.735e+02 9.206e+02, threshold=7.917e+02, percent-clipped=14.0 2023-04-29 10:34:37,914 INFO [train.py:904] (7/8) Epoch 11, batch 5550, loss[loss=0.2226, simple_loss=0.3098, pruned_loss=0.06768, over 16425.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2963, pruned_loss=0.06399, over 3131888.17 frames. ], batch size: 75, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:35:02,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1298, 3.3893, 3.5720, 3.5439, 3.5595, 3.3846, 3.3735, 3.3931], device='cuda:7'), covar=tensor([0.0378, 0.0626, 0.0425, 0.0460, 0.0542, 0.0522, 0.0895, 0.0555], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0335, 0.0338, 0.0325, 0.0384, 0.0359, 0.0461, 0.0288], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 10:35:11,958 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3117, 1.5085, 1.9551, 2.2862, 2.2899, 2.5219, 1.6878, 2.5341], device='cuda:7'), covar=tensor([0.0157, 0.0380, 0.0220, 0.0223, 0.0207, 0.0138, 0.0357, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0171, 0.0153, 0.0160, 0.0166, 0.0123, 0.0168, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 10:35:12,369 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-29 10:35:54,724 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6328, 2.5947, 1.7671, 2.6936, 2.1796, 2.7656, 2.0750, 2.4081], device='cuda:7'), covar=tensor([0.0219, 0.0328, 0.1143, 0.0187, 0.0595, 0.0448, 0.0983, 0.0495], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0165, 0.0190, 0.0127, 0.0169, 0.0207, 0.0196, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 10:35:57,911 INFO [train.py:904] (7/8) Epoch 11, batch 5600, loss[loss=0.2228, simple_loss=0.3035, pruned_loss=0.071, over 16649.00 frames. ], tot_loss[loss=0.2212, simple_loss=0.3029, pruned_loss=0.06971, over 3094428.03 frames. ], batch size: 57, lr: 6.21e-03, grad_scale: 8.0 2023-04-29 10:36:12,148 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107110.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:37:02,610 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107140.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:37:17,301 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.615e+02 3.503e+02 4.339e+02 5.380e+02 1.040e+03, threshold=8.679e+02, percent-clipped=3.0 2023-04-29 10:37:21,713 INFO [train.py:904] (7/8) Epoch 11, batch 5650, loss[loss=0.2038, simple_loss=0.2902, pruned_loss=0.05869, over 16746.00 frames. ], tot_loss[loss=0.227, simple_loss=0.3078, pruned_loss=0.07307, over 3082934.49 frames. ], batch size: 62, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:37:22,230 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107152.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:37:32,838 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:38:42,851 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107201.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 10:38:43,537 INFO [train.py:904] (7/8) Epoch 11, batch 5700, loss[loss=0.2271, simple_loss=0.3207, pruned_loss=0.06672, over 16772.00 frames. ], tot_loss[loss=0.23, simple_loss=0.3099, pruned_loss=0.07511, over 3070173.08 frames. ], batch size: 39, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:38:47,535 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-29 10:39:03,033 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107213.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 10:39:59,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.558e+02 3.709e+02 4.582e+02 6.080e+02 1.249e+03, threshold=9.164e+02, percent-clipped=1.0 2023-04-29 10:40:04,429 INFO [train.py:904] (7/8) Epoch 11, batch 5750, loss[loss=0.2207, simple_loss=0.3046, pruned_loss=0.06842, over 16876.00 frames. ], tot_loss[loss=0.2329, simple_loss=0.3121, pruned_loss=0.0768, over 3050999.93 frames. ], batch size: 109, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:40:35,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4272, 1.6212, 1.9958, 2.3036, 2.3675, 2.5358, 1.8165, 2.4756], device='cuda:7'), covar=tensor([0.0151, 0.0358, 0.0213, 0.0229, 0.0202, 0.0131, 0.0342, 0.0107], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0170, 0.0152, 0.0158, 0.0165, 0.0123, 0.0168, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 10:41:25,732 INFO [train.py:904] (7/8) Epoch 11, batch 5800, loss[loss=0.199, simple_loss=0.2974, pruned_loss=0.05032, over 16494.00 frames. ], tot_loss[loss=0.2318, simple_loss=0.3117, pruned_loss=0.07591, over 3041679.65 frames. ], batch size: 68, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:42:24,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7096, 2.5570, 2.4040, 3.7720, 2.7574, 3.8061, 1.4496, 2.8161], device='cuda:7'), covar=tensor([0.1253, 0.0659, 0.1102, 0.0145, 0.0251, 0.0393, 0.1490, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0158, 0.0181, 0.0143, 0.0198, 0.0208, 0.0180, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 10:42:39,029 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.001e+02 2.956e+02 3.560e+02 4.660e+02 9.709e+02, threshold=7.120e+02, percent-clipped=1.0 2023-04-29 10:42:43,690 INFO [train.py:904] (7/8) Epoch 11, batch 5850, loss[loss=0.2154, simple_loss=0.3026, pruned_loss=0.06413, over 16718.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.309, pruned_loss=0.07355, over 3056378.13 frames. ], batch size: 62, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:44:05,222 INFO [train.py:904] (7/8) Epoch 11, batch 5900, loss[loss=0.2109, simple_loss=0.2958, pruned_loss=0.06304, over 16736.00 frames. ], tot_loss[loss=0.2281, simple_loss=0.309, pruned_loss=0.07359, over 3069190.50 frames. ], batch size: 124, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:44:30,186 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4152, 2.0522, 2.1385, 4.2434, 1.9754, 2.5943, 2.1880, 2.2895], device='cuda:7'), covar=tensor([0.0903, 0.3186, 0.2249, 0.0324, 0.3798, 0.2084, 0.2943, 0.2866], device='cuda:7'), in_proj_covar=tensor([0.0354, 0.0382, 0.0321, 0.0313, 0.0402, 0.0439, 0.0347, 0.0449], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:45:21,997 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.914e+02 3.759e+02 4.619e+02 1.179e+03, threshold=7.519e+02, percent-clipped=2.0 2023-04-29 10:45:26,049 INFO [train.py:904] (7/8) Epoch 11, batch 5950, loss[loss=0.2411, simple_loss=0.3141, pruned_loss=0.08403, over 11356.00 frames. ], tot_loss[loss=0.2268, simple_loss=0.3092, pruned_loss=0.07215, over 3053784.82 frames. ], batch size: 247, lr: 6.20e-03, grad_scale: 4.0 2023-04-29 10:46:26,213 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8404, 3.1537, 2.5758, 4.9882, 3.8254, 4.3721, 1.6415, 3.1841], device='cuda:7'), covar=tensor([0.1311, 0.0695, 0.1299, 0.0175, 0.0438, 0.0377, 0.1585, 0.0857], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0158, 0.0181, 0.0143, 0.0199, 0.0208, 0.0180, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 10:46:40,557 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107496.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 10:46:45,955 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 10:46:48,951 INFO [train.py:904] (7/8) Epoch 11, batch 6000, loss[loss=0.2215, simple_loss=0.2953, pruned_loss=0.07386, over 11727.00 frames. ], tot_loss[loss=0.2259, simple_loss=0.3083, pruned_loss=0.07172, over 3070044.84 frames. ], batch size: 248, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:46:48,951 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 10:46:59,889 INFO [train.py:938] (7/8) Epoch 11, validation: loss=0.163, simple_loss=0.2761, pruned_loss=0.02492, over 944034.00 frames. 2023-04-29 10:46:59,890 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 10:47:10,250 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107508.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 10:47:13,688 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107510.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:47:22,736 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 10:47:35,717 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6284, 2.2631, 2.3788, 4.5039, 2.1974, 2.7405, 2.3826, 2.5718], device='cuda:7'), covar=tensor([0.0903, 0.3164, 0.2085, 0.0316, 0.3564, 0.2192, 0.2732, 0.2900], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0387, 0.0325, 0.0316, 0.0407, 0.0445, 0.0352, 0.0454], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:47:49,869 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4712, 5.3971, 5.2229, 4.4723, 5.2248, 1.8652, 4.9769, 5.1284], device='cuda:7'), covar=tensor([0.0070, 0.0058, 0.0133, 0.0382, 0.0075, 0.2245, 0.0121, 0.0153], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0114, 0.0162, 0.0155, 0.0133, 0.0175, 0.0149, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:48:12,721 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.272e+02 3.327e+02 3.927e+02 4.902e+02 9.680e+02, threshold=7.854e+02, percent-clipped=6.0 2023-04-29 10:48:18,425 INFO [train.py:904] (7/8) Epoch 11, batch 6050, loss[loss=0.1997, simple_loss=0.2916, pruned_loss=0.05388, over 16715.00 frames. ], tot_loss[loss=0.224, simple_loss=0.3066, pruned_loss=0.07071, over 3090305.02 frames. ], batch size: 83, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:48:48,412 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107571.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:49:21,960 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 10:49:34,940 INFO [train.py:904] (7/8) Epoch 11, batch 6100, loss[loss=0.2799, simple_loss=0.3491, pruned_loss=0.1053, over 11847.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3061, pruned_loss=0.07011, over 3078609.29 frames. ], batch size: 248, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:50:51,425 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.980e+02 3.646e+02 4.413e+02 7.935e+02, threshold=7.292e+02, percent-clipped=3.0 2023-04-29 10:50:56,527 INFO [train.py:904] (7/8) Epoch 11, batch 6150, loss[loss=0.232, simple_loss=0.3038, pruned_loss=0.08015, over 11163.00 frames. ], tot_loss[loss=0.2205, simple_loss=0.3038, pruned_loss=0.06864, over 3108211.11 frames. ], batch size: 248, lr: 6.19e-03, grad_scale: 8.0 2023-04-29 10:51:01,796 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 10:51:07,644 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:52:14,170 INFO [train.py:904] (7/8) Epoch 11, batch 6200, loss[loss=0.1791, simple_loss=0.2692, pruned_loss=0.04454, over 16586.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3021, pruned_loss=0.06855, over 3108613.43 frames. ], batch size: 68, lr: 6.19e-03, grad_scale: 4.0 2023-04-29 10:52:37,370 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-29 10:52:37,477 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 10:52:42,680 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107720.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:53:07,999 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107737.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:53:27,429 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.232e+02 3.321e+02 3.959e+02 5.175e+02 1.024e+03, threshold=7.919e+02, percent-clipped=5.0 2023-04-29 10:53:30,004 INFO [train.py:904] (7/8) Epoch 11, batch 6250, loss[loss=0.2689, simple_loss=0.3337, pruned_loss=0.1021, over 11819.00 frames. ], tot_loss[loss=0.2195, simple_loss=0.3021, pruned_loss=0.06851, over 3108244.63 frames. ], batch size: 247, lr: 6.19e-03, grad_scale: 4.0 2023-04-29 10:54:36,386 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107796.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 10:54:38,999 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107798.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:54:45,103 INFO [train.py:904] (7/8) Epoch 11, batch 6300, loss[loss=0.2148, simple_loss=0.2989, pruned_loss=0.06533, over 15267.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.3025, pruned_loss=0.06847, over 3118896.19 frames. ], batch size: 190, lr: 6.19e-03, grad_scale: 4.0 2023-04-29 10:54:54,740 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107808.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:55:52,955 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107844.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 10:56:00,517 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 3.066e+02 3.848e+02 4.977e+02 9.936e+02, threshold=7.696e+02, percent-clipped=3.0 2023-04-29 10:56:03,053 INFO [train.py:904] (7/8) Epoch 11, batch 6350, loss[loss=0.2615, simple_loss=0.3307, pruned_loss=0.09613, over 16321.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.3034, pruned_loss=0.07, over 3104013.62 frames. ], batch size: 165, lr: 6.18e-03, grad_scale: 4.0 2023-04-29 10:56:10,569 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=107856.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:56:14,912 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-29 10:56:18,683 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 10:56:25,831 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107866.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 10:57:20,963 INFO [train.py:904] (7/8) Epoch 11, batch 6400, loss[loss=0.2825, simple_loss=0.3406, pruned_loss=0.1122, over 11172.00 frames. ], tot_loss[loss=0.2239, simple_loss=0.3046, pruned_loss=0.07162, over 3085995.51 frames. ], batch size: 246, lr: 6.18e-03, grad_scale: 8.0 2023-04-29 10:58:35,867 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 3.293e+02 4.168e+02 5.156e+02 1.369e+03, threshold=8.335e+02, percent-clipped=6.0 2023-04-29 10:58:35,889 INFO [train.py:904] (7/8) Epoch 11, batch 6450, loss[loss=0.2091, simple_loss=0.2953, pruned_loss=0.06148, over 16589.00 frames. ], tot_loss[loss=0.2223, simple_loss=0.3036, pruned_loss=0.07057, over 3091859.45 frames. ], batch size: 68, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 10:58:38,864 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2856, 1.9963, 2.1311, 3.9306, 1.9702, 2.4622, 2.1319, 2.2021], device='cuda:7'), covar=tensor([0.0959, 0.3279, 0.2307, 0.0417, 0.3726, 0.2126, 0.2998, 0.3118], device='cuda:7'), in_proj_covar=tensor([0.0360, 0.0387, 0.0326, 0.0317, 0.0409, 0.0445, 0.0353, 0.0455], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 10:59:02,303 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2297, 3.4466, 3.5648, 3.5438, 3.5573, 3.3879, 3.4233, 3.4157], device='cuda:7'), covar=tensor([0.0374, 0.0585, 0.0488, 0.0486, 0.0532, 0.0549, 0.0798, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0340, 0.0343, 0.0322, 0.0390, 0.0360, 0.0460, 0.0292], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 10:59:54,953 INFO [train.py:904] (7/8) Epoch 11, batch 6500, loss[loss=0.188, simple_loss=0.2748, pruned_loss=0.05065, over 17125.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3008, pruned_loss=0.06917, over 3110672.25 frames. ], batch size: 47, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:00:14,836 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108015.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:01:12,921 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.008e+02 3.175e+02 3.633e+02 4.590e+02 1.124e+03, threshold=7.266e+02, percent-clipped=3.0 2023-04-29 11:01:12,936 INFO [train.py:904] (7/8) Epoch 11, batch 6550, loss[loss=0.246, simple_loss=0.3336, pruned_loss=0.07924, over 16860.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.3036, pruned_loss=0.06977, over 3110383.01 frames. ], batch size: 42, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:01:40,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5109, 3.4854, 3.4279, 2.7949, 3.3024, 2.0511, 3.1762, 2.8730], device='cuda:7'), covar=tensor([0.0112, 0.0099, 0.0157, 0.0221, 0.0090, 0.2029, 0.0117, 0.0192], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0115, 0.0162, 0.0153, 0.0133, 0.0177, 0.0148, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:02:13,650 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108093.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 11:02:25,859 INFO [train.py:904] (7/8) Epoch 11, batch 6600, loss[loss=0.221, simple_loss=0.3064, pruned_loss=0.06777, over 16426.00 frames. ], tot_loss[loss=0.2236, simple_loss=0.3062, pruned_loss=0.07049, over 3096933.90 frames. ], batch size: 146, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:02:45,759 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5860, 2.5774, 1.8578, 2.6761, 2.1776, 2.7424, 2.0186, 2.3318], device='cuda:7'), covar=tensor([0.0269, 0.0353, 0.1128, 0.0177, 0.0634, 0.0456, 0.1161, 0.0557], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0164, 0.0188, 0.0126, 0.0168, 0.0206, 0.0195, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:03:08,018 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6538, 2.6121, 2.0974, 2.4920, 3.0456, 2.6473, 3.3537, 3.2834], device='cuda:7'), covar=tensor([0.0058, 0.0269, 0.0400, 0.0301, 0.0178, 0.0302, 0.0159, 0.0162], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0199, 0.0196, 0.0197, 0.0200, 0.0202, 0.0204, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:03:30,278 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 11:03:41,607 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 3.251e+02 4.038e+02 4.757e+02 1.371e+03, threshold=8.077e+02, percent-clipped=5.0 2023-04-29 11:03:41,623 INFO [train.py:904] (7/8) Epoch 11, batch 6650, loss[loss=0.1992, simple_loss=0.2864, pruned_loss=0.05601, over 16282.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.3062, pruned_loss=0.07097, over 3109117.91 frames. ], batch size: 165, lr: 6.18e-03, grad_scale: 2.0 2023-04-29 11:03:45,523 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 11:04:03,250 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108166.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:04:56,962 INFO [train.py:904] (7/8) Epoch 11, batch 6700, loss[loss=0.2393, simple_loss=0.321, pruned_loss=0.07881, over 16568.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.305, pruned_loss=0.07121, over 3102717.91 frames. ], batch size: 75, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:05:15,641 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=108214.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:06:13,522 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 3.397e+02 4.413e+02 5.907e+02 2.679e+03, threshold=8.826e+02, percent-clipped=7.0 2023-04-29 11:06:13,543 INFO [train.py:904] (7/8) Epoch 11, batch 6750, loss[loss=0.2623, simple_loss=0.3257, pruned_loss=0.09946, over 11853.00 frames. ], tot_loss[loss=0.2235, simple_loss=0.3041, pruned_loss=0.07149, over 3082365.47 frames. ], batch size: 246, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:07:28,512 INFO [train.py:904] (7/8) Epoch 11, batch 6800, loss[loss=0.2217, simple_loss=0.3068, pruned_loss=0.06836, over 16736.00 frames. ], tot_loss[loss=0.2241, simple_loss=0.3049, pruned_loss=0.07167, over 3081195.94 frames. ], batch size: 83, lr: 6.17e-03, grad_scale: 4.0 2023-04-29 11:07:49,162 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108315.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:07:59,972 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.57 vs. limit=2.0 2023-04-29 11:08:05,637 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108325.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:08:14,803 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.67 vs. limit=5.0 2023-04-29 11:08:32,017 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.82 vs. limit=5.0 2023-04-29 11:08:45,535 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.125e+02 2.977e+02 3.381e+02 4.053e+02 7.016e+02, threshold=6.762e+02, percent-clipped=0.0 2023-04-29 11:08:45,551 INFO [train.py:904] (7/8) Epoch 11, batch 6850, loss[loss=0.2033, simple_loss=0.297, pruned_loss=0.05481, over 16768.00 frames. ], tot_loss[loss=0.2243, simple_loss=0.3055, pruned_loss=0.07159, over 3088707.05 frames. ], batch size: 39, lr: 6.17e-03, grad_scale: 4.0 2023-04-29 11:09:01,716 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=108363.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:09:28,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0672, 5.0313, 5.5221, 5.5092, 5.4437, 5.1478, 5.0441, 4.8100], device='cuda:7'), covar=tensor([0.0277, 0.0460, 0.0379, 0.0313, 0.0363, 0.0330, 0.1001, 0.0419], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0339, 0.0343, 0.0321, 0.0391, 0.0357, 0.0460, 0.0292], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 11:09:35,670 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108386.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:09:46,480 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108393.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:09:59,716 INFO [train.py:904] (7/8) Epoch 11, batch 6900, loss[loss=0.2621, simple_loss=0.3399, pruned_loss=0.09217, over 16381.00 frames. ], tot_loss[loss=0.2251, simple_loss=0.3078, pruned_loss=0.0712, over 3099324.67 frames. ], batch size: 35, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:10:02,375 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 11:10:45,923 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:10:59,696 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=108441.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:11:15,798 INFO [train.py:904] (7/8) Epoch 11, batch 6950, loss[loss=0.2091, simple_loss=0.2945, pruned_loss=0.06185, over 16714.00 frames. ], tot_loss[loss=0.2268, simple_loss=0.3089, pruned_loss=0.0723, over 3102932.29 frames. ], batch size: 83, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:11:17,889 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.069e+02 2.933e+02 3.744e+02 4.621e+02 9.342e+02, threshold=7.489e+02, percent-clipped=9.0 2023-04-29 11:12:20,908 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108493.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 11:12:33,551 INFO [train.py:904] (7/8) Epoch 11, batch 7000, loss[loss=0.2076, simple_loss=0.304, pruned_loss=0.05556, over 16961.00 frames. ], tot_loss[loss=0.2259, simple_loss=0.3086, pruned_loss=0.07159, over 3088709.14 frames. ], batch size: 41, lr: 6.17e-03, grad_scale: 2.0 2023-04-29 11:13:52,280 INFO [train.py:904] (7/8) Epoch 11, batch 7050, loss[loss=0.2317, simple_loss=0.3101, pruned_loss=0.0766, over 16339.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3091, pruned_loss=0.07107, over 3101352.94 frames. ], batch size: 146, lr: 6.16e-03, grad_scale: 2.0 2023-04-29 11:13:53,483 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.044e+02 3.927e+02 4.822e+02 1.171e+03, threshold=7.854e+02, percent-clipped=4.0 2023-04-29 11:14:33,451 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:15:02,474 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7409, 4.5576, 4.6001, 3.0717, 3.8533, 4.3966, 4.1675, 2.6236], device='cuda:7'), covar=tensor([0.0370, 0.0019, 0.0021, 0.0276, 0.0066, 0.0082, 0.0044, 0.0297], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0068, 0.0068, 0.0125, 0.0077, 0.0090, 0.0077, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 11:15:11,200 INFO [train.py:904] (7/8) Epoch 11, batch 7100, loss[loss=0.2035, simple_loss=0.2892, pruned_loss=0.05896, over 16223.00 frames. ], tot_loss[loss=0.2237, simple_loss=0.3069, pruned_loss=0.07021, over 3110672.81 frames. ], batch size: 165, lr: 6.16e-03, grad_scale: 2.0 2023-04-29 11:15:13,041 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-29 11:16:07,810 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108638.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:16:27,146 INFO [train.py:904] (7/8) Epoch 11, batch 7150, loss[loss=0.2113, simple_loss=0.2963, pruned_loss=0.06311, over 16692.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3044, pruned_loss=0.06984, over 3112146.62 frames. ], batch size: 89, lr: 6.16e-03, grad_scale: 2.0 2023-04-29 11:16:28,932 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.054e+02 3.516e+02 4.486e+02 8.068e+02, threshold=7.031e+02, percent-clipped=2.0 2023-04-29 11:16:51,007 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0372, 3.3060, 3.3940, 2.2511, 3.0534, 3.3028, 3.2202, 1.9412], device='cuda:7'), covar=tensor([0.0427, 0.0035, 0.0040, 0.0315, 0.0082, 0.0088, 0.0064, 0.0378], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0068, 0.0068, 0.0125, 0.0077, 0.0090, 0.0077, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 11:17:10,536 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108681.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:17:41,824 INFO [train.py:904] (7/8) Epoch 11, batch 7200, loss[loss=0.1982, simple_loss=0.29, pruned_loss=0.05314, over 16428.00 frames. ], tot_loss[loss=0.22, simple_loss=0.3029, pruned_loss=0.06856, over 3108351.65 frames. ], batch size: 146, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:19:02,070 INFO [train.py:904] (7/8) Epoch 11, batch 7250, loss[loss=0.2351, simple_loss=0.3031, pruned_loss=0.08361, over 11558.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.301, pruned_loss=0.06768, over 3086415.87 frames. ], batch size: 248, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:19:03,150 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.903e+02 3.497e+02 4.380e+02 7.660e+02, threshold=6.994e+02, percent-clipped=1.0 2023-04-29 11:19:10,257 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-29 11:19:45,869 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.27 vs. limit=5.0 2023-04-29 11:19:55,768 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108788.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 11:20:01,032 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108791.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:20:16,100 INFO [train.py:904] (7/8) Epoch 11, batch 7300, loss[loss=0.2044, simple_loss=0.2955, pruned_loss=0.05666, over 16766.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.2998, pruned_loss=0.06685, over 3078399.96 frames. ], batch size: 89, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:21:05,592 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5457, 2.5159, 2.2297, 4.0184, 3.1095, 3.9061, 1.3815, 2.8917], device='cuda:7'), covar=tensor([0.1348, 0.0776, 0.1343, 0.0120, 0.0311, 0.0389, 0.1587, 0.0779], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0160, 0.0183, 0.0144, 0.0201, 0.0209, 0.0182, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:21:34,179 INFO [train.py:904] (7/8) Epoch 11, batch 7350, loss[loss=0.1996, simple_loss=0.2859, pruned_loss=0.05666, over 16779.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3004, pruned_loss=0.06746, over 3058644.92 frames. ], batch size: 124, lr: 6.16e-03, grad_scale: 4.0 2023-04-29 11:21:34,693 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:21:35,290 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 3.104e+02 3.844e+02 4.749e+02 1.066e+03, threshold=7.688e+02, percent-clipped=6.0 2023-04-29 11:22:17,967 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7590, 4.1344, 2.9795, 2.2324, 3.1221, 2.5895, 4.6682, 3.8592], device='cuda:7'), covar=tensor([0.2685, 0.0682, 0.1582, 0.2043, 0.2299, 0.1642, 0.0347, 0.0877], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0258, 0.0284, 0.0277, 0.0283, 0.0222, 0.0267, 0.0295], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:22:54,332 INFO [train.py:904] (7/8) Epoch 11, batch 7400, loss[loss=0.208, simple_loss=0.2929, pruned_loss=0.06152, over 16639.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.3017, pruned_loss=0.06832, over 3061993.00 frames. ], batch size: 57, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:23:30,580 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9981, 5.0005, 4.8047, 4.1725, 4.7909, 1.8943, 4.5834, 4.7031], device='cuda:7'), covar=tensor([0.0066, 0.0061, 0.0139, 0.0326, 0.0072, 0.2321, 0.0112, 0.0146], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0114, 0.0162, 0.0152, 0.0133, 0.0179, 0.0149, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:23:42,585 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:24:11,399 INFO [train.py:904] (7/8) Epoch 11, batch 7450, loss[loss=0.2184, simple_loss=0.3106, pruned_loss=0.06315, over 16749.00 frames. ], tot_loss[loss=0.2215, simple_loss=0.3039, pruned_loss=0.06958, over 3074702.83 frames. ], batch size: 83, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:24:13,641 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.939e+02 3.048e+02 3.705e+02 5.111e+02 8.676e+02, threshold=7.410e+02, percent-clipped=2.0 2023-04-29 11:24:48,980 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9319, 3.1550, 3.3238, 1.9916, 3.1343, 3.2785, 3.2401, 1.8754], device='cuda:7'), covar=tensor([0.0459, 0.0063, 0.0051, 0.0384, 0.0073, 0.0089, 0.0059, 0.0397], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0068, 0.0069, 0.0127, 0.0078, 0.0090, 0.0077, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 11:24:59,156 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108981.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:25:31,638 INFO [train.py:904] (7/8) Epoch 11, batch 7500, loss[loss=0.2238, simple_loss=0.3118, pruned_loss=0.06792, over 16213.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.3047, pruned_loss=0.06955, over 3061901.07 frames. ], batch size: 165, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:26:16,355 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109029.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:26:16,947 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 11:26:28,603 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109037.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 11:26:51,083 INFO [train.py:904] (7/8) Epoch 11, batch 7550, loss[loss=0.2191, simple_loss=0.3002, pruned_loss=0.06896, over 16664.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.3038, pruned_loss=0.06972, over 3065893.30 frames. ], batch size: 62, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:26:52,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 3.123e+02 3.909e+02 4.676e+02 1.371e+03, threshold=7.817e+02, percent-clipped=2.0 2023-04-29 11:27:46,150 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:28:02,050 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109098.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 11:28:06,703 INFO [train.py:904] (7/8) Epoch 11, batch 7600, loss[loss=0.2239, simple_loss=0.3098, pruned_loss=0.06897, over 16431.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3031, pruned_loss=0.06989, over 3054150.52 frames. ], batch size: 146, lr: 6.15e-03, grad_scale: 8.0 2023-04-29 11:28:07,262 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5286, 3.7207, 3.9750, 1.9518, 4.1716, 4.1712, 3.0048, 3.0166], device='cuda:7'), covar=tensor([0.0793, 0.0189, 0.0173, 0.1154, 0.0046, 0.0125, 0.0395, 0.0435], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0098, 0.0087, 0.0137, 0.0068, 0.0100, 0.0118, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 11:28:15,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4447, 4.4739, 4.4940, 2.7580, 3.9717, 4.3694, 4.0543, 2.5749], device='cuda:7'), covar=tensor([0.0414, 0.0022, 0.0023, 0.0314, 0.0055, 0.0070, 0.0047, 0.0310], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0068, 0.0068, 0.0126, 0.0077, 0.0089, 0.0077, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 11:28:20,077 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 11:28:24,474 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 11:28:27,827 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8535, 4.1263, 3.9357, 3.9654, 3.6356, 3.7300, 3.8289, 4.1044], device='cuda:7'), covar=tensor([0.0985, 0.0819, 0.0877, 0.0661, 0.0770, 0.1503, 0.0893, 0.0931], device='cuda:7'), in_proj_covar=tensor([0.0536, 0.0661, 0.0549, 0.0455, 0.0417, 0.0434, 0.0551, 0.0502], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:28:56,879 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:29:12,285 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109147.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:29:20,704 INFO [train.py:904] (7/8) Epoch 11, batch 7650, loss[loss=0.2349, simple_loss=0.3088, pruned_loss=0.08046, over 16679.00 frames. ], tot_loss[loss=0.2225, simple_loss=0.3038, pruned_loss=0.07065, over 3064356.77 frames. ], batch size: 57, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:29:23,630 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 2.787e+02 3.621e+02 4.390e+02 9.045e+02, threshold=7.242e+02, percent-clipped=3.0 2023-04-29 11:30:25,792 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109194.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:30:36,534 INFO [train.py:904] (7/8) Epoch 11, batch 7700, loss[loss=0.2287, simple_loss=0.3133, pruned_loss=0.07202, over 16341.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3038, pruned_loss=0.0708, over 3070959.79 frames. ], batch size: 146, lr: 6.15e-03, grad_scale: 4.0 2023-04-29 11:31:25,265 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109233.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:31:53,781 INFO [train.py:904] (7/8) Epoch 11, batch 7750, loss[loss=0.2004, simple_loss=0.2891, pruned_loss=0.05585, over 17147.00 frames. ], tot_loss[loss=0.222, simple_loss=0.3039, pruned_loss=0.07006, over 3085859.52 frames. ], batch size: 46, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:31:56,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.029e+02 3.061e+02 3.488e+02 4.422e+02 1.299e+03, threshold=6.977e+02, percent-clipped=6.0 2023-04-29 11:31:59,744 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109255.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:32:39,672 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:32:58,572 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109294.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:33:03,200 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5689, 3.1203, 3.0764, 1.9019, 2.7409, 2.1049, 3.2097, 3.2611], device='cuda:7'), covar=tensor([0.0276, 0.0626, 0.0574, 0.1805, 0.0800, 0.0989, 0.0615, 0.0870], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0143, 0.0156, 0.0143, 0.0137, 0.0125, 0.0136, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:33:10,726 INFO [train.py:904] (7/8) Epoch 11, batch 7800, loss[loss=0.2114, simple_loss=0.3019, pruned_loss=0.06043, over 17177.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.305, pruned_loss=0.07073, over 3083656.29 frames. ], batch size: 44, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:33:22,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7223, 4.7261, 4.5766, 4.2952, 4.1503, 4.6448, 4.5488, 4.3186], device='cuda:7'), covar=tensor([0.0708, 0.0700, 0.0317, 0.0334, 0.1157, 0.0577, 0.0418, 0.0863], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0306, 0.0279, 0.0256, 0.0299, 0.0297, 0.0194, 0.0325], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:33:35,386 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109318.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:34:23,873 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3955, 2.1539, 2.2327, 4.1591, 2.0996, 2.6204, 2.2302, 2.2559], device='cuda:7'), covar=tensor([0.0960, 0.3203, 0.2219, 0.0379, 0.3680, 0.2079, 0.2907, 0.3145], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0390, 0.0327, 0.0318, 0.0413, 0.0446, 0.0353, 0.0456], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:34:25,578 INFO [train.py:904] (7/8) Epoch 11, batch 7850, loss[loss=0.2489, simple_loss=0.3126, pruned_loss=0.09264, over 11822.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3056, pruned_loss=0.07064, over 3080209.89 frames. ], batch size: 246, lr: 6.14e-03, grad_scale: 2.0 2023-04-29 11:34:30,496 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.254e+02 2.996e+02 3.807e+02 4.830e+02 8.310e+02, threshold=7.614e+02, percent-clipped=7.0 2023-04-29 11:34:31,674 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109355.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:34:46,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5582, 4.7418, 4.9562, 4.7254, 4.7584, 5.3491, 4.8082, 4.6152], device='cuda:7'), covar=tensor([0.1174, 0.1776, 0.1967, 0.1932, 0.2489, 0.1056, 0.1538, 0.2443], device='cuda:7'), in_proj_covar=tensor([0.0353, 0.0488, 0.0531, 0.0420, 0.0558, 0.0556, 0.0420, 0.0574], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 11:34:55,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1138, 1.9584, 2.1231, 3.6531, 2.0148, 2.3146, 2.1091, 2.0989], device='cuda:7'), covar=tensor([0.1034, 0.3188, 0.2212, 0.0452, 0.3738, 0.2199, 0.2931, 0.3057], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0389, 0.0326, 0.0318, 0.0412, 0.0445, 0.0353, 0.0456], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:35:07,064 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109379.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 11:35:19,540 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 11:35:27,304 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109393.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 11:35:37,647 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-29 11:35:41,206 INFO [train.py:904] (7/8) Epoch 11, batch 7900, loss[loss=0.2552, simple_loss=0.328, pruned_loss=0.09118, over 15452.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.3047, pruned_loss=0.07054, over 3085422.10 frames. ], batch size: 191, lr: 6.14e-03, grad_scale: 2.0 2023-04-29 11:36:06,934 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 11:36:52,319 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109447.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:37:00,099 INFO [train.py:904] (7/8) Epoch 11, batch 7950, loss[loss=0.2003, simple_loss=0.277, pruned_loss=0.06182, over 16675.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3054, pruned_loss=0.07144, over 3078502.94 frames. ], batch size: 62, lr: 6.14e-03, grad_scale: 2.0 2023-04-29 11:37:04,711 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.170e+02 3.029e+02 3.447e+02 4.404e+02 6.785e+02, threshold=6.894e+02, percent-clipped=0.0 2023-04-29 11:37:33,510 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9508, 3.0716, 2.5474, 4.7708, 3.1707, 4.1947, 1.7477, 2.9309], device='cuda:7'), covar=tensor([0.1281, 0.0746, 0.1313, 0.0202, 0.0504, 0.0475, 0.1544, 0.1035], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0159, 0.0182, 0.0145, 0.0201, 0.0209, 0.0181, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:37:35,709 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 11:38:04,155 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109495.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:38:14,165 INFO [train.py:904] (7/8) Epoch 11, batch 8000, loss[loss=0.2177, simple_loss=0.3028, pruned_loss=0.06635, over 16298.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.3056, pruned_loss=0.07175, over 3075825.99 frames. ], batch size: 165, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:38:48,558 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3822, 2.8397, 2.6016, 2.2374, 2.2595, 2.2136, 2.9149, 2.8681], device='cuda:7'), covar=tensor([0.2210, 0.0939, 0.1362, 0.2067, 0.1983, 0.1799, 0.0491, 0.1021], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0260, 0.0286, 0.0279, 0.0285, 0.0225, 0.0268, 0.0296], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:39:19,018 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2639, 4.3288, 4.1260, 3.9028, 3.8136, 4.2410, 3.9868, 3.9415], device='cuda:7'), covar=tensor([0.0604, 0.0423, 0.0285, 0.0287, 0.0892, 0.0455, 0.0577, 0.0652], device='cuda:7'), in_proj_covar=tensor([0.0235, 0.0304, 0.0277, 0.0256, 0.0295, 0.0296, 0.0191, 0.0322], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:39:24,980 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109550.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:39:27,482 INFO [train.py:904] (7/8) Epoch 11, batch 8050, loss[loss=0.2434, simple_loss=0.3171, pruned_loss=0.08485, over 16364.00 frames. ], tot_loss[loss=0.225, simple_loss=0.3056, pruned_loss=0.07222, over 3046766.38 frames. ], batch size: 68, lr: 6.14e-03, grad_scale: 4.0 2023-04-29 11:39:31,004 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.160e+02 2.984e+02 3.790e+02 4.579e+02 1.063e+03, threshold=7.580e+02, percent-clipped=3.0 2023-04-29 11:39:36,526 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4117, 3.4483, 1.9014, 3.7636, 2.5417, 3.7731, 2.0276, 2.8121], device='cuda:7'), covar=tensor([0.0234, 0.0346, 0.1667, 0.0168, 0.0817, 0.0537, 0.1555, 0.0678], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0164, 0.0188, 0.0125, 0.0167, 0.0204, 0.0196, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:39:50,626 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6430, 2.5963, 2.2892, 3.8827, 2.8293, 4.0014, 1.4274, 2.8312], device='cuda:7'), covar=tensor([0.1372, 0.0726, 0.1278, 0.0184, 0.0258, 0.0352, 0.1604, 0.0863], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0161, 0.0183, 0.0146, 0.0203, 0.0210, 0.0183, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:40:07,253 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109579.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:40:40,909 INFO [train.py:904] (7/8) Epoch 11, batch 8100, loss[loss=0.2113, simple_loss=0.2984, pruned_loss=0.06206, over 16523.00 frames. ], tot_loss[loss=0.2239, simple_loss=0.305, pruned_loss=0.07135, over 3048809.18 frames. ], batch size: 62, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:41:36,149 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-29 11:41:39,225 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109640.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:41:50,603 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 11:41:54,547 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:41:56,620 INFO [train.py:904] (7/8) Epoch 11, batch 8150, loss[loss=0.1845, simple_loss=0.2699, pruned_loss=0.04952, over 16912.00 frames. ], tot_loss[loss=0.2207, simple_loss=0.3019, pruned_loss=0.06977, over 3065226.84 frames. ], batch size: 109, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:42:01,362 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.997e+02 3.110e+02 3.865e+02 4.805e+02 7.636e+02, threshold=7.730e+02, percent-clipped=1.0 2023-04-29 11:42:05,488 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3935, 3.2693, 2.6606, 2.0901, 2.2962, 2.1460, 3.2617, 3.0236], device='cuda:7'), covar=tensor([0.2579, 0.0769, 0.1662, 0.2321, 0.2320, 0.1905, 0.0569, 0.1150], device='cuda:7'), in_proj_covar=tensor([0.0307, 0.0258, 0.0286, 0.0277, 0.0284, 0.0224, 0.0268, 0.0295], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:42:06,960 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5440, 4.6955, 4.8923, 4.7650, 4.8005, 5.3128, 4.8133, 4.5897], device='cuda:7'), covar=tensor([0.1114, 0.1762, 0.1836, 0.1688, 0.2320, 0.0970, 0.1426, 0.2401], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0484, 0.0526, 0.0420, 0.0550, 0.0552, 0.0415, 0.0568], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 11:42:08,255 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:42:30,929 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109674.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 11:42:49,761 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7320, 5.0236, 4.7518, 4.7513, 4.4637, 4.4743, 4.4190, 5.0548], device='cuda:7'), covar=tensor([0.0942, 0.0786, 0.1009, 0.0799, 0.0874, 0.0996, 0.1094, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0542, 0.0664, 0.0556, 0.0463, 0.0420, 0.0439, 0.0558, 0.0510], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:42:59,287 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109693.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 11:43:07,573 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3126, 4.3374, 4.2047, 3.5057, 4.2131, 1.7073, 3.9583, 3.9041], device='cuda:7'), covar=tensor([0.0086, 0.0068, 0.0141, 0.0310, 0.0081, 0.2431, 0.0126, 0.0201], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0114, 0.0164, 0.0154, 0.0133, 0.0181, 0.0149, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:43:12,479 INFO [train.py:904] (7/8) Epoch 11, batch 8200, loss[loss=0.2675, simple_loss=0.3188, pruned_loss=0.1081, over 11489.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.2992, pruned_loss=0.06876, over 3067267.75 frames. ], batch size: 248, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:43:23,189 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-29 11:43:42,558 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109720.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:44:16,310 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109741.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 11:44:33,559 INFO [train.py:904] (7/8) Epoch 11, batch 8250, loss[loss=0.1932, simple_loss=0.2856, pruned_loss=0.05042, over 16459.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2985, pruned_loss=0.06629, over 3067622.55 frames. ], batch size: 75, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:44:38,009 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.186e+02 3.150e+02 3.966e+02 4.988e+02 1.179e+03, threshold=7.932e+02, percent-clipped=8.0 2023-04-29 11:44:46,453 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 11:45:52,517 INFO [train.py:904] (7/8) Epoch 11, batch 8300, loss[loss=0.1893, simple_loss=0.265, pruned_loss=0.0568, over 12228.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2961, pruned_loss=0.06346, over 3061732.32 frames. ], batch size: 246, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:46:29,791 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9727, 1.9757, 2.2368, 3.1562, 2.0966, 2.2105, 2.1548, 2.0602], device='cuda:7'), covar=tensor([0.0898, 0.3492, 0.2228, 0.0532, 0.3841, 0.2324, 0.3112, 0.3326], device='cuda:7'), in_proj_covar=tensor([0.0352, 0.0384, 0.0322, 0.0311, 0.0406, 0.0436, 0.0347, 0.0448], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:46:57,462 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 11:47:09,195 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109850.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:47:11,577 INFO [train.py:904] (7/8) Epoch 11, batch 8350, loss[loss=0.2108, simple_loss=0.2882, pruned_loss=0.06672, over 11786.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2962, pruned_loss=0.06247, over 3056977.20 frames. ], batch size: 246, lr: 6.13e-03, grad_scale: 4.0 2023-04-29 11:47:16,955 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.495e+02 2.844e+02 3.376e+02 6.294e+02, threshold=5.687e+02, percent-clipped=0.0 2023-04-29 11:48:25,292 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109898.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:48:30,270 INFO [train.py:904] (7/8) Epoch 11, batch 8400, loss[loss=0.1882, simple_loss=0.2776, pruned_loss=0.04939, over 16748.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2929, pruned_loss=0.06015, over 3041582.36 frames. ], batch size: 124, lr: 6.13e-03, grad_scale: 8.0 2023-04-29 11:49:22,654 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109935.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:49:45,390 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109950.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:49:47,550 INFO [train.py:904] (7/8) Epoch 11, batch 8450, loss[loss=0.1908, simple_loss=0.2772, pruned_loss=0.05223, over 16053.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2904, pruned_loss=0.05789, over 3044946.38 frames. ], batch size: 35, lr: 6.13e-03, grad_scale: 8.0 2023-04-29 11:49:52,358 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.195e+02 2.664e+02 3.370e+02 8.341e+02, threshold=5.327e+02, percent-clipped=3.0 2023-04-29 11:50:23,543 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109974.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:00,501 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=109998.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:09,686 INFO [train.py:904] (7/8) Epoch 11, batch 8500, loss[loss=0.1795, simple_loss=0.2661, pruned_loss=0.04649, over 16180.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2855, pruned_loss=0.05497, over 3022880.36 frames. ], batch size: 165, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 11:51:31,238 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110015.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:34,674 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110017.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:51:41,429 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110022.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:52:12,090 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5149, 3.4720, 2.8154, 1.9974, 2.2804, 2.2148, 3.5636, 3.1426], device='cuda:7'), covar=tensor([0.2577, 0.0676, 0.1406, 0.2294, 0.2269, 0.1863, 0.0434, 0.1080], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0246, 0.0271, 0.0265, 0.0266, 0.0215, 0.0254, 0.0280], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:52:17,184 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110044.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:52:22,973 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4063, 3.0442, 3.0733, 1.7842, 2.7377, 2.2035, 3.0449, 3.1723], device='cuda:7'), covar=tensor([0.0338, 0.0681, 0.0466, 0.1860, 0.0748, 0.0884, 0.0652, 0.0841], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0138, 0.0152, 0.0138, 0.0132, 0.0121, 0.0131, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 11:52:31,121 INFO [train.py:904] (7/8) Epoch 11, batch 8550, loss[loss=0.2121, simple_loss=0.3033, pruned_loss=0.06048, over 16330.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2831, pruned_loss=0.05341, over 3021508.34 frames. ], batch size: 146, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 11:52:34,326 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-04-29 11:52:37,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.340e+02 2.855e+02 3.457e+02 5.505e+02, threshold=5.709e+02, percent-clipped=2.0 2023-04-29 11:53:19,577 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110078.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:53:35,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4168, 3.3064, 3.4895, 1.6885, 3.6381, 3.6996, 2.8903, 2.8205], device='cuda:7'), covar=tensor([0.0603, 0.0219, 0.0183, 0.1082, 0.0066, 0.0107, 0.0341, 0.0373], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0097, 0.0084, 0.0134, 0.0067, 0.0098, 0.0116, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 11:54:07,131 INFO [train.py:904] (7/8) Epoch 11, batch 8600, loss[loss=0.1892, simple_loss=0.2802, pruned_loss=0.0491, over 16894.00 frames. ], tot_loss[loss=0.1942, simple_loss=0.2835, pruned_loss=0.05244, over 3030139.62 frames. ], batch size: 116, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 11:54:14,611 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110105.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:54:23,156 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110110.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:54:25,134 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3161, 4.3434, 4.2090, 3.6881, 4.2041, 1.7031, 3.9957, 4.0422], device='cuda:7'), covar=tensor([0.0078, 0.0065, 0.0122, 0.0226, 0.0071, 0.2220, 0.0104, 0.0168], device='cuda:7'), in_proj_covar=tensor([0.0123, 0.0111, 0.0158, 0.0147, 0.0129, 0.0175, 0.0144, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:54:25,442 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 11:54:35,108 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110115.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:55:43,453 INFO [train.py:904] (7/8) Epoch 11, batch 8650, loss[loss=0.1874, simple_loss=0.287, pruned_loss=0.04388, over 16161.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2815, pruned_loss=0.05131, over 3015727.32 frames. ], batch size: 165, lr: 6.12e-03, grad_scale: 4.0 2023-04-29 11:55:45,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2234, 3.4173, 3.6152, 3.5827, 3.5984, 3.4292, 3.4574, 3.4936], device='cuda:7'), covar=tensor([0.0345, 0.0488, 0.0443, 0.0430, 0.0463, 0.0390, 0.0698, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0315, 0.0320, 0.0299, 0.0365, 0.0333, 0.0433, 0.0272], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-29 11:55:53,866 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.527e+02 3.196e+02 4.321e+02 7.577e+02, threshold=6.393e+02, percent-clipped=5.0 2023-04-29 11:56:28,799 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110171.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:56:40,487 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:57:04,642 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4073, 2.5844, 2.1617, 2.2475, 2.9332, 2.6475, 3.2349, 3.2074], device='cuda:7'), covar=tensor([0.0107, 0.0280, 0.0338, 0.0343, 0.0172, 0.0259, 0.0132, 0.0137], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0198, 0.0193, 0.0192, 0.0196, 0.0196, 0.0196, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:57:30,809 INFO [train.py:904] (7/8) Epoch 11, batch 8700, loss[loss=0.1773, simple_loss=0.2729, pruned_loss=0.04085, over 16864.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2788, pruned_loss=0.04992, over 3029750.49 frames. ], batch size: 102, lr: 6.12e-03, grad_scale: 4.0 2023-04-29 11:57:49,677 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3795, 3.0094, 2.6794, 2.1977, 2.1533, 2.2175, 3.0220, 2.8091], device='cuda:7'), covar=tensor([0.2269, 0.0716, 0.1370, 0.2190, 0.2362, 0.1890, 0.0450, 0.1166], device='cuda:7'), in_proj_covar=tensor([0.0297, 0.0250, 0.0276, 0.0270, 0.0268, 0.0217, 0.0258, 0.0284], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 11:58:24,707 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4052, 3.6252, 2.0840, 3.9667, 2.6262, 3.8773, 2.2436, 2.7795], device='cuda:7'), covar=tensor([0.0241, 0.0287, 0.1415, 0.0151, 0.0705, 0.0450, 0.1340, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0155, 0.0181, 0.0121, 0.0162, 0.0194, 0.0189, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 11:58:34,081 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110235.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 11:59:03,790 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-29 11:59:05,968 INFO [train.py:904] (7/8) Epoch 11, batch 8750, loss[loss=0.1764, simple_loss=0.2771, pruned_loss=0.03785, over 16715.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2778, pruned_loss=0.0489, over 3035591.22 frames. ], batch size: 76, lr: 6.12e-03, grad_scale: 4.0 2023-04-29 11:59:15,716 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.315e+02 2.719e+02 3.353e+02 7.427e+02, threshold=5.437e+02, percent-clipped=1.0 2023-04-29 11:59:28,385 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-29 12:00:22,510 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:00:22,650 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:01:00,291 INFO [train.py:904] (7/8) Epoch 11, batch 8800, loss[loss=0.1852, simple_loss=0.2688, pruned_loss=0.0508, over 12775.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2758, pruned_loss=0.04739, over 3048928.95 frames. ], batch size: 248, lr: 6.12e-03, grad_scale: 8.0 2023-04-29 12:01:03,746 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110303.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:01:28,150 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110315.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:01:33,054 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-29 12:01:34,201 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110318.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:02:05,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 12:02:29,444 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110344.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:02:45,259 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1992, 3.5342, 3.5382, 2.2980, 3.2408, 3.5459, 3.3491, 1.5918], device='cuda:7'), covar=tensor([0.0447, 0.0040, 0.0047, 0.0364, 0.0078, 0.0081, 0.0085, 0.0557], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0066, 0.0068, 0.0123, 0.0077, 0.0087, 0.0075, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 12:02:45,841 INFO [train.py:904] (7/8) Epoch 11, batch 8850, loss[loss=0.1971, simple_loss=0.2988, pruned_loss=0.04768, over 16798.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2778, pruned_loss=0.04684, over 3025462.05 frames. ], batch size: 124, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:02:52,482 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.510e+02 2.922e+02 3.907e+02 6.547e+02, threshold=5.844e+02, percent-clipped=7.0 2023-04-29 12:03:09,620 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110363.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:03:11,336 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110364.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:03:29,846 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110373.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:03:42,525 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110379.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:04:27,928 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110400.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:04:30,883 INFO [train.py:904] (7/8) Epoch 11, batch 8900, loss[loss=0.1784, simple_loss=0.2773, pruned_loss=0.03981, over 16859.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2778, pruned_loss=0.04603, over 3027340.09 frames. ], batch size: 96, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:05:08,450 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 12:06:21,886 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-29 12:06:36,601 INFO [train.py:904] (7/8) Epoch 11, batch 8950, loss[loss=0.1827, simple_loss=0.2739, pruned_loss=0.04575, over 16322.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2775, pruned_loss=0.04628, over 3046621.15 frames. ], batch size: 146, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:06:45,510 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.155e+02 2.674e+02 3.353e+02 7.252e+02, threshold=5.349e+02, percent-clipped=1.0 2023-04-29 12:07:08,194 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110466.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:07:18,587 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110471.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:08:26,532 INFO [train.py:904] (7/8) Epoch 11, batch 9000, loss[loss=0.177, simple_loss=0.2613, pruned_loss=0.04634, over 12583.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2743, pruned_loss=0.04475, over 3055815.42 frames. ], batch size: 248, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:08:26,532 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 12:08:36,936 INFO [train.py:938] (7/8) Epoch 11, validation: loss=0.1545, simple_loss=0.2586, pruned_loss=0.02523, over 944034.00 frames. 2023-04-29 12:08:36,936 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 12:10:21,915 INFO [train.py:904] (7/8) Epoch 11, batch 9050, loss[loss=0.2014, simple_loss=0.2824, pruned_loss=0.06027, over 12435.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2758, pruned_loss=0.04554, over 3073448.68 frames. ], batch size: 248, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:10:28,905 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.375e+02 3.006e+02 3.857e+02 1.104e+03, threshold=6.012e+02, percent-clipped=5.0 2023-04-29 12:12:06,404 INFO [train.py:904] (7/8) Epoch 11, batch 9100, loss[loss=0.1809, simple_loss=0.2767, pruned_loss=0.0425, over 16714.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2751, pruned_loss=0.04594, over 3058184.09 frames. ], batch size: 134, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:12:13,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8999, 1.7197, 1.5178, 1.3516, 1.7932, 1.4199, 1.6541, 1.8964], device='cuda:7'), covar=tensor([0.0134, 0.0302, 0.0411, 0.0371, 0.0205, 0.0280, 0.0145, 0.0204], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0199, 0.0193, 0.0193, 0.0197, 0.0197, 0.0195, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:12:25,261 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110611.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:13:11,207 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5392, 3.6038, 3.3601, 3.1956, 3.2532, 3.5248, 3.2851, 3.2680], device='cuda:7'), covar=tensor([0.0563, 0.0612, 0.0244, 0.0229, 0.0508, 0.0441, 0.1091, 0.0488], device='cuda:7'), in_proj_covar=tensor([0.0228, 0.0294, 0.0270, 0.0251, 0.0285, 0.0289, 0.0188, 0.0312], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:13:25,717 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 12:13:36,726 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110639.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:13:41,891 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0075, 2.6168, 2.6651, 1.7957, 2.8086, 2.8533, 2.4597, 2.3917], device='cuda:7'), covar=tensor([0.0696, 0.0194, 0.0166, 0.1007, 0.0070, 0.0170, 0.0386, 0.0457], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0096, 0.0083, 0.0136, 0.0065, 0.0097, 0.0116, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 12:14:06,386 INFO [train.py:904] (7/8) Epoch 11, batch 9150, loss[loss=0.1752, simple_loss=0.2681, pruned_loss=0.04113, over 16288.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2755, pruned_loss=0.04542, over 3059928.79 frames. ], batch size: 166, lr: 6.11e-03, grad_scale: 8.0 2023-04-29 12:14:15,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.395e+02 2.852e+02 3.560e+02 7.952e+02, threshold=5.704e+02, percent-clipped=1.0 2023-04-29 12:14:23,495 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:40,756 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4546, 3.7786, 3.9686, 3.9070, 3.9225, 3.7454, 3.4867, 3.7639], device='cuda:7'), covar=tensor([0.0595, 0.0744, 0.0628, 0.0755, 0.0772, 0.0705, 0.1299, 0.0512], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0316, 0.0320, 0.0300, 0.0362, 0.0334, 0.0430, 0.0271], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0002], device='cuda:7') 2023-04-29 12:14:52,228 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110672.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:54,527 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110673.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:14:56,561 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110674.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:15:46,104 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110700.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:15:48,936 INFO [train.py:904] (7/8) Epoch 11, batch 9200, loss[loss=0.168, simple_loss=0.2493, pruned_loss=0.04335, over 12126.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2707, pruned_loss=0.04434, over 3062826.12 frames. ], batch size: 248, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:16:24,211 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110721.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:16:24,411 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4209, 2.0495, 1.6914, 1.7002, 2.2844, 1.9752, 2.1885, 2.4089], device='cuda:7'), covar=tensor([0.0104, 0.0272, 0.0360, 0.0335, 0.0180, 0.0236, 0.0149, 0.0186], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0198, 0.0192, 0.0192, 0.0195, 0.0196, 0.0193, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:16:53,797 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110736.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:17:18,225 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110748.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:17:24,357 INFO [train.py:904] (7/8) Epoch 11, batch 9250, loss[loss=0.1577, simple_loss=0.2411, pruned_loss=0.03719, over 12317.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2705, pruned_loss=0.04412, over 3072184.93 frames. ], batch size: 247, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:17:32,959 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.583e+02 3.099e+02 3.621e+02 8.759e+02, threshold=6.198e+02, percent-clipped=4.0 2023-04-29 12:17:49,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4705, 2.9646, 3.1406, 1.8430, 2.8248, 2.2750, 3.0615, 3.1404], device='cuda:7'), covar=tensor([0.0301, 0.0690, 0.0495, 0.1878, 0.0736, 0.0900, 0.0708, 0.0797], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0135, 0.0153, 0.0140, 0.0133, 0.0122, 0.0132, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 12:17:53,889 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110766.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:18:05,350 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110771.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:19:06,544 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110797.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 12:19:16,920 INFO [train.py:904] (7/8) Epoch 11, batch 9300, loss[loss=0.1586, simple_loss=0.2521, pruned_loss=0.03254, over 16804.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2691, pruned_loss=0.04373, over 3059434.53 frames. ], batch size: 83, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:19:36,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1241, 4.1660, 4.5900, 4.5663, 4.5291, 4.2901, 4.2759, 4.1514], device='cuda:7'), covar=tensor([0.0268, 0.0508, 0.0313, 0.0345, 0.0501, 0.0319, 0.0751, 0.0363], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0311, 0.0315, 0.0293, 0.0354, 0.0330, 0.0422, 0.0267], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 12:19:45,792 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110814.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:19:59,853 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110819.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:21:01,165 INFO [train.py:904] (7/8) Epoch 11, batch 9350, loss[loss=0.2056, simple_loss=0.2882, pruned_loss=0.06152, over 12453.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2692, pruned_loss=0.04354, over 3077432.07 frames. ], batch size: 248, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:21:10,068 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.415e+02 2.201e+02 2.572e+02 3.084e+02 6.751e+02, threshold=5.144e+02, percent-clipped=1.0 2023-04-29 12:21:11,577 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-29 12:21:36,202 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110868.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:21:44,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8774, 5.2043, 5.3911, 5.1991, 5.2842, 5.8000, 5.3474, 5.0594], device='cuda:7'), covar=tensor([0.0814, 0.1502, 0.1521, 0.1566, 0.2063, 0.0857, 0.1270, 0.2070], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0463, 0.0508, 0.0401, 0.0528, 0.0540, 0.0405, 0.0541], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 12:22:31,173 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5632, 4.8255, 4.6521, 4.6451, 4.3439, 4.2676, 4.3519, 4.9012], device='cuda:7'), covar=tensor([0.1010, 0.0916, 0.0921, 0.0655, 0.0779, 0.1174, 0.0904, 0.0819], device='cuda:7'), in_proj_covar=tensor([0.0507, 0.0629, 0.0517, 0.0437, 0.0394, 0.0413, 0.0526, 0.0480], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:22:41,706 INFO [train.py:904] (7/8) Epoch 11, batch 9400, loss[loss=0.1737, simple_loss=0.2672, pruned_loss=0.04013, over 12453.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2689, pruned_loss=0.04335, over 3057044.06 frames. ], batch size: 248, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:23:11,271 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 12:23:36,515 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110929.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:23:57,892 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110939.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:24:21,281 INFO [train.py:904] (7/8) Epoch 11, batch 9450, loss[loss=0.189, simple_loss=0.2757, pruned_loss=0.05119, over 12769.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2711, pruned_loss=0.04373, over 3053441.27 frames. ], batch size: 248, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:24:27,347 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.422e+02 3.077e+02 3.855e+02 7.743e+02, threshold=6.155e+02, percent-clipped=6.0 2023-04-29 12:24:33,089 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110958.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:24:34,538 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110959.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:24:50,529 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110967.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:25:06,766 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110974.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:25:18,503 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9876, 4.2264, 4.0228, 4.0470, 3.7169, 3.7957, 3.8623, 4.2286], device='cuda:7'), covar=tensor([0.1046, 0.0929, 0.1037, 0.0655, 0.0773, 0.1616, 0.0894, 0.0930], device='cuda:7'), in_proj_covar=tensor([0.0514, 0.0638, 0.0523, 0.0442, 0.0400, 0.0418, 0.0531, 0.0485], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:25:33,577 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=110987.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:26:00,936 INFO [train.py:904] (7/8) Epoch 11, batch 9500, loss[loss=0.1682, simple_loss=0.2595, pruned_loss=0.03839, over 17041.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2707, pruned_loss=0.04356, over 3054424.81 frames. ], batch size: 55, lr: 6.10e-03, grad_scale: 8.0 2023-04-29 12:26:15,566 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111007.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:26:38,140 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111019.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:26:44,355 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111022.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:27:27,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0720, 4.1240, 4.5009, 4.4552, 4.4885, 4.2074, 4.2402, 4.0738], device='cuda:7'), covar=tensor([0.0301, 0.0599, 0.0410, 0.0438, 0.0441, 0.0378, 0.0805, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0302, 0.0313, 0.0316, 0.0295, 0.0357, 0.0333, 0.0424, 0.0268], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 12:27:46,821 INFO [train.py:904] (7/8) Epoch 11, batch 9550, loss[loss=0.2061, simple_loss=0.3078, pruned_loss=0.05215, over 15324.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2705, pruned_loss=0.04362, over 3075218.86 frames. ], batch size: 191, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:27:55,300 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.456e+02 2.878e+02 3.283e+02 5.727e+02, threshold=5.755e+02, percent-clipped=0.0 2023-04-29 12:28:26,862 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6159, 2.7585, 1.8002, 2.8570, 2.1258, 2.8544, 1.9842, 2.4314], device='cuda:7'), covar=tensor([0.0212, 0.0311, 0.1219, 0.0291, 0.0659, 0.0479, 0.1269, 0.0600], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0154, 0.0181, 0.0120, 0.0161, 0.0192, 0.0191, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 12:29:10,608 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111092.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 12:29:26,775 INFO [train.py:904] (7/8) Epoch 11, batch 9600, loss[loss=0.1596, simple_loss=0.2609, pruned_loss=0.02913, over 16920.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2721, pruned_loss=0.04452, over 3072903.25 frames. ], batch size: 96, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:30:14,523 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7027, 2.0206, 1.6522, 1.7252, 2.3663, 2.0535, 2.4440, 2.6020], device='cuda:7'), covar=tensor([0.0101, 0.0373, 0.0445, 0.0443, 0.0237, 0.0346, 0.0146, 0.0198], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0201, 0.0195, 0.0195, 0.0198, 0.0198, 0.0195, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:31:14,972 INFO [train.py:904] (7/8) Epoch 11, batch 9650, loss[loss=0.1919, simple_loss=0.2755, pruned_loss=0.05412, over 12359.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2737, pruned_loss=0.04462, over 3068573.29 frames. ], batch size: 247, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:31:24,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.367e+02 2.756e+02 3.328e+02 5.495e+02, threshold=5.512e+02, percent-clipped=0.0 2023-04-29 12:31:37,196 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.62 vs. limit=5.0 2023-04-29 12:33:03,910 INFO [train.py:904] (7/8) Epoch 11, batch 9700, loss[loss=0.1897, simple_loss=0.2772, pruned_loss=0.05106, over 16863.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.273, pruned_loss=0.04457, over 3081633.20 frames. ], batch size: 124, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:33:08,453 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 12:33:19,417 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6696, 6.0676, 5.8217, 5.8925, 5.4291, 5.2878, 5.5094, 6.1759], device='cuda:7'), covar=tensor([0.1090, 0.0874, 0.1049, 0.0616, 0.0736, 0.0586, 0.0898, 0.0805], device='cuda:7'), in_proj_covar=tensor([0.0512, 0.0631, 0.0517, 0.0439, 0.0395, 0.0413, 0.0527, 0.0482], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:33:41,200 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5323, 4.5151, 4.3149, 3.9653, 4.3474, 1.6834, 4.1500, 4.3151], device='cuda:7'), covar=tensor([0.0082, 0.0081, 0.0180, 0.0243, 0.0097, 0.2273, 0.0125, 0.0156], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0110, 0.0158, 0.0143, 0.0129, 0.0177, 0.0145, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:33:49,195 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111224.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:33:52,525 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111225.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:34:46,297 INFO [train.py:904] (7/8) Epoch 11, batch 9750, loss[loss=0.1925, simple_loss=0.2711, pruned_loss=0.05694, over 12443.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2715, pruned_loss=0.04489, over 3065714.34 frames. ], batch size: 248, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:34:53,748 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.704e+02 2.278e+02 2.798e+02 3.645e+02 8.933e+02, threshold=5.595e+02, percent-clipped=9.0 2023-04-29 12:35:17,236 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111267.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:35:58,487 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111286.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:36:09,554 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1372, 4.1403, 4.5994, 4.5546, 4.5578, 4.2973, 4.2758, 4.1258], device='cuda:7'), covar=tensor([0.0284, 0.0439, 0.0333, 0.0409, 0.0436, 0.0336, 0.0780, 0.0381], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0309, 0.0313, 0.0292, 0.0352, 0.0328, 0.0417, 0.0266], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:7') 2023-04-29 12:36:25,544 INFO [train.py:904] (7/8) Epoch 11, batch 9800, loss[loss=0.156, simple_loss=0.2407, pruned_loss=0.0356, over 12561.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2706, pruned_loss=0.04353, over 3071268.77 frames. ], batch size: 248, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:36:37,435 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 12:36:49,120 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111314.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:36:51,182 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111315.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:38:11,451 INFO [train.py:904] (7/8) Epoch 11, batch 9850, loss[loss=0.1958, simple_loss=0.2904, pruned_loss=0.05056, over 15455.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2724, pruned_loss=0.0437, over 3057756.14 frames. ], batch size: 190, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:38:20,190 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.271e+02 2.860e+02 3.428e+02 8.615e+02, threshold=5.720e+02, percent-clipped=1.0 2023-04-29 12:38:33,660 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-29 12:39:42,374 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111392.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 12:40:02,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9014, 3.8209, 3.9772, 4.1212, 4.2265, 3.8122, 4.2076, 4.2637], device='cuda:7'), covar=tensor([0.1487, 0.0979, 0.1393, 0.0707, 0.0525, 0.1380, 0.0668, 0.0560], device='cuda:7'), in_proj_covar=tensor([0.0486, 0.0606, 0.0728, 0.0616, 0.0469, 0.0477, 0.0492, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:40:02,799 INFO [train.py:904] (7/8) Epoch 11, batch 9900, loss[loss=0.1782, simple_loss=0.2818, pruned_loss=0.03735, over 16208.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2735, pruned_loss=0.04404, over 3045682.80 frames. ], batch size: 165, lr: 6.09e-03, grad_scale: 8.0 2023-04-29 12:41:34,351 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111440.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:42:01,111 INFO [train.py:904] (7/8) Epoch 11, batch 9950, loss[loss=0.1763, simple_loss=0.2725, pruned_loss=0.04009, over 16672.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2754, pruned_loss=0.04438, over 3048856.59 frames. ], batch size: 57, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:42:11,465 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.371e+02 2.780e+02 3.480e+02 6.168e+02, threshold=5.560e+02, percent-clipped=1.0 2023-04-29 12:42:26,791 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-29 12:42:41,120 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111468.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:43:52,424 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 12:44:02,038 INFO [train.py:904] (7/8) Epoch 11, batch 10000, loss[loss=0.1936, simple_loss=0.2921, pruned_loss=0.04757, over 16798.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2742, pruned_loss=0.04398, over 3066521.37 frames. ], batch size: 124, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:44:45,277 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111524.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:44:53,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9776, 4.2317, 4.0742, 4.0954, 3.7437, 3.8156, 3.8353, 4.2191], device='cuda:7'), covar=tensor([0.0882, 0.0792, 0.0785, 0.0588, 0.0685, 0.1368, 0.0791, 0.0870], device='cuda:7'), in_proj_covar=tensor([0.0505, 0.0627, 0.0508, 0.0434, 0.0392, 0.0411, 0.0521, 0.0478], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:44:55,738 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111529.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:45:18,965 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9992, 4.0472, 3.8347, 3.6194, 3.6046, 3.9847, 3.6790, 3.7100], device='cuda:7'), covar=tensor([0.0492, 0.0420, 0.0255, 0.0241, 0.0602, 0.0393, 0.0847, 0.0479], device='cuda:7'), in_proj_covar=tensor([0.0227, 0.0291, 0.0267, 0.0248, 0.0284, 0.0286, 0.0186, 0.0307], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:45:42,137 INFO [train.py:904] (7/8) Epoch 11, batch 10050, loss[loss=0.1738, simple_loss=0.2677, pruned_loss=0.04, over 16635.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2744, pruned_loss=0.04376, over 3077581.25 frames. ], batch size: 76, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:45:45,957 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-04-29 12:45:50,240 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.160e+02 2.693e+02 3.599e+02 6.171e+02, threshold=5.385e+02, percent-clipped=1.0 2023-04-29 12:46:22,430 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111572.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:46:38,799 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111581.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:47:14,834 INFO [train.py:904] (7/8) Epoch 11, batch 10100, loss[loss=0.1919, simple_loss=0.2759, pruned_loss=0.05397, over 16171.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2749, pruned_loss=0.04424, over 3073723.63 frames. ], batch size: 165, lr: 6.08e-03, grad_scale: 8.0 2023-04-29 12:47:35,996 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111614.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:48:15,269 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.89 vs. limit=5.0 2023-04-29 12:48:58,022 INFO [train.py:904] (7/8) Epoch 12, batch 0, loss[loss=0.227, simple_loss=0.3072, pruned_loss=0.07334, over 16626.00 frames. ], tot_loss[loss=0.227, simple_loss=0.3072, pruned_loss=0.07334, over 16626.00 frames. ], batch size: 62, lr: 5.82e-03, grad_scale: 8.0 2023-04-29 12:48:58,023 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 12:49:05,316 INFO [train.py:938] (7/8) Epoch 12, validation: loss=0.1543, simple_loss=0.2577, pruned_loss=0.0254, over 944034.00 frames. 2023-04-29 12:49:05,316 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 12:49:12,538 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.560e+02 3.107e+02 3.982e+02 7.820e+02, threshold=6.214e+02, percent-clipped=3.0 2023-04-29 12:49:21,580 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111662.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:49:22,845 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4049, 4.4372, 4.5561, 4.4380, 4.4205, 4.9817, 4.5783, 4.2994], device='cuda:7'), covar=tensor([0.1474, 0.2129, 0.2189, 0.2243, 0.2651, 0.1147, 0.1702, 0.2612], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0454, 0.0497, 0.0394, 0.0519, 0.0532, 0.0400, 0.0527], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:50:01,822 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7145, 4.0528, 4.3349, 2.8761, 3.6445, 4.1899, 3.8306, 2.5580], device='cuda:7'), covar=tensor([0.0423, 0.0043, 0.0027, 0.0311, 0.0082, 0.0070, 0.0057, 0.0353], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0066, 0.0067, 0.0124, 0.0076, 0.0085, 0.0075, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 12:50:15,968 INFO [train.py:904] (7/8) Epoch 12, batch 50, loss[loss=0.2165, simple_loss=0.2828, pruned_loss=0.07511, over 16902.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2907, pruned_loss=0.06691, over 737105.58 frames. ], batch size: 109, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:51:25,699 INFO [train.py:904] (7/8) Epoch 12, batch 100, loss[loss=0.177, simple_loss=0.2611, pruned_loss=0.04639, over 17188.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.283, pruned_loss=0.06103, over 1310162.26 frames. ], batch size: 46, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:51:34,345 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.444e+02 2.855e+02 3.643e+02 7.519e+02, threshold=5.710e+02, percent-clipped=2.0 2023-04-29 12:52:18,933 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111791.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 12:52:31,943 INFO [train.py:904] (7/8) Epoch 12, batch 150, loss[loss=0.2309, simple_loss=0.3006, pruned_loss=0.08062, over 16673.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2814, pruned_loss=0.0596, over 1760928.09 frames. ], batch size: 134, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:53:03,092 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111824.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:53:29,047 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.01 vs. limit=2.0 2023-04-29 12:53:33,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8592, 4.1119, 2.4909, 4.6259, 3.1588, 4.5366, 2.7366, 3.3733], device='cuda:7'), covar=tensor([0.0264, 0.0318, 0.1462, 0.0202, 0.0709, 0.0471, 0.1320, 0.0625], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0162, 0.0189, 0.0127, 0.0168, 0.0202, 0.0198, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 12:53:40,728 INFO [train.py:904] (7/8) Epoch 12, batch 200, loss[loss=0.2249, simple_loss=0.2892, pruned_loss=0.08033, over 16838.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2809, pruned_loss=0.05876, over 2107822.09 frames. ], batch size: 96, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:53:41,220 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111852.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 12:53:50,240 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.566e+02 3.072e+02 3.744e+02 9.632e+02, threshold=6.144e+02, percent-clipped=5.0 2023-04-29 12:54:21,387 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111881.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:54:49,918 INFO [train.py:904] (7/8) Epoch 12, batch 250, loss[loss=0.2128, simple_loss=0.2798, pruned_loss=0.07295, over 16912.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2785, pruned_loss=0.05901, over 2381800.40 frames. ], batch size: 109, lr: 5.82e-03, grad_scale: 2.0 2023-04-29 12:54:50,751 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-04-29 12:55:27,221 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=111929.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:55:28,553 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0011, 4.9249, 4.7953, 4.4659, 4.4476, 4.8448, 4.7901, 4.4864], device='cuda:7'), covar=tensor([0.0626, 0.0560, 0.0276, 0.0282, 0.0966, 0.0383, 0.0361, 0.0677], device='cuda:7'), in_proj_covar=tensor([0.0244, 0.0316, 0.0288, 0.0268, 0.0308, 0.0311, 0.0199, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 12:55:57,761 INFO [train.py:904] (7/8) Epoch 12, batch 300, loss[loss=0.1908, simple_loss=0.2627, pruned_loss=0.0594, over 16704.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2754, pruned_loss=0.05719, over 2598841.42 frames. ], batch size: 134, lr: 5.82e-03, grad_scale: 1.0 2023-04-29 12:56:09,471 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.376e+02 2.773e+02 3.192e+02 6.381e+02, threshold=5.545e+02, percent-clipped=1.0 2023-04-29 12:57:10,670 INFO [train.py:904] (7/8) Epoch 12, batch 350, loss[loss=0.1782, simple_loss=0.2627, pruned_loss=0.04679, over 17156.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2726, pruned_loss=0.05495, over 2765110.04 frames. ], batch size: 46, lr: 5.81e-03, grad_scale: 1.0 2023-04-29 12:58:17,747 INFO [train.py:904] (7/8) Epoch 12, batch 400, loss[loss=0.1909, simple_loss=0.261, pruned_loss=0.06043, over 16358.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2715, pruned_loss=0.05466, over 2887467.42 frames. ], batch size: 146, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 12:58:27,674 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 2.508e+02 2.981e+02 3.635e+02 6.653e+02, threshold=5.962e+02, percent-clipped=1.0 2023-04-29 12:58:37,593 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112066.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 12:58:53,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-29 12:59:26,048 INFO [train.py:904] (7/8) Epoch 12, batch 450, loss[loss=0.1909, simple_loss=0.278, pruned_loss=0.05184, over 16630.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2691, pruned_loss=0.05323, over 2994591.67 frames. ], batch size: 62, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 12:59:47,102 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7780, 5.0975, 5.2073, 5.0349, 5.0266, 5.6278, 5.1276, 4.8216], device='cuda:7'), covar=tensor([0.1079, 0.1807, 0.2134, 0.2176, 0.3184, 0.1185, 0.1557, 0.2446], device='cuda:7'), in_proj_covar=tensor([0.0351, 0.0501, 0.0550, 0.0435, 0.0581, 0.0579, 0.0436, 0.0583], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 12:59:56,558 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112124.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:00:00,704 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112127.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:00:27,070 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112147.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 13:00:33,824 INFO [train.py:904] (7/8) Epoch 12, batch 500, loss[loss=0.2083, simple_loss=0.2773, pruned_loss=0.0697, over 12078.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2671, pruned_loss=0.05229, over 3054978.37 frames. ], batch size: 247, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:00:45,214 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.270e+02 2.759e+02 3.532e+02 6.724e+02, threshold=5.519e+02, percent-clipped=2.0 2023-04-29 13:01:02,760 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=112172.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:01:19,701 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 13:01:30,076 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 13:01:32,814 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1355, 4.1507, 4.6059, 4.5762, 4.6115, 4.2902, 4.3066, 4.1750], device='cuda:7'), covar=tensor([0.0349, 0.0608, 0.0395, 0.0447, 0.0394, 0.0385, 0.0807, 0.0502], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0346, 0.0348, 0.0328, 0.0389, 0.0367, 0.0468, 0.0298], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 13:01:44,746 INFO [train.py:904] (7/8) Epoch 12, batch 550, loss[loss=0.204, simple_loss=0.2766, pruned_loss=0.06574, over 16860.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2649, pruned_loss=0.05075, over 3121110.73 frames. ], batch size: 96, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:02:55,017 INFO [train.py:904] (7/8) Epoch 12, batch 600, loss[loss=0.1747, simple_loss=0.2592, pruned_loss=0.04512, over 16643.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2637, pruned_loss=0.05048, over 3170015.13 frames. ], batch size: 62, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:03:06,855 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.658e+02 2.365e+02 2.773e+02 3.420e+02 1.272e+03, threshold=5.547e+02, percent-clipped=1.0 2023-04-29 13:04:04,841 INFO [train.py:904] (7/8) Epoch 12, batch 650, loss[loss=0.1549, simple_loss=0.2464, pruned_loss=0.03169, over 17121.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2619, pruned_loss=0.04994, over 3208662.82 frames. ], batch size: 49, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:04:50,467 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 13:05:04,988 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1060, 2.0093, 1.7535, 1.7905, 2.3215, 2.0231, 2.1792, 2.4373], device='cuda:7'), covar=tensor([0.0167, 0.0282, 0.0341, 0.0324, 0.0170, 0.0233, 0.0181, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0209, 0.0202, 0.0203, 0.0209, 0.0208, 0.0212, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:05:14,237 INFO [train.py:904] (7/8) Epoch 12, batch 700, loss[loss=0.2122, simple_loss=0.278, pruned_loss=0.07313, over 16778.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2624, pruned_loss=0.04997, over 3231497.63 frames. ], batch size: 116, lr: 5.81e-03, grad_scale: 2.0 2023-04-29 13:05:26,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.678e+02 2.509e+02 2.916e+02 3.539e+02 5.225e+02, threshold=5.832e+02, percent-clipped=0.0 2023-04-29 13:06:07,749 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112389.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:06:24,428 INFO [train.py:904] (7/8) Epoch 12, batch 750, loss[loss=0.2096, simple_loss=0.2774, pruned_loss=0.07092, over 16581.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2629, pruned_loss=0.04982, over 3257613.84 frames. ], batch size: 146, lr: 5.80e-03, grad_scale: 2.0 2023-04-29 13:06:52,553 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112422.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:07:18,363 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0164, 5.3864, 5.1425, 5.1633, 4.8876, 4.7541, 4.8402, 5.4442], device='cuda:7'), covar=tensor([0.1120, 0.0803, 0.0913, 0.0643, 0.0789, 0.0955, 0.0965, 0.0958], device='cuda:7'), in_proj_covar=tensor([0.0565, 0.0701, 0.0572, 0.0488, 0.0440, 0.0450, 0.0586, 0.0537], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:07:20,822 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-29 13:07:27,218 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1247, 5.5527, 5.7347, 5.5583, 5.5158, 6.1228, 5.6113, 5.3669], device='cuda:7'), covar=tensor([0.0829, 0.1742, 0.1984, 0.1838, 0.2931, 0.0909, 0.1392, 0.2384], device='cuda:7'), in_proj_covar=tensor([0.0355, 0.0508, 0.0557, 0.0441, 0.0588, 0.0584, 0.0437, 0.0588], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 13:07:29,124 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112447.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 13:07:32,800 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112450.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:07:34,642 INFO [train.py:904] (7/8) Epoch 12, batch 800, loss[loss=0.1963, simple_loss=0.2654, pruned_loss=0.06364, over 12051.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2621, pruned_loss=0.04937, over 3269822.07 frames. ], batch size: 246, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:07:45,052 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.488e+02 3.052e+02 3.720e+02 1.064e+03, threshold=6.105e+02, percent-clipped=2.0 2023-04-29 13:08:33,921 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=112495.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 13:08:42,881 INFO [train.py:904] (7/8) Epoch 12, batch 850, loss[loss=0.2072, simple_loss=0.2773, pruned_loss=0.06851, over 15461.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.262, pruned_loss=0.04942, over 3284056.64 frames. ], batch size: 191, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:09:52,006 INFO [train.py:904] (7/8) Epoch 12, batch 900, loss[loss=0.2002, simple_loss=0.2724, pruned_loss=0.06399, over 16958.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2609, pruned_loss=0.04854, over 3291118.01 frames. ], batch size: 109, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:09:53,362 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 13:10:02,365 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.271e+02 2.835e+02 3.426e+02 5.348e+02, threshold=5.671e+02, percent-clipped=0.0 2023-04-29 13:10:05,767 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112562.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:10:26,193 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:10:59,467 INFO [train.py:904] (7/8) Epoch 12, batch 950, loss[loss=0.1593, simple_loss=0.244, pruned_loss=0.03735, over 16830.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2607, pruned_loss=0.0486, over 3292518.23 frames. ], batch size: 42, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:11:01,695 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5630, 5.4390, 5.3815, 4.9835, 4.9846, 5.4529, 5.4034, 5.0825], device='cuda:7'), covar=tensor([0.0496, 0.0351, 0.0245, 0.0267, 0.0984, 0.0314, 0.0261, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0336, 0.0307, 0.0286, 0.0330, 0.0331, 0.0211, 0.0357], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:11:21,360 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112617.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:11:28,383 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112623.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:11:50,189 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112638.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 13:11:54,749 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7793, 3.9885, 1.9590, 4.4296, 2.8154, 4.2742, 2.1230, 3.0855], device='cuda:7'), covar=tensor([0.0248, 0.0283, 0.1867, 0.0169, 0.0740, 0.0440, 0.1784, 0.0632], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0166, 0.0189, 0.0133, 0.0166, 0.0207, 0.0198, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 13:12:07,860 INFO [train.py:904] (7/8) Epoch 12, batch 1000, loss[loss=0.1736, simple_loss=0.2419, pruned_loss=0.05267, over 16747.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.26, pruned_loss=0.04855, over 3299193.28 frames. ], batch size: 83, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:12:18,368 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.238e+02 2.624e+02 3.056e+02 7.407e+02, threshold=5.248e+02, percent-clipped=1.0 2023-04-29 13:12:43,299 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112678.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 13:13:15,640 INFO [train.py:904] (7/8) Epoch 12, batch 1050, loss[loss=0.1846, simple_loss=0.2664, pruned_loss=0.05138, over 16469.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2603, pruned_loss=0.0482, over 3305768.72 frames. ], batch size: 68, lr: 5.80e-03, grad_scale: 4.0 2023-04-29 13:13:42,890 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112722.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:14:14,576 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112745.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:14:23,344 INFO [train.py:904] (7/8) Epoch 12, batch 1100, loss[loss=0.1676, simple_loss=0.246, pruned_loss=0.04464, over 15370.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2596, pruned_loss=0.04813, over 3302764.45 frames. ], batch size: 190, lr: 5.79e-03, grad_scale: 4.0 2023-04-29 13:14:34,075 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.374e+02 2.893e+02 3.790e+02 9.509e+02, threshold=5.785e+02, percent-clipped=6.0 2023-04-29 13:14:43,329 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 13:14:48,742 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=112770.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:15:33,618 INFO [train.py:904] (7/8) Epoch 12, batch 1150, loss[loss=0.1561, simple_loss=0.2363, pruned_loss=0.03799, over 15771.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2582, pruned_loss=0.04759, over 3306235.93 frames. ], batch size: 35, lr: 5.79e-03, grad_scale: 4.0 2023-04-29 13:15:57,896 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3870, 2.5641, 2.1496, 2.2971, 2.8637, 2.6101, 3.1836, 3.0551], device='cuda:7'), covar=tensor([0.0113, 0.0302, 0.0387, 0.0360, 0.0214, 0.0302, 0.0201, 0.0207], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0215, 0.0206, 0.0207, 0.0213, 0.0213, 0.0217, 0.0203], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:16:42,808 INFO [train.py:904] (7/8) Epoch 12, batch 1200, loss[loss=0.2063, simple_loss=0.2723, pruned_loss=0.07012, over 16815.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2574, pruned_loss=0.04713, over 3303429.37 frames. ], batch size: 109, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:16:52,677 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.211e+02 2.743e+02 3.250e+02 7.821e+02, threshold=5.486e+02, percent-clipped=2.0 2023-04-29 13:17:49,339 INFO [train.py:904] (7/8) Epoch 12, batch 1250, loss[loss=0.1694, simple_loss=0.2618, pruned_loss=0.03855, over 17079.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2582, pruned_loss=0.04831, over 3297858.50 frames. ], batch size: 47, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:17:58,036 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6296, 4.6102, 4.5390, 3.9930, 4.5017, 1.9539, 4.3112, 4.3138], device='cuda:7'), covar=tensor([0.0088, 0.0082, 0.0138, 0.0288, 0.0083, 0.2242, 0.0126, 0.0170], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0124, 0.0176, 0.0162, 0.0144, 0.0191, 0.0162, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:18:12,326 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112918.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:18:16,624 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112921.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:18:32,357 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112933.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 13:18:57,928 INFO [train.py:904] (7/8) Epoch 12, batch 1300, loss[loss=0.1898, simple_loss=0.2805, pruned_loss=0.04959, over 17097.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2577, pruned_loss=0.04809, over 3301761.62 frames. ], batch size: 53, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:19:09,588 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.880e+02 2.412e+02 2.831e+02 3.329e+02 6.613e+02, threshold=5.661e+02, percent-clipped=2.0 2023-04-29 13:19:14,695 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3422, 2.0650, 2.1372, 4.0018, 2.0535, 2.5091, 2.1589, 2.2798], device='cuda:7'), covar=tensor([0.1082, 0.3523, 0.2522, 0.0440, 0.3719, 0.2280, 0.3389, 0.2951], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0397, 0.0336, 0.0324, 0.0416, 0.0455, 0.0361, 0.0465], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:19:27,116 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112973.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 13:19:41,128 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112982.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:20:08,547 INFO [train.py:904] (7/8) Epoch 12, batch 1350, loss[loss=0.1787, simple_loss=0.2713, pruned_loss=0.04306, over 17098.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2585, pruned_loss=0.04809, over 3300058.37 frames. ], batch size: 49, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:20:40,120 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1288, 2.5523, 2.1000, 2.4282, 2.9015, 2.6283, 3.0205, 3.0201], device='cuda:7'), covar=tensor([0.0128, 0.0272, 0.0370, 0.0323, 0.0183, 0.0264, 0.0280, 0.0217], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0213, 0.0205, 0.0204, 0.0212, 0.0211, 0.0217, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:21:08,007 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113045.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:21:16,378 INFO [train.py:904] (7/8) Epoch 12, batch 1400, loss[loss=0.1644, simple_loss=0.2571, pruned_loss=0.0358, over 17025.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2592, pruned_loss=0.04838, over 3299201.74 frames. ], batch size: 50, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:21:22,092 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7952, 4.8750, 5.4039, 5.3273, 5.3514, 4.9961, 4.9637, 4.6923], device='cuda:7'), covar=tensor([0.0306, 0.0510, 0.0318, 0.0454, 0.0406, 0.0347, 0.0843, 0.0414], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0359, 0.0360, 0.0339, 0.0403, 0.0380, 0.0480, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 13:21:26,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 2.440e+02 3.018e+02 3.818e+02 8.239e+02, threshold=6.035e+02, percent-clipped=8.0 2023-04-29 13:22:13,296 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113093.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:22:24,199 INFO [train.py:904] (7/8) Epoch 12, batch 1450, loss[loss=0.167, simple_loss=0.2339, pruned_loss=0.05008, over 16807.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.259, pruned_loss=0.04864, over 3312532.68 frames. ], batch size: 96, lr: 5.79e-03, grad_scale: 8.0 2023-04-29 13:23:35,120 INFO [train.py:904] (7/8) Epoch 12, batch 1500, loss[loss=0.1686, simple_loss=0.2564, pruned_loss=0.04043, over 17094.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2592, pruned_loss=0.04843, over 3316873.70 frames. ], batch size: 47, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:23:45,776 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.499e+02 2.892e+02 3.478e+02 9.992e+02, threshold=5.783e+02, percent-clipped=3.0 2023-04-29 13:24:26,592 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6679, 3.0080, 2.9290, 5.0103, 4.1729, 4.6128, 1.6562, 3.3983], device='cuda:7'), covar=tensor([0.1401, 0.0679, 0.0997, 0.0133, 0.0223, 0.0315, 0.1434, 0.0650], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0160, 0.0181, 0.0149, 0.0194, 0.0212, 0.0182, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 13:24:31,608 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5633, 3.4131, 3.9163, 2.8542, 3.5597, 3.9465, 3.6826, 2.5296], device='cuda:7'), covar=tensor([0.0385, 0.0221, 0.0044, 0.0260, 0.0092, 0.0088, 0.0074, 0.0324], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0073, 0.0072, 0.0128, 0.0081, 0.0092, 0.0081, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 13:24:43,453 INFO [train.py:904] (7/8) Epoch 12, batch 1550, loss[loss=0.228, simple_loss=0.3008, pruned_loss=0.07763, over 15388.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2609, pruned_loss=0.0499, over 3312218.17 frames. ], batch size: 191, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:25:06,608 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113218.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:25:17,720 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2023-04-29 13:25:21,530 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4213, 4.4026, 4.8980, 4.8936, 4.9100, 4.5778, 4.5699, 4.3667], device='cuda:7'), covar=tensor([0.0374, 0.0705, 0.0400, 0.0428, 0.0466, 0.0402, 0.0876, 0.0591], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0360, 0.0362, 0.0340, 0.0404, 0.0383, 0.0484, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 13:25:27,325 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113233.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:25:29,536 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113234.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:25:54,139 INFO [train.py:904] (7/8) Epoch 12, batch 1600, loss[loss=0.1936, simple_loss=0.276, pruned_loss=0.05562, over 16579.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2626, pruned_loss=0.04979, over 3311306.36 frames. ], batch size: 76, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:26:04,706 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.814e+02 2.404e+02 2.815e+02 3.461e+02 5.184e+02, threshold=5.631e+02, percent-clipped=0.0 2023-04-29 13:26:12,639 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113266.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:26:23,361 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113273.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 13:26:28,589 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113277.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:26:34,041 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113281.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:26:53,627 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113295.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:27:02,736 INFO [train.py:904] (7/8) Epoch 12, batch 1650, loss[loss=0.1606, simple_loss=0.2535, pruned_loss=0.03386, over 17240.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2643, pruned_loss=0.05068, over 3301849.80 frames. ], batch size: 44, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:27:18,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0421, 4.1100, 3.9688, 3.8394, 3.3934, 4.2165, 3.9567, 3.8678], device='cuda:7'), covar=tensor([0.0904, 0.0730, 0.0479, 0.0383, 0.1489, 0.0494, 0.0934, 0.0679], device='cuda:7'), in_proj_covar=tensor([0.0261, 0.0341, 0.0311, 0.0288, 0.0330, 0.0335, 0.0213, 0.0360], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 13:27:29,756 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:28:12,400 INFO [train.py:904] (7/8) Epoch 12, batch 1700, loss[loss=0.164, simple_loss=0.2495, pruned_loss=0.03928, over 16842.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2659, pruned_loss=0.05119, over 3304784.20 frames. ], batch size: 39, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:28:23,616 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.537e+02 3.070e+02 3.736e+02 6.116e+02, threshold=6.140e+02, percent-clipped=1.0 2023-04-29 13:29:22,275 INFO [train.py:904] (7/8) Epoch 12, batch 1750, loss[loss=0.1809, simple_loss=0.2601, pruned_loss=0.05086, over 16724.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2662, pruned_loss=0.05074, over 3309012.78 frames. ], batch size: 89, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:30:32,326 INFO [train.py:904] (7/8) Epoch 12, batch 1800, loss[loss=0.1747, simple_loss=0.2541, pruned_loss=0.04762, over 16788.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2668, pruned_loss=0.05063, over 3311677.25 frames. ], batch size: 39, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:30:43,448 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.362e+02 2.904e+02 3.594e+02 5.616e+02, threshold=5.809e+02, percent-clipped=0.0 2023-04-29 13:31:18,399 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1412, 5.2070, 5.6811, 5.6533, 5.6634, 5.2471, 5.2055, 5.0145], device='cuda:7'), covar=tensor([0.0277, 0.0501, 0.0277, 0.0366, 0.0439, 0.0314, 0.0906, 0.0409], device='cuda:7'), in_proj_covar=tensor([0.0346, 0.0358, 0.0360, 0.0339, 0.0402, 0.0382, 0.0482, 0.0307], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 13:31:42,921 INFO [train.py:904] (7/8) Epoch 12, batch 1850, loss[loss=0.1611, simple_loss=0.2474, pruned_loss=0.0374, over 17005.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2679, pruned_loss=0.05101, over 3302816.28 frames. ], batch size: 41, lr: 5.78e-03, grad_scale: 8.0 2023-04-29 13:32:03,292 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7326, 2.6675, 2.1866, 2.4772, 2.9235, 2.7813, 3.5160, 3.2507], device='cuda:7'), covar=tensor([0.0086, 0.0289, 0.0405, 0.0354, 0.0216, 0.0296, 0.0178, 0.0194], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0213, 0.0207, 0.0207, 0.0213, 0.0213, 0.0220, 0.0203], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:32:29,575 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3334, 2.2082, 1.7372, 1.9501, 2.5289, 2.3196, 2.6159, 2.7241], device='cuda:7'), covar=tensor([0.0137, 0.0275, 0.0393, 0.0358, 0.0167, 0.0243, 0.0173, 0.0204], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0215, 0.0208, 0.0207, 0.0214, 0.0214, 0.0222, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:32:53,600 INFO [train.py:904] (7/8) Epoch 12, batch 1900, loss[loss=0.1725, simple_loss=0.2663, pruned_loss=0.03937, over 17123.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2662, pruned_loss=0.04964, over 3299583.22 frames. ], batch size: 47, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:33:04,794 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.197e+02 2.521e+02 3.135e+02 9.394e+02, threshold=5.041e+02, percent-clipped=1.0 2023-04-29 13:33:28,986 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 13:33:29,865 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:33:30,188 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 13:33:49,014 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113590.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:34:05,636 INFO [train.py:904] (7/8) Epoch 12, batch 1950, loss[loss=0.2077, simple_loss=0.2745, pruned_loss=0.07043, over 16883.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2662, pruned_loss=0.04903, over 3298837.26 frames. ], batch size: 96, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:34:06,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9067, 4.6595, 4.9100, 5.1522, 5.3264, 4.6357, 5.2590, 5.2662], device='cuda:7'), covar=tensor([0.1471, 0.1214, 0.1647, 0.0621, 0.0475, 0.1032, 0.0542, 0.0555], device='cuda:7'), in_proj_covar=tensor([0.0571, 0.0712, 0.0859, 0.0727, 0.0546, 0.0561, 0.0563, 0.0651], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:34:27,505 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113617.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:34:39,274 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:35:08,081 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113646.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:35:13,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5048, 2.9511, 3.0299, 2.0056, 2.7521, 2.2082, 3.1273, 3.1592], device='cuda:7'), covar=tensor([0.0275, 0.0745, 0.0527, 0.1658, 0.0764, 0.0908, 0.0589, 0.0786], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0147, 0.0160, 0.0146, 0.0137, 0.0125, 0.0138, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 13:35:16,112 INFO [train.py:904] (7/8) Epoch 12, batch 2000, loss[loss=0.1721, simple_loss=0.2726, pruned_loss=0.03576, over 17031.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2657, pruned_loss=0.04847, over 3302057.36 frames. ], batch size: 50, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:35:27,902 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.257e+02 2.755e+02 3.569e+02 6.259e+02, threshold=5.509e+02, percent-clipped=3.0 2023-04-29 13:35:52,950 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113678.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:36:25,552 INFO [train.py:904] (7/8) Epoch 12, batch 2050, loss[loss=0.1654, simple_loss=0.251, pruned_loss=0.03984, over 16835.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2657, pruned_loss=0.0486, over 3305704.10 frames. ], batch size: 42, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:36:32,488 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:37:33,731 INFO [train.py:904] (7/8) Epoch 12, batch 2100, loss[loss=0.1542, simple_loss=0.2423, pruned_loss=0.03305, over 17175.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2678, pruned_loss=0.04974, over 3313091.55 frames. ], batch size: 40, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:37:39,837 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 13:37:45,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.602e+02 3.022e+02 3.669e+02 5.748e+02, threshold=6.044e+02, percent-clipped=1.0 2023-04-29 13:38:01,584 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4053, 3.6059, 3.2633, 1.9710, 2.7967, 2.3005, 3.7820, 3.8729], device='cuda:7'), covar=tensor([0.0197, 0.0636, 0.0636, 0.1939, 0.0881, 0.0997, 0.0468, 0.0730], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0147, 0.0160, 0.0145, 0.0137, 0.0125, 0.0138, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 13:38:27,584 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9512, 4.7306, 4.9875, 5.1971, 5.3389, 4.7254, 5.3367, 5.2859], device='cuda:7'), covar=tensor([0.1470, 0.1078, 0.1489, 0.0581, 0.0476, 0.0878, 0.0407, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0567, 0.0708, 0.0854, 0.0722, 0.0541, 0.0556, 0.0561, 0.0648], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:38:44,796 INFO [train.py:904] (7/8) Epoch 12, batch 2150, loss[loss=0.1919, simple_loss=0.2848, pruned_loss=0.04947, over 16735.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2687, pruned_loss=0.05077, over 3318861.95 frames. ], batch size: 57, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:39:52,653 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-04-29 13:39:54,125 INFO [train.py:904] (7/8) Epoch 12, batch 2200, loss[loss=0.2081, simple_loss=0.2779, pruned_loss=0.0691, over 16811.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2692, pruned_loss=0.05091, over 3316621.56 frames. ], batch size: 116, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:40:05,118 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.389e+02 2.779e+02 3.462e+02 6.586e+02, threshold=5.558e+02, percent-clipped=1.0 2023-04-29 13:40:43,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6147, 3.6964, 2.9600, 2.1541, 2.5639, 2.2497, 3.8862, 3.3985], device='cuda:7'), covar=tensor([0.2558, 0.0624, 0.1481, 0.2600, 0.2480, 0.1854, 0.0508, 0.1194], device='cuda:7'), in_proj_covar=tensor([0.0299, 0.0256, 0.0281, 0.0277, 0.0279, 0.0222, 0.0266, 0.0299], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:40:48,984 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113890.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:41:03,706 INFO [train.py:904] (7/8) Epoch 12, batch 2250, loss[loss=0.2147, simple_loss=0.2849, pruned_loss=0.07229, over 16809.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2696, pruned_loss=0.05097, over 3311193.72 frames. ], batch size: 124, lr: 5.77e-03, grad_scale: 8.0 2023-04-29 13:41:54,645 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=113938.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:42:14,197 INFO [train.py:904] (7/8) Epoch 12, batch 2300, loss[loss=0.205, simple_loss=0.2766, pruned_loss=0.06669, over 16325.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2696, pruned_loss=0.05143, over 3303568.59 frames. ], batch size: 165, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:42:24,219 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.413e+02 2.819e+02 3.398e+02 7.599e+02, threshold=5.637e+02, percent-clipped=2.0 2023-04-29 13:42:38,151 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 13:42:43,128 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113973.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:43:01,618 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2023-04-29 13:43:25,584 INFO [train.py:904] (7/8) Epoch 12, batch 2350, loss[loss=0.167, simple_loss=0.2561, pruned_loss=0.03899, over 16766.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2705, pruned_loss=0.05218, over 3300923.85 frames. ], batch size: 39, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:43:25,967 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:43:26,190 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0661, 1.9728, 2.4517, 2.9906, 2.8891, 3.4269, 2.2660, 3.2681], device='cuda:7'), covar=tensor([0.0159, 0.0361, 0.0240, 0.0206, 0.0211, 0.0114, 0.0327, 0.0123], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0177, 0.0158, 0.0164, 0.0172, 0.0129, 0.0175, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 13:44:35,988 INFO [train.py:904] (7/8) Epoch 12, batch 2400, loss[loss=0.2296, simple_loss=0.3131, pruned_loss=0.07309, over 11754.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2711, pruned_loss=0.05224, over 3305136.39 frames. ], batch size: 247, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:44:48,487 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.062e+02 2.572e+02 3.152e+02 4.032e+02 9.919e+02, threshold=6.305e+02, percent-clipped=6.0 2023-04-29 13:45:28,503 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 13:45:49,051 INFO [train.py:904] (7/8) Epoch 12, batch 2450, loss[loss=0.1801, simple_loss=0.2667, pruned_loss=0.04675, over 16813.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2714, pruned_loss=0.0516, over 3310931.11 frames. ], batch size: 42, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:46:23,352 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114127.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:46:55,429 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4114, 5.8076, 5.5415, 5.6145, 5.1793, 5.1899, 5.2618, 5.9384], device='cuda:7'), covar=tensor([0.1242, 0.0964, 0.1078, 0.0686, 0.0928, 0.0682, 0.0969, 0.0938], device='cuda:7'), in_proj_covar=tensor([0.0575, 0.0715, 0.0581, 0.0500, 0.0448, 0.0455, 0.0592, 0.0548], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:46:57,386 INFO [train.py:904] (7/8) Epoch 12, batch 2500, loss[loss=0.1846, simple_loss=0.2762, pruned_loss=0.04647, over 17045.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2703, pruned_loss=0.05058, over 3313258.90 frames. ], batch size: 50, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:47:09,686 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.290e+02 2.668e+02 3.158e+02 5.614e+02, threshold=5.335e+02, percent-clipped=0.0 2023-04-29 13:47:48,656 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114188.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:48:06,983 INFO [train.py:904] (7/8) Epoch 12, batch 2550, loss[loss=0.189, simple_loss=0.2664, pruned_loss=0.05581, over 16766.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2713, pruned_loss=0.05145, over 3322011.54 frames. ], batch size: 83, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:49:15,403 INFO [train.py:904] (7/8) Epoch 12, batch 2600, loss[loss=0.1827, simple_loss=0.2837, pruned_loss=0.04087, over 17247.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2713, pruned_loss=0.05121, over 3323298.44 frames. ], batch size: 52, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:49:18,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6937, 2.9581, 2.6684, 4.6713, 3.7447, 4.0753, 1.4849, 3.0465], device='cuda:7'), covar=tensor([0.1500, 0.0735, 0.1202, 0.0178, 0.0341, 0.0473, 0.1726, 0.0829], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0160, 0.0180, 0.0151, 0.0196, 0.0212, 0.0181, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 13:49:24,598 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 13:49:25,936 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.446e+02 2.880e+02 3.491e+02 5.288e+02, threshold=5.760e+02, percent-clipped=0.0 2023-04-29 13:49:46,569 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114273.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:50:24,316 INFO [train.py:904] (7/8) Epoch 12, batch 2650, loss[loss=0.1901, simple_loss=0.2716, pruned_loss=0.05429, over 16863.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2721, pruned_loss=0.05117, over 3326684.67 frames. ], batch size: 96, lr: 5.76e-03, grad_scale: 16.0 2023-04-29 13:50:25,261 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114302.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:50:51,495 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=114321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:50:57,193 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6254, 2.9078, 2.4370, 4.2201, 3.4807, 4.1279, 1.4861, 2.8015], device='cuda:7'), covar=tensor([0.1349, 0.0558, 0.1088, 0.0145, 0.0205, 0.0351, 0.1399, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0160, 0.0181, 0.0152, 0.0197, 0.0212, 0.0182, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 13:51:06,172 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8010, 5.1536, 4.8816, 4.9670, 4.6508, 4.6198, 4.6495, 5.2163], device='cuda:7'), covar=tensor([0.1064, 0.0872, 0.1015, 0.0664, 0.0811, 0.1030, 0.0952, 0.0974], device='cuda:7'), in_proj_covar=tensor([0.0577, 0.0718, 0.0588, 0.0502, 0.0451, 0.0458, 0.0595, 0.0553], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:51:32,414 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=114350.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:51:35,147 INFO [train.py:904] (7/8) Epoch 12, batch 2700, loss[loss=0.1718, simple_loss=0.2559, pruned_loss=0.0439, over 16770.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.272, pruned_loss=0.05062, over 3328402.60 frames. ], batch size: 83, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:51:45,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.425e+02 2.890e+02 3.523e+02 1.000e+03, threshold=5.781e+02, percent-clipped=5.0 2023-04-29 13:52:44,870 INFO [train.py:904] (7/8) Epoch 12, batch 2750, loss[loss=0.1597, simple_loss=0.252, pruned_loss=0.03373, over 16877.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2715, pruned_loss=0.05045, over 3328235.38 frames. ], batch size: 42, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:53:13,328 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5389, 3.5820, 2.1282, 3.8696, 2.6819, 3.8404, 2.0819, 2.8073], device='cuda:7'), covar=tensor([0.0218, 0.0337, 0.1383, 0.0259, 0.0804, 0.0653, 0.1484, 0.0708], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0169, 0.0190, 0.0141, 0.0171, 0.0215, 0.0199, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 13:53:16,398 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1002, 5.9050, 6.0297, 5.6333, 5.7575, 6.2920, 5.8484, 5.6463], device='cuda:7'), covar=tensor([0.1032, 0.1693, 0.2112, 0.1916, 0.2674, 0.0980, 0.1226, 0.2091], device='cuda:7'), in_proj_covar=tensor([0.0361, 0.0516, 0.0560, 0.0437, 0.0596, 0.0586, 0.0444, 0.0596], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 13:53:51,039 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9267, 4.6786, 4.9013, 5.1817, 5.2979, 4.6104, 5.3056, 5.2877], device='cuda:7'), covar=tensor([0.1515, 0.1093, 0.1785, 0.0550, 0.0529, 0.0899, 0.0436, 0.0511], device='cuda:7'), in_proj_covar=tensor([0.0570, 0.0712, 0.0864, 0.0726, 0.0547, 0.0561, 0.0566, 0.0652], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:53:54,633 INFO [train.py:904] (7/8) Epoch 12, batch 2800, loss[loss=0.1975, simple_loss=0.2673, pruned_loss=0.06392, over 16722.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2713, pruned_loss=0.05073, over 3326734.62 frames. ], batch size: 124, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:54:06,075 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.386e+02 2.760e+02 3.486e+02 6.287e+02, threshold=5.520e+02, percent-clipped=1.0 2023-04-29 13:54:38,714 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114483.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 13:54:53,284 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2216, 5.2080, 4.9325, 4.4599, 4.9990, 1.9317, 4.7504, 4.9550], device='cuda:7'), covar=tensor([0.0059, 0.0048, 0.0151, 0.0291, 0.0072, 0.2250, 0.0107, 0.0131], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0125, 0.0177, 0.0164, 0.0147, 0.0188, 0.0164, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 13:55:04,047 INFO [train.py:904] (7/8) Epoch 12, batch 2850, loss[loss=0.1634, simple_loss=0.2609, pruned_loss=0.03298, over 17015.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2704, pruned_loss=0.0505, over 3322905.19 frames. ], batch size: 50, lr: 5.75e-03, grad_scale: 16.0 2023-04-29 13:56:13,208 INFO [train.py:904] (7/8) Epoch 12, batch 2900, loss[loss=0.1955, simple_loss=0.2563, pruned_loss=0.06736, over 16878.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2683, pruned_loss=0.05055, over 3324702.59 frames. ], batch size: 116, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 13:56:21,392 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-29 13:56:24,540 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.459e+02 2.961e+02 3.423e+02 5.764e+02, threshold=5.923e+02, percent-clipped=1.0 2023-04-29 13:56:31,181 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 13:56:50,646 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0041, 2.5732, 2.6585, 1.8765, 2.7976, 2.7637, 2.3725, 2.3178], device='cuda:7'), covar=tensor([0.0588, 0.0189, 0.0175, 0.0784, 0.0088, 0.0206, 0.0390, 0.0377], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0101, 0.0088, 0.0137, 0.0070, 0.0109, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 13:57:20,462 INFO [train.py:904] (7/8) Epoch 12, batch 2950, loss[loss=0.157, simple_loss=0.2467, pruned_loss=0.03368, over 17202.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2688, pruned_loss=0.05173, over 3322283.68 frames. ], batch size: 44, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 13:58:28,574 INFO [train.py:904] (7/8) Epoch 12, batch 3000, loss[loss=0.1844, simple_loss=0.2679, pruned_loss=0.05045, over 16788.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2684, pruned_loss=0.05151, over 3320473.74 frames. ], batch size: 83, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 13:58:28,575 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 13:58:38,470 INFO [train.py:938] (7/8) Epoch 12, validation: loss=0.14, simple_loss=0.2459, pruned_loss=0.01708, over 944034.00 frames. 2023-04-29 13:58:38,471 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 13:58:50,229 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.323e+02 2.752e+02 3.480e+02 1.028e+03, threshold=5.504e+02, percent-clipped=2.0 2023-04-29 13:59:48,676 INFO [train.py:904] (7/8) Epoch 12, batch 3050, loss[loss=0.1468, simple_loss=0.2427, pruned_loss=0.02543, over 17198.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2673, pruned_loss=0.05089, over 3320674.93 frames. ], batch size: 46, lr: 5.75e-03, grad_scale: 8.0 2023-04-29 14:00:09,003 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114716.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:00:34,853 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4891, 4.4374, 4.4136, 3.9390, 4.3926, 1.8275, 4.1945, 4.1524], device='cuda:7'), covar=tensor([0.0093, 0.0070, 0.0129, 0.0269, 0.0081, 0.2213, 0.0112, 0.0164], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0128, 0.0178, 0.0168, 0.0148, 0.0190, 0.0166, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:00:50,636 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-29 14:00:56,640 INFO [train.py:904] (7/8) Epoch 12, batch 3100, loss[loss=0.1988, simple_loss=0.2687, pruned_loss=0.06448, over 16701.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2672, pruned_loss=0.05172, over 3315955.02 frames. ], batch size: 124, lr: 5.74e-03, grad_scale: 8.0 2023-04-29 14:01:08,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8899, 4.1733, 3.9518, 4.0416, 3.7315, 3.7365, 3.7759, 4.1379], device='cuda:7'), covar=tensor([0.0997, 0.0912, 0.0933, 0.0737, 0.0722, 0.1665, 0.0921, 0.1059], device='cuda:7'), in_proj_covar=tensor([0.0586, 0.0727, 0.0599, 0.0511, 0.0456, 0.0461, 0.0606, 0.0563], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:01:10,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.272e+02 2.804e+02 3.413e+02 6.523e+02, threshold=5.608e+02, percent-clipped=1.0 2023-04-29 14:01:32,705 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114777.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:01:40,357 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114783.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:02:03,476 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5144, 3.4924, 2.8209, 2.1076, 2.3467, 2.1610, 3.5010, 3.1848], device='cuda:7'), covar=tensor([0.2429, 0.0671, 0.1421, 0.2287, 0.2492, 0.1835, 0.0485, 0.1257], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0258, 0.0282, 0.0279, 0.0282, 0.0222, 0.0269, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:02:05,285 INFO [train.py:904] (7/8) Epoch 12, batch 3150, loss[loss=0.1696, simple_loss=0.2656, pruned_loss=0.03684, over 17086.00 frames. ], tot_loss[loss=0.184, simple_loss=0.266, pruned_loss=0.05097, over 3322614.95 frames. ], batch size: 47, lr: 5.74e-03, grad_scale: 8.0 2023-04-29 14:02:16,672 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6297, 2.5779, 2.2605, 2.3876, 2.9089, 2.6300, 3.3565, 3.0969], device='cuda:7'), covar=tensor([0.0111, 0.0336, 0.0385, 0.0389, 0.0209, 0.0332, 0.0207, 0.0233], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0211, 0.0203, 0.0205, 0.0213, 0.0213, 0.0221, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:02:25,656 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4079, 4.4914, 4.8464, 4.8476, 4.8433, 4.5169, 4.5454, 4.3346], device='cuda:7'), covar=tensor([0.0314, 0.0459, 0.0314, 0.0349, 0.0455, 0.0340, 0.0762, 0.0507], device='cuda:7'), in_proj_covar=tensor([0.0355, 0.0370, 0.0372, 0.0347, 0.0415, 0.0389, 0.0501, 0.0315], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 14:02:45,205 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=114831.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:03:14,012 INFO [train.py:904] (7/8) Epoch 12, batch 3200, loss[loss=0.2142, simple_loss=0.2821, pruned_loss=0.07313, over 12349.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2655, pruned_loss=0.05059, over 3324046.97 frames. ], batch size: 246, lr: 5.74e-03, grad_scale: 8.0 2023-04-29 14:03:26,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.649e+02 2.483e+02 2.978e+02 3.563e+02 6.234e+02, threshold=5.956e+02, percent-clipped=4.0 2023-04-29 14:03:40,799 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114871.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:04:22,407 INFO [train.py:904] (7/8) Epoch 12, batch 3250, loss[loss=0.2095, simple_loss=0.2785, pruned_loss=0.07025, over 16878.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2663, pruned_loss=0.05057, over 3323421.03 frames. ], batch size: 109, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:05:05,239 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114932.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:05:32,345 INFO [train.py:904] (7/8) Epoch 12, batch 3300, loss[loss=0.2244, simple_loss=0.3001, pruned_loss=0.0744, over 12018.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2678, pruned_loss=0.05065, over 3318059.99 frames. ], batch size: 246, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:05:45,380 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.351e+02 3.077e+02 3.881e+02 7.792e+02, threshold=6.153e+02, percent-clipped=7.0 2023-04-29 14:06:42,027 INFO [train.py:904] (7/8) Epoch 12, batch 3350, loss[loss=0.1858, simple_loss=0.2762, pruned_loss=0.04768, over 16833.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2684, pruned_loss=0.05084, over 3309072.53 frames. ], batch size: 102, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:07:50,800 INFO [train.py:904] (7/8) Epoch 12, batch 3400, loss[loss=0.185, simple_loss=0.2662, pruned_loss=0.05193, over 16741.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2686, pruned_loss=0.05078, over 3303529.64 frames. ], batch size: 83, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:08:04,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.344e+02 2.703e+02 3.237e+02 1.034e+03, threshold=5.406e+02, percent-clipped=1.0 2023-04-29 14:08:18,300 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115072.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:08:21,216 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4806, 4.4043, 4.3866, 4.1906, 4.1059, 4.4507, 4.2012, 4.2007], device='cuda:7'), covar=tensor([0.0542, 0.0601, 0.0285, 0.0254, 0.0824, 0.0383, 0.0561, 0.0599], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0356, 0.0324, 0.0300, 0.0343, 0.0346, 0.0218, 0.0377], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:08:37,789 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115086.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:09:00,041 INFO [train.py:904] (7/8) Epoch 12, batch 3450, loss[loss=0.1753, simple_loss=0.2693, pruned_loss=0.04067, over 17286.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2669, pruned_loss=0.0502, over 3306610.96 frames. ], batch size: 52, lr: 5.74e-03, grad_scale: 4.0 2023-04-29 14:10:02,225 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115147.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:10:08,329 INFO [train.py:904] (7/8) Epoch 12, batch 3500, loss[loss=0.164, simple_loss=0.2574, pruned_loss=0.03532, over 17229.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2651, pruned_loss=0.04928, over 3306889.69 frames. ], batch size: 52, lr: 5.73e-03, grad_scale: 4.0 2023-04-29 14:10:23,311 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.367e+02 2.772e+02 3.300e+02 1.193e+03, threshold=5.543e+02, percent-clipped=2.0 2023-04-29 14:11:19,471 INFO [train.py:904] (7/8) Epoch 12, batch 3550, loss[loss=0.1675, simple_loss=0.2465, pruned_loss=0.0443, over 16731.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2632, pruned_loss=0.04827, over 3313854.33 frames. ], batch size: 89, lr: 5.73e-03, grad_scale: 4.0 2023-04-29 14:11:23,931 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9702, 4.4082, 4.3504, 3.1096, 3.6761, 4.3315, 3.9666, 2.4784], device='cuda:7'), covar=tensor([0.0352, 0.0036, 0.0029, 0.0277, 0.0079, 0.0072, 0.0060, 0.0390], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0072, 0.0072, 0.0126, 0.0081, 0.0091, 0.0081, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:11:53,652 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115227.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:12:09,159 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7026, 3.7842, 2.8913, 2.2336, 2.5348, 2.3090, 3.7181, 3.4026], device='cuda:7'), covar=tensor([0.2313, 0.0509, 0.1374, 0.2319, 0.2316, 0.1678, 0.0503, 0.1144], device='cuda:7'), in_proj_covar=tensor([0.0299, 0.0257, 0.0283, 0.0279, 0.0282, 0.0222, 0.0267, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:12:28,666 INFO [train.py:904] (7/8) Epoch 12, batch 3600, loss[loss=0.1588, simple_loss=0.232, pruned_loss=0.04284, over 16824.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2619, pruned_loss=0.04748, over 3313190.03 frames. ], batch size: 96, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:12:43,851 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.756e+02 2.259e+02 2.631e+02 3.384e+02 1.021e+03, threshold=5.262e+02, percent-clipped=2.0 2023-04-29 14:13:40,320 INFO [train.py:904] (7/8) Epoch 12, batch 3650, loss[loss=0.1872, simple_loss=0.287, pruned_loss=0.04369, over 16764.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2613, pruned_loss=0.04795, over 3298473.49 frames. ], batch size: 57, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:14:04,760 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8107, 2.2454, 2.3674, 4.7347, 2.3489, 2.7426, 2.3313, 2.4343], device='cuda:7'), covar=tensor([0.0927, 0.3305, 0.2460, 0.0347, 0.3745, 0.2361, 0.2972, 0.3661], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0400, 0.0337, 0.0326, 0.0415, 0.0462, 0.0364, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:14:27,961 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1345, 2.6082, 2.0687, 2.4714, 3.0799, 2.8425, 3.2941, 3.1853], device='cuda:7'), covar=tensor([0.0131, 0.0264, 0.0370, 0.0296, 0.0148, 0.0216, 0.0157, 0.0181], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0214, 0.0205, 0.0206, 0.0214, 0.0213, 0.0223, 0.0205], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:14:55,148 INFO [train.py:904] (7/8) Epoch 12, batch 3700, loss[loss=0.1903, simple_loss=0.2639, pruned_loss=0.05834, over 16809.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2598, pruned_loss=0.04917, over 3251197.07 frames. ], batch size: 102, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:15:09,332 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.309e+02 2.720e+02 3.193e+02 6.265e+02, threshold=5.440e+02, percent-clipped=2.0 2023-04-29 14:15:23,473 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115370.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:15:24,754 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115371.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:15:25,933 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115372.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:15:33,409 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-29 14:16:10,085 INFO [train.py:904] (7/8) Epoch 12, batch 3750, loss[loss=0.1823, simple_loss=0.2538, pruned_loss=0.05536, over 16834.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.261, pruned_loss=0.05111, over 3246188.74 frames. ], batch size: 83, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:16:36,744 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=115420.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:16:54,168 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115431.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:16:55,878 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:17:09,522 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115442.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:17:23,449 INFO [train.py:904] (7/8) Epoch 12, batch 3800, loss[loss=0.1968, simple_loss=0.2792, pruned_loss=0.05715, over 15628.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2623, pruned_loss=0.05265, over 3252614.91 frames. ], batch size: 191, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:17:38,969 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 2.466e+02 2.817e+02 3.424e+02 6.065e+02, threshold=5.633e+02, percent-clipped=2.0 2023-04-29 14:18:37,595 INFO [train.py:904] (7/8) Epoch 12, batch 3850, loss[loss=0.2261, simple_loss=0.298, pruned_loss=0.07712, over 12307.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2629, pruned_loss=0.05328, over 3260714.43 frames. ], batch size: 246, lr: 5.73e-03, grad_scale: 8.0 2023-04-29 14:19:16,909 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115527.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:19:22,074 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1450, 4.2065, 4.5545, 4.5326, 4.5519, 4.2655, 4.3103, 4.1262], device='cuda:7'), covar=tensor([0.0334, 0.0563, 0.0316, 0.0390, 0.0439, 0.0341, 0.0693, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0354, 0.0370, 0.0369, 0.0344, 0.0414, 0.0387, 0.0494, 0.0312], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 14:19:25,606 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2141, 3.2964, 1.9266, 3.4903, 2.5613, 3.4811, 2.0259, 2.5913], device='cuda:7'), covar=tensor([0.0235, 0.0459, 0.1567, 0.0201, 0.0724, 0.0620, 0.1395, 0.0695], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0168, 0.0190, 0.0139, 0.0169, 0.0213, 0.0197, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 14:19:48,211 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 14:19:52,699 INFO [train.py:904] (7/8) Epoch 12, batch 3900, loss[loss=0.1871, simple_loss=0.2629, pruned_loss=0.05565, over 16407.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.262, pruned_loss=0.05366, over 3261969.18 frames. ], batch size: 75, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:20:01,608 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6104, 4.5429, 4.4905, 4.2522, 4.1846, 4.5514, 4.3373, 4.2790], device='cuda:7'), covar=tensor([0.0537, 0.0536, 0.0303, 0.0288, 0.0884, 0.0421, 0.0467, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0261, 0.0341, 0.0311, 0.0288, 0.0330, 0.0333, 0.0210, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:20:07,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.153e+02 2.652e+02 3.211e+02 6.333e+02, threshold=5.304e+02, percent-clipped=2.0 2023-04-29 14:20:29,517 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=115575.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:20:45,577 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5909, 3.6655, 2.7573, 2.1809, 2.3712, 2.1970, 3.6626, 3.1982], device='cuda:7'), covar=tensor([0.2314, 0.0580, 0.1434, 0.2371, 0.2478, 0.1843, 0.0497, 0.1169], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0257, 0.0285, 0.0280, 0.0286, 0.0223, 0.0268, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:21:08,914 INFO [train.py:904] (7/8) Epoch 12, batch 3950, loss[loss=0.1899, simple_loss=0.2563, pruned_loss=0.06177, over 16925.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2622, pruned_loss=0.05451, over 3270945.14 frames. ], batch size: 96, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:21:26,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2368, 5.5631, 5.3173, 5.2990, 5.0228, 4.8719, 5.0333, 5.6401], device='cuda:7'), covar=tensor([0.0989, 0.0761, 0.0911, 0.0730, 0.0733, 0.0891, 0.0890, 0.0779], device='cuda:7'), in_proj_covar=tensor([0.0582, 0.0726, 0.0593, 0.0514, 0.0456, 0.0464, 0.0601, 0.0562], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:22:21,470 INFO [train.py:904] (7/8) Epoch 12, batch 4000, loss[loss=0.2061, simple_loss=0.2872, pruned_loss=0.06248, over 15619.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2629, pruned_loss=0.05518, over 3270438.18 frames. ], batch size: 191, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:22:28,541 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-29 14:22:34,737 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.376e+02 2.718e+02 3.245e+02 5.190e+02, threshold=5.435e+02, percent-clipped=0.0 2023-04-29 14:23:03,443 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-04-29 14:23:10,866 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.16 vs. limit=5.0 2023-04-29 14:23:15,673 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0175, 5.1848, 5.5295, 5.5097, 5.5211, 5.1882, 5.0946, 4.8160], device='cuda:7'), covar=tensor([0.0248, 0.0350, 0.0266, 0.0350, 0.0361, 0.0274, 0.0718, 0.0367], device='cuda:7'), in_proj_covar=tensor([0.0357, 0.0372, 0.0371, 0.0349, 0.0417, 0.0391, 0.0498, 0.0314], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 14:23:20,654 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8532, 1.4362, 1.6775, 1.6878, 1.8679, 1.9326, 1.5522, 1.7332], device='cuda:7'), covar=tensor([0.0187, 0.0289, 0.0156, 0.0216, 0.0201, 0.0128, 0.0314, 0.0071], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0175, 0.0158, 0.0163, 0.0174, 0.0130, 0.0174, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 14:23:35,789 INFO [train.py:904] (7/8) Epoch 12, batch 4050, loss[loss=0.1839, simple_loss=0.2597, pruned_loss=0.05406, over 16678.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2626, pruned_loss=0.05381, over 3271431.84 frames. ], batch size: 62, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:24:12,149 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115726.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:24:13,877 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115727.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:24:36,757 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115742.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:24:50,503 INFO [train.py:904] (7/8) Epoch 12, batch 4100, loss[loss=0.1878, simple_loss=0.275, pruned_loss=0.05034, over 16676.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2643, pruned_loss=0.05337, over 3267634.45 frames. ], batch size: 62, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:25:05,531 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 1.888e+02 2.192e+02 2.609e+02 4.553e+02, threshold=4.384e+02, percent-clipped=0.0 2023-04-29 14:25:06,878 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4689, 1.7992, 2.0734, 2.4021, 2.5488, 2.6995, 1.7925, 2.5455], device='cuda:7'), covar=tensor([0.0171, 0.0343, 0.0254, 0.0206, 0.0209, 0.0133, 0.0361, 0.0090], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0177, 0.0160, 0.0164, 0.0176, 0.0131, 0.0176, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:25:14,319 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 14:25:48,338 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=115790.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:26:05,822 INFO [train.py:904] (7/8) Epoch 12, batch 4150, loss[loss=0.2601, simple_loss=0.3262, pruned_loss=0.09703, over 11189.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2721, pruned_loss=0.05667, over 3210224.45 frames. ], batch size: 247, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:26:30,878 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9541, 2.3932, 2.3244, 2.9092, 2.2040, 3.2181, 1.7183, 2.7323], device='cuda:7'), covar=tensor([0.1074, 0.0523, 0.0974, 0.0142, 0.0155, 0.0379, 0.1280, 0.0618], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0162, 0.0183, 0.0155, 0.0203, 0.0215, 0.0184, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 14:27:23,046 INFO [train.py:904] (7/8) Epoch 12, batch 4200, loss[loss=0.2299, simple_loss=0.3162, pruned_loss=0.07175, over 16866.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2795, pruned_loss=0.05852, over 3188010.62 frames. ], batch size: 116, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:27:37,145 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.546e+02 2.889e+02 3.538e+02 7.743e+02, threshold=5.778e+02, percent-clipped=11.0 2023-04-29 14:28:01,310 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 14:28:04,374 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3472, 4.6362, 4.4360, 4.4374, 4.1461, 4.1390, 4.0938, 4.6864], device='cuda:7'), covar=tensor([0.0975, 0.0796, 0.0843, 0.0652, 0.0705, 0.1284, 0.0982, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0558, 0.0695, 0.0572, 0.0490, 0.0437, 0.0448, 0.0576, 0.0538], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:28:22,410 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115892.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 14:28:31,800 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2053, 1.8356, 2.5511, 3.0901, 2.8648, 3.5474, 2.2115, 3.3952], device='cuda:7'), covar=tensor([0.0161, 0.0426, 0.0273, 0.0211, 0.0242, 0.0109, 0.0373, 0.0101], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0176, 0.0158, 0.0162, 0.0175, 0.0130, 0.0174, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 14:28:36,620 INFO [train.py:904] (7/8) Epoch 12, batch 4250, loss[loss=0.1972, simple_loss=0.287, pruned_loss=0.05368, over 16783.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2832, pruned_loss=0.05895, over 3161527.67 frames. ], batch size: 83, lr: 5.72e-03, grad_scale: 8.0 2023-04-29 14:29:00,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8203, 5.0489, 5.2858, 5.0022, 5.0735, 5.6545, 5.2061, 4.9915], device='cuda:7'), covar=tensor([0.0968, 0.1603, 0.1525, 0.1765, 0.2487, 0.0891, 0.1271, 0.2177], device='cuda:7'), in_proj_covar=tensor([0.0351, 0.0497, 0.0537, 0.0424, 0.0569, 0.0564, 0.0431, 0.0578], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:29:49,181 INFO [train.py:904] (7/8) Epoch 12, batch 4300, loss[loss=0.2157, simple_loss=0.3058, pruned_loss=0.06278, over 17024.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.284, pruned_loss=0.05761, over 3182224.19 frames. ], batch size: 50, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:29:50,994 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115953.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 14:30:04,742 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.520e+02 2.885e+02 3.320e+02 6.529e+02, threshold=5.769e+02, percent-clipped=1.0 2023-04-29 14:30:12,614 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 14:30:59,261 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0759, 2.9771, 3.1589, 1.6219, 3.3291, 3.3335, 2.6409, 2.5111], device='cuda:7'), covar=tensor([0.0782, 0.0217, 0.0183, 0.1118, 0.0058, 0.0114, 0.0416, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0100, 0.0087, 0.0138, 0.0070, 0.0106, 0.0121, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 14:31:07,415 INFO [train.py:904] (7/8) Epoch 12, batch 4350, loss[loss=0.2004, simple_loss=0.2941, pruned_loss=0.05341, over 16873.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2874, pruned_loss=0.05851, over 3188919.98 frames. ], batch size: 96, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:31:27,485 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5037, 4.7958, 4.5553, 4.5607, 4.2632, 4.2233, 4.2681, 4.8522], device='cuda:7'), covar=tensor([0.0907, 0.0758, 0.0928, 0.0745, 0.0731, 0.1176, 0.0911, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0557, 0.0693, 0.0570, 0.0488, 0.0436, 0.0447, 0.0574, 0.0536], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:31:27,549 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116014.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:31:46,174 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116026.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:31:48,046 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116027.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:32:22,081 INFO [train.py:904] (7/8) Epoch 12, batch 4400, loss[loss=0.2143, simple_loss=0.2987, pruned_loss=0.06495, over 16639.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2898, pruned_loss=0.05981, over 3159392.58 frames. ], batch size: 57, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:32:37,554 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.512e+02 3.045e+02 3.645e+02 6.621e+02, threshold=6.090e+02, percent-clipped=4.0 2023-04-29 14:32:55,647 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116074.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:32:57,359 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116075.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:32:57,502 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116075.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 14:33:35,432 INFO [train.py:904] (7/8) Epoch 12, batch 4450, loss[loss=0.202, simple_loss=0.2951, pruned_loss=0.05448, over 16720.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.2937, pruned_loss=0.06134, over 3178037.69 frames. ], batch size: 89, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:33:40,155 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116105.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:33:57,063 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116116.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:34:49,339 INFO [train.py:904] (7/8) Epoch 12, batch 4500, loss[loss=0.215, simple_loss=0.2976, pruned_loss=0.06619, over 15486.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2939, pruned_loss=0.06176, over 3182976.68 frames. ], batch size: 190, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:35:03,492 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.115e+02 2.423e+02 2.863e+02 5.089e+02, threshold=4.846e+02, percent-clipped=0.0 2023-04-29 14:35:10,866 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116166.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:35:26,184 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116177.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:36:02,117 INFO [train.py:904] (7/8) Epoch 12, batch 4550, loss[loss=0.2446, simple_loss=0.3214, pruned_loss=0.08391, over 15451.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2943, pruned_loss=0.06216, over 3203393.10 frames. ], batch size: 191, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:36:11,974 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-29 14:36:15,258 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1337, 3.0927, 3.3351, 1.6506, 3.5010, 3.4808, 2.7485, 2.6430], device='cuda:7'), covar=tensor([0.0840, 0.0211, 0.0162, 0.1179, 0.0062, 0.0117, 0.0421, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0100, 0.0088, 0.0137, 0.0069, 0.0106, 0.0121, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 14:36:15,312 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5602, 2.5049, 2.5528, 3.3244, 2.8526, 3.7252, 1.2561, 2.8881], device='cuda:7'), covar=tensor([0.1315, 0.0648, 0.0959, 0.0119, 0.0197, 0.0298, 0.1530, 0.0661], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0160, 0.0180, 0.0151, 0.0199, 0.0208, 0.0181, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 14:36:20,880 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-29 14:36:29,711 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5691, 4.6111, 4.8386, 4.6580, 4.7773, 5.2749, 4.6942, 4.4242], device='cuda:7'), covar=tensor([0.1107, 0.1774, 0.1748, 0.1935, 0.2360, 0.0836, 0.1377, 0.2483], device='cuda:7'), in_proj_covar=tensor([0.0348, 0.0485, 0.0525, 0.0416, 0.0558, 0.0553, 0.0421, 0.0568], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:37:08,871 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116248.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 14:37:14,082 INFO [train.py:904] (7/8) Epoch 12, batch 4600, loss[loss=0.1929, simple_loss=0.2814, pruned_loss=0.05223, over 17006.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.295, pruned_loss=0.06234, over 3209255.32 frames. ], batch size: 50, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:37:29,434 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 1.985e+02 2.258e+02 2.658e+02 3.690e+02, threshold=4.517e+02, percent-clipped=0.0 2023-04-29 14:37:49,560 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5416, 2.6686, 2.4849, 3.7307, 2.9547, 3.8029, 1.3792, 2.7811], device='cuda:7'), covar=tensor([0.1353, 0.0638, 0.1083, 0.0119, 0.0243, 0.0340, 0.1579, 0.0766], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0161, 0.0182, 0.0151, 0.0201, 0.0210, 0.0182, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 14:37:53,164 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 14:38:07,626 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0109, 3.9233, 4.1077, 4.2280, 4.3325, 3.9500, 4.2514, 4.3836], device='cuda:7'), covar=tensor([0.1189, 0.0951, 0.1132, 0.0536, 0.0437, 0.1277, 0.0691, 0.0481], device='cuda:7'), in_proj_covar=tensor([0.0525, 0.0659, 0.0786, 0.0671, 0.0499, 0.0520, 0.0522, 0.0603], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:38:26,068 INFO [train.py:904] (7/8) Epoch 12, batch 4650, loss[loss=0.1932, simple_loss=0.273, pruned_loss=0.05676, over 16814.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2941, pruned_loss=0.06238, over 3195592.38 frames. ], batch size: 42, lr: 5.71e-03, grad_scale: 8.0 2023-04-29 14:38:56,348 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116323.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:39:38,314 INFO [train.py:904] (7/8) Epoch 12, batch 4700, loss[loss=0.1928, simple_loss=0.2732, pruned_loss=0.05624, over 16642.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2912, pruned_loss=0.06149, over 3202826.43 frames. ], batch size: 62, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:39:53,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.029e+02 2.445e+02 3.086e+02 5.514e+02, threshold=4.890e+02, percent-clipped=4.0 2023-04-29 14:40:05,827 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116370.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 14:40:16,140 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1633, 3.7638, 3.7227, 2.3095, 3.3069, 3.7053, 3.5204, 1.9497], device='cuda:7'), covar=tensor([0.0454, 0.0029, 0.0031, 0.0372, 0.0069, 0.0079, 0.0062, 0.0409], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0070, 0.0071, 0.0126, 0.0081, 0.0090, 0.0079, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:40:27,557 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:40:54,113 INFO [train.py:904] (7/8) Epoch 12, batch 4750, loss[loss=0.1768, simple_loss=0.2608, pruned_loss=0.04641, over 16761.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.287, pruned_loss=0.05942, over 3215680.76 frames. ], batch size: 124, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:41:31,443 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6183, 4.3194, 4.2917, 2.9808, 3.7110, 4.2630, 3.8385, 2.2646], device='cuda:7'), covar=tensor([0.0385, 0.0022, 0.0025, 0.0275, 0.0063, 0.0070, 0.0065, 0.0367], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0070, 0.0071, 0.0126, 0.0081, 0.0090, 0.0080, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:41:42,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8668, 2.3027, 2.1598, 2.9024, 2.0316, 3.2137, 1.6771, 2.6687], device='cuda:7'), covar=tensor([0.1244, 0.0619, 0.1195, 0.0162, 0.0168, 0.0388, 0.1451, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0161, 0.0182, 0.0151, 0.0199, 0.0209, 0.0183, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 14:41:43,763 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5891, 2.6300, 1.6740, 2.7473, 2.1824, 2.8011, 1.8772, 2.3750], device='cuda:7'), covar=tensor([0.0221, 0.0298, 0.1305, 0.0138, 0.0574, 0.0407, 0.1210, 0.0514], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0163, 0.0187, 0.0131, 0.0164, 0.0206, 0.0194, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 14:41:59,902 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116446.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:42:07,125 INFO [train.py:904] (7/8) Epoch 12, batch 4800, loss[loss=0.1973, simple_loss=0.2879, pruned_loss=0.05338, over 16243.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2831, pruned_loss=0.05706, over 3220621.89 frames. ], batch size: 165, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:42:22,179 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116461.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:42:23,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.880e+02 2.223e+02 2.728e+02 4.441e+02, threshold=4.446e+02, percent-clipped=0.0 2023-04-29 14:42:39,678 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116472.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:42:49,943 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.67 vs. limit=5.0 2023-04-29 14:43:23,417 INFO [train.py:904] (7/8) Epoch 12, batch 4850, loss[loss=0.177, simple_loss=0.2573, pruned_loss=0.04835, over 16677.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2839, pruned_loss=0.05626, over 3202705.19 frames. ], batch size: 57, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:43:24,304 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 14:43:28,256 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116505.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:43:31,424 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116507.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 14:44:31,807 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116548.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 14:44:38,150 INFO [train.py:904] (7/8) Epoch 12, batch 4900, loss[loss=0.1742, simple_loss=0.262, pruned_loss=0.04316, over 17209.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2829, pruned_loss=0.05516, over 3185903.56 frames. ], batch size: 45, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:44:45,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6328, 2.4707, 2.1069, 3.2485, 2.0479, 3.5558, 1.3099, 2.8072], device='cuda:7'), covar=tensor([0.1369, 0.0677, 0.1341, 0.0132, 0.0137, 0.0349, 0.1727, 0.0719], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0160, 0.0180, 0.0149, 0.0197, 0.0207, 0.0182, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 14:44:52,630 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.192e+02 2.607e+02 3.082e+02 6.652e+02, threshold=5.215e+02, percent-clipped=3.0 2023-04-29 14:44:59,326 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116566.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:45:29,236 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0505, 1.4551, 1.8053, 2.0148, 2.2129, 2.3089, 1.6373, 2.2365], device='cuda:7'), covar=tensor([0.0193, 0.0365, 0.0211, 0.0274, 0.0222, 0.0132, 0.0362, 0.0103], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0172, 0.0155, 0.0160, 0.0169, 0.0126, 0.0174, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 14:45:42,830 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116596.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 14:45:52,022 INFO [train.py:904] (7/8) Epoch 12, batch 4950, loss[loss=0.2048, simple_loss=0.2834, pruned_loss=0.06309, over 16220.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.282, pruned_loss=0.05445, over 3195771.72 frames. ], batch size: 35, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:46:16,242 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116618.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:47:04,274 INFO [train.py:904] (7/8) Epoch 12, batch 5000, loss[loss=0.2092, simple_loss=0.3008, pruned_loss=0.05881, over 16783.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2838, pruned_loss=0.05471, over 3179132.38 frames. ], batch size: 134, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:47:17,036 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.254e+02 2.645e+02 3.532e+02 7.072e+02, threshold=5.290e+02, percent-clipped=1.0 2023-04-29 14:47:18,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5901, 2.7709, 2.3833, 4.9889, 3.4783, 4.1330, 1.5104, 2.9023], device='cuda:7'), covar=tensor([0.1423, 0.0799, 0.1340, 0.0095, 0.0429, 0.0391, 0.1604, 0.0967], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0160, 0.0180, 0.0149, 0.0197, 0.0206, 0.0181, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 14:47:20,437 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-29 14:47:30,370 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116670.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 14:47:43,242 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:47:43,414 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:48:15,699 INFO [train.py:904] (7/8) Epoch 12, batch 5050, loss[loss=0.1967, simple_loss=0.2831, pruned_loss=0.05513, over 16779.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2842, pruned_loss=0.05462, over 3181007.62 frames. ], batch size: 124, lr: 5.70e-03, grad_scale: 8.0 2023-04-29 14:48:18,396 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:48:38,187 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116718.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:49:24,617 INFO [train.py:904] (7/8) Epoch 12, batch 5100, loss[loss=0.1794, simple_loss=0.2574, pruned_loss=0.05063, over 16727.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2821, pruned_loss=0.0537, over 3194754.83 frames. ], batch size: 57, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:49:37,478 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116761.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:49:38,227 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.103e+02 2.626e+02 3.146e+02 5.078e+02, threshold=5.251e+02, percent-clipped=1.0 2023-04-29 14:49:42,519 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1716, 2.3390, 1.8929, 2.1893, 2.7673, 2.4055, 3.0313, 3.0478], device='cuda:7'), covar=tensor([0.0076, 0.0314, 0.0445, 0.0346, 0.0195, 0.0308, 0.0123, 0.0171], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0205, 0.0202, 0.0200, 0.0205, 0.0204, 0.0209, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:49:43,701 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116765.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:49:52,573 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116772.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:50:35,779 INFO [train.py:904] (7/8) Epoch 12, batch 5150, loss[loss=0.1943, simple_loss=0.2895, pruned_loss=0.04959, over 16411.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2826, pruned_loss=0.05275, over 3191782.13 frames. ], batch size: 146, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:50:36,712 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116802.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 14:50:36,794 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116802.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:50:46,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5849, 3.6697, 2.6839, 2.1097, 2.4699, 2.2737, 3.7210, 3.3812], device='cuda:7'), covar=tensor([0.2499, 0.0673, 0.1639, 0.2317, 0.2167, 0.1706, 0.0470, 0.0957], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0260, 0.0287, 0.0283, 0.0284, 0.0224, 0.0270, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:50:47,873 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116809.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:50:57,628 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 14:51:00,434 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-04-29 14:51:03,103 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=116820.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:51:47,941 INFO [train.py:904] (7/8) Epoch 12, batch 5200, loss[loss=0.1731, simple_loss=0.2655, pruned_loss=0.04031, over 16842.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.281, pruned_loss=0.0521, over 3173714.13 frames. ], batch size: 102, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:52:00,771 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116861.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:52:01,645 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.168e+02 2.438e+02 3.019e+02 6.927e+02, threshold=4.876e+02, percent-clipped=1.0 2023-04-29 14:52:03,376 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116863.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:52:07,455 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1228, 3.8767, 3.7854, 2.3963, 3.4512, 3.8083, 3.3817, 1.7095], device='cuda:7'), covar=tensor([0.0523, 0.0062, 0.0063, 0.0397, 0.0111, 0.0224, 0.0317, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0069, 0.0070, 0.0126, 0.0080, 0.0090, 0.0080, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:52:55,714 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1307, 5.4881, 5.1836, 5.2361, 4.8964, 4.8037, 4.8531, 5.5391], device='cuda:7'), covar=tensor([0.0992, 0.0716, 0.0925, 0.0632, 0.0759, 0.0651, 0.0938, 0.0829], device='cuda:7'), in_proj_covar=tensor([0.0547, 0.0676, 0.0561, 0.0479, 0.0433, 0.0441, 0.0562, 0.0529], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:52:58,582 INFO [train.py:904] (7/8) Epoch 12, batch 5250, loss[loss=0.1781, simple_loss=0.2737, pruned_loss=0.04125, over 16515.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2779, pruned_loss=0.0512, over 3187117.38 frames. ], batch size: 75, lr: 5.69e-03, grad_scale: 16.0 2023-04-29 14:54:11,426 INFO [train.py:904] (7/8) Epoch 12, batch 5300, loss[loss=0.1695, simple_loss=0.258, pruned_loss=0.04055, over 16358.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2753, pruned_loss=0.05061, over 3179656.67 frames. ], batch size: 165, lr: 5.69e-03, grad_scale: 8.0 2023-04-29 14:54:27,263 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.433e+02 2.208e+02 2.588e+02 3.045e+02 5.450e+02, threshold=5.175e+02, percent-clipped=1.0 2023-04-29 14:54:42,588 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116974.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:54:49,770 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116979.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:55:21,986 INFO [train.py:904] (7/8) Epoch 12, batch 5350, loss[loss=0.1938, simple_loss=0.2851, pruned_loss=0.0512, over 16379.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2729, pruned_loss=0.0496, over 3184388.33 frames. ], batch size: 146, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:55:58,442 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117027.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:56:31,821 INFO [train.py:904] (7/8) Epoch 12, batch 5400, loss[loss=0.2171, simple_loss=0.2961, pruned_loss=0.06907, over 11903.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2756, pruned_loss=0.05022, over 3189007.13 frames. ], batch size: 248, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:56:35,738 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7947, 3.7558, 3.9146, 3.7096, 3.7778, 4.2681, 3.9655, 3.6394], device='cuda:7'), covar=tensor([0.2151, 0.1904, 0.1681, 0.2103, 0.2484, 0.1454, 0.1321, 0.2409], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0480, 0.0521, 0.0417, 0.0564, 0.0553, 0.0419, 0.0573], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 14:56:43,956 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117060.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:56:48,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3977, 2.5329, 2.0817, 2.3341, 2.8684, 2.5221, 3.1456, 3.0985], device='cuda:7'), covar=tensor([0.0068, 0.0292, 0.0393, 0.0321, 0.0190, 0.0318, 0.0152, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0206, 0.0202, 0.0200, 0.0206, 0.0205, 0.0209, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 14:56:49,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 2.253e+02 2.451e+02 2.837e+02 5.241e+02, threshold=4.902e+02, percent-clipped=1.0 2023-04-29 14:57:01,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0860, 1.5012, 1.7744, 2.0327, 2.2423, 2.2500, 1.6351, 2.1902], device='cuda:7'), covar=tensor([0.0180, 0.0380, 0.0237, 0.0270, 0.0209, 0.0149, 0.0367, 0.0092], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0174, 0.0156, 0.0162, 0.0171, 0.0126, 0.0174, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 14:57:46,206 INFO [train.py:904] (7/8) Epoch 12, batch 5450, loss[loss=0.217, simple_loss=0.3038, pruned_loss=0.06512, over 16539.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2791, pruned_loss=0.05203, over 3182242.37 frames. ], batch size: 75, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:57:46,720 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117102.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:58:57,655 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117150.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:59:00,361 INFO [train.py:904] (7/8) Epoch 12, batch 5500, loss[loss=0.2284, simple_loss=0.3109, pruned_loss=0.07294, over 16639.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2873, pruned_loss=0.05682, over 3178812.81 frames. ], batch size: 68, lr: 5.69e-03, grad_scale: 4.0 2023-04-29 14:59:09,316 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117158.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:59:13,867 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117161.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 14:59:18,160 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.773e+02 3.440e+02 4.409e+02 8.971e+02, threshold=6.880e+02, percent-clipped=17.0 2023-04-29 15:00:18,016 INFO [train.py:904] (7/8) Epoch 12, batch 5550, loss[loss=0.2386, simple_loss=0.316, pruned_loss=0.08063, over 17013.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2953, pruned_loss=0.06298, over 3147626.78 frames. ], batch size: 55, lr: 5.68e-03, grad_scale: 4.0 2023-04-29 15:00:30,360 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117209.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:00:49,869 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117221.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:01:39,153 INFO [train.py:904] (7/8) Epoch 12, batch 5600, loss[loss=0.2596, simple_loss=0.3305, pruned_loss=0.0944, over 16684.00 frames. ], tot_loss[loss=0.2176, simple_loss=0.3005, pruned_loss=0.06737, over 3125148.03 frames. ], batch size: 134, lr: 5.68e-03, grad_scale: 8.0 2023-04-29 15:01:58,907 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.404e+02 3.697e+02 4.299e+02 5.006e+02 8.998e+02, threshold=8.599e+02, percent-clipped=4.0 2023-04-29 15:02:16,740 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117274.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:02:30,523 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117282.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:02:49,996 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6528, 3.0714, 2.7215, 5.1745, 3.8834, 4.2554, 1.9178, 3.0799], device='cuda:7'), covar=tensor([0.1343, 0.0681, 0.1153, 0.0117, 0.0428, 0.0391, 0.1446, 0.0835], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0162, 0.0182, 0.0151, 0.0198, 0.0208, 0.0184, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 15:03:02,074 INFO [train.py:904] (7/8) Epoch 12, batch 5650, loss[loss=0.2985, simple_loss=0.3456, pruned_loss=0.1258, over 11261.00 frames. ], tot_loss[loss=0.2264, simple_loss=0.3069, pruned_loss=0.07298, over 3059984.68 frames. ], batch size: 246, lr: 5.68e-03, grad_scale: 4.0 2023-04-29 15:03:08,371 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 15:03:33,953 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117322.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:03:45,030 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:04:18,931 INFO [train.py:904] (7/8) Epoch 12, batch 5700, loss[loss=0.2042, simple_loss=0.2961, pruned_loss=0.05619, over 16490.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.308, pruned_loss=0.07436, over 3051314.05 frames. ], batch size: 68, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:04:32,770 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117360.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:04:41,588 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.520e+02 3.684e+02 4.316e+02 5.371e+02 1.144e+03, threshold=8.631e+02, percent-clipped=1.0 2023-04-29 15:05:08,838 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0047, 2.0089, 2.2469, 3.5414, 1.9637, 2.3957, 2.1741, 2.1512], device='cuda:7'), covar=tensor([0.1054, 0.3014, 0.2116, 0.0455, 0.3814, 0.1995, 0.2689, 0.3263], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0395, 0.0331, 0.0316, 0.0412, 0.0454, 0.0359, 0.0460], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:05:21,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117390.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:05:39,188 INFO [train.py:904] (7/8) Epoch 12, batch 5750, loss[loss=0.2405, simple_loss=0.328, pruned_loss=0.07649, over 16910.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.3109, pruned_loss=0.07625, over 3027570.51 frames. ], batch size: 109, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:05:48,658 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117408.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:06:06,391 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 15:06:59,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0184, 3.0043, 2.5523, 2.9115, 3.3889, 3.0754, 3.8077, 3.6334], device='cuda:7'), covar=tensor([0.0059, 0.0252, 0.0355, 0.0266, 0.0163, 0.0235, 0.0157, 0.0141], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0203, 0.0198, 0.0198, 0.0203, 0.0201, 0.0208, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:07:00,144 INFO [train.py:904] (7/8) Epoch 12, batch 5800, loss[loss=0.2046, simple_loss=0.2789, pruned_loss=0.06513, over 16378.00 frames. ], tot_loss[loss=0.2303, simple_loss=0.3106, pruned_loss=0.07496, over 3044591.56 frames. ], batch size: 35, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:07:00,830 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7312, 3.8559, 2.7398, 2.2898, 2.8055, 2.3078, 4.1580, 3.5424], device='cuda:7'), covar=tensor([0.2774, 0.0828, 0.1871, 0.2245, 0.2313, 0.1975, 0.0517, 0.1118], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0255, 0.0282, 0.0278, 0.0279, 0.0220, 0.0265, 0.0296], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:07:09,848 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117458.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:07:15,423 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 15:07:21,330 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.077e+02 3.389e+02 3.943e+02 4.772e+02 8.236e+02, threshold=7.885e+02, percent-clipped=0.0 2023-04-29 15:08:16,570 INFO [train.py:904] (7/8) Epoch 12, batch 5850, loss[loss=0.2348, simple_loss=0.3121, pruned_loss=0.07871, over 16839.00 frames. ], tot_loss[loss=0.2272, simple_loss=0.3082, pruned_loss=0.07313, over 3051821.87 frames. ], batch size: 116, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:08:25,115 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117506.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:09:06,926 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7689, 3.3298, 3.3080, 1.9218, 2.7611, 2.2412, 3.3044, 3.4833], device='cuda:7'), covar=tensor([0.0257, 0.0631, 0.0539, 0.1810, 0.0807, 0.0930, 0.0643, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0147, 0.0161, 0.0144, 0.0138, 0.0125, 0.0139, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 15:09:37,163 INFO [train.py:904] (7/8) Epoch 12, batch 5900, loss[loss=0.1897, simple_loss=0.2815, pruned_loss=0.04896, over 16821.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3075, pruned_loss=0.0725, over 3067384.36 frames. ], batch size: 102, lr: 5.68e-03, grad_scale: 2.0 2023-04-29 15:10:01,567 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.797e+02 2.838e+02 3.606e+02 4.204e+02 7.799e+02, threshold=7.213e+02, percent-clipped=0.0 2023-04-29 15:10:14,538 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-29 15:10:18,959 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117577.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:10:55,837 INFO [train.py:904] (7/8) Epoch 12, batch 5950, loss[loss=0.2091, simple_loss=0.3001, pruned_loss=0.05905, over 17217.00 frames. ], tot_loss[loss=0.225, simple_loss=0.3082, pruned_loss=0.07096, over 3077693.54 frames. ], batch size: 52, lr: 5.67e-03, grad_scale: 2.0 2023-04-29 15:11:32,649 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-29 15:12:14,110 INFO [train.py:904] (7/8) Epoch 12, batch 6000, loss[loss=0.2247, simple_loss=0.2936, pruned_loss=0.07788, over 11420.00 frames. ], tot_loss[loss=0.2242, simple_loss=0.3072, pruned_loss=0.07057, over 3076370.92 frames. ], batch size: 247, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:12:14,110 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 15:12:25,317 INFO [train.py:938] (7/8) Epoch 12, validation: loss=0.161, simple_loss=0.2739, pruned_loss=0.02405, over 944034.00 frames. 2023-04-29 15:12:25,318 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 15:12:46,505 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.036e+02 2.944e+02 3.518e+02 4.227e+02 7.444e+02, threshold=7.035e+02, percent-clipped=1.0 2023-04-29 15:13:18,112 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117685.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:13:46,119 INFO [train.py:904] (7/8) Epoch 12, batch 6050, loss[loss=0.2035, simple_loss=0.3025, pruned_loss=0.05223, over 16650.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.3048, pruned_loss=0.0695, over 3086211.40 frames. ], batch size: 62, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:14:18,806 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6873, 3.8182, 2.9956, 2.2083, 2.6108, 2.2595, 4.1345, 3.5164], device='cuda:7'), covar=tensor([0.2555, 0.0658, 0.1538, 0.2294, 0.2304, 0.1867, 0.0419, 0.1021], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0258, 0.0286, 0.0282, 0.0282, 0.0222, 0.0268, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:15:02,183 INFO [train.py:904] (7/8) Epoch 12, batch 6100, loss[loss=0.2038, simple_loss=0.2904, pruned_loss=0.05863, over 15498.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.3044, pruned_loss=0.06868, over 3090179.75 frames. ], batch size: 190, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:15:24,828 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.907e+02 2.974e+02 3.801e+02 4.781e+02 1.516e+03, threshold=7.602e+02, percent-clipped=11.0 2023-04-29 15:16:19,687 INFO [train.py:904] (7/8) Epoch 12, batch 6150, loss[loss=0.2177, simple_loss=0.2941, pruned_loss=0.07065, over 15351.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.302, pruned_loss=0.06754, over 3112297.48 frames. ], batch size: 191, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:16:42,005 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8247, 3.2336, 3.1690, 1.9896, 3.0598, 3.2076, 3.0796, 1.9698], device='cuda:7'), covar=tensor([0.0479, 0.0035, 0.0042, 0.0370, 0.0071, 0.0083, 0.0065, 0.0337], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0068, 0.0070, 0.0124, 0.0080, 0.0090, 0.0079, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 15:17:38,954 INFO [train.py:904] (7/8) Epoch 12, batch 6200, loss[loss=0.2264, simple_loss=0.3115, pruned_loss=0.0706, over 16747.00 frames. ], tot_loss[loss=0.217, simple_loss=0.3001, pruned_loss=0.06692, over 3118332.65 frames. ], batch size: 76, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:18:00,667 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.030e+02 3.055e+02 3.626e+02 4.280e+02 7.203e+02, threshold=7.253e+02, percent-clipped=0.0 2023-04-29 15:18:17,990 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117877.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:18:52,875 INFO [train.py:904] (7/8) Epoch 12, batch 6250, loss[loss=0.1931, simple_loss=0.2853, pruned_loss=0.05049, over 16882.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2998, pruned_loss=0.06636, over 3139617.21 frames. ], batch size: 96, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:19:28,654 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=117925.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:20:06,464 INFO [train.py:904] (7/8) Epoch 12, batch 6300, loss[loss=0.199, simple_loss=0.2846, pruned_loss=0.05668, over 16517.00 frames. ], tot_loss[loss=0.2156, simple_loss=0.2999, pruned_loss=0.06561, over 3149685.78 frames. ], batch size: 68, lr: 5.67e-03, grad_scale: 4.0 2023-04-29 15:20:28,834 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.927e+02 3.047e+02 3.596e+02 4.345e+02 1.152e+03, threshold=7.193e+02, percent-clipped=4.0 2023-04-29 15:20:57,508 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117985.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:21:25,212 INFO [train.py:904] (7/8) Epoch 12, batch 6350, loss[loss=0.1961, simple_loss=0.2779, pruned_loss=0.05722, over 16952.00 frames. ], tot_loss[loss=0.218, simple_loss=0.3011, pruned_loss=0.06744, over 3127246.71 frames. ], batch size: 41, lr: 5.66e-03, grad_scale: 4.0 2023-04-29 15:22:12,495 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118033.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:22:18,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1630, 4.2006, 4.5845, 4.5294, 4.5375, 4.2676, 4.2583, 4.1219], device='cuda:7'), covar=tensor([0.0304, 0.0611, 0.0333, 0.0427, 0.0463, 0.0365, 0.0839, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0348, 0.0361, 0.0361, 0.0345, 0.0413, 0.0386, 0.0489, 0.0310], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 15:22:21,814 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:22:37,554 INFO [train.py:904] (7/8) Epoch 12, batch 6400, loss[loss=0.194, simple_loss=0.2797, pruned_loss=0.05413, over 16671.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.3006, pruned_loss=0.06824, over 3111887.91 frames. ], batch size: 62, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:22:38,528 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 15:22:57,664 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.235e+02 3.125e+02 4.036e+02 4.702e+02 8.359e+02, threshold=8.072e+02, percent-clipped=7.0 2023-04-29 15:23:50,400 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118101.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:23:50,965 INFO [train.py:904] (7/8) Epoch 12, batch 6450, loss[loss=0.2578, simple_loss=0.3132, pruned_loss=0.1012, over 11678.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2998, pruned_loss=0.0672, over 3114763.58 frames. ], batch size: 248, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:23:57,662 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0855, 3.4400, 3.4914, 1.8449, 2.9171, 2.3208, 3.4973, 3.6329], device='cuda:7'), covar=tensor([0.0299, 0.0706, 0.0548, 0.2068, 0.0864, 0.0962, 0.0623, 0.0897], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0148, 0.0161, 0.0146, 0.0139, 0.0127, 0.0140, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 15:24:25,622 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4634, 4.2531, 4.5113, 4.6890, 4.8059, 4.3577, 4.7753, 4.8167], device='cuda:7'), covar=tensor([0.1599, 0.1256, 0.1475, 0.0598, 0.0528, 0.0989, 0.0591, 0.0586], device='cuda:7'), in_proj_covar=tensor([0.0529, 0.0660, 0.0794, 0.0671, 0.0505, 0.0520, 0.0532, 0.0607], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:25:04,381 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7969, 1.7734, 2.2185, 2.6796, 2.6234, 3.0157, 1.7696, 2.9075], device='cuda:7'), covar=tensor([0.0149, 0.0359, 0.0237, 0.0204, 0.0210, 0.0128, 0.0399, 0.0093], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0171, 0.0154, 0.0158, 0.0169, 0.0125, 0.0172, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 15:25:08,002 INFO [train.py:904] (7/8) Epoch 12, batch 6500, loss[loss=0.2159, simple_loss=0.2842, pruned_loss=0.0738, over 12176.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2977, pruned_loss=0.06659, over 3115329.12 frames. ], batch size: 248, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:25:29,347 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.059e+02 2.874e+02 3.539e+02 4.309e+02 7.479e+02, threshold=7.078e+02, percent-clipped=0.0 2023-04-29 15:25:32,638 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118168.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:26:28,542 INFO [train.py:904] (7/8) Epoch 12, batch 6550, loss[loss=0.3142, simple_loss=0.3749, pruned_loss=0.1267, over 11416.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.3005, pruned_loss=0.06705, over 3131047.31 frames. ], batch size: 248, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:27:10,773 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118229.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 15:27:44,391 INFO [train.py:904] (7/8) Epoch 12, batch 6600, loss[loss=0.2044, simple_loss=0.2899, pruned_loss=0.05943, over 16729.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.3032, pruned_loss=0.06853, over 3091182.47 frames. ], batch size: 57, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:28:05,473 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 3.186e+02 3.899e+02 4.864e+02 8.341e+02, threshold=7.798e+02, percent-clipped=4.0 2023-04-29 15:28:10,867 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118269.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:28:26,023 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0366, 4.9998, 4.8039, 4.2402, 4.8851, 1.8995, 4.6069, 4.6151], device='cuda:7'), covar=tensor([0.0058, 0.0046, 0.0119, 0.0302, 0.0062, 0.2201, 0.0087, 0.0162], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0118, 0.0161, 0.0154, 0.0134, 0.0177, 0.0151, 0.0150], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:28:26,498 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-29 15:28:42,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118289.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:29:01,033 INFO [train.py:904] (7/8) Epoch 12, batch 6650, loss[loss=0.2151, simple_loss=0.2977, pruned_loss=0.06623, over 16896.00 frames. ], tot_loss[loss=0.22, simple_loss=0.3029, pruned_loss=0.06858, over 3117812.34 frames. ], batch size: 109, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:29:44,391 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118329.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:29:45,591 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118330.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:29:47,440 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-29 15:30:16,396 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118350.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:30:18,845 INFO [train.py:904] (7/8) Epoch 12, batch 6700, loss[loss=0.2048, simple_loss=0.2885, pruned_loss=0.06059, over 16518.00 frames. ], tot_loss[loss=0.2204, simple_loss=0.3025, pruned_loss=0.06919, over 3111578.16 frames. ], batch size: 75, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:30:39,919 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 3.120e+02 3.560e+02 4.149e+02 6.786e+02, threshold=7.119e+02, percent-clipped=0.0 2023-04-29 15:31:17,551 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118390.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:31:27,355 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118396.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:31:35,211 INFO [train.py:904] (7/8) Epoch 12, batch 6750, loss[loss=0.1951, simple_loss=0.2836, pruned_loss=0.05333, over 16444.00 frames. ], tot_loss[loss=0.2199, simple_loss=0.3013, pruned_loss=0.06928, over 3097828.89 frames. ], batch size: 75, lr: 5.66e-03, grad_scale: 8.0 2023-04-29 15:32:25,162 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2060, 2.1704, 2.3403, 4.1278, 1.9809, 2.5889, 2.2039, 2.3432], device='cuda:7'), covar=tensor([0.1157, 0.3255, 0.2347, 0.0368, 0.4021, 0.2236, 0.3076, 0.3051], device='cuda:7'), in_proj_covar=tensor([0.0365, 0.0394, 0.0331, 0.0316, 0.0413, 0.0454, 0.0361, 0.0461], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:32:49,917 INFO [train.py:904] (7/8) Epoch 12, batch 6800, loss[loss=0.2031, simple_loss=0.293, pruned_loss=0.05661, over 17035.00 frames. ], tot_loss[loss=0.2204, simple_loss=0.3016, pruned_loss=0.0696, over 3078112.05 frames. ], batch size: 55, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:33:11,649 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.917e+02 3.071e+02 3.777e+02 4.759e+02 7.416e+02, threshold=7.554e+02, percent-clipped=1.0 2023-04-29 15:34:04,798 INFO [train.py:904] (7/8) Epoch 12, batch 6850, loss[loss=0.2218, simple_loss=0.3308, pruned_loss=0.05635, over 16805.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.3031, pruned_loss=0.07012, over 3072647.25 frames. ], batch size: 83, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:34:36,279 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118524.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 15:35:15,895 INFO [train.py:904] (7/8) Epoch 12, batch 6900, loss[loss=0.2881, simple_loss=0.3398, pruned_loss=0.1182, over 11354.00 frames. ], tot_loss[loss=0.2222, simple_loss=0.3057, pruned_loss=0.06938, over 3102090.51 frames. ], batch size: 247, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:35:36,844 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 3.007e+02 3.634e+02 4.757e+02 1.105e+03, threshold=7.268e+02, percent-clipped=1.0 2023-04-29 15:35:58,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1157, 3.7446, 3.0432, 1.7168, 2.5459, 2.0698, 3.4497, 3.7691], device='cuda:7'), covar=tensor([0.0290, 0.0616, 0.0883, 0.2406, 0.1143, 0.1250, 0.0715, 0.0922], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0146, 0.0160, 0.0145, 0.0138, 0.0126, 0.0138, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 15:36:30,556 INFO [train.py:904] (7/8) Epoch 12, batch 6950, loss[loss=0.1945, simple_loss=0.2795, pruned_loss=0.05477, over 16681.00 frames. ], tot_loss[loss=0.2232, simple_loss=0.3063, pruned_loss=0.07007, over 3101515.94 frames. ], batch size: 57, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:37:04,590 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118625.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:37:15,306 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118631.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:37:26,175 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-29 15:37:35,445 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118645.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:37:44,793 INFO [train.py:904] (7/8) Epoch 12, batch 7000, loss[loss=0.2208, simple_loss=0.3134, pruned_loss=0.06415, over 16722.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.3064, pruned_loss=0.06974, over 3092773.41 frames. ], batch size: 134, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:38:05,443 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.982e+02 3.795e+02 4.664e+02 8.323e+02, threshold=7.591e+02, percent-clipped=2.0 2023-04-29 15:38:33,851 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118685.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:38:38,040 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-04-29 15:38:44,688 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:38:50,367 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118696.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:38:59,534 INFO [train.py:904] (7/8) Epoch 12, batch 7050, loss[loss=0.2592, simple_loss=0.3222, pruned_loss=0.09813, over 11160.00 frames. ], tot_loss[loss=0.2234, simple_loss=0.3074, pruned_loss=0.0697, over 3090869.24 frames. ], batch size: 246, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:39:07,334 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-29 15:40:01,790 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118744.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:40:14,607 INFO [train.py:904] (7/8) Epoch 12, batch 7100, loss[loss=0.277, simple_loss=0.3302, pruned_loss=0.1119, over 11069.00 frames. ], tot_loss[loss=0.2227, simple_loss=0.3058, pruned_loss=0.06978, over 3078530.86 frames. ], batch size: 248, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:40:36,858 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 3.120e+02 3.747e+02 4.649e+02 1.761e+03, threshold=7.495e+02, percent-clipped=5.0 2023-04-29 15:41:29,304 INFO [train.py:904] (7/8) Epoch 12, batch 7150, loss[loss=0.2191, simple_loss=0.3147, pruned_loss=0.06173, over 16674.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.3042, pruned_loss=0.06963, over 3073021.53 frames. ], batch size: 89, lr: 5.65e-03, grad_scale: 8.0 2023-04-29 15:42:01,942 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118824.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:42:16,543 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9058, 1.7577, 1.9985, 3.3231, 1.7876, 2.0190, 1.9351, 1.8499], device='cuda:7'), covar=tensor([0.1350, 0.4184, 0.2763, 0.0696, 0.5115, 0.2942, 0.3878, 0.4184], device='cuda:7'), in_proj_covar=tensor([0.0365, 0.0394, 0.0330, 0.0316, 0.0414, 0.0454, 0.0362, 0.0462], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:42:41,521 INFO [train.py:904] (7/8) Epoch 12, batch 7200, loss[loss=0.1944, simple_loss=0.2888, pruned_loss=0.05003, over 16680.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3019, pruned_loss=0.06772, over 3072783.01 frames. ], batch size: 134, lr: 5.64e-03, grad_scale: 8.0 2023-04-29 15:43:03,912 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.822e+02 2.566e+02 3.185e+02 3.996e+02 6.189e+02, threshold=6.369e+02, percent-clipped=0.0 2023-04-29 15:43:12,503 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:44:00,076 INFO [train.py:904] (7/8) Epoch 12, batch 7250, loss[loss=0.2212, simple_loss=0.2947, pruned_loss=0.07381, over 16179.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.2992, pruned_loss=0.06624, over 3070089.24 frames. ], batch size: 165, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:44:35,556 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118925.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:44:58,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1954, 3.5349, 3.7321, 1.9895, 3.0242, 2.4195, 3.5565, 3.7087], device='cuda:7'), covar=tensor([0.0269, 0.0719, 0.0501, 0.1937, 0.0809, 0.0928, 0.0664, 0.0991], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0147, 0.0161, 0.0145, 0.0139, 0.0126, 0.0139, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 15:45:04,540 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118945.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:45:15,114 INFO [train.py:904] (7/8) Epoch 12, batch 7300, loss[loss=0.206, simple_loss=0.2903, pruned_loss=0.06086, over 16587.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2989, pruned_loss=0.06607, over 3065609.49 frames. ], batch size: 68, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:45:36,388 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.034e+02 3.599e+02 4.380e+02 7.583e+02, threshold=7.199e+02, percent-clipped=5.0 2023-04-29 15:45:45,782 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118973.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:03,517 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118985.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:06,454 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118987.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:12,661 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 15:46:15,622 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=118993.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:46:28,512 INFO [train.py:904] (7/8) Epoch 12, batch 7350, loss[loss=0.2172, simple_loss=0.2999, pruned_loss=0.06722, over 16765.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2996, pruned_loss=0.06685, over 3065611.64 frames. ], batch size: 124, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:47:14,443 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119033.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:47:42,951 INFO [train.py:904] (7/8) Epoch 12, batch 7400, loss[loss=0.2205, simple_loss=0.3073, pruned_loss=0.06687, over 16686.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3015, pruned_loss=0.06796, over 3065457.63 frames. ], batch size: 134, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:48:06,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 3.304e+02 4.015e+02 4.796e+02 1.422e+03, threshold=8.030e+02, percent-clipped=7.0 2023-04-29 15:48:19,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119075.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:48:47,569 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119093.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:48:53,879 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7280, 3.7693, 4.1153, 4.1000, 4.1010, 3.8484, 3.8779, 3.8164], device='cuda:7'), covar=tensor([0.0341, 0.0596, 0.0408, 0.0432, 0.0445, 0.0432, 0.0850, 0.0471], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0354, 0.0354, 0.0333, 0.0401, 0.0375, 0.0472, 0.0301], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 15:49:01,209 INFO [train.py:904] (7/8) Epoch 12, batch 7450, loss[loss=0.242, simple_loss=0.3102, pruned_loss=0.08688, over 11290.00 frames. ], tot_loss[loss=0.2208, simple_loss=0.303, pruned_loss=0.06929, over 3055541.46 frames. ], batch size: 246, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:49:45,612 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7255, 1.2536, 1.6230, 1.6337, 1.7906, 1.7977, 1.5537, 1.7949], device='cuda:7'), covar=tensor([0.0192, 0.0283, 0.0151, 0.0204, 0.0211, 0.0142, 0.0273, 0.0095], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0154, 0.0157, 0.0169, 0.0124, 0.0171, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 15:49:55,972 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:50:20,478 INFO [train.py:904] (7/8) Epoch 12, batch 7500, loss[loss=0.1866, simple_loss=0.2737, pruned_loss=0.04979, over 16742.00 frames. ], tot_loss[loss=0.22, simple_loss=0.3031, pruned_loss=0.06846, over 3065520.30 frames. ], batch size: 124, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:50:24,102 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119154.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 15:50:42,266 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.952e+02 3.435e+02 4.438e+02 7.679e+02, threshold=6.870e+02, percent-clipped=0.0 2023-04-29 15:51:06,499 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4080, 2.9021, 2.6603, 2.2066, 2.2086, 2.2176, 2.9551, 2.8326], device='cuda:7'), covar=tensor([0.2270, 0.0734, 0.1398, 0.2205, 0.1938, 0.1783, 0.0477, 0.0993], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0258, 0.0289, 0.0283, 0.0282, 0.0224, 0.0270, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:51:35,619 INFO [train.py:904] (7/8) Epoch 12, batch 7550, loss[loss=0.2237, simple_loss=0.3052, pruned_loss=0.07105, over 16473.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.3022, pruned_loss=0.06859, over 3073832.61 frames. ], batch size: 35, lr: 5.64e-03, grad_scale: 4.0 2023-04-29 15:52:50,119 INFO [train.py:904] (7/8) Epoch 12, batch 7600, loss[loss=0.2297, simple_loss=0.3039, pruned_loss=0.07777, over 15256.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.301, pruned_loss=0.0686, over 3080307.69 frames. ], batch size: 190, lr: 5.64e-03, grad_scale: 8.0 2023-04-29 15:53:12,406 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.101e+02 2.984e+02 3.563e+02 4.291e+02 1.017e+03, threshold=7.126e+02, percent-clipped=2.0 2023-04-29 15:53:16,221 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8033, 2.5731, 2.6659, 4.6884, 2.4553, 3.0135, 2.6192, 2.8028], device='cuda:7'), covar=tensor([0.0914, 0.2861, 0.2114, 0.0346, 0.3528, 0.1911, 0.2660, 0.2773], device='cuda:7'), in_proj_covar=tensor([0.0365, 0.0396, 0.0331, 0.0318, 0.0416, 0.0454, 0.0361, 0.0460], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 15:53:43,872 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119287.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:54:05,863 INFO [train.py:904] (7/8) Epoch 12, batch 7650, loss[loss=0.2546, simple_loss=0.316, pruned_loss=0.09661, over 11293.00 frames. ], tot_loss[loss=0.2189, simple_loss=0.3009, pruned_loss=0.0684, over 3087911.19 frames. ], batch size: 247, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:54:55,654 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119335.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:55:00,199 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 15:55:20,167 INFO [train.py:904] (7/8) Epoch 12, batch 7700, loss[loss=0.2635, simple_loss=0.3228, pruned_loss=0.1021, over 12102.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3006, pruned_loss=0.06842, over 3102711.45 frames. ], batch size: 248, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:55:42,612 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.234e+02 3.396e+02 4.418e+02 5.495e+02 1.012e+03, threshold=8.835e+02, percent-clipped=5.0 2023-04-29 15:56:32,059 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8199, 3.2868, 2.8867, 5.1103, 4.0172, 4.4452, 1.6942, 3.4031], device='cuda:7'), covar=tensor([0.1313, 0.0614, 0.1078, 0.0105, 0.0359, 0.0358, 0.1492, 0.0706], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0162, 0.0182, 0.0151, 0.0199, 0.0208, 0.0184, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 15:56:36,114 INFO [train.py:904] (7/8) Epoch 12, batch 7750, loss[loss=0.1959, simple_loss=0.2834, pruned_loss=0.05425, over 17182.00 frames. ], tot_loss[loss=0.2184, simple_loss=0.3006, pruned_loss=0.06816, over 3117318.40 frames. ], batch size: 46, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:57:18,831 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119431.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 15:57:45,520 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119449.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 15:57:48,704 INFO [train.py:904] (7/8) Epoch 12, batch 7800, loss[loss=0.2186, simple_loss=0.3014, pruned_loss=0.06796, over 16278.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.3015, pruned_loss=0.06904, over 3113916.15 frames. ], batch size: 165, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 15:58:11,184 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.141e+02 3.879e+02 4.510e+02 7.221e+02, threshold=7.757e+02, percent-clipped=0.0 2023-04-29 15:59:04,885 INFO [train.py:904] (7/8) Epoch 12, batch 7850, loss[loss=0.2238, simple_loss=0.3017, pruned_loss=0.07291, over 16188.00 frames. ], tot_loss[loss=0.2201, simple_loss=0.3023, pruned_loss=0.06897, over 3105589.78 frames. ], batch size: 35, lr: 5.63e-03, grad_scale: 4.0 2023-04-29 15:59:24,420 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119514.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:00:21,534 INFO [train.py:904] (7/8) Epoch 12, batch 7900, loss[loss=0.2322, simple_loss=0.3002, pruned_loss=0.08213, over 11549.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3018, pruned_loss=0.06941, over 3078921.10 frames. ], batch size: 246, lr: 5.63e-03, grad_scale: 4.0 2023-04-29 16:00:45,732 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.963e+02 2.838e+02 3.537e+02 4.302e+02 9.445e+02, threshold=7.074e+02, percent-clipped=1.0 2023-04-29 16:00:56,336 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119575.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:01:38,577 INFO [train.py:904] (7/8) Epoch 12, batch 7950, loss[loss=0.2115, simple_loss=0.2928, pruned_loss=0.0651, over 16744.00 frames. ], tot_loss[loss=0.2217, simple_loss=0.3025, pruned_loss=0.07042, over 3051937.70 frames. ], batch size: 124, lr: 5.63e-03, grad_scale: 4.0 2023-04-29 16:02:39,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1052, 3.1043, 3.1661, 2.0832, 2.9625, 3.1832, 3.1071, 1.9083], device='cuda:7'), covar=tensor([0.0410, 0.0048, 0.0045, 0.0351, 0.0078, 0.0089, 0.0065, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0069, 0.0071, 0.0127, 0.0080, 0.0092, 0.0081, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 16:02:53,354 INFO [train.py:904] (7/8) Epoch 12, batch 8000, loss[loss=0.2727, simple_loss=0.3363, pruned_loss=0.1046, over 11300.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3022, pruned_loss=0.07024, over 3055267.67 frames. ], batch size: 247, lr: 5.63e-03, grad_scale: 8.0 2023-04-29 16:02:58,640 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7061, 4.7478, 4.5320, 4.2953, 4.1957, 4.6114, 4.5017, 4.2473], device='cuda:7'), covar=tensor([0.0617, 0.0505, 0.0276, 0.0272, 0.0939, 0.0471, 0.0402, 0.0655], device='cuda:7'), in_proj_covar=tensor([0.0240, 0.0324, 0.0285, 0.0264, 0.0303, 0.0307, 0.0197, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:03:13,397 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119665.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:03:17,120 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.898e+02 3.526e+02 5.291e+02 1.062e+03, threshold=7.052e+02, percent-clipped=9.0 2023-04-29 16:04:07,957 INFO [train.py:904] (7/8) Epoch 12, batch 8050, loss[loss=0.2225, simple_loss=0.3079, pruned_loss=0.06857, over 16833.00 frames. ], tot_loss[loss=0.2206, simple_loss=0.3018, pruned_loss=0.06969, over 3061815.57 frames. ], batch size: 116, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:04:26,769 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7861, 4.6564, 4.8069, 5.0111, 5.1521, 4.6358, 5.1455, 5.1109], device='cuda:7'), covar=tensor([0.1661, 0.1038, 0.1621, 0.0625, 0.0531, 0.0724, 0.0488, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0523, 0.0655, 0.0785, 0.0665, 0.0505, 0.0513, 0.0535, 0.0607], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:04:42,931 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119726.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:04:50,460 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119731.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:05:17,381 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119749.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 16:05:21,185 INFO [train.py:904] (7/8) Epoch 12, batch 8100, loss[loss=0.2308, simple_loss=0.2985, pruned_loss=0.08152, over 11284.00 frames. ], tot_loss[loss=0.219, simple_loss=0.301, pruned_loss=0.06854, over 3081511.89 frames. ], batch size: 248, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:05:47,745 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.274e+02 3.055e+02 3.528e+02 4.290e+02 7.598e+02, threshold=7.056e+02, percent-clipped=1.0 2023-04-29 16:05:48,750 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-29 16:05:52,296 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0224, 4.0273, 3.9355, 3.2892, 3.9345, 1.7235, 3.7625, 3.5112], device='cuda:7'), covar=tensor([0.0100, 0.0079, 0.0148, 0.0261, 0.0082, 0.2423, 0.0110, 0.0207], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0119, 0.0165, 0.0156, 0.0137, 0.0181, 0.0152, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:06:01,241 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119779.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:06:28,072 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=119797.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:06:34,773 INFO [train.py:904] (7/8) Epoch 12, batch 8150, loss[loss=0.2103, simple_loss=0.2885, pruned_loss=0.06605, over 16389.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2988, pruned_loss=0.06736, over 3089413.52 frames. ], batch size: 146, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:06:57,722 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2100, 5.5398, 5.2175, 5.2374, 4.9374, 4.7412, 4.9655, 5.6055], device='cuda:7'), covar=tensor([0.0967, 0.0695, 0.0907, 0.0666, 0.0751, 0.0848, 0.0989, 0.0835], device='cuda:7'), in_proj_covar=tensor([0.0544, 0.0672, 0.0559, 0.0478, 0.0429, 0.0447, 0.0566, 0.0523], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:07:50,610 INFO [train.py:904] (7/8) Epoch 12, batch 8200, loss[loss=0.2428, simple_loss=0.3052, pruned_loss=0.09023, over 11356.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2968, pruned_loss=0.06668, over 3100698.98 frames. ], batch size: 248, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:08:18,226 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 3.028e+02 3.797e+02 4.737e+02 9.915e+02, threshold=7.593e+02, percent-clipped=4.0 2023-04-29 16:08:19,312 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119870.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:08:37,009 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-29 16:08:38,875 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-29 16:09:09,081 INFO [train.py:904] (7/8) Epoch 12, batch 8250, loss[loss=0.2009, simple_loss=0.2937, pruned_loss=0.05406, over 15419.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2958, pruned_loss=0.06429, over 3080406.72 frames. ], batch size: 190, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:09:20,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7093, 2.1647, 1.8720, 1.9602, 2.4466, 2.1723, 2.4626, 2.6592], device='cuda:7'), covar=tensor([0.0115, 0.0293, 0.0361, 0.0342, 0.0200, 0.0269, 0.0174, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0201, 0.0197, 0.0196, 0.0202, 0.0199, 0.0204, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:10:28,024 INFO [train.py:904] (7/8) Epoch 12, batch 8300, loss[loss=0.186, simple_loss=0.2862, pruned_loss=0.04293, over 16564.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2932, pruned_loss=0.06189, over 3043372.71 frames. ], batch size: 75, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:10:47,029 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1028, 5.4214, 5.1328, 5.1814, 4.8670, 4.8200, 4.8105, 5.4885], device='cuda:7'), covar=tensor([0.0968, 0.0847, 0.0967, 0.0682, 0.0814, 0.0724, 0.1020, 0.0881], device='cuda:7'), in_proj_covar=tensor([0.0539, 0.0666, 0.0554, 0.0474, 0.0424, 0.0443, 0.0559, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:10:57,564 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.352e+02 2.901e+02 3.420e+02 5.863e+02, threshold=5.801e+02, percent-clipped=0.0 2023-04-29 16:11:17,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7360, 3.7502, 2.8892, 2.0237, 2.5181, 2.3074, 4.0673, 3.5518], device='cuda:7'), covar=tensor([0.2631, 0.0732, 0.1652, 0.2851, 0.2487, 0.1987, 0.0411, 0.1022], device='cuda:7'), in_proj_covar=tensor([0.0302, 0.0253, 0.0283, 0.0278, 0.0277, 0.0220, 0.0265, 0.0293], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:11:52,966 INFO [train.py:904] (7/8) Epoch 12, batch 8350, loss[loss=0.198, simple_loss=0.2769, pruned_loss=0.05961, over 12430.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2915, pruned_loss=0.05917, over 3051984.96 frames. ], batch size: 248, lr: 5.62e-03, grad_scale: 2.0 2023-04-29 16:12:03,964 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7426, 3.7571, 4.0918, 4.0583, 4.0936, 3.8562, 3.8522, 3.8753], device='cuda:7'), covar=tensor([0.0310, 0.0695, 0.0416, 0.0459, 0.0396, 0.0403, 0.0810, 0.0425], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0354, 0.0350, 0.0333, 0.0402, 0.0375, 0.0471, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 16:12:24,537 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=120021.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:13:14,450 INFO [train.py:904] (7/8) Epoch 12, batch 8400, loss[loss=0.1913, simple_loss=0.2744, pruned_loss=0.05408, over 11832.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2884, pruned_loss=0.05727, over 3015374.02 frames. ], batch size: 246, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:13:42,956 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.479e+02 2.965e+02 3.428e+02 6.852e+02, threshold=5.930e+02, percent-clipped=2.0 2023-04-29 16:14:01,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7288, 2.8531, 2.4370, 4.1357, 2.8628, 4.1035, 1.5132, 3.0691], device='cuda:7'), covar=tensor([0.1305, 0.0623, 0.1131, 0.0147, 0.0180, 0.0349, 0.1511, 0.0625], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0160, 0.0181, 0.0148, 0.0196, 0.0206, 0.0182, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 16:14:31,501 INFO [train.py:904] (7/8) Epoch 12, batch 8450, loss[loss=0.1895, simple_loss=0.2791, pruned_loss=0.04995, over 16386.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2861, pruned_loss=0.05503, over 3030677.28 frames. ], batch size: 146, lr: 5.62e-03, grad_scale: 4.0 2023-04-29 16:15:00,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4124, 3.3629, 3.4641, 3.5703, 3.5881, 3.2796, 3.5558, 3.6337], device='cuda:7'), covar=tensor([0.1158, 0.0869, 0.1022, 0.0609, 0.0591, 0.2581, 0.0858, 0.0657], device='cuda:7'), in_proj_covar=tensor([0.0509, 0.0639, 0.0766, 0.0652, 0.0493, 0.0503, 0.0522, 0.0594], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:15:24,273 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4519, 3.5123, 2.0109, 3.8227, 2.5815, 3.8026, 2.0949, 2.8982], device='cuda:7'), covar=tensor([0.0221, 0.0257, 0.1417, 0.0176, 0.0714, 0.0441, 0.1544, 0.0577], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0155, 0.0182, 0.0124, 0.0161, 0.0195, 0.0190, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 16:15:25,535 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0696, 1.4253, 1.7716, 2.1196, 2.1744, 2.2715, 1.7002, 2.2151], device='cuda:7'), covar=tensor([0.0173, 0.0371, 0.0219, 0.0207, 0.0216, 0.0137, 0.0316, 0.0096], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0171, 0.0155, 0.0156, 0.0169, 0.0122, 0.0168, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 16:15:47,459 INFO [train.py:904] (7/8) Epoch 12, batch 8500, loss[loss=0.1878, simple_loss=0.2775, pruned_loss=0.04904, over 16517.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2827, pruned_loss=0.05256, over 3046172.06 frames. ], batch size: 62, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:16:15,432 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.373e+02 2.767e+02 3.491e+02 7.213e+02, threshold=5.534e+02, percent-clipped=3.0 2023-04-29 16:16:15,850 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120170.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:17:07,726 INFO [train.py:904] (7/8) Epoch 12, batch 8550, loss[loss=0.2073, simple_loss=0.3018, pruned_loss=0.05646, over 16812.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2803, pruned_loss=0.05153, over 3026032.21 frames. ], batch size: 102, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:17:20,350 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-29 16:17:37,326 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=120218.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:18:39,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9826, 1.8216, 1.6225, 1.5115, 1.9472, 1.5631, 1.6981, 1.9655], device='cuda:7'), covar=tensor([0.0125, 0.0234, 0.0305, 0.0274, 0.0163, 0.0210, 0.0163, 0.0168], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0201, 0.0196, 0.0196, 0.0201, 0.0198, 0.0201, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:18:41,890 INFO [train.py:904] (7/8) Epoch 12, batch 8600, loss[loss=0.191, simple_loss=0.2874, pruned_loss=0.04726, over 16255.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2808, pruned_loss=0.05044, over 3031386.79 frames. ], batch size: 165, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:18:58,228 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8596, 4.8661, 5.3274, 5.3070, 5.2907, 4.9912, 4.9378, 4.7009], device='cuda:7'), covar=tensor([0.0268, 0.0569, 0.0337, 0.0335, 0.0431, 0.0330, 0.0883, 0.0383], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0353, 0.0346, 0.0331, 0.0400, 0.0372, 0.0466, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 16:19:00,396 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-04-29 16:19:19,443 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.485e+02 2.557e+02 3.073e+02 4.030e+02 6.956e+02, threshold=6.147e+02, percent-clipped=5.0 2023-04-29 16:20:19,200 INFO [train.py:904] (7/8) Epoch 12, batch 8650, loss[loss=0.1743, simple_loss=0.2634, pruned_loss=0.04263, over 12058.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2788, pruned_loss=0.04885, over 3034666.01 frames. ], batch size: 246, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:20:25,842 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-29 16:21:01,181 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=120321.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:22:02,152 INFO [train.py:904] (7/8) Epoch 12, batch 8700, loss[loss=0.1717, simple_loss=0.2611, pruned_loss=0.04117, over 12244.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2763, pruned_loss=0.0475, over 3050058.41 frames. ], batch size: 248, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:22:33,633 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=120369.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:22:34,436 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.303e+02 2.810e+02 3.260e+02 5.191e+02, threshold=5.620e+02, percent-clipped=0.0 2023-04-29 16:23:36,533 INFO [train.py:904] (7/8) Epoch 12, batch 8750, loss[loss=0.1681, simple_loss=0.2695, pruned_loss=0.03338, over 16849.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2755, pruned_loss=0.04686, over 3049917.23 frames. ], batch size: 90, lr: 5.61e-03, grad_scale: 4.0 2023-04-29 16:25:11,310 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-29 16:25:20,246 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0270, 1.8757, 1.6645, 1.5656, 1.9822, 1.6162, 1.7656, 1.9928], device='cuda:7'), covar=tensor([0.0114, 0.0200, 0.0285, 0.0251, 0.0162, 0.0205, 0.0139, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0202, 0.0197, 0.0196, 0.0202, 0.0200, 0.0201, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:25:30,738 INFO [train.py:904] (7/8) Epoch 12, batch 8800, loss[loss=0.1883, simple_loss=0.2809, pruned_loss=0.04787, over 16980.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2742, pruned_loss=0.04573, over 3060230.91 frames. ], batch size: 109, lr: 5.61e-03, grad_scale: 8.0 2023-04-29 16:25:44,284 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.93 vs. limit=5.0 2023-04-29 16:26:08,431 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.538e+02 3.135e+02 3.629e+02 6.620e+02, threshold=6.271e+02, percent-clipped=2.0 2023-04-29 16:27:17,127 INFO [train.py:904] (7/8) Epoch 12, batch 8850, loss[loss=0.1908, simple_loss=0.2717, pruned_loss=0.05495, over 12341.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2759, pruned_loss=0.04507, over 3052817.21 frames. ], batch size: 248, lr: 5.61e-03, grad_scale: 8.0 2023-04-29 16:27:41,102 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.39 vs. limit=5.0 2023-04-29 16:28:59,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0574, 3.1368, 1.8322, 3.3068, 2.3069, 3.2866, 2.0923, 2.5967], device='cuda:7'), covar=tensor([0.0239, 0.0290, 0.1369, 0.0201, 0.0713, 0.0509, 0.1298, 0.0619], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0155, 0.0182, 0.0125, 0.0163, 0.0195, 0.0191, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 16:29:04,162 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 16:29:04,468 INFO [train.py:904] (7/8) Epoch 12, batch 8900, loss[loss=0.1715, simple_loss=0.2696, pruned_loss=0.03667, over 16469.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2764, pruned_loss=0.04439, over 3055300.06 frames. ], batch size: 146, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:29:26,274 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7012, 4.6893, 4.5142, 4.1933, 4.2016, 4.5894, 4.4898, 4.2298], device='cuda:7'), covar=tensor([0.0570, 0.0603, 0.0310, 0.0335, 0.0969, 0.0487, 0.0362, 0.0754], device='cuda:7'), in_proj_covar=tensor([0.0234, 0.0313, 0.0278, 0.0258, 0.0293, 0.0296, 0.0193, 0.0323], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:29:37,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5057, 3.7931, 3.9692, 2.8581, 3.4651, 3.8337, 3.5985, 2.1890], device='cuda:7'), covar=tensor([0.0397, 0.0027, 0.0024, 0.0255, 0.0063, 0.0057, 0.0056, 0.0379], device='cuda:7'), in_proj_covar=tensor([0.0126, 0.0067, 0.0068, 0.0123, 0.0078, 0.0087, 0.0079, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 16:29:39,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.788e+02 2.488e+02 2.995e+02 3.537e+02 1.098e+03, threshold=5.991e+02, percent-clipped=1.0 2023-04-29 16:31:10,996 INFO [train.py:904] (7/8) Epoch 12, batch 8950, loss[loss=0.1662, simple_loss=0.2623, pruned_loss=0.03502, over 16610.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.276, pruned_loss=0.0443, over 3091931.12 frames. ], batch size: 83, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:33:00,379 INFO [train.py:904] (7/8) Epoch 12, batch 9000, loss[loss=0.1584, simple_loss=0.2477, pruned_loss=0.03448, over 16477.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2721, pruned_loss=0.04281, over 3091538.38 frames. ], batch size: 68, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:33:00,380 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 16:33:10,351 INFO [train.py:938] (7/8) Epoch 12, validation: loss=0.1532, simple_loss=0.2571, pruned_loss=0.02465, over 944034.00 frames. 2023-04-29 16:33:10,352 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 16:33:49,119 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.337e+02 2.622e+02 3.241e+02 7.245e+02, threshold=5.244e+02, percent-clipped=4.0 2023-04-29 16:34:01,171 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=120676.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:34:54,273 INFO [train.py:904] (7/8) Epoch 12, batch 9050, loss[loss=0.1744, simple_loss=0.2563, pruned_loss=0.0463, over 16686.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2733, pruned_loss=0.0435, over 3092028.71 frames. ], batch size: 134, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:35:27,443 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7974, 2.7280, 2.6730, 1.9530, 2.5319, 2.8188, 2.7217, 1.8342], device='cuda:7'), covar=tensor([0.0412, 0.0043, 0.0043, 0.0308, 0.0081, 0.0067, 0.0068, 0.0397], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0067, 0.0068, 0.0124, 0.0079, 0.0088, 0.0079, 0.0118], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 16:35:43,845 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1703, 4.2386, 4.0672, 3.7597, 3.7533, 4.1578, 3.9063, 3.8592], device='cuda:7'), covar=tensor([0.0557, 0.0515, 0.0306, 0.0307, 0.0785, 0.0452, 0.0610, 0.0681], device='cuda:7'), in_proj_covar=tensor([0.0232, 0.0312, 0.0275, 0.0256, 0.0291, 0.0296, 0.0191, 0.0320], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:36:04,481 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=120737.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:36:36,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1253, 5.5010, 5.2367, 5.2212, 4.9114, 4.8830, 4.8398, 5.5475], device='cuda:7'), covar=tensor([0.1122, 0.0883, 0.0864, 0.0628, 0.0719, 0.0703, 0.1087, 0.0834], device='cuda:7'), in_proj_covar=tensor([0.0528, 0.0654, 0.0539, 0.0465, 0.0419, 0.0433, 0.0546, 0.0510], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:36:39,410 INFO [train.py:904] (7/8) Epoch 12, batch 9100, loss[loss=0.1713, simple_loss=0.2621, pruned_loss=0.0402, over 12076.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.274, pruned_loss=0.04412, over 3097133.88 frames. ], batch size: 247, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:37:15,451 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.513e+02 2.910e+02 3.503e+02 5.095e+02, threshold=5.819e+02, percent-clipped=0.0 2023-04-29 16:37:27,118 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4766, 3.7993, 3.9396, 2.7658, 3.4513, 3.8293, 3.6756, 2.1488], device='cuda:7'), covar=tensor([0.0417, 0.0033, 0.0027, 0.0279, 0.0078, 0.0067, 0.0050, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0125, 0.0066, 0.0067, 0.0122, 0.0078, 0.0086, 0.0077, 0.0116], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 16:38:36,949 INFO [train.py:904] (7/8) Epoch 12, batch 9150, loss[loss=0.1747, simple_loss=0.2663, pruned_loss=0.04156, over 16488.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2746, pruned_loss=0.04392, over 3098877.37 frames. ], batch size: 68, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:40:21,528 INFO [train.py:904] (7/8) Epoch 12, batch 9200, loss[loss=0.157, simple_loss=0.2459, pruned_loss=0.03406, over 11963.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2707, pruned_loss=0.04325, over 3094524.30 frames. ], batch size: 248, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:40:55,540 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.751e+02 2.400e+02 2.881e+02 3.588e+02 6.775e+02, threshold=5.761e+02, percent-clipped=4.0 2023-04-29 16:41:01,498 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-29 16:41:03,853 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6224, 2.8304, 2.3577, 4.3429, 2.5690, 4.0147, 1.4820, 2.9314], device='cuda:7'), covar=tensor([0.1483, 0.0754, 0.1314, 0.0160, 0.0162, 0.0381, 0.1696, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0158, 0.0179, 0.0143, 0.0187, 0.0202, 0.0181, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 16:41:15,674 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-29 16:42:00,506 INFO [train.py:904] (7/8) Epoch 12, batch 9250, loss[loss=0.1543, simple_loss=0.2408, pruned_loss=0.03389, over 12566.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2697, pruned_loss=0.04333, over 3048704.99 frames. ], batch size: 248, lr: 5.60e-03, grad_scale: 8.0 2023-04-29 16:42:15,637 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3417, 5.6969, 5.4189, 5.4600, 5.0795, 4.9952, 5.0362, 5.7428], device='cuda:7'), covar=tensor([0.0957, 0.0924, 0.1006, 0.0704, 0.0712, 0.0735, 0.1001, 0.0924], device='cuda:7'), in_proj_covar=tensor([0.0526, 0.0655, 0.0538, 0.0463, 0.0418, 0.0435, 0.0545, 0.0510], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:42:24,486 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 16:43:49,110 INFO [train.py:904] (7/8) Epoch 12, batch 9300, loss[loss=0.1655, simple_loss=0.2465, pruned_loss=0.04221, over 16548.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.268, pruned_loss=0.04293, over 3030885.42 frames. ], batch size: 62, lr: 5.60e-03, grad_scale: 4.0 2023-04-29 16:44:07,084 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1705, 2.0440, 2.1944, 3.6145, 2.0879, 2.3591, 2.2014, 2.1977], device='cuda:7'), covar=tensor([0.0961, 0.3290, 0.2425, 0.0437, 0.3850, 0.2306, 0.3145, 0.3221], device='cuda:7'), in_proj_covar=tensor([0.0354, 0.0384, 0.0326, 0.0309, 0.0406, 0.0437, 0.0352, 0.0446], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:44:31,805 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.549e+02 3.003e+02 3.626e+02 5.994e+02, threshold=6.005e+02, percent-clipped=1.0 2023-04-29 16:45:32,162 INFO [train.py:904] (7/8) Epoch 12, batch 9350, loss[loss=0.1814, simple_loss=0.2636, pruned_loss=0.04964, over 12228.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2684, pruned_loss=0.04323, over 3038753.29 frames. ], batch size: 248, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:45:35,681 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121003.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:46:33,085 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:47:13,259 INFO [train.py:904] (7/8) Epoch 12, batch 9400, loss[loss=0.1763, simple_loss=0.2795, pruned_loss=0.03654, over 16343.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2688, pruned_loss=0.04311, over 3048567.36 frames. ], batch size: 146, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:47:26,348 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4709, 2.1481, 2.2193, 3.9721, 2.0260, 2.5813, 2.2763, 2.2883], device='cuda:7'), covar=tensor([0.0831, 0.3259, 0.2310, 0.0383, 0.3839, 0.2120, 0.2936, 0.3294], device='cuda:7'), in_proj_covar=tensor([0.0354, 0.0384, 0.0326, 0.0310, 0.0406, 0.0437, 0.0351, 0.0446], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:47:39,259 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121064.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:47:50,521 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.435e+02 2.897e+02 3.662e+02 7.614e+02, threshold=5.793e+02, percent-clipped=2.0 2023-04-29 16:48:04,427 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1674, 5.1283, 4.9167, 4.5050, 4.9181, 1.8875, 4.7070, 4.7831], device='cuda:7'), covar=tensor([0.0051, 0.0048, 0.0117, 0.0191, 0.0061, 0.2140, 0.0083, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0115, 0.0157, 0.0146, 0.0132, 0.0178, 0.0147, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 16:48:55,048 INFO [train.py:904] (7/8) Epoch 12, batch 9450, loss[loss=0.1647, simple_loss=0.2537, pruned_loss=0.03786, over 12370.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2699, pruned_loss=0.04318, over 3031432.48 frames. ], batch size: 248, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:49:39,766 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2428, 2.2247, 2.6294, 3.4693, 3.0528, 3.8102, 2.3463, 3.6156], device='cuda:7'), covar=tensor([0.0151, 0.0372, 0.0289, 0.0169, 0.0201, 0.0089, 0.0355, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0156, 0.0157, 0.0167, 0.0121, 0.0170, 0.0115], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 16:49:47,162 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5672, 3.0428, 3.2096, 1.8586, 2.7984, 2.1839, 3.1655, 3.0923], device='cuda:7'), covar=tensor([0.0267, 0.0745, 0.0540, 0.1927, 0.0729, 0.0948, 0.0691, 0.1006], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0138, 0.0155, 0.0141, 0.0133, 0.0123, 0.0133, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 16:50:34,763 INFO [train.py:904] (7/8) Epoch 12, batch 9500, loss[loss=0.1628, simple_loss=0.2571, pruned_loss=0.03424, over 16883.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2693, pruned_loss=0.04267, over 3047546.64 frames. ], batch size: 96, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:51:01,983 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121164.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:51:13,621 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.296e+02 2.757e+02 3.587e+02 6.751e+02, threshold=5.515e+02, percent-clipped=2.0 2023-04-29 16:52:20,005 INFO [train.py:904] (7/8) Epoch 12, batch 9550, loss[loss=0.1865, simple_loss=0.2725, pruned_loss=0.05027, over 12450.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2693, pruned_loss=0.04256, over 3071334.29 frames. ], batch size: 248, lr: 5.59e-03, grad_scale: 4.0 2023-04-29 16:53:08,753 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121225.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:53:59,938 INFO [train.py:904] (7/8) Epoch 12, batch 9600, loss[loss=0.1932, simple_loss=0.2938, pruned_loss=0.04624, over 16562.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2711, pruned_loss=0.04369, over 3068321.16 frames. ], batch size: 68, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:54:35,191 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.441e+02 3.037e+02 3.482e+02 7.490e+02, threshold=6.075e+02, percent-clipped=2.0 2023-04-29 16:55:12,448 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8855, 5.2794, 5.3884, 5.1981, 5.2247, 5.7914, 5.3091, 5.0504], device='cuda:7'), covar=tensor([0.0801, 0.1679, 0.1642, 0.2010, 0.2410, 0.0870, 0.1331, 0.2283], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0470, 0.0524, 0.0406, 0.0541, 0.0545, 0.0410, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 16:55:45,658 INFO [train.py:904] (7/8) Epoch 12, batch 9650, loss[loss=0.1873, simple_loss=0.274, pruned_loss=0.05027, over 16973.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2733, pruned_loss=0.04441, over 3052165.82 frames. ], batch size: 109, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:56:10,114 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121311.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 16:56:51,953 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121332.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:57:28,707 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7042, 3.7062, 3.9296, 1.8118, 4.0662, 4.0990, 3.0620, 3.1517], device='cuda:7'), covar=tensor([0.0636, 0.0172, 0.0131, 0.1223, 0.0052, 0.0098, 0.0353, 0.0389], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0098, 0.0084, 0.0135, 0.0067, 0.0102, 0.0119, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 16:57:30,750 INFO [train.py:904] (7/8) Epoch 12, batch 9700, loss[loss=0.1802, simple_loss=0.2643, pruned_loss=0.04804, over 12518.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2714, pruned_loss=0.04398, over 3045017.30 frames. ], batch size: 248, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:57:44,335 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121359.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:58:07,292 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.472e+02 3.175e+02 3.928e+02 9.920e+02, threshold=6.351e+02, percent-clipped=4.0 2023-04-29 16:58:11,351 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121372.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 16:58:31,089 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=121380.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:58:33,127 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-29 16:59:04,711 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5666, 4.6072, 4.4124, 4.0557, 4.0897, 4.5229, 4.3666, 4.1898], device='cuda:7'), covar=tensor([0.0594, 0.0530, 0.0308, 0.0303, 0.0937, 0.0420, 0.0413, 0.0689], device='cuda:7'), in_proj_covar=tensor([0.0232, 0.0304, 0.0273, 0.0252, 0.0287, 0.0293, 0.0188, 0.0317], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-29 16:59:14,344 INFO [train.py:904] (7/8) Epoch 12, batch 9750, loss[loss=0.171, simple_loss=0.2717, pruned_loss=0.03521, over 15428.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2702, pruned_loss=0.04422, over 3023844.19 frames. ], batch size: 191, lr: 5.59e-03, grad_scale: 8.0 2023-04-29 16:59:23,929 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121407.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 16:59:53,811 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-29 16:59:58,302 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-29 17:00:39,171 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5715, 1.6821, 2.0802, 2.6332, 2.4871, 2.7690, 1.9080, 2.7500], device='cuda:7'), covar=tensor([0.0172, 0.0377, 0.0258, 0.0195, 0.0224, 0.0131, 0.0345, 0.0104], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0168, 0.0154, 0.0156, 0.0166, 0.0120, 0.0167, 0.0113], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 17:00:53,753 INFO [train.py:904] (7/8) Epoch 12, batch 9800, loss[loss=0.1625, simple_loss=0.269, pruned_loss=0.02795, over 16686.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2701, pruned_loss=0.04302, over 3039250.12 frames. ], batch size: 89, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:01:24,635 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=121468.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:01:29,905 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.372e+02 2.708e+02 3.035e+02 5.819e+02, threshold=5.415e+02, percent-clipped=0.0 2023-04-29 17:02:38,505 INFO [train.py:904] (7/8) Epoch 12, batch 9850, loss[loss=0.1792, simple_loss=0.2697, pruned_loss=0.0443, over 16214.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2713, pruned_loss=0.04314, over 3042273.87 frames. ], batch size: 165, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:03:17,657 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121520.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:03:23,812 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1041, 4.0316, 4.2339, 4.3449, 4.4722, 3.9960, 4.4387, 4.4889], device='cuda:7'), covar=tensor([0.1499, 0.1031, 0.1299, 0.0648, 0.0492, 0.1190, 0.0575, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0500, 0.0625, 0.0750, 0.0640, 0.0482, 0.0495, 0.0508, 0.0581], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:04:30,202 INFO [train.py:904] (7/8) Epoch 12, batch 9900, loss[loss=0.1919, simple_loss=0.2919, pruned_loss=0.04594, over 16751.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2717, pruned_loss=0.04275, over 3047212.38 frames. ], batch size: 134, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:05:12,986 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.385e+02 2.886e+02 3.421e+02 6.283e+02, threshold=5.771e+02, percent-clipped=3.0 2023-04-29 17:05:56,129 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-29 17:06:26,649 INFO [train.py:904] (7/8) Epoch 12, batch 9950, loss[loss=0.1973, simple_loss=0.2787, pruned_loss=0.05799, over 12165.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2743, pruned_loss=0.04339, over 3048849.21 frames. ], batch size: 248, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:08:27,936 INFO [train.py:904] (7/8) Epoch 12, batch 10000, loss[loss=0.1778, simple_loss=0.2754, pruned_loss=0.04012, over 16223.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2726, pruned_loss=0.04278, over 3068917.80 frames. ], batch size: 165, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:08:44,240 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121659.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:08:59,919 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121667.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 17:09:06,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.322e+02 2.719e+02 3.378e+02 7.215e+02, threshold=5.438e+02, percent-clipped=2.0 2023-04-29 17:09:56,835 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7650, 3.7911, 2.4169, 4.4436, 2.8844, 4.3346, 2.6610, 3.0228], device='cuda:7'), covar=tensor([0.0217, 0.0335, 0.1384, 0.0117, 0.0821, 0.0360, 0.1294, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0155, 0.0180, 0.0123, 0.0162, 0.0191, 0.0191, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 17:10:10,832 INFO [train.py:904] (7/8) Epoch 12, batch 10050, loss[loss=0.1739, simple_loss=0.271, pruned_loss=0.03833, over 16799.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2729, pruned_loss=0.04273, over 3066859.59 frames. ], batch size: 76, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:10:21,646 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=121707.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:10:45,178 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4913, 1.5655, 2.0181, 2.4597, 2.3419, 2.6478, 1.9645, 2.6770], device='cuda:7'), covar=tensor([0.0198, 0.0418, 0.0302, 0.0277, 0.0276, 0.0180, 0.0353, 0.0131], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0156, 0.0158, 0.0169, 0.0122, 0.0170, 0.0114], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001], device='cuda:7') 2023-04-29 17:11:46,097 INFO [train.py:904] (7/8) Epoch 12, batch 10100, loss[loss=0.1648, simple_loss=0.2451, pruned_loss=0.04221, over 12598.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2734, pruned_loss=0.04327, over 3076958.34 frames. ], batch size: 249, lr: 5.58e-03, grad_scale: 8.0 2023-04-29 17:12:06,127 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=121763.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:12:23,725 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.328e+02 2.737e+02 3.349e+02 5.090e+02, threshold=5.474e+02, percent-clipped=0.0 2023-04-29 17:13:30,554 INFO [train.py:904] (7/8) Epoch 13, batch 0, loss[loss=0.2445, simple_loss=0.2963, pruned_loss=0.0963, over 16625.00 frames. ], tot_loss[loss=0.2445, simple_loss=0.2963, pruned_loss=0.0963, over 16625.00 frames. ], batch size: 76, lr: 5.36e-03, grad_scale: 8.0 2023-04-29 17:13:30,554 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 17:13:38,111 INFO [train.py:938] (7/8) Epoch 13, validation: loss=0.1523, simple_loss=0.2559, pruned_loss=0.02431, over 944034.00 frames. 2023-04-29 17:13:38,111 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 17:14:04,019 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121820.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:14:30,569 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1823, 5.7979, 5.9682, 5.6725, 5.7532, 6.3008, 5.8638, 5.6394], device='cuda:7'), covar=tensor([0.0736, 0.1673, 0.1901, 0.1881, 0.2331, 0.0836, 0.1398, 0.1921], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0476, 0.0525, 0.0413, 0.0548, 0.0547, 0.0416, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 17:14:49,509 INFO [train.py:904] (7/8) Epoch 13, batch 50, loss[loss=0.1686, simple_loss=0.2525, pruned_loss=0.04234, over 16756.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2826, pruned_loss=0.06267, over 759027.41 frames. ], batch size: 39, lr: 5.36e-03, grad_scale: 2.0 2023-04-29 17:15:01,512 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7568, 3.1881, 2.8926, 5.0795, 4.2374, 4.5028, 1.6250, 3.3276], device='cuda:7'), covar=tensor([0.1373, 0.0660, 0.1085, 0.0131, 0.0246, 0.0381, 0.1577, 0.0706], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0156, 0.0179, 0.0143, 0.0184, 0.0202, 0.0180, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 17:15:11,441 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=121868.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:15:18,825 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.864e+02 2.677e+02 3.187e+02 4.069e+02 9.250e+02, threshold=6.374e+02, percent-clipped=6.0 2023-04-29 17:15:41,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8291, 4.6151, 4.8590, 5.0429, 5.2416, 4.5974, 5.2124, 5.2314], device='cuda:7'), covar=tensor([0.1566, 0.1295, 0.1605, 0.0684, 0.0542, 0.0877, 0.0626, 0.0538], device='cuda:7'), in_proj_covar=tensor([0.0506, 0.0633, 0.0759, 0.0646, 0.0486, 0.0497, 0.0511, 0.0585], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:15:54,753 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3313, 2.1259, 1.6265, 1.8956, 2.4999, 2.2407, 2.4273, 2.5960], device='cuda:7'), covar=tensor([0.0139, 0.0310, 0.0412, 0.0383, 0.0173, 0.0275, 0.0158, 0.0191], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0209, 0.0203, 0.0202, 0.0207, 0.0207, 0.0206, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:15:58,271 INFO [train.py:904] (7/8) Epoch 13, batch 100, loss[loss=0.2079, simple_loss=0.2796, pruned_loss=0.06805, over 16866.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2807, pruned_loss=0.05978, over 1321110.63 frames. ], batch size: 90, lr: 5.36e-03, grad_scale: 2.0 2023-04-29 17:17:04,434 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=121950.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:17:07,075 INFO [train.py:904] (7/8) Epoch 13, batch 150, loss[loss=0.1563, simple_loss=0.2427, pruned_loss=0.03495, over 16796.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2775, pruned_loss=0.05782, over 1754432.01 frames. ], batch size: 39, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:17:27,840 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=121967.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 17:17:35,632 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.538e+02 3.283e+02 4.116e+02 1.264e+03, threshold=6.566e+02, percent-clipped=3.0 2023-04-29 17:18:18,203 INFO [train.py:904] (7/8) Epoch 13, batch 200, loss[loss=0.1858, simple_loss=0.285, pruned_loss=0.04329, over 17033.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2751, pruned_loss=0.05566, over 2107519.43 frames. ], batch size: 53, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:18:30,546 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122011.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:18:35,253 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122015.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 17:18:39,830 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 17:19:02,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6951, 3.7354, 2.0588, 3.9771, 2.8711, 3.8281, 2.1955, 2.8894], device='cuda:7'), covar=tensor([0.0228, 0.0332, 0.1636, 0.0286, 0.0759, 0.0753, 0.1469, 0.0678], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0160, 0.0185, 0.0130, 0.0168, 0.0199, 0.0194, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 17:19:20,001 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 17:19:26,248 INFO [train.py:904] (7/8) Epoch 13, batch 250, loss[loss=0.1619, simple_loss=0.2501, pruned_loss=0.03685, over 15926.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2724, pruned_loss=0.05432, over 2377652.74 frames. ], batch size: 35, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:19:30,326 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 17:19:41,371 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122063.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:19:54,203 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.354e+02 2.915e+02 3.517e+02 1.210e+03, threshold=5.831e+02, percent-clipped=2.0 2023-04-29 17:20:34,750 INFO [train.py:904] (7/8) Epoch 13, batch 300, loss[loss=0.173, simple_loss=0.262, pruned_loss=0.04199, over 17163.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2693, pruned_loss=0.05296, over 2582968.40 frames. ], batch size: 46, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:20:48,064 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122111.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:20:55,827 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122117.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:21:37,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8948, 1.9229, 2.3869, 2.8945, 2.7100, 3.3019, 2.1027, 3.1796], device='cuda:7'), covar=tensor([0.0175, 0.0369, 0.0233, 0.0229, 0.0236, 0.0146, 0.0359, 0.0133], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0175, 0.0161, 0.0164, 0.0174, 0.0127, 0.0176, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 17:21:42,734 INFO [train.py:904] (7/8) Epoch 13, batch 350, loss[loss=0.1996, simple_loss=0.2783, pruned_loss=0.06052, over 16451.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2671, pruned_loss=0.05157, over 2755918.71 frames. ], batch size: 146, lr: 5.35e-03, grad_scale: 2.0 2023-04-29 17:21:48,325 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7998, 1.8236, 2.3093, 2.6715, 2.6962, 2.6332, 1.8007, 2.8457], device='cuda:7'), covar=tensor([0.0142, 0.0365, 0.0239, 0.0221, 0.0210, 0.0189, 0.0389, 0.0109], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0174, 0.0161, 0.0163, 0.0174, 0.0127, 0.0175, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 17:21:52,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6159, 2.6789, 2.2521, 2.5752, 3.0459, 2.7775, 3.4711, 3.2224], device='cuda:7'), covar=tensor([0.0092, 0.0304, 0.0417, 0.0333, 0.0209, 0.0290, 0.0185, 0.0202], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0215, 0.0209, 0.0207, 0.0213, 0.0212, 0.0215, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:22:11,530 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6407, 2.4772, 1.9624, 2.3123, 2.8476, 2.6548, 2.9357, 2.9217], device='cuda:7'), covar=tensor([0.0177, 0.0298, 0.0433, 0.0335, 0.0183, 0.0260, 0.0204, 0.0212], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0215, 0.0209, 0.0207, 0.0214, 0.0212, 0.0215, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:22:13,943 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.554e+02 2.369e+02 2.759e+02 3.305e+02 7.145e+02, threshold=5.518e+02, percent-clipped=2.0 2023-04-29 17:22:20,637 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122178.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 17:22:44,386 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122195.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:22:53,146 INFO [train.py:904] (7/8) Epoch 13, batch 400, loss[loss=0.2009, simple_loss=0.2735, pruned_loss=0.06414, over 16714.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2659, pruned_loss=0.05165, over 2874733.38 frames. ], batch size: 89, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:23:43,184 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5999, 4.6367, 4.8575, 4.7599, 4.7281, 5.3322, 4.9546, 4.6133], device='cuda:7'), covar=tensor([0.1272, 0.1876, 0.2148, 0.2221, 0.2911, 0.1150, 0.1396, 0.2510], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0505, 0.0558, 0.0438, 0.0583, 0.0581, 0.0438, 0.0581], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 17:24:03,781 INFO [train.py:904] (7/8) Epoch 13, batch 450, loss[loss=0.1834, simple_loss=0.2558, pruned_loss=0.05556, over 16512.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2638, pruned_loss=0.05084, over 2970463.12 frames. ], batch size: 146, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:24:10,458 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122256.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:24:32,947 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.134e+02 2.674e+02 3.325e+02 9.122e+02, threshold=5.349e+02, percent-clipped=2.0 2023-04-29 17:24:45,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1645, 5.0947, 4.9712, 4.4914, 4.5519, 4.9867, 4.9977, 4.5957], device='cuda:7'), covar=tensor([0.0577, 0.0489, 0.0268, 0.0308, 0.1022, 0.0443, 0.0346, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0251, 0.0337, 0.0299, 0.0278, 0.0318, 0.0318, 0.0204, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:25:13,894 INFO [train.py:904] (7/8) Epoch 13, batch 500, loss[loss=0.1872, simple_loss=0.2589, pruned_loss=0.05781, over 16330.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2617, pruned_loss=0.04949, over 3052706.37 frames. ], batch size: 165, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:25:19,572 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122306.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:25:27,362 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6701, 2.1444, 2.2312, 4.5281, 2.2025, 2.7244, 2.3254, 2.3692], device='cuda:7'), covar=tensor([0.0916, 0.3471, 0.2525, 0.0371, 0.3908, 0.2274, 0.3008, 0.3323], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0397, 0.0336, 0.0321, 0.0414, 0.0454, 0.0363, 0.0463], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:25:36,874 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122319.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:26:21,448 INFO [train.py:904] (7/8) Epoch 13, batch 550, loss[loss=0.145, simple_loss=0.2295, pruned_loss=0.03024, over 16764.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2612, pruned_loss=0.04901, over 3101091.79 frames. ], batch size: 39, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:26:50,250 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.266e+02 2.671e+02 3.065e+02 5.763e+02, threshold=5.342e+02, percent-clipped=1.0 2023-04-29 17:27:00,813 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122380.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:27:04,242 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 17:27:30,043 INFO [train.py:904] (7/8) Epoch 13, batch 600, loss[loss=0.1795, simple_loss=0.2552, pruned_loss=0.05189, over 16411.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2614, pruned_loss=0.04907, over 3152987.16 frames. ], batch size: 146, lr: 5.35e-03, grad_scale: 4.0 2023-04-29 17:28:37,809 INFO [train.py:904] (7/8) Epoch 13, batch 650, loss[loss=0.1744, simple_loss=0.2507, pruned_loss=0.04902, over 16239.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2607, pruned_loss=0.04913, over 3188446.28 frames. ], batch size: 165, lr: 5.34e-03, grad_scale: 4.0 2023-04-29 17:29:01,205 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122468.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:29:07,724 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.494e+02 2.897e+02 3.411e+02 6.772e+02, threshold=5.794e+02, percent-clipped=6.0 2023-04-29 17:29:08,029 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122473.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 17:29:47,562 INFO [train.py:904] (7/8) Epoch 13, batch 700, loss[loss=0.1825, simple_loss=0.2582, pruned_loss=0.05343, over 16784.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.26, pruned_loss=0.04872, over 3216153.42 frames. ], batch size: 83, lr: 5.34e-03, grad_scale: 4.0 2023-04-29 17:30:26,039 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122529.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:30:56,880 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122551.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:30:57,601 INFO [train.py:904] (7/8) Epoch 13, batch 750, loss[loss=0.2062, simple_loss=0.274, pruned_loss=0.06916, over 16925.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2602, pruned_loss=0.04807, over 3239744.78 frames. ], batch size: 109, lr: 5.34e-03, grad_scale: 4.0 2023-04-29 17:31:00,531 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8951, 4.9589, 5.4112, 5.4237, 5.4275, 5.0729, 5.0198, 4.7820], device='cuda:7'), covar=tensor([0.0360, 0.0548, 0.0456, 0.0443, 0.0421, 0.0367, 0.0914, 0.0469], device='cuda:7'), in_proj_covar=tensor([0.0354, 0.0374, 0.0371, 0.0353, 0.0420, 0.0396, 0.0493, 0.0320], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 17:31:27,840 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.324e+02 2.801e+02 3.357e+02 5.795e+02, threshold=5.603e+02, percent-clipped=1.0 2023-04-29 17:32:03,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7978, 4.6998, 4.6663, 4.3684, 4.3043, 4.7177, 4.5381, 4.4393], device='cuda:7'), covar=tensor([0.0568, 0.0598, 0.0305, 0.0277, 0.0889, 0.0404, 0.0388, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0257, 0.0345, 0.0307, 0.0287, 0.0327, 0.0328, 0.0208, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:32:09,560 INFO [train.py:904] (7/8) Epoch 13, batch 800, loss[loss=0.1942, simple_loss=0.2671, pruned_loss=0.06066, over 16882.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2607, pruned_loss=0.04802, over 3261816.35 frames. ], batch size: 109, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:32:15,177 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122606.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:33:15,746 INFO [train.py:904] (7/8) Epoch 13, batch 850, loss[loss=0.175, simple_loss=0.2501, pruned_loss=0.04995, over 16874.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.26, pruned_loss=0.04717, over 3278850.57 frames. ], batch size: 96, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:33:18,970 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122654.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:33:23,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3641, 3.4927, 2.0822, 3.6618, 2.6719, 3.6023, 2.1720, 2.7024], device='cuda:7'), covar=tensor([0.0253, 0.0346, 0.1442, 0.0286, 0.0732, 0.0769, 0.1306, 0.0661], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0164, 0.0188, 0.0136, 0.0168, 0.0206, 0.0196, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 17:33:44,227 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.256e+02 2.755e+02 3.448e+02 5.006e+02, threshold=5.510e+02, percent-clipped=0.0 2023-04-29 17:33:46,992 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122675.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:34:23,501 INFO [train.py:904] (7/8) Epoch 13, batch 900, loss[loss=0.1676, simple_loss=0.2485, pruned_loss=0.04337, over 16716.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2596, pruned_loss=0.04684, over 3286791.94 frames. ], batch size: 89, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:34:35,559 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0877, 5.0954, 4.8316, 4.3049, 4.8964, 1.9494, 4.6805, 4.7766], device='cuda:7'), covar=tensor([0.0076, 0.0063, 0.0156, 0.0340, 0.0091, 0.2332, 0.0115, 0.0158], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0125, 0.0170, 0.0158, 0.0143, 0.0187, 0.0158, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:35:33,171 INFO [train.py:904] (7/8) Epoch 13, batch 950, loss[loss=0.1888, simple_loss=0.2614, pruned_loss=0.05811, over 16782.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2601, pruned_loss=0.04744, over 3297201.31 frames. ], batch size: 83, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:36:02,671 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.301e+02 2.627e+02 3.023e+02 6.278e+02, threshold=5.253e+02, percent-clipped=3.0 2023-04-29 17:36:03,717 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122773.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 17:36:40,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9009, 3.9863, 4.2797, 4.2782, 4.2889, 4.0049, 4.0622, 4.0149], device='cuda:7'), covar=tensor([0.0385, 0.0644, 0.0398, 0.0392, 0.0457, 0.0454, 0.0738, 0.0493], device='cuda:7'), in_proj_covar=tensor([0.0355, 0.0374, 0.0369, 0.0354, 0.0421, 0.0398, 0.0494, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 17:36:43,324 INFO [train.py:904] (7/8) Epoch 13, batch 1000, loss[loss=0.1814, simple_loss=0.277, pruned_loss=0.04288, over 17105.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2587, pruned_loss=0.04686, over 3300261.05 frames. ], batch size: 49, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:37:10,558 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122821.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:37:14,657 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=122824.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:37:46,897 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 17:37:52,058 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122851.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:37:52,816 INFO [train.py:904] (7/8) Epoch 13, batch 1050, loss[loss=0.2002, simple_loss=0.2666, pruned_loss=0.06691, over 16804.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2591, pruned_loss=0.04734, over 3307113.33 frames. ], batch size: 102, lr: 5.34e-03, grad_scale: 8.0 2023-04-29 17:38:20,514 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=122872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:38:21,177 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.183e+02 2.662e+02 3.059e+02 5.432e+02, threshold=5.323e+02, percent-clipped=2.0 2023-04-29 17:38:32,660 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9964, 4.2641, 4.3689, 3.3643, 3.7333, 4.2789, 4.0083, 2.4793], device='cuda:7'), covar=tensor([0.0364, 0.0057, 0.0039, 0.0244, 0.0094, 0.0075, 0.0058, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0073, 0.0072, 0.0129, 0.0084, 0.0092, 0.0082, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 17:38:58,265 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=122899.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:39:01,465 INFO [train.py:904] (7/8) Epoch 13, batch 1100, loss[loss=0.1783, simple_loss=0.249, pruned_loss=0.05384, over 16913.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2588, pruned_loss=0.04714, over 3309772.40 frames. ], batch size: 90, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:39:45,854 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=122933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:40:11,777 INFO [train.py:904] (7/8) Epoch 13, batch 1150, loss[loss=0.1976, simple_loss=0.2667, pruned_loss=0.06426, over 16825.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2581, pruned_loss=0.0467, over 3317355.15 frames. ], batch size: 102, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:40:39,515 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.197e+02 2.560e+02 3.227e+02 9.975e+02, threshold=5.120e+02, percent-clipped=3.0 2023-04-29 17:40:43,473 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=122975.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:41:20,390 INFO [train.py:904] (7/8) Epoch 13, batch 1200, loss[loss=0.1932, simple_loss=0.2525, pruned_loss=0.06697, over 16826.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2572, pruned_loss=0.04623, over 3319125.20 frames. ], batch size: 109, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:41:50,192 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=123023.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:42:28,691 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 17:42:30,142 INFO [train.py:904] (7/8) Epoch 13, batch 1250, loss[loss=0.1605, simple_loss=0.2483, pruned_loss=0.03638, over 17194.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2569, pruned_loss=0.04633, over 3321276.61 frames. ], batch size: 44, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:42:59,816 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.819e+02 2.369e+02 2.845e+02 3.353e+02 6.021e+02, threshold=5.690e+02, percent-clipped=1.0 2023-04-29 17:43:40,472 INFO [train.py:904] (7/8) Epoch 13, batch 1300, loss[loss=0.1337, simple_loss=0.2194, pruned_loss=0.02398, over 16817.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2571, pruned_loss=0.04682, over 3317099.84 frames. ], batch size: 39, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:44:02,948 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 17:44:12,448 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0610, 4.7012, 4.8985, 5.2803, 5.4180, 4.7281, 5.4411, 5.3979], device='cuda:7'), covar=tensor([0.1449, 0.1365, 0.2139, 0.0856, 0.0777, 0.0797, 0.0726, 0.0813], device='cuda:7'), in_proj_covar=tensor([0.0565, 0.0711, 0.0858, 0.0716, 0.0541, 0.0561, 0.0570, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:44:12,460 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123124.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:44:26,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1569, 4.1409, 4.0928, 3.4228, 4.1121, 1.6497, 3.8830, 3.6096], device='cuda:7'), covar=tensor([0.0120, 0.0111, 0.0169, 0.0344, 0.0091, 0.2783, 0.0139, 0.0239], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0128, 0.0174, 0.0162, 0.0146, 0.0189, 0.0163, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:44:49,696 INFO [train.py:904] (7/8) Epoch 13, batch 1350, loss[loss=0.1681, simple_loss=0.2632, pruned_loss=0.0365, over 17117.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2574, pruned_loss=0.04642, over 3317607.25 frames. ], batch size: 47, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:45:17,070 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=123172.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:45:17,979 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.284e+02 2.773e+02 3.220e+02 7.351e+02, threshold=5.547e+02, percent-clipped=2.0 2023-04-29 17:45:58,577 INFO [train.py:904] (7/8) Epoch 13, batch 1400, loss[loss=0.1934, simple_loss=0.2707, pruned_loss=0.05809, over 16804.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2575, pruned_loss=0.04635, over 3321407.92 frames. ], batch size: 102, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:46:26,143 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 17:46:35,868 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123228.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:47:09,282 INFO [train.py:904] (7/8) Epoch 13, batch 1450, loss[loss=0.1918, simple_loss=0.2745, pruned_loss=0.05461, over 16566.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2563, pruned_loss=0.046, over 3313909.49 frames. ], batch size: 62, lr: 5.33e-03, grad_scale: 8.0 2023-04-29 17:47:38,912 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.758e+02 2.301e+02 2.596e+02 3.249e+02 6.793e+02, threshold=5.192e+02, percent-clipped=2.0 2023-04-29 17:47:59,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3594, 3.3360, 3.3876, 3.4916, 3.5357, 3.2875, 3.4856, 3.5855], device='cuda:7'), covar=tensor([0.1058, 0.0841, 0.1118, 0.0603, 0.0577, 0.1998, 0.1108, 0.0693], device='cuda:7'), in_proj_covar=tensor([0.0571, 0.0716, 0.0865, 0.0723, 0.0546, 0.0567, 0.0578, 0.0666], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:48:02,606 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8038, 5.1395, 4.8918, 4.9113, 4.6395, 4.6131, 4.6457, 5.2139], device='cuda:7'), covar=tensor([0.1084, 0.0875, 0.0903, 0.0698, 0.0716, 0.0937, 0.0971, 0.0870], device='cuda:7'), in_proj_covar=tensor([0.0576, 0.0731, 0.0589, 0.0517, 0.0461, 0.0471, 0.0608, 0.0560], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:48:19,763 INFO [train.py:904] (7/8) Epoch 13, batch 1500, loss[loss=0.2113, simple_loss=0.2731, pruned_loss=0.07474, over 16534.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.256, pruned_loss=0.04603, over 3307167.48 frames. ], batch size: 146, lr: 5.33e-03, grad_scale: 4.0 2023-04-29 17:48:37,131 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123314.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:48:37,223 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5700, 2.7015, 2.2303, 2.3522, 2.9285, 2.6652, 3.3659, 3.1647], device='cuda:7'), covar=tensor([0.0088, 0.0274, 0.0380, 0.0338, 0.0195, 0.0293, 0.0165, 0.0183], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0218, 0.0209, 0.0209, 0.0218, 0.0216, 0.0223, 0.0206], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:49:30,723 INFO [train.py:904] (7/8) Epoch 13, batch 1550, loss[loss=0.16, simple_loss=0.2439, pruned_loss=0.03803, over 17204.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2575, pruned_loss=0.04734, over 3311235.67 frames. ], batch size: 45, lr: 5.32e-03, grad_scale: 4.0 2023-04-29 17:49:42,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5095, 3.4550, 3.7768, 1.8073, 3.9118, 3.8689, 3.0837, 2.9228], device='cuda:7'), covar=tensor([0.0729, 0.0207, 0.0165, 0.1163, 0.0063, 0.0144, 0.0363, 0.0405], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0102, 0.0091, 0.0139, 0.0070, 0.0111, 0.0122, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 17:50:00,255 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 2.426e+02 2.890e+02 3.629e+02 7.930e+02, threshold=5.780e+02, percent-clipped=5.0 2023-04-29 17:50:01,904 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123375.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:50:31,351 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9548, 2.0498, 2.5582, 2.9703, 2.7865, 3.3534, 2.2588, 3.4137], device='cuda:7'), covar=tensor([0.0198, 0.0363, 0.0256, 0.0250, 0.0252, 0.0163, 0.0338, 0.0109], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0176, 0.0161, 0.0166, 0.0175, 0.0129, 0.0176, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 17:50:39,394 INFO [train.py:904] (7/8) Epoch 13, batch 1600, loss[loss=0.1768, simple_loss=0.2685, pruned_loss=0.04252, over 16026.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2592, pruned_loss=0.04786, over 3313028.82 frames. ], batch size: 35, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:51:05,945 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5418, 3.8423, 4.0641, 2.2409, 3.3137, 2.6627, 4.0839, 3.9724], device='cuda:7'), covar=tensor([0.0256, 0.0782, 0.0441, 0.1755, 0.0709, 0.0860, 0.0566, 0.1114], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0148, 0.0158, 0.0144, 0.0137, 0.0124, 0.0136, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 17:51:08,919 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7795, 4.0427, 3.1921, 2.3128, 2.8764, 2.5115, 4.3655, 3.7753], device='cuda:7'), covar=tensor([0.2526, 0.0636, 0.1445, 0.2405, 0.2346, 0.1769, 0.0417, 0.1059], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0260, 0.0287, 0.0282, 0.0279, 0.0227, 0.0272, 0.0302], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:51:12,077 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6284, 6.0183, 5.7019, 5.7895, 5.3774, 5.3051, 5.4053, 6.1134], device='cuda:7'), covar=tensor([0.1187, 0.0840, 0.1088, 0.0646, 0.0800, 0.0639, 0.1061, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0587, 0.0742, 0.0600, 0.0526, 0.0469, 0.0478, 0.0617, 0.0566], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:51:15,113 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123428.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:51:47,311 INFO [train.py:904] (7/8) Epoch 13, batch 1650, loss[loss=0.1966, simple_loss=0.2783, pruned_loss=0.05742, over 16239.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2606, pruned_loss=0.04811, over 3326586.43 frames. ], batch size: 165, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:51:57,525 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 17:52:18,029 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.454e+02 2.938e+02 3.518e+02 6.758e+02, threshold=5.875e+02, percent-clipped=2.0 2023-04-29 17:52:38,878 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123489.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:52:55,812 INFO [train.py:904] (7/8) Epoch 13, batch 1700, loss[loss=0.1751, simple_loss=0.255, pruned_loss=0.04755, over 16289.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2626, pruned_loss=0.04852, over 3319081.85 frames. ], batch size: 165, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:53:27,699 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1108, 4.4198, 4.6445, 4.6475, 4.6828, 4.3865, 4.0210, 4.1758], device='cuda:7'), covar=tensor([0.0601, 0.0749, 0.0626, 0.0702, 0.0718, 0.0603, 0.1488, 0.0677], device='cuda:7'), in_proj_covar=tensor([0.0360, 0.0381, 0.0379, 0.0359, 0.0427, 0.0404, 0.0501, 0.0324], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 17:53:31,950 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123528.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:53:44,817 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2023-04-29 17:53:50,698 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 17:54:04,801 INFO [train.py:904] (7/8) Epoch 13, batch 1750, loss[loss=0.1697, simple_loss=0.2629, pruned_loss=0.03828, over 17019.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2639, pruned_loss=0.04868, over 3313988.85 frames. ], batch size: 50, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:54:34,134 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.358e+02 2.760e+02 3.262e+02 5.842e+02, threshold=5.520e+02, percent-clipped=0.0 2023-04-29 17:54:38,679 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=123576.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:55:14,406 INFO [train.py:904] (7/8) Epoch 13, batch 1800, loss[loss=0.1805, simple_loss=0.2586, pruned_loss=0.05121, over 16837.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2651, pruned_loss=0.04854, over 3310234.55 frames. ], batch size: 96, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:56:23,373 INFO [train.py:904] (7/8) Epoch 13, batch 1850, loss[loss=0.2006, simple_loss=0.2922, pruned_loss=0.05447, over 17081.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2656, pruned_loss=0.04874, over 3317731.50 frames. ], batch size: 53, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:56:44,909 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3692, 5.2047, 5.1116, 4.6851, 4.7110, 5.1376, 5.1742, 4.7666], device='cuda:7'), covar=tensor([0.0542, 0.0533, 0.0315, 0.0304, 0.1099, 0.0484, 0.0315, 0.0803], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0361, 0.0320, 0.0300, 0.0342, 0.0343, 0.0216, 0.0375], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 17:56:47,458 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123670.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:56:52,514 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.459e+02 2.923e+02 3.563e+02 7.345e+02, threshold=5.846e+02, percent-clipped=2.0 2023-04-29 17:57:33,314 INFO [train.py:904] (7/8) Epoch 13, batch 1900, loss[loss=0.1811, simple_loss=0.2692, pruned_loss=0.04648, over 17059.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2657, pruned_loss=0.048, over 3316744.63 frames. ], batch size: 53, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:58:39,339 INFO [train.py:904] (7/8) Epoch 13, batch 1950, loss[loss=0.1903, simple_loss=0.277, pruned_loss=0.05176, over 16153.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2654, pruned_loss=0.04755, over 3314026.70 frames. ], batch size: 165, lr: 5.32e-03, grad_scale: 8.0 2023-04-29 17:58:55,804 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 17:59:09,943 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.247e+02 2.704e+02 3.281e+02 7.395e+02, threshold=5.408e+02, percent-clipped=2.0 2023-04-29 17:59:24,290 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6768, 2.3024, 2.3869, 4.4746, 2.2384, 2.7494, 2.4233, 2.4897], device='cuda:7'), covar=tensor([0.0935, 0.3465, 0.2476, 0.0361, 0.3797, 0.2420, 0.2995, 0.3493], device='cuda:7'), in_proj_covar=tensor([0.0370, 0.0402, 0.0339, 0.0324, 0.0416, 0.0464, 0.0367, 0.0471], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:59:26,067 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=123784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 17:59:28,647 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7512, 2.6713, 2.5088, 4.0294, 3.1468, 4.0506, 1.4637, 2.7903], device='cuda:7'), covar=tensor([0.1361, 0.0655, 0.1088, 0.0171, 0.0189, 0.0342, 0.1555, 0.0822], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0161, 0.0183, 0.0155, 0.0196, 0.0211, 0.0183, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 17:59:46,510 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6152, 6.0204, 5.7538, 5.8257, 5.3375, 5.2615, 5.3710, 6.1022], device='cuda:7'), covar=tensor([0.1115, 0.0776, 0.0987, 0.0683, 0.0837, 0.0654, 0.0994, 0.0801], device='cuda:7'), in_proj_covar=tensor([0.0577, 0.0727, 0.0587, 0.0517, 0.0461, 0.0468, 0.0605, 0.0556], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 17:59:48,925 INFO [train.py:904] (7/8) Epoch 13, batch 2000, loss[loss=0.2499, simple_loss=0.3296, pruned_loss=0.08508, over 12109.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2653, pruned_loss=0.04768, over 3299225.36 frames. ], batch size: 247, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:00:59,631 INFO [train.py:904] (7/8) Epoch 13, batch 2050, loss[loss=0.2113, simple_loss=0.296, pruned_loss=0.06332, over 15347.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2642, pruned_loss=0.04755, over 3303942.43 frames. ], batch size: 190, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:01:28,870 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=123872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:01:30,782 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.367e+02 2.767e+02 3.495e+02 6.657e+02, threshold=5.534e+02, percent-clipped=3.0 2023-04-29 18:02:09,988 INFO [train.py:904] (7/8) Epoch 13, batch 2100, loss[loss=0.2026, simple_loss=0.2876, pruned_loss=0.0588, over 17110.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2654, pruned_loss=0.04802, over 3312254.17 frames. ], batch size: 55, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:02:23,998 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-04-29 18:02:54,427 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=123933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:03:20,335 INFO [train.py:904] (7/8) Epoch 13, batch 2150, loss[loss=0.1714, simple_loss=0.2592, pruned_loss=0.04174, over 15859.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2667, pruned_loss=0.04874, over 3312200.68 frames. ], batch size: 35, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:03:43,125 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2023-04-29 18:03:46,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3206, 4.1429, 4.3708, 4.5208, 4.6049, 4.1248, 4.3862, 4.6041], device='cuda:7'), covar=tensor([0.1344, 0.1021, 0.1212, 0.0577, 0.0523, 0.1256, 0.1686, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0572, 0.0717, 0.0863, 0.0724, 0.0543, 0.0568, 0.0577, 0.0662], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:03:46,110 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=123970.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:03:50,027 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.337e+02 2.985e+02 3.427e+02 6.976e+02, threshold=5.971e+02, percent-clipped=3.0 2023-04-29 18:04:01,208 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0149, 2.0026, 2.1483, 3.4920, 2.0432, 2.2795, 2.1292, 2.1511], device='cuda:7'), covar=tensor([0.1135, 0.3254, 0.2405, 0.0614, 0.3651, 0.2403, 0.3113, 0.3185], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0404, 0.0339, 0.0326, 0.0417, 0.0466, 0.0369, 0.0473], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:04:03,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7089, 4.8092, 4.9955, 4.7676, 4.7892, 5.4552, 4.9777, 4.6824], device='cuda:7'), covar=tensor([0.1197, 0.2023, 0.1871, 0.2186, 0.2902, 0.1037, 0.1500, 0.2400], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0537, 0.0583, 0.0465, 0.0624, 0.0610, 0.0463, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:04:12,782 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7884, 4.7591, 5.2645, 5.2467, 5.2509, 4.9073, 4.8493, 4.6420], device='cuda:7'), covar=tensor([0.0271, 0.0556, 0.0334, 0.0363, 0.0437, 0.0314, 0.0813, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0360, 0.0382, 0.0379, 0.0358, 0.0428, 0.0403, 0.0504, 0.0324], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 18:04:30,743 INFO [train.py:904] (7/8) Epoch 13, batch 2200, loss[loss=0.178, simple_loss=0.274, pruned_loss=0.04103, over 17018.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2678, pruned_loss=0.05009, over 3306589.37 frames. ], batch size: 50, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:04:53,417 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:05:00,679 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.05 vs. limit=5.0 2023-04-29 18:05:19,561 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:05:32,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2041, 3.8090, 3.8872, 2.1524, 3.1372, 2.4169, 3.7121, 3.8887], device='cuda:7'), covar=tensor([0.0317, 0.0781, 0.0440, 0.1871, 0.0748, 0.0927, 0.0684, 0.0957], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0151, 0.0161, 0.0147, 0.0139, 0.0127, 0.0139, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:05:40,223 INFO [train.py:904] (7/8) Epoch 13, batch 2250, loss[loss=0.1703, simple_loss=0.2645, pruned_loss=0.03804, over 17045.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2685, pruned_loss=0.05039, over 3298379.24 frames. ], batch size: 55, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:05:53,623 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124062.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:06:09,555 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.445e+02 2.929e+02 3.478e+02 7.502e+02, threshold=5.857e+02, percent-clipped=2.0 2023-04-29 18:06:22,794 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124084.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:06:42,917 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124097.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:06:46,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3098, 5.2210, 5.1331, 4.6685, 4.6522, 5.1709, 5.2167, 4.7212], device='cuda:7'), covar=tensor([0.0625, 0.0525, 0.0284, 0.0358, 0.1226, 0.0404, 0.0250, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0270, 0.0365, 0.0322, 0.0301, 0.0344, 0.0345, 0.0218, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:06:48,019 INFO [train.py:904] (7/8) Epoch 13, batch 2300, loss[loss=0.1915, simple_loss=0.2701, pruned_loss=0.0564, over 16761.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2691, pruned_loss=0.05059, over 3305249.03 frames. ], batch size: 124, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:07:17,816 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124123.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:07:29,438 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124132.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:07:56,751 INFO [train.py:904] (7/8) Epoch 13, batch 2350, loss[loss=0.1598, simple_loss=0.2453, pruned_loss=0.03712, over 17217.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2687, pruned_loss=0.05072, over 3309831.22 frames. ], batch size: 44, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:08:14,606 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5274, 3.8089, 4.2122, 1.8374, 4.3661, 4.4801, 3.2462, 3.2717], device='cuda:7'), covar=tensor([0.1034, 0.0187, 0.0207, 0.1275, 0.0094, 0.0157, 0.0381, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0102, 0.0090, 0.0138, 0.0070, 0.0111, 0.0122, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:08:26,418 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.342e+02 2.726e+02 3.219e+02 8.801e+02, threshold=5.452e+02, percent-clipped=1.0 2023-04-29 18:08:29,934 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3277, 3.9336, 4.0152, 2.1206, 3.1555, 2.5650, 4.0479, 4.0441], device='cuda:7'), covar=tensor([0.0272, 0.0705, 0.0447, 0.1820, 0.0735, 0.0881, 0.0560, 0.0904], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0151, 0.0161, 0.0146, 0.0138, 0.0126, 0.0138, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:09:06,169 INFO [train.py:904] (7/8) Epoch 13, batch 2400, loss[loss=0.1864, simple_loss=0.2633, pruned_loss=0.05476, over 16710.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2693, pruned_loss=0.05047, over 3311733.23 frames. ], batch size: 83, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:09:42,821 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124228.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:10:15,692 INFO [train.py:904] (7/8) Epoch 13, batch 2450, loss[loss=0.1552, simple_loss=0.2361, pruned_loss=0.03711, over 16793.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2711, pruned_loss=0.05107, over 3307073.77 frames. ], batch size: 39, lr: 5.31e-03, grad_scale: 8.0 2023-04-29 18:10:43,362 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124272.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:10:46,037 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.400e+02 2.819e+02 3.354e+02 7.582e+02, threshold=5.638e+02, percent-clipped=2.0 2023-04-29 18:11:24,244 INFO [train.py:904] (7/8) Epoch 13, batch 2500, loss[loss=0.153, simple_loss=0.2451, pruned_loss=0.03044, over 17196.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2691, pruned_loss=0.04984, over 3316906.47 frames. ], batch size: 46, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:11:44,850 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8595, 2.9259, 2.4950, 4.4736, 3.3019, 4.1130, 1.7941, 2.9037], device='cuda:7'), covar=tensor([0.1410, 0.0770, 0.1279, 0.0191, 0.0277, 0.0474, 0.1575, 0.0908], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0162, 0.0184, 0.0157, 0.0198, 0.0211, 0.0184, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:12:09,477 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124333.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:12:31,555 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124348.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 18:12:37,095 INFO [train.py:904] (7/8) Epoch 13, batch 2550, loss[loss=0.1924, simple_loss=0.2829, pruned_loss=0.05096, over 16630.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2686, pruned_loss=0.04966, over 3313028.62 frames. ], batch size: 62, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:12:41,081 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0647, 4.6519, 4.5712, 3.3850, 3.8102, 4.4636, 4.0233, 2.7763], device='cuda:7'), covar=tensor([0.0374, 0.0039, 0.0028, 0.0261, 0.0094, 0.0074, 0.0062, 0.0339], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0071, 0.0072, 0.0127, 0.0084, 0.0093, 0.0083, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:13:06,417 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.275e+02 2.738e+02 3.268e+02 6.282e+02, threshold=5.476e+02, percent-clipped=1.0 2023-04-29 18:13:31,395 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124392.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:13:44,304 INFO [train.py:904] (7/8) Epoch 13, batch 2600, loss[loss=0.18, simple_loss=0.2768, pruned_loss=0.04164, over 17223.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2674, pruned_loss=0.04862, over 3326947.39 frames. ], batch size: 52, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:13:54,903 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124409.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 18:14:06,945 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124418.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:14:15,387 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 18:14:26,645 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:14:54,857 INFO [train.py:904] (7/8) Epoch 13, batch 2650, loss[loss=0.2006, simple_loss=0.2936, pruned_loss=0.05384, over 16977.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2683, pruned_loss=0.04885, over 3325652.16 frames. ], batch size: 58, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:15:05,680 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9724, 2.7412, 2.7953, 2.1140, 2.6118, 2.1461, 2.7601, 2.9222], device='cuda:7'), covar=tensor([0.0290, 0.0726, 0.0489, 0.1585, 0.0740, 0.0846, 0.0591, 0.0676], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0152, 0.0162, 0.0147, 0.0138, 0.0127, 0.0139, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:15:25,988 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.199e+02 2.542e+02 3.057e+02 5.216e+02, threshold=5.084e+02, percent-clipped=0.0 2023-04-29 18:15:52,832 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124493.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:16:04,566 INFO [train.py:904] (7/8) Epoch 13, batch 2700, loss[loss=0.1904, simple_loss=0.2813, pruned_loss=0.04969, over 16577.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2694, pruned_loss=0.04878, over 3312096.99 frames. ], batch size: 62, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:16:05,398 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-29 18:16:26,461 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7318, 3.7249, 4.2333, 2.0201, 4.3363, 4.3558, 3.1278, 3.2405], device='cuda:7'), covar=tensor([0.0763, 0.0228, 0.0173, 0.1095, 0.0076, 0.0180, 0.0397, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0102, 0.0090, 0.0139, 0.0071, 0.0112, 0.0122, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:16:40,325 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124528.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:16:47,295 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1801, 5.1110, 4.9075, 4.4375, 4.9983, 1.9681, 4.7985, 4.8574], device='cuda:7'), covar=tensor([0.0063, 0.0061, 0.0156, 0.0308, 0.0074, 0.2372, 0.0111, 0.0152], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0132, 0.0179, 0.0168, 0.0151, 0.0191, 0.0168, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:17:13,331 INFO [train.py:904] (7/8) Epoch 13, batch 2750, loss[loss=0.187, simple_loss=0.2608, pruned_loss=0.05663, over 16346.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2694, pruned_loss=0.04849, over 3321305.19 frames. ], batch size: 146, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:17:44,014 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.177e+02 2.567e+02 3.156e+02 4.821e+02, threshold=5.134e+02, percent-clipped=0.0 2023-04-29 18:17:46,656 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124576.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:17:52,199 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4755, 2.2259, 2.3107, 4.2516, 2.1333, 2.6659, 2.3861, 2.4670], device='cuda:7'), covar=tensor([0.1172, 0.3465, 0.2401, 0.0466, 0.3906, 0.2348, 0.3141, 0.3062], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0406, 0.0339, 0.0325, 0.0417, 0.0468, 0.0369, 0.0476], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:18:22,961 INFO [train.py:904] (7/8) Epoch 13, batch 2800, loss[loss=0.1851, simple_loss=0.2612, pruned_loss=0.05446, over 16905.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2684, pruned_loss=0.04838, over 3329221.56 frames. ], batch size: 109, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:18:23,493 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8362, 3.1559, 2.7071, 5.0046, 4.0278, 4.6453, 1.8413, 3.2303], device='cuda:7'), covar=tensor([0.1371, 0.0712, 0.1196, 0.0143, 0.0267, 0.0342, 0.1484, 0.0750], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0163, 0.0185, 0.0159, 0.0200, 0.0212, 0.0186, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:18:59,159 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124628.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:19:12,871 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124638.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:19:16,361 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0546, 1.9356, 2.5081, 2.8618, 2.9255, 2.9569, 1.8219, 3.1482], device='cuda:7'), covar=tensor([0.0117, 0.0356, 0.0220, 0.0194, 0.0184, 0.0173, 0.0411, 0.0107], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0165, 0.0171, 0.0179, 0.0133, 0.0179, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:19:31,059 INFO [train.py:904] (7/8) Epoch 13, batch 2850, loss[loss=0.1669, simple_loss=0.263, pruned_loss=0.03536, over 17270.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2677, pruned_loss=0.0479, over 3332778.51 frames. ], batch size: 52, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:19:46,068 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124663.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:20:00,143 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.347e+02 2.816e+02 3.672e+02 8.891e+02, threshold=5.632e+02, percent-clipped=4.0 2023-04-29 18:20:05,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9000, 4.2824, 3.2121, 2.3158, 2.9269, 2.4498, 4.7046, 3.8398], device='cuda:7'), covar=tensor([0.2418, 0.0623, 0.1421, 0.2477, 0.2513, 0.1882, 0.0315, 0.1066], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0261, 0.0287, 0.0285, 0.0283, 0.0229, 0.0272, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:20:24,606 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124692.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:20:34,571 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124699.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:20:37,528 INFO [train.py:904] (7/8) Epoch 13, batch 2900, loss[loss=0.1821, simple_loss=0.2682, pruned_loss=0.04797, over 16713.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2663, pruned_loss=0.04823, over 3335359.94 frames. ], batch size: 57, lr: 5.30e-03, grad_scale: 8.0 2023-04-29 18:20:40,126 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124704.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 18:20:59,010 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124718.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:21:07,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124724.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:21:29,889 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:21:46,241 INFO [train.py:904] (7/8) Epoch 13, batch 2950, loss[loss=0.1764, simple_loss=0.2666, pruned_loss=0.04311, over 17218.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2661, pruned_loss=0.04873, over 3333428.87 frames. ], batch size: 43, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:22:05,843 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124766.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:22:16,915 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.794e+02 2.554e+02 2.996e+02 3.501e+02 7.058e+02, threshold=5.993e+02, percent-clipped=3.0 2023-04-29 18:22:36,241 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124788.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:22:55,201 INFO [train.py:904] (7/8) Epoch 13, batch 3000, loss[loss=0.1701, simple_loss=0.2604, pruned_loss=0.03988, over 17230.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2665, pruned_loss=0.0496, over 3330029.46 frames. ], batch size: 45, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:22:55,202 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 18:23:04,002 INFO [train.py:938] (7/8) Epoch 13, validation: loss=0.1391, simple_loss=0.2452, pruned_loss=0.01648, over 944034.00 frames. 2023-04-29 18:23:04,003 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 18:23:29,549 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=124820.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:24:07,926 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8208, 5.1234, 5.3300, 5.1717, 5.1886, 5.7895, 5.2880, 4.9591], device='cuda:7'), covar=tensor([0.1179, 0.1905, 0.2135, 0.1868, 0.2397, 0.0882, 0.1428, 0.2315], device='cuda:7'), in_proj_covar=tensor([0.0376, 0.0532, 0.0579, 0.0461, 0.0625, 0.0604, 0.0461, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:24:14,178 INFO [train.py:904] (7/8) Epoch 13, batch 3050, loss[loss=0.1702, simple_loss=0.2527, pruned_loss=0.04386, over 15928.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2658, pruned_loss=0.04888, over 3332323.72 frames. ], batch size: 35, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:24:44,776 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.296e+02 3.000e+02 3.565e+02 8.414e+02, threshold=5.999e+02, percent-clipped=2.0 2023-04-29 18:24:54,367 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1907, 3.6511, 3.7771, 1.9847, 2.9201, 2.3060, 3.6612, 3.7095], device='cuda:7'), covar=tensor([0.0313, 0.0785, 0.0477, 0.1974, 0.0833, 0.0939, 0.0738, 0.1064], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0152, 0.0160, 0.0146, 0.0138, 0.0126, 0.0139, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:24:55,428 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=124881.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:25:21,639 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8387, 2.7604, 2.7768, 2.1044, 2.5933, 2.1681, 2.6952, 2.8978], device='cuda:7'), covar=tensor([0.0252, 0.0677, 0.0456, 0.1605, 0.0716, 0.0824, 0.0574, 0.0599], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0151, 0.0160, 0.0145, 0.0137, 0.0125, 0.0138, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:25:25,282 INFO [train.py:904] (7/8) Epoch 13, batch 3100, loss[loss=0.1822, simple_loss=0.2548, pruned_loss=0.05474, over 16459.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2655, pruned_loss=0.0488, over 3313259.19 frames. ], batch size: 75, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:26:01,310 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=124928.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:26:30,469 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9075, 4.8812, 5.3667, 5.3689, 5.3636, 5.0328, 4.9754, 4.7838], device='cuda:7'), covar=tensor([0.0288, 0.0503, 0.0364, 0.0369, 0.0398, 0.0323, 0.0864, 0.0403], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0386, 0.0384, 0.0365, 0.0433, 0.0408, 0.0510, 0.0327], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 18:26:34,832 INFO [train.py:904] (7/8) Epoch 13, batch 3150, loss[loss=0.1879, simple_loss=0.2681, pruned_loss=0.05381, over 15519.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2651, pruned_loss=0.04852, over 3319379.97 frames. ], batch size: 191, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:26:45,146 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2023-04-29 18:26:49,727 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4170, 3.3268, 3.5897, 1.8339, 3.6268, 3.6605, 2.9470, 2.8470], device='cuda:7'), covar=tensor([0.0717, 0.0194, 0.0159, 0.1144, 0.0085, 0.0162, 0.0399, 0.0388], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0101, 0.0089, 0.0138, 0.0070, 0.0111, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:27:05,836 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.205e+02 2.626e+02 3.111e+02 5.476e+02, threshold=5.252e+02, percent-clipped=0.0 2023-04-29 18:27:09,147 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=124976.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:27:34,216 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=124994.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:27:45,738 INFO [train.py:904] (7/8) Epoch 13, batch 3200, loss[loss=0.1646, simple_loss=0.2473, pruned_loss=0.04097, over 15911.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2643, pruned_loss=0.0481, over 3307360.79 frames. ], batch size: 35, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:27:49,148 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125004.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 18:28:10,527 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125019.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:28:30,797 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-04-29 18:28:56,258 INFO [train.py:904] (7/8) Epoch 13, batch 3250, loss[loss=0.1465, simple_loss=0.2342, pruned_loss=0.0294, over 16804.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2644, pruned_loss=0.04809, over 3311519.27 frames. ], batch size: 39, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:28:56,587 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125052.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 18:29:02,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5682, 3.5322, 3.9512, 1.9748, 3.9862, 4.0413, 3.1786, 3.0128], device='cuda:7'), covar=tensor([0.0732, 0.0213, 0.0133, 0.1107, 0.0071, 0.0155, 0.0342, 0.0423], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0102, 0.0090, 0.0138, 0.0071, 0.0112, 0.0122, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:29:27,130 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.287e+02 2.914e+02 3.406e+02 7.385e+02, threshold=5.827e+02, percent-clipped=6.0 2023-04-29 18:29:46,802 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:30:05,569 INFO [train.py:904] (7/8) Epoch 13, batch 3300, loss[loss=0.2337, simple_loss=0.3058, pruned_loss=0.08076, over 16274.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2654, pruned_loss=0.04794, over 3318429.38 frames. ], batch size: 165, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:30:41,253 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6650, 3.6613, 4.0973, 2.0894, 4.1392, 4.1850, 3.1457, 3.1632], device='cuda:7'), covar=tensor([0.0712, 0.0196, 0.0121, 0.1116, 0.0070, 0.0147, 0.0343, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0101, 0.0089, 0.0137, 0.0071, 0.0111, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:30:42,351 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125128.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:30:53,994 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:31:00,670 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3957, 3.3960, 2.0246, 3.5783, 2.5757, 3.5785, 2.0679, 2.7510], device='cuda:7'), covar=tensor([0.0192, 0.0388, 0.1410, 0.0254, 0.0684, 0.0656, 0.1304, 0.0613], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0169, 0.0190, 0.0145, 0.0170, 0.0215, 0.0198, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:31:08,016 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7562, 3.1178, 2.8455, 4.9035, 4.0572, 4.4742, 1.5270, 3.3520], device='cuda:7'), covar=tensor([0.1402, 0.0676, 0.1071, 0.0204, 0.0253, 0.0340, 0.1605, 0.0640], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0162, 0.0183, 0.0159, 0.0198, 0.0211, 0.0184, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 18:31:15,070 INFO [train.py:904] (7/8) Epoch 13, batch 3350, loss[loss=0.1701, simple_loss=0.2675, pruned_loss=0.03632, over 17192.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.267, pruned_loss=0.049, over 3316832.59 frames. ], batch size: 46, lr: 5.29e-03, grad_scale: 8.0 2023-04-29 18:31:45,942 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.375e+02 2.737e+02 3.421e+02 8.798e+02, threshold=5.473e+02, percent-clipped=2.0 2023-04-29 18:31:49,120 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125176.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:32:06,985 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125189.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:32:24,432 INFO [train.py:904] (7/8) Epoch 13, batch 3400, loss[loss=0.1527, simple_loss=0.237, pruned_loss=0.03417, over 16827.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2669, pruned_loss=0.04915, over 3319602.33 frames. ], batch size: 42, lr: 5.29e-03, grad_scale: 4.0 2023-04-29 18:33:12,115 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1022, 5.7635, 5.9125, 5.6768, 5.8188, 6.3174, 5.7788, 5.5034], device='cuda:7'), covar=tensor([0.0955, 0.1830, 0.1866, 0.2186, 0.2475, 0.0875, 0.1322, 0.2406], device='cuda:7'), in_proj_covar=tensor([0.0381, 0.0537, 0.0583, 0.0466, 0.0631, 0.0609, 0.0464, 0.0618], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:33:35,396 INFO [train.py:904] (7/8) Epoch 13, batch 3450, loss[loss=0.1728, simple_loss=0.2459, pruned_loss=0.04983, over 16782.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2659, pruned_loss=0.04891, over 3317183.94 frames. ], batch size: 96, lr: 5.28e-03, grad_scale: 4.0 2023-04-29 18:34:04,763 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9308, 4.3606, 4.4959, 3.3312, 3.6632, 4.3568, 3.9206, 2.4771], device='cuda:7'), covar=tensor([0.0360, 0.0039, 0.0026, 0.0223, 0.0090, 0.0070, 0.0061, 0.0364], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0074, 0.0073, 0.0128, 0.0085, 0.0094, 0.0084, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:34:07,281 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.393e+02 2.670e+02 3.290e+02 7.203e+02, threshold=5.341e+02, percent-clipped=1.0 2023-04-29 18:34:35,517 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125294.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:34:41,447 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1854, 3.9653, 4.5399, 2.1693, 4.6249, 4.7477, 3.3770, 3.8866], device='cuda:7'), covar=tensor([0.0604, 0.0219, 0.0189, 0.1082, 0.0080, 0.0105, 0.0349, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0102, 0.0089, 0.0137, 0.0070, 0.0112, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:34:46,239 INFO [train.py:904] (7/8) Epoch 13, batch 3500, loss[loss=0.221, simple_loss=0.2948, pruned_loss=0.07361, over 16275.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2652, pruned_loss=0.04845, over 3316123.77 frames. ], batch size: 164, lr: 5.28e-03, grad_scale: 4.0 2023-04-29 18:35:09,858 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125319.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:35:43,535 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125342.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:35:57,856 INFO [train.py:904] (7/8) Epoch 13, batch 3550, loss[loss=0.1775, simple_loss=0.2706, pruned_loss=0.04221, over 17300.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2631, pruned_loss=0.04741, over 3322767.23 frames. ], batch size: 52, lr: 5.28e-03, grad_scale: 4.0 2023-04-29 18:36:03,821 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:36:19,304 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125367.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:36:30,597 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.719e+02 2.251e+02 2.708e+02 3.279e+02 8.250e+02, threshold=5.415e+02, percent-clipped=4.0 2023-04-29 18:36:32,896 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125376.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:36:39,833 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9281, 4.3194, 4.4258, 2.9821, 3.6411, 4.2968, 3.8485, 2.6103], device='cuda:7'), covar=tensor([0.0372, 0.0049, 0.0032, 0.0296, 0.0091, 0.0082, 0.0066, 0.0348], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0074, 0.0073, 0.0128, 0.0085, 0.0094, 0.0083, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:37:08,977 INFO [train.py:904] (7/8) Epoch 13, batch 3600, loss[loss=0.1847, simple_loss=0.2632, pruned_loss=0.05307, over 16787.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2623, pruned_loss=0.04714, over 3313793.73 frames. ], batch size: 134, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:37:31,350 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125417.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 18:37:36,719 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125421.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:37:59,830 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125437.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:38:21,938 INFO [train.py:904] (7/8) Epoch 13, batch 3650, loss[loss=0.1621, simple_loss=0.2347, pruned_loss=0.04475, over 16793.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2611, pruned_loss=0.04741, over 3294277.31 frames. ], batch size: 83, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:38:57,387 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.272e+02 2.828e+02 3.424e+02 6.489e+02, threshold=5.656e+02, percent-clipped=1.0 2023-04-29 18:38:59,682 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125476.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:39:07,809 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125482.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:39:11,402 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125484.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:39:36,246 INFO [train.py:904] (7/8) Epoch 13, batch 3700, loss[loss=0.1943, simple_loss=0.2625, pruned_loss=0.06302, over 16728.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.26, pruned_loss=0.04909, over 3285128.02 frames. ], batch size: 124, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:39:44,177 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-04-29 18:40:10,903 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125524.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:40:25,076 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3089, 5.1162, 5.3207, 5.5048, 5.6679, 4.9715, 5.6278, 5.6355], device='cuda:7'), covar=tensor([0.1165, 0.0984, 0.1342, 0.0541, 0.0363, 0.0582, 0.0390, 0.0435], device='cuda:7'), in_proj_covar=tensor([0.0571, 0.0721, 0.0869, 0.0728, 0.0545, 0.0572, 0.0577, 0.0668], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:40:51,187 INFO [train.py:904] (7/8) Epoch 13, batch 3750, loss[loss=0.1827, simple_loss=0.2519, pruned_loss=0.0568, over 16894.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2605, pruned_loss=0.05005, over 3282931.12 frames. ], batch size: 109, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:40:53,703 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.90 vs. limit=5.0 2023-04-29 18:41:16,950 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4628, 3.8954, 4.2181, 3.0148, 3.7341, 4.1801, 3.8730, 2.4294], device='cuda:7'), covar=tensor([0.0430, 0.0171, 0.0032, 0.0250, 0.0070, 0.0079, 0.0055, 0.0351], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0073, 0.0071, 0.0126, 0.0083, 0.0092, 0.0082, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:41:24,168 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.222e+02 2.735e+02 3.211e+02 5.083e+02, threshold=5.471e+02, percent-clipped=0.0 2023-04-29 18:42:05,180 INFO [train.py:904] (7/8) Epoch 13, batch 3800, loss[loss=0.1889, simple_loss=0.2699, pruned_loss=0.0539, over 17001.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2616, pruned_loss=0.05155, over 3287749.61 frames. ], batch size: 41, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:42:07,516 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125603.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:42:11,227 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5084, 3.4941, 1.8030, 3.6824, 2.4382, 3.6183, 1.8454, 2.8004], device='cuda:7'), covar=tensor([0.0166, 0.0265, 0.1420, 0.0192, 0.0679, 0.0461, 0.1271, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0168, 0.0189, 0.0144, 0.0169, 0.0215, 0.0198, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:42:32,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2167, 1.5600, 2.0052, 2.0648, 2.2716, 2.2459, 1.6219, 2.3053], device='cuda:7'), covar=tensor([0.0160, 0.0354, 0.0219, 0.0209, 0.0200, 0.0186, 0.0356, 0.0088], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0179, 0.0164, 0.0170, 0.0178, 0.0133, 0.0178, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:43:18,389 INFO [train.py:904] (7/8) Epoch 13, batch 3850, loss[loss=0.1718, simple_loss=0.2488, pruned_loss=0.04737, over 16899.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2623, pruned_loss=0.05253, over 3287139.31 frames. ], batch size: 90, lr: 5.28e-03, grad_scale: 8.0 2023-04-29 18:43:29,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-29 18:43:35,173 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125664.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:43:50,562 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.444e+02 2.805e+02 3.444e+02 9.574e+02, threshold=5.610e+02, percent-clipped=2.0 2023-04-29 18:44:05,497 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-29 18:44:29,767 INFO [train.py:904] (7/8) Epoch 13, batch 3900, loss[loss=0.1712, simple_loss=0.2464, pruned_loss=0.04802, over 16812.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2619, pruned_loss=0.05281, over 3280132.52 frames. ], batch size: 102, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:44:33,591 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.54 vs. limit=5.0 2023-04-29 18:44:45,145 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125712.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 18:45:03,892 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125725.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:45:12,424 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7512, 6.0550, 5.7579, 5.9061, 5.3944, 5.2265, 5.5292, 6.1289], device='cuda:7'), covar=tensor([0.1032, 0.0740, 0.0981, 0.0656, 0.0770, 0.0673, 0.0991, 0.0766], device='cuda:7'), in_proj_covar=tensor([0.0584, 0.0738, 0.0595, 0.0524, 0.0466, 0.0471, 0.0616, 0.0567], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:45:14,828 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125732.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:45:14,945 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4524, 4.0237, 4.3340, 2.8876, 3.8768, 4.2666, 3.8546, 2.2584], device='cuda:7'), covar=tensor([0.0451, 0.0121, 0.0035, 0.0298, 0.0058, 0.0061, 0.0056, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0072, 0.0072, 0.0127, 0.0083, 0.0092, 0.0083, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:45:42,340 INFO [train.py:904] (7/8) Epoch 13, batch 3950, loss[loss=0.176, simple_loss=0.2535, pruned_loss=0.04926, over 16890.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2613, pruned_loss=0.05345, over 3279544.40 frames. ], batch size: 102, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:46:16,759 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.736e+02 2.289e+02 2.629e+02 3.210e+02 7.394e+02, threshold=5.259e+02, percent-clipped=3.0 2023-04-29 18:46:18,466 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6697, 2.5153, 1.9586, 2.2976, 2.8488, 2.5874, 2.9583, 2.9537], device='cuda:7'), covar=tensor([0.0146, 0.0262, 0.0367, 0.0348, 0.0163, 0.0265, 0.0174, 0.0183], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0216, 0.0207, 0.0209, 0.0216, 0.0214, 0.0226, 0.0207], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:46:20,064 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125777.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:46:30,440 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=125784.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:46:32,768 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=125786.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:46:45,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4811, 4.3955, 4.5196, 2.9706, 3.8702, 4.3965, 3.9037, 2.4438], device='cuda:7'), covar=tensor([0.0434, 0.0017, 0.0018, 0.0280, 0.0060, 0.0060, 0.0056, 0.0345], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0073, 0.0072, 0.0128, 0.0084, 0.0094, 0.0084, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:46:55,505 INFO [train.py:904] (7/8) Epoch 13, batch 4000, loss[loss=0.1707, simple_loss=0.2578, pruned_loss=0.04184, over 17137.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2611, pruned_loss=0.0537, over 3278855.65 frames. ], batch size: 48, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:47:05,019 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7206, 5.0035, 5.1925, 4.9992, 5.0578, 5.6538, 5.1554, 4.8862], device='cuda:7'), covar=tensor([0.1052, 0.1629, 0.1681, 0.1951, 0.2497, 0.0933, 0.1346, 0.2157], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0528, 0.0571, 0.0456, 0.0614, 0.0594, 0.0453, 0.0607], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:47:38,001 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=125832.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:48:05,369 INFO [train.py:904] (7/8) Epoch 13, batch 4050, loss[loss=0.1703, simple_loss=0.259, pruned_loss=0.04083, over 16825.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2617, pruned_loss=0.05294, over 3273691.23 frames. ], batch size: 83, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:48:36,497 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 1.993e+02 2.274e+02 2.780e+02 5.444e+02, threshold=4.549e+02, percent-clipped=3.0 2023-04-29 18:48:49,499 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-29 18:49:15,116 INFO [train.py:904] (7/8) Epoch 13, batch 4100, loss[loss=0.1864, simple_loss=0.2727, pruned_loss=0.05008, over 16781.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2626, pruned_loss=0.05215, over 3286856.94 frames. ], batch size: 124, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:50:30,164 INFO [train.py:904] (7/8) Epoch 13, batch 4150, loss[loss=0.1923, simple_loss=0.2891, pruned_loss=0.04778, over 16659.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2706, pruned_loss=0.05509, over 3256400.74 frames. ], batch size: 89, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:50:42,636 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=125959.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:51:05,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.394e+02 2.865e+02 3.650e+02 6.775e+02, threshold=5.730e+02, percent-clipped=10.0 2023-04-29 18:51:39,656 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2023-04-29 18:51:40,676 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=125998.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:51:48,709 INFO [train.py:904] (7/8) Epoch 13, batch 4200, loss[loss=0.2521, simple_loss=0.3219, pruned_loss=0.09113, over 11917.00 frames. ], tot_loss[loss=0.196, simple_loss=0.278, pruned_loss=0.05704, over 3218439.51 frames. ], batch size: 248, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:52:04,539 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126012.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 18:52:08,856 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3706, 4.6246, 4.4607, 4.4938, 4.2174, 4.1103, 4.1633, 4.6884], device='cuda:7'), covar=tensor([0.0983, 0.0848, 0.0874, 0.0702, 0.0749, 0.1358, 0.1037, 0.0847], device='cuda:7'), in_proj_covar=tensor([0.0570, 0.0723, 0.0582, 0.0512, 0.0458, 0.0463, 0.0602, 0.0554], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:52:34,273 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:52:35,668 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3845, 5.6241, 5.3862, 5.5109, 5.1826, 4.9738, 4.9794, 5.7322], device='cuda:7'), covar=tensor([0.0840, 0.0659, 0.0790, 0.0601, 0.0744, 0.0717, 0.0997, 0.0690], device='cuda:7'), in_proj_covar=tensor([0.0569, 0.0722, 0.0581, 0.0510, 0.0457, 0.0463, 0.0600, 0.0554], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:53:02,532 INFO [train.py:904] (7/8) Epoch 13, batch 4250, loss[loss=0.2117, simple_loss=0.2806, pruned_loss=0.07136, over 12034.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.281, pruned_loss=0.05662, over 3209614.00 frames. ], batch size: 247, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:53:13,112 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126059.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:14,162 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126060.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:36,047 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.365e+02 2.942e+02 3.644e+02 5.722e+02, threshold=5.884e+02, percent-clipped=0.0 2023-04-29 18:53:39,746 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126077.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:39,814 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0081, 2.5265, 2.6487, 1.7819, 2.7056, 2.7916, 2.4130, 2.3976], device='cuda:7'), covar=tensor([0.0696, 0.0208, 0.0216, 0.0951, 0.0097, 0.0222, 0.0418, 0.0393], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0101, 0.0088, 0.0136, 0.0069, 0.0109, 0.0121, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 18:53:43,910 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126080.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:53:45,896 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126081.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:54:16,797 INFO [train.py:904] (7/8) Epoch 13, batch 4300, loss[loss=0.2237, simple_loss=0.304, pruned_loss=0.07169, over 11913.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2819, pruned_loss=0.05566, over 3184307.56 frames. ], batch size: 248, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:54:51,667 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126125.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:55:01,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5162, 4.7487, 4.9219, 4.7235, 4.7699, 5.3095, 4.8461, 4.5197], device='cuda:7'), covar=tensor([0.1082, 0.1514, 0.1667, 0.1585, 0.2159, 0.0861, 0.1244, 0.2262], device='cuda:7'), in_proj_covar=tensor([0.0363, 0.0509, 0.0552, 0.0436, 0.0589, 0.0578, 0.0440, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 18:55:09,272 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 18:55:26,549 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7058, 4.7167, 4.5090, 3.9248, 4.6514, 1.5820, 4.3988, 4.3041], device='cuda:7'), covar=tensor([0.0054, 0.0045, 0.0128, 0.0261, 0.0057, 0.2553, 0.0090, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0128, 0.0173, 0.0162, 0.0146, 0.0184, 0.0163, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 18:55:30,684 INFO [train.py:904] (7/8) Epoch 13, batch 4350, loss[loss=0.2171, simple_loss=0.3003, pruned_loss=0.06691, over 11700.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.285, pruned_loss=0.05666, over 3170663.34 frames. ], batch size: 247, lr: 5.27e-03, grad_scale: 8.0 2023-04-29 18:56:06,024 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.441e+02 2.808e+02 3.338e+02 6.392e+02, threshold=5.616e+02, percent-clipped=2.0 2023-04-29 18:56:46,674 INFO [train.py:904] (7/8) Epoch 13, batch 4400, loss[loss=0.2095, simple_loss=0.2975, pruned_loss=0.06069, over 16200.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2869, pruned_loss=0.05764, over 3182451.02 frames. ], batch size: 165, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 18:57:59,273 INFO [train.py:904] (7/8) Epoch 13, batch 4450, loss[loss=0.2092, simple_loss=0.3024, pruned_loss=0.05797, over 16245.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2907, pruned_loss=0.05917, over 3185841.08 frames. ], batch size: 165, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 18:58:10,182 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126259.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:58:18,378 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7588, 3.9948, 2.5381, 4.7528, 3.0509, 4.5854, 2.5712, 2.9344], device='cuda:7'), covar=tensor([0.0245, 0.0311, 0.1425, 0.0082, 0.0683, 0.0361, 0.1380, 0.0781], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0167, 0.0189, 0.0140, 0.0168, 0.0212, 0.0197, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 18:58:33,038 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.129e+02 2.519e+02 3.084e+02 5.887e+02, threshold=5.038e+02, percent-clipped=2.0 2023-04-29 18:59:14,454 INFO [train.py:904] (7/8) Epoch 13, batch 4500, loss[loss=0.1871, simple_loss=0.2769, pruned_loss=0.04866, over 16892.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2906, pruned_loss=0.05935, over 3196800.59 frames. ], batch size: 42, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 18:59:21,427 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126307.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 18:59:57,362 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6109, 1.5013, 2.1414, 2.5815, 2.5279, 2.7714, 1.4619, 2.7804], device='cuda:7'), covar=tensor([0.0138, 0.0418, 0.0200, 0.0180, 0.0179, 0.0121, 0.0477, 0.0085], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0176, 0.0159, 0.0165, 0.0176, 0.0132, 0.0175, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 19:00:10,390 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-29 19:00:25,580 INFO [train.py:904] (7/8) Epoch 13, batch 4550, loss[loss=0.2011, simple_loss=0.2944, pruned_loss=0.05394, over 16733.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.291, pruned_loss=0.06004, over 3192309.72 frames. ], batch size: 83, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:00:28,884 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126354.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:00:39,582 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0958, 3.8988, 4.6608, 2.2516, 4.8517, 4.8726, 3.2486, 3.7980], device='cuda:7'), covar=tensor([0.0676, 0.0221, 0.0126, 0.1060, 0.0032, 0.0060, 0.0390, 0.0331], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0103, 0.0089, 0.0138, 0.0070, 0.0110, 0.0123, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 19:00:59,196 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.430e+02 2.007e+02 2.258e+02 2.674e+02 4.772e+02, threshold=4.515e+02, percent-clipped=0.0 2023-04-29 19:01:06,178 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126380.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:01:07,321 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126381.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:01:35,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3361, 4.2993, 4.1142, 3.2398, 4.2213, 1.5605, 3.9438, 3.5713], device='cuda:7'), covar=tensor([0.0069, 0.0062, 0.0142, 0.0418, 0.0075, 0.2941, 0.0118, 0.0280], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0126, 0.0171, 0.0161, 0.0144, 0.0182, 0.0160, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:01:36,171 INFO [train.py:904] (7/8) Epoch 13, batch 4600, loss[loss=0.2091, simple_loss=0.2914, pruned_loss=0.06342, over 16328.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2919, pruned_loss=0.0602, over 3192283.37 frames. ], batch size: 146, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:02:00,583 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6717, 2.8440, 2.6405, 4.3789, 3.4934, 4.0529, 1.5510, 3.0089], device='cuda:7'), covar=tensor([0.1274, 0.0641, 0.1131, 0.0119, 0.0346, 0.0340, 0.1505, 0.0729], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0161, 0.0183, 0.0157, 0.0198, 0.0206, 0.0182, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 19:02:22,725 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126429.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:02:23,051 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5266, 2.5133, 2.3756, 4.6277, 2.1649, 2.9821, 2.5040, 2.6288], device='cuda:7'), covar=tensor([0.0967, 0.2807, 0.2130, 0.0289, 0.3596, 0.1861, 0.2542, 0.2842], device='cuda:7'), in_proj_covar=tensor([0.0372, 0.0405, 0.0336, 0.0319, 0.0419, 0.0470, 0.0369, 0.0473], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:02:27,631 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.01 vs. limit=5.0 2023-04-29 19:02:39,298 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126441.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:02:53,250 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126450.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:02:55,164 INFO [train.py:904] (7/8) Epoch 13, batch 4650, loss[loss=0.1915, simple_loss=0.2761, pruned_loss=0.05339, over 16397.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2907, pruned_loss=0.06012, over 3219704.16 frames. ], batch size: 68, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:03:27,527 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.885e+02 2.230e+02 2.796e+02 5.506e+02, threshold=4.460e+02, percent-clipped=0.0 2023-04-29 19:03:36,415 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 19:04:07,667 INFO [train.py:904] (7/8) Epoch 13, batch 4700, loss[loss=0.1888, simple_loss=0.2679, pruned_loss=0.05486, over 17062.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2877, pruned_loss=0.05883, over 3218015.27 frames. ], batch size: 55, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:04:21,551 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126511.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:05:13,257 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-04-29 19:05:21,236 INFO [train.py:904] (7/8) Epoch 13, batch 4750, loss[loss=0.2225, simple_loss=0.2913, pruned_loss=0.07688, over 11587.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2837, pruned_loss=0.05708, over 3209719.22 frames. ], batch size: 248, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:05:53,857 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 1.907e+02 2.465e+02 2.943e+02 4.218e+02, threshold=4.930e+02, percent-clipped=1.0 2023-04-29 19:06:32,271 INFO [train.py:904] (7/8) Epoch 13, batch 4800, loss[loss=0.1895, simple_loss=0.2827, pruned_loss=0.0481, over 16752.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.281, pruned_loss=0.05566, over 3199933.56 frames. ], batch size: 124, lr: 5.26e-03, grad_scale: 8.0 2023-04-29 19:07:26,361 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-29 19:07:43,558 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8998, 2.2928, 2.3249, 2.8298, 2.0937, 3.2415, 1.7221, 2.7059], device='cuda:7'), covar=tensor([0.1159, 0.0610, 0.1072, 0.0143, 0.0117, 0.0340, 0.1388, 0.0665], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0163, 0.0185, 0.0159, 0.0201, 0.0208, 0.0184, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 19:07:46,146 INFO [train.py:904] (7/8) Epoch 13, batch 4850, loss[loss=0.1554, simple_loss=0.2571, pruned_loss=0.0269, over 16873.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2824, pruned_loss=0.0555, over 3166117.48 frames. ], batch size: 102, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:07:50,112 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=126654.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:08:19,808 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 1.948e+02 2.386e+02 2.716e+02 6.913e+02, threshold=4.771e+02, percent-clipped=1.0 2023-04-29 19:08:59,244 INFO [train.py:904] (7/8) Epoch 13, batch 4900, loss[loss=0.1864, simple_loss=0.2794, pruned_loss=0.04669, over 16444.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2818, pruned_loss=0.05428, over 3162649.68 frames. ], batch size: 146, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:09:00,359 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=126702.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:09:21,146 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0052, 4.0211, 4.3957, 4.3687, 4.3351, 4.0781, 4.0435, 4.0202], device='cuda:7'), covar=tensor([0.0291, 0.0469, 0.0297, 0.0403, 0.0431, 0.0327, 0.0866, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0357, 0.0356, 0.0347, 0.0407, 0.0379, 0.0476, 0.0306], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 19:09:22,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0002, 2.7480, 2.7438, 1.9659, 2.6227, 2.1745, 2.7596, 2.9263], device='cuda:7'), covar=tensor([0.0262, 0.0732, 0.0506, 0.1715, 0.0738, 0.0833, 0.0588, 0.0604], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0150, 0.0159, 0.0145, 0.0138, 0.0125, 0.0138, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 19:09:44,219 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-29 19:09:48,683 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3574, 3.2988, 3.4089, 3.5152, 3.5579, 3.2406, 3.5384, 3.6005], device='cuda:7'), covar=tensor([0.1127, 0.0878, 0.1075, 0.0560, 0.0550, 0.2689, 0.0820, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0542, 0.0677, 0.0809, 0.0682, 0.0514, 0.0535, 0.0546, 0.0624], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:09:49,802 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126736.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:10:12,989 INFO [train.py:904] (7/8) Epoch 13, batch 4950, loss[loss=0.2132, simple_loss=0.2946, pruned_loss=0.06586, over 16652.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.281, pruned_loss=0.05341, over 3160639.43 frames. ], batch size: 57, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:10:38,168 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3247, 2.0132, 1.5701, 1.8466, 2.3558, 2.0001, 2.1177, 2.4796], device='cuda:7'), covar=tensor([0.0109, 0.0360, 0.0469, 0.0394, 0.0201, 0.0330, 0.0163, 0.0206], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0213, 0.0206, 0.0206, 0.0212, 0.0211, 0.0217, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:10:45,735 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.239e+02 2.710e+02 3.273e+02 4.681e+02, threshold=5.421e+02, percent-clipped=0.0 2023-04-29 19:11:26,271 INFO [train.py:904] (7/8) Epoch 13, batch 5000, loss[loss=0.1901, simple_loss=0.2828, pruned_loss=0.04875, over 16497.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2825, pruned_loss=0.05322, over 3188751.93 frames. ], batch size: 75, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:11:32,788 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=126806.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:11:48,052 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=126816.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:12:21,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7424, 4.7850, 5.0945, 5.1127, 5.0489, 4.7947, 4.7158, 4.5565], device='cuda:7'), covar=tensor([0.0228, 0.0330, 0.0341, 0.0348, 0.0465, 0.0280, 0.0814, 0.0354], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0356, 0.0356, 0.0345, 0.0405, 0.0379, 0.0475, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 19:12:38,207 INFO [train.py:904] (7/8) Epoch 13, batch 5050, loss[loss=0.1928, simple_loss=0.278, pruned_loss=0.05382, over 17055.00 frames. ], tot_loss[loss=0.1948, simple_loss=0.2834, pruned_loss=0.0531, over 3196507.64 frames. ], batch size: 55, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:13:11,128 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.161e+02 2.561e+02 3.247e+02 4.827e+02, threshold=5.121e+02, percent-clipped=0.0 2023-04-29 19:13:14,686 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=126877.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:13:24,156 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2513, 4.9723, 5.1995, 5.4256, 5.6228, 4.9474, 5.6215, 5.5767], device='cuda:7'), covar=tensor([0.1462, 0.1209, 0.1773, 0.0672, 0.0401, 0.0622, 0.0370, 0.0540], device='cuda:7'), in_proj_covar=tensor([0.0541, 0.0677, 0.0811, 0.0684, 0.0514, 0.0534, 0.0545, 0.0623], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:13:25,746 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-04-29 19:13:51,478 INFO [train.py:904] (7/8) Epoch 13, batch 5100, loss[loss=0.1961, simple_loss=0.2786, pruned_loss=0.05686, over 11829.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2806, pruned_loss=0.05167, over 3208725.23 frames. ], batch size: 247, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:14:06,067 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.54 vs. limit=5.0 2023-04-29 19:14:38,524 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1628, 4.0898, 4.0765, 3.3532, 4.0955, 1.5581, 3.8188, 3.6771], device='cuda:7'), covar=tensor([0.0085, 0.0092, 0.0126, 0.0384, 0.0082, 0.2544, 0.0131, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0124, 0.0168, 0.0159, 0.0141, 0.0180, 0.0158, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:14:49,018 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5435, 3.6200, 3.3687, 3.0566, 3.1634, 3.4642, 3.2356, 3.2816], device='cuda:7'), covar=tensor([0.0559, 0.0398, 0.0262, 0.0251, 0.0583, 0.0374, 0.1471, 0.0487], device='cuda:7'), in_proj_covar=tensor([0.0250, 0.0341, 0.0300, 0.0281, 0.0322, 0.0325, 0.0204, 0.0350], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:15:04,926 INFO [train.py:904] (7/8) Epoch 13, batch 5150, loss[loss=0.1699, simple_loss=0.2645, pruned_loss=0.03764, over 16743.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2811, pruned_loss=0.05123, over 3196862.31 frames. ], batch size: 83, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:15:37,793 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.071e+02 2.483e+02 2.841e+02 7.629e+02, threshold=4.965e+02, percent-clipped=2.0 2023-04-29 19:16:17,737 INFO [train.py:904] (7/8) Epoch 13, batch 5200, loss[loss=0.1761, simple_loss=0.2638, pruned_loss=0.0442, over 16882.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2798, pruned_loss=0.0508, over 3192275.86 frames. ], batch size: 109, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:16:21,158 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1201, 3.6734, 3.7464, 2.4529, 3.3370, 3.6733, 3.3810, 2.1966], device='cuda:7'), covar=tensor([0.0459, 0.0031, 0.0028, 0.0309, 0.0077, 0.0084, 0.0074, 0.0345], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0070, 0.0071, 0.0127, 0.0084, 0.0092, 0.0082, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 19:17:08,376 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127036.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:17:31,272 INFO [train.py:904] (7/8) Epoch 13, batch 5250, loss[loss=0.2279, simple_loss=0.2967, pruned_loss=0.07955, over 12434.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2767, pruned_loss=0.05026, over 3200991.70 frames. ], batch size: 247, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:17:34,745 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.68 vs. limit=5.0 2023-04-29 19:18:04,191 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.082e+02 2.366e+02 2.761e+02 5.233e+02, threshold=4.733e+02, percent-clipped=1.0 2023-04-29 19:18:18,869 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127084.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:18:44,213 INFO [train.py:904] (7/8) Epoch 13, batch 5300, loss[loss=0.1793, simple_loss=0.2614, pruned_loss=0.04864, over 16810.00 frames. ], tot_loss[loss=0.1853, simple_loss=0.2731, pruned_loss=0.04878, over 3206595.19 frames. ], batch size: 116, lr: 5.25e-03, grad_scale: 8.0 2023-04-29 19:18:50,760 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127106.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:19:23,089 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6050, 4.4172, 4.3980, 2.9584, 3.7757, 4.3638, 3.7156, 2.6369], device='cuda:7'), covar=tensor([0.0424, 0.0020, 0.0023, 0.0292, 0.0074, 0.0062, 0.0066, 0.0323], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0069, 0.0070, 0.0126, 0.0083, 0.0090, 0.0082, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 19:19:58,323 INFO [train.py:904] (7/8) Epoch 13, batch 5350, loss[loss=0.1992, simple_loss=0.2816, pruned_loss=0.05843, over 12181.00 frames. ], tot_loss[loss=0.1841, simple_loss=0.2716, pruned_loss=0.04825, over 3203014.15 frames. ], batch size: 246, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:20:01,843 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127154.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:20:27,673 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127172.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:20:32,664 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.449e+02 2.172e+02 2.538e+02 2.960e+02 5.660e+02, threshold=5.076e+02, percent-clipped=1.0 2023-04-29 19:20:39,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9287, 4.1690, 3.1735, 2.4443, 3.0547, 2.5569, 4.5792, 4.0011], device='cuda:7'), covar=tensor([0.2356, 0.0627, 0.1456, 0.2228, 0.2223, 0.1592, 0.0388, 0.0835], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0259, 0.0286, 0.0284, 0.0279, 0.0226, 0.0271, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:21:10,978 INFO [train.py:904] (7/8) Epoch 13, batch 5400, loss[loss=0.2114, simple_loss=0.3, pruned_loss=0.06142, over 15440.00 frames. ], tot_loss[loss=0.187, simple_loss=0.2751, pruned_loss=0.04949, over 3186837.35 frames. ], batch size: 190, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:22:20,967 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127249.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:22:26,312 INFO [train.py:904] (7/8) Epoch 13, batch 5450, loss[loss=0.2815, simple_loss=0.3402, pruned_loss=0.1114, over 12159.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2778, pruned_loss=0.05093, over 3193073.62 frames. ], batch size: 247, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:22:37,309 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8705, 2.6560, 2.6022, 1.9239, 2.5363, 2.7356, 2.5645, 1.8889], device='cuda:7'), covar=tensor([0.0334, 0.0061, 0.0052, 0.0290, 0.0094, 0.0099, 0.0087, 0.0321], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0069, 0.0070, 0.0125, 0.0083, 0.0090, 0.0081, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 19:23:02,329 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.258e+02 2.675e+02 3.738e+02 1.100e+03, threshold=5.349e+02, percent-clipped=10.0 2023-04-29 19:23:43,876 INFO [train.py:904] (7/8) Epoch 13, batch 5500, loss[loss=0.2069, simple_loss=0.2902, pruned_loss=0.06182, over 16329.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2857, pruned_loss=0.05618, over 3169457.60 frames. ], batch size: 35, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:23:57,150 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127310.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:24:02,899 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0463, 3.3815, 3.4598, 2.1332, 3.1680, 3.4850, 3.2329, 1.8028], device='cuda:7'), covar=tensor([0.0478, 0.0048, 0.0042, 0.0394, 0.0089, 0.0094, 0.0083, 0.0454], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0070, 0.0070, 0.0126, 0.0083, 0.0091, 0.0082, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 19:25:00,719 INFO [train.py:904] (7/8) Epoch 13, batch 5550, loss[loss=0.3381, simple_loss=0.3761, pruned_loss=0.1501, over 11291.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2936, pruned_loss=0.06158, over 3155932.78 frames. ], batch size: 248, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:25:38,574 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.215e+02 3.416e+02 4.230e+02 5.078e+02 9.045e+02, threshold=8.460e+02, percent-clipped=18.0 2023-04-29 19:26:20,886 INFO [train.py:904] (7/8) Epoch 13, batch 5600, loss[loss=0.1969, simple_loss=0.2808, pruned_loss=0.05645, over 17017.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2984, pruned_loss=0.06585, over 3133297.80 frames. ], batch size: 50, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:27:02,287 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127427.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 19:27:41,576 INFO [train.py:904] (7/8) Epoch 13, batch 5650, loss[loss=0.2252, simple_loss=0.3133, pruned_loss=0.0685, over 16790.00 frames. ], tot_loss[loss=0.2233, simple_loss=0.3046, pruned_loss=0.07098, over 3097833.22 frames. ], batch size: 39, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:28:12,734 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127472.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:28:17,740 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.064e+02 3.489e+02 4.637e+02 5.886e+02 1.352e+03, threshold=9.273e+02, percent-clipped=4.0 2023-04-29 19:28:36,934 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127488.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 19:28:58,373 INFO [train.py:904] (7/8) Epoch 13, batch 5700, loss[loss=0.2788, simple_loss=0.3371, pruned_loss=0.1103, over 11945.00 frames. ], tot_loss[loss=0.2256, simple_loss=0.3061, pruned_loss=0.07252, over 3085215.66 frames. ], batch size: 247, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:29:27,720 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127520.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:29:42,129 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127529.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:30:18,050 INFO [train.py:904] (7/8) Epoch 13, batch 5750, loss[loss=0.2463, simple_loss=0.3083, pruned_loss=0.09213, over 11150.00 frames. ], tot_loss[loss=0.2284, simple_loss=0.3085, pruned_loss=0.07413, over 3040173.62 frames. ], batch size: 248, lr: 5.24e-03, grad_scale: 8.0 2023-04-29 19:30:40,910 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5302, 1.5771, 2.1116, 2.4422, 2.4732, 2.7647, 1.7114, 2.7589], device='cuda:7'), covar=tensor([0.0150, 0.0436, 0.0258, 0.0231, 0.0222, 0.0157, 0.0422, 0.0101], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0176, 0.0159, 0.0162, 0.0173, 0.0129, 0.0174, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 19:30:56,350 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.152e+02 3.042e+02 3.668e+02 4.519e+02 8.025e+02, threshold=7.337e+02, percent-clipped=0.0 2023-04-29 19:31:21,790 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127590.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:31:39,689 INFO [train.py:904] (7/8) Epoch 13, batch 5800, loss[loss=0.2144, simple_loss=0.2973, pruned_loss=0.06575, over 16438.00 frames. ], tot_loss[loss=0.2247, simple_loss=0.307, pruned_loss=0.07122, over 3073000.86 frames. ], batch size: 68, lr: 5.24e-03, grad_scale: 4.0 2023-04-29 19:31:44,463 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127605.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:32:57,621 INFO [train.py:904] (7/8) Epoch 13, batch 5850, loss[loss=0.1908, simple_loss=0.2761, pruned_loss=0.05275, over 17201.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.3049, pruned_loss=0.06969, over 3075504.61 frames. ], batch size: 45, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:33:05,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6218, 4.9210, 4.6683, 4.6928, 4.4271, 4.4365, 4.4178, 5.0079], device='cuda:7'), covar=tensor([0.1013, 0.0772, 0.1055, 0.0733, 0.0743, 0.0985, 0.0976, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0563, 0.0698, 0.0573, 0.0497, 0.0441, 0.0450, 0.0584, 0.0539], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:33:15,134 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127662.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:33:22,814 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1490, 3.9066, 4.1570, 4.3023, 4.4435, 4.0009, 4.4034, 4.4452], device='cuda:7'), covar=tensor([0.1455, 0.1284, 0.1485, 0.0709, 0.0520, 0.1269, 0.0679, 0.0705], device='cuda:7'), in_proj_covar=tensor([0.0540, 0.0676, 0.0804, 0.0684, 0.0516, 0.0531, 0.0542, 0.0623], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:33:36,798 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.121e+02 3.118e+02 3.715e+02 4.396e+02 1.301e+03, threshold=7.431e+02, percent-clipped=3.0 2023-04-29 19:34:07,825 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5231, 3.5334, 2.8367, 2.1473, 2.3771, 2.2679, 3.5951, 3.3456], device='cuda:7'), covar=tensor([0.2565, 0.0600, 0.1478, 0.2500, 0.2270, 0.1751, 0.0440, 0.0989], device='cuda:7'), in_proj_covar=tensor([0.0307, 0.0258, 0.0287, 0.0285, 0.0281, 0.0227, 0.0271, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:34:19,538 INFO [train.py:904] (7/8) Epoch 13, batch 5900, loss[loss=0.2572, simple_loss=0.3146, pruned_loss=0.09984, over 11550.00 frames. ], tot_loss[loss=0.2221, simple_loss=0.3047, pruned_loss=0.06974, over 3082504.09 frames. ], batch size: 246, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:34:57,897 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127723.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:35:41,979 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5027, 4.5657, 4.3618, 4.1187, 4.0139, 4.4477, 4.2506, 4.1891], device='cuda:7'), covar=tensor([0.0597, 0.0581, 0.0280, 0.0263, 0.0942, 0.0405, 0.0458, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0251, 0.0340, 0.0297, 0.0277, 0.0318, 0.0323, 0.0201, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:35:42,729 INFO [train.py:904] (7/8) Epoch 13, batch 5950, loss[loss=0.224, simple_loss=0.3121, pruned_loss=0.06791, over 16856.00 frames. ], tot_loss[loss=0.2214, simple_loss=0.3054, pruned_loss=0.06868, over 3084518.90 frames. ], batch size: 116, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:36:09,359 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 19:36:21,829 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.141e+02 2.934e+02 3.631e+02 4.257e+02 8.531e+02, threshold=7.262e+02, percent-clipped=1.0 2023-04-29 19:36:26,821 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127779.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:36:33,434 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127783.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 19:36:50,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4141, 5.7468, 5.4806, 5.4899, 5.1793, 5.0811, 5.2114, 5.8413], device='cuda:7'), covar=tensor([0.1120, 0.0857, 0.0939, 0.0729, 0.0804, 0.0723, 0.0988, 0.0850], device='cuda:7'), in_proj_covar=tensor([0.0569, 0.0705, 0.0579, 0.0503, 0.0444, 0.0455, 0.0591, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:36:52,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 19:37:04,087 INFO [train.py:904] (7/8) Epoch 13, batch 6000, loss[loss=0.1959, simple_loss=0.2786, pruned_loss=0.0566, over 16904.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.3034, pruned_loss=0.06741, over 3097349.46 frames. ], batch size: 109, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:37:04,087 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 19:37:14,217 INFO [train.py:938] (7/8) Epoch 13, validation: loss=0.1599, simple_loss=0.2726, pruned_loss=0.02359, over 944034.00 frames. 2023-04-29 19:37:14,218 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 19:37:49,299 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-29 19:38:13,868 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127840.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:38:32,719 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127850.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:38:35,172 INFO [train.py:904] (7/8) Epoch 13, batch 6050, loss[loss=0.2416, simple_loss=0.3119, pruned_loss=0.08565, over 11665.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3022, pruned_loss=0.06717, over 3094844.89 frames. ], batch size: 246, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:39:07,473 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=127872.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:39:14,940 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.979e+02 3.574e+02 4.345e+02 1.015e+03, threshold=7.148e+02, percent-clipped=2.0 2023-04-29 19:39:28,469 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=127885.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:39:40,829 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-29 19:39:55,368 INFO [train.py:904] (7/8) Epoch 13, batch 6100, loss[loss=0.2046, simple_loss=0.2956, pruned_loss=0.05679, over 16877.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.3017, pruned_loss=0.06586, over 3109119.46 frames. ], batch size: 109, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:40:01,635 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=127905.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:40:12,718 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127911.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:40:35,787 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3341, 3.4308, 2.6107, 2.0501, 2.2964, 2.1900, 3.4788, 3.2220], device='cuda:7'), covar=tensor([0.2879, 0.0749, 0.1763, 0.2416, 0.2261, 0.1821, 0.0516, 0.1043], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0259, 0.0287, 0.0285, 0.0283, 0.0226, 0.0272, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:40:47,002 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=127933.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:41:15,204 INFO [train.py:904] (7/8) Epoch 13, batch 6150, loss[loss=0.1855, simple_loss=0.2764, pruned_loss=0.04727, over 17212.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.3002, pruned_loss=0.066, over 3098086.99 frames. ], batch size: 45, lr: 5.23e-03, grad_scale: 8.0 2023-04-29 19:41:17,426 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=127953.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:41:30,575 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 19:41:56,818 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.069e+02 3.037e+02 3.583e+02 4.334e+02 1.010e+03, threshold=7.167e+02, percent-clipped=2.0 2023-04-29 19:41:57,469 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5100, 4.4906, 4.4201, 3.6999, 4.4189, 1.6897, 4.1747, 4.1829], device='cuda:7'), covar=tensor([0.0087, 0.0070, 0.0129, 0.0328, 0.0082, 0.2460, 0.0123, 0.0194], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0125, 0.0171, 0.0161, 0.0143, 0.0184, 0.0159, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:42:39,406 INFO [train.py:904] (7/8) Epoch 13, batch 6200, loss[loss=0.2072, simple_loss=0.2953, pruned_loss=0.05955, over 16810.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2981, pruned_loss=0.06546, over 3100291.72 frames. ], batch size: 83, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:43:00,872 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8743, 5.1567, 4.8951, 4.9024, 4.6624, 4.6469, 4.6594, 5.2263], device='cuda:7'), covar=tensor([0.1148, 0.0860, 0.1106, 0.0835, 0.0893, 0.0879, 0.1046, 0.0903], device='cuda:7'), in_proj_covar=tensor([0.0567, 0.0702, 0.0578, 0.0504, 0.0443, 0.0453, 0.0588, 0.0542], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:43:01,171 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2023-04-29 19:43:05,166 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128018.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:43:37,053 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 19:43:57,685 INFO [train.py:904] (7/8) Epoch 13, batch 6250, loss[loss=0.2089, simple_loss=0.3001, pruned_loss=0.05886, over 16471.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2971, pruned_loss=0.06451, over 3120262.20 frames. ], batch size: 68, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:44:04,261 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3362, 4.3131, 4.2399, 3.5124, 4.2454, 1.8092, 4.0445, 3.9462], device='cuda:7'), covar=tensor([0.0094, 0.0077, 0.0143, 0.0318, 0.0089, 0.2250, 0.0126, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0125, 0.0171, 0.0162, 0.0143, 0.0184, 0.0159, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:44:37,189 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.989e+02 2.817e+02 3.684e+02 4.282e+02 1.153e+03, threshold=7.367e+02, percent-clipped=4.0 2023-04-29 19:44:45,121 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128083.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 19:45:13,036 INFO [train.py:904] (7/8) Epoch 13, batch 6300, loss[loss=0.2557, simple_loss=0.3151, pruned_loss=0.09812, over 11748.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2971, pruned_loss=0.06425, over 3111701.91 frames. ], batch size: 246, lr: 5.23e-03, grad_scale: 4.0 2023-04-29 19:45:34,410 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1121, 5.0937, 4.9188, 4.2979, 4.9891, 1.9604, 4.7843, 4.8038], device='cuda:7'), covar=tensor([0.0059, 0.0050, 0.0119, 0.0307, 0.0065, 0.2147, 0.0097, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0125, 0.0171, 0.0161, 0.0143, 0.0184, 0.0159, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:46:02,875 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128131.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 19:46:03,010 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0889, 3.2424, 3.5583, 1.6463, 3.6895, 3.7353, 2.7144, 2.6358], device='cuda:7'), covar=tensor([0.0899, 0.0211, 0.0181, 0.1248, 0.0061, 0.0140, 0.0445, 0.0474], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0101, 0.0088, 0.0138, 0.0069, 0.0109, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 19:46:08,203 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128135.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:46:28,923 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1804, 1.4464, 1.8513, 2.0271, 2.2061, 2.3165, 1.5348, 2.2581], device='cuda:7'), covar=tensor([0.0158, 0.0384, 0.0213, 0.0222, 0.0231, 0.0159, 0.0401, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0175, 0.0159, 0.0162, 0.0173, 0.0130, 0.0175, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 19:46:33,880 INFO [train.py:904] (7/8) Epoch 13, batch 6350, loss[loss=0.2499, simple_loss=0.3342, pruned_loss=0.08279, over 16229.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2978, pruned_loss=0.06545, over 3104670.43 frames. ], batch size: 165, lr: 5.22e-03, grad_scale: 4.0 2023-04-29 19:46:49,594 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4101, 4.2122, 4.4160, 4.5808, 4.7334, 4.3089, 4.6634, 4.6870], device='cuda:7'), covar=tensor([0.1546, 0.1156, 0.1401, 0.0617, 0.0507, 0.0869, 0.0604, 0.0668], device='cuda:7'), in_proj_covar=tensor([0.0538, 0.0671, 0.0800, 0.0683, 0.0518, 0.0528, 0.0547, 0.0629], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:47:13,680 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.177e+02 3.833e+02 4.854e+02 7.993e+02, threshold=7.666e+02, percent-clipped=3.0 2023-04-29 19:47:25,521 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128185.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:47:49,930 INFO [train.py:904] (7/8) Epoch 13, batch 6400, loss[loss=0.1942, simple_loss=0.2823, pruned_loss=0.05311, over 16614.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2982, pruned_loss=0.06636, over 3102104.46 frames. ], batch size: 57, lr: 5.22e-03, grad_scale: 8.0 2023-04-29 19:47:56,324 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128206.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:48:30,156 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128228.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:48:36,411 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128233.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:49:04,031 INFO [train.py:904] (7/8) Epoch 13, batch 6450, loss[loss=0.2335, simple_loss=0.3069, pruned_loss=0.08007, over 11495.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2978, pruned_loss=0.06523, over 3118372.17 frames. ], batch size: 247, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:49:08,644 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-29 19:49:34,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-29 19:49:45,307 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0195, 2.4464, 2.0393, 2.1429, 2.7311, 2.3710, 2.8971, 2.9450], device='cuda:7'), covar=tensor([0.0117, 0.0293, 0.0384, 0.0366, 0.0199, 0.0310, 0.0185, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0209, 0.0202, 0.0204, 0.0208, 0.0207, 0.0214, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:49:48,884 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.912e+02 3.353e+02 4.052e+02 7.402e+02, threshold=6.705e+02, percent-clipped=0.0 2023-04-29 19:49:59,229 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0195, 2.4757, 2.6110, 1.8918, 2.7460, 2.7994, 2.3805, 2.3805], device='cuda:7'), covar=tensor([0.0739, 0.0229, 0.0208, 0.0979, 0.0098, 0.0225, 0.0446, 0.0406], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0102, 0.0089, 0.0138, 0.0069, 0.0109, 0.0121, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 19:50:21,896 INFO [train.py:904] (7/8) Epoch 13, batch 6500, loss[loss=0.2283, simple_loss=0.3045, pruned_loss=0.07607, over 15290.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2958, pruned_loss=0.06462, over 3117766.71 frames. ], batch size: 190, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:50:29,884 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 19:50:45,404 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128318.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:50:48,251 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128320.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:51:15,709 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1225, 4.4144, 4.0681, 3.8933, 3.3543, 4.3183, 4.0994, 3.8679], device='cuda:7'), covar=tensor([0.1296, 0.1297, 0.0528, 0.0468, 0.2091, 0.0616, 0.0910, 0.1022], device='cuda:7'), in_proj_covar=tensor([0.0249, 0.0341, 0.0298, 0.0278, 0.0318, 0.0323, 0.0202, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 19:51:39,460 INFO [train.py:904] (7/8) Epoch 13, batch 6550, loss[loss=0.2063, simple_loss=0.3028, pruned_loss=0.05491, over 15178.00 frames. ], tot_loss[loss=0.2155, simple_loss=0.2993, pruned_loss=0.06583, over 3097448.13 frames. ], batch size: 190, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:52:01,655 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128366.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:52:22,964 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 2.847e+02 3.449e+02 4.250e+02 8.173e+02, threshold=6.898e+02, percent-clipped=4.0 2023-04-29 19:52:25,115 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128381.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:52:56,146 INFO [train.py:904] (7/8) Epoch 13, batch 6600, loss[loss=0.2492, simple_loss=0.3239, pruned_loss=0.08724, over 16249.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3019, pruned_loss=0.06673, over 3102163.57 frames. ], batch size: 165, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:53:14,112 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128413.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:53:48,826 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128435.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:54:14,490 INFO [train.py:904] (7/8) Epoch 13, batch 6650, loss[loss=0.1763, simple_loss=0.2669, pruned_loss=0.04282, over 17159.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.3014, pruned_loss=0.06686, over 3108364.62 frames. ], batch size: 46, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:54:48,653 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128474.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:54:56,952 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.058e+02 3.713e+02 4.648e+02 8.477e+02, threshold=7.425e+02, percent-clipped=4.0 2023-04-29 19:55:02,280 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128483.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:55:30,392 INFO [train.py:904] (7/8) Epoch 13, batch 6700, loss[loss=0.1991, simple_loss=0.2851, pruned_loss=0.0565, over 16413.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.3006, pruned_loss=0.06723, over 3105588.06 frames. ], batch size: 146, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:55:36,534 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128506.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:56:10,903 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128528.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:56:46,045 INFO [train.py:904] (7/8) Epoch 13, batch 6750, loss[loss=0.2005, simple_loss=0.28, pruned_loss=0.06046, over 17195.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2992, pruned_loss=0.06695, over 3109364.93 frames. ], batch size: 46, lr: 5.22e-03, grad_scale: 2.0 2023-04-29 19:56:49,446 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128554.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:56:52,681 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9583, 3.3744, 3.3904, 2.1933, 3.0947, 3.3056, 3.1650, 1.9234], device='cuda:7'), covar=tensor([0.0494, 0.0036, 0.0040, 0.0355, 0.0086, 0.0115, 0.0079, 0.0413], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0070, 0.0072, 0.0127, 0.0083, 0.0094, 0.0082, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 19:57:22,974 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=128576.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:57:23,066 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7058, 3.7874, 4.0876, 4.0664, 4.0692, 3.8128, 3.8263, 3.8109], device='cuda:7'), covar=tensor([0.0367, 0.0601, 0.0409, 0.0450, 0.0543, 0.0445, 0.0925, 0.0529], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0368, 0.0369, 0.0353, 0.0417, 0.0389, 0.0490, 0.0313], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 19:57:28,180 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.101e+02 3.923e+02 4.752e+02 1.055e+03, threshold=7.847e+02, percent-clipped=2.0 2023-04-29 19:58:01,820 INFO [train.py:904] (7/8) Epoch 13, batch 6800, loss[loss=0.2075, simple_loss=0.2907, pruned_loss=0.06217, over 16623.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2999, pruned_loss=0.06713, over 3108793.06 frames. ], batch size: 57, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 19:59:19,108 INFO [train.py:904] (7/8) Epoch 13, batch 6850, loss[loss=0.2093, simple_loss=0.3042, pruned_loss=0.05719, over 16687.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.3006, pruned_loss=0.06716, over 3102521.73 frames. ], batch size: 134, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 19:59:41,289 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 19:59:55,668 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128676.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 19:59:58,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8797, 2.6300, 2.5929, 1.9902, 2.4997, 2.6735, 2.5978, 1.9441], device='cuda:7'), covar=tensor([0.0353, 0.0062, 0.0061, 0.0289, 0.0096, 0.0106, 0.0078, 0.0320], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0069, 0.0072, 0.0127, 0.0083, 0.0094, 0.0082, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 20:00:00,830 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.799e+02 3.352e+02 4.100e+02 8.801e+02, threshold=6.704e+02, percent-clipped=1.0 2023-04-29 20:00:34,571 INFO [train.py:904] (7/8) Epoch 13, batch 6900, loss[loss=0.2073, simple_loss=0.2979, pruned_loss=0.05836, over 16861.00 frames. ], tot_loss[loss=0.2197, simple_loss=0.3035, pruned_loss=0.06793, over 3089676.56 frames. ], batch size: 116, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:01:52,399 INFO [train.py:904] (7/8) Epoch 13, batch 6950, loss[loss=0.2804, simple_loss=0.3384, pruned_loss=0.1112, over 11428.00 frames. ], tot_loss[loss=0.2224, simple_loss=0.3052, pruned_loss=0.06979, over 3081884.80 frames. ], batch size: 248, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:02:20,040 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=128769.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:02:36,174 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.242e+02 3.339e+02 3.931e+02 4.735e+02 7.907e+02, threshold=7.862e+02, percent-clipped=3.0 2023-04-29 20:02:38,420 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128781.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:03:08,760 INFO [train.py:904] (7/8) Epoch 13, batch 7000, loss[loss=0.2817, simple_loss=0.3309, pruned_loss=0.1163, over 11590.00 frames. ], tot_loss[loss=0.2219, simple_loss=0.3052, pruned_loss=0.06929, over 3084953.66 frames. ], batch size: 247, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:04:07,261 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128842.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:04:20,579 INFO [train.py:904] (7/8) Epoch 13, batch 7050, loss[loss=0.2029, simple_loss=0.2951, pruned_loss=0.05538, over 16845.00 frames. ], tot_loss[loss=0.2209, simple_loss=0.305, pruned_loss=0.06844, over 3088246.95 frames. ], batch size: 96, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:04:35,635 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-29 20:05:01,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6905, 4.6483, 4.5319, 3.8445, 4.5617, 1.6757, 4.3451, 4.3441], device='cuda:7'), covar=tensor([0.0077, 0.0069, 0.0155, 0.0348, 0.0090, 0.2451, 0.0121, 0.0182], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0123, 0.0168, 0.0159, 0.0141, 0.0183, 0.0157, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:05:02,603 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 2.845e+02 3.584e+02 4.333e+02 9.135e+02, threshold=7.167e+02, percent-clipped=3.0 2023-04-29 20:05:03,851 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7187, 4.7450, 4.6047, 4.2907, 4.2371, 4.6554, 4.5299, 4.3515], device='cuda:7'), covar=tensor([0.0586, 0.0405, 0.0277, 0.0268, 0.0947, 0.0419, 0.0397, 0.0612], device='cuda:7'), in_proj_covar=tensor([0.0243, 0.0332, 0.0292, 0.0272, 0.0310, 0.0315, 0.0198, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:05:19,838 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=128890.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:05:38,305 INFO [train.py:904] (7/8) Epoch 13, batch 7100, loss[loss=0.2065, simple_loss=0.2895, pruned_loss=0.06171, over 16545.00 frames. ], tot_loss[loss=0.2198, simple_loss=0.3034, pruned_loss=0.0681, over 3084005.89 frames. ], batch size: 75, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:06:21,001 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3118, 4.2959, 4.1921, 3.5414, 4.2169, 1.6488, 4.0173, 3.8655], device='cuda:7'), covar=tensor([0.0088, 0.0075, 0.0151, 0.0325, 0.0088, 0.2540, 0.0125, 0.0218], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0124, 0.0169, 0.0160, 0.0141, 0.0184, 0.0158, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:06:56,811 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=128951.0, num_to_drop=1, layers_to_drop={3} 2023-04-29 20:06:57,450 INFO [train.py:904] (7/8) Epoch 13, batch 7150, loss[loss=0.2104, simple_loss=0.3073, pruned_loss=0.05677, over 16925.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.3015, pruned_loss=0.06804, over 3055677.72 frames. ], batch size: 96, lr: 5.21e-03, grad_scale: 4.0 2023-04-29 20:07:19,471 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0801, 2.3300, 2.4067, 2.8739, 2.1500, 3.2344, 1.8245, 2.7176], device='cuda:7'), covar=tensor([0.1031, 0.0532, 0.0951, 0.0169, 0.0134, 0.0369, 0.1243, 0.0659], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0163, 0.0183, 0.0157, 0.0203, 0.0208, 0.0186, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 20:07:34,694 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=128976.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:07:39,438 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.000e+02 3.228e+02 3.806e+02 5.170e+02 8.621e+02, threshold=7.612e+02, percent-clipped=3.0 2023-04-29 20:08:12,085 INFO [train.py:904] (7/8) Epoch 13, batch 7200, loss[loss=0.193, simple_loss=0.2906, pruned_loss=0.0477, over 16376.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2999, pruned_loss=0.06682, over 3038330.36 frames. ], batch size: 165, lr: 5.21e-03, grad_scale: 8.0 2023-04-29 20:08:12,881 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-29 20:08:45,131 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129024.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:09:29,936 INFO [train.py:904] (7/8) Epoch 13, batch 7250, loss[loss=0.2122, simple_loss=0.2946, pruned_loss=0.06493, over 16442.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2974, pruned_loss=0.06541, over 3038713.06 frames. ], batch size: 146, lr: 5.21e-03, grad_scale: 8.0 2023-04-29 20:09:57,001 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129069.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:10:12,185 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 2.885e+02 3.444e+02 4.352e+02 7.294e+02, threshold=6.888e+02, percent-clipped=0.0 2023-04-29 20:10:45,724 INFO [train.py:904] (7/8) Epoch 13, batch 7300, loss[loss=0.2171, simple_loss=0.2926, pruned_loss=0.07084, over 16397.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2967, pruned_loss=0.06515, over 3054553.16 frames. ], batch size: 68, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:11:09,712 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129117.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:11:40,204 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129137.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:11:45,910 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6060, 3.9484, 3.5598, 3.7826, 3.4285, 3.6052, 3.5715, 3.8867], device='cuda:7'), covar=tensor([0.2389, 0.1405, 0.2456, 0.1530, 0.1884, 0.2501, 0.2032, 0.1848], device='cuda:7'), in_proj_covar=tensor([0.0567, 0.0699, 0.0574, 0.0499, 0.0441, 0.0452, 0.0585, 0.0537], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:12:02,359 INFO [train.py:904] (7/8) Epoch 13, batch 7350, loss[loss=0.1887, simple_loss=0.2739, pruned_loss=0.05172, over 17101.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2978, pruned_loss=0.06652, over 3034807.67 frames. ], batch size: 47, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:12:46,237 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.924e+02 3.417e+02 4.096e+02 1.600e+03, threshold=6.834e+02, percent-clipped=4.0 2023-04-29 20:13:21,047 INFO [train.py:904] (7/8) Epoch 13, batch 7400, loss[loss=0.1966, simple_loss=0.2877, pruned_loss=0.05275, over 17206.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2984, pruned_loss=0.06722, over 3039191.19 frames. ], batch size: 45, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:14:32,923 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129246.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 20:14:41,692 INFO [train.py:904] (7/8) Epoch 13, batch 7450, loss[loss=0.2315, simple_loss=0.3169, pruned_loss=0.07306, over 15321.00 frames. ], tot_loss[loss=0.2191, simple_loss=0.3, pruned_loss=0.06907, over 3010969.44 frames. ], batch size: 191, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:15:30,917 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.914e+02 3.121e+02 3.842e+02 4.432e+02 7.351e+02, threshold=7.685e+02, percent-clipped=1.0 2023-04-29 20:16:05,640 INFO [train.py:904] (7/8) Epoch 13, batch 7500, loss[loss=0.213, simple_loss=0.2943, pruned_loss=0.06583, over 16911.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3002, pruned_loss=0.06761, over 3033393.07 frames. ], batch size: 116, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:17:24,565 INFO [train.py:904] (7/8) Epoch 13, batch 7550, loss[loss=0.2104, simple_loss=0.2911, pruned_loss=0.06487, over 16399.00 frames. ], tot_loss[loss=0.217, simple_loss=0.2993, pruned_loss=0.06738, over 3049690.08 frames. ], batch size: 146, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:17:47,399 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3264, 3.2700, 3.3799, 3.4753, 3.5006, 3.2347, 3.4634, 3.5319], device='cuda:7'), covar=tensor([0.1152, 0.0892, 0.1019, 0.0588, 0.0645, 0.2668, 0.1046, 0.0815], device='cuda:7'), in_proj_covar=tensor([0.0537, 0.0666, 0.0797, 0.0681, 0.0518, 0.0528, 0.0546, 0.0631], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:18:07,708 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.820e+02 3.717e+02 4.924e+02 9.310e+02, threshold=7.434e+02, percent-clipped=3.0 2023-04-29 20:18:40,663 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8412, 4.8377, 5.3148, 5.2932, 5.2886, 4.9483, 4.9384, 4.6721], device='cuda:7'), covar=tensor([0.0275, 0.0493, 0.0266, 0.0329, 0.0369, 0.0297, 0.0888, 0.0421], device='cuda:7'), in_proj_covar=tensor([0.0347, 0.0364, 0.0366, 0.0349, 0.0414, 0.0387, 0.0483, 0.0311], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 20:18:41,436 INFO [train.py:904] (7/8) Epoch 13, batch 7600, loss[loss=0.1993, simple_loss=0.2824, pruned_loss=0.05813, over 16691.00 frames. ], tot_loss[loss=0.217, simple_loss=0.2989, pruned_loss=0.0676, over 3042535.32 frames. ], batch size: 57, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:18:59,789 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129413.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:19:11,185 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6328, 2.8606, 2.6300, 4.5806, 3.3627, 4.0964, 1.6259, 2.9901], device='cuda:7'), covar=tensor([0.1454, 0.0764, 0.1199, 0.0151, 0.0353, 0.0409, 0.1572, 0.0859], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0186, 0.0159, 0.0205, 0.0210, 0.0189, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 20:19:37,492 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129437.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:20:00,006 INFO [train.py:904] (7/8) Epoch 13, batch 7650, loss[loss=0.1867, simple_loss=0.2843, pruned_loss=0.04451, over 16803.00 frames. ], tot_loss[loss=0.2181, simple_loss=0.3, pruned_loss=0.06811, over 3058075.56 frames. ], batch size: 102, lr: 5.20e-03, grad_scale: 8.0 2023-04-29 20:20:36,641 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129474.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:20:39,979 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-29 20:20:44,166 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.211e+02 3.428e+02 4.160e+02 4.831e+02 1.059e+03, threshold=8.320e+02, percent-clipped=3.0 2023-04-29 20:20:47,665 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 20:20:53,104 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129485.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:21:18,577 INFO [train.py:904] (7/8) Epoch 13, batch 7700, loss[loss=0.1848, simple_loss=0.2807, pruned_loss=0.04442, over 16901.00 frames. ], tot_loss[loss=0.217, simple_loss=0.299, pruned_loss=0.06753, over 3083066.52 frames. ], batch size: 96, lr: 5.20e-03, grad_scale: 4.0 2023-04-29 20:22:27,013 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=129546.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:22:35,890 INFO [train.py:904] (7/8) Epoch 13, batch 7750, loss[loss=0.2489, simple_loss=0.3118, pruned_loss=0.09305, over 11708.00 frames. ], tot_loss[loss=0.2173, simple_loss=0.2993, pruned_loss=0.0676, over 3067313.85 frames. ], batch size: 246, lr: 5.20e-03, grad_scale: 4.0 2023-04-29 20:22:40,091 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7908, 2.2658, 1.7963, 2.0860, 2.6380, 2.2956, 2.6785, 2.8232], device='cuda:7'), covar=tensor([0.0134, 0.0323, 0.0414, 0.0377, 0.0210, 0.0307, 0.0158, 0.0191], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0207, 0.0200, 0.0200, 0.0205, 0.0204, 0.0208, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:22:40,131 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8352, 3.3021, 3.2481, 1.8382, 2.7043, 2.1223, 3.3398, 3.4914], device='cuda:7'), covar=tensor([0.0309, 0.0757, 0.0610, 0.2126, 0.0973, 0.1085, 0.0778, 0.0911], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0150, 0.0160, 0.0146, 0.0139, 0.0127, 0.0139, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 20:23:20,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 3.197e+02 3.787e+02 4.587e+02 8.661e+02, threshold=7.574e+02, percent-clipped=1.0 2023-04-29 20:23:27,567 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0204, 2.4589, 2.5982, 1.8272, 2.6551, 2.8185, 2.3990, 2.3728], device='cuda:7'), covar=tensor([0.0728, 0.0223, 0.0228, 0.1025, 0.0109, 0.0252, 0.0457, 0.0437], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0102, 0.0089, 0.0139, 0.0069, 0.0110, 0.0122, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 20:23:40,232 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=129594.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 20:23:52,188 INFO [train.py:904] (7/8) Epoch 13, batch 7800, loss[loss=0.2216, simple_loss=0.3069, pruned_loss=0.06818, over 17180.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3006, pruned_loss=0.06843, over 3076164.21 frames. ], batch size: 46, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:24:09,332 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129612.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:25:09,856 INFO [train.py:904] (7/8) Epoch 13, batch 7850, loss[loss=0.2476, simple_loss=0.3183, pruned_loss=0.08848, over 11778.00 frames. ], tot_loss[loss=0.2194, simple_loss=0.3014, pruned_loss=0.06869, over 3043822.97 frames. ], batch size: 247, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:25:15,726 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-29 20:25:16,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6015, 3.0901, 3.0598, 1.9096, 2.7255, 2.1216, 3.1690, 3.2438], device='cuda:7'), covar=tensor([0.0298, 0.0685, 0.0558, 0.1941, 0.0845, 0.0979, 0.0697, 0.0834], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0150, 0.0160, 0.0145, 0.0138, 0.0127, 0.0138, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 20:25:40,285 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129673.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:25:52,273 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 3.020e+02 3.492e+02 4.287e+02 1.158e+03, threshold=6.983e+02, percent-clipped=4.0 2023-04-29 20:26:12,477 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9899, 4.7643, 5.0302, 5.1917, 5.3953, 4.7065, 5.3489, 5.3334], device='cuda:7'), covar=tensor([0.1695, 0.1337, 0.1517, 0.0622, 0.0478, 0.0869, 0.0525, 0.0621], device='cuda:7'), in_proj_covar=tensor([0.0545, 0.0672, 0.0804, 0.0685, 0.0523, 0.0534, 0.0550, 0.0636], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:26:22,715 INFO [train.py:904] (7/8) Epoch 13, batch 7900, loss[loss=0.2236, simple_loss=0.2934, pruned_loss=0.07696, over 11485.00 frames. ], tot_loss[loss=0.2176, simple_loss=0.3, pruned_loss=0.06764, over 3073811.51 frames. ], batch size: 247, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:27:36,782 INFO [train.py:904] (7/8) Epoch 13, batch 7950, loss[loss=0.2391, simple_loss=0.3173, pruned_loss=0.08044, over 16461.00 frames. ], tot_loss[loss=0.219, simple_loss=0.3008, pruned_loss=0.06861, over 3059367.72 frames. ], batch size: 146, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:27:49,191 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0553, 2.0078, 2.1565, 3.4905, 2.0551, 2.3410, 2.1498, 2.1477], device='cuda:7'), covar=tensor([0.1114, 0.3293, 0.2374, 0.0563, 0.3988, 0.2260, 0.3093, 0.3043], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0403, 0.0335, 0.0316, 0.0417, 0.0462, 0.0368, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:28:02,341 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129769.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 20:28:11,954 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5155, 4.4788, 4.3634, 3.6968, 4.3755, 1.5436, 4.1666, 4.1472], device='cuda:7'), covar=tensor([0.0112, 0.0083, 0.0165, 0.0382, 0.0102, 0.2676, 0.0137, 0.0207], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0123, 0.0169, 0.0159, 0.0141, 0.0184, 0.0157, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:28:18,209 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.009e+02 2.840e+02 3.431e+02 4.053e+02 8.396e+02, threshold=6.863e+02, percent-clipped=2.0 2023-04-29 20:28:33,876 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=129791.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:28:49,439 INFO [train.py:904] (7/8) Epoch 13, batch 8000, loss[loss=0.2451, simple_loss=0.3144, pruned_loss=0.08791, over 11616.00 frames. ], tot_loss[loss=0.2188, simple_loss=0.3011, pruned_loss=0.06828, over 3078300.30 frames. ], batch size: 248, lr: 5.19e-03, grad_scale: 8.0 2023-04-29 20:29:36,575 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3475, 2.3314, 2.2949, 4.3138, 2.1830, 2.6837, 2.3498, 2.5779], device='cuda:7'), covar=tensor([0.1052, 0.3297, 0.2563, 0.0379, 0.3822, 0.2268, 0.3118, 0.3050], device='cuda:7'), in_proj_covar=tensor([0.0367, 0.0404, 0.0336, 0.0317, 0.0419, 0.0464, 0.0368, 0.0471], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:30:02,157 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-04-29 20:30:02,347 INFO [train.py:904] (7/8) Epoch 13, batch 8050, loss[loss=0.2208, simple_loss=0.3004, pruned_loss=0.07056, over 16568.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.3008, pruned_loss=0.06812, over 3077057.21 frames. ], batch size: 68, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:30:03,008 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=129852.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:30:45,835 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.353e+02 3.065e+02 3.748e+02 4.862e+02 1.232e+03, threshold=7.497e+02, percent-clipped=5.0 2023-04-29 20:31:15,205 INFO [train.py:904] (7/8) Epoch 13, batch 8100, loss[loss=0.2158, simple_loss=0.303, pruned_loss=0.06427, over 15197.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3005, pruned_loss=0.06791, over 3067926.99 frames. ], batch size: 190, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:32:19,756 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-29 20:32:29,510 INFO [train.py:904] (7/8) Epoch 13, batch 8150, loss[loss=0.1843, simple_loss=0.2699, pruned_loss=0.04932, over 16190.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2983, pruned_loss=0.06698, over 3073027.24 frames. ], batch size: 165, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:32:53,078 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=129968.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:33:14,449 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.827e+02 2.878e+02 3.437e+02 4.165e+02 9.532e+02, threshold=6.873e+02, percent-clipped=2.0 2023-04-29 20:33:46,586 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130000.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:33:48,631 INFO [train.py:904] (7/8) Epoch 13, batch 8200, loss[loss=0.2191, simple_loss=0.2852, pruned_loss=0.0765, over 11720.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2956, pruned_loss=0.06602, over 3088924.42 frames. ], batch size: 246, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:33:49,230 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130002.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:35:09,265 INFO [train.py:904] (7/8) Epoch 13, batch 8250, loss[loss=0.209, simple_loss=0.3007, pruned_loss=0.05864, over 15367.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2944, pruned_loss=0.06308, over 3090872.46 frames. ], batch size: 191, lr: 5.19e-03, grad_scale: 4.0 2023-04-29 20:35:24,057 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130061.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:35:27,829 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130063.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:35:37,702 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130069.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 20:35:57,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.705e+02 3.349e+02 4.037e+02 8.257e+02, threshold=6.697e+02, percent-clipped=3.0 2023-04-29 20:36:04,342 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1218, 3.8820, 4.2784, 2.1612, 4.4455, 4.4907, 3.3112, 3.4093], device='cuda:7'), covar=tensor([0.0535, 0.0174, 0.0147, 0.1056, 0.0042, 0.0081, 0.0302, 0.0351], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0100, 0.0088, 0.0137, 0.0068, 0.0108, 0.0119, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 20:36:29,986 INFO [train.py:904] (7/8) Epoch 13, batch 8300, loss[loss=0.1836, simple_loss=0.2648, pruned_loss=0.05117, over 12347.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2917, pruned_loss=0.06056, over 3063337.26 frames. ], batch size: 247, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:36:41,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3081, 4.3731, 4.1709, 3.9482, 3.8485, 4.3023, 4.0043, 4.0212], device='cuda:7'), covar=tensor([0.0555, 0.0597, 0.0266, 0.0246, 0.0805, 0.0438, 0.0636, 0.0606], device='cuda:7'), in_proj_covar=tensor([0.0245, 0.0336, 0.0294, 0.0272, 0.0307, 0.0315, 0.0200, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:36:55,892 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130117.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:37:24,461 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 20:37:43,849 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130147.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:37:51,997 INFO [train.py:904] (7/8) Epoch 13, batch 8350, loss[loss=0.2213, simple_loss=0.3138, pruned_loss=0.06445, over 16238.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2914, pruned_loss=0.05906, over 3062870.32 frames. ], batch size: 165, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:38:39,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.345e+02 2.793e+02 3.218e+02 5.548e+02, threshold=5.587e+02, percent-clipped=0.0 2023-04-29 20:39:12,415 INFO [train.py:904] (7/8) Epoch 13, batch 8400, loss[loss=0.1769, simple_loss=0.2599, pruned_loss=0.04699, over 12197.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2883, pruned_loss=0.0567, over 3041145.73 frames. ], batch size: 246, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:39:25,394 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-29 20:40:29,234 INFO [train.py:904] (7/8) Epoch 13, batch 8450, loss[loss=0.1602, simple_loss=0.2639, pruned_loss=0.02825, over 16936.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2863, pruned_loss=0.05482, over 3041704.54 frames. ], batch size: 90, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:40:55,959 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130268.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:41:17,434 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.422e+02 2.811e+02 3.625e+02 1.309e+03, threshold=5.622e+02, percent-clipped=6.0 2023-04-29 20:41:49,398 INFO [train.py:904] (7/8) Epoch 13, batch 8500, loss[loss=0.1672, simple_loss=0.2495, pruned_loss=0.04241, over 11977.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.2823, pruned_loss=0.05237, over 3038599.58 frames. ], batch size: 246, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:42:13,190 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130316.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:42:29,691 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7471, 1.2325, 1.6923, 1.6673, 1.7579, 1.7905, 1.5835, 1.7910], device='cuda:7'), covar=tensor([0.0203, 0.0325, 0.0166, 0.0209, 0.0229, 0.0180, 0.0307, 0.0098], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0173, 0.0156, 0.0157, 0.0170, 0.0126, 0.0172, 0.0117], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-29 20:43:10,838 INFO [train.py:904] (7/8) Epoch 13, batch 8550, loss[loss=0.1989, simple_loss=0.2982, pruned_loss=0.04978, over 16685.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2804, pruned_loss=0.05123, over 3033620.41 frames. ], batch size: 83, lr: 5.18e-03, grad_scale: 8.0 2023-04-29 20:43:13,771 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6552, 2.6370, 1.8566, 2.7874, 2.1281, 2.7998, 2.1286, 2.3889], device='cuda:7'), covar=tensor([0.0249, 0.0287, 0.1163, 0.0187, 0.0594, 0.0416, 0.1094, 0.0528], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0159, 0.0183, 0.0131, 0.0162, 0.0199, 0.0190, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 20:43:19,869 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:43:23,785 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130358.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:44:07,715 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.204e+02 2.928e+02 3.716e+02 7.366e+02, threshold=5.857e+02, percent-clipped=6.0 2023-04-29 20:44:12,757 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130384.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:44:12,998 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-29 20:44:50,610 INFO [train.py:904] (7/8) Epoch 13, batch 8600, loss[loss=0.183, simple_loss=0.2766, pruned_loss=0.04472, over 16767.00 frames. ], tot_loss[loss=0.1909, simple_loss=0.2808, pruned_loss=0.05049, over 3021169.30 frames. ], batch size: 76, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:44:58,501 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 20:45:51,770 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130432.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:45:59,616 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0450, 2.2567, 1.9238, 2.1305, 2.6137, 2.3978, 2.8101, 2.8858], device='cuda:7'), covar=tensor([0.0118, 0.0381, 0.0427, 0.0392, 0.0246, 0.0311, 0.0183, 0.0224], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0205, 0.0198, 0.0199, 0.0203, 0.0203, 0.0204, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:46:04,165 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130438.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:16,942 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130445.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:20,642 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130447.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:46:30,000 INFO [train.py:904] (7/8) Epoch 13, batch 8650, loss[loss=0.1926, simple_loss=0.2945, pruned_loss=0.04532, over 16574.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2781, pruned_loss=0.04873, over 2998691.36 frames. ], batch size: 62, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:47:09,193 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-29 20:47:40,992 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.301e+02 2.678e+02 3.280e+02 8.282e+02, threshold=5.356e+02, percent-clipped=3.0 2023-04-29 20:48:01,242 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130493.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:48:05,678 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130495.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:48:12,891 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130499.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:48:17,482 INFO [train.py:904] (7/8) Epoch 13, batch 8700, loss[loss=0.1702, simple_loss=0.262, pruned_loss=0.03924, over 16195.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2752, pruned_loss=0.04691, over 3029609.00 frames. ], batch size: 165, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:49:04,391 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6567, 2.6991, 2.2662, 3.8215, 2.4609, 3.7779, 1.4910, 2.7079], device='cuda:7'), covar=tensor([0.1451, 0.0675, 0.1309, 0.0180, 0.0150, 0.0402, 0.1654, 0.0821], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0162, 0.0182, 0.0154, 0.0198, 0.0205, 0.0185, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 20:49:47,048 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130548.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:49:54,316 INFO [train.py:904] (7/8) Epoch 13, batch 8750, loss[loss=0.1542, simple_loss=0.2424, pruned_loss=0.03302, over 12250.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2757, pruned_loss=0.04683, over 3028353.24 frames. ], batch size: 247, lr: 5.18e-03, grad_scale: 4.0 2023-04-29 20:50:30,316 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=130565.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:51:07,897 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.672e+02 2.544e+02 3.252e+02 4.020e+02 9.168e+02, threshold=6.504e+02, percent-clipped=9.0 2023-04-29 20:51:13,035 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1175, 2.6046, 2.6600, 1.9116, 2.8442, 2.9263, 2.5209, 2.5003], device='cuda:7'), covar=tensor([0.0627, 0.0191, 0.0207, 0.0892, 0.0070, 0.0157, 0.0374, 0.0375], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0096, 0.0084, 0.0132, 0.0065, 0.0103, 0.0115, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 20:51:48,217 INFO [train.py:904] (7/8) Epoch 13, batch 8800, loss[loss=0.1716, simple_loss=0.2688, pruned_loss=0.03718, over 16852.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2733, pruned_loss=0.04524, over 3021450.60 frames. ], batch size: 96, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:52:02,547 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130609.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:52:12,157 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-29 20:52:39,863 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=130626.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:53:32,468 INFO [train.py:904] (7/8) Epoch 13, batch 8850, loss[loss=0.1962, simple_loss=0.2974, pruned_loss=0.04749, over 16769.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2758, pruned_loss=0.04481, over 3027799.56 frames. ], batch size: 134, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:53:41,571 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130656.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:53:45,818 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=130658.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 20:53:54,539 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7560, 4.2841, 4.2572, 3.0776, 3.6624, 4.2363, 3.9709, 2.3173], device='cuda:7'), covar=tensor([0.0375, 0.0023, 0.0024, 0.0276, 0.0072, 0.0052, 0.0048, 0.0396], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0068, 0.0071, 0.0127, 0.0083, 0.0091, 0.0080, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 20:54:37,531 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.288e+02 2.852e+02 3.513e+02 7.607e+02, threshold=5.705e+02, percent-clipped=2.0 2023-04-29 20:55:17,114 INFO [train.py:904] (7/8) Epoch 13, batch 8900, loss[loss=0.1722, simple_loss=0.2611, pruned_loss=0.04168, over 16563.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2763, pruned_loss=0.04425, over 3042391.57 frames. ], batch size: 62, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:55:22,814 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:55:26,797 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=130706.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:56:54,232 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130740.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:56:56,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0906, 3.9173, 4.1601, 4.2378, 4.3803, 3.9378, 4.3779, 4.3851], device='cuda:7'), covar=tensor([0.1373, 0.1033, 0.1349, 0.0695, 0.0552, 0.1196, 0.0504, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0522, 0.0645, 0.0769, 0.0662, 0.0501, 0.0507, 0.0526, 0.0609], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 20:57:21,729 INFO [train.py:904] (7/8) Epoch 13, batch 8950, loss[loss=0.1739, simple_loss=0.2602, pruned_loss=0.04379, over 16797.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2755, pruned_loss=0.04437, over 3048882.30 frames. ], batch size: 124, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:58:29,179 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-04-29 20:58:29,465 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.201e+02 2.684e+02 3.167e+02 7.959e+02, threshold=5.368e+02, percent-clipped=1.0 2023-04-29 20:58:41,713 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130788.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:58:54,891 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130794.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 20:59:11,372 INFO [train.py:904] (7/8) Epoch 13, batch 9000, loss[loss=0.1464, simple_loss=0.241, pruned_loss=0.02589, over 17273.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.272, pruned_loss=0.04305, over 3066338.36 frames. ], batch size: 52, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 20:59:11,372 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 20:59:22,054 INFO [train.py:938] (7/8) Epoch 13, validation: loss=0.1517, simple_loss=0.2561, pruned_loss=0.02371, over 944034.00 frames. 2023-04-29 20:59:22,055 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 21:01:06,021 INFO [train.py:904] (7/8) Epoch 13, batch 9050, loss[loss=0.1902, simple_loss=0.2764, pruned_loss=0.05201, over 16790.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2734, pruned_loss=0.04384, over 3084498.00 frames. ], batch size: 124, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:02:07,081 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.403e+02 2.798e+02 3.395e+02 5.131e+02, threshold=5.596e+02, percent-clipped=0.0 2023-04-29 21:02:14,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1912, 3.2709, 1.8841, 3.5079, 2.4058, 3.4945, 2.1257, 2.6891], device='cuda:7'), covar=tensor([0.0237, 0.0327, 0.1553, 0.0198, 0.0799, 0.0497, 0.1460, 0.0689], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0160, 0.0185, 0.0129, 0.0163, 0.0198, 0.0191, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 21:02:52,490 INFO [train.py:904] (7/8) Epoch 13, batch 9100, loss[loss=0.1752, simple_loss=0.2788, pruned_loss=0.03581, over 16676.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.273, pruned_loss=0.04416, over 3070060.44 frames. ], batch size: 76, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:02:58,161 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130904.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:03:33,382 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=130921.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:04:49,582 INFO [train.py:904] (7/8) Epoch 13, batch 9150, loss[loss=0.1746, simple_loss=0.2667, pruned_loss=0.04126, over 16637.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2729, pruned_loss=0.04365, over 3062003.45 frames. ], batch size: 134, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:05:52,923 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.726e+02 2.558e+02 2.960e+02 3.606e+02 6.781e+02, threshold=5.920e+02, percent-clipped=4.0 2023-04-29 21:06:31,924 INFO [train.py:904] (7/8) Epoch 13, batch 9200, loss[loss=0.1776, simple_loss=0.2653, pruned_loss=0.04499, over 16654.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.269, pruned_loss=0.04288, over 3061112.49 frames. ], batch size: 62, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:06:33,939 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4074, 3.3738, 3.4808, 3.5601, 3.6009, 3.3335, 3.5887, 3.6406], device='cuda:7'), covar=tensor([0.1120, 0.0839, 0.0940, 0.0606, 0.0556, 0.2083, 0.0804, 0.0696], device='cuda:7'), in_proj_covar=tensor([0.0526, 0.0646, 0.0772, 0.0663, 0.0503, 0.0509, 0.0529, 0.0611], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:07:31,107 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5794, 2.6604, 1.8175, 2.8242, 2.1660, 2.7633, 2.1259, 2.3577], device='cuda:7'), covar=tensor([0.0261, 0.0305, 0.1238, 0.0272, 0.0615, 0.0541, 0.1062, 0.0584], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0159, 0.0183, 0.0128, 0.0162, 0.0197, 0.0190, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 21:07:44,240 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131040.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:08:07,057 INFO [train.py:904] (7/8) Epoch 13, batch 9250, loss[loss=0.1742, simple_loss=0.275, pruned_loss=0.03671, over 16867.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2685, pruned_loss=0.04299, over 3033837.61 frames. ], batch size: 102, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:08:25,911 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-04-29 21:09:00,295 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-29 21:09:12,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.311e+02 2.375e+02 2.887e+02 3.471e+02 9.563e+02, threshold=5.774e+02, percent-clipped=3.0 2023-04-29 21:09:24,456 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:09:24,525 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131088.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:09:40,196 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131094.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:09:55,051 INFO [train.py:904] (7/8) Epoch 13, batch 9300, loss[loss=0.1724, simple_loss=0.251, pruned_loss=0.04686, over 11961.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2666, pruned_loss=0.04227, over 3015769.27 frames. ], batch size: 247, lr: 5.17e-03, grad_scale: 8.0 2023-04-29 21:10:00,935 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 21:10:29,877 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6871, 2.8099, 2.3492, 4.2262, 2.7316, 3.9829, 1.3566, 2.9922], device='cuda:7'), covar=tensor([0.1454, 0.0745, 0.1308, 0.0187, 0.0162, 0.0408, 0.1786, 0.0751], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0162, 0.0184, 0.0154, 0.0193, 0.0206, 0.0187, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 21:10:40,402 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6739, 4.0146, 3.0212, 2.2282, 2.6118, 2.4364, 4.3578, 3.5364], device='cuda:7'), covar=tensor([0.2913, 0.0646, 0.1683, 0.2607, 0.2427, 0.1899, 0.0346, 0.1039], device='cuda:7'), in_proj_covar=tensor([0.0302, 0.0252, 0.0284, 0.0280, 0.0265, 0.0225, 0.0266, 0.0294], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:11:11,224 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131136.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:11:24,003 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131142.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:11:40,558 INFO [train.py:904] (7/8) Epoch 13, batch 9350, loss[loss=0.1825, simple_loss=0.2708, pruned_loss=0.04709, over 16360.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2664, pruned_loss=0.04217, over 3028889.30 frames. ], batch size: 146, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:12:25,580 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131174.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:12:41,636 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.118e+02 2.389e+02 2.842e+02 4.349e+02, threshold=4.777e+02, percent-clipped=0.0 2023-04-29 21:13:20,705 INFO [train.py:904] (7/8) Epoch 13, batch 9400, loss[loss=0.1895, simple_loss=0.2782, pruned_loss=0.05039, over 16459.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2665, pruned_loss=0.04211, over 3020742.94 frames. ], batch size: 62, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:13:25,622 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131204.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:13:59,480 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131221.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:14:27,987 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131235.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:14:59,721 INFO [train.py:904] (7/8) Epoch 13, batch 9450, loss[loss=0.1672, simple_loss=0.2634, pruned_loss=0.03544, over 16272.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2682, pruned_loss=0.04202, over 3024156.65 frames. ], batch size: 165, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:15:00,865 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131252.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:15:34,999 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131269.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:15:42,273 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-04-29 21:16:03,422 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.757e+02 2.351e+02 2.704e+02 3.479e+02 5.453e+02, threshold=5.408e+02, percent-clipped=2.0 2023-04-29 21:16:19,283 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-29 21:16:40,124 INFO [train.py:904] (7/8) Epoch 13, batch 9500, loss[loss=0.1579, simple_loss=0.2536, pruned_loss=0.03109, over 16784.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2676, pruned_loss=0.04159, over 3036148.73 frames. ], batch size: 83, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:17:14,028 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-04-29 21:18:25,263 INFO [train.py:904] (7/8) Epoch 13, batch 9550, loss[loss=0.1867, simple_loss=0.2778, pruned_loss=0.04774, over 16856.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2678, pruned_loss=0.04184, over 3065476.79 frames. ], batch size: 116, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:19:29,580 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.317e+02 2.711e+02 3.253e+02 6.482e+02, threshold=5.422e+02, percent-clipped=4.0 2023-04-29 21:20:03,684 INFO [train.py:904] (7/8) Epoch 13, batch 9600, loss[loss=0.1727, simple_loss=0.2604, pruned_loss=0.0425, over 12665.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2689, pruned_loss=0.04243, over 3052492.04 frames. ], batch size: 247, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:20:18,018 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131409.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:20:30,383 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131416.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:21:12,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5888, 4.9179, 4.7280, 4.7552, 4.4025, 4.3304, 4.3675, 5.0079], device='cuda:7'), covar=tensor([0.1043, 0.0944, 0.1086, 0.0679, 0.0766, 0.1213, 0.1120, 0.0875], device='cuda:7'), in_proj_covar=tensor([0.0547, 0.0686, 0.0549, 0.0486, 0.0428, 0.0440, 0.0569, 0.0521], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:21:49,845 INFO [train.py:904] (7/8) Epoch 13, batch 9650, loss[loss=0.1952, simple_loss=0.2855, pruned_loss=0.05249, over 15440.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2709, pruned_loss=0.04282, over 3053822.63 frames. ], batch size: 191, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:22:34,305 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131470.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:22:48,216 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131477.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:22:58,036 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.434e+02 2.934e+02 3.737e+02 7.668e+02, threshold=5.867e+02, percent-clipped=4.0 2023-04-29 21:23:16,247 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-29 21:23:18,234 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131492.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:23:38,334 INFO [train.py:904] (7/8) Epoch 13, batch 9700, loss[loss=0.1783, simple_loss=0.2652, pruned_loss=0.04571, over 12102.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2696, pruned_loss=0.04233, over 3061828.52 frames. ], batch size: 248, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:23:49,505 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1250, 4.1532, 4.4924, 2.2990, 4.6763, 4.7852, 3.3319, 3.4655], device='cuda:7'), covar=tensor([0.0637, 0.0169, 0.0163, 0.0961, 0.0041, 0.0066, 0.0310, 0.0351], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0096, 0.0083, 0.0132, 0.0065, 0.0103, 0.0116, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 21:24:14,138 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-04-29 21:24:36,247 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131530.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 21:25:20,240 INFO [train.py:904] (7/8) Epoch 13, batch 9750, loss[loss=0.1743, simple_loss=0.2534, pruned_loss=0.04761, over 11891.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2688, pruned_loss=0.04263, over 3054282.60 frames. ], batch size: 247, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:25:22,378 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131553.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:26:22,849 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.333e+02 2.960e+02 3.674e+02 6.318e+02, threshold=5.921e+02, percent-clipped=2.0 2023-04-29 21:26:57,820 INFO [train.py:904] (7/8) Epoch 13, batch 9800, loss[loss=0.1741, simple_loss=0.2771, pruned_loss=0.03552, over 16346.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2689, pruned_loss=0.04162, over 3071652.01 frames. ], batch size: 146, lr: 5.16e-03, grad_scale: 8.0 2023-04-29 21:28:39,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0289, 3.0763, 3.1640, 2.2483, 2.8729, 3.2379, 3.1413, 1.8842], device='cuda:7'), covar=tensor([0.0423, 0.0045, 0.0043, 0.0296, 0.0104, 0.0069, 0.0066, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0067, 0.0069, 0.0125, 0.0081, 0.0088, 0.0078, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 21:28:39,807 INFO [train.py:904] (7/8) Epoch 13, batch 9850, loss[loss=0.1693, simple_loss=0.2642, pruned_loss=0.03722, over 15408.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2695, pruned_loss=0.04123, over 3044673.37 frames. ], batch size: 191, lr: 5.15e-03, grad_scale: 4.0 2023-04-29 21:28:46,630 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5899, 3.6266, 3.4307, 3.1286, 3.2477, 3.5487, 3.3211, 3.3424], device='cuda:7'), covar=tensor([0.0504, 0.0423, 0.0252, 0.0217, 0.0445, 0.0382, 0.1212, 0.0396], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0320, 0.0286, 0.0263, 0.0293, 0.0306, 0.0193, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-29 21:29:46,605 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.305e+02 2.925e+02 3.439e+02 6.228e+02, threshold=5.850e+02, percent-clipped=2.0 2023-04-29 21:30:18,631 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-29 21:30:20,061 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8545, 5.1496, 4.9758, 4.9803, 4.6901, 4.6448, 4.6062, 5.2173], device='cuda:7'), covar=tensor([0.1037, 0.0830, 0.0791, 0.0623, 0.0706, 0.0871, 0.0943, 0.0867], device='cuda:7'), in_proj_covar=tensor([0.0548, 0.0688, 0.0552, 0.0486, 0.0431, 0.0441, 0.0569, 0.0524], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:30:32,569 INFO [train.py:904] (7/8) Epoch 13, batch 9900, loss[loss=0.1675, simple_loss=0.2733, pruned_loss=0.03083, over 16724.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2701, pruned_loss=0.04137, over 3050518.04 frames. ], batch size: 83, lr: 5.15e-03, grad_scale: 4.0 2023-04-29 21:32:30,680 INFO [train.py:904] (7/8) Epoch 13, batch 9950, loss[loss=0.1617, simple_loss=0.261, pruned_loss=0.03122, over 16698.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2719, pruned_loss=0.04164, over 3049418.55 frames. ], batch size: 76, lr: 5.15e-03, grad_scale: 4.0 2023-04-29 21:33:02,018 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131765.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:33:02,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4623, 4.4606, 4.2875, 3.8315, 4.3533, 1.6518, 4.0933, 4.1199], device='cuda:7'), covar=tensor([0.0088, 0.0084, 0.0160, 0.0248, 0.0089, 0.2387, 0.0123, 0.0198], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0119, 0.0160, 0.0147, 0.0135, 0.0181, 0.0152, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:33:20,318 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131772.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:33:47,206 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.347e+02 2.831e+02 3.741e+02 5.862e+02, threshold=5.662e+02, percent-clipped=1.0 2023-04-29 21:34:31,157 INFO [train.py:904] (7/8) Epoch 13, batch 10000, loss[loss=0.1653, simple_loss=0.2638, pruned_loss=0.03339, over 16757.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2709, pruned_loss=0.0414, over 3068791.99 frames. ], batch size: 83, lr: 5.15e-03, grad_scale: 8.0 2023-04-29 21:35:27,710 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=131830.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:36:00,319 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 21:36:06,300 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=131848.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:36:11,642 INFO [train.py:904] (7/8) Epoch 13, batch 10050, loss[loss=0.1939, simple_loss=0.2859, pruned_loss=0.051, over 16969.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.271, pruned_loss=0.04138, over 3077648.25 frames. ], batch size: 109, lr: 5.15e-03, grad_scale: 8.0 2023-04-29 21:36:23,301 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=131857.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:37:04,852 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=131878.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:37:11,055 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-29 21:37:14,091 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.136e+02 2.458e+02 2.977e+02 5.348e+02, threshold=4.916e+02, percent-clipped=0.0 2023-04-29 21:37:45,833 INFO [train.py:904] (7/8) Epoch 13, batch 10100, loss[loss=0.1625, simple_loss=0.254, pruned_loss=0.03552, over 16146.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2714, pruned_loss=0.0416, over 3083073.44 frames. ], batch size: 165, lr: 5.15e-03, grad_scale: 8.0 2023-04-29 21:38:16,887 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=131918.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:39:29,405 INFO [train.py:904] (7/8) Epoch 14, batch 0, loss[loss=0.1768, simple_loss=0.2587, pruned_loss=0.04746, over 16737.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2587, pruned_loss=0.04746, over 16737.00 frames. ], batch size: 39, lr: 4.96e-03, grad_scale: 8.0 2023-04-29 21:39:29,406 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 21:39:36,900 INFO [train.py:938] (7/8) Epoch 14, validation: loss=0.1515, simple_loss=0.2551, pruned_loss=0.024, over 944034.00 frames. 2023-04-29 21:39:36,901 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 21:40:03,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6189, 3.6280, 3.9360, 2.9486, 3.5751, 3.9980, 3.7143, 2.3069], device='cuda:7'), covar=tensor([0.0413, 0.0220, 0.0030, 0.0279, 0.0097, 0.0077, 0.0077, 0.0430], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0069, 0.0070, 0.0127, 0.0082, 0.0089, 0.0080, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 21:40:04,687 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8124, 3.7925, 2.0077, 4.2232, 2.7801, 4.1464, 2.1782, 3.0912], device='cuda:7'), covar=tensor([0.0193, 0.0306, 0.1721, 0.0199, 0.0782, 0.0477, 0.1650, 0.0628], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0159, 0.0185, 0.0130, 0.0164, 0.0196, 0.0194, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-29 21:40:22,375 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.510e+02 3.083e+02 4.151e+02 9.991e+02, threshold=6.165e+02, percent-clipped=13.0 2023-04-29 21:40:46,681 INFO [train.py:904] (7/8) Epoch 14, batch 50, loss[loss=0.1753, simple_loss=0.2709, pruned_loss=0.03985, over 17116.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2805, pruned_loss=0.05781, over 746378.68 frames. ], batch size: 47, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:41:27,241 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132031.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:41:55,089 INFO [train.py:904] (7/8) Epoch 14, batch 100, loss[loss=0.1917, simple_loss=0.2804, pruned_loss=0.05153, over 17062.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2775, pruned_loss=0.05582, over 1309519.20 frames. ], batch size: 55, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:41:58,534 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132054.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:42:14,147 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132065.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:42:24,479 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132072.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:42:30,142 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-29 21:42:44,832 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.577e+02 2.362e+02 2.775e+02 3.378e+02 5.633e+02, threshold=5.550e+02, percent-clipped=0.0 2023-04-29 21:42:51,131 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132092.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:43:04,256 INFO [train.py:904] (7/8) Epoch 14, batch 150, loss[loss=0.1889, simple_loss=0.2869, pruned_loss=0.04542, over 17262.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2752, pruned_loss=0.05385, over 1763880.81 frames. ], batch size: 52, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:43:21,433 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132113.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:43:24,101 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132115.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 21:43:30,635 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132120.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:44:09,340 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132148.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:44:15,119 INFO [train.py:904] (7/8) Epoch 14, batch 200, loss[loss=0.175, simple_loss=0.2685, pruned_loss=0.04079, over 17146.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2759, pruned_loss=0.05389, over 2104019.15 frames. ], batch size: 48, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:45:02,151 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.381e+02 2.781e+02 3.514e+02 5.258e+02, threshold=5.562e+02, percent-clipped=0.0 2023-04-29 21:45:15,519 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132196.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:45:23,108 INFO [train.py:904] (7/8) Epoch 14, batch 250, loss[loss=0.1988, simple_loss=0.258, pruned_loss=0.06981, over 16786.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2742, pruned_loss=0.05508, over 2362896.42 frames. ], batch size: 83, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:45:38,859 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132213.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:46:28,024 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0613, 3.9709, 4.4482, 2.1406, 4.5620, 4.6711, 3.2251, 3.4746], device='cuda:7'), covar=tensor([0.0651, 0.0196, 0.0132, 0.1092, 0.0055, 0.0127, 0.0380, 0.0362], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0101, 0.0086, 0.0137, 0.0068, 0.0109, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 21:46:31,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8831, 2.3247, 2.3925, 4.8232, 2.3148, 2.8095, 2.4293, 2.6002], device='cuda:7'), covar=tensor([0.0904, 0.3619, 0.2483, 0.0300, 0.3876, 0.2373, 0.3177, 0.3176], device='cuda:7'), in_proj_covar=tensor([0.0368, 0.0401, 0.0338, 0.0316, 0.0415, 0.0458, 0.0366, 0.0468], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:46:33,334 INFO [train.py:904] (7/8) Epoch 14, batch 300, loss[loss=0.1753, simple_loss=0.2506, pruned_loss=0.05, over 16446.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2699, pruned_loss=0.05239, over 2572648.92 frames. ], batch size: 75, lr: 4.96e-03, grad_scale: 1.0 2023-04-29 21:47:22,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.346e+02 2.777e+02 3.394e+02 6.651e+02, threshold=5.554e+02, percent-clipped=2.0 2023-04-29 21:47:43,467 INFO [train.py:904] (7/8) Epoch 14, batch 350, loss[loss=0.182, simple_loss=0.2538, pruned_loss=0.05512, over 16792.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2672, pruned_loss=0.05108, over 2729918.43 frames. ], batch size: 102, lr: 4.95e-03, grad_scale: 1.0 2023-04-29 21:48:48,088 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-29 21:48:51,152 INFO [train.py:904] (7/8) Epoch 14, batch 400, loss[loss=0.2293, simple_loss=0.2977, pruned_loss=0.08043, over 15592.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.265, pruned_loss=0.05077, over 2850590.68 frames. ], batch size: 191, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:49:38,428 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.217e+02 2.653e+02 3.223e+02 5.439e+02, threshold=5.306e+02, percent-clipped=0.0 2023-04-29 21:49:39,470 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132387.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:50:00,134 INFO [train.py:904] (7/8) Epoch 14, batch 450, loss[loss=0.1622, simple_loss=0.2446, pruned_loss=0.03986, over 17233.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2627, pruned_loss=0.04969, over 2942237.15 frames. ], batch size: 44, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:50:12,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3288, 5.7328, 5.4572, 5.5465, 5.1126, 5.0792, 5.1529, 5.8617], device='cuda:7'), covar=tensor([0.1322, 0.1062, 0.1123, 0.0719, 0.0900, 0.0792, 0.1084, 0.1052], device='cuda:7'), in_proj_covar=tensor([0.0589, 0.0736, 0.0597, 0.0521, 0.0467, 0.0469, 0.0615, 0.0561], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:50:12,158 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=132410.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 21:51:11,610 INFO [train.py:904] (7/8) Epoch 14, batch 500, loss[loss=0.1986, simple_loss=0.2702, pruned_loss=0.06349, over 16722.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2621, pruned_loss=0.04921, over 3016616.73 frames. ], batch size: 124, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:51:15,675 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4541, 3.3031, 2.6634, 2.1216, 2.2289, 2.2366, 3.3873, 3.0776], device='cuda:7'), covar=tensor([0.2705, 0.0713, 0.1603, 0.2464, 0.2487, 0.2006, 0.0496, 0.1313], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0257, 0.0288, 0.0283, 0.0273, 0.0229, 0.0271, 0.0303], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:51:58,584 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.377e+02 2.811e+02 3.534e+02 1.282e+03, threshold=5.622e+02, percent-clipped=6.0 2023-04-29 21:52:19,306 INFO [train.py:904] (7/8) Epoch 14, batch 550, loss[loss=0.1875, simple_loss=0.2714, pruned_loss=0.05183, over 16471.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.261, pruned_loss=0.04823, over 3087350.91 frames. ], batch size: 68, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:52:34,414 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132513.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:53:28,147 INFO [train.py:904] (7/8) Epoch 14, batch 600, loss[loss=0.1773, simple_loss=0.2723, pruned_loss=0.04109, over 17023.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2602, pruned_loss=0.04832, over 3142909.58 frames. ], batch size: 55, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:53:41,854 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132561.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:54:12,435 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-29 21:54:17,822 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.200e+02 2.653e+02 3.173e+02 6.384e+02, threshold=5.306e+02, percent-clipped=1.0 2023-04-29 21:54:39,164 INFO [train.py:904] (7/8) Epoch 14, batch 650, loss[loss=0.1748, simple_loss=0.2554, pruned_loss=0.04709, over 16866.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2592, pruned_loss=0.04817, over 3187161.33 frames. ], batch size: 116, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:55:24,705 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2936, 5.9202, 5.9840, 5.7570, 5.7863, 6.3367, 5.8607, 5.5700], device='cuda:7'), covar=tensor([0.0814, 0.1726, 0.2171, 0.1789, 0.2750, 0.0948, 0.1488, 0.2655], device='cuda:7'), in_proj_covar=tensor([0.0370, 0.0522, 0.0575, 0.0443, 0.0602, 0.0598, 0.0452, 0.0594], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 21:55:49,509 INFO [train.py:904] (7/8) Epoch 14, batch 700, loss[loss=0.1522, simple_loss=0.246, pruned_loss=0.02919, over 17223.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2594, pruned_loss=0.04788, over 3209110.11 frames. ], batch size: 44, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:56:01,382 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132660.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:56:38,038 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.642e+02 2.258e+02 2.690e+02 3.168e+02 6.172e+02, threshold=5.381e+02, percent-clipped=1.0 2023-04-29 21:56:39,018 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132687.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:56:41,421 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4567, 4.3913, 4.4776, 3.8846, 4.4260, 1.6886, 4.1705, 4.1332], device='cuda:7'), covar=tensor([0.0121, 0.0106, 0.0142, 0.0340, 0.0098, 0.2632, 0.0148, 0.0212], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0130, 0.0174, 0.0161, 0.0146, 0.0192, 0.0165, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:56:58,545 INFO [train.py:904] (7/8) Epoch 14, batch 750, loss[loss=0.1639, simple_loss=0.2476, pruned_loss=0.04013, over 17231.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2592, pruned_loss=0.04789, over 3239949.46 frames. ], batch size: 44, lr: 4.95e-03, grad_scale: 2.0 2023-04-29 21:57:09,177 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=132710.0, num_to_drop=1, layers_to_drop={2} 2023-04-29 21:57:22,240 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7451, 4.3518, 3.1006, 2.2480, 2.6965, 2.5049, 4.6544, 3.6989], device='cuda:7'), covar=tensor([0.2672, 0.0541, 0.1621, 0.2645, 0.2736, 0.1851, 0.0350, 0.1079], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0259, 0.0291, 0.0286, 0.0277, 0.0231, 0.0274, 0.0306], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 21:57:24,380 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132721.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:57:43,962 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132735.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:57:59,772 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=132746.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:58:07,429 INFO [train.py:904] (7/8) Epoch 14, batch 800, loss[loss=0.1947, simple_loss=0.2725, pruned_loss=0.05844, over 17009.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2593, pruned_loss=0.04718, over 3264846.96 frames. ], batch size: 55, lr: 4.95e-03, grad_scale: 4.0 2023-04-29 21:58:16,067 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=132758.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 21:58:27,274 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2023-04-29 21:58:54,911 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.382e+02 2.716e+02 3.142e+02 6.467e+02, threshold=5.432e+02, percent-clipped=2.0 2023-04-29 21:59:15,846 INFO [train.py:904] (7/8) Epoch 14, batch 850, loss[loss=0.1538, simple_loss=0.2373, pruned_loss=0.03517, over 15564.00 frames. ], tot_loss[loss=0.176, simple_loss=0.259, pruned_loss=0.04653, over 3275435.81 frames. ], batch size: 191, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 21:59:16,842 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 21:59:23,600 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=132807.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:00:24,479 INFO [train.py:904] (7/8) Epoch 14, batch 900, loss[loss=0.1581, simple_loss=0.252, pruned_loss=0.03213, over 17134.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2586, pruned_loss=0.04608, over 3293095.17 frames. ], batch size: 47, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:01:14,373 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.740e+02 2.299e+02 2.623e+02 3.116e+02 6.405e+02, threshold=5.245e+02, percent-clipped=2.0 2023-04-29 22:01:34,140 INFO [train.py:904] (7/8) Epoch 14, batch 950, loss[loss=0.1672, simple_loss=0.2436, pruned_loss=0.0454, over 16267.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2579, pruned_loss=0.04574, over 3288890.55 frames. ], batch size: 165, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:02:41,757 INFO [train.py:904] (7/8) Epoch 14, batch 1000, loss[loss=0.1596, simple_loss=0.2477, pruned_loss=0.0358, over 17202.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2568, pruned_loss=0.04604, over 3300923.86 frames. ], batch size: 46, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:03:29,372 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.118e+02 2.524e+02 3.195e+02 6.689e+02, threshold=5.047e+02, percent-clipped=1.0 2023-04-29 22:03:50,001 INFO [train.py:904] (7/8) Epoch 14, batch 1050, loss[loss=0.1842, simple_loss=0.2791, pruned_loss=0.0446, over 17268.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2559, pruned_loss=0.04583, over 3303382.21 frames. ], batch size: 52, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:04:10,799 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133016.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:04:59,797 INFO [train.py:904] (7/8) Epoch 14, batch 1100, loss[loss=0.1616, simple_loss=0.2564, pruned_loss=0.03343, over 17253.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2559, pruned_loss=0.04528, over 3313179.82 frames. ], batch size: 52, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:05:39,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3297, 3.9019, 3.8648, 2.1521, 3.0173, 2.5998, 3.5515, 3.9472], device='cuda:7'), covar=tensor([0.0349, 0.0742, 0.0461, 0.1809, 0.0884, 0.0878, 0.0859, 0.1031], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0147, 0.0157, 0.0145, 0.0137, 0.0125, 0.0135, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:05:47,409 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.252e+02 2.587e+02 3.047e+02 6.829e+02, threshold=5.174e+02, percent-clipped=2.0 2023-04-29 22:06:05,387 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6529, 3.6925, 2.1631, 4.0421, 2.7755, 4.0035, 2.4595, 3.0804], device='cuda:7'), covar=tensor([0.0251, 0.0392, 0.1563, 0.0265, 0.0757, 0.0590, 0.1296, 0.0616], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0192, 0.0145, 0.0170, 0.0211, 0.0200, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:06:08,355 INFO [train.py:904] (7/8) Epoch 14, batch 1150, loss[loss=0.1654, simple_loss=0.2344, pruned_loss=0.04817, over 16825.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2557, pruned_loss=0.04518, over 3314956.48 frames. ], batch size: 96, lr: 4.94e-03, grad_scale: 4.0 2023-04-29 22:06:08,680 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133102.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:06:35,924 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2686, 5.7388, 5.8415, 5.5976, 5.7638, 6.2653, 5.7821, 5.4892], device='cuda:7'), covar=tensor([0.0823, 0.1811, 0.1958, 0.2084, 0.2590, 0.0886, 0.1390, 0.2320], device='cuda:7'), in_proj_covar=tensor([0.0372, 0.0528, 0.0580, 0.0450, 0.0611, 0.0603, 0.0459, 0.0600], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:07:16,507 INFO [train.py:904] (7/8) Epoch 14, batch 1200, loss[loss=0.1893, simple_loss=0.2804, pruned_loss=0.04914, over 16648.00 frames. ], tot_loss[loss=0.173, simple_loss=0.256, pruned_loss=0.04495, over 3308945.38 frames. ], batch size: 62, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:07:35,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5034, 3.6209, 2.1591, 3.7837, 2.7549, 3.7747, 2.3708, 2.8968], device='cuda:7'), covar=tensor([0.0249, 0.0340, 0.1455, 0.0345, 0.0726, 0.0616, 0.1288, 0.0630], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0169, 0.0191, 0.0145, 0.0170, 0.0210, 0.0199, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:07:48,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8718, 2.9598, 2.5371, 2.7900, 3.2568, 3.0907, 3.7110, 3.4852], device='cuda:7'), covar=tensor([0.0103, 0.0311, 0.0394, 0.0326, 0.0211, 0.0282, 0.0165, 0.0192], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0219, 0.0212, 0.0211, 0.0218, 0.0220, 0.0224, 0.0211], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:08:05,984 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.506e+02 2.966e+02 3.654e+02 5.614e+02, threshold=5.932e+02, percent-clipped=4.0 2023-04-29 22:08:28,193 INFO [train.py:904] (7/8) Epoch 14, batch 1250, loss[loss=0.1754, simple_loss=0.2557, pruned_loss=0.04756, over 12359.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2559, pruned_loss=0.04537, over 3295824.53 frames. ], batch size: 247, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:09:37,986 INFO [train.py:904] (7/8) Epoch 14, batch 1300, loss[loss=0.1993, simple_loss=0.2693, pruned_loss=0.06462, over 16732.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2558, pruned_loss=0.04542, over 3292909.57 frames. ], batch size: 83, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:10:27,128 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.451e+02 2.160e+02 2.631e+02 3.090e+02 7.401e+02, threshold=5.263e+02, percent-clipped=1.0 2023-04-29 22:10:48,345 INFO [train.py:904] (7/8) Epoch 14, batch 1350, loss[loss=0.1622, simple_loss=0.2552, pruned_loss=0.03461, over 17116.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2559, pruned_loss=0.04511, over 3306214.90 frames. ], batch size: 47, lr: 4.94e-03, grad_scale: 8.0 2023-04-29 22:11:06,818 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133316.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:11:21,400 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6018, 3.4549, 3.7810, 1.6642, 3.8641, 3.8693, 2.9639, 2.8130], device='cuda:7'), covar=tensor([0.0726, 0.0185, 0.0155, 0.1267, 0.0075, 0.0149, 0.0404, 0.0452], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0102, 0.0088, 0.0136, 0.0070, 0.0112, 0.0121, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 22:11:56,462 INFO [train.py:904] (7/8) Epoch 14, batch 1400, loss[loss=0.1984, simple_loss=0.2685, pruned_loss=0.06421, over 16921.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2559, pruned_loss=0.04532, over 3308289.97 frames. ], batch size: 109, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:12:13,483 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=133364.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:12:26,447 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133374.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 22:12:30,111 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133376.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:12:44,805 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.235e+02 2.580e+02 3.131e+02 5.456e+02, threshold=5.161e+02, percent-clipped=1.0 2023-04-29 22:12:56,544 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0892, 2.0821, 2.2554, 3.5273, 2.0968, 2.3710, 2.2561, 2.2078], device='cuda:7'), covar=tensor([0.1217, 0.3196, 0.2524, 0.0617, 0.3719, 0.2244, 0.3225, 0.3231], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0410, 0.0343, 0.0326, 0.0420, 0.0470, 0.0375, 0.0480], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:13:06,250 INFO [train.py:904] (7/8) Epoch 14, batch 1450, loss[loss=0.1602, simple_loss=0.249, pruned_loss=0.03565, over 17093.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.255, pruned_loss=0.04501, over 3310061.43 frames. ], batch size: 47, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:13:06,590 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=133402.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:13:23,276 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9212, 4.9589, 5.4174, 5.4052, 5.3787, 5.0428, 5.0100, 4.7486], device='cuda:7'), covar=tensor([0.0319, 0.0477, 0.0421, 0.0399, 0.0435, 0.0386, 0.0851, 0.0407], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0392, 0.0392, 0.0373, 0.0433, 0.0418, 0.0509, 0.0332], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-29 22:13:37,061 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2065, 3.5503, 3.7409, 1.9673, 3.1753, 2.5923, 3.6664, 3.7525], device='cuda:7'), covar=tensor([0.0318, 0.0797, 0.0473, 0.1894, 0.0729, 0.0871, 0.0672, 0.0967], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0149, 0.0159, 0.0146, 0.0138, 0.0126, 0.0137, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:13:48,632 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-29 22:13:52,492 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133435.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 22:13:55,260 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133437.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:14:12,936 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=133450.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:14:14,819 INFO [train.py:904] (7/8) Epoch 14, batch 1500, loss[loss=0.1574, simple_loss=0.2325, pruned_loss=0.04119, over 15698.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2552, pruned_loss=0.04553, over 3317095.81 frames. ], batch size: 35, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:14:36,222 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9281, 4.6554, 4.9297, 5.1433, 5.3148, 4.6236, 5.2803, 5.2885], device='cuda:7'), covar=tensor([0.1670, 0.1362, 0.1745, 0.0718, 0.0505, 0.0892, 0.0508, 0.0566], device='cuda:7'), in_proj_covar=tensor([0.0588, 0.0727, 0.0873, 0.0743, 0.0560, 0.0572, 0.0586, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:14:46,568 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133474.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:14:46,871 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 22:15:03,759 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.705e+02 2.267e+02 2.734e+02 3.162e+02 6.055e+02, threshold=5.467e+02, percent-clipped=3.0 2023-04-29 22:15:23,314 INFO [train.py:904] (7/8) Epoch 14, batch 1550, loss[loss=0.1592, simple_loss=0.2374, pruned_loss=0.04049, over 16771.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.257, pruned_loss=0.04674, over 3321751.16 frames. ], batch size: 39, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:16:10,840 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133535.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:16:27,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8895, 2.2072, 2.3330, 4.5473, 2.2144, 2.7368, 2.3377, 2.4644], device='cuda:7'), covar=tensor([0.0909, 0.3614, 0.2653, 0.0404, 0.3929, 0.2326, 0.3234, 0.3429], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0411, 0.0344, 0.0327, 0.0421, 0.0472, 0.0377, 0.0482], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:16:33,568 INFO [train.py:904] (7/8) Epoch 14, batch 1600, loss[loss=0.1863, simple_loss=0.2571, pruned_loss=0.05774, over 16813.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2589, pruned_loss=0.04768, over 3314598.19 frames. ], batch size: 124, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:17:21,879 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.312e+02 2.830e+02 3.166e+02 6.146e+02, threshold=5.659e+02, percent-clipped=1.0 2023-04-29 22:17:43,040 INFO [train.py:904] (7/8) Epoch 14, batch 1650, loss[loss=0.1785, simple_loss=0.2521, pruned_loss=0.05247, over 16723.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2601, pruned_loss=0.04834, over 3319983.25 frames. ], batch size: 124, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:18:01,424 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8218, 3.9713, 2.4752, 4.5071, 3.0409, 4.4543, 2.3649, 3.1372], device='cuda:7'), covar=tensor([0.0267, 0.0351, 0.1449, 0.0227, 0.0777, 0.0451, 0.1549, 0.0756], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0191, 0.0147, 0.0169, 0.0212, 0.0199, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:18:52,518 INFO [train.py:904] (7/8) Epoch 14, batch 1700, loss[loss=0.175, simple_loss=0.2578, pruned_loss=0.04607, over 15831.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2627, pruned_loss=0.04893, over 3318065.94 frames. ], batch size: 35, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:18:52,986 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133652.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:19:36,392 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1428, 3.2903, 3.2399, 5.2155, 4.4553, 4.6638, 1.9726, 3.6830], device='cuda:7'), covar=tensor([0.1209, 0.0648, 0.0920, 0.0161, 0.0241, 0.0342, 0.1387, 0.0631], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0162, 0.0180, 0.0160, 0.0196, 0.0208, 0.0185, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:19:42,413 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.547e+02 3.320e+02 4.199e+02 1.315e+03, threshold=6.641e+02, percent-clipped=10.0 2023-04-29 22:20:03,033 INFO [train.py:904] (7/8) Epoch 14, batch 1750, loss[loss=0.1686, simple_loss=0.2736, pruned_loss=0.03182, over 17267.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2632, pruned_loss=0.04842, over 3312154.33 frames. ], batch size: 52, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:20:19,417 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=133713.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:20:42,476 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133730.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 22:20:45,415 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133732.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:21:05,346 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-29 22:21:12,612 INFO [train.py:904] (7/8) Epoch 14, batch 1800, loss[loss=0.2129, simple_loss=0.2859, pruned_loss=0.07, over 16811.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2642, pruned_loss=0.0484, over 3309300.66 frames. ], batch size: 124, lr: 4.93e-03, grad_scale: 8.0 2023-04-29 22:21:28,477 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-29 22:21:48,347 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4192, 2.9484, 2.6098, 2.1807, 2.2658, 2.1773, 2.8363, 2.8133], device='cuda:7'), covar=tensor([0.2215, 0.0825, 0.1523, 0.2193, 0.2087, 0.1787, 0.0571, 0.1062], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0261, 0.0289, 0.0287, 0.0280, 0.0231, 0.0275, 0.0311], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:22:00,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.275e+02 2.807e+02 3.369e+02 7.697e+02, threshold=5.615e+02, percent-clipped=1.0 2023-04-29 22:22:22,399 INFO [train.py:904] (7/8) Epoch 14, batch 1850, loss[loss=0.1531, simple_loss=0.241, pruned_loss=0.03266, over 16853.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2649, pruned_loss=0.04813, over 3309269.28 frames. ], batch size: 42, lr: 4.93e-03, grad_scale: 4.0 2023-04-29 22:23:03,000 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=133830.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:23:32,967 INFO [train.py:904] (7/8) Epoch 14, batch 1900, loss[loss=0.1753, simple_loss=0.2689, pruned_loss=0.04081, over 16689.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2639, pruned_loss=0.04739, over 3305704.05 frames. ], batch size: 57, lr: 4.93e-03, grad_scale: 4.0 2023-04-29 22:23:45,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4410, 5.3536, 5.2813, 4.8123, 4.8826, 5.3264, 5.3366, 4.9282], device='cuda:7'), covar=tensor([0.0569, 0.0406, 0.0242, 0.0278, 0.0974, 0.0363, 0.0237, 0.0672], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0372, 0.0328, 0.0308, 0.0343, 0.0357, 0.0222, 0.0382], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:24:15,904 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0328, 4.0347, 3.9044, 3.6351, 3.6830, 4.0080, 3.6874, 3.7990], device='cuda:7'), covar=tensor([0.0594, 0.0508, 0.0237, 0.0241, 0.0705, 0.0416, 0.0971, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0272, 0.0373, 0.0328, 0.0308, 0.0344, 0.0358, 0.0222, 0.0383], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:24:23,484 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.126e+02 2.452e+02 2.920e+02 9.757e+02, threshold=4.904e+02, percent-clipped=2.0 2023-04-29 22:24:42,160 INFO [train.py:904] (7/8) Epoch 14, batch 1950, loss[loss=0.1493, simple_loss=0.248, pruned_loss=0.02526, over 17116.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2643, pruned_loss=0.04744, over 3308403.27 frames. ], batch size: 47, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:25:27,356 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-29 22:25:52,918 INFO [train.py:904] (7/8) Epoch 14, batch 2000, loss[loss=0.1961, simple_loss=0.2639, pruned_loss=0.06414, over 16871.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2639, pruned_loss=0.04735, over 3310377.17 frames. ], batch size: 116, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:25:53,818 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-29 22:26:18,058 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-29 22:26:29,320 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.02 vs. limit=5.0 2023-04-29 22:26:43,623 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.687e+02 2.429e+02 2.777e+02 3.533e+02 6.577e+02, threshold=5.554e+02, percent-clipped=5.0 2023-04-29 22:26:46,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0430, 4.7310, 5.0839, 5.2812, 5.4715, 4.7193, 5.4197, 5.4102], device='cuda:7'), covar=tensor([0.1737, 0.1323, 0.1547, 0.0694, 0.0556, 0.0905, 0.0487, 0.0537], device='cuda:7'), in_proj_covar=tensor([0.0597, 0.0739, 0.0887, 0.0757, 0.0568, 0.0586, 0.0597, 0.0700], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:26:57,047 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=133997.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:27:06,911 INFO [train.py:904] (7/8) Epoch 14, batch 2050, loss[loss=0.1815, simple_loss=0.266, pruned_loss=0.04851, over 16761.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2644, pruned_loss=0.04804, over 3318917.05 frames. ], batch size: 57, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:27:15,557 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134008.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:27:21,684 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1644, 5.1262, 4.8725, 4.3321, 4.9709, 1.8884, 4.7282, 4.8323], device='cuda:7'), covar=tensor([0.0087, 0.0087, 0.0198, 0.0409, 0.0092, 0.2538, 0.0141, 0.0187], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0136, 0.0184, 0.0170, 0.0154, 0.0196, 0.0173, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:27:46,413 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134030.0, num_to_drop=1, layers_to_drop={1} 2023-04-29 22:27:49,378 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134032.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:28:16,836 INFO [train.py:904] (7/8) Epoch 14, batch 2100, loss[loss=0.1525, simple_loss=0.2426, pruned_loss=0.03116, over 17184.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2646, pruned_loss=0.04837, over 3326391.16 frames. ], batch size: 46, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:28:25,745 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134058.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:28:54,066 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134078.0, num_to_drop=1, layers_to_drop={0} 2023-04-29 22:28:56,310 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134080.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:29:09,015 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.284e+02 2.721e+02 3.276e+02 7.722e+02, threshold=5.442e+02, percent-clipped=1.0 2023-04-29 22:29:26,686 INFO [train.py:904] (7/8) Epoch 14, batch 2150, loss[loss=0.1753, simple_loss=0.2612, pruned_loss=0.04466, over 17012.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2649, pruned_loss=0.0485, over 3331743.28 frames. ], batch size: 53, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:29:59,890 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-29 22:30:05,119 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6373, 3.0761, 2.7176, 5.0051, 4.1013, 4.4689, 1.4841, 3.1828], device='cuda:7'), covar=tensor([0.1411, 0.0684, 0.1111, 0.0183, 0.0168, 0.0349, 0.1586, 0.0705], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0163, 0.0181, 0.0161, 0.0198, 0.0210, 0.0186, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:30:06,171 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134130.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:30:30,237 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8166, 2.2486, 2.3242, 4.6189, 2.2107, 2.7147, 2.3134, 2.5096], device='cuda:7'), covar=tensor([0.1014, 0.3672, 0.2713, 0.0360, 0.4066, 0.2556, 0.3374, 0.3710], device='cuda:7'), in_proj_covar=tensor([0.0380, 0.0412, 0.0345, 0.0328, 0.0421, 0.0476, 0.0378, 0.0484], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:30:35,961 INFO [train.py:904] (7/8) Epoch 14, batch 2200, loss[loss=0.1853, simple_loss=0.2699, pruned_loss=0.05038, over 16847.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.265, pruned_loss=0.04832, over 3333244.94 frames. ], batch size: 96, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:31:12,759 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134178.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:31:27,737 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.617e+02 2.496e+02 2.998e+02 3.803e+02 8.330e+02, threshold=5.996e+02, percent-clipped=6.0 2023-04-29 22:31:45,972 INFO [train.py:904] (7/8) Epoch 14, batch 2250, loss[loss=0.1666, simple_loss=0.2371, pruned_loss=0.04807, over 16480.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2657, pruned_loss=0.04887, over 3334505.37 frames. ], batch size: 75, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:31:53,040 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2134, 5.5179, 5.2355, 5.3531, 4.9871, 4.7859, 4.9845, 5.6047], device='cuda:7'), covar=tensor([0.1048, 0.0931, 0.1009, 0.0683, 0.0800, 0.0914, 0.1004, 0.0901], device='cuda:7'), in_proj_covar=tensor([0.0600, 0.0753, 0.0608, 0.0536, 0.0475, 0.0476, 0.0624, 0.0575], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:32:56,464 INFO [train.py:904] (7/8) Epoch 14, batch 2300, loss[loss=0.2174, simple_loss=0.2955, pruned_loss=0.0697, over 16732.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.266, pruned_loss=0.04862, over 3337788.80 frames. ], batch size: 124, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:33:03,272 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0988, 4.4344, 3.1222, 2.5051, 3.1101, 2.5798, 4.7630, 3.9983], device='cuda:7'), covar=tensor([0.2448, 0.0705, 0.1710, 0.2242, 0.2566, 0.1907, 0.0387, 0.0981], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0263, 0.0290, 0.0289, 0.0283, 0.0232, 0.0277, 0.0313], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:33:48,104 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.438e+02 2.937e+02 3.292e+02 7.337e+02, threshold=5.875e+02, percent-clipped=1.0 2023-04-29 22:33:57,297 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7474, 2.7155, 2.3939, 4.2756, 3.6114, 4.1744, 1.5452, 2.9341], device='cuda:7'), covar=tensor([0.1288, 0.0652, 0.1130, 0.0172, 0.0169, 0.0366, 0.1402, 0.0724], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0162, 0.0181, 0.0161, 0.0197, 0.0209, 0.0185, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-29 22:34:06,357 INFO [train.py:904] (7/8) Epoch 14, batch 2350, loss[loss=0.1565, simple_loss=0.2486, pruned_loss=0.03217, over 16855.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2651, pruned_loss=0.04812, over 3344930.02 frames. ], batch size: 42, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:34:09,689 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134304.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:34:15,882 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134308.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:34:16,994 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134309.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:34:58,197 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3301, 5.2958, 5.1668, 4.6768, 4.8008, 5.2393, 5.2028, 4.7694], device='cuda:7'), covar=tensor([0.0557, 0.0347, 0.0266, 0.0304, 0.0978, 0.0336, 0.0263, 0.0707], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0370, 0.0328, 0.0308, 0.0342, 0.0354, 0.0223, 0.0381], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:35:17,013 INFO [train.py:904] (7/8) Epoch 14, batch 2400, loss[loss=0.1544, simple_loss=0.2532, pruned_loss=0.0278, over 17150.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2664, pruned_loss=0.04872, over 3324260.00 frames. ], batch size: 46, lr: 4.92e-03, grad_scale: 8.0 2023-04-29 22:35:19,204 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134353.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:35:23,413 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134356.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:35:35,683 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134365.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:35:42,623 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134370.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:36:08,843 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.303e+02 2.657e+02 3.180e+02 6.909e+02, threshold=5.314e+02, percent-clipped=1.0 2023-04-29 22:36:25,727 INFO [train.py:904] (7/8) Epoch 14, batch 2450, loss[loss=0.2127, simple_loss=0.2905, pruned_loss=0.06747, over 15475.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2672, pruned_loss=0.04844, over 3333185.21 frames. ], batch size: 190, lr: 4.92e-03, grad_scale: 4.0 2023-04-29 22:37:35,056 INFO [train.py:904] (7/8) Epoch 14, batch 2500, loss[loss=0.2273, simple_loss=0.302, pruned_loss=0.0763, over 15474.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2667, pruned_loss=0.04838, over 3328325.84 frames. ], batch size: 190, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:38:05,196 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-29 22:38:28,208 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.356e+02 2.745e+02 3.374e+02 4.761e+02, threshold=5.489e+02, percent-clipped=0.0 2023-04-29 22:38:37,268 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2023-04-29 22:38:41,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1182, 2.0934, 2.6521, 3.1097, 2.9216, 3.5725, 2.3643, 3.5556], device='cuda:7'), covar=tensor([0.0175, 0.0361, 0.0252, 0.0224, 0.0210, 0.0124, 0.0351, 0.0095], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0166, 0.0171, 0.0180, 0.0136, 0.0181, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:38:45,455 INFO [train.py:904] (7/8) Epoch 14, batch 2550, loss[loss=0.1602, simple_loss=0.2573, pruned_loss=0.03155, over 17201.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.266, pruned_loss=0.04757, over 3334735.98 frames. ], batch size: 44, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:38:49,899 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1591, 3.9943, 4.2089, 4.3530, 4.4455, 3.9826, 4.1783, 4.4438], device='cuda:7'), covar=tensor([0.1422, 0.1088, 0.1237, 0.0648, 0.0578, 0.1410, 0.1971, 0.0649], device='cuda:7'), in_proj_covar=tensor([0.0607, 0.0749, 0.0902, 0.0770, 0.0576, 0.0595, 0.0608, 0.0708], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:39:06,235 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7917, 4.7690, 4.7044, 4.1504, 4.7500, 1.8645, 4.5006, 4.4966], device='cuda:7'), covar=tensor([0.0103, 0.0087, 0.0178, 0.0329, 0.0096, 0.2423, 0.0154, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0135, 0.0182, 0.0169, 0.0153, 0.0194, 0.0172, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 22:39:55,155 INFO [train.py:904] (7/8) Epoch 14, batch 2600, loss[loss=0.2173, simple_loss=0.2932, pruned_loss=0.07066, over 16518.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.266, pruned_loss=0.0474, over 3331381.86 frames. ], batch size: 146, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:40:19,850 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.43 vs. limit=5.0 2023-04-29 22:40:46,904 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.364e+02 2.744e+02 3.247e+02 7.768e+02, threshold=5.488e+02, percent-clipped=4.0 2023-04-29 22:41:03,695 INFO [train.py:904] (7/8) Epoch 14, batch 2650, loss[loss=0.184, simple_loss=0.2803, pruned_loss=0.04382, over 17073.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2656, pruned_loss=0.04659, over 3331372.66 frames. ], batch size: 53, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:42:12,180 INFO [train.py:904] (7/8) Epoch 14, batch 2700, loss[loss=0.1828, simple_loss=0.2791, pruned_loss=0.04324, over 17235.00 frames. ], tot_loss[loss=0.179, simple_loss=0.266, pruned_loss=0.04598, over 3336098.03 frames. ], batch size: 52, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:42:13,662 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134653.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:42:16,891 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8068, 3.0680, 3.0757, 2.0410, 2.6994, 2.2381, 3.2915, 3.3125], device='cuda:7'), covar=tensor([0.0247, 0.0799, 0.0611, 0.1703, 0.0810, 0.0972, 0.0602, 0.0853], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0152, 0.0160, 0.0147, 0.0139, 0.0126, 0.0138, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:42:23,638 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134660.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:42:31,300 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=134665.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:43:04,246 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.219e+02 2.559e+02 3.067e+02 4.567e+02, threshold=5.119e+02, percent-clipped=0.0 2023-04-29 22:43:05,843 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6515, 3.7269, 2.2195, 4.2636, 2.8056, 4.2634, 2.3403, 3.0166], device='cuda:7'), covar=tensor([0.0257, 0.0379, 0.1620, 0.0288, 0.0832, 0.0403, 0.1477, 0.0723], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0172, 0.0191, 0.0148, 0.0170, 0.0214, 0.0200, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:43:20,435 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=134701.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:43:21,293 INFO [train.py:904] (7/8) Epoch 14, batch 2750, loss[loss=0.172, simple_loss=0.2554, pruned_loss=0.0443, over 16803.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2663, pruned_loss=0.04602, over 3327256.11 frames. ], batch size: 102, lr: 4.91e-03, grad_scale: 4.0 2023-04-29 22:43:27,336 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6354, 3.7716, 2.7655, 2.1529, 2.5498, 2.2921, 3.8502, 3.3810], device='cuda:7'), covar=tensor([0.2594, 0.0581, 0.1591, 0.2618, 0.2504, 0.1862, 0.0512, 0.1227], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0265, 0.0293, 0.0292, 0.0287, 0.0234, 0.0279, 0.0317], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:43:48,044 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-29 22:43:52,833 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-29 22:44:04,420 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-29 22:44:29,015 INFO [train.py:904] (7/8) Epoch 14, batch 2800, loss[loss=0.1803, simple_loss=0.2772, pruned_loss=0.04172, over 17023.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2666, pruned_loss=0.04615, over 3328026.59 frames. ], batch size: 50, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:44:29,370 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7441, 4.8106, 4.9311, 4.7797, 4.7312, 5.4069, 4.8903, 4.5321], device='cuda:7'), covar=tensor([0.1331, 0.2063, 0.2179, 0.2134, 0.2892, 0.0983, 0.1632, 0.2810], device='cuda:7'), in_proj_covar=tensor([0.0380, 0.0536, 0.0590, 0.0458, 0.0618, 0.0614, 0.0468, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:45:20,157 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.461e+02 2.848e+02 3.331e+02 6.185e+02, threshold=5.697e+02, percent-clipped=2.0 2023-04-29 22:45:37,186 INFO [train.py:904] (7/8) Epoch 14, batch 2850, loss[loss=0.1469, simple_loss=0.2383, pruned_loss=0.02773, over 17018.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2667, pruned_loss=0.04643, over 3330837.21 frames. ], batch size: 41, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:46:03,369 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2023-04-29 22:46:45,302 INFO [train.py:904] (7/8) Epoch 14, batch 2900, loss[loss=0.1535, simple_loss=0.2472, pruned_loss=0.02991, over 17106.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2653, pruned_loss=0.04678, over 3334410.86 frames. ], batch size: 47, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:46:58,766 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=134862.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:47:29,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8941, 3.0338, 2.6362, 4.2482, 3.5637, 4.2115, 1.5697, 2.8605], device='cuda:7'), covar=tensor([0.1280, 0.0530, 0.0918, 0.0169, 0.0172, 0.0356, 0.1393, 0.0784], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0163, 0.0182, 0.0163, 0.0200, 0.0211, 0.0186, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 22:47:36,340 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.493e+02 2.912e+02 3.442e+02 5.810e+02, threshold=5.824e+02, percent-clipped=2.0 2023-04-29 22:47:54,148 INFO [train.py:904] (7/8) Epoch 14, batch 2950, loss[loss=0.1394, simple_loss=0.2191, pruned_loss=0.02984, over 16817.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2641, pruned_loss=0.04715, over 3334862.41 frames. ], batch size: 39, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:48:24,051 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=134923.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:49:02,687 INFO [train.py:904] (7/8) Epoch 14, batch 3000, loss[loss=0.1774, simple_loss=0.2619, pruned_loss=0.04649, over 16526.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.265, pruned_loss=0.04816, over 3321528.31 frames. ], batch size: 68, lr: 4.91e-03, grad_scale: 8.0 2023-04-29 22:49:02,687 INFO [train.py:929] (7/8) Computing validation loss 2023-04-29 22:49:12,424 INFO [train.py:938] (7/8) Epoch 14, validation: loss=0.1382, simple_loss=0.244, pruned_loss=0.01621, over 944034.00 frames. 2023-04-29 22:49:12,424 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-29 22:49:24,947 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134960.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:49:31,407 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=134965.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:49:37,812 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-04-29 22:50:06,982 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.394e+02 2.911e+02 3.370e+02 7.406e+02, threshold=5.821e+02, percent-clipped=1.0 2023-04-29 22:50:19,818 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.40 vs. limit=5.0 2023-04-29 22:50:24,235 INFO [train.py:904] (7/8) Epoch 14, batch 3050, loss[loss=0.1807, simple_loss=0.26, pruned_loss=0.05075, over 16387.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2651, pruned_loss=0.04863, over 3315196.97 frames. ], batch size: 146, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:50:33,251 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135008.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:50:40,114 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135013.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:51:26,178 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2423, 3.3931, 3.6881, 2.4517, 3.3887, 3.7717, 3.4753, 2.0729], device='cuda:7'), covar=tensor([0.0424, 0.0121, 0.0048, 0.0307, 0.0081, 0.0076, 0.0074, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0073, 0.0073, 0.0128, 0.0085, 0.0094, 0.0083, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:51:32,227 INFO [train.py:904] (7/8) Epoch 14, batch 3100, loss[loss=0.1916, simple_loss=0.2855, pruned_loss=0.0489, over 16771.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.265, pruned_loss=0.04902, over 3315574.20 frames. ], batch size: 62, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:52:24,973 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.254e+02 2.683e+02 3.212e+02 6.924e+02, threshold=5.366e+02, percent-clipped=1.0 2023-04-29 22:52:40,986 INFO [train.py:904] (7/8) Epoch 14, batch 3150, loss[loss=0.2101, simple_loss=0.2843, pruned_loss=0.06797, over 16318.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.264, pruned_loss=0.04858, over 3311732.82 frames. ], batch size: 165, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:53:50,748 INFO [train.py:904] (7/8) Epoch 14, batch 3200, loss[loss=0.1715, simple_loss=0.2578, pruned_loss=0.04256, over 16475.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2636, pruned_loss=0.04797, over 3317617.68 frames. ], batch size: 68, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:54:17,421 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135171.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:54:41,150 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.733e+02 2.470e+02 2.854e+02 3.612e+02 7.577e+02, threshold=5.709e+02, percent-clipped=5.0 2023-04-29 22:54:59,025 INFO [train.py:904] (7/8) Epoch 14, batch 3250, loss[loss=0.2479, simple_loss=0.3212, pruned_loss=0.08732, over 12008.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2636, pruned_loss=0.04796, over 3316699.06 frames. ], batch size: 247, lr: 4.90e-03, grad_scale: 8.0 2023-04-29 22:55:22,963 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135218.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:55:41,826 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135232.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 22:56:07,848 INFO [train.py:904] (7/8) Epoch 14, batch 3300, loss[loss=0.1572, simple_loss=0.2561, pruned_loss=0.02917, over 17124.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2649, pruned_loss=0.04812, over 3319275.11 frames. ], batch size: 48, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:56:11,648 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-29 22:57:01,619 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.198e+02 2.720e+02 3.178e+02 6.678e+02, threshold=5.440e+02, percent-clipped=1.0 2023-04-29 22:57:16,995 INFO [train.py:904] (7/8) Epoch 14, batch 3350, loss[loss=0.1726, simple_loss=0.2485, pruned_loss=0.04832, over 16740.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2647, pruned_loss=0.04783, over 3314409.51 frames. ], batch size: 102, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:57:40,991 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2837, 4.0235, 4.3522, 2.1956, 4.6249, 4.6851, 3.3984, 3.5500], device='cuda:7'), covar=tensor([0.0631, 0.0193, 0.0195, 0.1077, 0.0052, 0.0119, 0.0345, 0.0373], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0103, 0.0090, 0.0136, 0.0072, 0.0114, 0.0121, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 22:57:48,095 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1277, 5.6367, 5.8016, 5.4740, 5.5586, 6.1289, 5.6957, 5.3300], device='cuda:7'), covar=tensor([0.0850, 0.1722, 0.1834, 0.1881, 0.2535, 0.0897, 0.1264, 0.2288], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0544, 0.0591, 0.0462, 0.0626, 0.0621, 0.0471, 0.0615], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 22:58:24,256 INFO [train.py:904] (7/8) Epoch 14, batch 3400, loss[loss=0.1834, simple_loss=0.2755, pruned_loss=0.0456, over 17053.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2649, pruned_loss=0.04758, over 3326034.35 frames. ], batch size: 55, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:59:16,989 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.205e+02 2.551e+02 3.042e+02 5.815e+02, threshold=5.103e+02, percent-clipped=2.0 2023-04-29 22:59:32,661 INFO [train.py:904] (7/8) Epoch 14, batch 3450, loss[loss=0.1795, simple_loss=0.2645, pruned_loss=0.04722, over 16806.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2632, pruned_loss=0.0475, over 3330969.05 frames. ], batch size: 42, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 22:59:38,168 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9665, 5.3616, 5.0900, 5.1603, 4.8623, 4.8202, 4.7737, 5.4303], device='cuda:7'), covar=tensor([0.1300, 0.0968, 0.1054, 0.0733, 0.0874, 0.0895, 0.1081, 0.0976], device='cuda:7'), in_proj_covar=tensor([0.0611, 0.0761, 0.0623, 0.0548, 0.0486, 0.0482, 0.0633, 0.0585], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:00:10,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5266, 2.5609, 2.1949, 2.3549, 2.8706, 2.6727, 3.2457, 3.1049], device='cuda:7'), covar=tensor([0.0127, 0.0356, 0.0431, 0.0397, 0.0234, 0.0317, 0.0238, 0.0211], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0220, 0.0215, 0.0215, 0.0222, 0.0222, 0.0232, 0.0216], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:00:18,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6056, 3.9649, 4.1229, 2.7794, 3.6322, 4.1556, 3.7863, 2.3267], device='cuda:7'), covar=tensor([0.0372, 0.0153, 0.0044, 0.0313, 0.0091, 0.0093, 0.0066, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0073, 0.0073, 0.0127, 0.0085, 0.0094, 0.0083, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-29 23:00:41,082 INFO [train.py:904] (7/8) Epoch 14, batch 3500, loss[loss=0.1976, simple_loss=0.2755, pruned_loss=0.05992, over 15505.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2626, pruned_loss=0.04772, over 3322131.84 frames. ], batch size: 191, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 23:01:37,087 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.212e+02 2.743e+02 3.344e+02 6.080e+02, threshold=5.486e+02, percent-clipped=2.0 2023-04-29 23:01:51,742 INFO [train.py:904] (7/8) Epoch 14, batch 3550, loss[loss=0.1656, simple_loss=0.261, pruned_loss=0.03507, over 17020.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.261, pruned_loss=0.04703, over 3323157.53 frames. ], batch size: 50, lr: 4.90e-03, grad_scale: 4.0 2023-04-29 23:02:15,202 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135518.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:02:27,917 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135527.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:03:02,250 INFO [train.py:904] (7/8) Epoch 14, batch 3600, loss[loss=0.1738, simple_loss=0.2654, pruned_loss=0.04112, over 16732.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2605, pruned_loss=0.04636, over 3326476.47 frames. ], batch size: 57, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:03:22,563 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135566.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:03:58,635 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.279e+02 2.683e+02 3.136e+02 7.106e+02, threshold=5.366e+02, percent-clipped=1.0 2023-04-29 23:04:14,273 INFO [train.py:904] (7/8) Epoch 14, batch 3650, loss[loss=0.1828, simple_loss=0.2615, pruned_loss=0.05207, over 15301.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2596, pruned_loss=0.04686, over 3321097.28 frames. ], batch size: 190, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:04:57,411 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135630.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:05:31,040 INFO [train.py:904] (7/8) Epoch 14, batch 3700, loss[loss=0.1891, simple_loss=0.271, pruned_loss=0.05363, over 16656.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2581, pruned_loss=0.04818, over 3298518.27 frames. ], batch size: 57, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:06:31,565 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=135691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:06:32,149 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.199e+02 2.631e+02 3.126e+02 8.071e+02, threshold=5.262e+02, percent-clipped=3.0 2023-04-29 23:06:39,476 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-29 23:06:46,315 INFO [train.py:904] (7/8) Epoch 14, batch 3750, loss[loss=0.1755, simple_loss=0.2524, pruned_loss=0.04933, over 16512.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2588, pruned_loss=0.04971, over 3286209.16 frames. ], batch size: 75, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:08:01,579 INFO [train.py:904] (7/8) Epoch 14, batch 3800, loss[loss=0.203, simple_loss=0.2809, pruned_loss=0.06254, over 16470.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2611, pruned_loss=0.05147, over 3272752.62 frames. ], batch size: 35, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:08:39,201 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7286, 4.8030, 4.9933, 4.8906, 4.8560, 5.4295, 4.9671, 4.6702], device='cuda:7'), covar=tensor([0.1291, 0.1833, 0.1866, 0.1939, 0.2686, 0.0987, 0.1358, 0.2215], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0536, 0.0579, 0.0454, 0.0614, 0.0608, 0.0463, 0.0602], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 23:09:00,855 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.772e+02 2.313e+02 2.548e+02 3.011e+02 5.523e+02, threshold=5.096e+02, percent-clipped=2.0 2023-04-29 23:09:16,028 INFO [train.py:904] (7/8) Epoch 14, batch 3850, loss[loss=0.1614, simple_loss=0.2388, pruned_loss=0.042, over 16858.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2615, pruned_loss=0.05222, over 3275268.82 frames. ], batch size: 90, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:09:53,626 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=135827.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:10:30,033 INFO [train.py:904] (7/8) Epoch 14, batch 3900, loss[loss=0.1843, simple_loss=0.2571, pruned_loss=0.05571, over 16882.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2608, pruned_loss=0.0527, over 3266911.01 frames. ], batch size: 109, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:11:05,208 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=135875.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:11:30,221 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.144e+02 2.448e+02 3.272e+02 6.091e+02, threshold=4.896e+02, percent-clipped=2.0 2023-04-29 23:11:40,171 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4566, 4.3698, 4.5281, 4.4335, 4.4366, 4.9963, 4.5408, 4.2780], device='cuda:7'), covar=tensor([0.1672, 0.2235, 0.2376, 0.2299, 0.2907, 0.1148, 0.1557, 0.2604], device='cuda:7'), in_proj_covar=tensor([0.0379, 0.0541, 0.0585, 0.0457, 0.0618, 0.0612, 0.0467, 0.0608], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 23:11:45,254 INFO [train.py:904] (7/8) Epoch 14, batch 3950, loss[loss=0.2294, simple_loss=0.2958, pruned_loss=0.08148, over 12763.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2611, pruned_loss=0.0532, over 3256167.53 frames. ], batch size: 247, lr: 4.89e-03, grad_scale: 4.0 2023-04-29 23:12:58,435 INFO [train.py:904] (7/8) Epoch 14, batch 4000, loss[loss=0.2017, simple_loss=0.2865, pruned_loss=0.05849, over 16340.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2603, pruned_loss=0.05326, over 3271809.49 frames. ], batch size: 165, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:13:36,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9206, 4.1751, 2.4228, 4.8988, 3.1592, 4.8746, 2.8487, 3.1836], device='cuda:7'), covar=tensor([0.0225, 0.0269, 0.1590, 0.0079, 0.0732, 0.0188, 0.1258, 0.0666], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0172, 0.0191, 0.0149, 0.0171, 0.0215, 0.0201, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 23:13:41,801 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135982.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:13:47,669 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=135986.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:13:55,966 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.249e+02 2.823e+02 3.453e+02 5.778e+02, threshold=5.647e+02, percent-clipped=7.0 2023-04-29 23:13:58,956 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=135994.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:14:13,538 INFO [train.py:904] (7/8) Epoch 14, batch 4050, loss[loss=0.1877, simple_loss=0.2652, pruned_loss=0.05507, over 16911.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2604, pruned_loss=0.0523, over 3273983.23 frames. ], batch size: 109, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:15:11,374 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-29 23:15:12,454 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136043.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:15:24,950 INFO [train.py:904] (7/8) Epoch 14, batch 4100, loss[loss=0.1851, simple_loss=0.2715, pruned_loss=0.04932, over 16985.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2622, pruned_loss=0.05156, over 3267033.93 frames. ], batch size: 55, lr: 4.89e-03, grad_scale: 8.0 2023-04-29 23:15:30,650 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136055.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:16:22,009 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-04-29 23:16:24,202 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 1.983e+02 2.350e+02 2.941e+02 4.738e+02, threshold=4.699e+02, percent-clipped=0.0 2023-04-29 23:16:40,533 INFO [train.py:904] (7/8) Epoch 14, batch 4150, loss[loss=0.2258, simple_loss=0.3152, pruned_loss=0.06826, over 16915.00 frames. ], tot_loss[loss=0.1888, simple_loss=0.2692, pruned_loss=0.05416, over 3227891.27 frames. ], batch size: 109, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:17:41,234 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136142.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:17:55,773 INFO [train.py:904] (7/8) Epoch 14, batch 4200, loss[loss=0.2393, simple_loss=0.3138, pruned_loss=0.08238, over 11591.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2763, pruned_loss=0.05594, over 3201827.97 frames. ], batch size: 247, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:18:28,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7747, 4.9210, 5.1774, 4.9231, 4.9222, 5.5638, 5.0371, 4.7379], device='cuda:7'), covar=tensor([0.0903, 0.1708, 0.1388, 0.1806, 0.2346, 0.0738, 0.1172, 0.2187], device='cuda:7'), in_proj_covar=tensor([0.0371, 0.0529, 0.0569, 0.0446, 0.0602, 0.0597, 0.0457, 0.0594], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 23:18:55,505 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.476e+02 2.870e+02 3.638e+02 7.412e+02, threshold=5.739e+02, percent-clipped=7.0 2023-04-29 23:19:10,110 INFO [train.py:904] (7/8) Epoch 14, batch 4250, loss[loss=0.1631, simple_loss=0.2653, pruned_loss=0.03042, over 16892.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2798, pruned_loss=0.05598, over 3172836.60 frames. ], batch size: 96, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:19:12,638 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136203.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:20:23,655 INFO [train.py:904] (7/8) Epoch 14, batch 4300, loss[loss=0.2191, simple_loss=0.2936, pruned_loss=0.07229, over 11537.00 frames. ], tot_loss[loss=0.1953, simple_loss=0.2808, pruned_loss=0.05492, over 3184375.76 frames. ], batch size: 247, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:21:14,658 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136286.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:21:23,663 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.749e+02 2.283e+02 2.653e+02 3.073e+02 4.214e+02, threshold=5.307e+02, percent-clipped=0.0 2023-04-29 23:21:38,132 INFO [train.py:904] (7/8) Epoch 14, batch 4350, loss[loss=0.2149, simple_loss=0.2956, pruned_loss=0.06712, over 12001.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2841, pruned_loss=0.05617, over 3175640.93 frames. ], batch size: 247, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:22:27,142 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136334.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:29,188 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136335.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:33,169 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136338.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:51,138 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136350.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:22:51,389 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0337, 3.1056, 2.4944, 2.8393, 3.4353, 3.0978, 3.6885, 3.5942], device='cuda:7'), covar=tensor([0.0047, 0.0267, 0.0379, 0.0295, 0.0144, 0.0256, 0.0149, 0.0154], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0214, 0.0210, 0.0209, 0.0216, 0.0216, 0.0223, 0.0209], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:22:53,926 INFO [train.py:904] (7/8) Epoch 14, batch 4400, loss[loss=0.2086, simple_loss=0.2983, pruned_loss=0.05944, over 16718.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2861, pruned_loss=0.05692, over 3186382.12 frames. ], batch size: 89, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:23:46,784 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4623, 2.6443, 2.2384, 2.3247, 2.9937, 2.5762, 3.1741, 3.1037], device='cuda:7'), covar=tensor([0.0060, 0.0270, 0.0362, 0.0328, 0.0167, 0.0270, 0.0146, 0.0193], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0213, 0.0209, 0.0208, 0.0215, 0.0216, 0.0221, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:23:51,150 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.242e+02 2.686e+02 3.350e+02 5.764e+02, threshold=5.372e+02, percent-clipped=1.0 2023-04-29 23:23:58,093 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136396.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:24:06,506 INFO [train.py:904] (7/8) Epoch 14, batch 4450, loss[loss=0.2162, simple_loss=0.3052, pruned_loss=0.06358, over 16464.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2898, pruned_loss=0.05837, over 3196113.06 frames. ], batch size: 68, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:25:18,614 INFO [train.py:904] (7/8) Epoch 14, batch 4500, loss[loss=0.1936, simple_loss=0.2799, pruned_loss=0.05369, over 16450.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2897, pruned_loss=0.0585, over 3208380.45 frames. ], batch size: 68, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:25:52,987 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9213, 5.1931, 4.9943, 5.0066, 4.7188, 4.5604, 4.6406, 5.3198], device='cuda:7'), covar=tensor([0.0961, 0.0793, 0.0888, 0.0646, 0.0730, 0.0873, 0.0914, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0581, 0.0719, 0.0588, 0.0522, 0.0463, 0.0463, 0.0604, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:26:18,141 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.032e+02 2.340e+02 2.822e+02 5.483e+02, threshold=4.680e+02, percent-clipped=1.0 2023-04-29 23:26:27,084 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136498.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:26:31,831 INFO [train.py:904] (7/8) Epoch 14, batch 4550, loss[loss=0.2068, simple_loss=0.2891, pruned_loss=0.06225, over 16673.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2904, pruned_loss=0.05931, over 3218486.41 frames. ], batch size: 57, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:26:46,178 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-29 23:27:39,821 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-04-29 23:27:45,703 INFO [train.py:904] (7/8) Epoch 14, batch 4600, loss[loss=0.2053, simple_loss=0.2908, pruned_loss=0.0599, over 17029.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2912, pruned_loss=0.05917, over 3235908.49 frames. ], batch size: 55, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:28:15,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0242, 2.5351, 2.6434, 1.9076, 2.7344, 2.8292, 2.4517, 2.3882], device='cuda:7'), covar=tensor([0.0722, 0.0232, 0.0200, 0.0913, 0.0087, 0.0205, 0.0441, 0.0419], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0105, 0.0089, 0.0139, 0.0072, 0.0115, 0.0123, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 23:28:42,908 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.145e+02 2.405e+02 2.956e+02 4.611e+02, threshold=4.810e+02, percent-clipped=0.0 2023-04-29 23:28:57,113 INFO [train.py:904] (7/8) Epoch 14, batch 4650, loss[loss=0.239, simple_loss=0.3093, pruned_loss=0.08442, over 11708.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2901, pruned_loss=0.05928, over 3218276.78 frames. ], batch size: 246, lr: 4.88e-03, grad_scale: 8.0 2023-04-29 23:29:03,176 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6829, 2.8799, 2.7696, 4.8690, 3.8635, 4.2266, 1.5894, 2.9791], device='cuda:7'), covar=tensor([0.1292, 0.0707, 0.1049, 0.0100, 0.0242, 0.0312, 0.1547, 0.0792], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0164, 0.0184, 0.0162, 0.0201, 0.0211, 0.0187, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 23:29:52,236 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136638.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:29:58,861 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136643.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:30:08,598 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136650.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:30:10,588 INFO [train.py:904] (7/8) Epoch 14, batch 4700, loss[loss=0.1724, simple_loss=0.2575, pruned_loss=0.04362, over 16278.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2867, pruned_loss=0.05805, over 3218361.21 frames. ], batch size: 35, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:30:17,282 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7346, 3.9184, 2.0621, 4.6920, 2.9408, 4.4814, 2.3393, 2.9868], device='cuda:7'), covar=tensor([0.0265, 0.0328, 0.1900, 0.0086, 0.0808, 0.0363, 0.1562, 0.0823], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0190, 0.0142, 0.0170, 0.0212, 0.0198, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 23:31:01,339 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136686.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:31:09,105 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136691.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:31:09,689 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-29 23:31:09,910 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.115e+02 2.410e+02 3.015e+02 5.394e+02, threshold=4.821e+02, percent-clipped=2.0 2023-04-29 23:31:18,635 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136698.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:31:23,576 INFO [train.py:904] (7/8) Epoch 14, batch 4750, loss[loss=0.1736, simple_loss=0.2586, pruned_loss=0.04426, over 16452.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.283, pruned_loss=0.0562, over 3212956.37 frames. ], batch size: 75, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:31:27,875 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136704.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:32:36,927 INFO [train.py:904] (7/8) Epoch 14, batch 4800, loss[loss=0.1831, simple_loss=0.2798, pruned_loss=0.04319, over 16846.00 frames. ], tot_loss[loss=0.1938, simple_loss=0.2793, pruned_loss=0.05418, over 3207702.39 frames. ], batch size: 102, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:33:36,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 1.970e+02 2.266e+02 2.773e+02 4.879e+02, threshold=4.532e+02, percent-clipped=1.0 2023-04-29 23:33:47,189 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136798.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:33:52,194 INFO [train.py:904] (7/8) Epoch 14, batch 4850, loss[loss=0.1759, simple_loss=0.2697, pruned_loss=0.04103, over 16560.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2814, pruned_loss=0.05441, over 3165600.49 frames. ], batch size: 62, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:33:59,189 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=136806.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:34:13,175 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5634, 4.6510, 4.8494, 4.7206, 4.6553, 5.2618, 4.7215, 4.4618], device='cuda:7'), covar=tensor([0.1104, 0.1586, 0.1327, 0.1757, 0.2403, 0.0779, 0.1222, 0.2192], device='cuda:7'), in_proj_covar=tensor([0.0372, 0.0522, 0.0561, 0.0443, 0.0598, 0.0592, 0.0452, 0.0593], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-29 23:34:20,031 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-29 23:34:56,907 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=136846.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:35:05,058 INFO [train.py:904] (7/8) Epoch 14, batch 4900, loss[loss=0.1972, simple_loss=0.2796, pruned_loss=0.05743, over 17028.00 frames. ], tot_loss[loss=0.1931, simple_loss=0.2804, pruned_loss=0.05292, over 3153311.59 frames. ], batch size: 55, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:35:27,602 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2884, 5.2838, 5.0886, 4.0585, 5.1593, 1.5022, 4.8523, 4.8299], device='cuda:7'), covar=tensor([0.0070, 0.0074, 0.0150, 0.0531, 0.0093, 0.2990, 0.0136, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0130, 0.0175, 0.0165, 0.0148, 0.0188, 0.0165, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:35:28,893 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=136867.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:35:33,739 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-29 23:35:43,234 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8269, 3.7163, 3.8768, 4.0131, 4.0832, 3.6539, 4.0402, 4.1096], device='cuda:7'), covar=tensor([0.1241, 0.0995, 0.1055, 0.0535, 0.0516, 0.1718, 0.0682, 0.0547], device='cuda:7'), in_proj_covar=tensor([0.0561, 0.0693, 0.0827, 0.0709, 0.0534, 0.0547, 0.0554, 0.0654], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:35:53,035 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3301, 5.3304, 5.1277, 4.4121, 5.1988, 1.6892, 4.9473, 5.0136], device='cuda:7'), covar=tensor([0.0054, 0.0052, 0.0122, 0.0420, 0.0074, 0.2476, 0.0093, 0.0158], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0130, 0.0174, 0.0165, 0.0148, 0.0188, 0.0165, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:36:02,821 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.076e+02 2.406e+02 2.771e+02 5.710e+02, threshold=4.811e+02, percent-clipped=1.0 2023-04-29 23:36:18,500 INFO [train.py:904] (7/8) Epoch 14, batch 4950, loss[loss=0.2011, simple_loss=0.2917, pruned_loss=0.05529, over 16481.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2797, pruned_loss=0.05215, over 3161434.05 frames. ], batch size: 146, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:37:30,154 INFO [train.py:904] (7/8) Epoch 14, batch 5000, loss[loss=0.1868, simple_loss=0.2719, pruned_loss=0.0509, over 17123.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2814, pruned_loss=0.05217, over 3173017.62 frames. ], batch size: 49, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:38:24,239 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=136991.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:38:25,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.271e+02 2.523e+02 3.075e+02 5.093e+02, threshold=5.046e+02, percent-clipped=1.0 2023-04-29 23:38:36,221 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=136999.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:38:39,701 INFO [train.py:904] (7/8) Epoch 14, batch 5050, loss[loss=0.2131, simple_loss=0.2994, pruned_loss=0.06337, over 16415.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2818, pruned_loss=0.05194, over 3182791.36 frames. ], batch size: 146, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:39:31,294 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=137039.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:39:49,981 INFO [train.py:904] (7/8) Epoch 14, batch 5100, loss[loss=0.1784, simple_loss=0.277, pruned_loss=0.03985, over 16500.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2794, pruned_loss=0.05095, over 3203892.35 frames. ], batch size: 68, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:40:45,953 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.206e+02 2.593e+02 3.193e+02 6.541e+02, threshold=5.187e+02, percent-clipped=2.0 2023-04-29 23:41:00,264 INFO [train.py:904] (7/8) Epoch 14, batch 5150, loss[loss=0.1851, simple_loss=0.2897, pruned_loss=0.04027, over 16434.00 frames. ], tot_loss[loss=0.1901, simple_loss=0.2797, pruned_loss=0.05019, over 3206430.03 frames. ], batch size: 146, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:42:05,698 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9501, 2.0444, 2.3599, 3.1976, 2.0716, 2.2875, 2.2379, 2.2140], device='cuda:7'), covar=tensor([0.1095, 0.3223, 0.2039, 0.0596, 0.3678, 0.2199, 0.2997, 0.2819], device='cuda:7'), in_proj_covar=tensor([0.0372, 0.0411, 0.0341, 0.0322, 0.0417, 0.0473, 0.0373, 0.0480], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:42:08,571 INFO [train.py:904] (7/8) Epoch 14, batch 5200, loss[loss=0.1787, simple_loss=0.2601, pruned_loss=0.04866, over 17251.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2776, pruned_loss=0.04966, over 3215057.76 frames. ], batch size: 52, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:42:18,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4015, 3.0065, 2.6960, 2.1664, 2.1714, 2.2851, 2.9673, 2.8530], device='cuda:7'), covar=tensor([0.2153, 0.0679, 0.1381, 0.2193, 0.2108, 0.1716, 0.0529, 0.0998], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0261, 0.0290, 0.0289, 0.0284, 0.0230, 0.0277, 0.0310], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:42:22,363 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=137162.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:43:01,739 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2023-04-29 23:43:03,328 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.223e+02 2.609e+02 3.158e+02 5.537e+02, threshold=5.218e+02, percent-clipped=1.0 2023-04-29 23:43:18,147 INFO [train.py:904] (7/8) Epoch 14, batch 5250, loss[loss=0.1592, simple_loss=0.2531, pruned_loss=0.03265, over 16883.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2748, pruned_loss=0.04903, over 3223446.73 frames. ], batch size: 96, lr: 4.87e-03, grad_scale: 8.0 2023-04-29 23:44:26,552 INFO [train.py:904] (7/8) Epoch 14, batch 5300, loss[loss=0.1711, simple_loss=0.2602, pruned_loss=0.04099, over 16670.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2714, pruned_loss=0.04805, over 3227384.88 frames. ], batch size: 89, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:44:27,774 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-04-29 23:44:32,846 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7051, 3.9062, 2.1792, 4.4959, 2.8470, 4.3941, 2.4407, 3.0017], device='cuda:7'), covar=tensor([0.0240, 0.0281, 0.1745, 0.0098, 0.0800, 0.0302, 0.1507, 0.0800], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0168, 0.0189, 0.0138, 0.0169, 0.0207, 0.0197, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-29 23:45:23,603 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.002e+02 2.371e+02 2.932e+02 5.680e+02, threshold=4.741e+02, percent-clipped=1.0 2023-04-29 23:45:33,694 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137299.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:45:36,788 INFO [train.py:904] (7/8) Epoch 14, batch 5350, loss[loss=0.1906, simple_loss=0.2729, pruned_loss=0.05409, over 16651.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2698, pruned_loss=0.04735, over 3224966.14 frames. ], batch size: 57, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:46:40,463 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=137347.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:46:46,693 INFO [train.py:904] (7/8) Epoch 14, batch 5400, loss[loss=0.205, simple_loss=0.2899, pruned_loss=0.06004, over 16833.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.273, pruned_loss=0.04804, over 3219388.07 frames. ], batch size: 116, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:47:06,684 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-29 23:47:46,870 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.156e+02 2.492e+02 3.031e+02 4.895e+02, threshold=4.984e+02, percent-clipped=1.0 2023-04-29 23:48:02,474 INFO [train.py:904] (7/8) Epoch 14, batch 5450, loss[loss=0.2099, simple_loss=0.2969, pruned_loss=0.06148, over 16739.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.276, pruned_loss=0.04944, over 3220447.70 frames. ], batch size: 89, lr: 4.86e-03, grad_scale: 4.0 2023-04-29 23:49:19,973 INFO [train.py:904] (7/8) Epoch 14, batch 5500, loss[loss=0.225, simple_loss=0.3095, pruned_loss=0.07019, over 16815.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2843, pruned_loss=0.05507, over 3167488.88 frames. ], batch size: 102, lr: 4.86e-03, grad_scale: 4.0 2023-04-29 23:49:35,274 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=137462.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:49:42,393 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4681, 3.5302, 3.2535, 2.9187, 3.1277, 3.4172, 3.3054, 3.1986], device='cuda:7'), covar=tensor([0.0537, 0.0535, 0.0232, 0.0240, 0.0463, 0.0407, 0.0935, 0.0442], device='cuda:7'), in_proj_covar=tensor([0.0255, 0.0353, 0.0306, 0.0289, 0.0325, 0.0340, 0.0207, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:49:53,715 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3979, 4.6596, 4.3986, 4.4354, 4.1924, 4.1700, 4.1585, 4.6937], device='cuda:7'), covar=tensor([0.1023, 0.0775, 0.0975, 0.0746, 0.0799, 0.1266, 0.1000, 0.0865], device='cuda:7'), in_proj_covar=tensor([0.0574, 0.0712, 0.0587, 0.0518, 0.0455, 0.0459, 0.0597, 0.0554], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:50:23,836 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.042e+02 3.226e+02 3.760e+02 4.566e+02 8.180e+02, threshold=7.520e+02, percent-clipped=15.0 2023-04-29 23:50:37,003 INFO [train.py:904] (7/8) Epoch 14, batch 5550, loss[loss=0.3268, simple_loss=0.3731, pruned_loss=0.1402, over 11124.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2922, pruned_loss=0.061, over 3138975.04 frames. ], batch size: 248, lr: 4.86e-03, grad_scale: 4.0 2023-04-29 23:50:51,297 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=137510.0, num_to_drop=0, layers_to_drop=set() 2023-04-29 23:50:56,365 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-29 23:51:28,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8843, 4.1303, 3.9014, 3.9730, 3.7102, 3.7082, 3.8436, 4.0829], device='cuda:7'), covar=tensor([0.1063, 0.0854, 0.1101, 0.0799, 0.0748, 0.1649, 0.0925, 0.1047], device='cuda:7'), in_proj_covar=tensor([0.0575, 0.0713, 0.0587, 0.0518, 0.0455, 0.0460, 0.0597, 0.0554], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:51:46,825 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9297, 2.7971, 2.8585, 2.1340, 2.6654, 2.2281, 2.7470, 2.9486], device='cuda:7'), covar=tensor([0.0262, 0.0625, 0.0473, 0.1632, 0.0712, 0.0890, 0.0520, 0.0607], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0152, 0.0162, 0.0147, 0.0139, 0.0127, 0.0139, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-29 23:51:55,245 INFO [train.py:904] (7/8) Epoch 14, batch 5600, loss[loss=0.2264, simple_loss=0.3056, pruned_loss=0.07353, over 16842.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2974, pruned_loss=0.06534, over 3115133.99 frames. ], batch size: 116, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:52:06,248 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0369, 2.4911, 2.6484, 1.9416, 2.6834, 2.7893, 2.4360, 2.3691], device='cuda:7'), covar=tensor([0.0659, 0.0205, 0.0205, 0.0816, 0.0089, 0.0216, 0.0422, 0.0393], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0104, 0.0089, 0.0138, 0.0071, 0.0113, 0.0123, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-29 23:53:03,061 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.471e+02 3.513e+02 4.095e+02 5.435e+02 1.324e+03, threshold=8.190e+02, percent-clipped=8.0 2023-04-29 23:53:17,476 INFO [train.py:904] (7/8) Epoch 14, batch 5650, loss[loss=0.3076, simple_loss=0.3579, pruned_loss=0.1286, over 11385.00 frames. ], tot_loss[loss=0.2218, simple_loss=0.303, pruned_loss=0.0703, over 3071971.20 frames. ], batch size: 248, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:53:48,241 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-04-29 23:53:49,562 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-29 23:53:51,025 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9307, 4.9183, 4.7145, 4.0931, 4.8063, 1.8363, 4.5761, 4.5211], device='cuda:7'), covar=tensor([0.0068, 0.0059, 0.0139, 0.0321, 0.0073, 0.2437, 0.0092, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0130, 0.0176, 0.0166, 0.0148, 0.0189, 0.0165, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:54:24,909 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4889, 3.4569, 3.4273, 2.7782, 3.2904, 2.1355, 3.0837, 2.8085], device='cuda:7'), covar=tensor([0.0133, 0.0098, 0.0147, 0.0215, 0.0087, 0.2049, 0.0121, 0.0203], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0130, 0.0176, 0.0165, 0.0148, 0.0189, 0.0165, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:54:38,724 INFO [train.py:904] (7/8) Epoch 14, batch 5700, loss[loss=0.2771, simple_loss=0.3335, pruned_loss=0.1104, over 11264.00 frames. ], tot_loss[loss=0.2245, simple_loss=0.3046, pruned_loss=0.07218, over 3038037.86 frames. ], batch size: 246, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:55:26,607 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6627, 2.1902, 1.7893, 1.9702, 2.5022, 2.1589, 2.5760, 2.7386], device='cuda:7'), covar=tensor([0.0110, 0.0294, 0.0410, 0.0362, 0.0176, 0.0291, 0.0147, 0.0181], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0210, 0.0206, 0.0206, 0.0212, 0.0213, 0.0217, 0.0206], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:55:45,159 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.087e+02 3.198e+02 3.940e+02 4.824e+02 8.425e+02, threshold=7.880e+02, percent-clipped=1.0 2023-04-29 23:55:51,571 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8051, 5.0670, 4.8296, 4.8606, 4.5672, 4.4809, 4.5280, 5.1642], device='cuda:7'), covar=tensor([0.1036, 0.0817, 0.1011, 0.0765, 0.0788, 0.1028, 0.1023, 0.0888], device='cuda:7'), in_proj_covar=tensor([0.0576, 0.0713, 0.0591, 0.0518, 0.0454, 0.0462, 0.0599, 0.0551], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-29 23:55:59,937 INFO [train.py:904] (7/8) Epoch 14, batch 5750, loss[loss=0.2272, simple_loss=0.309, pruned_loss=0.07269, over 16911.00 frames. ], tot_loss[loss=0.2262, simple_loss=0.3069, pruned_loss=0.07278, over 3049378.88 frames. ], batch size: 109, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:57:21,163 INFO [train.py:904] (7/8) Epoch 14, batch 5800, loss[loss=0.2084, simple_loss=0.298, pruned_loss=0.05939, over 16751.00 frames. ], tot_loss[loss=0.2249, simple_loss=0.3067, pruned_loss=0.07156, over 3045049.01 frames. ], batch size: 124, lr: 4.86e-03, grad_scale: 8.0 2023-04-29 23:57:39,472 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.97 vs. limit=5.0 2023-04-29 23:58:26,272 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.930e+02 3.076e+02 3.658e+02 4.339e+02 6.692e+02, threshold=7.316e+02, percent-clipped=0.0 2023-04-29 23:58:41,181 INFO [train.py:904] (7/8) Epoch 14, batch 5850, loss[loss=0.2326, simple_loss=0.2942, pruned_loss=0.08552, over 10900.00 frames. ], tot_loss[loss=0.2213, simple_loss=0.3041, pruned_loss=0.06925, over 3069424.08 frames. ], batch size: 247, lr: 4.85e-03, grad_scale: 8.0 2023-04-29 23:59:33,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-04-30 00:00:03,510 INFO [train.py:904] (7/8) Epoch 14, batch 5900, loss[loss=0.2314, simple_loss=0.3026, pruned_loss=0.08013, over 11338.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.3042, pruned_loss=0.06951, over 3060036.70 frames. ], batch size: 246, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:00:51,945 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-30 00:00:52,645 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1460, 5.1560, 4.9510, 4.6110, 4.5695, 5.0394, 4.9936, 4.6995], device='cuda:7'), covar=tensor([0.0640, 0.0716, 0.0288, 0.0292, 0.1064, 0.0451, 0.0299, 0.0793], device='cuda:7'), in_proj_covar=tensor([0.0255, 0.0350, 0.0304, 0.0285, 0.0323, 0.0334, 0.0207, 0.0359], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:01:10,261 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137892.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:01:10,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.749e+02 3.330e+02 4.231e+02 8.888e+02, threshold=6.660e+02, percent-clipped=2.0 2023-04-30 00:01:15,197 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=137895.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:01:24,306 INFO [train.py:904] (7/8) Epoch 14, batch 5950, loss[loss=0.2167, simple_loss=0.2966, pruned_loss=0.0684, over 11649.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.3046, pruned_loss=0.06785, over 3085660.78 frames. ], batch size: 248, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:02:44,554 INFO [train.py:904] (7/8) Epoch 14, batch 6000, loss[loss=0.1937, simple_loss=0.283, pruned_loss=0.05218, over 16235.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3031, pruned_loss=0.06698, over 3088795.89 frames. ], batch size: 165, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:02:44,554 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 00:02:55,348 INFO [train.py:938] (7/8) Epoch 14, validation: loss=0.1574, simple_loss=0.2706, pruned_loss=0.0221, over 944034.00 frames. 2023-04-30 00:02:55,349 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 00:02:57,523 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137953.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:03:01,783 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=137956.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:03:57,701 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-30 00:03:59,085 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.967e+02 2.895e+02 3.490e+02 4.190e+02 7.002e+02, threshold=6.980e+02, percent-clipped=3.0 2023-04-30 00:04:18,775 INFO [train.py:904] (7/8) Epoch 14, batch 6050, loss[loss=0.224, simple_loss=0.2958, pruned_loss=0.07605, over 11578.00 frames. ], tot_loss[loss=0.2177, simple_loss=0.3022, pruned_loss=0.06663, over 3080838.45 frames. ], batch size: 248, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:04:20,164 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5889, 3.6910, 2.7721, 2.1561, 2.4721, 2.2996, 3.8111, 3.3132], device='cuda:7'), covar=tensor([0.2718, 0.0650, 0.1655, 0.2505, 0.2496, 0.1908, 0.0550, 0.1175], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0260, 0.0288, 0.0289, 0.0283, 0.0230, 0.0276, 0.0308], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:04:35,871 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 00:04:52,153 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 00:05:40,166 INFO [train.py:904] (7/8) Epoch 14, batch 6100, loss[loss=0.1928, simple_loss=0.2786, pruned_loss=0.05351, over 17060.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.3008, pruned_loss=0.06494, over 3096010.46 frames. ], batch size: 55, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:05:58,726 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0622, 3.3452, 3.5224, 3.4851, 3.4813, 3.3058, 3.1748, 3.3922], device='cuda:7'), covar=tensor([0.0633, 0.0775, 0.0523, 0.0654, 0.0722, 0.0717, 0.1526, 0.0659], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0373, 0.0370, 0.0356, 0.0419, 0.0397, 0.0488, 0.0316], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 00:06:16,271 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2077, 4.5086, 4.1284, 3.9917, 3.4501, 4.4255, 4.2133, 3.9451], device='cuda:7'), covar=tensor([0.1058, 0.0610, 0.0509, 0.0404, 0.1925, 0.0479, 0.0717, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0255, 0.0349, 0.0303, 0.0284, 0.0322, 0.0333, 0.0206, 0.0359], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:06:45,112 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.919e+02 3.731e+02 4.841e+02 8.594e+02, threshold=7.462e+02, percent-clipped=5.0 2023-04-30 00:06:59,048 INFO [train.py:904] (7/8) Epoch 14, batch 6150, loss[loss=0.207, simple_loss=0.298, pruned_loss=0.05803, over 16726.00 frames. ], tot_loss[loss=0.2143, simple_loss=0.2992, pruned_loss=0.06471, over 3094028.27 frames. ], batch size: 89, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:08:16,196 INFO [train.py:904] (7/8) Epoch 14, batch 6200, loss[loss=0.2385, simple_loss=0.3187, pruned_loss=0.07914, over 15517.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2971, pruned_loss=0.06419, over 3100612.77 frames. ], batch size: 191, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:09:17,890 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.947e+02 3.424e+02 4.172e+02 1.390e+03, threshold=6.848e+02, percent-clipped=3.0 2023-04-30 00:09:32,468 INFO [train.py:904] (7/8) Epoch 14, batch 6250, loss[loss=0.1919, simple_loss=0.2857, pruned_loss=0.04903, over 16741.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.296, pruned_loss=0.06333, over 3121505.65 frames. ], batch size: 83, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:10:42,359 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138248.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:10:46,673 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138251.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:10:47,503 INFO [train.py:904] (7/8) Epoch 14, batch 6300, loss[loss=0.1943, simple_loss=0.2747, pruned_loss=0.05699, over 16677.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2957, pruned_loss=0.06304, over 3106817.05 frames. ], batch size: 57, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:11:17,697 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138270.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:11:31,967 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 00:11:43,463 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 00:11:52,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.960e+02 2.966e+02 3.499e+02 4.315e+02 9.401e+02, threshold=6.998e+02, percent-clipped=7.0 2023-04-30 00:11:59,799 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9460, 2.6567, 2.6107, 1.8808, 2.4513, 2.6713, 2.5603, 1.8147], device='cuda:7'), covar=tensor([0.0356, 0.0064, 0.0067, 0.0294, 0.0107, 0.0098, 0.0085, 0.0378], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0072, 0.0073, 0.0128, 0.0086, 0.0096, 0.0083, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 00:12:05,889 INFO [train.py:904] (7/8) Epoch 14, batch 6350, loss[loss=0.2403, simple_loss=0.3113, pruned_loss=0.08461, over 15474.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2968, pruned_loss=0.06394, over 3110864.67 frames. ], batch size: 190, lr: 4.85e-03, grad_scale: 8.0 2023-04-30 00:12:37,616 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138322.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:12:45,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4983, 3.5494, 3.3325, 3.0674, 3.1312, 3.4420, 3.2909, 3.2740], device='cuda:7'), covar=tensor([0.0581, 0.0631, 0.0256, 0.0243, 0.0474, 0.0419, 0.1057, 0.0495], device='cuda:7'), in_proj_covar=tensor([0.0252, 0.0347, 0.0301, 0.0281, 0.0319, 0.0330, 0.0205, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:12:50,069 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138331.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:13:22,157 INFO [train.py:904] (7/8) Epoch 14, batch 6400, loss[loss=0.2096, simple_loss=0.2922, pruned_loss=0.06349, over 17241.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2975, pruned_loss=0.06547, over 3104015.34 frames. ], batch size: 44, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:14:09,373 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138383.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:14:23,619 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.151e+02 3.125e+02 3.740e+02 4.766e+02 1.292e+03, threshold=7.481e+02, percent-clipped=6.0 2023-04-30 00:14:37,174 INFO [train.py:904] (7/8) Epoch 14, batch 6450, loss[loss=0.2478, simple_loss=0.3387, pruned_loss=0.07842, over 16271.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2972, pruned_loss=0.06448, over 3111451.20 frames. ], batch size: 35, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:15:22,849 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-04-30 00:15:44,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9202, 4.7705, 4.9708, 5.1416, 5.2939, 4.7035, 5.2958, 5.2960], device='cuda:7'), covar=tensor([0.1648, 0.1167, 0.1400, 0.0622, 0.0506, 0.0739, 0.0493, 0.0509], device='cuda:7'), in_proj_covar=tensor([0.0557, 0.0689, 0.0827, 0.0705, 0.0534, 0.0550, 0.0554, 0.0652], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:15:54,742 INFO [train.py:904] (7/8) Epoch 14, batch 6500, loss[loss=0.2097, simple_loss=0.2929, pruned_loss=0.06325, over 16707.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2952, pruned_loss=0.06365, over 3114383.88 frames. ], batch size: 134, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:16:01,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7560, 1.7894, 1.5724, 1.5362, 1.8642, 1.5697, 1.7293, 1.8999], device='cuda:7'), covar=tensor([0.0151, 0.0207, 0.0303, 0.0256, 0.0157, 0.0197, 0.0174, 0.0157], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0209, 0.0204, 0.0204, 0.0210, 0.0210, 0.0215, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:16:55,391 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-30 00:16:59,452 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.094e+02 2.847e+02 3.325e+02 4.301e+02 7.980e+02, threshold=6.650e+02, percent-clipped=1.0 2023-04-30 00:17:12,092 INFO [train.py:904] (7/8) Epoch 14, batch 6550, loss[loss=0.2133, simple_loss=0.309, pruned_loss=0.05883, over 15196.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2977, pruned_loss=0.06477, over 3096515.78 frames. ], batch size: 190, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:18:22,249 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138548.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:18:25,821 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138551.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:18:26,461 INFO [train.py:904] (7/8) Epoch 14, batch 6600, loss[loss=0.1877, simple_loss=0.2877, pruned_loss=0.04385, over 16895.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2999, pruned_loss=0.06521, over 3106271.04 frames. ], batch size: 102, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:19:29,571 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.028e+02 2.801e+02 3.547e+02 4.436e+02 1.538e+03, threshold=7.095e+02, percent-clipped=5.0 2023-04-30 00:19:33,987 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=138596.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:19:34,199 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4763, 2.1103, 1.6600, 1.9465, 2.4462, 2.1589, 2.3919, 2.6369], device='cuda:7'), covar=tensor([0.0137, 0.0326, 0.0448, 0.0377, 0.0210, 0.0290, 0.0180, 0.0213], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0209, 0.0205, 0.0205, 0.0211, 0.0210, 0.0216, 0.0204], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:19:38,212 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=138599.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:19:42,719 INFO [train.py:904] (7/8) Epoch 14, batch 6650, loss[loss=0.1954, simple_loss=0.2761, pruned_loss=0.0573, over 16765.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.3001, pruned_loss=0.06573, over 3113337.77 frames. ], batch size: 76, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:20:19,737 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138626.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:20:58,175 INFO [train.py:904] (7/8) Epoch 14, batch 6700, loss[loss=0.24, simple_loss=0.31, pruned_loss=0.08502, over 11694.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2996, pruned_loss=0.06682, over 3088755.91 frames. ], batch size: 246, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:21:06,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9925, 2.5411, 2.6803, 1.8424, 2.7371, 2.8235, 2.4392, 2.3805], device='cuda:7'), covar=tensor([0.0775, 0.0221, 0.0206, 0.0983, 0.0107, 0.0207, 0.0466, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0105, 0.0089, 0.0139, 0.0073, 0.0113, 0.0124, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 00:21:09,628 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8033, 3.9664, 2.4223, 4.6783, 2.9478, 4.5563, 2.5766, 2.9780], device='cuda:7'), covar=tensor([0.0253, 0.0333, 0.1567, 0.0134, 0.0756, 0.0418, 0.1416, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0168, 0.0188, 0.0138, 0.0166, 0.0208, 0.0196, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 00:21:12,759 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6745, 1.7757, 2.1386, 2.5675, 2.5849, 2.9764, 1.7054, 2.8486], device='cuda:7'), covar=tensor([0.0163, 0.0394, 0.0292, 0.0238, 0.0230, 0.0141, 0.0465, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0174, 0.0159, 0.0165, 0.0172, 0.0131, 0.0177, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 00:21:40,477 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=138678.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:21:44,966 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138681.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:21:52,994 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:22:03,785 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.354e+02 4.045e+02 5.286e+02 1.404e+03, threshold=8.090e+02, percent-clipped=8.0 2023-04-30 00:22:09,179 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138697.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:22:15,582 INFO [train.py:904] (7/8) Epoch 14, batch 6750, loss[loss=0.1954, simple_loss=0.276, pruned_loss=0.05738, over 16682.00 frames. ], tot_loss[loss=0.2167, simple_loss=0.2991, pruned_loss=0.06711, over 3078708.95 frames. ], batch size: 76, lr: 4.84e-03, grad_scale: 4.0 2023-04-30 00:22:32,547 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2023-04-30 00:23:14,953 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138742.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:23:22,974 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138747.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:23:29,055 INFO [train.py:904] (7/8) Epoch 14, batch 6800, loss[loss=0.2286, simple_loss=0.3099, pruned_loss=0.0737, over 16283.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2988, pruned_loss=0.06721, over 3067112.45 frames. ], batch size: 165, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:23:38,503 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138758.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:24:34,695 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.812e+02 3.433e+02 4.059e+02 1.030e+03, threshold=6.866e+02, percent-clipped=2.0 2023-04-30 00:24:46,920 INFO [train.py:904] (7/8) Epoch 14, batch 6850, loss[loss=0.2083, simple_loss=0.3145, pruned_loss=0.05109, over 16510.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2998, pruned_loss=0.06696, over 3082663.91 frames. ], batch size: 75, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:26:02,485 INFO [train.py:904] (7/8) Epoch 14, batch 6900, loss[loss=0.2136, simple_loss=0.3039, pruned_loss=0.06164, over 16892.00 frames. ], tot_loss[loss=0.2185, simple_loss=0.3024, pruned_loss=0.06729, over 3075714.49 frames. ], batch size: 109, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:26:39,200 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5251, 2.6639, 2.3794, 4.2626, 3.0739, 4.0513, 1.4689, 2.8096], device='cuda:7'), covar=tensor([0.1438, 0.0746, 0.1254, 0.0156, 0.0280, 0.0401, 0.1639, 0.0869], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0163, 0.0185, 0.0161, 0.0202, 0.0209, 0.0187, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 00:27:03,019 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6086, 3.7024, 2.8240, 2.2111, 2.5730, 2.3063, 3.9230, 3.4360], device='cuda:7'), covar=tensor([0.2724, 0.0684, 0.1619, 0.2375, 0.2371, 0.1879, 0.0421, 0.1032], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0261, 0.0291, 0.0292, 0.0286, 0.0231, 0.0278, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:27:10,354 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.725e+02 2.815e+02 3.844e+02 4.824e+02 1.140e+03, threshold=7.689e+02, percent-clipped=2.0 2023-04-30 00:27:22,850 INFO [train.py:904] (7/8) Epoch 14, batch 6950, loss[loss=0.2255, simple_loss=0.3043, pruned_loss=0.07336, over 16371.00 frames. ], tot_loss[loss=0.221, simple_loss=0.3042, pruned_loss=0.06896, over 3079718.86 frames. ], batch size: 165, lr: 4.84e-03, grad_scale: 8.0 2023-04-30 00:27:33,833 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=138909.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:28:00,056 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138926.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:28:38,225 INFO [train.py:904] (7/8) Epoch 14, batch 7000, loss[loss=0.2232, simple_loss=0.3175, pruned_loss=0.06447, over 16476.00 frames. ], tot_loss[loss=0.2196, simple_loss=0.3037, pruned_loss=0.06778, over 3088445.63 frames. ], batch size: 68, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:28:51,399 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3899, 2.9344, 2.7171, 2.2302, 2.3042, 2.3004, 2.8662, 2.8800], device='cuda:7'), covar=tensor([0.2334, 0.0811, 0.1457, 0.2247, 0.2149, 0.1860, 0.0501, 0.1102], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0262, 0.0292, 0.0292, 0.0286, 0.0232, 0.0278, 0.0310], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:29:06,781 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=138970.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 00:29:12,851 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=138974.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:29:19,234 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=138978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:29:42,509 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.027e+02 2.958e+02 3.558e+02 4.270e+02 7.816e+02, threshold=7.117e+02, percent-clipped=1.0 2023-04-30 00:29:52,590 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3680, 3.2361, 2.5264, 2.0768, 2.3419, 2.0649, 3.3477, 3.0490], device='cuda:7'), covar=tensor([0.2908, 0.0953, 0.1934, 0.2458, 0.2619, 0.2274, 0.0586, 0.1274], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0262, 0.0291, 0.0292, 0.0286, 0.0232, 0.0278, 0.0310], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:29:55,156 INFO [train.py:904] (7/8) Epoch 14, batch 7050, loss[loss=0.2374, simple_loss=0.3145, pruned_loss=0.08016, over 15305.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.3039, pruned_loss=0.06722, over 3077312.47 frames. ], batch size: 190, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:30:14,392 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8329, 2.7253, 2.1525, 2.5313, 3.1900, 2.7909, 3.5482, 3.4347], device='cuda:7'), covar=tensor([0.0073, 0.0330, 0.0446, 0.0346, 0.0192, 0.0329, 0.0157, 0.0179], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0211, 0.0205, 0.0207, 0.0212, 0.0212, 0.0217, 0.0203], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:30:32,423 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5074, 3.6225, 4.1890, 1.8029, 4.3234, 4.3597, 3.0293, 3.1164], device='cuda:7'), covar=tensor([0.0817, 0.0216, 0.0146, 0.1251, 0.0048, 0.0101, 0.0411, 0.0413], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0103, 0.0089, 0.0137, 0.0071, 0.0112, 0.0123, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 00:30:33,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:30:50,684 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139037.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:30:58,274 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139042.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:31:13,428 INFO [train.py:904] (7/8) Epoch 14, batch 7100, loss[loss=0.1946, simple_loss=0.2884, pruned_loss=0.05036, over 16808.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3026, pruned_loss=0.06737, over 3068316.00 frames. ], batch size: 102, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:31:15,012 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139053.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:31:37,171 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-04-30 00:31:52,633 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 00:32:18,348 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.051e+02 2.828e+02 3.507e+02 4.047e+02 5.820e+02, threshold=7.014e+02, percent-clipped=0.0 2023-04-30 00:32:31,526 INFO [train.py:904] (7/8) Epoch 14, batch 7150, loss[loss=0.1814, simple_loss=0.2659, pruned_loss=0.04842, over 16522.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2997, pruned_loss=0.0662, over 3089516.79 frames. ], batch size: 68, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:33:47,882 INFO [train.py:904] (7/8) Epoch 14, batch 7200, loss[loss=0.1894, simple_loss=0.2764, pruned_loss=0.05118, over 16459.00 frames. ], tot_loss[loss=0.2138, simple_loss=0.2981, pruned_loss=0.06472, over 3079774.39 frames. ], batch size: 68, lr: 4.83e-03, grad_scale: 8.0 2023-04-30 00:34:55,355 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.878e+02 2.722e+02 3.151e+02 3.906e+02 1.100e+03, threshold=6.302e+02, percent-clipped=3.0 2023-04-30 00:35:07,126 INFO [train.py:904] (7/8) Epoch 14, batch 7250, loss[loss=0.2108, simple_loss=0.2888, pruned_loss=0.0664, over 16295.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2963, pruned_loss=0.06381, over 3084262.12 frames. ], batch size: 165, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:36:01,818 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7368, 5.0645, 5.3124, 5.0598, 5.1048, 5.6800, 5.1385, 4.8822], device='cuda:7'), covar=tensor([0.1067, 0.1724, 0.1782, 0.1774, 0.2272, 0.0889, 0.1546, 0.2472], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0523, 0.0572, 0.0442, 0.0592, 0.0595, 0.0452, 0.0599], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 00:36:01,941 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139238.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:36:22,186 INFO [train.py:904] (7/8) Epoch 14, batch 7300, loss[loss=0.201, simple_loss=0.2923, pruned_loss=0.05486, over 16762.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2959, pruned_loss=0.06334, over 3102588.72 frames. ], batch size: 76, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:36:43,295 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139265.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 00:37:00,527 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139276.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:37:03,853 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 00:37:20,446 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139289.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:37:29,634 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.078e+02 2.984e+02 3.521e+02 4.495e+02 8.579e+02, threshold=7.042e+02, percent-clipped=5.0 2023-04-30 00:37:37,680 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139299.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:37:40,550 INFO [train.py:904] (7/8) Epoch 14, batch 7350, loss[loss=0.2218, simple_loss=0.3028, pruned_loss=0.07042, over 16911.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2962, pruned_loss=0.06428, over 3088269.16 frames. ], batch size: 109, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:38:36,954 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139337.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:38:36,993 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139337.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 00:38:45,012 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139342.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:38:57,426 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139350.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:38:59,366 INFO [train.py:904] (7/8) Epoch 14, batch 7400, loss[loss=0.1899, simple_loss=0.2849, pruned_loss=0.0474, over 17122.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2977, pruned_loss=0.06472, over 3096986.13 frames. ], batch size: 47, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:39:01,661 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139353.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:39:52,989 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139385.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:40:00,779 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139390.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:40:08,332 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.794e+02 3.378e+02 4.039e+02 6.978e+02, threshold=6.757e+02, percent-clipped=0.0 2023-04-30 00:40:18,655 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139401.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:40:19,484 INFO [train.py:904] (7/8) Epoch 14, batch 7450, loss[loss=0.2183, simple_loss=0.3039, pruned_loss=0.06637, over 16953.00 frames. ], tot_loss[loss=0.2151, simple_loss=0.2985, pruned_loss=0.06582, over 3081352.07 frames. ], batch size: 109, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:41:43,395 INFO [train.py:904] (7/8) Epoch 14, batch 7500, loss[loss=0.2361, simple_loss=0.303, pruned_loss=0.08464, over 11575.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.2999, pruned_loss=0.06655, over 3058334.89 frames. ], batch size: 247, lr: 4.83e-03, grad_scale: 4.0 2023-04-30 00:41:53,186 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0547, 5.7915, 6.0120, 5.6971, 5.7811, 6.2747, 5.7265, 5.5617], device='cuda:7'), covar=tensor([0.0832, 0.1603, 0.1754, 0.1635, 0.2051, 0.0821, 0.1466, 0.2274], device='cuda:7'), in_proj_covar=tensor([0.0377, 0.0525, 0.0582, 0.0447, 0.0599, 0.0603, 0.0457, 0.0608], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 00:42:53,291 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.056e+02 3.028e+02 3.537e+02 4.351e+02 9.459e+02, threshold=7.073e+02, percent-clipped=3.0 2023-04-30 00:43:02,953 INFO [train.py:904] (7/8) Epoch 14, batch 7550, loss[loss=0.1865, simple_loss=0.27, pruned_loss=0.05145, over 16757.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2983, pruned_loss=0.06628, over 3061340.04 frames. ], batch size: 83, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:43:14,380 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-30 00:44:19,048 INFO [train.py:904] (7/8) Epoch 14, batch 7600, loss[loss=0.2284, simple_loss=0.305, pruned_loss=0.07586, over 16422.00 frames. ], tot_loss[loss=0.2144, simple_loss=0.2969, pruned_loss=0.06596, over 3080184.10 frames. ], batch size: 68, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:44:28,874 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3474, 3.3396, 3.3906, 3.4723, 3.5035, 3.2726, 3.4839, 3.5429], device='cuda:7'), covar=tensor([0.1106, 0.0851, 0.0963, 0.0542, 0.0650, 0.2005, 0.0925, 0.0773], device='cuda:7'), in_proj_covar=tensor([0.0550, 0.0683, 0.0815, 0.0697, 0.0532, 0.0546, 0.0553, 0.0651], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:44:39,679 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139565.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 00:44:49,023 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139571.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:06,329 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139582.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:23,478 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139594.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:45:23,660 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7711, 1.2937, 1.6184, 1.5698, 1.7483, 1.9152, 1.5239, 1.7132], device='cuda:7'), covar=tensor([0.0192, 0.0317, 0.0165, 0.0223, 0.0220, 0.0145, 0.0326, 0.0107], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0175, 0.0158, 0.0164, 0.0173, 0.0130, 0.0177, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 00:45:24,243 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.745e+02 3.435e+02 4.599e+02 7.932e+02, threshold=6.869e+02, percent-clipped=2.0 2023-04-30 00:45:34,173 INFO [train.py:904] (7/8) Epoch 14, batch 7650, loss[loss=0.2028, simple_loss=0.2878, pruned_loss=0.05889, over 16401.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2982, pruned_loss=0.06716, over 3070667.77 frames. ], batch size: 68, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:45:50,728 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139613.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:19,717 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139632.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 00:46:19,895 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:37,231 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:39,470 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139645.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:46:49,456 INFO [train.py:904] (7/8) Epoch 14, batch 7700, loss[loss=0.2077, simple_loss=0.2926, pruned_loss=0.06147, over 16684.00 frames. ], tot_loss[loss=0.2161, simple_loss=0.2981, pruned_loss=0.06702, over 3067202.90 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:47:13,212 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-30 00:47:49,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6029, 2.7174, 2.3142, 3.9534, 2.9410, 3.9634, 1.3682, 2.7025], device='cuda:7'), covar=tensor([0.1372, 0.0703, 0.1271, 0.0167, 0.0250, 0.0374, 0.1640, 0.0855], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0164, 0.0186, 0.0162, 0.0204, 0.0210, 0.0188, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 00:47:57,376 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.201e+02 3.391e+02 3.997e+02 4.581e+02 8.153e+02, threshold=7.994e+02, percent-clipped=6.0 2023-04-30 00:48:06,813 INFO [train.py:904] (7/8) Epoch 14, batch 7750, loss[loss=0.2365, simple_loss=0.3217, pruned_loss=0.07563, over 16926.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2984, pruned_loss=0.06681, over 3071079.24 frames. ], batch size: 109, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:49:24,618 INFO [train.py:904] (7/8) Epoch 14, batch 7800, loss[loss=0.1884, simple_loss=0.2775, pruned_loss=0.04961, over 16687.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2992, pruned_loss=0.06727, over 3088476.75 frames. ], batch size: 62, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:49:37,695 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-30 00:49:51,615 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139769.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:50:22,508 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139789.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:50:32,560 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.773e+02 2.870e+02 3.646e+02 4.608e+02 1.262e+03, threshold=7.291e+02, percent-clipped=3.0 2023-04-30 00:50:36,369 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.84 vs. limit=5.0 2023-04-30 00:50:41,159 INFO [train.py:904] (7/8) Epoch 14, batch 7850, loss[loss=0.2059, simple_loss=0.308, pruned_loss=0.05192, over 16901.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.2995, pruned_loss=0.06704, over 3076593.92 frames. ], batch size: 90, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:50:46,086 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139804.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:50:47,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3615, 5.6784, 5.4088, 5.4156, 5.0282, 4.9698, 5.1176, 5.7665], device='cuda:7'), covar=tensor([0.1063, 0.0776, 0.1004, 0.0851, 0.0912, 0.0784, 0.1055, 0.0828], device='cuda:7'), in_proj_covar=tensor([0.0584, 0.0721, 0.0595, 0.0519, 0.0454, 0.0470, 0.0604, 0.0551], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:50:49,020 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139806.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:24,785 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139830.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:53,683 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:53,702 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139850.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:51:56,194 INFO [train.py:904] (7/8) Epoch 14, batch 7900, loss[loss=0.2338, simple_loss=0.3192, pruned_loss=0.07422, over 16679.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2986, pruned_loss=0.06605, over 3099235.45 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:52:08,825 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-30 00:52:15,864 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139865.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:52:18,324 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139867.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:52:28,049 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4706, 3.4442, 2.5366, 2.1728, 2.3116, 2.0740, 3.5262, 3.1410], device='cuda:7'), covar=tensor([0.2927, 0.0804, 0.1944, 0.2559, 0.2904, 0.2225, 0.0574, 0.1255], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0262, 0.0291, 0.0292, 0.0286, 0.0233, 0.0275, 0.0310], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 00:53:01,584 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139894.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:53:03,658 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.073e+02 2.924e+02 3.560e+02 4.104e+02 6.335e+02, threshold=7.120e+02, percent-clipped=0.0 2023-04-30 00:53:12,903 INFO [train.py:904] (7/8) Epoch 14, batch 7950, loss[loss=0.1956, simple_loss=0.2843, pruned_loss=0.05346, over 16386.00 frames. ], tot_loss[loss=0.2165, simple_loss=0.2995, pruned_loss=0.06674, over 3091201.85 frames. ], batch size: 146, lr: 4.82e-03, grad_scale: 4.0 2023-04-30 00:53:28,129 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=139911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:53:32,616 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9023, 2.8897, 3.0423, 1.6157, 3.0867, 3.3009, 2.6593, 2.3668], device='cuda:7'), covar=tensor([0.1141, 0.0206, 0.0177, 0.1402, 0.0095, 0.0164, 0.0499, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0104, 0.0090, 0.0139, 0.0072, 0.0112, 0.0124, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 00:53:37,400 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 00:53:51,412 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139927.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:53:59,091 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139932.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 00:54:08,420 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=139938.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:54:13,671 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:54:18,373 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=139945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:54:28,357 INFO [train.py:904] (7/8) Epoch 14, batch 8000, loss[loss=0.2402, simple_loss=0.3066, pruned_loss=0.08686, over 11097.00 frames. ], tot_loss[loss=0.218, simple_loss=0.3003, pruned_loss=0.06785, over 3078610.39 frames. ], batch size: 248, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:55:09,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9373, 5.1652, 4.9717, 4.9180, 4.6951, 4.6398, 4.5895, 5.2470], device='cuda:7'), covar=tensor([0.1053, 0.0765, 0.0876, 0.0764, 0.0758, 0.0870, 0.1063, 0.0797], device='cuda:7'), in_proj_covar=tensor([0.0573, 0.0706, 0.0583, 0.0508, 0.0447, 0.0461, 0.0590, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:55:11,102 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139980.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:55:30,359 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=139993.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:55:34,108 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 3.021e+02 3.514e+02 4.508e+02 9.326e+02, threshold=7.028e+02, percent-clipped=1.0 2023-04-30 00:55:46,025 INFO [train.py:904] (7/8) Epoch 14, batch 8050, loss[loss=0.1948, simple_loss=0.2867, pruned_loss=0.0514, over 16684.00 frames. ], tot_loss[loss=0.2178, simple_loss=0.3003, pruned_loss=0.06766, over 3076345.32 frames. ], batch size: 134, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:55:47,248 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4457, 1.3941, 2.0469, 2.3204, 2.4404, 2.6432, 1.4845, 2.6531], device='cuda:7'), covar=tensor([0.0153, 0.0512, 0.0244, 0.0236, 0.0225, 0.0155, 0.0532, 0.0093], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0175, 0.0159, 0.0165, 0.0174, 0.0132, 0.0178, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 00:55:54,724 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-30 00:56:52,705 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8898, 5.1632, 4.9898, 4.9000, 4.6785, 4.6323, 4.5937, 5.2689], device='cuda:7'), covar=tensor([0.0987, 0.0708, 0.0806, 0.0679, 0.0721, 0.0860, 0.0929, 0.0709], device='cuda:7'), in_proj_covar=tensor([0.0579, 0.0712, 0.0590, 0.0513, 0.0451, 0.0464, 0.0596, 0.0545], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 00:56:59,044 INFO [train.py:904] (7/8) Epoch 14, batch 8100, loss[loss=0.2036, simple_loss=0.2917, pruned_loss=0.05772, over 16816.00 frames. ], tot_loss[loss=0.2163, simple_loss=0.2994, pruned_loss=0.06662, over 3087493.26 frames. ], batch size: 76, lr: 4.82e-03, grad_scale: 8.0 2023-04-30 00:58:06,574 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.001e+02 3.052e+02 3.743e+02 5.058e+02 1.059e+03, threshold=7.487e+02, percent-clipped=6.0 2023-04-30 00:58:15,162 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 00:58:17,064 INFO [train.py:904] (7/8) Epoch 14, batch 8150, loss[loss=0.2081, simple_loss=0.2898, pruned_loss=0.0632, over 15368.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2972, pruned_loss=0.06609, over 3073150.31 frames. ], batch size: 190, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 00:58:18,900 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140103.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:58:29,469 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140110.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:58:39,296 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 00:58:52,675 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:17,687 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9480, 1.9837, 2.4647, 2.8796, 2.8465, 3.3140, 1.9075, 3.1865], device='cuda:7'), covar=tensor([0.0189, 0.0425, 0.0279, 0.0263, 0.0231, 0.0131, 0.0461, 0.0110], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0176, 0.0160, 0.0166, 0.0176, 0.0133, 0.0179, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 00:59:23,825 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:34,244 INFO [train.py:904] (7/8) Epoch 14, batch 8200, loss[loss=0.2297, simple_loss=0.3126, pruned_loss=0.07337, over 16737.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2943, pruned_loss=0.06499, over 3097927.87 frames. ], batch size: 124, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 00:59:47,468 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140160.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:50,574 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140162.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 00:59:53,842 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140164.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:00:05,615 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140171.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:00:26,413 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4003, 3.3302, 3.4495, 3.5429, 3.5905, 3.2848, 3.5434, 3.6145], device='cuda:7'), covar=tensor([0.1221, 0.0956, 0.1141, 0.0617, 0.0645, 0.2036, 0.0833, 0.0819], device='cuda:7'), in_proj_covar=tensor([0.0547, 0.0676, 0.0808, 0.0691, 0.0529, 0.0541, 0.0550, 0.0647], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:00:44,848 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.704e+02 2.531e+02 3.267e+02 4.112e+02 9.606e+02, threshold=6.535e+02, percent-clipped=4.0 2023-04-30 01:00:55,632 INFO [train.py:904] (7/8) Epoch 14, batch 8250, loss[loss=0.2057, simple_loss=0.2945, pruned_loss=0.05845, over 16707.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2927, pruned_loss=0.0623, over 3074083.52 frames. ], batch size: 134, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:01:02,348 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140206.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:01:37,289 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140227.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:01:55,217 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140238.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:02:16,388 INFO [train.py:904] (7/8) Epoch 14, batch 8300, loss[loss=0.1835, simple_loss=0.2776, pruned_loss=0.04472, over 16737.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2902, pruned_loss=0.05915, over 3080583.46 frames. ], batch size: 124, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:02:20,871 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5692, 2.5830, 2.2711, 2.4355, 2.9330, 2.6774, 3.2702, 3.2276], device='cuda:7'), covar=tensor([0.0098, 0.0327, 0.0406, 0.0348, 0.0255, 0.0298, 0.0188, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0207, 0.0203, 0.0203, 0.0208, 0.0208, 0.0209, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:02:29,800 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140259.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:02:55,101 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140275.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:02:58,762 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-30 01:03:12,858 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:03:27,888 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.340e+02 2.728e+02 3.506e+02 7.480e+02, threshold=5.455e+02, percent-clipped=2.0 2023-04-30 01:03:30,716 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 01:03:36,681 INFO [train.py:904] (7/8) Epoch 14, batch 8350, loss[loss=0.2082, simple_loss=0.298, pruned_loss=0.05924, over 16868.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2895, pruned_loss=0.05738, over 3077088.16 frames. ], batch size: 116, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:04:08,077 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140320.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:04:27,989 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6021, 3.6466, 2.7681, 2.1096, 2.2592, 2.2321, 3.7134, 3.2703], device='cuda:7'), covar=tensor([0.2713, 0.0527, 0.1716, 0.2635, 0.2778, 0.2096, 0.0395, 0.1143], device='cuda:7'), in_proj_covar=tensor([0.0302, 0.0253, 0.0282, 0.0283, 0.0275, 0.0227, 0.0267, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:04:57,946 INFO [train.py:904] (7/8) Epoch 14, batch 8400, loss[loss=0.1691, simple_loss=0.2539, pruned_loss=0.04214, over 12220.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2866, pruned_loss=0.05518, over 3069628.11 frames. ], batch size: 246, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:05:16,506 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 01:05:54,786 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140387.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:05:56,943 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0313, 2.8196, 2.7938, 1.9588, 2.5725, 2.0247, 2.8096, 2.9485], device='cuda:7'), covar=tensor([0.0321, 0.0856, 0.0592, 0.2099, 0.1012, 0.1304, 0.0645, 0.0786], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0147, 0.0156, 0.0143, 0.0135, 0.0123, 0.0135, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 01:06:06,343 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4544, 3.0328, 3.2594, 1.9822, 2.7549, 2.2983, 2.9215, 3.1581], device='cuda:7'), covar=tensor([0.0425, 0.0865, 0.0455, 0.1899, 0.0831, 0.0911, 0.0942, 0.0893], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0147, 0.0156, 0.0143, 0.0135, 0.0123, 0.0135, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 01:06:08,149 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.349e+02 2.839e+02 3.572e+02 6.227e+02, threshold=5.678e+02, percent-clipped=3.0 2023-04-30 01:06:18,713 INFO [train.py:904] (7/8) Epoch 14, batch 8450, loss[loss=0.1574, simple_loss=0.2539, pruned_loss=0.03038, over 16445.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2847, pruned_loss=0.05349, over 3069658.34 frames. ], batch size: 68, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:06:56,635 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140425.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:24,124 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 01:07:26,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1029, 2.0383, 2.1439, 3.5703, 2.0322, 2.3345, 2.2059, 2.1846], device='cuda:7'), covar=tensor([0.1074, 0.3726, 0.2811, 0.0503, 0.4361, 0.2562, 0.3479, 0.3664], device='cuda:7'), in_proj_covar=tensor([0.0364, 0.0404, 0.0336, 0.0311, 0.0412, 0.0461, 0.0367, 0.0469], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:07:27,344 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140445.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:32,374 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140448.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:37,964 INFO [train.py:904] (7/8) Epoch 14, batch 8500, loss[loss=0.1767, simple_loss=0.2687, pruned_loss=0.04235, over 16628.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2808, pruned_loss=0.05074, over 3077023.38 frames. ], batch size: 62, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:07:50,060 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140459.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:51,452 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:07:54,731 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140462.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:01,382 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140466.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:13,347 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140473.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:38,709 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140489.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:46,172 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140493.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:08:51,333 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.295e+02 2.729e+02 3.409e+02 8.118e+02, threshold=5.459e+02, percent-clipped=4.0 2023-04-30 01:08:57,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1473, 3.3015, 3.2896, 2.3910, 3.0428, 3.3713, 3.2342, 1.8583], device='cuda:7'), covar=tensor([0.0427, 0.0044, 0.0043, 0.0303, 0.0082, 0.0063, 0.0063, 0.0448], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0070, 0.0071, 0.0126, 0.0082, 0.0093, 0.0081, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 01:09:01,493 INFO [train.py:904] (7/8) Epoch 14, batch 8550, loss[loss=0.1857, simple_loss=0.2879, pruned_loss=0.04179, over 16830.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2787, pruned_loss=0.04963, over 3069255.70 frames. ], batch size: 102, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:09:12,149 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140506.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:09:15,778 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140508.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:09:17,279 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140509.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:09:19,130 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140510.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:10:08,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5540, 1.6651, 2.0801, 2.5407, 2.4728, 2.8099, 1.7561, 2.7827], device='cuda:7'), covar=tensor([0.0166, 0.0449, 0.0283, 0.0259, 0.0259, 0.0153, 0.0433, 0.0118], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0174, 0.0157, 0.0162, 0.0173, 0.0130, 0.0176, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 01:10:39,284 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140550.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:10:41,526 INFO [train.py:904] (7/8) Epoch 14, batch 8600, loss[loss=0.2121, simple_loss=0.2873, pruned_loss=0.06842, over 12919.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2786, pruned_loss=0.04865, over 3057756.18 frames. ], batch size: 250, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:10:46,550 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140554.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:10:54,468 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6736, 3.6463, 2.8018, 2.1268, 2.3399, 2.3183, 3.8947, 3.2811], device='cuda:7'), covar=tensor([0.2654, 0.0688, 0.1684, 0.2905, 0.2765, 0.1869, 0.0392, 0.1256], device='cuda:7'), in_proj_covar=tensor([0.0303, 0.0254, 0.0284, 0.0284, 0.0275, 0.0228, 0.0268, 0.0300], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:11:19,628 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:12:06,770 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 2.341e+02 2.953e+02 3.525e+02 9.042e+02, threshold=5.906e+02, percent-clipped=3.0 2023-04-30 01:12:18,457 INFO [train.py:904] (7/8) Epoch 14, batch 8650, loss[loss=0.1841, simple_loss=0.2701, pruned_loss=0.0491, over 12250.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2762, pruned_loss=0.04692, over 3051325.04 frames. ], batch size: 247, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:12:52,267 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:14:05,329 INFO [train.py:904] (7/8) Epoch 14, batch 8700, loss[loss=0.181, simple_loss=0.273, pruned_loss=0.04451, over 16897.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2735, pruned_loss=0.04548, over 3054292.11 frames. ], batch size: 102, lr: 4.81e-03, grad_scale: 8.0 2023-04-30 01:15:12,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-04-30 01:15:24,092 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.380e+02 2.814e+02 3.680e+02 6.881e+02, threshold=5.627e+02, percent-clipped=2.0 2023-04-30 01:15:38,753 INFO [train.py:904] (7/8) Epoch 14, batch 8750, loss[loss=0.1709, simple_loss=0.2556, pruned_loss=0.04312, over 11957.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2734, pruned_loss=0.04494, over 3063101.09 frames. ], batch size: 247, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:16:31,551 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140723.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:16:45,776 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4170, 3.5137, 2.1392, 3.8864, 2.5008, 3.8644, 2.1644, 2.7854], device='cuda:7'), covar=tensor([0.0259, 0.0303, 0.1501, 0.0151, 0.0892, 0.0355, 0.1482, 0.0714], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0159, 0.0182, 0.0131, 0.0161, 0.0196, 0.0189, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-30 01:16:49,152 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9988, 3.8646, 4.0727, 4.1611, 4.2793, 3.8685, 4.2472, 4.3023], device='cuda:7'), covar=tensor([0.1478, 0.1030, 0.1224, 0.0662, 0.0529, 0.1404, 0.0564, 0.0542], device='cuda:7'), in_proj_covar=tensor([0.0537, 0.0659, 0.0784, 0.0682, 0.0515, 0.0532, 0.0536, 0.0631], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:17:11,101 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140743.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:17:28,587 INFO [train.py:904] (7/8) Epoch 14, batch 8800, loss[loss=0.1877, simple_loss=0.274, pruned_loss=0.05069, over 15311.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2718, pruned_loss=0.0437, over 3082327.10 frames. ], batch size: 190, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:17:43,133 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140759.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:17:51,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4366, 3.0960, 2.6922, 2.1889, 2.2523, 2.2367, 2.9184, 2.8793], device='cuda:7'), covar=tensor([0.2462, 0.0828, 0.1494, 0.2452, 0.2287, 0.1955, 0.0447, 0.1278], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0252, 0.0281, 0.0281, 0.0271, 0.0226, 0.0265, 0.0296], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:17:57,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140766.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:18:36,015 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140784.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:19:00,062 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.211e+02 2.894e+02 3.689e+02 7.406e+02, threshold=5.789e+02, percent-clipped=2.0 2023-04-30 01:19:12,361 INFO [train.py:904] (7/8) Epoch 14, batch 8850, loss[loss=0.1569, simple_loss=0.2512, pruned_loss=0.03135, over 12356.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2742, pruned_loss=0.04335, over 3063516.48 frames. ], batch size: 246, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:19:23,526 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140807.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:19:27,243 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140809.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:19:37,857 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140814.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:20:44,022 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140845.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:20:58,229 INFO [train.py:904] (7/8) Epoch 14, batch 8900, loss[loss=0.1911, simple_loss=0.2837, pruned_loss=0.04926, over 16626.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2753, pruned_loss=0.04273, over 3082627.34 frames. ], batch size: 134, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:21:25,586 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=140865.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:21:26,070 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-30 01:21:34,818 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140870.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:22:46,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.342e+02 2.775e+02 3.236e+02 6.448e+02, threshold=5.549e+02, percent-clipped=1.0 2023-04-30 01:23:01,553 INFO [train.py:904] (7/8) Epoch 14, batch 8950, loss[loss=0.1953, simple_loss=0.2761, pruned_loss=0.05727, over 12398.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2748, pruned_loss=0.04333, over 3064687.72 frames. ], batch size: 247, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:23:29,072 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=140915.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:23:32,901 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140917.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:24:38,415 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=140947.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:24:48,139 INFO [train.py:904] (7/8) Epoch 14, batch 9000, loss[loss=0.1514, simple_loss=0.2462, pruned_loss=0.02828, over 16884.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.272, pruned_loss=0.04221, over 3073195.78 frames. ], batch size: 102, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:24:48,140 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 01:24:58,094 INFO [train.py:938] (7/8) Epoch 14, validation: loss=0.1514, simple_loss=0.2553, pruned_loss=0.0237, over 944034.00 frames. 2023-04-30 01:24:58,095 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 01:25:11,102 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-30 01:25:21,318 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=140963.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:25:47,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8823, 4.6633, 4.8943, 5.0406, 5.2172, 4.6141, 5.1973, 5.2044], device='cuda:7'), covar=tensor([0.1614, 0.1182, 0.1322, 0.0651, 0.0514, 0.0869, 0.0487, 0.0586], device='cuda:7'), in_proj_covar=tensor([0.0533, 0.0656, 0.0775, 0.0676, 0.0510, 0.0528, 0.0530, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:25:52,579 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=140978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:26:29,792 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.199e+02 2.701e+02 3.272e+02 7.252e+02, threshold=5.402e+02, percent-clipped=4.0 2023-04-30 01:26:39,933 INFO [train.py:904] (7/8) Epoch 14, batch 9050, loss[loss=0.1798, simple_loss=0.263, pruned_loss=0.04832, over 16263.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2724, pruned_loss=0.04219, over 3101784.40 frames. ], batch size: 165, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:26:55,093 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141008.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:28:08,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141043.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:28:25,843 INFO [train.py:904] (7/8) Epoch 14, batch 9100, loss[loss=0.1757, simple_loss=0.2583, pruned_loss=0.04659, over 12035.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2718, pruned_loss=0.04256, over 3088923.90 frames. ], batch size: 248, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:29:30,791 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141079.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:29:59,037 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141091.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:30:13,180 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.366e+02 2.876e+02 3.628e+02 8.857e+02, threshold=5.752e+02, percent-clipped=2.0 2023-04-30 01:30:22,849 INFO [train.py:904] (7/8) Epoch 14, batch 9150, loss[loss=0.1577, simple_loss=0.2555, pruned_loss=0.02994, over 16848.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2722, pruned_loss=0.0425, over 3086827.81 frames. ], batch size: 42, lr: 4.80e-03, grad_scale: 4.0 2023-04-30 01:31:51,352 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141143.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:31:54,456 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:32:04,842 INFO [train.py:904] (7/8) Epoch 14, batch 9200, loss[loss=0.1769, simple_loss=0.267, pruned_loss=0.04341, over 16703.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2675, pruned_loss=0.04142, over 3097802.65 frames. ], batch size: 134, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:32:29,701 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:32:29,732 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:33:23,833 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141193.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:33:31,525 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.509e+02 2.979e+02 4.006e+02 1.027e+03, threshold=5.958e+02, percent-clipped=7.0 2023-04-30 01:33:34,817 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4157, 3.0176, 2.6795, 2.1761, 2.1701, 2.2155, 2.9539, 2.8491], device='cuda:7'), covar=tensor([0.2553, 0.0775, 0.1476, 0.2694, 0.2697, 0.1997, 0.0488, 0.1236], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0252, 0.0283, 0.0282, 0.0268, 0.0227, 0.0264, 0.0295], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:33:41,002 INFO [train.py:904] (7/8) Epoch 14, batch 9250, loss[loss=0.1765, simple_loss=0.2676, pruned_loss=0.04268, over 16380.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2677, pruned_loss=0.04183, over 3080962.88 frames. ], batch size: 146, lr: 4.80e-03, grad_scale: 8.0 2023-04-30 01:33:46,107 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141204.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:34:03,406 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141213.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:35:30,129 INFO [train.py:904] (7/8) Epoch 14, batch 9300, loss[loss=0.1794, simple_loss=0.2745, pruned_loss=0.04214, over 16720.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2665, pruned_loss=0.04138, over 3068644.79 frames. ], batch size: 57, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:36:20,471 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141273.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:37:05,692 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.032e+02 2.564e+02 3.502e+02 9.246e+02, threshold=5.129e+02, percent-clipped=2.0 2023-04-30 01:37:14,525 INFO [train.py:904] (7/8) Epoch 14, batch 9350, loss[loss=0.1666, simple_loss=0.2641, pruned_loss=0.03453, over 16888.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2658, pruned_loss=0.04101, over 3077654.62 frames. ], batch size: 102, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:37:17,106 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141303.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:37:40,871 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8466, 1.3973, 1.6729, 1.7500, 1.8612, 1.8778, 1.6103, 1.8666], device='cuda:7'), covar=tensor([0.0198, 0.0341, 0.0171, 0.0254, 0.0229, 0.0163, 0.0326, 0.0113], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0174, 0.0156, 0.0160, 0.0171, 0.0128, 0.0175, 0.0119], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 01:38:09,343 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141329.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:38:47,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5910, 3.6549, 3.4272, 3.1591, 3.2372, 3.5291, 3.3206, 3.3402], device='cuda:7'), covar=tensor([0.0521, 0.0491, 0.0267, 0.0239, 0.0529, 0.0456, 0.1083, 0.0396], device='cuda:7'), in_proj_covar=tensor([0.0245, 0.0332, 0.0293, 0.0272, 0.0304, 0.0319, 0.0202, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:38:54,843 INFO [train.py:904] (7/8) Epoch 14, batch 9400, loss[loss=0.1519, simple_loss=0.2409, pruned_loss=0.0315, over 12413.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2649, pruned_loss=0.0406, over 3065538.45 frames. ], batch size: 248, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:39:49,461 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141379.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:40:12,118 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141390.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:40:25,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.208e+02 2.590e+02 3.171e+02 6.862e+02, threshold=5.180e+02, percent-clipped=4.0 2023-04-30 01:40:33,418 INFO [train.py:904] (7/8) Epoch 14, batch 9450, loss[loss=0.1591, simple_loss=0.2527, pruned_loss=0.03271, over 16496.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2667, pruned_loss=0.04076, over 3065502.59 frames. ], batch size: 68, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:41:03,378 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141417.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 01:41:24,441 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141427.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:41:54,156 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141441.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:42:14,326 INFO [train.py:904] (7/8) Epoch 14, batch 9500, loss[loss=0.1721, simple_loss=0.2701, pruned_loss=0.03701, over 16650.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2658, pruned_loss=0.04001, over 3058457.27 frames. ], batch size: 76, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:42:43,593 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141465.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:43:08,477 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141478.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 01:43:47,720 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.299e+02 2.699e+02 3.093e+02 4.944e+02, threshold=5.398e+02, percent-clipped=0.0 2023-04-30 01:43:55,342 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141499.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:43:59,699 INFO [train.py:904] (7/8) Epoch 14, batch 9550, loss[loss=0.1838, simple_loss=0.2795, pruned_loss=0.04402, over 15229.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.266, pruned_loss=0.04051, over 3058600.54 frames. ], batch size: 190, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:44:01,055 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141502.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:44:24,362 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141513.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:44:51,693 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-30 01:45:38,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5491, 4.3488, 4.5918, 4.7395, 4.9147, 4.3823, 4.9106, 4.9025], device='cuda:7'), covar=tensor([0.1832, 0.1176, 0.1524, 0.0687, 0.0554, 0.1001, 0.0487, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0531, 0.0655, 0.0772, 0.0670, 0.0506, 0.0523, 0.0528, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 01:45:41,084 INFO [train.py:904] (7/8) Epoch 14, batch 9600, loss[loss=0.1987, simple_loss=0.2972, pruned_loss=0.05008, over 16244.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2684, pruned_loss=0.04185, over 3055393.71 frames. ], batch size: 165, lr: 4.79e-03, grad_scale: 8.0 2023-04-30 01:46:00,717 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7999, 3.7284, 3.8962, 3.7213, 3.9149, 4.2890, 4.0240, 3.7429], device='cuda:7'), covar=tensor([0.1889, 0.2356, 0.2413, 0.2489, 0.2774, 0.1567, 0.1457, 0.2848], device='cuda:7'), in_proj_covar=tensor([0.0351, 0.0500, 0.0550, 0.0421, 0.0568, 0.0578, 0.0435, 0.0573], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 01:46:22,464 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141573.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:46:24,699 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-30 01:47:18,947 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-04-30 01:47:19,267 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.296e+02 2.650e+02 3.286e+02 9.159e+02, threshold=5.299e+02, percent-clipped=3.0 2023-04-30 01:47:30,288 INFO [train.py:904] (7/8) Epoch 14, batch 9650, loss[loss=0.1793, simple_loss=0.2627, pruned_loss=0.04796, over 12184.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.27, pruned_loss=0.04176, over 3055804.83 frames. ], batch size: 248, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:47:34,815 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141603.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:48:16,938 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141621.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:49:17,046 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141651.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:49:17,849 INFO [train.py:904] (7/8) Epoch 14, batch 9700, loss[loss=0.1675, simple_loss=0.2667, pruned_loss=0.03416, over 16829.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.269, pruned_loss=0.04184, over 3046639.78 frames. ], batch size: 102, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:50:27,998 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141685.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:50:53,325 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.310e+02 2.782e+02 3.506e+02 1.114e+03, threshold=5.564e+02, percent-clipped=5.0 2023-04-30 01:51:00,301 INFO [train.py:904] (7/8) Epoch 14, batch 9750, loss[loss=0.1725, simple_loss=0.2669, pruned_loss=0.03905, over 16375.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2674, pruned_loss=0.04133, over 3067560.96 frames. ], batch size: 146, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:51:32,952 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 01:51:54,429 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0680, 5.4695, 5.6545, 5.4029, 5.5161, 6.0029, 5.5360, 5.2904], device='cuda:7'), covar=tensor([0.0814, 0.1908, 0.1913, 0.1876, 0.2104, 0.0912, 0.1374, 0.2248], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0496, 0.0549, 0.0419, 0.0564, 0.0576, 0.0432, 0.0569], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 01:52:05,984 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2023-04-30 01:52:38,206 INFO [train.py:904] (7/8) Epoch 14, batch 9800, loss[loss=0.1794, simple_loss=0.2885, pruned_loss=0.03509, over 17167.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2682, pruned_loss=0.04075, over 3073466.03 frames. ], batch size: 46, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:53:18,671 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141773.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 01:53:51,649 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 01:54:13,245 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=141797.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:54:13,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.347e+02 2.793e+02 3.311e+02 9.229e+02, threshold=5.586e+02, percent-clipped=3.0 2023-04-30 01:54:17,289 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141799.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:54:21,354 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-04-30 01:54:21,459 INFO [train.py:904] (7/8) Epoch 14, batch 9850, loss[loss=0.1777, simple_loss=0.2747, pruned_loss=0.04038, over 16133.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.269, pruned_loss=0.04065, over 3055608.68 frames. ], batch size: 165, lr: 4.79e-03, grad_scale: 4.0 2023-04-30 01:54:31,072 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1439, 3.6319, 3.5644, 2.4758, 3.2021, 3.6179, 3.4101, 1.9776], device='cuda:7'), covar=tensor([0.0472, 0.0031, 0.0039, 0.0332, 0.0098, 0.0061, 0.0069, 0.0468], device='cuda:7'), in_proj_covar=tensor([0.0127, 0.0069, 0.0071, 0.0126, 0.0084, 0.0092, 0.0081, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 01:56:05,076 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=141847.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:56:14,794 INFO [train.py:904] (7/8) Epoch 14, batch 9900, loss[loss=0.1717, simple_loss=0.278, pruned_loss=0.03275, over 16766.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2694, pruned_loss=0.04058, over 3046204.99 frames. ], batch size: 83, lr: 4.78e-03, grad_scale: 4.0 2023-04-30 01:58:03,751 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.142e+02 2.825e+02 3.311e+02 8.255e+02, threshold=5.650e+02, percent-clipped=2.0 2023-04-30 01:58:13,946 INFO [train.py:904] (7/8) Epoch 14, batch 9950, loss[loss=0.1793, simple_loss=0.2875, pruned_loss=0.03558, over 16844.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2718, pruned_loss=0.04103, over 3047912.68 frames. ], batch size: 96, lr: 4.78e-03, grad_scale: 4.0 2023-04-30 01:58:48,424 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141916.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 01:59:21,077 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=141928.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:00:16,386 INFO [train.py:904] (7/8) Epoch 14, batch 10000, loss[loss=0.1937, simple_loss=0.2711, pruned_loss=0.05814, over 12757.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2699, pruned_loss=0.04062, over 3058371.25 frames. ], batch size: 250, lr: 4.78e-03, grad_scale: 8.0 2023-04-30 02:00:17,602 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1366, 4.3782, 4.0728, 3.7832, 3.4225, 4.3071, 4.1309, 3.9254], device='cuda:7'), covar=tensor([0.0961, 0.0627, 0.0561, 0.0487, 0.1791, 0.0601, 0.0670, 0.0744], device='cuda:7'), in_proj_covar=tensor([0.0240, 0.0326, 0.0286, 0.0267, 0.0295, 0.0311, 0.0196, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-30 02:00:43,670 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1812, 4.1670, 4.0625, 3.5803, 4.1327, 1.7139, 3.8976, 3.8483], device='cuda:7'), covar=tensor([0.0091, 0.0080, 0.0147, 0.0219, 0.0079, 0.2517, 0.0126, 0.0205], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0124, 0.0167, 0.0150, 0.0141, 0.0185, 0.0156, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:00:49,257 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.68 vs. limit=5.0 2023-04-30 02:01:07,118 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141977.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:01:26,018 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=141985.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:01:34,598 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=141989.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:01:50,044 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2036, 5.7267, 5.8453, 5.6002, 5.6396, 6.2013, 5.6864, 5.4332], device='cuda:7'), covar=tensor([0.0658, 0.1481, 0.1498, 0.1675, 0.2121, 0.0792, 0.1107, 0.2013], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0486, 0.0539, 0.0413, 0.0556, 0.0565, 0.0425, 0.0557], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:01:50,850 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.061e+02 2.551e+02 3.100e+02 8.241e+02, threshold=5.102e+02, percent-clipped=1.0 2023-04-30 02:02:01,165 INFO [train.py:904] (7/8) Epoch 14, batch 10050, loss[loss=0.1872, simple_loss=0.2821, pruned_loss=0.04616, over 16348.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2702, pruned_loss=0.04038, over 3066751.92 frames. ], batch size: 146, lr: 4.78e-03, grad_scale: 8.0 2023-04-30 02:02:05,873 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 02:03:00,120 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-30 02:03:01,574 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142033.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:03:34,651 INFO [train.py:904] (7/8) Epoch 14, batch 10100, loss[loss=0.1738, simple_loss=0.267, pruned_loss=0.04029, over 16367.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2708, pruned_loss=0.04103, over 3060383.71 frames. ], batch size: 146, lr: 4.78e-03, grad_scale: 8.0 2023-04-30 02:04:05,431 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1003, 4.0863, 3.9921, 3.0321, 4.0442, 1.5922, 3.7354, 3.6793], device='cuda:7'), covar=tensor([0.0175, 0.0166, 0.0215, 0.0559, 0.0167, 0.3290, 0.0207, 0.0358], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0124, 0.0166, 0.0149, 0.0140, 0.0184, 0.0155, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:04:19,933 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142073.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:04:49,709 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142097.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:04:50,458 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.492e+02 2.995e+02 3.648e+02 7.218e+02, threshold=5.990e+02, percent-clipped=8.0 2023-04-30 02:05:19,922 INFO [train.py:904] (7/8) Epoch 15, batch 0, loss[loss=0.2367, simple_loss=0.3044, pruned_loss=0.08449, over 16804.00 frames. ], tot_loss[loss=0.2367, simple_loss=0.3044, pruned_loss=0.08449, over 16804.00 frames. ], batch size: 102, lr: 4.62e-03, grad_scale: 8.0 2023-04-30 02:05:19,922 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 02:05:27,353 INFO [train.py:938] (7/8) Epoch 15, validation: loss=0.1501, simple_loss=0.2536, pruned_loss=0.02333, over 944034.00 frames. 2023-04-30 02:05:27,354 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 02:05:53,996 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142121.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:06:27,854 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:06:37,646 INFO [train.py:904] (7/8) Epoch 15, batch 50, loss[loss=0.1821, simple_loss=0.2529, pruned_loss=0.05568, over 16799.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.28, pruned_loss=0.05917, over 761049.40 frames. ], batch size: 102, lr: 4.62e-03, grad_scale: 2.0 2023-04-30 02:07:24,368 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-04-30 02:07:44,552 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.683e+02 3.202e+02 3.967e+02 8.381e+02, threshold=6.403e+02, percent-clipped=5.0 2023-04-30 02:07:47,926 INFO [train.py:904] (7/8) Epoch 15, batch 100, loss[loss=0.1767, simple_loss=0.2631, pruned_loss=0.04516, over 16602.00 frames. ], tot_loss[loss=0.1937, simple_loss=0.2755, pruned_loss=0.0559, over 1325913.11 frames. ], batch size: 68, lr: 4.62e-03, grad_scale: 2.0 2023-04-30 02:08:56,697 INFO [train.py:904] (7/8) Epoch 15, batch 150, loss[loss=0.1591, simple_loss=0.2514, pruned_loss=0.03342, over 17230.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2731, pruned_loss=0.05292, over 1772662.96 frames. ], batch size: 45, lr: 4.62e-03, grad_scale: 2.0 2023-04-30 02:09:11,314 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9642, 1.8675, 2.3915, 2.7551, 2.7035, 2.8372, 2.0796, 3.0771], device='cuda:7'), covar=tensor([0.0160, 0.0397, 0.0286, 0.0270, 0.0246, 0.0205, 0.0396, 0.0119], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0176, 0.0160, 0.0164, 0.0173, 0.0130, 0.0177, 0.0120], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 02:09:25,352 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142272.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:09:42,958 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142284.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:10:04,597 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.354e+02 2.662e+02 3.206e+02 6.929e+02, threshold=5.325e+02, percent-clipped=1.0 2023-04-30 02:10:07,774 INFO [train.py:904] (7/8) Epoch 15, batch 200, loss[loss=0.1983, simple_loss=0.2711, pruned_loss=0.06273, over 16653.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2726, pruned_loss=0.05258, over 2118827.24 frames. ], batch size: 89, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:10:26,177 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-30 02:11:17,109 INFO [train.py:904] (7/8) Epoch 15, batch 250, loss[loss=0.1893, simple_loss=0.2683, pruned_loss=0.05518, over 16479.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2699, pruned_loss=0.05156, over 2389192.46 frames. ], batch size: 68, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:12:07,550 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9214, 5.3124, 5.4376, 5.2849, 5.3300, 5.8794, 5.4015, 5.1185], device='cuda:7'), covar=tensor([0.1047, 0.1753, 0.2197, 0.1959, 0.2425, 0.0980, 0.1386, 0.2377], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0518, 0.0570, 0.0439, 0.0593, 0.0594, 0.0450, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:12:22,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.346e+02 2.797e+02 3.590e+02 6.679e+02, threshold=5.594e+02, percent-clipped=6.0 2023-04-30 02:12:25,440 INFO [train.py:904] (7/8) Epoch 15, batch 300, loss[loss=0.166, simple_loss=0.2624, pruned_loss=0.03487, over 17134.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2674, pruned_loss=0.05012, over 2595814.21 frames. ], batch size: 46, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:13:36,237 INFO [train.py:904] (7/8) Epoch 15, batch 350, loss[loss=0.1755, simple_loss=0.257, pruned_loss=0.04701, over 16745.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2642, pruned_loss=0.04853, over 2757172.26 frames. ], batch size: 39, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:13:45,319 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-04-30 02:14:42,902 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.497e+02 2.253e+02 2.722e+02 3.272e+02 5.602e+02, threshold=5.443e+02, percent-clipped=1.0 2023-04-30 02:14:45,198 INFO [train.py:904] (7/8) Epoch 15, batch 400, loss[loss=0.1559, simple_loss=0.2566, pruned_loss=0.02758, over 17029.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2628, pruned_loss=0.04789, over 2888692.79 frames. ], batch size: 50, lr: 4.61e-03, grad_scale: 4.0 2023-04-30 02:15:41,941 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0124, 4.9777, 5.5029, 5.5289, 5.5564, 5.1807, 5.0888, 4.8445], device='cuda:7'), covar=tensor([0.0372, 0.0575, 0.0481, 0.0486, 0.0494, 0.0387, 0.1005, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0379, 0.0380, 0.0357, 0.0422, 0.0403, 0.0487, 0.0320], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 02:15:54,200 INFO [train.py:904] (7/8) Epoch 15, batch 450, loss[loss=0.1593, simple_loss=0.2414, pruned_loss=0.03867, over 16896.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2607, pruned_loss=0.04674, over 2990780.06 frames. ], batch size: 90, lr: 4.61e-03, grad_scale: 4.0 2023-04-30 02:16:23,438 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142572.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:16:24,723 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142573.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:16:32,708 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4777, 3.5555, 3.8915, 1.9070, 3.9757, 3.9281, 3.1682, 2.8546], device='cuda:7'), covar=tensor([0.0773, 0.0174, 0.0113, 0.1130, 0.0063, 0.0156, 0.0350, 0.0423], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0102, 0.0087, 0.0139, 0.0071, 0.0113, 0.0124, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 02:16:41,581 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=142584.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:16:42,933 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4361, 2.4739, 2.0131, 2.3807, 2.8895, 2.6040, 3.1326, 3.1051], device='cuda:7'), covar=tensor([0.0140, 0.0398, 0.0524, 0.0410, 0.0248, 0.0366, 0.0255, 0.0234], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0219, 0.0211, 0.0212, 0.0218, 0.0218, 0.0222, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:16:50,555 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3341, 3.7820, 4.2241, 2.2717, 3.2856, 2.7131, 3.8418, 3.8935], device='cuda:7'), covar=tensor([0.0307, 0.0763, 0.0390, 0.1811, 0.0716, 0.0873, 0.0639, 0.1012], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0149, 0.0160, 0.0147, 0.0138, 0.0126, 0.0138, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:17:03,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.083e+02 2.627e+02 3.131e+02 6.454e+02, threshold=5.253e+02, percent-clipped=1.0 2023-04-30 02:17:05,230 INFO [train.py:904] (7/8) Epoch 15, batch 500, loss[loss=0.158, simple_loss=0.2408, pruned_loss=0.03759, over 16804.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2599, pruned_loss=0.04607, over 3070358.46 frames. ], batch size: 124, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:17:28,782 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142620.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:17:46,433 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=142632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:17:49,290 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142634.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:18:13,797 INFO [train.py:904] (7/8) Epoch 15, batch 550, loss[loss=0.1948, simple_loss=0.2675, pruned_loss=0.06101, over 16273.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2585, pruned_loss=0.04571, over 3125293.08 frames. ], batch size: 165, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:18:20,789 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 02:19:22,094 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.274e+02 2.667e+02 3.434e+02 1.165e+03, threshold=5.335e+02, percent-clipped=9.0 2023-04-30 02:19:23,207 INFO [train.py:904] (7/8) Epoch 15, batch 600, loss[loss=0.1781, simple_loss=0.2643, pruned_loss=0.04596, over 16706.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2584, pruned_loss=0.04573, over 3177782.97 frames. ], batch size: 62, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:19:53,056 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=142724.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:20:32,862 INFO [train.py:904] (7/8) Epoch 15, batch 650, loss[loss=0.143, simple_loss=0.2221, pruned_loss=0.03195, over 17013.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2576, pruned_loss=0.04578, over 3211910.21 frames. ], batch size: 41, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:21:17,838 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=142785.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:21:40,879 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.547e+02 2.460e+02 3.000e+02 3.809e+02 2.810e+03, threshold=6.001e+02, percent-clipped=15.0 2023-04-30 02:21:42,135 INFO [train.py:904] (7/8) Epoch 15, batch 700, loss[loss=0.1601, simple_loss=0.2414, pruned_loss=0.03944, over 16764.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2575, pruned_loss=0.04506, over 3245394.32 frames. ], batch size: 39, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:22:11,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2654, 3.4758, 3.6118, 3.6094, 3.6176, 3.4409, 3.4540, 3.4776], device='cuda:7'), covar=tensor([0.0382, 0.0682, 0.0440, 0.0415, 0.0492, 0.0505, 0.0744, 0.0505], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0392, 0.0392, 0.0367, 0.0435, 0.0416, 0.0502, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 02:22:21,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1122, 5.6564, 5.8254, 5.5696, 5.6185, 6.2217, 5.7980, 5.5126], device='cuda:7'), covar=tensor([0.0868, 0.2084, 0.2279, 0.2100, 0.2606, 0.1111, 0.1442, 0.2285], device='cuda:7'), in_proj_covar=tensor([0.0384, 0.0542, 0.0598, 0.0460, 0.0618, 0.0620, 0.0470, 0.0612], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:22:50,295 INFO [train.py:904] (7/8) Epoch 15, batch 750, loss[loss=0.1324, simple_loss=0.2239, pruned_loss=0.02044, over 17214.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2582, pruned_loss=0.04542, over 3265486.98 frames. ], batch size: 45, lr: 4.61e-03, grad_scale: 2.0 2023-04-30 02:23:57,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.171e+02 2.529e+02 2.994e+02 6.163e+02, threshold=5.058e+02, percent-clipped=1.0 2023-04-30 02:23:59,537 INFO [train.py:904] (7/8) Epoch 15, batch 800, loss[loss=0.1247, simple_loss=0.2096, pruned_loss=0.0199, over 16779.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2574, pruned_loss=0.04527, over 3271152.14 frames. ], batch size: 39, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:24:36,439 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=142929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:24:37,911 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9028, 3.3097, 2.9954, 5.1275, 4.3539, 4.6560, 2.0963, 3.4962], device='cuda:7'), covar=tensor([0.1249, 0.0623, 0.1007, 0.0146, 0.0240, 0.0329, 0.1295, 0.0632], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0164, 0.0185, 0.0164, 0.0195, 0.0211, 0.0189, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:24:55,999 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9799, 4.0341, 2.4161, 4.6277, 3.0766, 4.5885, 2.6616, 3.2477], device='cuda:7'), covar=tensor([0.0258, 0.0340, 0.1489, 0.0191, 0.0759, 0.0462, 0.1401, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0170, 0.0192, 0.0145, 0.0169, 0.0211, 0.0199, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:25:08,646 INFO [train.py:904] (7/8) Epoch 15, batch 850, loss[loss=0.154, simple_loss=0.2431, pruned_loss=0.03246, over 17207.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2562, pruned_loss=0.0446, over 3285097.14 frames. ], batch size: 45, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:25:39,078 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1710, 2.4915, 1.5796, 2.0860, 2.8833, 2.6308, 3.1237, 3.0669], device='cuda:7'), covar=tensor([0.0200, 0.0369, 0.0663, 0.0459, 0.0222, 0.0329, 0.0270, 0.0231], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0220, 0.0211, 0.0213, 0.0219, 0.0219, 0.0225, 0.0211], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:26:15,123 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.170e+02 2.499e+02 2.974e+02 6.190e+02, threshold=4.998e+02, percent-clipped=2.0 2023-04-30 02:26:16,300 INFO [train.py:904] (7/8) Epoch 15, batch 900, loss[loss=0.1658, simple_loss=0.2587, pruned_loss=0.03648, over 17101.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2552, pruned_loss=0.04319, over 3304167.78 frames. ], batch size: 53, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:26:26,029 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7831, 2.4826, 1.9129, 2.2714, 2.8612, 2.6552, 2.8853, 2.9431], device='cuda:7'), covar=tensor([0.0192, 0.0344, 0.0460, 0.0417, 0.0189, 0.0303, 0.0235, 0.0233], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0220, 0.0211, 0.0213, 0.0219, 0.0219, 0.0225, 0.0211], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:26:32,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3239, 5.2411, 5.0833, 4.5639, 4.6617, 5.1382, 5.1480, 4.7450], device='cuda:7'), covar=tensor([0.0531, 0.0449, 0.0290, 0.0349, 0.1199, 0.0432, 0.0310, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0366, 0.0318, 0.0300, 0.0334, 0.0349, 0.0219, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:27:24,778 INFO [train.py:904] (7/8) Epoch 15, batch 950, loss[loss=0.1665, simple_loss=0.2504, pruned_loss=0.04124, over 16851.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2559, pruned_loss=0.0439, over 3307372.16 frames. ], batch size: 42, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:28:02,447 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143080.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:28:23,207 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.58 vs. limit=5.0 2023-04-30 02:28:30,230 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.127e+02 2.534e+02 2.926e+02 6.813e+02, threshold=5.068e+02, percent-clipped=2.0 2023-04-30 02:28:31,473 INFO [train.py:904] (7/8) Epoch 15, batch 1000, loss[loss=0.1587, simple_loss=0.2624, pruned_loss=0.0275, over 17134.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2556, pruned_loss=0.04434, over 3308564.55 frames. ], batch size: 47, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:28:52,171 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4706, 1.6249, 2.0660, 2.2711, 2.4154, 2.3994, 1.6904, 2.4218], device='cuda:7'), covar=tensor([0.0155, 0.0414, 0.0252, 0.0234, 0.0226, 0.0205, 0.0425, 0.0127], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0180, 0.0164, 0.0169, 0.0177, 0.0134, 0.0181, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:29:09,548 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6818, 4.6910, 4.8791, 4.7265, 4.7138, 5.3554, 4.8996, 4.5520], device='cuda:7'), covar=tensor([0.1255, 0.2059, 0.2299, 0.2168, 0.2883, 0.1226, 0.1562, 0.2504], device='cuda:7'), in_proj_covar=tensor([0.0385, 0.0545, 0.0603, 0.0465, 0.0624, 0.0627, 0.0477, 0.0618], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:29:31,620 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7297, 1.8297, 2.2262, 2.5097, 2.6358, 2.4620, 1.7917, 2.7599], device='cuda:7'), covar=tensor([0.0136, 0.0397, 0.0271, 0.0239, 0.0224, 0.0237, 0.0460, 0.0110], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0180, 0.0165, 0.0169, 0.0178, 0.0135, 0.0181, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:29:41,406 INFO [train.py:904] (7/8) Epoch 15, batch 1050, loss[loss=0.1562, simple_loss=0.2476, pruned_loss=0.03241, over 17220.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2556, pruned_loss=0.04483, over 3311982.03 frames. ], batch size: 45, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:30:01,434 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5096, 3.3362, 2.5810, 2.1756, 2.3097, 2.1397, 3.2552, 3.0160], device='cuda:7'), covar=tensor([0.2625, 0.0792, 0.1851, 0.2658, 0.2633, 0.2306, 0.0593, 0.1370], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0262, 0.0293, 0.0291, 0.0281, 0.0237, 0.0276, 0.0313], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:30:13,740 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 02:30:46,963 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.203e+02 2.568e+02 3.063e+02 1.237e+03, threshold=5.135e+02, percent-clipped=3.0 2023-04-30 02:30:49,005 INFO [train.py:904] (7/8) Epoch 15, batch 1100, loss[loss=0.1708, simple_loss=0.257, pruned_loss=0.04225, over 16464.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2552, pruned_loss=0.04425, over 3319367.65 frames. ], batch size: 68, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:31:26,643 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143229.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:31:49,797 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 02:31:58,395 INFO [train.py:904] (7/8) Epoch 15, batch 1150, loss[loss=0.1779, simple_loss=0.2509, pruned_loss=0.05243, over 16833.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2551, pruned_loss=0.04384, over 3327885.69 frames. ], batch size: 102, lr: 4.60e-03, grad_scale: 4.0 2023-04-30 02:32:26,380 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-04-30 02:32:34,543 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=143277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:33:07,203 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.162e+02 2.475e+02 3.140e+02 4.983e+02, threshold=4.950e+02, percent-clipped=0.0 2023-04-30 02:33:08,293 INFO [train.py:904] (7/8) Epoch 15, batch 1200, loss[loss=0.1549, simple_loss=0.2488, pruned_loss=0.03053, over 17114.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2539, pruned_loss=0.04299, over 3334365.23 frames. ], batch size: 47, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:34:16,295 INFO [train.py:904] (7/8) Epoch 15, batch 1250, loss[loss=0.1595, simple_loss=0.2473, pruned_loss=0.03589, over 17136.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2536, pruned_loss=0.04413, over 3322878.89 frames. ], batch size: 48, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:34:57,314 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=143380.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:35:08,614 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-30 02:35:25,448 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143400.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:35:26,173 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.199e+02 2.560e+02 2.965e+02 4.851e+02, threshold=5.120e+02, percent-clipped=0.0 2023-04-30 02:35:27,820 INFO [train.py:904] (7/8) Epoch 15, batch 1300, loss[loss=0.1595, simple_loss=0.2542, pruned_loss=0.03239, over 17147.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2531, pruned_loss=0.04382, over 3321982.17 frames. ], batch size: 47, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:36:03,858 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=143428.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:36:37,211 INFO [train.py:904] (7/8) Epoch 15, batch 1350, loss[loss=0.1531, simple_loss=0.2505, pruned_loss=0.02788, over 17123.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2535, pruned_loss=0.04368, over 3323502.05 frames. ], batch size: 47, lr: 4.60e-03, grad_scale: 8.0 2023-04-30 02:36:50,779 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143461.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 02:36:59,317 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0366, 5.0468, 4.8736, 4.3025, 4.9070, 1.8653, 4.6245, 4.7447], device='cuda:7'), covar=tensor([0.0086, 0.0075, 0.0153, 0.0359, 0.0088, 0.2621, 0.0121, 0.0189], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0135, 0.0181, 0.0165, 0.0154, 0.0195, 0.0169, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:37:06,634 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=143473.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:37:45,744 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.266e+02 2.683e+02 3.217e+02 5.542e+02, threshold=5.366e+02, percent-clipped=1.0 2023-04-30 02:37:47,564 INFO [train.py:904] (7/8) Epoch 15, batch 1400, loss[loss=0.1565, simple_loss=0.2343, pruned_loss=0.03941, over 15446.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2535, pruned_loss=0.04438, over 3314551.15 frames. ], batch size: 191, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:38:32,157 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=143534.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 02:38:33,308 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9724, 2.0945, 2.4418, 2.8833, 2.7034, 3.3881, 2.0675, 3.2152], device='cuda:7'), covar=tensor([0.0222, 0.0425, 0.0303, 0.0257, 0.0285, 0.0148, 0.0447, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0184, 0.0168, 0.0173, 0.0181, 0.0136, 0.0184, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:38:55,992 INFO [train.py:904] (7/8) Epoch 15, batch 1450, loss[loss=0.1514, simple_loss=0.228, pruned_loss=0.03745, over 16285.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2521, pruned_loss=0.04339, over 3310532.36 frames. ], batch size: 165, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:39:56,103 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7931, 4.1285, 3.1235, 2.2937, 2.7166, 2.6241, 4.3968, 3.6004], device='cuda:7'), covar=tensor([0.2673, 0.0589, 0.1550, 0.2477, 0.2616, 0.1865, 0.0366, 0.1226], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0264, 0.0294, 0.0291, 0.0285, 0.0239, 0.0277, 0.0314], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:40:04,880 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7076, 1.9400, 2.2280, 2.5515, 2.6612, 2.6258, 1.8677, 2.8351], device='cuda:7'), covar=tensor([0.0147, 0.0357, 0.0275, 0.0206, 0.0219, 0.0220, 0.0394, 0.0105], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0183, 0.0168, 0.0172, 0.0181, 0.0136, 0.0183, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:40:05,553 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.183e+02 2.517e+02 3.188e+02 7.484e+02, threshold=5.034e+02, percent-clipped=1.0 2023-04-30 02:40:06,049 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3559, 5.3849, 5.1167, 4.5847, 5.1587, 1.9748, 4.8799, 5.0388], device='cuda:7'), covar=tensor([0.0068, 0.0054, 0.0148, 0.0362, 0.0078, 0.2426, 0.0124, 0.0177], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0135, 0.0181, 0.0167, 0.0154, 0.0195, 0.0170, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:40:06,735 INFO [train.py:904] (7/8) Epoch 15, batch 1500, loss[loss=0.1557, simple_loss=0.2426, pruned_loss=0.0344, over 16813.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2521, pruned_loss=0.04409, over 3313949.08 frames. ], batch size: 42, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:40:55,221 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 02:41:13,200 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 02:41:14,540 INFO [train.py:904] (7/8) Epoch 15, batch 1550, loss[loss=0.1985, simple_loss=0.2679, pruned_loss=0.06454, over 16880.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2533, pruned_loss=0.04521, over 3322524.73 frames. ], batch size: 90, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:41:36,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9228, 4.6785, 4.8344, 5.1306, 5.3088, 4.6549, 5.2263, 5.2566], device='cuda:7'), covar=tensor([0.1638, 0.1450, 0.2170, 0.0896, 0.0667, 0.1008, 0.0690, 0.0743], device='cuda:7'), in_proj_covar=tensor([0.0605, 0.0748, 0.0897, 0.0769, 0.0572, 0.0596, 0.0603, 0.0713], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:41:45,784 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.01 vs. limit=5.0 2023-04-30 02:41:58,220 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 02:42:22,872 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.352e+02 2.733e+02 3.255e+02 5.583e+02, threshold=5.465e+02, percent-clipped=1.0 2023-04-30 02:42:24,072 INFO [train.py:904] (7/8) Epoch 15, batch 1600, loss[loss=0.1724, simple_loss=0.2607, pruned_loss=0.04207, over 17227.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2559, pruned_loss=0.04563, over 3325839.49 frames. ], batch size: 44, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:43:35,420 INFO [train.py:904] (7/8) Epoch 15, batch 1650, loss[loss=0.1845, simple_loss=0.2615, pruned_loss=0.05378, over 16435.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2576, pruned_loss=0.04627, over 3322420.86 frames. ], batch size: 75, lr: 4.59e-03, grad_scale: 8.0 2023-04-30 02:43:40,913 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143756.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 02:44:46,128 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.448e+02 2.826e+02 3.842e+02 9.668e+02, threshold=5.653e+02, percent-clipped=8.0 2023-04-30 02:44:46,143 INFO [train.py:904] (7/8) Epoch 15, batch 1700, loss[loss=0.1921, simple_loss=0.2771, pruned_loss=0.05358, over 17221.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2597, pruned_loss=0.04697, over 3312450.66 frames. ], batch size: 46, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:45:22,436 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=143829.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:45:54,347 INFO [train.py:904] (7/8) Epoch 15, batch 1750, loss[loss=0.1773, simple_loss=0.2716, pruned_loss=0.04144, over 17049.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2604, pruned_loss=0.04686, over 3311209.09 frames. ], batch size: 55, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:46:05,182 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 02:46:23,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3728, 3.5343, 3.8673, 2.0727, 3.1303, 2.4391, 3.7904, 3.8340], device='cuda:7'), covar=tensor([0.0267, 0.0858, 0.0490, 0.1873, 0.0775, 0.1001, 0.0558, 0.0921], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0151, 0.0161, 0.0146, 0.0138, 0.0125, 0.0138, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:46:44,631 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-30 02:47:05,603 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.250e+02 2.638e+02 3.028e+02 6.204e+02, threshold=5.275e+02, percent-clipped=1.0 2023-04-30 02:47:05,618 INFO [train.py:904] (7/8) Epoch 15, batch 1800, loss[loss=0.2019, simple_loss=0.2832, pruned_loss=0.06034, over 16807.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2618, pruned_loss=0.04726, over 3295021.06 frames. ], batch size: 102, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:47:33,484 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8479, 2.3791, 2.5166, 4.7085, 2.3774, 2.8646, 2.5907, 2.7114], device='cuda:7'), covar=tensor([0.1016, 0.3647, 0.2546, 0.0389, 0.3969, 0.2587, 0.3216, 0.3429], device='cuda:7'), in_proj_covar=tensor([0.0379, 0.0417, 0.0349, 0.0327, 0.0423, 0.0480, 0.0383, 0.0488], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:48:15,545 INFO [train.py:904] (7/8) Epoch 15, batch 1850, loss[loss=0.2134, simple_loss=0.2844, pruned_loss=0.07119, over 16704.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2627, pruned_loss=0.04786, over 3281460.49 frames. ], batch size: 134, lr: 4.59e-03, grad_scale: 2.0 2023-04-30 02:48:20,008 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4502, 4.3314, 4.2895, 3.9932, 4.0413, 4.3694, 4.1231, 4.1095], device='cuda:7'), covar=tensor([0.0561, 0.0648, 0.0293, 0.0294, 0.0805, 0.0485, 0.0574, 0.0604], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0378, 0.0328, 0.0309, 0.0344, 0.0358, 0.0224, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:49:04,438 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7328, 2.8696, 2.4767, 4.7486, 3.6153, 4.3845, 1.5710, 2.8490], device='cuda:7'), covar=tensor([0.1459, 0.0802, 0.1317, 0.0184, 0.0304, 0.0413, 0.1665, 0.0974], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0164, 0.0185, 0.0167, 0.0198, 0.0212, 0.0188, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:49:30,767 INFO [train.py:904] (7/8) Epoch 15, batch 1900, loss[loss=0.1441, simple_loss=0.2311, pruned_loss=0.02856, over 17049.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2627, pruned_loss=0.04752, over 3290670.69 frames. ], batch size: 41, lr: 4.59e-03, grad_scale: 2.0 2023-04-30 02:49:31,848 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.211e+02 2.636e+02 2.995e+02 6.158e+02, threshold=5.272e+02, percent-clipped=2.0 2023-04-30 02:50:39,898 INFO [train.py:904] (7/8) Epoch 15, batch 1950, loss[loss=0.1572, simple_loss=0.2384, pruned_loss=0.03799, over 15849.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.262, pruned_loss=0.04678, over 3288366.21 frames. ], batch size: 35, lr: 4.59e-03, grad_scale: 2.0 2023-04-30 02:50:46,724 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144056.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 02:51:44,751 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5282, 5.9813, 5.7273, 5.7923, 5.2788, 5.2642, 5.3919, 6.1309], device='cuda:7'), covar=tensor([0.1390, 0.0912, 0.1120, 0.0743, 0.0997, 0.0636, 0.1174, 0.0844], device='cuda:7'), in_proj_covar=tensor([0.0618, 0.0768, 0.0632, 0.0551, 0.0486, 0.0492, 0.0642, 0.0582], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:51:49,552 INFO [train.py:904] (7/8) Epoch 15, batch 2000, loss[loss=0.1537, simple_loss=0.2364, pruned_loss=0.0355, over 17264.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2613, pruned_loss=0.04604, over 3288721.11 frames. ], batch size: 43, lr: 4.59e-03, grad_scale: 4.0 2023-04-30 02:51:51,364 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.196e+02 2.584e+02 2.966e+02 4.475e+02, threshold=5.169e+02, percent-clipped=0.0 2023-04-30 02:51:52,658 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=144104.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 02:52:27,381 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=144129.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 02:52:58,028 INFO [train.py:904] (7/8) Epoch 15, batch 2050, loss[loss=0.1841, simple_loss=0.2762, pruned_loss=0.04598, over 17120.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2627, pruned_loss=0.04675, over 3288199.97 frames. ], batch size: 48, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:53:32,910 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=144177.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 02:53:54,753 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6145, 3.8461, 4.0030, 2.8247, 3.5355, 4.0113, 3.7232, 2.3580], device='cuda:7'), covar=tensor([0.0423, 0.0191, 0.0043, 0.0319, 0.0093, 0.0091, 0.0079, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0076, 0.0075, 0.0131, 0.0088, 0.0099, 0.0086, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 02:54:07,934 INFO [train.py:904] (7/8) Epoch 15, batch 2100, loss[loss=0.2003, simple_loss=0.273, pruned_loss=0.06373, over 16850.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2628, pruned_loss=0.0467, over 3298521.50 frames. ], batch size: 109, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:54:08,981 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.513e+02 2.931e+02 3.819e+02 1.829e+03, threshold=5.862e+02, percent-clipped=10.0 2023-04-30 02:54:44,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8831, 4.0127, 2.5166, 4.5608, 2.8531, 4.5577, 2.5155, 3.1604], device='cuda:7'), covar=tensor([0.0266, 0.0349, 0.1481, 0.0243, 0.0904, 0.0495, 0.1484, 0.0729], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0171, 0.0192, 0.0150, 0.0171, 0.0213, 0.0200, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:55:17,941 INFO [train.py:904] (7/8) Epoch 15, batch 2150, loss[loss=0.1818, simple_loss=0.26, pruned_loss=0.05176, over 16774.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2633, pruned_loss=0.0468, over 3304919.42 frames. ], batch size: 96, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:55:27,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2565, 3.4957, 3.3636, 2.2161, 2.9204, 2.5090, 3.7111, 3.7666], device='cuda:7'), covar=tensor([0.0238, 0.0793, 0.0640, 0.1722, 0.0790, 0.0913, 0.0480, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0154, 0.0162, 0.0148, 0.0140, 0.0126, 0.0139, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:56:25,367 INFO [train.py:904] (7/8) Epoch 15, batch 2200, loss[loss=0.1637, simple_loss=0.2498, pruned_loss=0.0388, over 17193.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2637, pruned_loss=0.04771, over 3302324.43 frames. ], batch size: 46, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:56:27,077 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.322e+02 2.712e+02 3.377e+02 6.214e+02, threshold=5.423e+02, percent-clipped=1.0 2023-04-30 02:57:05,582 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5498, 3.8346, 4.1938, 2.3764, 3.3098, 2.7422, 4.0572, 4.0635], device='cuda:7'), covar=tensor([0.0272, 0.0870, 0.0428, 0.1693, 0.0750, 0.0868, 0.0518, 0.0930], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0154, 0.0162, 0.0147, 0.0140, 0.0126, 0.0139, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 02:57:36,233 INFO [train.py:904] (7/8) Epoch 15, batch 2250, loss[loss=0.2024, simple_loss=0.2826, pruned_loss=0.0611, over 16729.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2643, pruned_loss=0.04715, over 3317001.09 frames. ], batch size: 83, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:58:46,693 INFO [train.py:904] (7/8) Epoch 15, batch 2300, loss[loss=0.1707, simple_loss=0.2533, pruned_loss=0.04403, over 15955.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2635, pruned_loss=0.04643, over 3321748.30 frames. ], batch size: 35, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 02:58:47,879 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.350e+02 2.907e+02 3.506e+02 6.150e+02, threshold=5.814e+02, percent-clipped=3.0 2023-04-30 02:58:59,092 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144411.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 02:59:45,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0249, 5.3347, 5.0971, 5.0990, 4.8030, 4.7179, 4.7133, 5.4158], device='cuda:7'), covar=tensor([0.1159, 0.0943, 0.1033, 0.0769, 0.0971, 0.1066, 0.1247, 0.0916], device='cuda:7'), in_proj_covar=tensor([0.0623, 0.0776, 0.0632, 0.0553, 0.0488, 0.0494, 0.0643, 0.0587], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 02:59:53,224 INFO [train.py:904] (7/8) Epoch 15, batch 2350, loss[loss=0.1929, simple_loss=0.2703, pruned_loss=0.05776, over 16452.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2642, pruned_loss=0.04682, over 3329467.30 frames. ], batch size: 146, lr: 4.58e-03, grad_scale: 4.0 2023-04-30 03:00:20,645 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144472.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:01:02,762 INFO [train.py:904] (7/8) Epoch 15, batch 2400, loss[loss=0.1848, simple_loss=0.2719, pruned_loss=0.04882, over 16115.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2653, pruned_loss=0.04687, over 3329077.58 frames. ], batch size: 35, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:01:04,727 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.391e+02 2.805e+02 3.317e+02 7.772e+02, threshold=5.609e+02, percent-clipped=1.0 2023-04-30 03:01:14,982 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6126, 2.3762, 1.8577, 2.1091, 2.7622, 2.5147, 2.8570, 2.8536], device='cuda:7'), covar=tensor([0.0197, 0.0355, 0.0465, 0.0432, 0.0208, 0.0293, 0.0219, 0.0246], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0223, 0.0213, 0.0214, 0.0223, 0.0222, 0.0229, 0.0216], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:02:10,319 INFO [train.py:904] (7/8) Epoch 15, batch 2450, loss[loss=0.1735, simple_loss=0.2484, pruned_loss=0.04933, over 16427.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2658, pruned_loss=0.04689, over 3330391.75 frames. ], batch size: 146, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:02:26,690 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4236, 2.9178, 2.5867, 2.2226, 2.2135, 2.1744, 2.9297, 2.7959], device='cuda:7'), covar=tensor([0.2293, 0.0718, 0.1448, 0.2074, 0.2206, 0.1937, 0.0487, 0.1066], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0263, 0.0294, 0.0291, 0.0285, 0.0237, 0.0277, 0.0316], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:02:52,647 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 03:03:17,680 INFO [train.py:904] (7/8) Epoch 15, batch 2500, loss[loss=0.2071, simple_loss=0.28, pruned_loss=0.06712, over 16293.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2655, pruned_loss=0.04686, over 3321958.87 frames. ], batch size: 165, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:03:18,673 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.258e+02 2.678e+02 3.424e+02 5.626e+02, threshold=5.355e+02, percent-clipped=1.0 2023-04-30 03:04:26,858 INFO [train.py:904] (7/8) Epoch 15, batch 2550, loss[loss=0.1862, simple_loss=0.2779, pruned_loss=0.04723, over 16706.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2667, pruned_loss=0.04753, over 3319048.27 frames. ], batch size: 62, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:05:11,641 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 03:05:29,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-04-30 03:05:34,895 INFO [train.py:904] (7/8) Epoch 15, batch 2600, loss[loss=0.1877, simple_loss=0.2679, pruned_loss=0.05375, over 16390.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.266, pruned_loss=0.04694, over 3320290.47 frames. ], batch size: 146, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:05:36,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.335e+02 2.561e+02 3.164e+02 7.288e+02, threshold=5.122e+02, percent-clipped=2.0 2023-04-30 03:06:43,557 INFO [train.py:904] (7/8) Epoch 15, batch 2650, loss[loss=0.1784, simple_loss=0.2754, pruned_loss=0.04074, over 16735.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2664, pruned_loss=0.04703, over 3316217.81 frames. ], batch size: 57, lr: 4.58e-03, grad_scale: 8.0 2023-04-30 03:07:05,886 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=144767.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:07:14,156 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-04-30 03:07:53,567 INFO [train.py:904] (7/8) Epoch 15, batch 2700, loss[loss=0.1681, simple_loss=0.2563, pruned_loss=0.03989, over 16855.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2667, pruned_loss=0.04627, over 3323630.44 frames. ], batch size: 90, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:07:54,730 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.168e+02 2.529e+02 3.023e+02 4.642e+02, threshold=5.059e+02, percent-clipped=0.0 2023-04-30 03:08:19,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4311, 5.7658, 5.2493, 5.7056, 5.2742, 4.9041, 5.2516, 5.9420], device='cuda:7'), covar=tensor([0.2646, 0.1670, 0.3119, 0.1048, 0.1490, 0.1475, 0.2397, 0.1855], device='cuda:7'), in_proj_covar=tensor([0.0619, 0.0773, 0.0630, 0.0551, 0.0485, 0.0490, 0.0640, 0.0583], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:08:53,733 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-30 03:09:02,449 INFO [train.py:904] (7/8) Epoch 15, batch 2750, loss[loss=0.1759, simple_loss=0.267, pruned_loss=0.04235, over 16760.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2664, pruned_loss=0.04606, over 3320572.29 frames. ], batch size: 57, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:10:11,032 INFO [train.py:904] (7/8) Epoch 15, batch 2800, loss[loss=0.1653, simple_loss=0.2608, pruned_loss=0.03493, over 16863.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2664, pruned_loss=0.04612, over 3323025.41 frames. ], batch size: 42, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:10:12,143 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.181e+02 2.488e+02 3.014e+02 5.995e+02, threshold=4.976e+02, percent-clipped=2.0 2023-04-30 03:10:35,359 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8087, 4.6801, 4.6515, 4.3339, 4.3471, 4.7075, 4.6114, 4.4150], device='cuda:7'), covar=tensor([0.0579, 0.0581, 0.0286, 0.0323, 0.0981, 0.0472, 0.0450, 0.0735], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0383, 0.0332, 0.0315, 0.0347, 0.0361, 0.0225, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:10:47,661 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9044, 1.8674, 2.3421, 2.7530, 2.7380, 2.8807, 2.0475, 3.0419], device='cuda:7'), covar=tensor([0.0172, 0.0412, 0.0311, 0.0255, 0.0228, 0.0211, 0.0402, 0.0119], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0184, 0.0170, 0.0173, 0.0183, 0.0139, 0.0184, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:11:21,064 INFO [train.py:904] (7/8) Epoch 15, batch 2850, loss[loss=0.196, simple_loss=0.2877, pruned_loss=0.05215, over 16679.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2663, pruned_loss=0.04629, over 3324360.57 frames. ], batch size: 57, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:11:28,600 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-30 03:12:22,171 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=144996.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:12:31,819 INFO [train.py:904] (7/8) Epoch 15, batch 2900, loss[loss=0.2468, simple_loss=0.3062, pruned_loss=0.0937, over 11448.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2653, pruned_loss=0.04639, over 3325781.23 frames. ], batch size: 247, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:12:33,011 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.445e+02 2.844e+02 3.300e+02 6.709e+02, threshold=5.687e+02, percent-clipped=6.0 2023-04-30 03:12:33,712 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 03:12:40,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2937, 2.3619, 1.7628, 1.9476, 2.6840, 2.4506, 3.1098, 2.9865], device='cuda:7'), covar=tensor([0.0176, 0.0437, 0.0606, 0.0554, 0.0319, 0.0397, 0.0246, 0.0267], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0223, 0.0213, 0.0214, 0.0224, 0.0222, 0.0230, 0.0217], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:13:06,901 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145027.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:13:40,922 INFO [train.py:904] (7/8) Epoch 15, batch 2950, loss[loss=0.1908, simple_loss=0.269, pruned_loss=0.05631, over 16867.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2647, pruned_loss=0.04675, over 3324692.54 frames. ], batch size: 96, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:13:47,657 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145057.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:14:01,018 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145066.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:14:02,163 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145067.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:14:31,933 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:14:49,684 INFO [train.py:904] (7/8) Epoch 15, batch 3000, loss[loss=0.1677, simple_loss=0.2457, pruned_loss=0.04484, over 16853.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2647, pruned_loss=0.0476, over 3329761.16 frames. ], batch size: 90, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:14:49,685 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 03:14:58,803 INFO [train.py:938] (7/8) Epoch 15, validation: loss=0.138, simple_loss=0.2438, pruned_loss=0.01616, over 944034.00 frames. 2023-04-30 03:14:58,804 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 03:15:00,800 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.402e+02 2.841e+02 3.286e+02 6.614e+02, threshold=5.681e+02, percent-clipped=1.0 2023-04-30 03:15:17,420 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145115.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:15:34,610 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145127.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:16:07,553 INFO [train.py:904] (7/8) Epoch 15, batch 3050, loss[loss=0.1778, simple_loss=0.2522, pruned_loss=0.05173, over 16823.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.264, pruned_loss=0.04729, over 3333785.58 frames. ], batch size: 102, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:17:18,170 INFO [train.py:904] (7/8) Epoch 15, batch 3100, loss[loss=0.2017, simple_loss=0.2679, pruned_loss=0.06774, over 16418.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2637, pruned_loss=0.04724, over 3331588.90 frames. ], batch size: 146, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:17:19,345 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 2.445e+02 2.806e+02 3.389e+02 5.168e+02, threshold=5.611e+02, percent-clipped=0.0 2023-04-30 03:18:28,441 INFO [train.py:904] (7/8) Epoch 15, batch 3150, loss[loss=0.1564, simple_loss=0.2517, pruned_loss=0.0305, over 17181.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2627, pruned_loss=0.04709, over 3328768.74 frames. ], batch size: 46, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:19:37,256 INFO [train.py:904] (7/8) Epoch 15, batch 3200, loss[loss=0.1708, simple_loss=0.2668, pruned_loss=0.03741, over 17056.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2621, pruned_loss=0.04641, over 3330712.62 frames. ], batch size: 55, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:19:38,468 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.241e+02 2.735e+02 3.234e+02 5.514e+02, threshold=5.469e+02, percent-clipped=0.0 2023-04-30 03:19:47,933 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2534, 5.8543, 5.9464, 5.7006, 5.8532, 6.3222, 5.9108, 5.6081], device='cuda:7'), covar=tensor([0.0867, 0.1720, 0.2005, 0.1985, 0.2405, 0.0925, 0.1256, 0.2223], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0552, 0.0606, 0.0469, 0.0630, 0.0633, 0.0479, 0.0624], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:20:46,514 INFO [train.py:904] (7/8) Epoch 15, batch 3250, loss[loss=0.2007, simple_loss=0.2875, pruned_loss=0.05697, over 17080.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2616, pruned_loss=0.04631, over 3337006.22 frames. ], batch size: 53, lr: 4.57e-03, grad_scale: 8.0 2023-04-30 03:20:46,797 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145352.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:21:13,391 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145371.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:21:30,558 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145383.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:21:57,384 INFO [train.py:904] (7/8) Epoch 15, batch 3300, loss[loss=0.1985, simple_loss=0.2895, pruned_loss=0.05377, over 16708.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2627, pruned_loss=0.04669, over 3327955.41 frames. ], batch size: 62, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:21:58,628 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.241e+02 2.723e+02 3.266e+02 5.157e+02, threshold=5.447e+02, percent-clipped=0.0 2023-04-30 03:22:15,387 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4202, 5.8013, 5.5600, 5.6734, 5.2068, 5.2008, 5.1890, 5.9503], device='cuda:7'), covar=tensor([0.1317, 0.0956, 0.1015, 0.0716, 0.0857, 0.0670, 0.1061, 0.0943], device='cuda:7'), in_proj_covar=tensor([0.0634, 0.0792, 0.0646, 0.0565, 0.0500, 0.0503, 0.0658, 0.0599], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:22:25,245 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145422.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:22:38,879 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145432.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:23:06,159 INFO [train.py:904] (7/8) Epoch 15, batch 3350, loss[loss=0.1641, simple_loss=0.2574, pruned_loss=0.03544, over 17107.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2629, pruned_loss=0.0461, over 3329957.96 frames. ], batch size: 48, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:23:08,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9892, 5.0637, 5.4473, 5.4310, 5.4400, 5.1117, 5.0489, 4.8443], device='cuda:7'), covar=tensor([0.0266, 0.0526, 0.0399, 0.0423, 0.0431, 0.0343, 0.0870, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0404, 0.0400, 0.0383, 0.0448, 0.0423, 0.0520, 0.0337], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 03:23:13,894 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4789, 2.3116, 2.4196, 4.3272, 2.2812, 2.7773, 2.3903, 2.5549], device='cuda:7'), covar=tensor([0.1159, 0.3233, 0.2529, 0.0489, 0.3857, 0.2307, 0.3203, 0.3132], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0419, 0.0350, 0.0328, 0.0424, 0.0484, 0.0384, 0.0492], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:23:18,946 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1488, 3.9431, 4.0562, 4.3360, 4.4424, 4.0615, 4.2732, 4.4281], device='cuda:7'), covar=tensor([0.1534, 0.1328, 0.1841, 0.0870, 0.0799, 0.1358, 0.1815, 0.0909], device='cuda:7'), in_proj_covar=tensor([0.0619, 0.0766, 0.0915, 0.0785, 0.0584, 0.0611, 0.0615, 0.0724], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:24:05,764 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 03:24:17,516 INFO [train.py:904] (7/8) Epoch 15, batch 3400, loss[loss=0.1627, simple_loss=0.2588, pruned_loss=0.03335, over 17035.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2628, pruned_loss=0.04633, over 3319117.52 frames. ], batch size: 50, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:24:18,607 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.153e+02 2.635e+02 3.209e+02 5.771e+02, threshold=5.270e+02, percent-clipped=2.0 2023-04-30 03:24:47,858 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-30 03:25:28,517 INFO [train.py:904] (7/8) Epoch 15, batch 3450, loss[loss=0.1728, simple_loss=0.2682, pruned_loss=0.03869, over 16659.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2617, pruned_loss=0.0461, over 3315643.55 frames. ], batch size: 62, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:25:32,268 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-04-30 03:25:56,671 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6817, 3.8210, 2.2570, 4.4606, 2.8741, 4.4142, 2.5411, 3.0879], device='cuda:7'), covar=tensor([0.0279, 0.0368, 0.1575, 0.0239, 0.0816, 0.0411, 0.1355, 0.0684], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0173, 0.0192, 0.0152, 0.0171, 0.0216, 0.0201, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 03:26:27,203 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145594.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:26:38,103 INFO [train.py:904] (7/8) Epoch 15, batch 3500, loss[loss=0.1653, simple_loss=0.2587, pruned_loss=0.03597, over 17259.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2609, pruned_loss=0.046, over 3307450.01 frames. ], batch size: 52, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:26:39,236 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.268e+02 2.638e+02 3.199e+02 5.613e+02, threshold=5.276e+02, percent-clipped=1.0 2023-04-30 03:26:39,840 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5265, 1.7615, 2.1586, 2.3393, 2.5099, 2.4913, 1.7147, 2.5759], device='cuda:7'), covar=tensor([0.0164, 0.0421, 0.0270, 0.0249, 0.0233, 0.0248, 0.0439, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0186, 0.0171, 0.0176, 0.0185, 0.0141, 0.0187, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:26:40,134 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 03:27:35,963 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:27:47,276 INFO [train.py:904] (7/8) Epoch 15, batch 3550, loss[loss=0.1722, simple_loss=0.2526, pruned_loss=0.04588, over 16795.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2594, pruned_loss=0.04586, over 3316578.10 frames. ], batch size: 83, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:27:48,338 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145652.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:27:53,475 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145655.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:28:32,422 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145683.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:28:45,890 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-04-30 03:28:54,823 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145700.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:28:57,599 INFO [train.py:904] (7/8) Epoch 15, batch 3600, loss[loss=0.1682, simple_loss=0.2581, pruned_loss=0.03914, over 16875.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2588, pruned_loss=0.04532, over 3318661.21 frames. ], batch size: 42, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:28:58,730 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.177e+02 2.263e+02 2.507e+02 3.007e+02 5.256e+02, threshold=5.015e+02, percent-clipped=0.0 2023-04-30 03:29:01,405 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145704.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:27,706 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=145722.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:34,390 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145727.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:29:39,859 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145731.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:30:10,697 INFO [train.py:904] (7/8) Epoch 15, batch 3650, loss[loss=0.181, simple_loss=0.2599, pruned_loss=0.05103, over 11898.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2581, pruned_loss=0.04533, over 3303429.30 frames. ], batch size: 247, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:30:17,901 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6768, 2.7610, 2.5076, 4.0099, 3.4538, 4.0938, 1.5574, 2.7604], device='cuda:7'), covar=tensor([0.1397, 0.0626, 0.1070, 0.0173, 0.0163, 0.0347, 0.1481, 0.0842], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0165, 0.0185, 0.0171, 0.0201, 0.0212, 0.0188, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 03:30:24,169 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8854, 4.1449, 3.1734, 2.3903, 2.8104, 2.5967, 4.2341, 3.7233], device='cuda:7'), covar=tensor([0.2407, 0.0513, 0.1455, 0.2547, 0.2550, 0.1855, 0.0476, 0.1197], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0268, 0.0296, 0.0296, 0.0291, 0.0241, 0.0281, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:30:38,223 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=145770.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:31:19,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1987, 3.3288, 3.4786, 2.5305, 3.1982, 3.6330, 3.3924, 2.0925], device='cuda:7'), covar=tensor([0.0449, 0.0105, 0.0058, 0.0312, 0.0097, 0.0089, 0.0081, 0.0406], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0076, 0.0075, 0.0130, 0.0088, 0.0100, 0.0086, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:31:24,500 INFO [train.py:904] (7/8) Epoch 15, batch 3700, loss[loss=0.1727, simple_loss=0.248, pruned_loss=0.04866, over 16723.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.257, pruned_loss=0.04697, over 3292650.67 frames. ], batch size: 124, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:31:26,281 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.174e+02 2.701e+02 3.170e+02 5.249e+02, threshold=5.403e+02, percent-clipped=2.0 2023-04-30 03:31:53,367 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=145821.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 03:32:38,909 INFO [train.py:904] (7/8) Epoch 15, batch 3750, loss[loss=0.1757, simple_loss=0.2502, pruned_loss=0.05058, over 16804.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2585, pruned_loss=0.04873, over 3272898.28 frames. ], batch size: 102, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:32:46,241 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9748, 4.9939, 4.8767, 4.3783, 4.9559, 2.0253, 4.7126, 4.7174], device='cuda:7'), covar=tensor([0.0102, 0.0091, 0.0178, 0.0313, 0.0092, 0.2358, 0.0137, 0.0154], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0138, 0.0187, 0.0173, 0.0158, 0.0197, 0.0175, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:33:23,521 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=145882.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:33:27,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4437, 4.4228, 4.5583, 4.4399, 4.3630, 4.9849, 4.5335, 4.2213], device='cuda:7'), covar=tensor([0.1509, 0.1892, 0.2028, 0.2134, 0.2945, 0.1119, 0.1610, 0.2776], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0549, 0.0600, 0.0470, 0.0628, 0.0625, 0.0479, 0.0620], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:33:51,808 INFO [train.py:904] (7/8) Epoch 15, batch 3800, loss[loss=0.1714, simple_loss=0.2483, pruned_loss=0.04718, over 16433.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2599, pruned_loss=0.05016, over 3266530.87 frames. ], batch size: 75, lr: 4.56e-03, grad_scale: 8.0 2023-04-30 03:33:53,683 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.221e+02 2.643e+02 3.115e+02 5.182e+02, threshold=5.286e+02, percent-clipped=0.0 2023-04-30 03:34:35,555 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8097, 1.8981, 2.3239, 2.6875, 2.7345, 2.7049, 1.9328, 2.9355], device='cuda:7'), covar=tensor([0.0119, 0.0377, 0.0274, 0.0211, 0.0210, 0.0217, 0.0391, 0.0116], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0184, 0.0169, 0.0175, 0.0183, 0.0140, 0.0185, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:35:01,661 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145950.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 03:35:04,376 INFO [train.py:904] (7/8) Epoch 15, batch 3850, loss[loss=0.1846, simple_loss=0.257, pruned_loss=0.05612, over 16487.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2601, pruned_loss=0.05103, over 3267906.11 frames. ], batch size: 146, lr: 4.56e-03, grad_scale: 16.0 2023-04-30 03:35:13,489 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7521, 3.5792, 2.6241, 2.3251, 2.5307, 2.2352, 3.5763, 3.2578], device='cuda:7'), covar=tensor([0.2206, 0.0736, 0.1589, 0.2436, 0.2260, 0.2031, 0.0589, 0.1253], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0265, 0.0293, 0.0293, 0.0289, 0.0238, 0.0279, 0.0317], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:36:13,429 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=145999.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:36:21,022 INFO [train.py:904] (7/8) Epoch 15, batch 3900, loss[loss=0.1829, simple_loss=0.265, pruned_loss=0.05035, over 16589.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2591, pruned_loss=0.05111, over 3269291.78 frames. ], batch size: 57, lr: 4.56e-03, grad_scale: 16.0 2023-04-30 03:36:22,208 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.255e+02 2.647e+02 3.185e+02 6.041e+02, threshold=5.295e+02, percent-clipped=2.0 2023-04-30 03:36:39,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8743, 1.9561, 2.3836, 2.7175, 2.7938, 2.8397, 2.0917, 3.0204], device='cuda:7'), covar=tensor([0.0135, 0.0375, 0.0253, 0.0221, 0.0200, 0.0185, 0.0353, 0.0100], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0183, 0.0169, 0.0174, 0.0182, 0.0139, 0.0184, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:36:57,939 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146027.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:37:09,271 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.67 vs. limit=5.0 2023-04-30 03:37:22,222 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7784, 2.9856, 2.6465, 4.5095, 3.7220, 4.2601, 1.7323, 3.0670], device='cuda:7'), covar=tensor([0.1319, 0.0669, 0.1089, 0.0173, 0.0263, 0.0362, 0.1437, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0165, 0.0185, 0.0170, 0.0201, 0.0212, 0.0187, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 03:37:32,707 INFO [train.py:904] (7/8) Epoch 15, batch 3950, loss[loss=0.174, simple_loss=0.2493, pruned_loss=0.04937, over 16937.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2585, pruned_loss=0.05147, over 3266774.33 frames. ], batch size: 116, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:37:45,402 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 03:38:06,971 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146075.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:38:09,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8158, 4.8038, 4.7062, 4.4435, 4.3961, 4.7811, 4.6040, 4.5220], device='cuda:7'), covar=tensor([0.0585, 0.0561, 0.0272, 0.0249, 0.0814, 0.0425, 0.0385, 0.0546], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0381, 0.0327, 0.0311, 0.0344, 0.0358, 0.0220, 0.0389], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:38:33,718 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9309, 3.9658, 4.5395, 2.3265, 4.7496, 4.8594, 3.3058, 3.4771], device='cuda:7'), covar=tensor([0.0714, 0.0241, 0.0124, 0.1107, 0.0040, 0.0056, 0.0357, 0.0442], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0106, 0.0091, 0.0139, 0.0074, 0.0118, 0.0124, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 03:38:46,134 INFO [train.py:904] (7/8) Epoch 15, batch 4000, loss[loss=0.159, simple_loss=0.2439, pruned_loss=0.0371, over 16486.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2574, pruned_loss=0.05116, over 3281153.74 frames. ], batch size: 75, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:38:47,410 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.290e+02 2.701e+02 3.084e+02 7.730e+02, threshold=5.403e+02, percent-clipped=2.0 2023-04-30 03:38:54,856 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2281, 3.7310, 3.7780, 2.6011, 3.4335, 3.9000, 3.4813, 2.0268], device='cuda:7'), covar=tensor([0.0449, 0.0078, 0.0038, 0.0320, 0.0086, 0.0087, 0.0089, 0.0412], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0074, 0.0074, 0.0129, 0.0088, 0.0098, 0.0086, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:39:56,192 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146149.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:39:59,986 INFO [train.py:904] (7/8) Epoch 15, batch 4050, loss[loss=0.1772, simple_loss=0.2599, pruned_loss=0.04723, over 17261.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2581, pruned_loss=0.05017, over 3281190.63 frames. ], batch size: 52, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:40:36,978 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146177.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:40:58,485 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4137, 4.5248, 4.2571, 4.0040, 3.9630, 4.3846, 4.0932, 4.1032], device='cuda:7'), covar=tensor([0.0574, 0.0428, 0.0254, 0.0252, 0.0748, 0.0398, 0.0650, 0.0530], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0379, 0.0327, 0.0310, 0.0344, 0.0357, 0.0220, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:41:13,958 INFO [train.py:904] (7/8) Epoch 15, batch 4100, loss[loss=0.2171, simple_loss=0.3046, pruned_loss=0.06473, over 15303.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2599, pruned_loss=0.04953, over 3278986.58 frames. ], batch size: 190, lr: 4.55e-03, grad_scale: 16.0 2023-04-30 03:41:15,751 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 1.975e+02 2.401e+02 2.875e+02 5.931e+02, threshold=4.803e+02, percent-clipped=1.0 2023-04-30 03:41:19,728 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 03:41:27,478 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146210.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:42:30,154 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146250.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:42:32,945 INFO [train.py:904] (7/8) Epoch 15, batch 4150, loss[loss=0.2099, simple_loss=0.3024, pruned_loss=0.05873, over 16361.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2672, pruned_loss=0.05244, over 3234019.81 frames. ], batch size: 146, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:43:45,580 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146298.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:43:46,720 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146299.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:43:50,406 INFO [train.py:904] (7/8) Epoch 15, batch 4200, loss[loss=0.2266, simple_loss=0.3141, pruned_loss=0.06953, over 16829.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.274, pruned_loss=0.05427, over 3191009.72 frames. ], batch size: 116, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:43:53,466 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.342e+02 2.800e+02 3.448e+02 4.997e+02, threshold=5.600e+02, percent-clipped=3.0 2023-04-30 03:44:21,176 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1973, 3.4299, 3.5486, 3.5369, 3.5467, 3.3996, 3.4031, 3.4194], device='cuda:7'), covar=tensor([0.0383, 0.0590, 0.0485, 0.0465, 0.0544, 0.0523, 0.0850, 0.0563], device='cuda:7'), in_proj_covar=tensor([0.0367, 0.0396, 0.0390, 0.0369, 0.0437, 0.0409, 0.0504, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 03:44:59,533 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146347.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:45:06,386 INFO [train.py:904] (7/8) Epoch 15, batch 4250, loss[loss=0.201, simple_loss=0.292, pruned_loss=0.05498, over 16591.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2775, pruned_loss=0.05422, over 3182621.55 frames. ], batch size: 57, lr: 4.55e-03, grad_scale: 4.0 2023-04-30 03:45:23,937 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 03:45:40,131 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8611, 3.0281, 2.5223, 4.8968, 3.8365, 4.1848, 1.6298, 2.8486], device='cuda:7'), covar=tensor([0.1253, 0.0652, 0.1243, 0.0121, 0.0319, 0.0450, 0.1474, 0.0946], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0185, 0.0168, 0.0201, 0.0211, 0.0188, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 03:46:19,358 INFO [train.py:904] (7/8) Epoch 15, batch 4300, loss[loss=0.2051, simple_loss=0.2956, pruned_loss=0.05729, over 16899.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.279, pruned_loss=0.05335, over 3195151.27 frames. ], batch size: 96, lr: 4.55e-03, grad_scale: 4.0 2023-04-30 03:46:23,348 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.412e+02 2.971e+02 3.359e+02 7.082e+02, threshold=5.941e+02, percent-clipped=4.0 2023-04-30 03:46:23,949 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3895, 2.3206, 2.2184, 4.1804, 2.1328, 2.6681, 2.3739, 2.4480], device='cuda:7'), covar=tensor([0.1071, 0.3278, 0.2624, 0.0430, 0.3748, 0.2210, 0.2979, 0.3119], device='cuda:7'), in_proj_covar=tensor([0.0381, 0.0419, 0.0347, 0.0323, 0.0422, 0.0482, 0.0383, 0.0491], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:47:31,155 INFO [train.py:904] (7/8) Epoch 15, batch 4350, loss[loss=0.1892, simple_loss=0.2829, pruned_loss=0.04773, over 16575.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2827, pruned_loss=0.05458, over 3206650.16 frames. ], batch size: 62, lr: 4.55e-03, grad_scale: 4.0 2023-04-30 03:48:08,923 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146477.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 03:48:45,766 INFO [train.py:904] (7/8) Epoch 15, batch 4400, loss[loss=0.1964, simple_loss=0.2857, pruned_loss=0.05359, over 17028.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.284, pruned_loss=0.05535, over 3210717.39 frames. ], batch size: 53, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:48:50,397 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.716e+02 2.574e+02 2.972e+02 3.574e+02 6.742e+02, threshold=5.944e+02, percent-clipped=2.0 2023-04-30 03:48:51,430 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146505.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:49:21,116 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146525.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:49:35,960 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6587, 4.4895, 4.7292, 4.8423, 4.9626, 4.5181, 4.9626, 4.9983], device='cuda:7'), covar=tensor([0.1354, 0.1052, 0.1138, 0.0495, 0.0416, 0.0746, 0.0422, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0572, 0.0710, 0.0848, 0.0725, 0.0543, 0.0569, 0.0573, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:49:58,646 INFO [train.py:904] (7/8) Epoch 15, batch 4450, loss[loss=0.2025, simple_loss=0.2883, pruned_loss=0.05832, over 16444.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2881, pruned_loss=0.05701, over 3214112.65 frames. ], batch size: 68, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:50:05,803 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-30 03:50:57,113 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146591.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:51:12,446 INFO [train.py:904] (7/8) Epoch 15, batch 4500, loss[loss=0.2043, simple_loss=0.291, pruned_loss=0.05879, over 16329.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2881, pruned_loss=0.05743, over 3204898.95 frames. ], batch size: 35, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:51:16,068 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 1.915e+02 2.182e+02 2.470e+02 4.715e+02, threshold=4.363e+02, percent-clipped=0.0 2023-04-30 03:51:47,952 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146626.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:52:25,495 INFO [train.py:904] (7/8) Epoch 15, batch 4550, loss[loss=0.2142, simple_loss=0.2945, pruned_loss=0.06695, over 16868.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2889, pruned_loss=0.05811, over 3223125.57 frames. ], batch size: 42, lr: 4.55e-03, grad_scale: 8.0 2023-04-30 03:52:25,995 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146652.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:53:00,583 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3040, 2.2423, 2.2641, 4.0492, 2.1292, 2.6309, 2.2951, 2.3731], device='cuda:7'), covar=tensor([0.1120, 0.3195, 0.2546, 0.0449, 0.3972, 0.2257, 0.3232, 0.3124], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0418, 0.0345, 0.0321, 0.0423, 0.0481, 0.0382, 0.0488], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:53:03,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9396, 3.2183, 3.3215, 1.9217, 2.7664, 2.1255, 3.3878, 3.3349], device='cuda:7'), covar=tensor([0.0237, 0.0734, 0.0570, 0.1980, 0.0856, 0.1052, 0.0643, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0154, 0.0161, 0.0146, 0.0138, 0.0126, 0.0139, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 03:53:10,675 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9631, 3.3531, 3.4167, 2.0295, 2.8208, 2.2447, 3.3972, 3.4579], device='cuda:7'), covar=tensor([0.0234, 0.0656, 0.0576, 0.1922, 0.0848, 0.1008, 0.0622, 0.0823], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0154, 0.0161, 0.0146, 0.0138, 0.0126, 0.0139, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 03:53:16,722 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146687.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 03:53:37,398 INFO [train.py:904] (7/8) Epoch 15, batch 4600, loss[loss=0.1962, simple_loss=0.2795, pruned_loss=0.05647, over 16508.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2892, pruned_loss=0.05767, over 3239652.12 frames. ], batch size: 75, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:53:41,731 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 1.940e+02 2.192e+02 2.697e+02 4.440e+02, threshold=4.384e+02, percent-clipped=1.0 2023-04-30 03:53:44,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0180, 1.7472, 2.4817, 2.8735, 2.6953, 3.2800, 1.8535, 3.2592], device='cuda:7'), covar=tensor([0.0143, 0.0474, 0.0269, 0.0220, 0.0232, 0.0140, 0.0504, 0.0109], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0184, 0.0169, 0.0175, 0.0183, 0.0139, 0.0185, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:54:03,251 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3663, 2.9304, 2.6154, 2.2207, 2.1666, 2.1363, 2.8246, 2.7882], device='cuda:7'), covar=tensor([0.2393, 0.0752, 0.1459, 0.2258, 0.2309, 0.2018, 0.0503, 0.1070], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0265, 0.0293, 0.0294, 0.0288, 0.0237, 0.0277, 0.0316], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:54:49,174 INFO [train.py:904] (7/8) Epoch 15, batch 4650, loss[loss=0.1996, simple_loss=0.2835, pruned_loss=0.05781, over 16300.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2887, pruned_loss=0.05808, over 3225229.89 frames. ], batch size: 165, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:54:54,414 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2006, 3.9402, 3.8654, 2.4665, 3.5851, 3.9944, 3.6687, 2.2295], device='cuda:7'), covar=tensor([0.0523, 0.0031, 0.0038, 0.0404, 0.0064, 0.0079, 0.0072, 0.0398], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0073, 0.0073, 0.0129, 0.0087, 0.0097, 0.0085, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:55:26,830 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5853, 4.6429, 4.4477, 4.1582, 4.1831, 4.5336, 4.2108, 4.2480], device='cuda:7'), covar=tensor([0.0468, 0.0296, 0.0215, 0.0221, 0.0701, 0.0284, 0.0596, 0.0498], device='cuda:7'), in_proj_covar=tensor([0.0262, 0.0358, 0.0312, 0.0295, 0.0327, 0.0337, 0.0210, 0.0367], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:56:03,160 INFO [train.py:904] (7/8) Epoch 15, batch 4700, loss[loss=0.1852, simple_loss=0.2636, pruned_loss=0.05334, over 11292.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2862, pruned_loss=0.05692, over 3223320.01 frames. ], batch size: 246, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:56:07,891 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.906e+02 2.324e+02 2.777e+02 7.777e+02, threshold=4.648e+02, percent-clipped=3.0 2023-04-30 03:56:08,856 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=146805.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:56:50,172 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1215, 4.0059, 3.9627, 2.5961, 3.5863, 3.9952, 3.6180, 2.0192], device='cuda:7'), covar=tensor([0.0549, 0.0027, 0.0032, 0.0349, 0.0069, 0.0075, 0.0078, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0073, 0.0073, 0.0129, 0.0087, 0.0097, 0.0085, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 03:56:52,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6550, 4.6992, 4.5438, 4.2075, 4.1274, 4.6104, 4.3887, 4.3283], device='cuda:7'), covar=tensor([0.0537, 0.0451, 0.0244, 0.0261, 0.0960, 0.0444, 0.0449, 0.0550], device='cuda:7'), in_proj_covar=tensor([0.0262, 0.0359, 0.0313, 0.0295, 0.0326, 0.0338, 0.0210, 0.0367], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:57:07,706 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=146844.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:57:18,598 INFO [train.py:904] (7/8) Epoch 15, batch 4750, loss[loss=0.1824, simple_loss=0.2729, pruned_loss=0.046, over 15412.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.282, pruned_loss=0.05467, over 3228774.28 frames. ], batch size: 190, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:57:20,574 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=146853.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:57:37,414 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7656, 4.8046, 4.6832, 3.9688, 4.7454, 1.6833, 4.4425, 4.4138], device='cuda:7'), covar=tensor([0.0096, 0.0088, 0.0132, 0.0452, 0.0090, 0.2734, 0.0135, 0.0219], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0133, 0.0182, 0.0168, 0.0153, 0.0192, 0.0170, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 03:58:31,082 INFO [train.py:904] (7/8) Epoch 15, batch 4800, loss[loss=0.1725, simple_loss=0.2638, pruned_loss=0.04063, over 16538.00 frames. ], tot_loss[loss=0.1913, simple_loss=0.2778, pruned_loss=0.05238, over 3239513.64 frames. ], batch size: 75, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 03:58:36,187 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.051e+02 1.856e+02 2.167e+02 2.629e+02 4.962e+02, threshold=4.334e+02, percent-clipped=1.0 2023-04-30 03:58:36,793 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=146905.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:59:40,650 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146947.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 03:59:47,643 INFO [train.py:904] (7/8) Epoch 15, batch 4850, loss[loss=0.1854, simple_loss=0.2778, pruned_loss=0.04655, over 16809.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2783, pruned_loss=0.05181, over 3212741.51 frames. ], batch size: 89, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:00:34,906 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=146982.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 04:01:03,668 INFO [train.py:904] (7/8) Epoch 15, batch 4900, loss[loss=0.2018, simple_loss=0.2902, pruned_loss=0.05672, over 16631.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.278, pruned_loss=0.05122, over 3195618.45 frames. ], batch size: 57, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:01:08,002 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 1.989e+02 2.229e+02 2.704e+02 6.823e+02, threshold=4.458e+02, percent-clipped=4.0 2023-04-30 04:01:45,369 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 04:01:50,196 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 04:02:16,329 INFO [train.py:904] (7/8) Epoch 15, batch 4950, loss[loss=0.2179, simple_loss=0.3072, pruned_loss=0.06431, over 16887.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2776, pruned_loss=0.05058, over 3196589.05 frames. ], batch size: 116, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:02:34,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2438, 3.4730, 3.6382, 3.6097, 3.6036, 3.4124, 3.4302, 3.4871], device='cuda:7'), covar=tensor([0.0362, 0.0510, 0.0380, 0.0414, 0.0456, 0.0429, 0.0762, 0.0490], device='cuda:7'), in_proj_covar=tensor([0.0362, 0.0388, 0.0384, 0.0365, 0.0430, 0.0405, 0.0499, 0.0323], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 04:03:03,386 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.03 vs. limit=5.0 2023-04-30 04:03:25,715 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1299, 2.0816, 2.1592, 3.8381, 1.9856, 2.4605, 2.1980, 2.2606], device='cuda:7'), covar=tensor([0.1246, 0.3493, 0.2638, 0.0494, 0.3914, 0.2312, 0.3337, 0.2964], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0417, 0.0345, 0.0320, 0.0421, 0.0478, 0.0381, 0.0487], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:03:26,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8776, 5.2523, 5.4149, 5.2147, 5.2226, 5.7762, 5.2722, 4.9801], device='cuda:7'), covar=tensor([0.0907, 0.1679, 0.1557, 0.1728, 0.2208, 0.0867, 0.1400, 0.2262], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0526, 0.0571, 0.0449, 0.0596, 0.0602, 0.0456, 0.0598], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 04:03:28,736 INFO [train.py:904] (7/8) Epoch 15, batch 5000, loss[loss=0.1846, simple_loss=0.2747, pruned_loss=0.04722, over 16527.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2794, pruned_loss=0.0507, over 3198957.34 frames. ], batch size: 62, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:03:32,272 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.177e+02 2.680e+02 3.002e+02 5.827e+02, threshold=5.360e+02, percent-clipped=3.0 2023-04-30 04:03:52,107 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6086, 3.6321, 2.0676, 4.2447, 2.6558, 4.1042, 2.2764, 2.7848], device='cuda:7'), covar=tensor([0.0262, 0.0318, 0.1619, 0.0078, 0.0799, 0.0395, 0.1418, 0.0761], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0170, 0.0190, 0.0143, 0.0168, 0.0209, 0.0197, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 04:04:12,565 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-30 04:04:39,304 INFO [train.py:904] (7/8) Epoch 15, batch 5050, loss[loss=0.1855, simple_loss=0.2803, pruned_loss=0.04532, over 16486.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2788, pruned_loss=0.05018, over 3201590.10 frames. ], batch size: 146, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:05:46,851 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147200.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:05:48,955 INFO [train.py:904] (7/8) Epoch 15, batch 5100, loss[loss=0.1866, simple_loss=0.2763, pruned_loss=0.04841, over 15472.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2774, pruned_loss=0.04983, over 3208468.06 frames. ], batch size: 190, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:05:52,962 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.025e+02 2.374e+02 2.786e+02 3.985e+02, threshold=4.748e+02, percent-clipped=0.0 2023-04-30 04:06:28,318 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8469, 5.0685, 5.2319, 5.0602, 5.0458, 5.6367, 5.1296, 4.8393], device='cuda:7'), covar=tensor([0.0866, 0.1697, 0.1471, 0.1775, 0.2244, 0.0789, 0.1272, 0.2327], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0527, 0.0571, 0.0448, 0.0595, 0.0604, 0.0456, 0.0599], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 04:06:45,593 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147241.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:06:53,121 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147247.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:06:59,926 INFO [train.py:904] (7/8) Epoch 15, batch 5150, loss[loss=0.1869, simple_loss=0.2838, pruned_loss=0.04497, over 16490.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2772, pruned_loss=0.04905, over 3206579.52 frames. ], batch size: 75, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:07:09,052 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 04:07:42,571 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3789, 4.4224, 4.7770, 4.7569, 4.7454, 4.4835, 4.4559, 4.3178], device='cuda:7'), covar=tensor([0.0259, 0.0490, 0.0286, 0.0328, 0.0357, 0.0317, 0.0715, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0358, 0.0385, 0.0381, 0.0361, 0.0427, 0.0401, 0.0495, 0.0322], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 04:07:45,018 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:03,069 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147295.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:08,382 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2799, 4.2690, 4.2078, 3.4904, 4.1965, 1.6715, 3.9854, 3.9353], device='cuda:7'), covar=tensor([0.0097, 0.0093, 0.0141, 0.0377, 0.0099, 0.2687, 0.0138, 0.0196], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0130, 0.0178, 0.0167, 0.0150, 0.0189, 0.0167, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:08:14,142 INFO [train.py:904] (7/8) Epoch 15, batch 5200, loss[loss=0.1809, simple_loss=0.2699, pruned_loss=0.04593, over 15374.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2756, pruned_loss=0.04841, over 3215553.87 frames. ], batch size: 190, lr: 4.54e-03, grad_scale: 8.0 2023-04-30 04:08:14,609 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147302.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:17,789 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.189e+02 2.515e+02 3.060e+02 5.067e+02, threshold=5.031e+02, percent-clipped=2.0 2023-04-30 04:08:24,380 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 04:08:26,102 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147310.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:08:54,079 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4277, 2.1542, 2.2660, 4.2660, 2.1192, 2.4558, 2.3116, 2.3895], device='cuda:7'), covar=tensor([0.1177, 0.3447, 0.2639, 0.0416, 0.4024, 0.2456, 0.3299, 0.3014], device='cuda:7'), in_proj_covar=tensor([0.0373, 0.0411, 0.0341, 0.0316, 0.0416, 0.0472, 0.0376, 0.0480], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:08:55,676 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147330.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:09:27,029 INFO [train.py:904] (7/8) Epoch 15, batch 5250, loss[loss=0.1882, simple_loss=0.2737, pruned_loss=0.0513, over 15390.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2733, pruned_loss=0.04781, over 3230743.89 frames. ], batch size: 190, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:09:54,933 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147371.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:10:37,514 INFO [train.py:904] (7/8) Epoch 15, batch 5300, loss[loss=0.2006, simple_loss=0.2764, pruned_loss=0.06245, over 12121.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2701, pruned_loss=0.047, over 3215238.98 frames. ], batch size: 246, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:10:40,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.533e+02 1.977e+02 2.284e+02 2.754e+02 4.909e+02, threshold=4.569e+02, percent-clipped=0.0 2023-04-30 04:10:45,242 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147407.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:11:23,835 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147434.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:11:49,964 INFO [train.py:904] (7/8) Epoch 15, batch 5350, loss[loss=0.1874, simple_loss=0.276, pruned_loss=0.04942, over 16664.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.2694, pruned_loss=0.04671, over 3213968.94 frames. ], batch size: 134, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:12:14,803 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147468.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:12:51,775 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2412, 3.4331, 3.6046, 3.5763, 3.5759, 3.4057, 3.4378, 3.4651], device='cuda:7'), covar=tensor([0.0328, 0.0607, 0.0419, 0.0430, 0.0461, 0.0442, 0.0704, 0.0456], device='cuda:7'), in_proj_covar=tensor([0.0360, 0.0386, 0.0381, 0.0362, 0.0430, 0.0402, 0.0498, 0.0323], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 04:12:53,687 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147495.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:13:01,232 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147500.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:13:03,291 INFO [train.py:904] (7/8) Epoch 15, batch 5400, loss[loss=0.1817, simple_loss=0.2723, pruned_loss=0.04558, over 16584.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2725, pruned_loss=0.04807, over 3181494.35 frames. ], batch size: 62, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:13:07,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.104e+02 2.571e+02 3.291e+02 5.861e+02, threshold=5.143e+02, percent-clipped=4.0 2023-04-30 04:13:34,916 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7488, 1.3185, 1.6843, 1.6783, 1.7293, 1.9750, 1.5481, 1.8318], device='cuda:7'), covar=tensor([0.0196, 0.0331, 0.0178, 0.0241, 0.0226, 0.0148, 0.0356, 0.0108], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0181, 0.0167, 0.0172, 0.0179, 0.0136, 0.0184, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:14:13,939 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147548.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:14:20,734 INFO [train.py:904] (7/8) Epoch 15, batch 5450, loss[loss=0.2455, simple_loss=0.3282, pruned_loss=0.0814, over 15316.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2757, pruned_loss=0.04951, over 3166571.97 frames. ], batch size: 190, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:15:32,500 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147597.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:15:40,038 INFO [train.py:904] (7/8) Epoch 15, batch 5500, loss[loss=0.2843, simple_loss=0.351, pruned_loss=0.1088, over 11794.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2836, pruned_loss=0.05479, over 3133864.87 frames. ], batch size: 247, lr: 4.53e-03, grad_scale: 4.0 2023-04-30 04:15:45,687 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.456e+02 3.016e+02 4.198e+02 6.055e+02, threshold=6.032e+02, percent-clipped=7.0 2023-04-30 04:16:02,479 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0736, 5.3467, 5.0642, 5.0915, 4.8306, 4.6575, 4.7666, 5.4353], device='cuda:7'), covar=tensor([0.1073, 0.0804, 0.0975, 0.0777, 0.0778, 0.0815, 0.1071, 0.0803], device='cuda:7'), in_proj_covar=tensor([0.0593, 0.0736, 0.0611, 0.0528, 0.0465, 0.0473, 0.0610, 0.0559], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:16:03,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6931, 1.8229, 2.3455, 2.6587, 2.6493, 3.1291, 1.9566, 3.0388], device='cuda:7'), covar=tensor([0.0191, 0.0386, 0.0281, 0.0269, 0.0229, 0.0126, 0.0406, 0.0092], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0180, 0.0166, 0.0171, 0.0179, 0.0136, 0.0183, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:16:58,378 INFO [train.py:904] (7/8) Epoch 15, batch 5550, loss[loss=0.2108, simple_loss=0.3016, pruned_loss=0.06004, over 16769.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2905, pruned_loss=0.05953, over 3113937.15 frames. ], batch size: 83, lr: 4.53e-03, grad_scale: 4.0 2023-04-30 04:17:11,403 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147659.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:17:22,989 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147666.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:17:58,780 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5556, 1.8026, 2.1797, 2.5625, 2.5220, 2.9275, 1.8468, 2.7836], device='cuda:7'), covar=tensor([0.0167, 0.0392, 0.0267, 0.0243, 0.0240, 0.0144, 0.0422, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0181, 0.0167, 0.0171, 0.0180, 0.0137, 0.0184, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:18:21,671 INFO [train.py:904] (7/8) Epoch 15, batch 5600, loss[loss=0.1846, simple_loss=0.2722, pruned_loss=0.04847, over 17106.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2956, pruned_loss=0.06377, over 3080768.19 frames. ], batch size: 47, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:18:28,276 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.312e+02 3.387e+02 4.165e+02 5.212e+02 1.017e+03, threshold=8.330e+02, percent-clipped=15.0 2023-04-30 04:18:53,091 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147720.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:19:35,071 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7682, 1.8268, 2.3703, 2.7383, 2.6679, 3.0930, 1.9429, 2.9758], device='cuda:7'), covar=tensor([0.0155, 0.0428, 0.0238, 0.0233, 0.0221, 0.0131, 0.0422, 0.0105], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0181, 0.0166, 0.0170, 0.0178, 0.0136, 0.0183, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:19:46,192 INFO [train.py:904] (7/8) Epoch 15, batch 5650, loss[loss=0.2239, simple_loss=0.3066, pruned_loss=0.07062, over 16752.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.3006, pruned_loss=0.06802, over 3051892.04 frames. ], batch size: 89, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:20:04,666 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147763.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:20:47,732 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=147790.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:21:05,470 INFO [train.py:904] (7/8) Epoch 15, batch 5700, loss[loss=0.3128, simple_loss=0.3574, pruned_loss=0.134, over 11342.00 frames. ], tot_loss[loss=0.2216, simple_loss=0.3027, pruned_loss=0.07023, over 3027516.05 frames. ], batch size: 248, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:21:11,568 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.489e+02 3.643e+02 4.297e+02 4.973e+02 1.168e+03, threshold=8.593e+02, percent-clipped=2.0 2023-04-30 04:21:22,317 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 04:21:26,351 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:22:23,104 INFO [train.py:904] (7/8) Epoch 15, batch 5750, loss[loss=0.2184, simple_loss=0.3074, pruned_loss=0.06472, over 16345.00 frames. ], tot_loss[loss=0.225, simple_loss=0.306, pruned_loss=0.07199, over 3010925.73 frames. ], batch size: 68, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:22:32,848 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147857.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:23:03,247 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147876.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:23:37,366 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147897.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:23:44,682 INFO [train.py:904] (7/8) Epoch 15, batch 5800, loss[loss=0.2264, simple_loss=0.3082, pruned_loss=0.0723, over 16865.00 frames. ], tot_loss[loss=0.223, simple_loss=0.3052, pruned_loss=0.07042, over 3023353.31 frames. ], batch size: 116, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:23:51,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.104e+02 2.959e+02 3.306e+02 4.369e+02 1.266e+03, threshold=6.612e+02, percent-clipped=1.0 2023-04-30 04:24:00,786 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=147911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:24:12,256 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147918.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:24:55,226 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=147945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:25:05,481 INFO [train.py:904] (7/8) Epoch 15, batch 5850, loss[loss=0.1892, simple_loss=0.2829, pruned_loss=0.04776, over 16576.00 frames. ], tot_loss[loss=0.2187, simple_loss=0.3019, pruned_loss=0.06776, over 3041743.76 frames. ], batch size: 62, lr: 4.53e-03, grad_scale: 8.0 2023-04-30 04:25:29,215 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=147966.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:25:38,427 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=147972.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:26:30,243 INFO [train.py:904] (7/8) Epoch 15, batch 5900, loss[loss=0.2244, simple_loss=0.2955, pruned_loss=0.07669, over 11447.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3017, pruned_loss=0.06781, over 3050441.21 frames. ], batch size: 246, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:26:39,377 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.034e+02 2.736e+02 3.208e+02 4.005e+02 8.372e+02, threshold=6.416e+02, percent-clipped=2.0 2023-04-30 04:26:52,639 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148014.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:26:54,060 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148015.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:27:50,065 INFO [train.py:904] (7/8) Epoch 15, batch 5950, loss[loss=0.2491, simple_loss=0.338, pruned_loss=0.08009, over 16125.00 frames. ], tot_loss[loss=0.2183, simple_loss=0.303, pruned_loss=0.06677, over 3058860.69 frames. ], batch size: 35, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:28:09,470 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148063.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:28:19,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1695, 3.6365, 3.6029, 2.1200, 2.9028, 2.4128, 3.5287, 3.8341], device='cuda:7'), covar=tensor([0.0232, 0.0605, 0.0519, 0.1834, 0.0820, 0.0894, 0.0582, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0153, 0.0160, 0.0146, 0.0138, 0.0126, 0.0139, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 04:28:49,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:29:08,198 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5345, 3.6083, 2.0732, 4.1706, 2.6601, 4.1272, 2.2711, 2.8374], device='cuda:7'), covar=tensor([0.0268, 0.0384, 0.1684, 0.0175, 0.0820, 0.0456, 0.1491, 0.0720], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0171, 0.0189, 0.0143, 0.0169, 0.0210, 0.0196, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 04:29:08,855 INFO [train.py:904] (7/8) Epoch 15, batch 6000, loss[loss=0.1843, simple_loss=0.2774, pruned_loss=0.04562, over 16723.00 frames. ], tot_loss[loss=0.2174, simple_loss=0.3018, pruned_loss=0.06649, over 3060937.28 frames. ], batch size: 83, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:29:08,855 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 04:29:19,438 INFO [train.py:938] (7/8) Epoch 15, validation: loss=0.1559, simple_loss=0.2691, pruned_loss=0.0214, over 944034.00 frames. 2023-04-30 04:29:19,438 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 04:29:26,136 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.753e+02 3.314e+02 4.296e+02 7.936e+02, threshold=6.628e+02, percent-clipped=2.0 2023-04-30 04:29:33,717 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148111.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:30:15,359 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148138.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:30:23,174 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-04-30 04:30:36,863 INFO [train.py:904] (7/8) Epoch 15, batch 6050, loss[loss=0.2201, simple_loss=0.2899, pruned_loss=0.07514, over 11700.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.3004, pruned_loss=0.06599, over 3064826.92 frames. ], batch size: 246, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:30:54,925 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-04-30 04:31:08,634 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148171.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:31:59,327 INFO [train.py:904] (7/8) Epoch 15, batch 6100, loss[loss=0.2484, simple_loss=0.3123, pruned_loss=0.09229, over 11559.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.3001, pruned_loss=0.06538, over 3068598.13 frames. ], batch size: 248, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:32:08,601 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.651e+02 3.170e+02 3.946e+02 8.387e+02, threshold=6.339e+02, percent-clipped=2.0 2023-04-30 04:32:18,637 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148213.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:33:18,254 INFO [train.py:904] (7/8) Epoch 15, batch 6150, loss[loss=0.2023, simple_loss=0.2875, pruned_loss=0.05855, over 16685.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2968, pruned_loss=0.06395, over 3087235.17 frames. ], batch size: 76, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:33:43,179 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148267.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:34:15,559 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148287.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:34:34,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3159, 3.9030, 3.8752, 2.5375, 3.5182, 3.9129, 3.6896, 2.0371], device='cuda:7'), covar=tensor([0.0519, 0.0036, 0.0039, 0.0387, 0.0088, 0.0082, 0.0066, 0.0461], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0075, 0.0074, 0.0131, 0.0088, 0.0098, 0.0086, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 04:34:37,602 INFO [train.py:904] (7/8) Epoch 15, batch 6200, loss[loss=0.2134, simple_loss=0.2913, pruned_loss=0.06771, over 16643.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2951, pruned_loss=0.0632, over 3093963.71 frames. ], batch size: 134, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:34:46,168 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.083e+02 3.031e+02 3.685e+02 4.615e+02 1.155e+03, threshold=7.370e+02, percent-clipped=8.0 2023-04-30 04:34:58,629 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148315.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:35:15,655 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6280, 4.6339, 5.0914, 5.0402, 5.0265, 4.7117, 4.6949, 4.4446], device='cuda:7'), covar=tensor([0.0269, 0.0525, 0.0307, 0.0347, 0.0449, 0.0380, 0.0919, 0.0479], device='cuda:7'), in_proj_covar=tensor([0.0365, 0.0392, 0.0385, 0.0368, 0.0438, 0.0409, 0.0504, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 04:35:49,256 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148348.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:35:54,866 INFO [train.py:904] (7/8) Epoch 15, batch 6250, loss[loss=0.1994, simple_loss=0.2892, pruned_loss=0.05477, over 16714.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2948, pruned_loss=0.06269, over 3093403.77 frames. ], batch size: 134, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:35:57,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5548, 1.6689, 2.1514, 2.4786, 2.5230, 2.8153, 1.7736, 2.7447], device='cuda:7'), covar=tensor([0.0144, 0.0431, 0.0251, 0.0230, 0.0225, 0.0147, 0.0438, 0.0097], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0179, 0.0163, 0.0167, 0.0176, 0.0135, 0.0181, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:36:11,839 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148363.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:37:11,938 INFO [train.py:904] (7/8) Epoch 15, batch 6300, loss[loss=0.2161, simple_loss=0.2971, pruned_loss=0.06757, over 16663.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2947, pruned_loss=0.06262, over 3089133.04 frames. ], batch size: 134, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:37:21,869 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 2.809e+02 3.245e+02 4.323e+02 1.479e+03, threshold=6.491e+02, percent-clipped=2.0 2023-04-30 04:37:57,785 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148429.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:38:33,533 INFO [train.py:904] (7/8) Epoch 15, batch 6350, loss[loss=0.2656, simple_loss=0.3235, pruned_loss=0.1038, over 11361.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2958, pruned_loss=0.06428, over 3069260.61 frames. ], batch size: 250, lr: 4.52e-03, grad_scale: 4.0 2023-04-30 04:39:03,850 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148471.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:39:10,501 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-30 04:39:34,196 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148490.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:39:45,787 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:39:51,478 INFO [train.py:904] (7/8) Epoch 15, batch 6400, loss[loss=0.2711, simple_loss=0.3355, pruned_loss=0.1034, over 10998.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2963, pruned_loss=0.06553, over 3053871.37 frames. ], batch size: 247, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:39:59,987 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.110e+02 3.263e+02 3.782e+02 4.494e+02 8.205e+02, threshold=7.565e+02, percent-clipped=3.0 2023-04-30 04:40:09,591 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148513.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:40:18,146 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148519.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:41:08,007 INFO [train.py:904] (7/8) Epoch 15, batch 6450, loss[loss=0.2067, simple_loss=0.2724, pruned_loss=0.07053, over 11601.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2956, pruned_loss=0.06393, over 3080615.07 frames. ], batch size: 247, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:41:12,480 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8288, 4.6691, 4.8395, 5.0409, 5.1669, 4.6187, 5.1523, 5.1653], device='cuda:7'), covar=tensor([0.1625, 0.1085, 0.1410, 0.0664, 0.0536, 0.0820, 0.0619, 0.0526], device='cuda:7'), in_proj_covar=tensor([0.0567, 0.0700, 0.0841, 0.0712, 0.0537, 0.0563, 0.0569, 0.0666], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:41:15,042 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0412, 2.4466, 2.6194, 1.9379, 2.6971, 2.7607, 2.4782, 2.4096], device='cuda:7'), covar=tensor([0.0681, 0.0223, 0.0196, 0.1008, 0.0115, 0.0267, 0.0390, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0105, 0.0091, 0.0138, 0.0073, 0.0115, 0.0123, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 04:41:19,824 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148559.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:41:22,186 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148561.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:41:32,442 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148567.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:42:04,653 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7489, 3.9372, 2.7803, 2.2816, 2.7636, 2.2923, 4.0354, 3.4872], device='cuda:7'), covar=tensor([0.2668, 0.0668, 0.1820, 0.2366, 0.2436, 0.1946, 0.0502, 0.1120], device='cuda:7'), in_proj_covar=tensor([0.0307, 0.0259, 0.0288, 0.0289, 0.0282, 0.0233, 0.0273, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:42:27,004 INFO [train.py:904] (7/8) Epoch 15, batch 6500, loss[loss=0.2048, simple_loss=0.2826, pruned_loss=0.06352, over 15161.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2947, pruned_loss=0.06424, over 3086351.74 frames. ], batch size: 190, lr: 4.52e-03, grad_scale: 8.0 2023-04-30 04:42:37,001 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 2.869e+02 3.331e+02 4.011e+02 8.248e+02, threshold=6.661e+02, percent-clipped=1.0 2023-04-30 04:42:47,765 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:43:33,198 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:43:47,254 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148651.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:43:47,993 INFO [train.py:904] (7/8) Epoch 15, batch 6550, loss[loss=0.1945, simple_loss=0.3011, pruned_loss=0.04392, over 16699.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2974, pruned_loss=0.06544, over 3071749.01 frames. ], batch size: 89, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:44:14,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0640, 2.9685, 3.1769, 1.8061, 3.3061, 3.3421, 2.6887, 2.5967], device='cuda:7'), covar=tensor([0.0769, 0.0213, 0.0196, 0.1052, 0.0077, 0.0196, 0.0398, 0.0447], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0105, 0.0091, 0.0137, 0.0073, 0.0114, 0.0123, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 04:44:47,735 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-30 04:45:05,100 INFO [train.py:904] (7/8) Epoch 15, batch 6600, loss[loss=0.2549, simple_loss=0.3182, pruned_loss=0.09582, over 11617.00 frames. ], tot_loss[loss=0.2153, simple_loss=0.2995, pruned_loss=0.06556, over 3084199.35 frames. ], batch size: 248, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:45:13,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 3.041e+02 3.810e+02 4.934e+02 1.256e+03, threshold=7.620e+02, percent-clipped=11.0 2023-04-30 04:45:19,921 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148712.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:45:27,034 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=148716.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:46:00,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9986, 2.0478, 2.2150, 3.4720, 2.0339, 2.3790, 2.1792, 2.1584], device='cuda:7'), covar=tensor([0.1296, 0.3408, 0.2479, 0.0566, 0.3858, 0.2326, 0.3312, 0.3149], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0413, 0.0343, 0.0317, 0.0421, 0.0475, 0.0380, 0.0484], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:46:22,124 INFO [train.py:904] (7/8) Epoch 15, batch 6650, loss[loss=0.2389, simple_loss=0.3255, pruned_loss=0.0761, over 15324.00 frames. ], tot_loss[loss=0.215, simple_loss=0.299, pruned_loss=0.06548, over 3098160.30 frames. ], batch size: 191, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:46:34,196 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5377, 3.6691, 2.7869, 2.0510, 2.3981, 2.2597, 3.8522, 3.2683], device='cuda:7'), covar=tensor([0.2877, 0.0714, 0.1764, 0.2549, 0.2625, 0.1991, 0.0512, 0.1180], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0262, 0.0291, 0.0292, 0.0285, 0.0235, 0.0276, 0.0312], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 04:47:00,950 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=148777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:47:13,485 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148785.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:47:39,118 INFO [train.py:904] (7/8) Epoch 15, batch 6700, loss[loss=0.1902, simple_loss=0.2801, pruned_loss=0.05019, over 16925.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2975, pruned_loss=0.06545, over 3089639.17 frames. ], batch size: 116, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:47:47,147 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.803e+02 3.506e+02 4.354e+02 9.571e+02, threshold=7.012e+02, percent-clipped=2.0 2023-04-30 04:47:57,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2865, 3.3553, 2.0142, 3.7691, 2.4855, 3.7249, 2.2122, 2.6859], device='cuda:7'), covar=tensor([0.0283, 0.0443, 0.1680, 0.0176, 0.0897, 0.0579, 0.1544, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0188, 0.0142, 0.0168, 0.0209, 0.0195, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 04:48:54,929 INFO [train.py:904] (7/8) Epoch 15, batch 6750, loss[loss=0.1843, simple_loss=0.2739, pruned_loss=0.0474, over 16912.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2962, pruned_loss=0.0651, over 3096285.92 frames. ], batch size: 90, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:48:59,145 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=148854.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:49:01,698 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8342, 4.8500, 4.6753, 4.3286, 4.3078, 4.7334, 4.6036, 4.4348], device='cuda:7'), covar=tensor([0.0535, 0.0364, 0.0271, 0.0286, 0.0967, 0.0387, 0.0363, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0262, 0.0363, 0.0310, 0.0293, 0.0325, 0.0340, 0.0209, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:49:08,882 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-04-30 04:50:09,787 INFO [train.py:904] (7/8) Epoch 15, batch 6800, loss[loss=0.2049, simple_loss=0.2958, pruned_loss=0.05699, over 16852.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2967, pruned_loss=0.06525, over 3096831.19 frames. ], batch size: 102, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:50:21,238 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.768e+02 2.837e+02 3.542e+02 4.429e+02 7.153e+02, threshold=7.083e+02, percent-clipped=1.0 2023-04-30 04:51:15,917 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=148943.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:51:28,460 INFO [train.py:904] (7/8) Epoch 15, batch 6850, loss[loss=0.1853, simple_loss=0.2888, pruned_loss=0.04084, over 16406.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2975, pruned_loss=0.06487, over 3101934.85 frames. ], batch size: 68, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:51:59,502 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8089, 1.3642, 1.7046, 1.7207, 1.8034, 1.9199, 1.5710, 1.8481], device='cuda:7'), covar=tensor([0.0197, 0.0348, 0.0180, 0.0220, 0.0239, 0.0156, 0.0362, 0.0106], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0179, 0.0164, 0.0168, 0.0177, 0.0135, 0.0181, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:52:26,511 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=148991.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:52:43,277 INFO [train.py:904] (7/8) Epoch 15, batch 6900, loss[loss=0.2742, simple_loss=0.3332, pruned_loss=0.1075, over 11147.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2992, pruned_loss=0.06379, over 3114894.35 frames. ], batch size: 246, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:52:52,145 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149007.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:52:54,701 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.687e+02 3.079e+02 3.976e+02 7.116e+02, threshold=6.157e+02, percent-clipped=1.0 2023-04-30 04:53:13,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5424, 4.4802, 4.8973, 4.8851, 4.8905, 4.5946, 4.5763, 4.4313], device='cuda:7'), covar=tensor([0.0282, 0.0508, 0.0386, 0.0402, 0.0423, 0.0361, 0.0888, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0371, 0.0395, 0.0387, 0.0372, 0.0440, 0.0412, 0.0511, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 04:53:29,706 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8058, 5.1109, 4.8857, 4.8679, 4.6190, 4.6247, 4.5060, 5.2082], device='cuda:7'), covar=tensor([0.1181, 0.0819, 0.0984, 0.0778, 0.0838, 0.0983, 0.1119, 0.0842], device='cuda:7'), in_proj_covar=tensor([0.0596, 0.0736, 0.0612, 0.0534, 0.0465, 0.0479, 0.0614, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:53:58,323 INFO [train.py:904] (7/8) Epoch 15, batch 6950, loss[loss=0.2336, simple_loss=0.3225, pruned_loss=0.07236, over 15481.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.3011, pruned_loss=0.06586, over 3100009.01 frames. ], batch size: 191, lr: 4.51e-03, grad_scale: 8.0 2023-04-30 04:54:27,432 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149072.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:54:43,118 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9936, 2.7422, 2.7883, 2.0560, 2.6159, 2.1464, 2.6640, 2.9127], device='cuda:7'), covar=tensor([0.0290, 0.0680, 0.0503, 0.1626, 0.0739, 0.0883, 0.0580, 0.0652], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0154, 0.0163, 0.0147, 0.0139, 0.0127, 0.0140, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 04:54:45,986 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149085.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:55:02,999 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 04:55:10,882 INFO [train.py:904] (7/8) Epoch 15, batch 7000, loss[loss=0.2097, simple_loss=0.3111, pruned_loss=0.05414, over 16825.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.3016, pruned_loss=0.06558, over 3083261.72 frames. ], batch size: 83, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:55:23,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.893e+02 3.767e+02 4.998e+02 1.151e+03, threshold=7.533e+02, percent-clipped=10.0 2023-04-30 04:55:58,208 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149133.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:56:25,569 INFO [train.py:904] (7/8) Epoch 15, batch 7050, loss[loss=0.2106, simple_loss=0.3017, pruned_loss=0.05973, over 15397.00 frames. ], tot_loss[loss=0.216, simple_loss=0.3019, pruned_loss=0.06505, over 3086141.64 frames. ], batch size: 190, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:56:29,221 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149154.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:57:02,727 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 04:57:40,357 INFO [train.py:904] (7/8) Epoch 15, batch 7100, loss[loss=0.217, simple_loss=0.3008, pruned_loss=0.06658, over 16424.00 frames. ], tot_loss[loss=0.2159, simple_loss=0.3011, pruned_loss=0.06537, over 3092675.11 frames. ], batch size: 146, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:57:40,782 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149202.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 04:57:53,984 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 2.990e+02 3.583e+02 4.304e+02 1.223e+03, threshold=7.166e+02, percent-clipped=1.0 2023-04-30 04:58:55,148 INFO [train.py:904] (7/8) Epoch 15, batch 7150, loss[loss=0.2048, simple_loss=0.2882, pruned_loss=0.06067, over 15484.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2986, pruned_loss=0.06469, over 3110778.58 frames. ], batch size: 191, lr: 4.51e-03, grad_scale: 2.0 2023-04-30 04:59:22,684 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8265, 2.0838, 2.3826, 3.1088, 2.1546, 2.3056, 2.2772, 2.1733], device='cuda:7'), covar=tensor([0.1183, 0.3186, 0.2183, 0.0621, 0.3749, 0.2066, 0.2772, 0.3242], device='cuda:7'), in_proj_covar=tensor([0.0376, 0.0413, 0.0343, 0.0318, 0.0422, 0.0475, 0.0381, 0.0484], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 04:59:30,599 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-30 05:00:05,796 INFO [train.py:904] (7/8) Epoch 15, batch 7200, loss[loss=0.1822, simple_loss=0.2775, pruned_loss=0.04351, over 16712.00 frames. ], tot_loss[loss=0.2111, simple_loss=0.2962, pruned_loss=0.06298, over 3094515.45 frames. ], batch size: 89, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:00:13,331 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149307.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:00:17,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.693e+02 3.090e+02 3.791e+02 7.265e+02, threshold=6.181e+02, percent-clipped=1.0 2023-04-30 05:00:27,608 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-30 05:00:39,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0011, 2.4144, 2.6315, 1.8251, 2.7056, 2.8216, 2.4266, 2.4002], device='cuda:7'), covar=tensor([0.0713, 0.0209, 0.0225, 0.1058, 0.0107, 0.0206, 0.0443, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0104, 0.0090, 0.0137, 0.0073, 0.0114, 0.0122, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 05:01:26,144 INFO [train.py:904] (7/8) Epoch 15, batch 7250, loss[loss=0.232, simple_loss=0.2999, pruned_loss=0.08209, over 15339.00 frames. ], tot_loss[loss=0.209, simple_loss=0.294, pruned_loss=0.06198, over 3096486.94 frames. ], batch size: 191, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:01:32,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149355.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:01:46,894 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 05:01:54,572 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0229, 2.3254, 2.3078, 2.7039, 2.0479, 3.2244, 1.8372, 2.7154], device='cuda:7'), covar=tensor([0.1118, 0.0581, 0.1077, 0.0157, 0.0122, 0.0387, 0.1367, 0.0642], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0167, 0.0189, 0.0168, 0.0204, 0.0213, 0.0192, 0.0188], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 05:01:57,337 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=149372.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:02:13,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2192, 2.1285, 2.2079, 4.0278, 2.0857, 2.5581, 2.2247, 2.3621], device='cuda:7'), covar=tensor([0.1154, 0.3520, 0.2721, 0.0447, 0.3968, 0.2404, 0.3448, 0.3239], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0411, 0.0341, 0.0316, 0.0420, 0.0472, 0.0379, 0.0480], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:02:42,286 INFO [train.py:904] (7/8) Epoch 15, batch 7300, loss[loss=0.1899, simple_loss=0.2834, pruned_loss=0.04821, over 16424.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2935, pruned_loss=0.06206, over 3077378.67 frames. ], batch size: 75, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:02:55,639 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 2.888e+02 3.399e+02 4.220e+02 6.338e+02, threshold=6.799e+02, percent-clipped=2.0 2023-04-30 05:03:01,298 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9736, 2.3979, 2.6101, 1.8850, 2.7236, 2.7484, 2.3536, 2.3816], device='cuda:7'), covar=tensor([0.0714, 0.0238, 0.0217, 0.0891, 0.0102, 0.0209, 0.0454, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0105, 0.0091, 0.0137, 0.0073, 0.0114, 0.0122, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 05:03:10,597 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=149420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:03:20,068 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6767, 2.5459, 2.5918, 4.5287, 2.3645, 3.0043, 2.5561, 2.7701], device='cuda:7'), covar=tensor([0.1012, 0.2936, 0.2254, 0.0386, 0.3494, 0.1986, 0.2797, 0.2985], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0411, 0.0341, 0.0316, 0.0420, 0.0472, 0.0380, 0.0481], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:03:58,899 INFO [train.py:904] (7/8) Epoch 15, batch 7350, loss[loss=0.1812, simple_loss=0.2727, pruned_loss=0.04486, over 16984.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2941, pruned_loss=0.06191, over 3089138.62 frames. ], batch size: 50, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:05:17,927 INFO [train.py:904] (7/8) Epoch 15, batch 7400, loss[loss=0.237, simple_loss=0.3071, pruned_loss=0.08349, over 11505.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2948, pruned_loss=0.06232, over 3100713.98 frames. ], batch size: 246, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:05:32,174 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.760e+02 2.749e+02 3.166e+02 4.051e+02 9.173e+02, threshold=6.332e+02, percent-clipped=4.0 2023-04-30 05:05:38,069 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149514.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:05:57,333 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 05:06:37,053 INFO [train.py:904] (7/8) Epoch 15, batch 7450, loss[loss=0.2083, simple_loss=0.2957, pruned_loss=0.06047, over 16850.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2959, pruned_loss=0.06385, over 3088365.21 frames. ], batch size: 116, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:07:00,516 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149565.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:07:16,516 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149575.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:07:56,101 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5316, 3.5659, 3.3580, 3.1097, 3.2054, 3.4607, 3.2898, 3.2871], device='cuda:7'), covar=tensor([0.0582, 0.0561, 0.0252, 0.0251, 0.0509, 0.0404, 0.1314, 0.0493], device='cuda:7'), in_proj_covar=tensor([0.0256, 0.0355, 0.0303, 0.0288, 0.0319, 0.0333, 0.0207, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:08:00,173 INFO [train.py:904] (7/8) Epoch 15, batch 7500, loss[loss=0.2151, simple_loss=0.2996, pruned_loss=0.06527, over 16691.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2964, pruned_loss=0.06353, over 3076852.14 frames. ], batch size: 76, lr: 4.50e-03, grad_scale: 2.0 2023-04-30 05:08:16,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 2.987e+02 3.634e+02 4.459e+02 1.121e+03, threshold=7.267e+02, percent-clipped=5.0 2023-04-30 05:08:40,060 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149626.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 05:09:18,804 INFO [train.py:904] (7/8) Epoch 15, batch 7550, loss[loss=0.2307, simple_loss=0.2948, pruned_loss=0.08329, over 11361.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2955, pruned_loss=0.06385, over 3076440.31 frames. ], batch size: 248, lr: 4.50e-03, grad_scale: 2.0 2023-04-30 05:09:20,257 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7295, 3.7841, 2.4198, 4.4638, 2.9881, 4.3632, 2.4730, 3.0515], device='cuda:7'), covar=tensor([0.0245, 0.0360, 0.1494, 0.0143, 0.0697, 0.0465, 0.1436, 0.0687], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0189, 0.0142, 0.0168, 0.0209, 0.0197, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 05:09:27,755 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0813, 4.1096, 2.7674, 4.9625, 3.3734, 4.8248, 2.6494, 3.2791], device='cuda:7'), covar=tensor([0.0243, 0.0360, 0.1440, 0.0124, 0.0647, 0.0416, 0.1486, 0.0677], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0190, 0.0142, 0.0169, 0.0210, 0.0198, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 05:10:35,818 INFO [train.py:904] (7/8) Epoch 15, batch 7600, loss[loss=0.2187, simple_loss=0.2987, pruned_loss=0.0693, over 16454.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2946, pruned_loss=0.06435, over 3071888.59 frames. ], batch size: 146, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:10:50,668 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 2.884e+02 3.508e+02 4.101e+02 6.446e+02, threshold=7.017e+02, percent-clipped=0.0 2023-04-30 05:11:21,680 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9186, 4.8880, 4.7487, 4.1195, 4.8085, 1.7817, 4.5610, 4.5725], device='cuda:7'), covar=tensor([0.0075, 0.0062, 0.0151, 0.0325, 0.0078, 0.2452, 0.0091, 0.0179], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0131, 0.0178, 0.0165, 0.0149, 0.0189, 0.0166, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:11:45,941 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=149746.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 05:11:55,075 INFO [train.py:904] (7/8) Epoch 15, batch 7650, loss[loss=0.2317, simple_loss=0.3142, pruned_loss=0.07461, over 15311.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2953, pruned_loss=0.065, over 3071282.04 frames. ], batch size: 190, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:13:13,697 INFO [train.py:904] (7/8) Epoch 15, batch 7700, loss[loss=0.2032, simple_loss=0.2809, pruned_loss=0.06279, over 16668.00 frames. ], tot_loss[loss=0.2124, simple_loss=0.2952, pruned_loss=0.06483, over 3087068.41 frames. ], batch size: 57, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:13:22,689 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=149807.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 05:13:29,267 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.235e+02 4.085e+02 5.006e+02 1.161e+03, threshold=8.169e+02, percent-clipped=3.0 2023-04-30 05:13:54,600 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5105, 1.6060, 2.0877, 2.4095, 2.3861, 2.7719, 1.6500, 2.6709], device='cuda:7'), covar=tensor([0.0188, 0.0491, 0.0276, 0.0289, 0.0278, 0.0165, 0.0541, 0.0126], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0181, 0.0165, 0.0170, 0.0179, 0.0137, 0.0184, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:14:05,420 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3214, 2.0584, 1.5965, 1.8708, 2.3738, 2.0892, 2.1798, 2.5389], device='cuda:7'), covar=tensor([0.0171, 0.0372, 0.0478, 0.0427, 0.0200, 0.0306, 0.0168, 0.0217], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0216, 0.0211, 0.0212, 0.0216, 0.0216, 0.0217, 0.0210], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:14:10,657 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-30 05:14:33,190 INFO [train.py:904] (7/8) Epoch 15, batch 7750, loss[loss=0.1917, simple_loss=0.2802, pruned_loss=0.05156, over 17170.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2957, pruned_loss=0.06484, over 3099478.27 frames. ], batch size: 46, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:15:02,335 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149870.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:15:30,650 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 05:15:35,105 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1479, 4.2258, 4.0114, 3.7812, 3.7178, 4.1062, 3.8606, 3.8503], device='cuda:7'), covar=tensor([0.0685, 0.0604, 0.0302, 0.0299, 0.0843, 0.0546, 0.0765, 0.0715], device='cuda:7'), in_proj_covar=tensor([0.0257, 0.0355, 0.0303, 0.0288, 0.0319, 0.0334, 0.0208, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:15:51,919 INFO [train.py:904] (7/8) Epoch 15, batch 7800, loss[loss=0.1961, simple_loss=0.2843, pruned_loss=0.05397, over 17254.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2974, pruned_loss=0.06626, over 3079393.47 frames. ], batch size: 52, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:16:07,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.059e+02 3.237e+02 3.890e+02 4.639e+02 7.802e+02, threshold=7.781e+02, percent-clipped=0.0 2023-04-30 05:16:21,593 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=149921.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 05:17:06,634 INFO [train.py:904] (7/8) Epoch 15, batch 7850, loss[loss=0.2635, simple_loss=0.341, pruned_loss=0.09304, over 16674.00 frames. ], tot_loss[loss=0.214, simple_loss=0.2975, pruned_loss=0.06525, over 3097964.56 frames. ], batch size: 134, lr: 4.50e-03, grad_scale: 4.0 2023-04-30 05:18:25,136 INFO [train.py:904] (7/8) Epoch 15, batch 7900, loss[loss=0.205, simple_loss=0.2928, pruned_loss=0.05862, over 17121.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2968, pruned_loss=0.06435, over 3113851.31 frames. ], batch size: 47, lr: 4.49e-03, grad_scale: 4.0 2023-04-30 05:18:40,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.219e+02 2.824e+02 3.475e+02 4.207e+02 7.504e+02, threshold=6.949e+02, percent-clipped=0.0 2023-04-30 05:19:43,695 INFO [train.py:904] (7/8) Epoch 15, batch 7950, loss[loss=0.2015, simple_loss=0.2879, pruned_loss=0.05755, over 16798.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.298, pruned_loss=0.06574, over 3087291.87 frames. ], batch size: 124, lr: 4.49e-03, grad_scale: 4.0 2023-04-30 05:19:50,697 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-30 05:20:25,865 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150079.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:20:58,678 INFO [train.py:904] (7/8) Epoch 15, batch 8000, loss[loss=0.2188, simple_loss=0.3086, pruned_loss=0.06451, over 16714.00 frames. ], tot_loss[loss=0.2169, simple_loss=0.2995, pruned_loss=0.06716, over 3071522.20 frames. ], batch size: 124, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:20:59,769 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150102.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 05:21:14,671 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.969e+02 2.932e+02 3.545e+02 4.036e+02 7.060e+02, threshold=7.089e+02, percent-clipped=1.0 2023-04-30 05:21:56,026 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150140.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:22:14,257 INFO [train.py:904] (7/8) Epoch 15, batch 8050, loss[loss=0.2214, simple_loss=0.3133, pruned_loss=0.06472, over 16730.00 frames. ], tot_loss[loss=0.2166, simple_loss=0.2994, pruned_loss=0.06694, over 3054317.47 frames. ], batch size: 134, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:22:28,649 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9979, 3.9423, 3.8962, 3.0121, 3.9383, 1.5791, 3.7083, 3.4631], device='cuda:7'), covar=tensor([0.0145, 0.0129, 0.0194, 0.0449, 0.0117, 0.3218, 0.0165, 0.0330], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0132, 0.0180, 0.0167, 0.0152, 0.0192, 0.0167, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:22:34,089 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8743, 2.7370, 2.6196, 1.9810, 2.5426, 2.7551, 2.6206, 1.8976], device='cuda:7'), covar=tensor([0.0376, 0.0071, 0.0060, 0.0315, 0.0110, 0.0091, 0.0091, 0.0359], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0074, 0.0073, 0.0132, 0.0088, 0.0097, 0.0085, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 05:22:41,322 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:23:30,586 INFO [train.py:904] (7/8) Epoch 15, batch 8100, loss[loss=0.1904, simple_loss=0.277, pruned_loss=0.05192, over 17038.00 frames. ], tot_loss[loss=0.2158, simple_loss=0.2987, pruned_loss=0.06645, over 3051934.09 frames. ], batch size: 50, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:23:45,521 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.677e+02 3.415e+02 4.132e+02 1.188e+03, threshold=6.830e+02, percent-clipped=3.0 2023-04-30 05:23:56,139 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150218.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:23:59,912 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:24:45,711 INFO [train.py:904] (7/8) Epoch 15, batch 8150, loss[loss=0.2143, simple_loss=0.2847, pruned_loss=0.07197, over 15304.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2956, pruned_loss=0.0652, over 3061276.89 frames. ], batch size: 190, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:25:11,109 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150269.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:25:19,767 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150274.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:25:28,417 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4557, 4.4721, 4.3105, 4.0223, 3.9907, 4.3873, 4.1616, 4.1037], device='cuda:7'), covar=tensor([0.0572, 0.0434, 0.0277, 0.0303, 0.0934, 0.0423, 0.0550, 0.0698], device='cuda:7'), in_proj_covar=tensor([0.0259, 0.0357, 0.0304, 0.0288, 0.0320, 0.0336, 0.0210, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:26:00,377 INFO [train.py:904] (7/8) Epoch 15, batch 8200, loss[loss=0.2317, simple_loss=0.3083, pruned_loss=0.07753, over 16906.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2926, pruned_loss=0.06399, over 3085156.68 frames. ], batch size: 116, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:26:16,345 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.058e+02 2.882e+02 3.369e+02 3.879e+02 7.748e+02, threshold=6.737e+02, percent-clipped=3.0 2023-04-30 05:26:54,568 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150335.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:27:19,305 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-04-30 05:27:22,327 INFO [train.py:904] (7/8) Epoch 15, batch 8250, loss[loss=0.199, simple_loss=0.293, pruned_loss=0.05248, over 16234.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2918, pruned_loss=0.06144, over 3060236.77 frames. ], batch size: 165, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:28:45,016 INFO [train.py:904] (7/8) Epoch 15, batch 8300, loss[loss=0.1945, simple_loss=0.286, pruned_loss=0.05149, over 15296.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2886, pruned_loss=0.05815, over 3057422.21 frames. ], batch size: 190, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:28:45,721 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150402.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 05:28:57,620 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4506, 2.0935, 2.2174, 4.2038, 2.0688, 2.5529, 2.2352, 2.3248], device='cuda:7'), covar=tensor([0.0958, 0.3834, 0.2829, 0.0372, 0.4291, 0.2528, 0.3686, 0.3304], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0409, 0.0340, 0.0313, 0.0418, 0.0469, 0.0377, 0.0478], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:29:01,273 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.394e+02 2.913e+02 3.586e+02 6.520e+02, threshold=5.826e+02, percent-clipped=0.0 2023-04-30 05:29:39,977 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150435.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:30:05,307 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150450.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 05:30:07,764 INFO [train.py:904] (7/8) Epoch 15, batch 8350, loss[loss=0.1978, simple_loss=0.2948, pruned_loss=0.05043, over 16331.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2878, pruned_loss=0.05589, over 3080377.11 frames. ], batch size: 146, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:30:49,899 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 05:31:29,422 INFO [train.py:904] (7/8) Epoch 15, batch 8400, loss[loss=0.1687, simple_loss=0.262, pruned_loss=0.03775, over 16827.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2849, pruned_loss=0.05391, over 3058359.12 frames. ], batch size: 116, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:31:46,290 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.409e+02 2.912e+02 3.516e+02 5.302e+02, threshold=5.823e+02, percent-clipped=0.0 2023-04-30 05:32:27,439 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150537.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:32:49,413 INFO [train.py:904] (7/8) Epoch 15, batch 8450, loss[loss=0.1803, simple_loss=0.283, pruned_loss=0.0388, over 16674.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2827, pruned_loss=0.05154, over 3064581.57 frames. ], batch size: 89, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:32:51,959 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 05:33:35,618 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7346, 4.6602, 5.1301, 5.1368, 5.0884, 4.8265, 4.7642, 4.5878], device='cuda:7'), covar=tensor([0.0346, 0.0901, 0.0431, 0.0395, 0.0544, 0.0397, 0.1209, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0363, 0.0391, 0.0379, 0.0366, 0.0434, 0.0404, 0.0499, 0.0322], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 05:34:02,193 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9444, 3.9183, 4.2888, 4.2722, 4.2715, 4.0346, 4.0151, 4.0391], device='cuda:7'), covar=tensor([0.0370, 0.0769, 0.0454, 0.0479, 0.0494, 0.0465, 0.1043, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0363, 0.0391, 0.0380, 0.0365, 0.0434, 0.0404, 0.0498, 0.0322], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 05:34:02,278 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150598.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:34:08,635 INFO [train.py:904] (7/8) Epoch 15, batch 8500, loss[loss=0.1775, simple_loss=0.2652, pruned_loss=0.0449, over 16442.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2794, pruned_loss=0.04936, over 3080389.25 frames. ], batch size: 68, lr: 4.49e-03, grad_scale: 8.0 2023-04-30 05:34:18,825 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2744, 4.3208, 4.4818, 4.2996, 4.2526, 4.8238, 4.4135, 4.1391], device='cuda:7'), covar=tensor([0.1509, 0.1962, 0.1991, 0.2031, 0.2615, 0.1104, 0.1546, 0.2410], device='cuda:7'), in_proj_covar=tensor([0.0372, 0.0522, 0.0573, 0.0440, 0.0588, 0.0603, 0.0455, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 05:34:25,006 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 2.200e+02 2.703e+02 3.335e+02 6.908e+02, threshold=5.407e+02, percent-clipped=1.0 2023-04-30 05:34:54,222 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150630.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:35:30,242 INFO [train.py:904] (7/8) Epoch 15, batch 8550, loss[loss=0.2092, simple_loss=0.3061, pruned_loss=0.05617, over 15358.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2769, pruned_loss=0.04813, over 3060924.48 frames. ], batch size: 190, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:35:56,554 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0797, 5.4299, 5.2099, 5.2320, 4.9116, 4.8713, 4.8344, 5.4918], device='cuda:7'), covar=tensor([0.1131, 0.0857, 0.0924, 0.0685, 0.0867, 0.0692, 0.1124, 0.0799], device='cuda:7'), in_proj_covar=tensor([0.0584, 0.0722, 0.0598, 0.0520, 0.0452, 0.0467, 0.0601, 0.0552], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:36:22,899 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2424, 5.8808, 6.0864, 5.7846, 5.8141, 6.3753, 6.0103, 5.7044], device='cuda:7'), covar=tensor([0.0683, 0.1658, 0.1548, 0.1693, 0.1965, 0.0753, 0.1182, 0.1928], device='cuda:7'), in_proj_covar=tensor([0.0367, 0.0516, 0.0566, 0.0434, 0.0579, 0.0595, 0.0448, 0.0583], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 05:37:12,312 INFO [train.py:904] (7/8) Epoch 15, batch 8600, loss[loss=0.178, simple_loss=0.2601, pruned_loss=0.0479, over 12257.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2767, pruned_loss=0.04728, over 3036010.85 frames. ], batch size: 247, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:37:32,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.969e+02 2.521e+02 3.034e+02 3.543e+02 4.971e+02, threshold=6.069e+02, percent-clipped=0.0 2023-04-30 05:38:11,791 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-30 05:38:15,005 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150732.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:38:20,247 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150735.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:38:45,611 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6914, 3.5142, 4.0126, 1.9379, 4.1922, 4.2631, 3.1009, 3.3162], device='cuda:7'), covar=tensor([0.0669, 0.0272, 0.0191, 0.1146, 0.0046, 0.0094, 0.0363, 0.0355], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0102, 0.0087, 0.0134, 0.0070, 0.0110, 0.0119, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 05:38:51,971 INFO [train.py:904] (7/8) Epoch 15, batch 8650, loss[loss=0.1722, simple_loss=0.2704, pruned_loss=0.03704, over 16638.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2753, pruned_loss=0.04605, over 3047998.22 frames. ], batch size: 134, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:39:25,138 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0293, 4.0270, 3.9498, 3.3341, 3.9726, 1.7842, 3.7677, 3.6941], device='cuda:7'), covar=tensor([0.0098, 0.0087, 0.0146, 0.0258, 0.0096, 0.2568, 0.0126, 0.0209], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0130, 0.0178, 0.0162, 0.0150, 0.0191, 0.0164, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:40:03,719 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150783.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:40:23,700 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150793.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 05:40:39,991 INFO [train.py:904] (7/8) Epoch 15, batch 8700, loss[loss=0.1684, simple_loss=0.2548, pruned_loss=0.04103, over 12245.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2729, pruned_loss=0.04523, over 3064396.95 frames. ], batch size: 248, lr: 4.48e-03, grad_scale: 4.0 2023-04-30 05:41:01,902 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.169e+02 2.751e+02 3.261e+02 6.645e+02, threshold=5.502e+02, percent-clipped=1.0 2023-04-30 05:41:20,604 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9573, 2.0855, 2.2910, 3.2809, 2.1763, 2.3434, 2.2961, 2.1966], device='cuda:7'), covar=tensor([0.1106, 0.3658, 0.2575, 0.0585, 0.4297, 0.2373, 0.3351, 0.3438], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0409, 0.0342, 0.0312, 0.0418, 0.0467, 0.0376, 0.0476], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:42:16,618 INFO [train.py:904] (7/8) Epoch 15, batch 8750, loss[loss=0.2127, simple_loss=0.3098, pruned_loss=0.05774, over 16876.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2726, pruned_loss=0.0447, over 3056072.76 frames. ], batch size: 116, lr: 4.48e-03, grad_scale: 4.0 2023-04-30 05:43:06,123 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9881, 5.3646, 5.1403, 5.1198, 4.8747, 4.7595, 4.7246, 5.4535], device='cuda:7'), covar=tensor([0.1109, 0.0791, 0.0794, 0.0703, 0.0701, 0.0806, 0.1109, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0580, 0.0718, 0.0589, 0.0516, 0.0450, 0.0465, 0.0595, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:43:52,662 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=150893.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:44:06,429 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=150900.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:44:08,767 INFO [train.py:904] (7/8) Epoch 15, batch 8800, loss[loss=0.1725, simple_loss=0.2707, pruned_loss=0.03717, over 15405.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2704, pruned_loss=0.04348, over 3031973.93 frames. ], batch size: 191, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:44:28,835 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.483e+02 3.172e+02 3.684e+02 7.481e+02, threshold=6.344e+02, percent-clipped=2.0 2023-04-30 05:45:06,453 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=150930.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:45:50,526 INFO [train.py:904] (7/8) Epoch 15, batch 8850, loss[loss=0.1575, simple_loss=0.2572, pruned_loss=0.02889, over 17181.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2719, pruned_loss=0.04294, over 3009416.96 frames. ], batch size: 46, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:46:10,288 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=150961.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:46:46,307 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=150978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:47:36,874 INFO [train.py:904] (7/8) Epoch 15, batch 8900, loss[loss=0.1611, simple_loss=0.2593, pruned_loss=0.03142, over 16768.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2724, pruned_loss=0.04217, over 3018385.25 frames. ], batch size: 76, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:47:54,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4306, 3.3610, 3.4864, 3.5510, 3.5932, 3.2949, 3.5656, 3.6320], device='cuda:7'), covar=tensor([0.1177, 0.0843, 0.0893, 0.0593, 0.0558, 0.2128, 0.0684, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0542, 0.0673, 0.0795, 0.0687, 0.0518, 0.0542, 0.0550, 0.0636], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:47:59,490 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.270e+02 2.672e+02 3.305e+02 7.174e+02, threshold=5.344e+02, percent-clipped=2.0 2023-04-30 05:48:55,621 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1679, 3.2677, 1.9096, 3.5139, 2.3783, 3.4909, 2.0854, 2.6586], device='cuda:7'), covar=tensor([0.0263, 0.0305, 0.1533, 0.0155, 0.0837, 0.0494, 0.1498, 0.0652], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0160, 0.0181, 0.0134, 0.0162, 0.0197, 0.0191, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-30 05:49:42,116 INFO [train.py:904] (7/8) Epoch 15, batch 8950, loss[loss=0.1792, simple_loss=0.2651, pruned_loss=0.04669, over 12850.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2725, pruned_loss=0.04264, over 3038578.37 frames. ], batch size: 248, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:50:48,714 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-04-30 05:51:02,411 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151088.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 05:51:31,001 INFO [train.py:904] (7/8) Epoch 15, batch 9000, loss[loss=0.1795, simple_loss=0.2633, pruned_loss=0.0478, over 12064.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2691, pruned_loss=0.04105, over 3054794.93 frames. ], batch size: 248, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:51:31,002 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 05:51:40,826 INFO [train.py:938] (7/8) Epoch 15, validation: loss=0.15, simple_loss=0.2539, pruned_loss=0.02307, over 944034.00 frames. 2023-04-30 05:51:40,827 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 05:51:41,747 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9976, 3.1761, 3.1929, 2.2403, 2.9631, 3.2145, 3.1321, 1.9740], device='cuda:7'), covar=tensor([0.0466, 0.0046, 0.0048, 0.0355, 0.0083, 0.0072, 0.0069, 0.0433], device='cuda:7'), in_proj_covar=tensor([0.0128, 0.0072, 0.0071, 0.0129, 0.0085, 0.0094, 0.0082, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 05:51:49,977 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2379, 5.5779, 5.3617, 5.3671, 5.0904, 5.0010, 5.0025, 5.6664], device='cuda:7'), covar=tensor([0.1200, 0.0774, 0.0788, 0.0677, 0.0716, 0.0675, 0.0971, 0.0713], device='cuda:7'), in_proj_covar=tensor([0.0577, 0.0715, 0.0585, 0.0513, 0.0447, 0.0463, 0.0590, 0.0543], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:51:53,717 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9461, 4.0281, 4.3418, 4.2879, 4.3474, 4.0926, 4.0164, 4.0467], device='cuda:7'), covar=tensor([0.0532, 0.0820, 0.0590, 0.0794, 0.0733, 0.0607, 0.1276, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0350, 0.0375, 0.0365, 0.0351, 0.0419, 0.0388, 0.0476, 0.0312], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 05:52:03,720 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.164e+02 2.577e+02 3.241e+02 6.734e+02, threshold=5.154e+02, percent-clipped=2.0 2023-04-30 05:53:23,285 INFO [train.py:904] (7/8) Epoch 15, batch 9050, loss[loss=0.1573, simple_loss=0.2506, pruned_loss=0.03204, over 16870.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.27, pruned_loss=0.04152, over 3055903.70 frames. ], batch size: 102, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:53:42,584 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 05:54:28,336 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151184.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:54:51,887 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151193.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:55:09,981 INFO [train.py:904] (7/8) Epoch 15, batch 9100, loss[loss=0.1698, simple_loss=0.2585, pruned_loss=0.04054, over 12245.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2703, pruned_loss=0.04216, over 3062331.21 frames. ], batch size: 246, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:55:31,073 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.276e+02 2.684e+02 3.281e+02 5.650e+02, threshold=5.369e+02, percent-clipped=3.0 2023-04-30 05:56:44,650 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151241.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:56:54,403 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151245.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:57:08,972 INFO [train.py:904] (7/8) Epoch 15, batch 9150, loss[loss=0.1582, simple_loss=0.2499, pruned_loss=0.03325, over 11879.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2701, pruned_loss=0.04167, over 3041660.79 frames. ], batch size: 249, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:57:18,603 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151256.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:58:12,774 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:58:39,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2344, 4.0447, 3.9688, 4.3604, 4.5254, 4.1087, 4.4634, 4.5310], device='cuda:7'), covar=tensor([0.1454, 0.1175, 0.2156, 0.0943, 0.0729, 0.1416, 0.0996, 0.0945], device='cuda:7'), in_proj_covar=tensor([0.0538, 0.0668, 0.0790, 0.0682, 0.0513, 0.0537, 0.0546, 0.0632], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 05:58:46,581 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151298.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 05:58:51,685 INFO [train.py:904] (7/8) Epoch 15, batch 9200, loss[loss=0.1657, simple_loss=0.2622, pruned_loss=0.03459, over 15501.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2657, pruned_loss=0.04069, over 3037327.13 frames. ], batch size: 191, lr: 4.48e-03, grad_scale: 8.0 2023-04-30 05:59:12,241 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.499e+02 3.414e+02 4.138e+02 7.678e+02, threshold=6.827e+02, percent-clipped=7.0 2023-04-30 06:00:09,405 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151343.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:00:24,891 INFO [train.py:904] (7/8) Epoch 15, batch 9250, loss[loss=0.157, simple_loss=0.2401, pruned_loss=0.03698, over 12069.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2655, pruned_loss=0.04073, over 3045081.32 frames. ], batch size: 247, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:00:41,064 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151359.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:01:07,617 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5484, 3.5217, 3.5195, 2.6095, 3.4452, 1.9576, 3.2340, 2.8321], device='cuda:7'), covar=tensor([0.0177, 0.0145, 0.0192, 0.0325, 0.0128, 0.2818, 0.0175, 0.0314], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0129, 0.0175, 0.0158, 0.0149, 0.0189, 0.0163, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:01:43,155 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151388.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:02:14,993 INFO [train.py:904] (7/8) Epoch 15, batch 9300, loss[loss=0.175, simple_loss=0.2526, pruned_loss=0.0487, over 12077.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2642, pruned_loss=0.04024, over 3036126.77 frames. ], batch size: 250, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:02:37,913 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.225e+02 2.604e+02 3.286e+02 5.862e+02, threshold=5.207e+02, percent-clipped=0.0 2023-04-30 06:03:16,637 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3257, 2.1072, 2.1911, 3.9620, 2.0964, 2.5477, 2.2612, 2.2884], device='cuda:7'), covar=tensor([0.1037, 0.3666, 0.2903, 0.0433, 0.4206, 0.2459, 0.3455, 0.3524], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0406, 0.0343, 0.0312, 0.0419, 0.0464, 0.0374, 0.0475], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:03:21,590 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 06:03:29,841 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151436.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:03:59,122 INFO [train.py:904] (7/8) Epoch 15, batch 9350, loss[loss=0.1726, simple_loss=0.2613, pruned_loss=0.04197, over 16622.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2632, pruned_loss=0.03984, over 3033480.64 frames. ], batch size: 57, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:04:04,567 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8560, 2.8208, 2.6859, 2.0173, 2.5345, 2.8100, 2.7413, 1.8823], device='cuda:7'), covar=tensor([0.0356, 0.0054, 0.0049, 0.0302, 0.0098, 0.0084, 0.0074, 0.0398], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0072, 0.0072, 0.0130, 0.0086, 0.0094, 0.0084, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:05:21,692 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 06:05:36,968 INFO [train.py:904] (7/8) Epoch 15, batch 9400, loss[loss=0.1781, simple_loss=0.2573, pruned_loss=0.04946, over 12618.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2626, pruned_loss=0.03944, over 3018750.55 frames. ], batch size: 248, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:05:59,181 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.126e+02 2.557e+02 3.057e+02 4.455e+02, threshold=5.114e+02, percent-clipped=0.0 2023-04-30 06:06:23,617 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-30 06:06:55,107 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151540.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:07:17,456 INFO [train.py:904] (7/8) Epoch 15, batch 9450, loss[loss=0.1775, simple_loss=0.2696, pruned_loss=0.04267, over 12666.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2647, pruned_loss=0.03967, over 3019819.64 frames. ], batch size: 250, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:07:24,807 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151556.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:08:58,274 INFO [train.py:904] (7/8) Epoch 15, batch 9500, loss[loss=0.1544, simple_loss=0.2527, pruned_loss=0.02809, over 16917.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2644, pruned_loss=0.03936, over 3041591.07 frames. ], batch size: 96, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:09:04,140 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151604.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:09:22,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.144e+02 2.700e+02 3.388e+02 6.768e+02, threshold=5.400e+02, percent-clipped=4.0 2023-04-30 06:10:12,037 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151638.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:10:44,454 INFO [train.py:904] (7/8) Epoch 15, batch 9550, loss[loss=0.1823, simple_loss=0.285, pruned_loss=0.03974, over 15389.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2649, pruned_loss=0.03953, over 3050879.79 frames. ], batch size: 191, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:10:49,461 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=151654.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:11:59,346 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151688.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:12:24,740 INFO [train.py:904] (7/8) Epoch 15, batch 9600, loss[loss=0.1743, simple_loss=0.2625, pruned_loss=0.04303, over 16626.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2666, pruned_loss=0.0407, over 3053364.31 frames. ], batch size: 57, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:12:44,299 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.311e+02 2.682e+02 3.396e+02 6.074e+02, threshold=5.365e+02, percent-clipped=2.0 2023-04-30 06:14:04,287 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151749.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:14:11,704 INFO [train.py:904] (7/8) Epoch 15, batch 9650, loss[loss=0.1804, simple_loss=0.2648, pruned_loss=0.04796, over 12004.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2682, pruned_loss=0.04094, over 3049816.62 frames. ], batch size: 248, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:14:41,226 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5854, 4.5811, 4.4460, 4.1301, 4.1697, 4.5573, 4.4086, 4.2187], device='cuda:7'), covar=tensor([0.0536, 0.0600, 0.0313, 0.0284, 0.0873, 0.0455, 0.0396, 0.0642], device='cuda:7'), in_proj_covar=tensor([0.0248, 0.0339, 0.0291, 0.0275, 0.0301, 0.0320, 0.0200, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-30 06:14:55,813 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9358, 2.7168, 2.8747, 1.9875, 2.6086, 2.0868, 2.6282, 2.7670], device='cuda:7'), covar=tensor([0.0323, 0.0844, 0.0463, 0.1983, 0.0805, 0.0988, 0.0664, 0.0812], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0145, 0.0157, 0.0144, 0.0136, 0.0123, 0.0136, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 06:15:07,635 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5492, 3.5359, 3.5544, 3.0341, 3.5144, 2.0864, 3.2827, 3.1351], device='cuda:7'), covar=tensor([0.0123, 0.0106, 0.0131, 0.0206, 0.0094, 0.2012, 0.0121, 0.0225], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0126, 0.0171, 0.0155, 0.0146, 0.0187, 0.0160, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:15:20,903 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3800, 3.0084, 2.6927, 2.2405, 2.1484, 2.2640, 2.9675, 2.8234], device='cuda:7'), covar=tensor([0.2537, 0.0773, 0.1543, 0.2634, 0.2417, 0.1957, 0.0485, 0.1310], device='cuda:7'), in_proj_covar=tensor([0.0302, 0.0252, 0.0283, 0.0282, 0.0267, 0.0228, 0.0266, 0.0297], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:15:57,994 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1117, 5.5152, 5.7205, 5.4713, 5.5355, 6.0548, 5.5765, 5.2881], device='cuda:7'), covar=tensor([0.0801, 0.1724, 0.1788, 0.1609, 0.2140, 0.0803, 0.1398, 0.2177], device='cuda:7'), in_proj_covar=tensor([0.0359, 0.0509, 0.0554, 0.0426, 0.0568, 0.0588, 0.0438, 0.0565], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:15:58,141 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=151801.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:15:58,774 INFO [train.py:904] (7/8) Epoch 15, batch 9700, loss[loss=0.1771, simple_loss=0.2652, pruned_loss=0.04451, over 16624.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2669, pruned_loss=0.0405, over 3056133.56 frames. ], batch size: 57, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:16:19,908 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.752e+02 2.225e+02 2.684e+02 3.461e+02 6.863e+02, threshold=5.368e+02, percent-clipped=3.0 2023-04-30 06:16:52,988 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4768, 1.7414, 2.0795, 2.4651, 2.4787, 2.7363, 1.8478, 2.6684], device='cuda:7'), covar=tensor([0.0206, 0.0472, 0.0311, 0.0266, 0.0276, 0.0191, 0.0460, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0175, 0.0160, 0.0164, 0.0173, 0.0131, 0.0176, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:7') 2023-04-30 06:17:06,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1317, 2.0844, 2.2315, 3.6143, 1.9967, 2.3453, 2.2031, 2.2369], device='cuda:7'), covar=tensor([0.1010, 0.3383, 0.2628, 0.0498, 0.4172, 0.2401, 0.3273, 0.3127], device='cuda:7'), in_proj_covar=tensor([0.0363, 0.0399, 0.0337, 0.0306, 0.0412, 0.0456, 0.0369, 0.0465], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:17:17,544 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3170, 3.4087, 3.6381, 3.6206, 3.6558, 3.4348, 3.5032, 3.5199], device='cuda:7'), covar=tensor([0.0396, 0.0928, 0.0617, 0.0648, 0.0628, 0.0719, 0.0856, 0.0478], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0369, 0.0361, 0.0347, 0.0412, 0.0386, 0.0469, 0.0307], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 06:17:18,827 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151840.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:17:21,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1458, 2.5375, 2.6579, 1.9226, 2.7608, 2.8982, 2.5188, 2.4667], device='cuda:7'), covar=tensor([0.0655, 0.0218, 0.0189, 0.0938, 0.0088, 0.0182, 0.0405, 0.0425], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0101, 0.0085, 0.0134, 0.0069, 0.0109, 0.0118, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 06:17:41,659 INFO [train.py:904] (7/8) Epoch 15, batch 9750, loss[loss=0.1698, simple_loss=0.2647, pruned_loss=0.03752, over 16523.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2657, pruned_loss=0.04061, over 3038180.85 frames. ], batch size: 68, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:17:56,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5053, 4.5100, 4.3298, 3.8281, 4.4164, 1.6501, 4.2001, 4.1537], device='cuda:7'), covar=tensor([0.0069, 0.0069, 0.0147, 0.0232, 0.0082, 0.2519, 0.0103, 0.0190], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0127, 0.0171, 0.0155, 0.0147, 0.0187, 0.0161, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:18:01,692 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=151862.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:18:10,608 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7689, 3.6154, 3.7331, 3.9332, 3.9713, 3.6224, 3.9863, 4.0202], device='cuda:7'), covar=tensor([0.1485, 0.1228, 0.1512, 0.0806, 0.0818, 0.1932, 0.0769, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0537, 0.0666, 0.0783, 0.0681, 0.0513, 0.0534, 0.0546, 0.0632], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:18:56,066 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151888.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:19:10,239 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0378, 3.3781, 3.6291, 2.0783, 2.9667, 2.2588, 3.6349, 3.4390], device='cuda:7'), covar=tensor([0.0231, 0.0685, 0.0459, 0.1847, 0.0719, 0.0920, 0.0543, 0.0824], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0144, 0.0157, 0.0145, 0.0136, 0.0123, 0.0136, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 06:19:18,831 INFO [train.py:904] (7/8) Epoch 15, batch 9800, loss[loss=0.1892, simple_loss=0.2833, pruned_loss=0.04758, over 16987.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2662, pruned_loss=0.0396, over 3055508.52 frames. ], batch size: 109, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:19:40,663 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.231e+02 2.655e+02 3.314e+02 7.232e+02, threshold=5.310e+02, percent-clipped=1.0 2023-04-30 06:20:31,064 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151938.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:21:03,089 INFO [train.py:904] (7/8) Epoch 15, batch 9850, loss[loss=0.2045, simple_loss=0.2935, pruned_loss=0.05778, over 16793.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2672, pruned_loss=0.03951, over 3049716.63 frames. ], batch size: 124, lr: 4.47e-03, grad_scale: 8.0 2023-04-30 06:21:08,519 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=151954.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:21:16,821 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-30 06:22:08,368 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3517, 4.6795, 4.5257, 4.5337, 4.2142, 4.1708, 4.1712, 4.7355], device='cuda:7'), covar=tensor([0.0990, 0.0839, 0.0845, 0.0623, 0.0757, 0.1325, 0.0955, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0572, 0.0708, 0.0576, 0.0507, 0.0444, 0.0459, 0.0584, 0.0538], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:22:17,655 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=151986.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:22:57,945 INFO [train.py:904] (7/8) Epoch 15, batch 9900, loss[loss=0.1685, simple_loss=0.251, pruned_loss=0.043, over 12511.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2674, pruned_loss=0.03922, over 3061965.03 frames. ], batch size: 246, lr: 4.46e-03, grad_scale: 8.0 2023-04-30 06:22:58,859 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152002.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:23:24,826 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.244e+02 2.678e+02 3.249e+02 7.284e+02, threshold=5.355e+02, percent-clipped=2.0 2023-04-30 06:24:37,932 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152044.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:24:55,590 INFO [train.py:904] (7/8) Epoch 15, batch 9950, loss[loss=0.1771, simple_loss=0.2755, pruned_loss=0.03935, over 15356.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2693, pruned_loss=0.03947, over 3066105.02 frames. ], batch size: 190, lr: 4.46e-03, grad_scale: 8.0 2023-04-30 06:25:59,898 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 06:26:56,978 INFO [train.py:904] (7/8) Epoch 15, batch 10000, loss[loss=0.1677, simple_loss=0.2534, pruned_loss=0.04103, over 12612.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2678, pruned_loss=0.03915, over 3080157.46 frames. ], batch size: 248, lr: 4.46e-03, grad_scale: 8.0 2023-04-30 06:27:18,768 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.235e+02 2.836e+02 3.494e+02 9.282e+02, threshold=5.672e+02, percent-clipped=5.0 2023-04-30 06:28:35,919 INFO [train.py:904] (7/8) Epoch 15, batch 10050, loss[loss=0.1929, simple_loss=0.2833, pruned_loss=0.05124, over 16907.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2681, pruned_loss=0.03904, over 3082041.99 frames. ], batch size: 116, lr: 4.46e-03, grad_scale: 4.0 2023-04-30 06:28:45,899 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152157.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:28:56,206 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4586, 3.9150, 4.1163, 2.8160, 3.6116, 4.0913, 3.8143, 2.4312], device='cuda:7'), covar=tensor([0.0469, 0.0042, 0.0030, 0.0346, 0.0098, 0.0062, 0.0057, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0071, 0.0071, 0.0128, 0.0086, 0.0093, 0.0082, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:30:08,501 INFO [train.py:904] (7/8) Epoch 15, batch 10100, loss[loss=0.1904, simple_loss=0.2792, pruned_loss=0.05078, over 16177.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2684, pruned_loss=0.03916, over 3089242.31 frames. ], batch size: 165, lr: 4.46e-03, grad_scale: 4.0 2023-04-30 06:30:28,185 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.142e+02 2.576e+02 3.219e+02 1.174e+03, threshold=5.152e+02, percent-clipped=3.0 2023-04-30 06:30:55,434 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152225.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:31:53,402 INFO [train.py:904] (7/8) Epoch 16, batch 0, loss[loss=0.2541, simple_loss=0.3023, pruned_loss=0.1029, over 16939.00 frames. ], tot_loss[loss=0.2541, simple_loss=0.3023, pruned_loss=0.1029, over 16939.00 frames. ], batch size: 109, lr: 4.32e-03, grad_scale: 8.0 2023-04-30 06:31:53,402 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 06:32:00,892 INFO [train.py:938] (7/8) Epoch 16, validation: loss=0.1492, simple_loss=0.2525, pruned_loss=0.02288, over 944034.00 frames. 2023-04-30 06:32:00,893 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 06:32:12,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2006, 5.2071, 4.9872, 4.6631, 4.9768, 2.1092, 4.8190, 4.9036], device='cuda:7'), covar=tensor([0.0063, 0.0061, 0.0169, 0.0256, 0.0094, 0.2120, 0.0116, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0127, 0.0171, 0.0153, 0.0147, 0.0187, 0.0161, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:32:29,787 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152273.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:32:29,898 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8478, 2.0832, 2.3557, 3.1426, 2.1181, 2.3073, 2.2021, 2.1763], device='cuda:7'), covar=tensor([0.1104, 0.2961, 0.2077, 0.0637, 0.3641, 0.2121, 0.2905, 0.2872], device='cuda:7'), in_proj_covar=tensor([0.0366, 0.0401, 0.0339, 0.0308, 0.0414, 0.0457, 0.0371, 0.0468], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:32:49,356 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:33:09,968 INFO [train.py:904] (7/8) Epoch 16, batch 50, loss[loss=0.1812, simple_loss=0.2727, pruned_loss=0.04482, over 17025.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2803, pruned_loss=0.05845, over 751188.54 frames. ], batch size: 50, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:33:29,877 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.475e+02 3.034e+02 3.841e+02 6.460e+02, threshold=6.067e+02, percent-clipped=5.0 2023-04-30 06:33:53,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0910, 4.0029, 4.4814, 2.4272, 4.6634, 4.7411, 3.4079, 3.7136], device='cuda:7'), covar=tensor([0.0619, 0.0223, 0.0181, 0.1042, 0.0057, 0.0118, 0.0367, 0.0318], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0101, 0.0086, 0.0135, 0.0070, 0.0110, 0.0120, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 06:33:53,596 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152334.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:34:08,321 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:34:18,731 INFO [train.py:904] (7/8) Epoch 16, batch 100, loss[loss=0.1909, simple_loss=0.2692, pruned_loss=0.05634, over 16341.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2725, pruned_loss=0.05377, over 1324985.48 frames. ], batch size: 165, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:34:56,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8078, 3.0958, 2.8331, 5.0131, 4.1999, 4.5851, 1.5830, 3.3428], device='cuda:7'), covar=tensor([0.1326, 0.0665, 0.1081, 0.0189, 0.0212, 0.0356, 0.1569, 0.0656], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0162, 0.0185, 0.0161, 0.0189, 0.0205, 0.0188, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-30 06:35:10,713 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8331, 4.9599, 5.3663, 5.3423, 5.4217, 5.0683, 4.8749, 4.7923], device='cuda:7'), covar=tensor([0.0556, 0.0681, 0.0587, 0.0651, 0.0599, 0.0660, 0.1190, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0345, 0.0373, 0.0364, 0.0350, 0.0413, 0.0387, 0.0470, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 06:35:14,879 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152392.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:35:26,683 INFO [train.py:904] (7/8) Epoch 16, batch 150, loss[loss=0.1543, simple_loss=0.2414, pruned_loss=0.03364, over 17012.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.272, pruned_loss=0.05234, over 1758002.27 frames. ], batch size: 41, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:35:28,926 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6442, 3.8362, 4.1314, 2.8681, 3.6386, 4.0913, 3.7999, 2.3499], device='cuda:7'), covar=tensor([0.0422, 0.0236, 0.0039, 0.0330, 0.0096, 0.0073, 0.0073, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0074, 0.0072, 0.0131, 0.0088, 0.0095, 0.0084, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:35:48,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.454e+02 2.817e+02 3.397e+02 1.160e+03, threshold=5.634e+02, percent-clipped=3.0 2023-04-30 06:36:25,014 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152445.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:36:35,190 INFO [train.py:904] (7/8) Epoch 16, batch 200, loss[loss=0.1851, simple_loss=0.2541, pruned_loss=0.05801, over 16844.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2712, pruned_loss=0.05157, over 2105137.91 frames. ], batch size: 102, lr: 4.32e-03, grad_scale: 1.0 2023-04-30 06:36:42,911 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152457.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:37:44,243 INFO [train.py:904] (7/8) Epoch 16, batch 250, loss[loss=0.1748, simple_loss=0.2655, pruned_loss=0.04206, over 17284.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2685, pruned_loss=0.05084, over 2373028.38 frames. ], batch size: 52, lr: 4.31e-03, grad_scale: 1.0 2023-04-30 06:37:48,059 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152505.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:37:50,063 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152506.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:38:05,712 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.341e+02 2.836e+02 3.695e+02 5.924e+02, threshold=5.672e+02, percent-clipped=1.0 2023-04-30 06:38:19,905 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 06:38:24,752 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7652, 2.8927, 2.5655, 4.8661, 4.0182, 4.4338, 1.5304, 3.0603], device='cuda:7'), covar=tensor([0.1334, 0.0728, 0.1293, 0.0188, 0.0219, 0.0378, 0.1663, 0.0772], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0163, 0.0186, 0.0163, 0.0191, 0.0207, 0.0189, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 06:38:53,027 INFO [train.py:904] (7/8) Epoch 16, batch 300, loss[loss=0.1648, simple_loss=0.2621, pruned_loss=0.03378, over 17145.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2661, pruned_loss=0.04932, over 2570910.48 frames. ], batch size: 49, lr: 4.31e-03, grad_scale: 1.0 2023-04-30 06:39:23,836 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9883, 3.0818, 3.2251, 2.0829, 3.0864, 3.3246, 3.1907, 1.8185], device='cuda:7'), covar=tensor([0.0536, 0.0163, 0.0065, 0.0426, 0.0107, 0.0109, 0.0102, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0075, 0.0073, 0.0131, 0.0088, 0.0096, 0.0085, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:39:29,815 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3874, 5.2530, 5.1488, 4.6968, 4.7958, 5.2231, 5.1731, 4.7891], device='cuda:7'), covar=tensor([0.0509, 0.0402, 0.0286, 0.0292, 0.0997, 0.0344, 0.0271, 0.0714], device='cuda:7'), in_proj_covar=tensor([0.0263, 0.0360, 0.0309, 0.0293, 0.0323, 0.0339, 0.0211, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:39:33,872 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152581.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:39:35,055 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152582.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:39:42,997 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152589.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:39:57,787 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1596, 3.2574, 1.9549, 3.4153, 2.5891, 3.4391, 2.0201, 2.6073], device='cuda:7'), covar=tensor([0.0301, 0.0400, 0.1588, 0.0304, 0.0757, 0.0661, 0.1456, 0.0778], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0169, 0.0191, 0.0145, 0.0172, 0.0209, 0.0201, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 06:40:01,545 INFO [train.py:904] (7/8) Epoch 16, batch 350, loss[loss=0.1693, simple_loss=0.2525, pruned_loss=0.04303, over 16064.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2632, pruned_loss=0.04851, over 2735964.66 frames. ], batch size: 35, lr: 4.31e-03, grad_scale: 1.0 2023-04-30 06:40:20,733 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.239e+02 2.608e+02 3.067e+02 7.666e+02, threshold=5.216e+02, percent-clipped=3.0 2023-04-30 06:40:38,492 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152629.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:40:58,669 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:41:06,996 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152650.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:41:08,837 INFO [train.py:904] (7/8) Epoch 16, batch 400, loss[loss=0.1932, simple_loss=0.277, pruned_loss=0.05472, over 17055.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2614, pruned_loss=0.04832, over 2867560.65 frames. ], batch size: 55, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:41:24,354 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-04-30 06:41:36,027 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2023-04-30 06:42:16,096 INFO [train.py:904] (7/8) Epoch 16, batch 450, loss[loss=0.1732, simple_loss=0.2518, pruned_loss=0.04736, over 15617.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2592, pruned_loss=0.04707, over 2967933.40 frames. ], batch size: 190, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:42:36,235 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.224e+02 2.572e+02 2.956e+02 7.682e+02, threshold=5.144e+02, percent-clipped=1.0 2023-04-30 06:42:50,653 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=152726.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:43:25,841 INFO [train.py:904] (7/8) Epoch 16, batch 500, loss[loss=0.1764, simple_loss=0.2679, pruned_loss=0.04245, over 16665.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2579, pruned_loss=0.04627, over 3040337.64 frames. ], batch size: 57, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:43:50,543 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.41 vs. limit=5.0 2023-04-30 06:44:14,757 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=152787.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:44:33,481 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152801.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:44:34,233 INFO [train.py:904] (7/8) Epoch 16, batch 550, loss[loss=0.1443, simple_loss=0.2232, pruned_loss=0.03265, over 16811.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.257, pruned_loss=0.0452, over 3113114.87 frames. ], batch size: 39, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:44:55,530 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 2.218e+02 2.743e+02 3.343e+02 5.376e+02, threshold=5.487e+02, percent-clipped=2.0 2023-04-30 06:45:11,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5241, 3.5940, 2.2653, 3.7686, 2.8047, 3.7522, 2.2417, 2.8731], device='cuda:7'), covar=tensor([0.0234, 0.0393, 0.1350, 0.0306, 0.0701, 0.0688, 0.1268, 0.0666], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0170, 0.0191, 0.0147, 0.0172, 0.0210, 0.0200, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 06:45:17,559 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5700, 3.5965, 3.9426, 2.9211, 3.6001, 3.9883, 3.7448, 2.2594], device='cuda:7'), covar=tensor([0.0424, 0.0220, 0.0047, 0.0295, 0.0087, 0.0089, 0.0083, 0.0436], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0075, 0.0073, 0.0131, 0.0088, 0.0097, 0.0085, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:45:46,475 INFO [train.py:904] (7/8) Epoch 16, batch 600, loss[loss=0.1511, simple_loss=0.2358, pruned_loss=0.03323, over 17230.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2557, pruned_loss=0.04542, over 3155381.62 frames. ], batch size: 45, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:46:25,193 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152881.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:46:53,574 INFO [train.py:904] (7/8) Epoch 16, batch 650, loss[loss=0.1559, simple_loss=0.2445, pruned_loss=0.03368, over 17239.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2552, pruned_loss=0.04519, over 3193117.48 frames. ], batch size: 45, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:47:14,338 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.133e+02 2.528e+02 3.105e+02 5.746e+02, threshold=5.056e+02, percent-clipped=1.0 2023-04-30 06:47:30,643 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:47:30,818 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=152929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:47:43,499 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152938.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:47:52,172 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=152945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:48:00,902 INFO [train.py:904] (7/8) Epoch 16, batch 700, loss[loss=0.1792, simple_loss=0.2622, pruned_loss=0.04806, over 16573.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2548, pruned_loss=0.04487, over 3211662.22 frames. ], batch size: 68, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:48:37,712 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=152977.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:49:10,743 INFO [train.py:904] (7/8) Epoch 16, batch 750, loss[loss=0.1702, simple_loss=0.2693, pruned_loss=0.03553, over 17290.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2565, pruned_loss=0.04586, over 3231081.14 frames. ], batch size: 52, lr: 4.31e-03, grad_scale: 2.0 2023-04-30 06:49:31,101 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.303e+02 2.598e+02 3.090e+02 5.870e+02, threshold=5.196e+02, percent-clipped=1.0 2023-04-30 06:50:06,265 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 06:50:17,795 INFO [train.py:904] (7/8) Epoch 16, batch 800, loss[loss=0.1443, simple_loss=0.2281, pruned_loss=0.03019, over 16977.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2558, pruned_loss=0.04575, over 3257178.30 frames. ], batch size: 41, lr: 4.31e-03, grad_scale: 4.0 2023-04-30 06:51:00,662 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153082.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:51:25,738 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153101.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:51:26,417 INFO [train.py:904] (7/8) Epoch 16, batch 850, loss[loss=0.172, simple_loss=0.2529, pruned_loss=0.04553, over 16577.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2551, pruned_loss=0.0449, over 3281534.62 frames. ], batch size: 68, lr: 4.31e-03, grad_scale: 4.0 2023-04-30 06:51:46,529 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.045e+02 2.529e+02 2.957e+02 4.458e+02, threshold=5.058e+02, percent-clipped=0.0 2023-04-30 06:51:46,938 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153117.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:52:03,091 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0870, 4.9975, 4.8742, 4.4690, 4.5383, 4.9543, 4.8682, 4.6163], device='cuda:7'), covar=tensor([0.0539, 0.0530, 0.0315, 0.0322, 0.1074, 0.0455, 0.0357, 0.0733], device='cuda:7'), in_proj_covar=tensor([0.0277, 0.0380, 0.0325, 0.0311, 0.0341, 0.0359, 0.0222, 0.0389], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:52:32,207 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153149.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:52:35,403 INFO [train.py:904] (7/8) Epoch 16, batch 900, loss[loss=0.1711, simple_loss=0.2431, pruned_loss=0.04955, over 16860.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2542, pruned_loss=0.04401, over 3299023.62 frames. ], batch size: 116, lr: 4.31e-03, grad_scale: 4.0 2023-04-30 06:52:52,780 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 06:52:56,763 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 06:53:11,078 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153178.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:53:36,543 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4274, 4.3175, 4.2956, 4.0616, 4.0721, 4.3985, 4.1209, 4.1501], device='cuda:7'), covar=tensor([0.0661, 0.0809, 0.0329, 0.0300, 0.0834, 0.0494, 0.0667, 0.0710], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0383, 0.0327, 0.0314, 0.0344, 0.0362, 0.0223, 0.0392], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:53:43,189 INFO [train.py:904] (7/8) Epoch 16, batch 950, loss[loss=0.1557, simple_loss=0.2366, pruned_loss=0.03745, over 16776.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2549, pruned_loss=0.04427, over 3305727.10 frames. ], batch size: 39, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:54:04,606 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.182e+02 2.578e+02 3.357e+02 5.954e+02, threshold=5.156e+02, percent-clipped=2.0 2023-04-30 06:54:33,249 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153238.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:54:34,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2418, 5.2945, 5.7024, 5.6923, 5.7053, 5.3764, 5.2838, 5.1026], device='cuda:7'), covar=tensor([0.0293, 0.0573, 0.0345, 0.0373, 0.0447, 0.0332, 0.0872, 0.0380], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0406, 0.0394, 0.0378, 0.0446, 0.0419, 0.0511, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 06:54:42,474 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153245.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:54:53,050 INFO [train.py:904] (7/8) Epoch 16, batch 1000, loss[loss=0.1724, simple_loss=0.243, pruned_loss=0.05087, over 16690.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2541, pruned_loss=0.04355, over 3310962.83 frames. ], batch size: 134, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:55:29,881 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153278.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:55:40,267 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:55:51,055 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153293.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:56:02,859 INFO [train.py:904] (7/8) Epoch 16, batch 1050, loss[loss=0.1776, simple_loss=0.2525, pruned_loss=0.05135, over 16599.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2541, pruned_loss=0.04379, over 3313159.46 frames. ], batch size: 68, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:56:24,079 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6394, 1.8634, 2.2822, 2.4560, 2.5760, 2.3915, 1.8529, 2.7273], device='cuda:7'), covar=tensor([0.0140, 0.0381, 0.0236, 0.0254, 0.0238, 0.0284, 0.0425, 0.0150], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0184, 0.0170, 0.0173, 0.0182, 0.0140, 0.0185, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 06:56:24,689 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.258e+02 2.836e+02 3.297e+02 6.244e+02, threshold=5.672e+02, percent-clipped=2.0 2023-04-30 06:56:55,517 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153339.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:56:55,601 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0404, 4.5412, 3.3038, 2.4465, 2.8599, 2.6209, 4.8538, 3.8381], device='cuda:7'), covar=tensor([0.2495, 0.0504, 0.1543, 0.2531, 0.2676, 0.1925, 0.0314, 0.1239], device='cuda:7'), in_proj_covar=tensor([0.0313, 0.0263, 0.0294, 0.0291, 0.0282, 0.0237, 0.0278, 0.0315], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 06:57:12,832 INFO [train.py:904] (7/8) Epoch 16, batch 1100, loss[loss=0.1658, simple_loss=0.2655, pruned_loss=0.03305, over 17025.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.254, pruned_loss=0.04344, over 3311347.51 frames. ], batch size: 55, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:57:54,289 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153382.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:58:21,567 INFO [train.py:904] (7/8) Epoch 16, batch 1150, loss[loss=0.1786, simple_loss=0.2531, pruned_loss=0.05201, over 16533.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2538, pruned_loss=0.04305, over 3312960.29 frames. ], batch size: 75, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 06:58:42,015 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.113e+02 2.608e+02 3.086e+02 6.684e+02, threshold=5.217e+02, percent-clipped=1.0 2023-04-30 06:58:46,831 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153421.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:58:58,600 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153430.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 06:59:27,894 INFO [train.py:904] (7/8) Epoch 16, batch 1200, loss[loss=0.1734, simple_loss=0.2614, pruned_loss=0.04272, over 17035.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2523, pruned_loss=0.04227, over 3311666.65 frames. ], batch size: 53, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 06:59:57,507 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153473.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:00:10,791 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153482.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:00:34,438 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3329, 4.1229, 4.2801, 4.5068, 4.5619, 4.2214, 4.4481, 4.6138], device='cuda:7'), covar=tensor([0.1581, 0.1307, 0.1810, 0.0890, 0.0773, 0.1152, 0.1995, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0611, 0.0759, 0.0899, 0.0770, 0.0578, 0.0603, 0.0618, 0.0712], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:00:37,164 INFO [train.py:904] (7/8) Epoch 16, batch 1250, loss[loss=0.1884, simple_loss=0.2682, pruned_loss=0.05428, over 16540.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2523, pruned_loss=0.04285, over 3314254.58 frames. ], batch size: 62, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:00:57,397 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 2.233e+02 2.659e+02 3.231e+02 7.143e+02, threshold=5.319e+02, percent-clipped=2.0 2023-04-30 07:01:43,343 INFO [train.py:904] (7/8) Epoch 16, batch 1300, loss[loss=0.1549, simple_loss=0.2495, pruned_loss=0.0301, over 17128.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2523, pruned_loss=0.04292, over 3314345.86 frames. ], batch size: 47, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:01:55,058 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8136, 3.8233, 2.9339, 2.2693, 2.5782, 2.3418, 3.8847, 3.3672], device='cuda:7'), covar=tensor([0.2338, 0.0570, 0.1517, 0.2544, 0.2311, 0.1834, 0.0494, 0.1388], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0264, 0.0293, 0.0291, 0.0282, 0.0237, 0.0278, 0.0316], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:02:52,655 INFO [train.py:904] (7/8) Epoch 16, batch 1350, loss[loss=0.1593, simple_loss=0.2441, pruned_loss=0.03718, over 17208.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2518, pruned_loss=0.04271, over 3320327.30 frames. ], batch size: 46, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:03:12,846 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.319e+02 2.697e+02 3.051e+02 7.500e+02, threshold=5.394e+02, percent-clipped=1.0 2023-04-30 07:03:37,104 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153634.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:04:02,091 INFO [train.py:904] (7/8) Epoch 16, batch 1400, loss[loss=0.1714, simple_loss=0.2452, pruned_loss=0.04883, over 16849.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2516, pruned_loss=0.04293, over 3318502.51 frames. ], batch size: 109, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:05:12,116 INFO [train.py:904] (7/8) Epoch 16, batch 1450, loss[loss=0.163, simple_loss=0.2582, pruned_loss=0.03391, over 16717.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2518, pruned_loss=0.04283, over 3329067.15 frames. ], batch size: 57, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 07:05:34,104 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.274e+02 2.601e+02 3.378e+02 7.170e+02, threshold=5.202e+02, percent-clipped=2.0 2023-04-30 07:06:11,392 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7709, 3.9331, 2.4754, 4.5560, 2.9805, 4.4652, 2.5329, 3.1630], device='cuda:7'), covar=tensor([0.0247, 0.0312, 0.1456, 0.0185, 0.0748, 0.0458, 0.1349, 0.0649], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0172, 0.0193, 0.0152, 0.0174, 0.0214, 0.0202, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:06:22,467 INFO [train.py:904] (7/8) Epoch 16, batch 1500, loss[loss=0.1599, simple_loss=0.2481, pruned_loss=0.03584, over 17218.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2522, pruned_loss=0.04325, over 3319627.35 frames. ], batch size: 44, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 07:06:50,970 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153773.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:06:56,033 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=153777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:06:59,012 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2664, 3.4668, 3.6310, 3.6078, 3.6172, 3.4444, 3.4561, 3.4893], device='cuda:7'), covar=tensor([0.0422, 0.0746, 0.0545, 0.0547, 0.0589, 0.0547, 0.0781, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0384, 0.0415, 0.0403, 0.0385, 0.0455, 0.0430, 0.0524, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 07:07:01,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9656, 4.3407, 3.2636, 2.2945, 2.7959, 2.5804, 4.6514, 3.7597], device='cuda:7'), covar=tensor([0.2541, 0.0589, 0.1642, 0.2817, 0.2793, 0.1870, 0.0395, 0.1174], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0264, 0.0293, 0.0292, 0.0283, 0.0237, 0.0279, 0.0316], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:07:10,668 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6949, 2.5982, 1.8887, 2.7888, 2.1624, 2.7818, 2.1645, 2.3948], device='cuda:7'), covar=tensor([0.0235, 0.0330, 0.1278, 0.0223, 0.0599, 0.0386, 0.1067, 0.0565], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0172, 0.0192, 0.0152, 0.0173, 0.0214, 0.0201, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:07:25,530 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 07:07:30,673 INFO [train.py:904] (7/8) Epoch 16, batch 1550, loss[loss=0.196, simple_loss=0.26, pruned_loss=0.06602, over 16798.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2539, pruned_loss=0.04516, over 3312515.84 frames. ], batch size: 124, lr: 4.30e-03, grad_scale: 4.0 2023-04-30 07:07:53,746 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.189e+02 2.739e+02 3.070e+02 6.481e+02, threshold=5.478e+02, percent-clipped=4.0 2023-04-30 07:07:58,990 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153821.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:08:40,386 INFO [train.py:904] (7/8) Epoch 16, batch 1600, loss[loss=0.1601, simple_loss=0.2475, pruned_loss=0.03641, over 16843.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2564, pruned_loss=0.04565, over 3314250.89 frames. ], batch size: 42, lr: 4.30e-03, grad_scale: 8.0 2023-04-30 07:09:47,773 INFO [train.py:904] (7/8) Epoch 16, batch 1650, loss[loss=0.2057, simple_loss=0.2776, pruned_loss=0.06694, over 16432.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2577, pruned_loss=0.04633, over 3313796.29 frames. ], batch size: 146, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:09:48,518 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.60 vs. limit=5.0 2023-04-30 07:10:08,860 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 07:10:09,060 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.549e+02 2.974e+02 3.703e+02 7.236e+02, threshold=5.949e+02, percent-clipped=5.0 2023-04-30 07:10:16,061 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=153923.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:10:32,636 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=153934.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:10:56,045 INFO [train.py:904] (7/8) Epoch 16, batch 1700, loss[loss=0.1784, simple_loss=0.262, pruned_loss=0.0474, over 16474.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2588, pruned_loss=0.04615, over 3320027.71 frames. ], batch size: 146, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:11:17,981 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 07:11:38,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=153982.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:11:40,892 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=153984.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:12:09,305 INFO [train.py:904] (7/8) Epoch 16, batch 1750, loss[loss=0.1782, simple_loss=0.2534, pruned_loss=0.05146, over 16860.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2601, pruned_loss=0.04625, over 3316854.41 frames. ], batch size: 109, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:12:21,051 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5088, 4.6097, 4.7570, 4.5374, 4.4821, 5.2071, 4.6803, 4.3177], device='cuda:7'), covar=tensor([0.1759, 0.2104, 0.2267, 0.2300, 0.3517, 0.1268, 0.1673, 0.2751], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0568, 0.0617, 0.0474, 0.0637, 0.0649, 0.0483, 0.0629], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:12:33,141 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.261e+02 2.728e+02 3.333e+02 9.430e+02, threshold=5.456e+02, percent-clipped=2.0 2023-04-30 07:13:15,113 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154049.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:13:18,927 INFO [train.py:904] (7/8) Epoch 16, batch 1800, loss[loss=0.1731, simple_loss=0.2673, pruned_loss=0.03949, over 17051.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2622, pruned_loss=0.04686, over 3304955.75 frames. ], batch size: 55, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:13:53,304 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154076.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:13:54,287 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154077.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:14:03,085 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154084.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:14:06,365 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7073, 4.4457, 4.7225, 4.8828, 5.0290, 4.4704, 4.9465, 5.0289], device='cuda:7'), covar=tensor([0.1510, 0.1196, 0.1403, 0.0646, 0.0519, 0.1031, 0.1094, 0.0589], device='cuda:7'), in_proj_covar=tensor([0.0613, 0.0761, 0.0902, 0.0770, 0.0577, 0.0603, 0.0615, 0.0714], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:14:28,055 INFO [train.py:904] (7/8) Epoch 16, batch 1850, loss[loss=0.1733, simple_loss=0.2581, pruned_loss=0.0442, over 16846.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2626, pruned_loss=0.04665, over 3311817.97 frames. ], batch size: 96, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:14:39,378 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154110.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:14:39,724 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-30 07:14:50,220 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.254e+02 2.561e+02 3.083e+02 7.849e+02, threshold=5.121e+02, percent-clipped=3.0 2023-04-30 07:14:59,482 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:15:16,110 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154137.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 07:15:21,281 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9115, 3.9999, 2.5612, 4.6434, 3.1928, 4.5907, 2.5681, 3.1597], device='cuda:7'), covar=tensor([0.0261, 0.0329, 0.1490, 0.0257, 0.0747, 0.0471, 0.1523, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0174, 0.0195, 0.0154, 0.0174, 0.0216, 0.0203, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:15:25,849 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154145.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:15:33,370 INFO [train.py:904] (7/8) Epoch 16, batch 1900, loss[loss=0.1913, simple_loss=0.265, pruned_loss=0.05882, over 16797.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2617, pruned_loss=0.04585, over 3306057.98 frames. ], batch size: 102, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:16:08,937 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.51 vs. limit=5.0 2023-04-30 07:16:44,652 INFO [train.py:904] (7/8) Epoch 16, batch 1950, loss[loss=0.1768, simple_loss=0.2762, pruned_loss=0.03868, over 16661.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2616, pruned_loss=0.04548, over 3301428.14 frames. ], batch size: 62, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:16:49,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1034, 5.0961, 5.6120, 5.6018, 5.6383, 5.2224, 5.1464, 4.9306], device='cuda:7'), covar=tensor([0.0313, 0.0476, 0.0307, 0.0409, 0.0391, 0.0330, 0.0868, 0.0375], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0412, 0.0400, 0.0382, 0.0451, 0.0429, 0.0523, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 07:16:50,772 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3810, 3.6024, 3.8195, 2.1315, 3.1253, 2.4855, 3.8049, 3.8562], device='cuda:7'), covar=tensor([0.0247, 0.0842, 0.0456, 0.1873, 0.0785, 0.0942, 0.0579, 0.1009], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0157, 0.0163, 0.0150, 0.0141, 0.0127, 0.0142, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:17:04,525 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.235e+02 2.551e+02 3.026e+02 6.742e+02, threshold=5.103e+02, percent-clipped=2.0 2023-04-30 07:17:10,571 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:17:51,474 INFO [train.py:904] (7/8) Epoch 16, batch 2000, loss[loss=0.1636, simple_loss=0.2428, pruned_loss=0.04218, over 16861.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2611, pruned_loss=0.0451, over 3302686.12 frames. ], batch size: 102, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:18:27,774 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154279.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:18:33,613 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154282.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:18:38,801 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154286.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:18:44,585 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154291.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:18:59,382 INFO [train.py:904] (7/8) Epoch 16, batch 2050, loss[loss=0.1759, simple_loss=0.2649, pruned_loss=0.04347, over 17023.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2614, pruned_loss=0.04585, over 3299503.29 frames. ], batch size: 50, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:19:19,611 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.241e+02 2.674e+02 3.116e+02 4.900e+02, threshold=5.347e+02, percent-clipped=0.0 2023-04-30 07:19:46,308 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154337.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:19:51,385 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 07:20:01,074 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154347.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:20:04,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9512, 2.4564, 2.0073, 2.2456, 2.8448, 2.6339, 3.0124, 2.9702], device='cuda:7'), covar=tensor([0.0163, 0.0339, 0.0445, 0.0418, 0.0208, 0.0294, 0.0191, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0225, 0.0218, 0.0218, 0.0228, 0.0229, 0.0232, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:20:06,433 INFO [train.py:904] (7/8) Epoch 16, batch 2100, loss[loss=0.1594, simple_loss=0.2433, pruned_loss=0.03773, over 16736.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2616, pruned_loss=0.04603, over 3311952.42 frames. ], batch size: 83, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:20:07,001 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154352.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:20:57,963 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6251, 2.2127, 2.2482, 4.4916, 2.2163, 2.6465, 2.2847, 2.4406], device='cuda:7'), covar=tensor([0.1022, 0.3541, 0.2848, 0.0397, 0.3980, 0.2578, 0.3378, 0.3523], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0420, 0.0352, 0.0323, 0.0424, 0.0483, 0.0389, 0.0493], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:21:09,612 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154398.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:21:14,703 INFO [train.py:904] (7/8) Epoch 16, batch 2150, loss[loss=0.1814, simple_loss=0.2638, pruned_loss=0.04953, over 16819.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2623, pruned_loss=0.04662, over 3316546.27 frames. ], batch size: 102, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:21:19,325 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154405.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 07:21:36,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 2.339e+02 2.706e+02 3.339e+02 5.041e+02, threshold=5.411e+02, percent-clipped=0.0 2023-04-30 07:21:57,914 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154432.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:22:07,249 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154440.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:22:25,091 INFO [train.py:904] (7/8) Epoch 16, batch 2200, loss[loss=0.1828, simple_loss=0.2585, pruned_loss=0.05353, over 16772.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2627, pruned_loss=0.04651, over 3324903.45 frames. ], batch size: 83, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:22:55,770 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154474.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:23:01,851 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6584, 2.3603, 1.8696, 2.1527, 2.7855, 2.5740, 2.8677, 2.8997], device='cuda:7'), covar=tensor([0.0162, 0.0329, 0.0458, 0.0390, 0.0188, 0.0268, 0.0205, 0.0218], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0225, 0.0218, 0.0218, 0.0227, 0.0228, 0.0231, 0.0222], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:23:16,621 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4679, 4.4131, 4.4116, 3.9111, 4.4222, 1.9098, 4.1490, 4.0970], device='cuda:7'), covar=tensor([0.0127, 0.0091, 0.0159, 0.0301, 0.0087, 0.2409, 0.0145, 0.0180], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0141, 0.0188, 0.0173, 0.0161, 0.0200, 0.0177, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:23:32,245 INFO [train.py:904] (7/8) Epoch 16, batch 2250, loss[loss=0.1915, simple_loss=0.2665, pruned_loss=0.05826, over 16772.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2625, pruned_loss=0.04637, over 3334894.42 frames. ], batch size: 124, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:23:45,874 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9123, 4.3903, 4.3746, 3.1464, 3.6481, 4.3471, 3.9628, 2.6811], device='cuda:7'), covar=tensor([0.0433, 0.0055, 0.0033, 0.0327, 0.0108, 0.0078, 0.0085, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0077, 0.0076, 0.0133, 0.0091, 0.0100, 0.0089, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:23:55,136 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.236e+02 2.597e+02 3.012e+02 8.725e+02, threshold=5.194e+02, percent-clipped=2.0 2023-04-30 07:24:18,870 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154535.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:24:20,659 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8501, 4.4055, 4.3936, 3.1414, 3.6946, 4.3738, 4.0064, 2.5809], device='cuda:7'), covar=tensor([0.0425, 0.0054, 0.0034, 0.0307, 0.0100, 0.0073, 0.0077, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0077, 0.0076, 0.0133, 0.0090, 0.0099, 0.0089, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:24:40,028 INFO [train.py:904] (7/8) Epoch 16, batch 2300, loss[loss=0.151, simple_loss=0.2417, pruned_loss=0.03013, over 16871.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2626, pruned_loss=0.04597, over 3332971.41 frames. ], batch size: 42, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:25:07,253 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1534, 5.7341, 5.9106, 5.5122, 5.7436, 6.2330, 5.7883, 5.4429], device='cuda:7'), covar=tensor([0.0955, 0.1836, 0.2257, 0.2152, 0.2468, 0.1044, 0.1448, 0.2234], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0566, 0.0614, 0.0476, 0.0632, 0.0646, 0.0483, 0.0627], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:25:16,762 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154577.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 07:25:20,034 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154579.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:25:25,812 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4854, 5.8975, 5.6264, 5.7437, 5.3079, 5.2867, 5.2270, 6.0176], device='cuda:7'), covar=tensor([0.1407, 0.0935, 0.1136, 0.0746, 0.1064, 0.0756, 0.1206, 0.0940], device='cuda:7'), in_proj_covar=tensor([0.0637, 0.0786, 0.0641, 0.0566, 0.0495, 0.0503, 0.0655, 0.0603], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:25:33,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7130, 3.7854, 2.3679, 4.3837, 2.9165, 4.3641, 2.6480, 3.0776], device='cuda:7'), covar=tensor([0.0282, 0.0378, 0.1620, 0.0315, 0.0850, 0.0498, 0.1291, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0176, 0.0195, 0.0155, 0.0175, 0.0218, 0.0204, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:25:49,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1449, 3.4052, 3.3985, 2.3084, 3.0777, 2.5384, 3.5592, 3.6947], device='cuda:7'), covar=tensor([0.0307, 0.0835, 0.0586, 0.1631, 0.0749, 0.0882, 0.0631, 0.0824], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0157, 0.0163, 0.0149, 0.0141, 0.0127, 0.0142, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:25:51,261 INFO [train.py:904] (7/8) Epoch 16, batch 2350, loss[loss=0.177, simple_loss=0.2706, pruned_loss=0.0417, over 16706.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2631, pruned_loss=0.04655, over 3328775.75 frames. ], batch size: 57, lr: 4.29e-03, grad_scale: 8.0 2023-04-30 07:26:11,868 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.266e+02 2.563e+02 3.089e+02 4.962e+02, threshold=5.126e+02, percent-clipped=0.0 2023-04-30 07:26:14,928 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3879, 3.6598, 4.0102, 2.2202, 3.2567, 2.4816, 3.9359, 3.9004], device='cuda:7'), covar=tensor([0.0286, 0.0882, 0.0461, 0.1830, 0.0760, 0.0940, 0.0550, 0.0908], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0157, 0.0163, 0.0149, 0.0141, 0.0127, 0.0142, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:26:25,330 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154627.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:26,738 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154628.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:45,841 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154642.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:52,336 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154647.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:26:57,950 INFO [train.py:904] (7/8) Epoch 16, batch 2400, loss[loss=0.154, simple_loss=0.2463, pruned_loss=0.03085, over 17228.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2637, pruned_loss=0.04599, over 3340718.05 frames. ], batch size: 45, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:27:46,413 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 07:27:51,080 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154689.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:27:53,237 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1647, 5.7586, 5.9175, 5.5099, 5.6253, 6.2030, 5.7297, 5.4123], device='cuda:7'), covar=tensor([0.0880, 0.1831, 0.1936, 0.2119, 0.2432, 0.0929, 0.1291, 0.2196], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0563, 0.0611, 0.0473, 0.0631, 0.0644, 0.0480, 0.0626], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:27:55,529 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154693.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:28:06,464 INFO [train.py:904] (7/8) Epoch 16, batch 2450, loss[loss=0.1588, simple_loss=0.2517, pruned_loss=0.03296, over 17229.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2648, pruned_loss=0.04612, over 3335107.79 frames. ], batch size: 45, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:28:12,181 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154705.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:28:16,108 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5342, 2.5882, 2.2361, 2.3069, 2.9043, 2.5857, 3.1931, 3.1260], device='cuda:7'), covar=tensor([0.0119, 0.0365, 0.0431, 0.0387, 0.0229, 0.0340, 0.0235, 0.0236], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0226, 0.0219, 0.0220, 0.0229, 0.0230, 0.0234, 0.0224], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:28:20,229 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4016, 3.2205, 3.6425, 1.8954, 3.7208, 3.7391, 3.0639, 2.7596], device='cuda:7'), covar=tensor([0.0746, 0.0226, 0.0134, 0.1082, 0.0077, 0.0162, 0.0360, 0.0436], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0105, 0.0092, 0.0138, 0.0074, 0.0118, 0.0125, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 07:28:28,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.204e+02 2.505e+02 3.030e+02 6.585e+02, threshold=5.010e+02, percent-clipped=3.0 2023-04-30 07:28:31,844 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154720.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:28:50,223 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154732.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:28:59,903 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154740.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:29:16,068 INFO [train.py:904] (7/8) Epoch 16, batch 2500, loss[loss=0.2232, simple_loss=0.2978, pruned_loss=0.07431, over 16448.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2658, pruned_loss=0.04614, over 3331816.10 frames. ], batch size: 146, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:29:18,089 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154753.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:29:53,836 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154780.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:29:55,131 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=154781.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:30:05,087 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154788.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:30:24,153 INFO [train.py:904] (7/8) Epoch 16, batch 2550, loss[loss=0.2134, simple_loss=0.291, pruned_loss=0.06794, over 11813.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2649, pruned_loss=0.04567, over 3325848.89 frames. ], batch size: 246, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:30:47,018 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.136e+02 2.491e+02 3.165e+02 6.684e+02, threshold=4.981e+02, percent-clipped=5.0 2023-04-30 07:31:01,671 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6056, 3.6457, 4.1309, 2.2654, 4.2251, 4.3211, 3.2019, 3.1506], device='cuda:7'), covar=tensor([0.0796, 0.0244, 0.0186, 0.1060, 0.0085, 0.0181, 0.0398, 0.0434], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0106, 0.0093, 0.0138, 0.0074, 0.0118, 0.0126, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 07:31:02,608 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154830.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:31:32,685 INFO [train.py:904] (7/8) Epoch 16, batch 2600, loss[loss=0.1858, simple_loss=0.2659, pruned_loss=0.05287, over 16726.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2653, pruned_loss=0.04604, over 3326657.22 frames. ], batch size: 134, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:31:46,984 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9249, 5.3186, 5.4622, 5.1962, 5.2514, 5.8825, 5.3972, 5.1060], device='cuda:7'), covar=tensor([0.1056, 0.1967, 0.2291, 0.2233, 0.2800, 0.1031, 0.1442, 0.2452], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0567, 0.0614, 0.0480, 0.0635, 0.0647, 0.0484, 0.0631], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:32:07,885 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154877.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:32:40,415 INFO [train.py:904] (7/8) Epoch 16, batch 2650, loss[loss=0.1589, simple_loss=0.2471, pruned_loss=0.03539, over 17207.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2655, pruned_loss=0.04568, over 3330571.38 frames. ], batch size: 44, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:33:01,395 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.264e+02 2.671e+02 3.159e+02 6.185e+02, threshold=5.341e+02, percent-clipped=3.0 2023-04-30 07:33:12,872 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154925.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:33:18,902 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3215, 3.5056, 3.6240, 2.0819, 2.9225, 2.3851, 3.8435, 3.7839], device='cuda:7'), covar=tensor([0.0261, 0.0892, 0.0568, 0.1899, 0.0878, 0.0945, 0.0583, 0.0996], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0156, 0.0162, 0.0148, 0.0139, 0.0126, 0.0140, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:33:35,801 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:33:43,598 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154947.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:33:47,365 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0641, 4.4538, 3.2467, 2.4071, 2.9060, 2.6583, 4.6834, 3.7471], device='cuda:7'), covar=tensor([0.2564, 0.0636, 0.1592, 0.2818, 0.2784, 0.1872, 0.0440, 0.1276], device='cuda:7'), in_proj_covar=tensor([0.0315, 0.0267, 0.0295, 0.0296, 0.0289, 0.0240, 0.0283, 0.0323], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:33:49,084 INFO [train.py:904] (7/8) Epoch 16, batch 2700, loss[loss=0.1696, simple_loss=0.2675, pruned_loss=0.03589, over 16506.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.265, pruned_loss=0.04512, over 3326447.32 frames. ], batch size: 68, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:34:22,426 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=154976.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:30,161 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 07:34:32,712 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=154984.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:41,400 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154990.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:45,872 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=154993.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 07:34:47,985 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=154995.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:34:56,803 INFO [train.py:904] (7/8) Epoch 16, batch 2750, loss[loss=0.1795, simple_loss=0.2669, pruned_loss=0.04603, over 17219.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2656, pruned_loss=0.04454, over 3331021.79 frames. ], batch size: 45, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:35:18,259 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.250e+02 2.149e+02 2.425e+02 2.902e+02 6.172e+02, threshold=4.850e+02, percent-clipped=1.0 2023-04-30 07:35:46,137 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155037.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:35:50,516 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155041.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 07:36:04,763 INFO [train.py:904] (7/8) Epoch 16, batch 2800, loss[loss=0.1624, simple_loss=0.2561, pruned_loss=0.03437, over 17010.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2656, pruned_loss=0.04459, over 3322914.91 frames. ], batch size: 50, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:36:29,118 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155069.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:36:38,308 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155076.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:37:15,725 INFO [train.py:904] (7/8) Epoch 16, batch 2850, loss[loss=0.1535, simple_loss=0.2486, pruned_loss=0.02918, over 17128.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2637, pruned_loss=0.04409, over 3330414.64 frames. ], batch size: 47, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:37:36,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.238e+02 2.628e+02 3.275e+02 9.061e+02, threshold=5.256e+02, percent-clipped=3.0 2023-04-30 07:37:54,370 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155130.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:37:54,466 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155130.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:38:23,957 INFO [train.py:904] (7/8) Epoch 16, batch 2900, loss[loss=0.179, simple_loss=0.2614, pruned_loss=0.04826, over 16523.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2641, pruned_loss=0.04541, over 3317483.43 frames. ], batch size: 68, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:38:51,543 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5965, 2.3638, 2.3785, 4.5901, 2.3271, 2.8154, 2.4445, 2.5411], device='cuda:7'), covar=tensor([0.1083, 0.3387, 0.2541, 0.0371, 0.3811, 0.2360, 0.3085, 0.3217], device='cuda:7'), in_proj_covar=tensor([0.0385, 0.0422, 0.0352, 0.0324, 0.0425, 0.0487, 0.0391, 0.0494], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:39:01,142 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155178.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:39:33,229 INFO [train.py:904] (7/8) Epoch 16, batch 2950, loss[loss=0.1936, simple_loss=0.2841, pruned_loss=0.05161, over 17063.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2639, pruned_loss=0.04626, over 3308669.94 frames. ], batch size: 53, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:39:54,114 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.372e+02 2.734e+02 3.300e+02 7.430e+02, threshold=5.468e+02, percent-clipped=3.0 2023-04-30 07:40:09,275 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-04-30 07:40:40,762 INFO [train.py:904] (7/8) Epoch 16, batch 3000, loss[loss=0.1855, simple_loss=0.2824, pruned_loss=0.04433, over 17171.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2656, pruned_loss=0.04764, over 3297678.56 frames. ], batch size: 46, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:40:40,763 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 07:40:49,856 INFO [train.py:938] (7/8) Epoch 16, validation: loss=0.1368, simple_loss=0.2429, pruned_loss=0.01541, over 944034.00 frames. 2023-04-30 07:40:49,857 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 07:41:09,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5866, 3.5425, 3.5220, 2.9756, 3.4161, 1.9528, 3.1664, 2.8356], device='cuda:7'), covar=tensor([0.0131, 0.0101, 0.0162, 0.0204, 0.0078, 0.2326, 0.0117, 0.0200], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0142, 0.0191, 0.0175, 0.0162, 0.0202, 0.0179, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:41:34,716 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155284.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:41:48,534 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-30 07:41:59,655 INFO [train.py:904] (7/8) Epoch 16, batch 3050, loss[loss=0.1662, simple_loss=0.2576, pruned_loss=0.03742, over 17211.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2655, pruned_loss=0.04773, over 3293921.50 frames. ], batch size: 46, lr: 4.28e-03, grad_scale: 8.0 2023-04-30 07:42:21,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.286e+02 2.843e+02 3.408e+02 8.038e+02, threshold=5.686e+02, percent-clipped=2.0 2023-04-30 07:42:31,848 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4957, 4.2776, 4.4713, 4.6405, 4.7729, 4.3522, 4.6112, 4.7584], device='cuda:7'), covar=tensor([0.1408, 0.1146, 0.1426, 0.0706, 0.0583, 0.1027, 0.1808, 0.0711], device='cuda:7'), in_proj_covar=tensor([0.0620, 0.0770, 0.0906, 0.0783, 0.0582, 0.0612, 0.0616, 0.0725], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:42:37,610 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 07:42:42,402 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155332.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:42:42,425 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155332.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:42:45,148 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 07:43:10,101 INFO [train.py:904] (7/8) Epoch 16, batch 3100, loss[loss=0.1847, simple_loss=0.261, pruned_loss=0.05425, over 15397.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2653, pruned_loss=0.04784, over 3284783.81 frames. ], batch size: 190, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:43:28,198 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4858, 5.8906, 5.3494, 5.7066, 5.1977, 5.0416, 5.4056, 5.8875], device='cuda:7'), covar=tensor([0.2065, 0.1505, 0.2628, 0.1279, 0.1776, 0.1459, 0.2115, 0.2050], device='cuda:7'), in_proj_covar=tensor([0.0646, 0.0797, 0.0647, 0.0575, 0.0501, 0.0508, 0.0663, 0.0612], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:43:42,798 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155376.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:44:17,698 INFO [train.py:904] (7/8) Epoch 16, batch 3150, loss[loss=0.163, simple_loss=0.2406, pruned_loss=0.0427, over 16853.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2643, pruned_loss=0.04703, over 3299807.68 frames. ], batch size: 42, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:44:18,519 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 07:44:39,891 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.373e+02 2.704e+02 3.278e+02 4.749e+02, threshold=5.408e+02, percent-clipped=0.0 2023-04-30 07:44:43,560 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9784, 2.3465, 2.3654, 2.8210, 2.2201, 3.2714, 1.7120, 2.7028], device='cuda:7'), covar=tensor([0.1058, 0.0657, 0.1015, 0.0180, 0.0150, 0.0357, 0.1413, 0.0693], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0187, 0.0173, 0.0200, 0.0212, 0.0189, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:44:47,922 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155424.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:44:49,870 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155425.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:45:13,329 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-04-30 07:45:17,686 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 07:45:19,893 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-30 07:45:24,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7512, 5.1275, 4.8950, 4.9051, 4.6727, 4.6026, 4.5823, 5.2427], device='cuda:7'), covar=tensor([0.1131, 0.0869, 0.1041, 0.0803, 0.0811, 0.1197, 0.1040, 0.0946], device='cuda:7'), in_proj_covar=tensor([0.0648, 0.0803, 0.0651, 0.0579, 0.0504, 0.0512, 0.0668, 0.0616], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:45:27,254 INFO [train.py:904] (7/8) Epoch 16, batch 3200, loss[loss=0.1731, simple_loss=0.2688, pruned_loss=0.03874, over 17246.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2635, pruned_loss=0.04672, over 3296279.91 frames. ], batch size: 52, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:45:34,168 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155457.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:46:36,135 INFO [train.py:904] (7/8) Epoch 16, batch 3250, loss[loss=0.1941, simple_loss=0.2727, pruned_loss=0.05773, over 16880.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2629, pruned_loss=0.04708, over 3294614.26 frames. ], batch size: 116, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:46:58,465 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.206e+02 2.546e+02 3.042e+02 5.764e+02, threshold=5.092e+02, percent-clipped=1.0 2023-04-30 07:46:58,922 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155518.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:47:35,028 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2023-04-30 07:47:45,925 INFO [train.py:904] (7/8) Epoch 16, batch 3300, loss[loss=0.2017, simple_loss=0.2809, pruned_loss=0.06125, over 16880.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2643, pruned_loss=0.04782, over 3303079.61 frames. ], batch size: 109, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:47:57,477 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4428, 5.4225, 5.2689, 4.8407, 4.9018, 5.3721, 5.2702, 5.0191], device='cuda:7'), covar=tensor([0.0521, 0.0439, 0.0295, 0.0339, 0.1151, 0.0428, 0.0291, 0.0753], device='cuda:7'), in_proj_covar=tensor([0.0289, 0.0401, 0.0340, 0.0331, 0.0359, 0.0381, 0.0234, 0.0411], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:48:29,583 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9791, 1.9771, 2.1863, 3.5227, 2.0855, 2.2495, 2.1149, 2.0961], device='cuda:7'), covar=tensor([0.1354, 0.3761, 0.2627, 0.0636, 0.3911, 0.2520, 0.3619, 0.3425], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0425, 0.0355, 0.0326, 0.0428, 0.0490, 0.0393, 0.0497], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:48:46,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-30 07:48:47,784 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0984, 4.1672, 4.4873, 4.4522, 4.4751, 4.1486, 4.2008, 4.0495], device='cuda:7'), covar=tensor([0.0339, 0.0568, 0.0415, 0.0429, 0.0520, 0.0472, 0.0804, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0423, 0.0412, 0.0392, 0.0466, 0.0441, 0.0537, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 07:48:56,953 INFO [train.py:904] (7/8) Epoch 16, batch 3350, loss[loss=0.1623, simple_loss=0.2596, pruned_loss=0.03248, over 17127.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2643, pruned_loss=0.04707, over 3303482.07 frames. ], batch size: 49, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:49:01,211 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155605.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:49:19,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.838e+02 2.398e+02 2.739e+02 3.162e+02 6.710e+02, threshold=5.477e+02, percent-clipped=3.0 2023-04-30 07:49:25,976 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8866, 3.8473, 4.1987, 1.9258, 4.2394, 4.4836, 3.2430, 3.4265], device='cuda:7'), covar=tensor([0.0777, 0.0242, 0.0266, 0.1352, 0.0134, 0.0176, 0.0436, 0.0405], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0107, 0.0094, 0.0140, 0.0076, 0.0122, 0.0127, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 07:49:39,809 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:50:08,446 INFO [train.py:904] (7/8) Epoch 16, batch 3400, loss[loss=0.2011, simple_loss=0.2786, pruned_loss=0.06187, over 16416.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2638, pruned_loss=0.04687, over 3298595.06 frames. ], batch size: 146, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:50:16,673 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155657.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:50:28,088 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155666.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:50:49,344 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155680.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:51:19,270 INFO [train.py:904] (7/8) Epoch 16, batch 3450, loss[loss=0.1671, simple_loss=0.2618, pruned_loss=0.03618, over 17030.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2622, pruned_loss=0.046, over 3303267.56 frames. ], batch size: 55, lr: 4.27e-03, grad_scale: 16.0 2023-04-30 07:51:41,376 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.296e+02 2.647e+02 3.295e+02 8.789e+02, threshold=5.295e+02, percent-clipped=1.0 2023-04-30 07:51:41,889 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155718.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:51:52,578 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=155725.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:52:28,787 INFO [train.py:904] (7/8) Epoch 16, batch 3500, loss[loss=0.1565, simple_loss=0.2342, pruned_loss=0.03939, over 16917.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2598, pruned_loss=0.04508, over 3307551.87 frames. ], batch size: 96, lr: 4.27e-03, grad_scale: 16.0 2023-04-30 07:52:58,256 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=155773.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:53:14,796 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=155784.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:53:40,590 INFO [train.py:904] (7/8) Epoch 16, batch 3550, loss[loss=0.182, simple_loss=0.2768, pruned_loss=0.04364, over 16720.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2584, pruned_loss=0.04427, over 3312718.03 frames. ], batch size: 57, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:53:56,193 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155813.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:54:04,791 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.310e+02 2.739e+02 3.224e+02 5.856e+02, threshold=5.477e+02, percent-clipped=2.0 2023-04-30 07:54:15,900 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3620, 3.6482, 3.8007, 2.1171, 2.9952, 2.4047, 3.7426, 3.7223], device='cuda:7'), covar=tensor([0.0248, 0.0776, 0.0499, 0.1840, 0.0829, 0.0945, 0.0599, 0.0949], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0159, 0.0164, 0.0150, 0.0141, 0.0127, 0.0143, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:54:42,494 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=155845.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 07:54:52,022 INFO [train.py:904] (7/8) Epoch 16, batch 3600, loss[loss=0.1371, simple_loss=0.2242, pruned_loss=0.02498, over 17006.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2572, pruned_loss=0.04302, over 3314429.46 frames. ], batch size: 41, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:56:03,895 INFO [train.py:904] (7/8) Epoch 16, batch 3650, loss[loss=0.1876, simple_loss=0.2574, pruned_loss=0.05888, over 16676.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.256, pruned_loss=0.04372, over 3306259.50 frames. ], batch size: 134, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:56:08,922 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7710, 3.9411, 2.2958, 4.1984, 3.0215, 4.1602, 2.4550, 3.1617], device='cuda:7'), covar=tensor([0.0261, 0.0360, 0.1560, 0.0280, 0.0711, 0.0491, 0.1303, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0176, 0.0193, 0.0157, 0.0174, 0.0218, 0.0203, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 07:56:28,567 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.258e+02 2.651e+02 3.350e+02 6.145e+02, threshold=5.302e+02, percent-clipped=1.0 2023-04-30 07:56:43,181 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-04-30 07:56:43,897 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3982, 5.7149, 5.4166, 5.5225, 5.1512, 5.0389, 5.1389, 5.8059], device='cuda:7'), covar=tensor([0.0977, 0.0685, 0.0904, 0.0679, 0.0748, 0.0698, 0.0922, 0.0809], device='cuda:7'), in_proj_covar=tensor([0.0637, 0.0792, 0.0640, 0.0572, 0.0494, 0.0504, 0.0656, 0.0606], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 07:57:17,740 INFO [train.py:904] (7/8) Epoch 16, batch 3700, loss[loss=0.1816, simple_loss=0.2533, pruned_loss=0.05494, over 16680.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2547, pruned_loss=0.04542, over 3300399.99 frames. ], batch size: 124, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:57:32,545 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=155961.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:58:36,928 INFO [train.py:904] (7/8) Epoch 16, batch 3750, loss[loss=0.202, simple_loss=0.2929, pruned_loss=0.05555, over 16710.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2556, pruned_loss=0.04702, over 3294242.52 frames. ], batch size: 57, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 07:58:43,955 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3340, 4.2551, 4.3862, 2.9669, 3.6915, 4.3408, 3.8768, 2.3693], device='cuda:7'), covar=tensor([0.0483, 0.0041, 0.0029, 0.0327, 0.0085, 0.0057, 0.0073, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0076, 0.0076, 0.0130, 0.0089, 0.0099, 0.0088, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 07:58:53,139 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156013.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 07:59:02,687 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.313e+02 2.777e+02 3.166e+02 7.362e+02, threshold=5.555e+02, percent-clipped=3.0 2023-04-30 07:59:51,249 INFO [train.py:904] (7/8) Epoch 16, batch 3800, loss[loss=0.1765, simple_loss=0.2603, pruned_loss=0.04634, over 16904.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2572, pruned_loss=0.04866, over 3290045.20 frames. ], batch size: 58, lr: 4.27e-03, grad_scale: 8.0 2023-04-30 08:00:17,801 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156070.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:01:03,118 INFO [train.py:904] (7/8) Epoch 16, batch 3850, loss[loss=0.1687, simple_loss=0.2393, pruned_loss=0.04907, over 16735.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2571, pruned_loss=0.0492, over 3286203.28 frames. ], batch size: 83, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:01:12,784 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5278, 4.8379, 4.6243, 4.6298, 4.3413, 4.2921, 4.3382, 4.9000], device='cuda:7'), covar=tensor([0.1116, 0.0774, 0.0927, 0.0729, 0.0725, 0.1312, 0.0972, 0.0808], device='cuda:7'), in_proj_covar=tensor([0.0632, 0.0782, 0.0636, 0.0566, 0.0489, 0.0500, 0.0649, 0.0598], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:01:20,124 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156113.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:01:28,705 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.683e+02 2.283e+02 2.734e+02 3.408e+02 7.535e+02, threshold=5.468e+02, percent-clipped=3.0 2023-04-30 08:01:47,434 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156131.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:01:59,184 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156140.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:02:16,231 INFO [train.py:904] (7/8) Epoch 16, batch 3900, loss[loss=0.1662, simple_loss=0.2437, pruned_loss=0.04438, over 15500.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2575, pruned_loss=0.05008, over 3275114.51 frames. ], batch size: 190, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:02:29,483 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156161.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:03:26,079 INFO [train.py:904] (7/8) Epoch 16, batch 3950, loss[loss=0.1738, simple_loss=0.2459, pruned_loss=0.0509, over 16782.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2572, pruned_loss=0.05052, over 3278904.58 frames. ], batch size: 76, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:03:31,191 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 08:03:47,419 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:03:50,344 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.580e+02 2.349e+02 2.934e+02 3.426e+02 7.028e+02, threshold=5.868e+02, percent-clipped=1.0 2023-04-30 08:04:37,102 INFO [train.py:904] (7/8) Epoch 16, batch 4000, loss[loss=0.1896, simple_loss=0.2646, pruned_loss=0.05732, over 16430.00 frames. ], tot_loss[loss=0.18, simple_loss=0.2578, pruned_loss=0.05112, over 3273130.07 frames. ], batch size: 146, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:04:45,904 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6670, 3.9837, 2.9981, 2.2838, 2.8621, 2.7217, 4.2208, 3.5883], device='cuda:7'), covar=tensor([0.2833, 0.0626, 0.1701, 0.2364, 0.2233, 0.1633, 0.0457, 0.0921], device='cuda:7'), in_proj_covar=tensor([0.0315, 0.0266, 0.0295, 0.0295, 0.0292, 0.0241, 0.0282, 0.0322], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:04:49,881 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156261.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:05:12,184 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:05:31,825 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156290.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:05:34,525 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0706, 3.2473, 3.4587, 1.9537, 3.0071, 2.2467, 3.5677, 3.5227], device='cuda:7'), covar=tensor([0.0216, 0.0772, 0.0540, 0.1942, 0.0787, 0.0949, 0.0509, 0.0770], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0158, 0.0162, 0.0149, 0.0140, 0.0127, 0.0141, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 08:05:49,252 INFO [train.py:904] (7/8) Epoch 16, batch 4050, loss[loss=0.1796, simple_loss=0.2643, pruned_loss=0.0474, over 16452.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2586, pruned_loss=0.05002, over 3264670.65 frames. ], batch size: 146, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:05:59,657 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156309.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:06:05,016 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156313.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:06:14,227 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 1.898e+02 2.203e+02 2.661e+02 4.268e+02, threshold=4.405e+02, percent-clipped=0.0 2023-04-30 08:07:01,348 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156351.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:07:02,006 INFO [train.py:904] (7/8) Epoch 16, batch 4100, loss[loss=0.233, simple_loss=0.3091, pruned_loss=0.07849, over 11574.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2599, pruned_loss=0.04955, over 3259299.19 frames. ], batch size: 246, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:07:15,199 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156361.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:07:31,240 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6231, 2.6225, 2.4471, 3.8934, 2.9024, 3.8568, 1.5342, 2.7682], device='cuda:7'), covar=tensor([0.1394, 0.0793, 0.1231, 0.0183, 0.0288, 0.0401, 0.1650, 0.0865], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0169, 0.0188, 0.0174, 0.0204, 0.0214, 0.0191, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 08:08:15,981 INFO [train.py:904] (7/8) Epoch 16, batch 4150, loss[loss=0.2124, simple_loss=0.2981, pruned_loss=0.06332, over 16428.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2667, pruned_loss=0.05209, over 3232737.53 frames. ], batch size: 35, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:08:21,463 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 08:08:40,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 2.118e+02 2.582e+02 3.252e+02 7.284e+02, threshold=5.164e+02, percent-clipped=7.0 2023-04-30 08:08:52,212 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156426.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:09:13,291 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156440.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:09:30,683 INFO [train.py:904] (7/8) Epoch 16, batch 4200, loss[loss=0.2222, simple_loss=0.3103, pruned_loss=0.06703, over 16602.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2742, pruned_loss=0.05429, over 3213339.55 frames. ], batch size: 75, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:10:24,425 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156488.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:10:27,437 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-30 08:10:27,623 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-30 08:10:43,958 INFO [train.py:904] (7/8) Epoch 16, batch 4250, loss[loss=0.1817, simple_loss=0.2785, pruned_loss=0.0425, over 16450.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2773, pruned_loss=0.05401, over 3186273.65 frames. ], batch size: 75, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:11:09,165 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.212e+02 2.674e+02 3.089e+02 6.979e+02, threshold=5.347e+02, percent-clipped=2.0 2023-04-30 08:11:35,579 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1991, 3.2833, 1.6628, 3.5268, 2.4629, 3.5343, 1.8907, 2.4766], device='cuda:7'), covar=tensor([0.0275, 0.0346, 0.1974, 0.0182, 0.0818, 0.0443, 0.1694, 0.0806], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0173, 0.0191, 0.0150, 0.0173, 0.0213, 0.0200, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 08:11:56,450 INFO [train.py:904] (7/8) Epoch 16, batch 4300, loss[loss=0.1947, simple_loss=0.2812, pruned_loss=0.05414, over 15240.00 frames. ], tot_loss[loss=0.1925, simple_loss=0.2789, pruned_loss=0.0531, over 3182756.22 frames. ], batch size: 190, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:12:16,029 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5233, 2.4919, 2.6154, 3.9485, 3.3153, 3.9042, 1.2858, 2.9820], device='cuda:7'), covar=tensor([0.1334, 0.0755, 0.1067, 0.0191, 0.0244, 0.0353, 0.1643, 0.0716], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0185, 0.0171, 0.0200, 0.0210, 0.0188, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:7') 2023-04-30 08:12:19,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1855, 4.0267, 4.0523, 2.6342, 3.5484, 4.0217, 3.6532, 2.2133], device='cuda:7'), covar=tensor([0.0494, 0.0026, 0.0034, 0.0356, 0.0078, 0.0062, 0.0073, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0076, 0.0077, 0.0131, 0.0090, 0.0099, 0.0088, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:12:27,170 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156572.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:13:09,406 INFO [train.py:904] (7/8) Epoch 16, batch 4350, loss[loss=0.2214, simple_loss=0.3021, pruned_loss=0.0703, over 11520.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2825, pruned_loss=0.0546, over 3180339.02 frames. ], batch size: 246, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:13:34,599 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.777e+02 2.270e+02 2.623e+02 3.147e+02 7.539e+02, threshold=5.245e+02, percent-clipped=2.0 2023-04-30 08:14:14,561 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=156646.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:14:22,116 INFO [train.py:904] (7/8) Epoch 16, batch 4400, loss[loss=0.2001, simple_loss=0.292, pruned_loss=0.05416, over 16691.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2842, pruned_loss=0.05502, over 3196481.01 frames. ], batch size: 57, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:15:06,851 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5062, 1.6193, 2.1505, 2.4086, 2.3903, 2.7814, 1.7356, 2.7352], device='cuda:7'), covar=tensor([0.0163, 0.0463, 0.0258, 0.0251, 0.0262, 0.0143, 0.0462, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0184, 0.0170, 0.0174, 0.0185, 0.0140, 0.0183, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:15:32,119 INFO [train.py:904] (7/8) Epoch 16, batch 4450, loss[loss=0.1941, simple_loss=0.2876, pruned_loss=0.05032, over 16523.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2869, pruned_loss=0.05578, over 3207244.18 frames. ], batch size: 68, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:15:57,564 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.034e+02 2.328e+02 2.907e+02 4.699e+02, threshold=4.656e+02, percent-clipped=0.0 2023-04-30 08:16:04,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7705, 3.7835, 2.3735, 4.5678, 2.8842, 4.4716, 2.5490, 2.9993], device='cuda:7'), covar=tensor([0.0268, 0.0399, 0.1599, 0.0119, 0.0867, 0.0420, 0.1371, 0.0796], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0172, 0.0190, 0.0149, 0.0172, 0.0212, 0.0199, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 08:16:09,711 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156726.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:16:41,475 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156749.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:16:45,647 INFO [train.py:904] (7/8) Epoch 16, batch 4500, loss[loss=0.191, simple_loss=0.2816, pruned_loss=0.05016, over 16901.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.287, pruned_loss=0.05602, over 3212929.76 frames. ], batch size: 116, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:16:47,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1534, 2.0499, 1.7166, 1.8070, 2.3044, 1.9301, 2.1010, 2.3868], device='cuda:7'), covar=tensor([0.0135, 0.0299, 0.0420, 0.0338, 0.0183, 0.0293, 0.0168, 0.0204], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0217, 0.0210, 0.0211, 0.0220, 0.0221, 0.0224, 0.0217], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:16:49,116 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-30 08:17:16,201 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156774.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:17:19,786 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 08:17:54,747 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6894, 2.6849, 2.4603, 3.6104, 2.7714, 3.7654, 1.5000, 2.8921], device='cuda:7'), covar=tensor([0.1263, 0.0647, 0.1100, 0.0140, 0.0240, 0.0368, 0.1558, 0.0718], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0168, 0.0187, 0.0172, 0.0203, 0.0213, 0.0190, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 08:17:56,565 INFO [train.py:904] (7/8) Epoch 16, batch 4550, loss[loss=0.2175, simple_loss=0.2856, pruned_loss=0.07463, over 11942.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2876, pruned_loss=0.05689, over 3214978.86 frames. ], batch size: 246, lr: 4.26e-03, grad_scale: 8.0 2023-04-30 08:18:07,894 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156810.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:18:16,518 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3781, 2.9643, 2.6217, 2.2712, 2.2109, 2.1458, 2.9041, 2.8259], device='cuda:7'), covar=tensor([0.2276, 0.0692, 0.1412, 0.2013, 0.2139, 0.2029, 0.0439, 0.1024], device='cuda:7'), in_proj_covar=tensor([0.0315, 0.0264, 0.0295, 0.0295, 0.0291, 0.0240, 0.0282, 0.0320], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:18:19,419 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.005e+02 2.274e+02 2.770e+02 4.941e+02, threshold=4.548e+02, percent-clipped=1.0 2023-04-30 08:19:06,336 INFO [train.py:904] (7/8) Epoch 16, batch 4600, loss[loss=0.186, simple_loss=0.2778, pruned_loss=0.04707, over 17179.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.289, pruned_loss=0.05763, over 3209106.00 frames. ], batch size: 46, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:19:37,069 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156872.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:19:55,783 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=156885.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:20:19,278 INFO [train.py:904] (7/8) Epoch 16, batch 4650, loss[loss=0.1766, simple_loss=0.2623, pruned_loss=0.0455, over 16535.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2877, pruned_loss=0.05737, over 3220897.79 frames. ], batch size: 68, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:20:37,659 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-30 08:20:39,128 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-04-30 08:20:45,037 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 1.893e+02 2.205e+02 2.654e+02 4.694e+02, threshold=4.410e+02, percent-clipped=1.0 2023-04-30 08:20:46,968 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156920.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:20:48,410 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6660, 3.5732, 3.9478, 1.9848, 4.2245, 4.2412, 3.0490, 3.0385], device='cuda:7'), covar=tensor([0.0782, 0.0275, 0.0264, 0.1218, 0.0064, 0.0117, 0.0443, 0.0475], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0139, 0.0074, 0.0119, 0.0126, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 08:21:12,277 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-30 08:21:25,726 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=156946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:21:25,816 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=156946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:21:33,080 INFO [train.py:904] (7/8) Epoch 16, batch 4700, loss[loss=0.1835, simple_loss=0.272, pruned_loss=0.04751, over 16765.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.2849, pruned_loss=0.05625, over 3232311.19 frames. ], batch size: 83, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:22:36,836 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=156994.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:22:47,496 INFO [train.py:904] (7/8) Epoch 16, batch 4750, loss[loss=0.163, simple_loss=0.2558, pruned_loss=0.03515, over 16843.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2818, pruned_loss=0.05473, over 3219724.56 frames. ], batch size: 102, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:22:59,901 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157010.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:23:11,424 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 1.885e+02 2.255e+02 2.719e+02 5.243e+02, threshold=4.511e+02, percent-clipped=2.0 2023-04-30 08:23:59,611 INFO [train.py:904] (7/8) Epoch 16, batch 4800, loss[loss=0.2229, simple_loss=0.2971, pruned_loss=0.07437, over 11909.00 frames. ], tot_loss[loss=0.191, simple_loss=0.2779, pruned_loss=0.05208, over 3221514.32 frames. ], batch size: 246, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:24:29,044 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157071.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:25:14,791 INFO [train.py:904] (7/8) Epoch 16, batch 4850, loss[loss=0.188, simple_loss=0.2747, pruned_loss=0.05064, over 12462.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2779, pruned_loss=0.05104, over 3205803.26 frames. ], batch size: 246, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:25:20,721 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157105.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:25:39,365 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8949, 5.1749, 5.4076, 5.1547, 5.2049, 5.7708, 5.2607, 5.0534], device='cuda:7'), covar=tensor([0.0885, 0.1649, 0.1693, 0.1856, 0.2271, 0.0785, 0.1209, 0.2112], device='cuda:7'), in_proj_covar=tensor([0.0384, 0.0544, 0.0586, 0.0460, 0.0613, 0.0622, 0.0464, 0.0614], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:25:41,921 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.885e+02 2.190e+02 2.625e+02 3.825e+02, threshold=4.379e+02, percent-clipped=0.0 2023-04-30 08:26:17,000 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157142.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:26:30,916 INFO [train.py:904] (7/8) Epoch 16, batch 4900, loss[loss=0.1632, simple_loss=0.255, pruned_loss=0.03567, over 16642.00 frames. ], tot_loss[loss=0.1881, simple_loss=0.2773, pruned_loss=0.04949, over 3205664.82 frames. ], batch size: 62, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:27:45,030 INFO [train.py:904] (7/8) Epoch 16, batch 4950, loss[loss=0.2177, simple_loss=0.2982, pruned_loss=0.06866, over 12127.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2765, pruned_loss=0.049, over 3186770.03 frames. ], batch size: 247, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:27:46,813 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157203.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:27:59,803 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6555, 4.6783, 4.4929, 4.2153, 4.1363, 4.5880, 4.3855, 4.2977], device='cuda:7'), covar=tensor([0.0505, 0.0448, 0.0276, 0.0276, 0.0992, 0.0465, 0.0447, 0.0636], device='cuda:7'), in_proj_covar=tensor([0.0264, 0.0367, 0.0313, 0.0302, 0.0329, 0.0349, 0.0214, 0.0374], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:28:08,811 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.143e+02 2.414e+02 2.690e+02 6.197e+02, threshold=4.827e+02, percent-clipped=2.0 2023-04-30 08:28:41,079 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157241.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:28:57,717 INFO [train.py:904] (7/8) Epoch 16, batch 5000, loss[loss=0.1948, simple_loss=0.2901, pruned_loss=0.04977, over 16470.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2785, pruned_loss=0.04969, over 3187559.35 frames. ], batch size: 146, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:30:10,077 INFO [train.py:904] (7/8) Epoch 16, batch 5050, loss[loss=0.1778, simple_loss=0.275, pruned_loss=0.04026, over 16794.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2789, pruned_loss=0.04981, over 3188633.08 frames. ], batch size: 83, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:30:33,956 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.188e+02 2.518e+02 3.068e+02 6.843e+02, threshold=5.037e+02, percent-clipped=3.0 2023-04-30 08:31:22,541 INFO [train.py:904] (7/8) Epoch 16, batch 5100, loss[loss=0.1897, simple_loss=0.2799, pruned_loss=0.04976, over 15504.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2774, pruned_loss=0.04916, over 3193297.22 frames. ], batch size: 191, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:31:29,231 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157356.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:31:44,280 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157366.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:32:07,534 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157382.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:32:35,788 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8245, 3.6568, 3.9409, 3.6239, 3.8444, 4.2723, 3.9597, 3.5681], device='cuda:7'), covar=tensor([0.2128, 0.2140, 0.1757, 0.2446, 0.2859, 0.1837, 0.1295, 0.2522], device='cuda:7'), in_proj_covar=tensor([0.0381, 0.0536, 0.0577, 0.0449, 0.0606, 0.0618, 0.0457, 0.0607], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:32:38,192 INFO [train.py:904] (7/8) Epoch 16, batch 5150, loss[loss=0.2078, simple_loss=0.3011, pruned_loss=0.05729, over 16759.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2773, pruned_loss=0.04821, over 3186351.79 frames. ], batch size: 124, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:32:42,310 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157404.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:32:44,009 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157405.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:33:02,272 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157417.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:33:04,663 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 1.981e+02 2.319e+02 2.667e+02 4.266e+02, threshold=4.638e+02, percent-clipped=0.0 2023-04-30 08:33:14,463 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 08:33:40,909 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157443.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:33:53,102 INFO [train.py:904] (7/8) Epoch 16, batch 5200, loss[loss=0.2031, simple_loss=0.2805, pruned_loss=0.06284, over 16882.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2754, pruned_loss=0.04766, over 3202611.63 frames. ], batch size: 109, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:33:54,778 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157453.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:34:13,199 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157465.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:35:01,959 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:35:05,997 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 08:35:08,195 INFO [train.py:904] (7/8) Epoch 16, batch 5250, loss[loss=0.179, simple_loss=0.267, pruned_loss=0.0455, over 16914.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2729, pruned_loss=0.04705, over 3209187.50 frames. ], batch size: 109, lr: 4.25e-03, grad_scale: 8.0 2023-04-30 08:35:14,335 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157506.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:35:32,956 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.109e+02 2.306e+02 2.689e+02 3.834e+02, threshold=4.611e+02, percent-clipped=0.0 2023-04-30 08:36:00,292 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2149, 4.0699, 4.0593, 2.7252, 3.5610, 4.0570, 3.6705, 2.1273], device='cuda:7'), covar=tensor([0.0564, 0.0037, 0.0040, 0.0355, 0.0084, 0.0097, 0.0081, 0.0445], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0074, 0.0076, 0.0129, 0.0089, 0.0098, 0.0087, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:36:03,345 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8489, 5.1829, 5.3559, 5.1837, 5.1644, 5.7336, 5.1756, 4.8799], device='cuda:7'), covar=tensor([0.0995, 0.1414, 0.1317, 0.1667, 0.2294, 0.0778, 0.1363, 0.2441], device='cuda:7'), in_proj_covar=tensor([0.0382, 0.0538, 0.0578, 0.0451, 0.0608, 0.0618, 0.0459, 0.0608], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:36:06,359 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157541.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:36:21,504 INFO [train.py:904] (7/8) Epoch 16, batch 5300, loss[loss=0.1672, simple_loss=0.2576, pruned_loss=0.03844, over 16308.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2695, pruned_loss=0.0458, over 3207510.86 frames. ], batch size: 165, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:36:45,031 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157567.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:36:46,226 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157568.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:37:17,232 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157589.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:37:35,572 INFO [train.py:904] (7/8) Epoch 16, batch 5350, loss[loss=0.1966, simple_loss=0.2877, pruned_loss=0.05278, over 16389.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2681, pruned_loss=0.04563, over 3198818.17 frames. ], batch size: 146, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:38:00,211 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 1.990e+02 2.338e+02 2.844e+02 4.721e+02, threshold=4.676e+02, percent-clipped=1.0 2023-04-30 08:38:10,830 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-30 08:38:16,607 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157629.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:38:49,047 INFO [train.py:904] (7/8) Epoch 16, batch 5400, loss[loss=0.2011, simple_loss=0.2927, pruned_loss=0.05472, over 16594.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2712, pruned_loss=0.0464, over 3196866.16 frames. ], batch size: 134, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:38:52,804 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1627, 3.5794, 3.1751, 1.6693, 2.7958, 2.0161, 3.5947, 3.7092], device='cuda:7'), covar=tensor([0.0256, 0.0641, 0.0721, 0.2252, 0.1027, 0.1115, 0.0585, 0.0852], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0156, 0.0163, 0.0149, 0.0140, 0.0126, 0.0141, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 08:39:09,687 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157666.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:39:30,854 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4173, 5.3644, 5.1486, 4.3490, 5.3446, 1.8407, 4.9930, 5.0132], device='cuda:7'), covar=tensor([0.0081, 0.0087, 0.0162, 0.0470, 0.0082, 0.2645, 0.0119, 0.0215], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0136, 0.0183, 0.0171, 0.0155, 0.0194, 0.0171, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:39:45,115 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 08:40:04,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4651, 5.8298, 5.4874, 5.5473, 5.2208, 5.1144, 5.1979, 5.9224], device='cuda:7'), covar=tensor([0.1227, 0.0751, 0.0973, 0.0857, 0.0809, 0.0626, 0.1081, 0.0815], device='cuda:7'), in_proj_covar=tensor([0.0613, 0.0758, 0.0623, 0.0550, 0.0476, 0.0483, 0.0628, 0.0580], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:40:04,952 INFO [train.py:904] (7/8) Epoch 16, batch 5450, loss[loss=0.186, simple_loss=0.2714, pruned_loss=0.05027, over 17054.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2742, pruned_loss=0.04797, over 3193179.45 frames. ], batch size: 55, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:40:20,462 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157712.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:40:23,523 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157714.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:40:29,849 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 2.375e+02 2.743e+02 3.382e+02 6.991e+02, threshold=5.486e+02, percent-clipped=5.0 2023-04-30 08:41:00,232 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157738.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:41:20,987 INFO [train.py:904] (7/8) Epoch 16, batch 5500, loss[loss=0.2581, simple_loss=0.3287, pruned_loss=0.0938, over 11805.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.282, pruned_loss=0.05332, over 3145780.57 frames. ], batch size: 248, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:41:33,715 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157760.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:42:30,942 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=157798.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:42:35,970 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4215, 1.6362, 2.1204, 2.3639, 2.4358, 2.6464, 1.7635, 2.6081], device='cuda:7'), covar=tensor([0.0165, 0.0439, 0.0252, 0.0268, 0.0258, 0.0174, 0.0426, 0.0117], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0186, 0.0172, 0.0175, 0.0185, 0.0142, 0.0185, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:42:36,608 INFO [train.py:904] (7/8) Epoch 16, batch 5550, loss[loss=0.2146, simple_loss=0.2962, pruned_loss=0.06651, over 16458.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2899, pruned_loss=0.0588, over 3126607.57 frames. ], batch size: 146, lr: 4.24e-03, grad_scale: 16.0 2023-04-30 08:43:02,163 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 08:43:04,451 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.202e+02 3.935e+02 4.988e+02 8.755e+02, threshold=7.870e+02, percent-clipped=18.0 2023-04-30 08:43:31,899 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.08 vs. limit=5.0 2023-04-30 08:43:47,867 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=157846.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:43:57,056 INFO [train.py:904] (7/8) Epoch 16, batch 5600, loss[loss=0.2003, simple_loss=0.2805, pruned_loss=0.06004, over 16612.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.295, pruned_loss=0.06316, over 3106547.68 frames. ], batch size: 57, lr: 4.24e-03, grad_scale: 16.0 2023-04-30 08:44:00,072 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=157853.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:44:07,212 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 08:44:14,256 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157862.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:45:01,284 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1808, 3.3983, 3.5767, 3.5415, 3.5501, 3.3559, 3.3792, 3.4477], device='cuda:7'), covar=tensor([0.0445, 0.0783, 0.0525, 0.0551, 0.0551, 0.0651, 0.0939, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0402, 0.0391, 0.0371, 0.0442, 0.0417, 0.0513, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 08:45:21,498 INFO [train.py:904] (7/8) Epoch 16, batch 5650, loss[loss=0.2903, simple_loss=0.3438, pruned_loss=0.1184, over 11309.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3006, pruned_loss=0.0679, over 3054768.88 frames. ], batch size: 248, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:45:36,533 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4397, 2.3977, 2.4732, 4.2286, 2.1085, 2.7234, 2.4325, 2.5082], device='cuda:7'), covar=tensor([0.1080, 0.3012, 0.2330, 0.0434, 0.3811, 0.2028, 0.2882, 0.2982], device='cuda:7'), in_proj_covar=tensor([0.0376, 0.0417, 0.0346, 0.0317, 0.0420, 0.0481, 0.0385, 0.0486], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:45:41,563 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=157914.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:45:50,675 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.188e+02 3.549e+02 4.142e+02 4.836e+02 1.033e+03, threshold=8.283e+02, percent-clipped=2.0 2023-04-30 08:45:56,919 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=157924.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:46:39,307 INFO [train.py:904] (7/8) Epoch 16, batch 5700, loss[loss=0.2015, simple_loss=0.3004, pruned_loss=0.05136, over 16708.00 frames. ], tot_loss[loss=0.2202, simple_loss=0.3024, pruned_loss=0.06899, over 3066081.68 frames. ], batch size: 83, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:48:01,944 INFO [train.py:904] (7/8) Epoch 16, batch 5750, loss[loss=0.2577, simple_loss=0.3172, pruned_loss=0.09913, over 11056.00 frames. ], tot_loss[loss=0.2229, simple_loss=0.305, pruned_loss=0.0704, over 3065185.89 frames. ], batch size: 247, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:48:17,389 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158012.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:48:27,438 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 08:48:30,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 3.002e+02 3.593e+02 4.729e+02 9.320e+02, threshold=7.185e+02, percent-clipped=2.0 2023-04-30 08:49:00,628 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158038.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:49:22,786 INFO [train.py:904] (7/8) Epoch 16, batch 5800, loss[loss=0.2426, simple_loss=0.3153, pruned_loss=0.08496, over 11958.00 frames. ], tot_loss[loss=0.223, simple_loss=0.3054, pruned_loss=0.07024, over 3030733.93 frames. ], batch size: 248, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:49:35,951 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158060.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:49:36,008 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158060.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:50:16,665 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158086.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:50:41,306 INFO [train.py:904] (7/8) Epoch 16, batch 5850, loss[loss=0.2027, simple_loss=0.2886, pruned_loss=0.05846, over 17187.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.3025, pruned_loss=0.06802, over 3045061.52 frames. ], batch size: 45, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:50:44,888 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8972, 3.1236, 3.1538, 2.1072, 2.9698, 3.1955, 3.0447, 1.7868], device='cuda:7'), covar=tensor([0.0538, 0.0058, 0.0063, 0.0388, 0.0088, 0.0095, 0.0082, 0.0465], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0074, 0.0075, 0.0128, 0.0088, 0.0097, 0.0086, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 08:50:51,715 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158108.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 08:50:54,092 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7061, 4.7231, 5.1083, 5.0841, 5.0921, 4.7866, 4.7557, 4.5725], device='cuda:7'), covar=tensor([0.0302, 0.0498, 0.0384, 0.0419, 0.0439, 0.0341, 0.0886, 0.0461], device='cuda:7'), in_proj_covar=tensor([0.0375, 0.0404, 0.0394, 0.0373, 0.0446, 0.0419, 0.0514, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 08:51:08,529 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.851e+02 3.457e+02 4.318e+02 8.019e+02, threshold=6.915e+02, percent-clipped=1.0 2023-04-30 08:51:13,080 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158122.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:52:03,769 INFO [train.py:904] (7/8) Epoch 16, batch 5900, loss[loss=0.1817, simple_loss=0.2765, pruned_loss=0.04339, over 16724.00 frames. ], tot_loss[loss=0.2192, simple_loss=0.3023, pruned_loss=0.06805, over 3046870.96 frames. ], batch size: 89, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:52:24,572 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158162.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 08:52:55,990 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158183.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:53:25,274 INFO [train.py:904] (7/8) Epoch 16, batch 5950, loss[loss=0.2119, simple_loss=0.2992, pruned_loss=0.06234, over 16238.00 frames. ], tot_loss[loss=0.2186, simple_loss=0.3034, pruned_loss=0.06695, over 3042017.53 frames. ], batch size: 165, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:53:37,057 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=158209.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:53:38,263 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158210.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 08:53:55,271 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.924e+02 3.226e+02 4.367e+02 1.058e+03, threshold=6.453e+02, percent-clipped=4.0 2023-04-30 08:54:01,835 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158224.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:54:45,450 INFO [train.py:904] (7/8) Epoch 16, batch 6000, loss[loss=0.2122, simple_loss=0.2924, pruned_loss=0.06602, over 11662.00 frames. ], tot_loss[loss=0.217, simple_loss=0.3016, pruned_loss=0.06622, over 3056032.67 frames. ], batch size: 248, lr: 4.24e-03, grad_scale: 8.0 2023-04-30 08:54:45,451 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 08:54:56,477 INFO [train.py:938] (7/8) Epoch 16, validation: loss=0.1553, simple_loss=0.2682, pruned_loss=0.0212, over 944034.00 frames. 2023-04-30 08:54:56,478 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 08:55:14,500 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.02 vs. limit=5.0 2023-04-30 08:55:27,088 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158272.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:56:13,212 INFO [train.py:904] (7/8) Epoch 16, batch 6050, loss[loss=0.2103, simple_loss=0.3092, pruned_loss=0.05571, over 17280.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2996, pruned_loss=0.0647, over 3085069.58 frames. ], batch size: 52, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 08:56:40,248 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.935e+02 2.678e+02 3.324e+02 4.253e+02 8.310e+02, threshold=6.647e+02, percent-clipped=4.0 2023-04-30 08:56:43,022 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8001, 4.7722, 4.5790, 3.8991, 4.6809, 1.6848, 4.4143, 4.4281], device='cuda:7'), covar=tensor([0.0098, 0.0087, 0.0187, 0.0401, 0.0111, 0.2796, 0.0142, 0.0206], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0136, 0.0183, 0.0170, 0.0155, 0.0193, 0.0169, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:57:31,988 INFO [train.py:904] (7/8) Epoch 16, batch 6100, loss[loss=0.2067, simple_loss=0.289, pruned_loss=0.06218, over 16391.00 frames. ], tot_loss[loss=0.2127, simple_loss=0.2989, pruned_loss=0.0633, over 3109334.24 frames. ], batch size: 146, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 08:57:47,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1886, 4.5513, 4.0179, 4.3440, 4.0857, 4.0412, 4.0297, 4.5463], device='cuda:7'), covar=tensor([0.2036, 0.1397, 0.2343, 0.1427, 0.1570, 0.2232, 0.2287, 0.1671], device='cuda:7'), in_proj_covar=tensor([0.0604, 0.0746, 0.0610, 0.0544, 0.0466, 0.0476, 0.0618, 0.0571], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:58:05,658 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 08:58:24,020 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158384.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 08:58:49,721 INFO [train.py:904] (7/8) Epoch 16, batch 6150, loss[loss=0.1915, simple_loss=0.2774, pruned_loss=0.05276, over 17053.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.297, pruned_loss=0.06269, over 3118906.10 frames. ], batch size: 53, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 08:59:18,316 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.901e+02 2.793e+02 3.279e+02 3.951e+02 7.811e+02, threshold=6.558e+02, percent-clipped=1.0 2023-04-30 08:59:57,728 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2221, 3.9929, 3.8210, 4.3249, 4.3956, 4.1727, 4.3829, 4.4241], device='cuda:7'), covar=tensor([0.1645, 0.1340, 0.2577, 0.1038, 0.1116, 0.1730, 0.1214, 0.1262], device='cuda:7'), in_proj_covar=tensor([0.0584, 0.0720, 0.0846, 0.0728, 0.0544, 0.0572, 0.0579, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 08:59:57,787 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158445.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:00:08,000 INFO [train.py:904] (7/8) Epoch 16, batch 6200, loss[loss=0.2099, simple_loss=0.2954, pruned_loss=0.06215, over 15443.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.2954, pruned_loss=0.06254, over 3119967.19 frames. ], batch size: 191, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 09:00:12,924 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8727, 2.0974, 2.4211, 3.1600, 2.1647, 2.3387, 2.2791, 2.2252], device='cuda:7'), covar=tensor([0.1147, 0.3253, 0.2115, 0.0622, 0.3666, 0.2200, 0.3032, 0.3060], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0419, 0.0348, 0.0319, 0.0423, 0.0483, 0.0387, 0.0489], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:00:20,384 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4801, 3.4771, 3.4437, 2.7479, 3.3241, 2.1464, 3.1526, 2.8260], device='cuda:7'), covar=tensor([0.0160, 0.0119, 0.0177, 0.0234, 0.0108, 0.2155, 0.0140, 0.0215], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0137, 0.0184, 0.0171, 0.0156, 0.0195, 0.0170, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:00:48,501 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=158478.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:00:58,089 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6740, 2.6985, 2.3108, 2.5133, 3.1382, 2.8012, 3.3327, 3.3973], device='cuda:7'), covar=tensor([0.0097, 0.0375, 0.0471, 0.0417, 0.0219, 0.0349, 0.0206, 0.0208], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0219, 0.0213, 0.0212, 0.0220, 0.0221, 0.0224, 0.0215], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:01:21,949 INFO [train.py:904] (7/8) Epoch 16, batch 6250, loss[loss=0.2048, simple_loss=0.2888, pruned_loss=0.06038, over 15538.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2947, pruned_loss=0.06223, over 3119878.49 frames. ], batch size: 191, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:01:34,528 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158509.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:01:50,906 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.618e+02 3.109e+02 4.242e+02 1.066e+03, threshold=6.218e+02, percent-clipped=4.0 2023-04-30 09:02:08,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8785, 2.0944, 2.4065, 3.1362, 2.1787, 2.2994, 2.2523, 2.1895], device='cuda:7'), covar=tensor([0.1149, 0.3180, 0.2136, 0.0642, 0.3747, 0.2368, 0.3186, 0.3121], device='cuda:7'), in_proj_covar=tensor([0.0377, 0.0416, 0.0346, 0.0317, 0.0421, 0.0480, 0.0385, 0.0485], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:02:38,751 INFO [train.py:904] (7/8) Epoch 16, batch 6300, loss[loss=0.1855, simple_loss=0.2764, pruned_loss=0.04734, over 16683.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2942, pruned_loss=0.06137, over 3124296.29 frames. ], batch size: 134, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:02:45,574 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158557.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:02:59,490 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6587, 4.4717, 4.6537, 4.8599, 5.0126, 4.5489, 5.0475, 5.0219], device='cuda:7'), covar=tensor([0.1897, 0.1351, 0.1965, 0.0812, 0.0647, 0.0936, 0.0589, 0.0672], device='cuda:7'), in_proj_covar=tensor([0.0583, 0.0721, 0.0846, 0.0726, 0.0546, 0.0572, 0.0580, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:03:00,577 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9202, 4.9205, 4.7333, 4.0391, 4.7974, 1.8192, 4.5714, 4.5727], device='cuda:7'), covar=tensor([0.0079, 0.0069, 0.0155, 0.0373, 0.0094, 0.2465, 0.0119, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0137, 0.0184, 0.0171, 0.0156, 0.0195, 0.0170, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:03:55,409 INFO [train.py:904] (7/8) Epoch 16, batch 6350, loss[loss=0.2263, simple_loss=0.3031, pruned_loss=0.07473, over 16353.00 frames. ], tot_loss[loss=0.2096, simple_loss=0.2947, pruned_loss=0.06222, over 3116982.68 frames. ], batch size: 165, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:04:24,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.016e+02 2.970e+02 3.577e+02 4.444e+02 9.031e+02, threshold=7.154e+02, percent-clipped=4.0 2023-04-30 09:05:11,913 INFO [train.py:904] (7/8) Epoch 16, batch 6400, loss[loss=0.2668, simple_loss=0.3313, pruned_loss=0.1012, over 11699.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2947, pruned_loss=0.0632, over 3116460.44 frames. ], batch size: 248, lr: 4.23e-03, grad_scale: 8.0 2023-04-30 09:06:09,559 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6166, 4.6424, 4.5116, 3.1981, 4.5495, 1.5883, 4.1680, 4.1025], device='cuda:7'), covar=tensor([0.0198, 0.0136, 0.0236, 0.0762, 0.0154, 0.3501, 0.0234, 0.0411], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0137, 0.0184, 0.0171, 0.0156, 0.0195, 0.0171, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:06:17,653 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-30 09:06:27,728 INFO [train.py:904] (7/8) Epoch 16, batch 6450, loss[loss=0.2226, simple_loss=0.3182, pruned_loss=0.06352, over 17016.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2945, pruned_loss=0.06264, over 3115869.87 frames. ], batch size: 41, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:06:31,341 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-04-30 09:06:56,390 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.032e+02 2.821e+02 3.663e+02 4.597e+02 9.189e+02, threshold=7.326e+02, percent-clipped=6.0 2023-04-30 09:07:11,739 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 09:07:26,497 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=158740.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:07:43,547 INFO [train.py:904] (7/8) Epoch 16, batch 6500, loss[loss=0.2175, simple_loss=0.2906, pruned_loss=0.07213, over 11732.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2919, pruned_loss=0.06185, over 3106663.94 frames. ], batch size: 247, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:08:21,763 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:08:22,826 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=158778.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:08:31,735 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158783.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:08:54,703 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-04-30 09:09:01,982 INFO [train.py:904] (7/8) Epoch 16, batch 6550, loss[loss=0.1936, simple_loss=0.297, pruned_loss=0.04512, over 16881.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2947, pruned_loss=0.0623, over 3112603.54 frames. ], batch size: 96, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:09:33,176 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.010e+02 2.836e+02 3.301e+02 3.944e+02 7.483e+02, threshold=6.603e+02, percent-clipped=1.0 2023-04-30 09:09:39,690 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=158826.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:09:53,256 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=158835.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 09:09:58,238 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158838.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:10:06,330 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158844.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:10:18,016 INFO [train.py:904] (7/8) Epoch 16, batch 6600, loss[loss=0.205, simple_loss=0.2939, pruned_loss=0.05801, over 16243.00 frames. ], tot_loss[loss=0.2117, simple_loss=0.2972, pruned_loss=0.06307, over 3103479.78 frames. ], batch size: 165, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:10:24,190 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8911, 4.1525, 3.9129, 3.9887, 3.7421, 3.8152, 3.7927, 4.1371], device='cuda:7'), covar=tensor([0.1140, 0.0912, 0.1091, 0.0783, 0.0801, 0.1381, 0.0963, 0.1053], device='cuda:7'), in_proj_covar=tensor([0.0605, 0.0744, 0.0611, 0.0547, 0.0467, 0.0478, 0.0619, 0.0571], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:10:38,277 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2595, 3.8012, 3.3738, 1.8231, 2.8455, 2.1294, 3.6160, 3.9020], device='cuda:7'), covar=tensor([0.0260, 0.0644, 0.0731, 0.2282, 0.1077, 0.1174, 0.0655, 0.0889], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0156, 0.0161, 0.0148, 0.0139, 0.0126, 0.0140, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 09:10:45,152 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 09:10:52,535 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6462, 3.7104, 1.7219, 4.2135, 2.7086, 4.1118, 1.9896, 2.7548], device='cuda:7'), covar=tensor([0.0251, 0.0332, 0.2215, 0.0223, 0.0866, 0.0486, 0.1975, 0.0845], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0170, 0.0193, 0.0147, 0.0173, 0.0209, 0.0200, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 09:11:26,056 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=158896.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 09:11:34,221 INFO [train.py:904] (7/8) Epoch 16, batch 6650, loss[loss=0.1868, simple_loss=0.2851, pruned_loss=0.04429, over 16805.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2973, pruned_loss=0.06355, over 3102379.09 frames. ], batch size: 83, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:12:04,651 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 3.031e+02 3.577e+02 4.481e+02 9.334e+02, threshold=7.154e+02, percent-clipped=3.0 2023-04-30 09:12:50,525 INFO [train.py:904] (7/8) Epoch 16, batch 6700, loss[loss=0.2619, simple_loss=0.3242, pruned_loss=0.09984, over 11374.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2964, pruned_loss=0.06411, over 3086422.82 frames. ], batch size: 247, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:13:46,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4688, 4.0438, 4.4349, 1.9365, 4.6430, 4.7472, 3.3133, 3.2191], device='cuda:7'), covar=tensor([0.1107, 0.0165, 0.0175, 0.1288, 0.0056, 0.0083, 0.0369, 0.0531], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0106, 0.0092, 0.0138, 0.0073, 0.0117, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 09:14:07,737 INFO [train.py:904] (7/8) Epoch 16, batch 6750, loss[loss=0.1876, simple_loss=0.2722, pruned_loss=0.05154, over 16452.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2939, pruned_loss=0.06296, over 3105075.42 frames. ], batch size: 68, lr: 4.23e-03, grad_scale: 4.0 2023-04-30 09:14:37,801 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.101e+02 2.962e+02 3.496e+02 4.058e+02 1.383e+03, threshold=6.992e+02, percent-clipped=2.0 2023-04-30 09:15:05,837 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159040.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:15:23,313 INFO [train.py:904] (7/8) Epoch 16, batch 6800, loss[loss=0.2271, simple_loss=0.3118, pruned_loss=0.07116, over 16367.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2947, pruned_loss=0.06367, over 3094427.23 frames. ], batch size: 165, lr: 4.22e-03, grad_scale: 8.0 2023-04-30 09:15:50,763 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159069.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:16:21,092 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:16:40,967 INFO [train.py:904] (7/8) Epoch 16, batch 6850, loss[loss=0.2303, simple_loss=0.3211, pruned_loss=0.06982, over 15441.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.296, pruned_loss=0.06386, over 3104230.82 frames. ], batch size: 191, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:17:12,962 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 2.851e+02 3.389e+02 4.147e+02 9.414e+02, threshold=6.778e+02, percent-clipped=4.0 2023-04-30 09:17:22,496 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159130.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:17:26,721 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159133.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:17:35,445 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159139.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:17:54,572 INFO [train.py:904] (7/8) Epoch 16, batch 6900, loss[loss=0.2119, simple_loss=0.2989, pruned_loss=0.06245, over 16206.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2986, pruned_loss=0.06392, over 3096048.37 frames. ], batch size: 165, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:18:07,175 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159159.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:18:10,868 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.19 vs. limit=5.0 2023-04-30 09:18:57,092 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159191.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 09:19:13,874 INFO [train.py:904] (7/8) Epoch 16, batch 6950, loss[loss=0.2198, simple_loss=0.2964, pruned_loss=0.07157, over 15661.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.3006, pruned_loss=0.06542, over 3087620.28 frames. ], batch size: 191, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:19:42,681 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159220.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:19:48,723 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 3.353e+02 4.093e+02 5.038e+02 1.287e+03, threshold=8.186e+02, percent-clipped=11.0 2023-04-30 09:20:29,903 INFO [train.py:904] (7/8) Epoch 16, batch 7000, loss[loss=0.2123, simple_loss=0.3069, pruned_loss=0.05884, over 16535.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.3011, pruned_loss=0.06468, over 3100154.44 frames. ], batch size: 68, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:21:17,269 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 09:21:17,490 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 09:21:28,083 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159290.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:21:45,229 INFO [train.py:904] (7/8) Epoch 16, batch 7050, loss[loss=0.2577, simple_loss=0.3188, pruned_loss=0.09831, over 11206.00 frames. ], tot_loss[loss=0.216, simple_loss=0.302, pruned_loss=0.06498, over 3089956.81 frames. ], batch size: 246, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:22:18,896 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.833e+02 2.988e+02 3.653e+02 4.362e+02 9.002e+02, threshold=7.306e+02, percent-clipped=3.0 2023-04-30 09:23:00,665 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159351.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:23:01,361 INFO [train.py:904] (7/8) Epoch 16, batch 7100, loss[loss=0.1989, simple_loss=0.2759, pruned_loss=0.06096, over 11311.00 frames. ], tot_loss[loss=0.2148, simple_loss=0.3003, pruned_loss=0.06467, over 3089738.10 frames. ], batch size: 247, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:24:17,490 INFO [train.py:904] (7/8) Epoch 16, batch 7150, loss[loss=0.1881, simple_loss=0.2784, pruned_loss=0.04885, over 16518.00 frames. ], tot_loss[loss=0.2136, simple_loss=0.2982, pruned_loss=0.06454, over 3102018.88 frames. ], batch size: 68, lr: 4.22e-03, grad_scale: 2.0 2023-04-30 09:24:49,380 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 3.206e+02 4.048e+02 4.680e+02 7.501e+02, threshold=8.096e+02, percent-clipped=1.0 2023-04-30 09:24:51,595 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159425.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:25:02,916 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:25:11,803 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159439.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:25:30,666 INFO [train.py:904] (7/8) Epoch 16, batch 7200, loss[loss=0.1898, simple_loss=0.2803, pruned_loss=0.04964, over 16734.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2955, pruned_loss=0.06269, over 3084082.24 frames. ], batch size: 83, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:26:15,872 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:26:27,648 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159487.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:26:33,391 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159491.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 09:26:50,674 INFO [train.py:904] (7/8) Epoch 16, batch 7250, loss[loss=0.1895, simple_loss=0.2759, pruned_loss=0.0516, over 16872.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2927, pruned_loss=0.06101, over 3105972.37 frames. ], batch size: 109, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:26:51,202 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2447, 4.2336, 4.1257, 3.4461, 4.1607, 1.7742, 3.9510, 3.6919], device='cuda:7'), covar=tensor([0.0093, 0.0082, 0.0153, 0.0301, 0.0080, 0.2556, 0.0108, 0.0206], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0135, 0.0182, 0.0168, 0.0154, 0.0194, 0.0168, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:27:11,562 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159515.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:27:21,049 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4386, 2.2135, 1.7463, 1.9645, 2.4698, 2.1477, 2.3635, 2.6290], device='cuda:7'), covar=tensor([0.0156, 0.0333, 0.0461, 0.0390, 0.0229, 0.0340, 0.0180, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0217, 0.0211, 0.0211, 0.0217, 0.0217, 0.0220, 0.0212], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:27:23,392 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.928e+02 2.845e+02 3.560e+02 4.449e+02 7.164e+02, threshold=7.119e+02, percent-clipped=0.0 2023-04-30 09:27:29,574 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2482, 2.1384, 2.3649, 4.0785, 1.9856, 2.4834, 2.2549, 2.2705], device='cuda:7'), covar=tensor([0.1420, 0.3865, 0.2686, 0.0544, 0.4920, 0.2853, 0.3527, 0.3902], device='cuda:7'), in_proj_covar=tensor([0.0378, 0.0418, 0.0347, 0.0315, 0.0423, 0.0481, 0.0387, 0.0487], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:27:40,642 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2089, 2.0670, 1.6922, 1.7496, 2.2367, 1.9022, 2.0996, 2.3684], device='cuda:7'), covar=tensor([0.0160, 0.0311, 0.0413, 0.0380, 0.0230, 0.0318, 0.0175, 0.0220], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0216, 0.0210, 0.0211, 0.0217, 0.0217, 0.0219, 0.0212], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:27:47,293 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159539.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 09:28:06,282 INFO [train.py:904] (7/8) Epoch 16, batch 7300, loss[loss=0.2134, simple_loss=0.3067, pruned_loss=0.06005, over 16698.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2921, pruned_loss=0.0608, over 3098782.79 frames. ], batch size: 134, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:29:22,399 INFO [train.py:904] (7/8) Epoch 16, batch 7350, loss[loss=0.1648, simple_loss=0.2595, pruned_loss=0.03504, over 17097.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2922, pruned_loss=0.0613, over 3092479.42 frames. ], batch size: 47, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:29:33,209 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0059, 4.9724, 4.8611, 4.5258, 4.5368, 4.9165, 4.8108, 4.6163], device='cuda:7'), covar=tensor([0.0541, 0.0364, 0.0249, 0.0286, 0.0864, 0.0356, 0.0318, 0.0613], device='cuda:7'), in_proj_covar=tensor([0.0261, 0.0364, 0.0305, 0.0294, 0.0318, 0.0342, 0.0211, 0.0364], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:29:56,736 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 3.169e+02 3.768e+02 4.450e+02 9.358e+02, threshold=7.536e+02, percent-clipped=3.0 2023-04-30 09:30:06,782 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 09:30:07,800 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159631.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:30:28,672 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5307, 3.4636, 2.7971, 2.2174, 2.5485, 2.3631, 3.7135, 3.2966], device='cuda:7'), covar=tensor([0.2926, 0.0868, 0.1769, 0.2681, 0.2321, 0.1943, 0.0568, 0.1245], device='cuda:7'), in_proj_covar=tensor([0.0317, 0.0264, 0.0297, 0.0298, 0.0290, 0.0241, 0.0282, 0.0319], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 09:30:31,263 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159646.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:30:39,224 INFO [train.py:904] (7/8) Epoch 16, batch 7400, loss[loss=0.2017, simple_loss=0.3008, pruned_loss=0.05132, over 16476.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.2937, pruned_loss=0.06257, over 3067746.88 frames. ], batch size: 75, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:31:41,345 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159692.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:31:57,486 INFO [train.py:904] (7/8) Epoch 16, batch 7450, loss[loss=0.2043, simple_loss=0.2909, pruned_loss=0.05881, over 16445.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2949, pruned_loss=0.06385, over 3068175.11 frames. ], batch size: 75, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:32:33,424 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 3.124e+02 3.664e+02 4.558e+02 8.463e+02, threshold=7.327e+02, percent-clipped=4.0 2023-04-30 09:32:36,502 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159725.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:33:17,756 INFO [train.py:904] (7/8) Epoch 16, batch 7500, loss[loss=0.2041, simple_loss=0.2863, pruned_loss=0.06097, over 16673.00 frames. ], tot_loss[loss=0.2112, simple_loss=0.2957, pruned_loss=0.06338, over 3068203.85 frames. ], batch size: 57, lr: 4.22e-03, grad_scale: 4.0 2023-04-30 09:33:51,103 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159773.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:33:53,071 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3013, 2.0431, 2.6055, 3.1989, 3.0002, 3.6575, 2.2006, 3.5771], device='cuda:7'), covar=tensor([0.0176, 0.0476, 0.0323, 0.0252, 0.0268, 0.0127, 0.0491, 0.0106], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0184, 0.0168, 0.0173, 0.0183, 0.0140, 0.0183, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:33:54,977 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4398, 3.4344, 3.4192, 2.4962, 3.3733, 2.0208, 3.0917, 2.6314], device='cuda:7'), covar=tensor([0.0229, 0.0164, 0.0224, 0.0397, 0.0141, 0.2769, 0.0173, 0.0330], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0133, 0.0180, 0.0165, 0.0152, 0.0191, 0.0166, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:34:37,019 INFO [train.py:904] (7/8) Epoch 16, batch 7550, loss[loss=0.1927, simple_loss=0.2731, pruned_loss=0.05614, over 16601.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2945, pruned_loss=0.06337, over 3068518.73 frames. ], batch size: 62, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:34:39,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6588, 1.7217, 2.2337, 2.5652, 2.5798, 2.9017, 1.9051, 2.8532], device='cuda:7'), covar=tensor([0.0160, 0.0483, 0.0276, 0.0256, 0.0265, 0.0154, 0.0477, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0184, 0.0168, 0.0172, 0.0183, 0.0140, 0.0183, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:34:58,285 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:35:10,144 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.970e+02 2.719e+02 3.178e+02 3.820e+02 7.109e+02, threshold=6.356e+02, percent-clipped=0.0 2023-04-30 09:35:22,065 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 09:35:53,840 INFO [train.py:904] (7/8) Epoch 16, batch 7600, loss[loss=0.2103, simple_loss=0.297, pruned_loss=0.06184, over 16612.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2932, pruned_loss=0.06316, over 3075264.07 frames. ], batch size: 62, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:35:59,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159855.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:36:10,414 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159863.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:37:09,877 INFO [train.py:904] (7/8) Epoch 16, batch 7650, loss[loss=0.2377, simple_loss=0.3199, pruned_loss=0.07774, over 16719.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2943, pruned_loss=0.06417, over 3080040.49 frames. ], batch size: 76, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:37:21,803 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=159910.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:37:30,451 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159916.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:37:43,264 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.780e+02 3.207e+02 4.068e+02 5.310e+02 2.281e+03, threshold=8.135e+02, percent-clipped=15.0 2023-04-30 09:38:13,642 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=159946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:38:22,099 INFO [train.py:904] (7/8) Epoch 16, batch 7700, loss[loss=0.2114, simple_loss=0.3034, pruned_loss=0.05968, over 16628.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2951, pruned_loss=0.0649, over 3080571.21 frames. ], batch size: 57, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:38:42,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0182, 2.3867, 2.3889, 2.7988, 2.0895, 3.2420, 1.8209, 2.6803], device='cuda:7'), covar=tensor([0.1102, 0.0570, 0.0995, 0.0158, 0.0132, 0.0390, 0.1384, 0.0710], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0166, 0.0188, 0.0169, 0.0203, 0.0212, 0.0193, 0.0188], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 09:38:51,283 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=159971.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:39:15,452 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=159987.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:39:26,411 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=159994.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:39:40,215 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0925, 5.5393, 5.7842, 5.4400, 5.5073, 6.0785, 5.5406, 5.3589], device='cuda:7'), covar=tensor([0.0801, 0.1737, 0.2199, 0.1836, 0.2137, 0.0859, 0.1409, 0.2028], device='cuda:7'), in_proj_covar=tensor([0.0384, 0.0551, 0.0601, 0.0463, 0.0613, 0.0630, 0.0473, 0.0621], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 09:39:41,050 INFO [train.py:904] (7/8) Epoch 16, batch 7750, loss[loss=0.1829, simple_loss=0.2743, pruned_loss=0.04571, over 17021.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.2949, pruned_loss=0.0642, over 3087117.10 frames. ], batch size: 53, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:40:13,569 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 3.102e+02 3.551e+02 4.185e+02 8.391e+02, threshold=7.102e+02, percent-clipped=1.0 2023-04-30 09:40:33,226 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2333, 3.4283, 3.5870, 3.5640, 3.5776, 3.3779, 3.4188, 3.4572], device='cuda:7'), covar=tensor([0.0434, 0.0742, 0.0524, 0.0526, 0.0530, 0.0598, 0.0898, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0377, 0.0408, 0.0397, 0.0377, 0.0449, 0.0422, 0.0516, 0.0334], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 09:40:41,624 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-04-30 09:40:53,473 INFO [train.py:904] (7/8) Epoch 16, batch 7800, loss[loss=0.2245, simple_loss=0.3135, pruned_loss=0.06774, over 15347.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2956, pruned_loss=0.06452, over 3101015.52 frames. ], batch size: 191, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:42:08,850 INFO [train.py:904] (7/8) Epoch 16, batch 7850, loss[loss=0.1942, simple_loss=0.2806, pruned_loss=0.05389, over 16901.00 frames. ], tot_loss[loss=0.2121, simple_loss=0.2962, pruned_loss=0.06401, over 3102442.79 frames. ], batch size: 116, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:42:43,408 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.758e+02 3.305e+02 4.051e+02 1.069e+03, threshold=6.609e+02, percent-clipped=3.0 2023-04-30 09:42:46,990 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8060, 1.3451, 1.6425, 1.6414, 1.7573, 1.9203, 1.5552, 1.8284], device='cuda:7'), covar=tensor([0.0208, 0.0350, 0.0184, 0.0260, 0.0228, 0.0158, 0.0373, 0.0126], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0183, 0.0167, 0.0172, 0.0182, 0.0140, 0.0183, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:43:25,068 INFO [train.py:904] (7/8) Epoch 16, batch 7900, loss[loss=0.2408, simple_loss=0.3081, pruned_loss=0.08671, over 11525.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2958, pruned_loss=0.06365, over 3096799.53 frames. ], batch size: 246, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:44:43,663 INFO [train.py:904] (7/8) Epoch 16, batch 7950, loss[loss=0.242, simple_loss=0.3078, pruned_loss=0.08807, over 11914.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2972, pruned_loss=0.06507, over 3078561.77 frames. ], batch size: 248, lr: 4.21e-03, grad_scale: 4.0 2023-04-30 09:44:56,971 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=160211.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:44:57,060 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7011, 4.6674, 4.5020, 3.8277, 4.5736, 1.6343, 4.3233, 4.3465], device='cuda:7'), covar=tensor([0.0101, 0.0079, 0.0178, 0.0399, 0.0100, 0.2663, 0.0136, 0.0194], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0133, 0.0180, 0.0165, 0.0151, 0.0191, 0.0166, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:45:00,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3970, 5.6850, 5.4556, 5.5073, 5.1709, 5.0215, 5.1014, 5.8322], device='cuda:7'), covar=tensor([0.1102, 0.0783, 0.0942, 0.0682, 0.0846, 0.0753, 0.1087, 0.0703], device='cuda:7'), in_proj_covar=tensor([0.0598, 0.0733, 0.0600, 0.0537, 0.0461, 0.0474, 0.0608, 0.0561], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:45:16,544 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 2.700e+02 3.245e+02 4.105e+02 1.062e+03, threshold=6.489e+02, percent-clipped=6.0 2023-04-30 09:45:56,356 INFO [train.py:904] (7/8) Epoch 16, batch 8000, loss[loss=0.2582, simple_loss=0.3271, pruned_loss=0.09464, over 11107.00 frames. ], tot_loss[loss=0.2147, simple_loss=0.2976, pruned_loss=0.06587, over 3069310.58 frames. ], batch size: 247, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:46:18,475 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=160266.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:46:50,333 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160287.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:47:04,171 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6843, 3.7141, 2.2226, 4.3237, 2.8687, 4.3046, 2.5048, 2.9821], device='cuda:7'), covar=tensor([0.0254, 0.0368, 0.1663, 0.0216, 0.0780, 0.0560, 0.1401, 0.0735], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0170, 0.0192, 0.0147, 0.0173, 0.0211, 0.0201, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 09:47:12,599 INFO [train.py:904] (7/8) Epoch 16, batch 8050, loss[loss=0.2169, simple_loss=0.2984, pruned_loss=0.06775, over 16401.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2969, pruned_loss=0.06569, over 3050595.08 frames. ], batch size: 146, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:47:13,179 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6081, 5.6014, 5.4386, 5.0974, 5.0171, 5.4569, 5.3874, 5.1580], device='cuda:7'), covar=tensor([0.0535, 0.0337, 0.0259, 0.0270, 0.1051, 0.0375, 0.0236, 0.0645], device='cuda:7'), in_proj_covar=tensor([0.0264, 0.0370, 0.0308, 0.0297, 0.0321, 0.0347, 0.0213, 0.0369], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:47:20,986 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=160307.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 09:47:41,181 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1330, 2.0557, 1.6867, 1.7598, 2.2995, 2.0113, 1.9852, 2.3827], device='cuda:7'), covar=tensor([0.0152, 0.0306, 0.0446, 0.0388, 0.0199, 0.0284, 0.0178, 0.0213], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0216, 0.0209, 0.0210, 0.0215, 0.0215, 0.0219, 0.0212], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:47:47,810 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.763e+02 2.953e+02 3.727e+02 4.663e+02 8.321e+02, threshold=7.453e+02, percent-clipped=3.0 2023-04-30 09:48:04,060 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=160335.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:48:29,614 INFO [train.py:904] (7/8) Epoch 16, batch 8100, loss[loss=0.2422, simple_loss=0.3086, pruned_loss=0.08788, over 11275.00 frames. ], tot_loss[loss=0.2126, simple_loss=0.2958, pruned_loss=0.06473, over 3052074.80 frames. ], batch size: 247, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:48:54,883 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=160368.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 09:48:56,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4615, 4.0968, 4.1233, 2.8793, 3.6271, 4.0864, 3.7167, 2.1649], device='cuda:7'), covar=tensor([0.0481, 0.0040, 0.0036, 0.0319, 0.0083, 0.0084, 0.0067, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0074, 0.0075, 0.0130, 0.0088, 0.0098, 0.0086, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 09:49:37,378 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7879, 3.6911, 4.1571, 2.0022, 4.3056, 4.3025, 3.0987, 3.1430], device='cuda:7'), covar=tensor([0.0761, 0.0239, 0.0154, 0.1218, 0.0050, 0.0132, 0.0378, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0092, 0.0138, 0.0074, 0.0117, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 09:49:46,012 INFO [train.py:904] (7/8) Epoch 16, batch 8150, loss[loss=0.2193, simple_loss=0.2865, pruned_loss=0.07605, over 11630.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2933, pruned_loss=0.06335, over 3070085.56 frames. ], batch size: 248, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:49:55,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0555, 5.5184, 5.7315, 5.4395, 5.5708, 6.0551, 5.5608, 5.3686], device='cuda:7'), covar=tensor([0.0814, 0.1848, 0.2059, 0.1868, 0.2376, 0.1019, 0.1469, 0.2361], device='cuda:7'), in_proj_covar=tensor([0.0384, 0.0553, 0.0602, 0.0464, 0.0616, 0.0633, 0.0473, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 09:50:21,739 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.821e+02 3.128e+02 3.748e+02 4.647e+02 9.176e+02, threshold=7.497e+02, percent-clipped=1.0 2023-04-30 09:50:49,982 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-04-30 09:51:05,078 INFO [train.py:904] (7/8) Epoch 16, batch 8200, loss[loss=0.2169, simple_loss=0.3044, pruned_loss=0.06471, over 16701.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2916, pruned_loss=0.06311, over 3073942.87 frames. ], batch size: 134, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:51:25,039 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9373, 2.4163, 2.4833, 2.9746, 2.3766, 3.1895, 1.7268, 2.8258], device='cuda:7'), covar=tensor([0.1219, 0.0591, 0.0961, 0.0247, 0.0162, 0.0400, 0.1416, 0.0634], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0167, 0.0189, 0.0171, 0.0204, 0.0212, 0.0193, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 09:52:27,321 INFO [train.py:904] (7/8) Epoch 16, batch 8250, loss[loss=0.1783, simple_loss=0.2774, pruned_loss=0.03962, over 16670.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2903, pruned_loss=0.06014, over 3074711.68 frames. ], batch size: 57, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:52:42,478 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160511.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:53:00,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3600, 1.6947, 1.9332, 2.3887, 2.3535, 2.6718, 1.7923, 2.4711], device='cuda:7'), covar=tensor([0.0203, 0.0473, 0.0327, 0.0276, 0.0299, 0.0187, 0.0469, 0.0202], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0184, 0.0169, 0.0173, 0.0183, 0.0141, 0.0184, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:53:04,606 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.122e+02 2.811e+02 3.275e+02 4.165e+02 1.278e+03, threshold=6.551e+02, percent-clipped=3.0 2023-04-30 09:53:17,725 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.75 vs. limit=5.0 2023-04-30 09:53:49,474 INFO [train.py:904] (7/8) Epoch 16, batch 8300, loss[loss=0.1776, simple_loss=0.2758, pruned_loss=0.0397, over 16184.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2877, pruned_loss=0.05702, over 3078429.26 frames. ], batch size: 165, lr: 4.21e-03, grad_scale: 8.0 2023-04-30 09:54:01,476 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=160559.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:54:13,092 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160566.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:54:31,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3802, 3.3285, 2.7581, 2.0717, 2.1320, 2.3156, 3.4182, 3.0084], device='cuda:7'), covar=tensor([0.3024, 0.0685, 0.1654, 0.3067, 0.2569, 0.2068, 0.0440, 0.1289], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0259, 0.0292, 0.0295, 0.0285, 0.0237, 0.0277, 0.0314], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 09:54:41,217 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-04-30 09:55:10,570 INFO [train.py:904] (7/8) Epoch 16, batch 8350, loss[loss=0.1788, simple_loss=0.2831, pruned_loss=0.03725, over 16697.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2873, pruned_loss=0.05526, over 3070273.55 frames. ], batch size: 134, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:55:31,667 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=160614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 09:55:33,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0484, 3.9091, 4.1261, 4.2386, 4.3783, 3.9577, 4.3110, 4.3932], device='cuda:7'), covar=tensor([0.1610, 0.1201, 0.1405, 0.0702, 0.0536, 0.1281, 0.0697, 0.0674], device='cuda:7'), in_proj_covar=tensor([0.0565, 0.0701, 0.0822, 0.0706, 0.0535, 0.0558, 0.0570, 0.0666], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:55:48,786 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.217e+02 2.608e+02 3.326e+02 6.898e+02, threshold=5.216e+02, percent-clipped=1.0 2023-04-30 09:55:58,144 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2413, 3.3974, 3.6215, 3.6037, 3.6180, 3.4605, 3.4637, 3.5131], device='cuda:7'), covar=tensor([0.0422, 0.0856, 0.0498, 0.0538, 0.0530, 0.0585, 0.0904, 0.0538], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0405, 0.0395, 0.0374, 0.0443, 0.0418, 0.0512, 0.0332], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 09:56:33,057 INFO [train.py:904] (7/8) Epoch 16, batch 8400, loss[loss=0.1967, simple_loss=0.2779, pruned_loss=0.0578, over 12085.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2841, pruned_loss=0.05332, over 3042476.63 frames. ], batch size: 246, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:56:51,252 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=160663.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 09:57:09,606 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2707, 2.1176, 2.0981, 3.9554, 2.0514, 2.5065, 2.2284, 2.2490], device='cuda:7'), covar=tensor([0.1086, 0.3659, 0.2911, 0.0431, 0.4412, 0.2514, 0.3552, 0.3623], device='cuda:7'), in_proj_covar=tensor([0.0368, 0.0410, 0.0339, 0.0308, 0.0415, 0.0469, 0.0378, 0.0476], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:57:54,570 INFO [train.py:904] (7/8) Epoch 16, batch 8450, loss[loss=0.1728, simple_loss=0.2635, pruned_loss=0.04105, over 17064.00 frames. ], tot_loss[loss=0.1923, simple_loss=0.282, pruned_loss=0.05127, over 3052812.72 frames. ], batch size: 55, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:58:31,819 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.257e+02 2.696e+02 3.215e+02 7.413e+02, threshold=5.391e+02, percent-clipped=3.0 2023-04-30 09:58:50,575 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3248, 3.4666, 3.6647, 3.6383, 3.6503, 3.4928, 3.5127, 3.5542], device='cuda:7'), covar=tensor([0.0402, 0.0715, 0.0460, 0.0495, 0.0519, 0.0495, 0.0758, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0370, 0.0401, 0.0390, 0.0369, 0.0438, 0.0414, 0.0505, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 09:59:15,557 INFO [train.py:904] (7/8) Epoch 16, batch 8500, loss[loss=0.2052, simple_loss=0.2829, pruned_loss=0.06373, over 15158.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2782, pruned_loss=0.04866, over 3067050.72 frames. ], batch size: 190, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 09:59:44,013 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3315, 2.0240, 1.6214, 1.6124, 2.1960, 1.8773, 2.0493, 2.3318], device='cuda:7'), covar=tensor([0.0191, 0.0401, 0.0561, 0.0478, 0.0279, 0.0379, 0.0211, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0213, 0.0207, 0.0207, 0.0212, 0.0213, 0.0215, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 09:59:58,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0146, 4.0075, 3.9204, 3.2638, 3.9614, 1.8177, 3.7819, 3.5903], device='cuda:7'), covar=tensor([0.0111, 0.0101, 0.0174, 0.0269, 0.0106, 0.2446, 0.0140, 0.0224], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0131, 0.0177, 0.0161, 0.0149, 0.0188, 0.0164, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:00:42,447 INFO [train.py:904] (7/8) Epoch 16, batch 8550, loss[loss=0.2025, simple_loss=0.2993, pruned_loss=0.05283, over 15369.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2762, pruned_loss=0.04775, over 3049431.39 frames. ], batch size: 191, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:00:49,853 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0679, 3.8293, 3.8079, 4.1690, 4.3458, 4.0030, 4.2948, 4.3280], device='cuda:7'), covar=tensor([0.1628, 0.1368, 0.2408, 0.1115, 0.0823, 0.1737, 0.1028, 0.1067], device='cuda:7'), in_proj_covar=tensor([0.0560, 0.0694, 0.0815, 0.0700, 0.0529, 0.0552, 0.0564, 0.0661], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:01:27,144 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.280e+02 2.208e+02 2.559e+02 3.052e+02 5.214e+02, threshold=5.118e+02, percent-clipped=0.0 2023-04-30 10:01:44,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9921, 2.6986, 2.6349, 2.0199, 2.4900, 2.7138, 2.6086, 1.8470], device='cuda:7'), covar=tensor([0.0338, 0.0066, 0.0062, 0.0306, 0.0106, 0.0093, 0.0091, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0074, 0.0075, 0.0129, 0.0088, 0.0098, 0.0087, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 10:01:56,572 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9625, 1.8266, 1.6581, 1.4936, 1.9835, 1.6033, 1.5770, 1.9500], device='cuda:7'), covar=tensor([0.0173, 0.0290, 0.0399, 0.0352, 0.0230, 0.0276, 0.0185, 0.0219], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0213, 0.0207, 0.0207, 0.0211, 0.0213, 0.0215, 0.0207], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:02:19,938 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0726, 3.0436, 1.8489, 3.2626, 2.2709, 3.2720, 1.9964, 2.5921], device='cuda:7'), covar=tensor([0.0320, 0.0425, 0.1637, 0.0287, 0.0880, 0.0699, 0.1617, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0166, 0.0187, 0.0143, 0.0168, 0.0204, 0.0197, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 10:02:22,612 INFO [train.py:904] (7/8) Epoch 16, batch 8600, loss[loss=0.1723, simple_loss=0.2698, pruned_loss=0.03739, over 16890.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2765, pruned_loss=0.04681, over 3051023.98 frames. ], batch size: 90, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:02:25,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2922, 4.1387, 4.3297, 4.4400, 4.6346, 4.1728, 4.5695, 4.5945], device='cuda:7'), covar=tensor([0.1601, 0.1080, 0.1415, 0.0706, 0.0464, 0.1214, 0.0578, 0.0586], device='cuda:7'), in_proj_covar=tensor([0.0557, 0.0690, 0.0810, 0.0696, 0.0526, 0.0550, 0.0560, 0.0658], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:04:02,565 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=160901.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:04:03,219 INFO [train.py:904] (7/8) Epoch 16, batch 8650, loss[loss=0.1636, simple_loss=0.2639, pruned_loss=0.03162, over 16754.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2742, pruned_loss=0.04514, over 3059837.14 frames. ], batch size: 134, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:04:56,267 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.325e+02 2.764e+02 3.397e+02 5.575e+02, threshold=5.528e+02, percent-clipped=1.0 2023-04-30 10:05:52,487 INFO [train.py:904] (7/8) Epoch 16, batch 8700, loss[loss=0.1502, simple_loss=0.2502, pruned_loss=0.02513, over 16835.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2711, pruned_loss=0.04364, over 3062863.49 frames. ], batch size: 102, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:06:14,062 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=160962.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:06:15,225 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=160963.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 10:06:18,643 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=160965.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:07:29,347 INFO [train.py:904] (7/8) Epoch 16, batch 8750, loss[loss=0.1819, simple_loss=0.2819, pruned_loss=0.041, over 16154.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2714, pruned_loss=0.04316, over 3056346.99 frames. ], batch size: 165, lr: 4.20e-03, grad_scale: 4.0 2023-04-30 10:07:53,660 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161011.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 10:08:27,686 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.172e+02 2.627e+02 3.227e+02 5.767e+02, threshold=5.254e+02, percent-clipped=1.0 2023-04-30 10:08:28,321 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:09:20,783 INFO [train.py:904] (7/8) Epoch 16, batch 8800, loss[loss=0.1686, simple_loss=0.2642, pruned_loss=0.03653, over 16568.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2694, pruned_loss=0.0419, over 3055925.53 frames. ], batch size: 62, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:11:05,650 INFO [train.py:904] (7/8) Epoch 16, batch 8850, loss[loss=0.188, simple_loss=0.2879, pruned_loss=0.04407, over 16332.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2725, pruned_loss=0.0415, over 3059124.10 frames. ], batch size: 146, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:11:55,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.310e+02 2.876e+02 3.561e+02 5.593e+02, threshold=5.752e+02, percent-clipped=2.0 2023-04-30 10:12:53,109 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2023-04-30 10:12:53,432 INFO [train.py:904] (7/8) Epoch 16, batch 8900, loss[loss=0.1893, simple_loss=0.2774, pruned_loss=0.05061, over 16808.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2723, pruned_loss=0.04081, over 3054719.86 frames. ], batch size: 124, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:13:01,530 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2286, 3.2996, 2.0334, 3.4718, 2.4774, 3.5027, 2.1189, 2.7232], device='cuda:7'), covar=tensor([0.0254, 0.0333, 0.1447, 0.0243, 0.0769, 0.0485, 0.1490, 0.0645], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0163, 0.0185, 0.0141, 0.0167, 0.0200, 0.0195, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-30 10:14:59,663 INFO [train.py:904] (7/8) Epoch 16, batch 8950, loss[loss=0.1627, simple_loss=0.2535, pruned_loss=0.03593, over 15289.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2718, pruned_loss=0.04147, over 3054663.51 frames. ], batch size: 191, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:15:07,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7278, 2.6875, 1.8882, 2.8462, 2.1506, 2.8405, 2.1051, 2.4356], device='cuda:7'), covar=tensor([0.0295, 0.0351, 0.1268, 0.0215, 0.0692, 0.0579, 0.1197, 0.0628], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0163, 0.0185, 0.0140, 0.0167, 0.0200, 0.0195, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-30 10:15:30,115 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:15:47,472 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.276e+02 2.813e+02 3.338e+02 5.711e+02, threshold=5.626e+02, percent-clipped=0.0 2023-04-30 10:16:02,657 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161231.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:16:04,846 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4235, 1.6666, 2.0231, 2.3558, 2.4018, 2.6203, 1.7530, 2.6558], device='cuda:7'), covar=tensor([0.0195, 0.0480, 0.0306, 0.0262, 0.0303, 0.0181, 0.0482, 0.0132], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0181, 0.0167, 0.0168, 0.0181, 0.0137, 0.0182, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:16:46,889 INFO [train.py:904] (7/8) Epoch 16, batch 9000, loss[loss=0.1619, simple_loss=0.2524, pruned_loss=0.0357, over 15362.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2688, pruned_loss=0.04036, over 3052829.24 frames. ], batch size: 191, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:16:46,889 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 10:16:56,912 INFO [train.py:938] (7/8) Epoch 16, validation: loss=0.1491, simple_loss=0.2531, pruned_loss=0.02259, over 944034.00 frames. 2023-04-30 10:16:56,913 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 10:17:08,098 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:17:49,203 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:18:20,867 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161292.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:18:23,621 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 10:18:40,787 INFO [train.py:904] (7/8) Epoch 16, batch 9050, loss[loss=0.158, simple_loss=0.2508, pruned_loss=0.03264, over 15444.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2694, pruned_loss=0.04082, over 3047749.47 frames. ], batch size: 191, lr: 4.20e-03, grad_scale: 8.0 2023-04-30 10:18:53,371 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7343, 4.8876, 5.0482, 4.8567, 4.9175, 5.4376, 4.8736, 4.6246], device='cuda:7'), covar=tensor([0.1000, 0.1716, 0.1698, 0.1796, 0.2071, 0.0897, 0.1400, 0.2283], device='cuda:7'), in_proj_covar=tensor([0.0364, 0.0530, 0.0573, 0.0442, 0.0585, 0.0609, 0.0453, 0.0594], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 10:19:19,471 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161321.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:19:27,693 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.317e+02 2.661e+02 3.212e+02 6.846e+02, threshold=5.322e+02, percent-clipped=5.0 2023-04-30 10:19:52,575 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0231, 2.3507, 1.9354, 2.0784, 2.6715, 2.3762, 2.6767, 2.8807], device='cuda:7'), covar=tensor([0.0125, 0.0383, 0.0499, 0.0464, 0.0235, 0.0357, 0.0184, 0.0221], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0215, 0.0209, 0.0208, 0.0213, 0.0214, 0.0215, 0.0207], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:20:22,818 INFO [train.py:904] (7/8) Epoch 16, batch 9100, loss[loss=0.1722, simple_loss=0.2568, pruned_loss=0.04384, over 12057.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2687, pruned_loss=0.04104, over 3047456.89 frames. ], batch size: 247, lr: 4.19e-03, grad_scale: 4.0 2023-04-30 10:22:19,397 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161401.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:22:19,988 INFO [train.py:904] (7/8) Epoch 16, batch 9150, loss[loss=0.1805, simple_loss=0.265, pruned_loss=0.04799, over 12045.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2692, pruned_loss=0.04082, over 3047145.98 frames. ], batch size: 248, lr: 4.19e-03, grad_scale: 4.0 2023-04-30 10:23:13,974 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.279e+02 2.699e+02 3.193e+02 4.519e+02, threshold=5.398e+02, percent-clipped=0.0 2023-04-30 10:24:04,532 INFO [train.py:904] (7/8) Epoch 16, batch 9200, loss[loss=0.1704, simple_loss=0.2658, pruned_loss=0.03751, over 16673.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2648, pruned_loss=0.03989, over 3065663.90 frames. ], batch size: 134, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:24:20,982 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:24:24,479 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161462.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:25:08,338 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 10:25:42,687 INFO [train.py:904] (7/8) Epoch 16, batch 9250, loss[loss=0.1522, simple_loss=0.2362, pruned_loss=0.03416, over 12249.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2642, pruned_loss=0.03992, over 3060416.73 frames. ], batch size: 246, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:26:23,327 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161521.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:26:36,795 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.444e+02 2.938e+02 3.576e+02 6.685e+02, threshold=5.875e+02, percent-clipped=5.0 2023-04-30 10:27:34,803 INFO [train.py:904] (7/8) Epoch 16, batch 9300, loss[loss=0.1649, simple_loss=0.2583, pruned_loss=0.03573, over 16280.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.263, pruned_loss=0.03943, over 3066558.58 frames. ], batch size: 165, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:27:46,186 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161557.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:28:23,211 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161572.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:28:42,149 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-04-30 10:28:54,788 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161587.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:29:21,691 INFO [train.py:904] (7/8) Epoch 16, batch 9350, loss[loss=0.1882, simple_loss=0.274, pruned_loss=0.05123, over 16852.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2637, pruned_loss=0.03965, over 3101014.85 frames. ], batch size: 116, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:29:28,475 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161605.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:30:03,217 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161621.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:30:12,563 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 2.307e+02 2.790e+02 3.265e+02 7.040e+02, threshold=5.579e+02, percent-clipped=1.0 2023-04-30 10:30:13,647 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4552, 2.0423, 1.7970, 1.8016, 2.2913, 1.9574, 1.9773, 2.3760], device='cuda:7'), covar=tensor([0.0126, 0.0339, 0.0460, 0.0413, 0.0239, 0.0337, 0.0172, 0.0228], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0218, 0.0211, 0.0211, 0.0216, 0.0217, 0.0217, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:31:04,003 INFO [train.py:904] (7/8) Epoch 16, batch 9400, loss[loss=0.1786, simple_loss=0.2775, pruned_loss=0.03982, over 16368.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2641, pruned_loss=0.03938, over 3097750.00 frames. ], batch size: 146, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:31:21,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4307, 4.5020, 4.3443, 4.0361, 4.0465, 4.4202, 4.1305, 4.2035], device='cuda:7'), covar=tensor([0.0538, 0.0448, 0.0247, 0.0253, 0.0718, 0.0456, 0.0559, 0.0532], device='cuda:7'), in_proj_covar=tensor([0.0256, 0.0357, 0.0299, 0.0289, 0.0309, 0.0335, 0.0207, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:31:39,351 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161669.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:32:18,386 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 10:32:44,168 INFO [train.py:904] (7/8) Epoch 16, batch 9450, loss[loss=0.1642, simple_loss=0.2574, pruned_loss=0.03549, over 12413.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2652, pruned_loss=0.03898, over 3092737.28 frames. ], batch size: 246, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:32:49,857 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2508, 4.2889, 4.4849, 4.2925, 4.3665, 4.8251, 4.4079, 4.1153], device='cuda:7'), covar=tensor([0.1467, 0.1983, 0.1843, 0.2081, 0.2400, 0.1065, 0.1442, 0.2532], device='cuda:7'), in_proj_covar=tensor([0.0357, 0.0520, 0.0565, 0.0435, 0.0579, 0.0602, 0.0446, 0.0582], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 10:33:33,850 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.284e+02 2.576e+02 3.195e+02 6.155e+02, threshold=5.152e+02, percent-clipped=3.0 2023-04-30 10:33:55,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5693, 3.5256, 3.5063, 2.8629, 3.4242, 2.0012, 3.2496, 2.8354], device='cuda:7'), covar=tensor([0.0122, 0.0108, 0.0170, 0.0190, 0.0112, 0.2251, 0.0142, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0128, 0.0172, 0.0155, 0.0147, 0.0186, 0.0160, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:34:23,904 INFO [train.py:904] (7/8) Epoch 16, batch 9500, loss[loss=0.1502, simple_loss=0.24, pruned_loss=0.03024, over 17024.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2648, pruned_loss=0.03877, over 3100796.48 frames. ], batch size: 50, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:34:37,550 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161757.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:36:08,220 INFO [train.py:904] (7/8) Epoch 16, batch 9550, loss[loss=0.1859, simple_loss=0.2796, pruned_loss=0.04616, over 16684.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2647, pruned_loss=0.03923, over 3088982.76 frames. ], batch size: 76, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:36:38,441 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=161815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:36:40,765 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=161816.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:37:00,501 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.146e+02 2.559e+02 3.093e+02 5.125e+02, threshold=5.118e+02, percent-clipped=0.0 2023-04-30 10:37:25,670 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9821, 3.1584, 2.8747, 4.8808, 3.7280, 4.4467, 1.7657, 3.3140], device='cuda:7'), covar=tensor([0.1270, 0.0655, 0.1050, 0.0144, 0.0198, 0.0294, 0.1505, 0.0654], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0160, 0.0184, 0.0162, 0.0189, 0.0204, 0.0189, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:7') 2023-04-30 10:37:51,449 INFO [train.py:904] (7/8) Epoch 16, batch 9600, loss[loss=0.1927, simple_loss=0.2884, pruned_loss=0.04851, over 16868.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2662, pruned_loss=0.04012, over 3072694.54 frames. ], batch size: 116, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:38:28,419 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7935, 3.6399, 3.6848, 3.9491, 3.9841, 3.6626, 4.0470, 4.0526], device='cuda:7'), covar=tensor([0.1486, 0.1256, 0.1762, 0.0834, 0.0826, 0.1846, 0.0759, 0.0858], device='cuda:7'), in_proj_covar=tensor([0.0560, 0.0690, 0.0808, 0.0697, 0.0527, 0.0552, 0.0562, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:38:29,830 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161872.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:38:38,451 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=161876.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:39:02,439 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=161887.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:39:37,501 INFO [train.py:904] (7/8) Epoch 16, batch 9650, loss[loss=0.1769, simple_loss=0.2636, pruned_loss=0.04515, over 16994.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2676, pruned_loss=0.04039, over 3063269.30 frames. ], batch size: 109, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:40:09,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1169, 2.0860, 2.5142, 3.1239, 2.8381, 3.5749, 2.0923, 3.4539], device='cuda:7'), covar=tensor([0.0171, 0.0432, 0.0329, 0.0231, 0.0273, 0.0133, 0.0470, 0.0114], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0177, 0.0164, 0.0166, 0.0178, 0.0134, 0.0179, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:40:22,072 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161920.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:40:36,070 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.311e+02 2.824e+02 3.460e+02 5.730e+02, threshold=5.647e+02, percent-clipped=1.0 2023-04-30 10:40:49,470 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=161935.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:41:23,394 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-30 10:41:23,670 INFO [train.py:904] (7/8) Epoch 16, batch 9700, loss[loss=0.1736, simple_loss=0.2683, pruned_loss=0.03942, over 16825.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2667, pruned_loss=0.04006, over 3074512.42 frames. ], batch size: 124, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:43:08,506 INFO [train.py:904] (7/8) Epoch 16, batch 9750, loss[loss=0.1752, simple_loss=0.2717, pruned_loss=0.03938, over 16887.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2653, pruned_loss=0.03984, over 3080853.73 frames. ], batch size: 116, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:43:58,498 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.156e+02 2.615e+02 3.105e+02 5.589e+02, threshold=5.230e+02, percent-clipped=0.0 2023-04-30 10:44:46,289 INFO [train.py:904] (7/8) Epoch 16, batch 9800, loss[loss=0.1896, simple_loss=0.2934, pruned_loss=0.04294, over 16755.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2653, pruned_loss=0.03883, over 3098976.28 frames. ], batch size: 134, lr: 4.19e-03, grad_scale: 8.0 2023-04-30 10:44:56,987 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162057.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:44:57,342 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-04-30 10:45:34,362 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162077.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:45:34,377 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8086, 1.3163, 1.6849, 1.7140, 1.7717, 1.9421, 1.5841, 1.8189], device='cuda:7'), covar=tensor([0.0267, 0.0372, 0.0202, 0.0280, 0.0283, 0.0187, 0.0389, 0.0126], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0177, 0.0164, 0.0167, 0.0179, 0.0135, 0.0179, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:46:31,260 INFO [train.py:904] (7/8) Epoch 16, batch 9850, loss[loss=0.1807, simple_loss=0.2724, pruned_loss=0.04446, over 16933.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2658, pruned_loss=0.03823, over 3092131.72 frames. ], batch size: 109, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:46:38,640 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162105.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:46:59,788 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162116.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:47:23,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.292e+02 2.812e+02 3.288e+02 4.754e+02, threshold=5.623e+02, percent-clipped=0.0 2023-04-30 10:47:51,549 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162138.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:48:22,264 INFO [train.py:904] (7/8) Epoch 16, batch 9900, loss[loss=0.182, simple_loss=0.2878, pruned_loss=0.03811, over 16297.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2665, pruned_loss=0.03823, over 3080811.89 frames. ], batch size: 146, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:48:36,124 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7781, 4.5648, 4.7861, 4.9591, 5.1099, 4.5805, 5.1217, 5.1228], device='cuda:7'), covar=tensor([0.1547, 0.1239, 0.1535, 0.0650, 0.0476, 0.0910, 0.0449, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0556, 0.0683, 0.0801, 0.0691, 0.0522, 0.0546, 0.0557, 0.0651], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:48:52,743 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162164.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:49:08,598 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162171.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:49:29,382 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0750, 2.5609, 2.6674, 1.8801, 2.7751, 2.8732, 2.4869, 2.4533], device='cuda:7'), covar=tensor([0.0690, 0.0236, 0.0199, 0.0980, 0.0094, 0.0219, 0.0437, 0.0399], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0102, 0.0087, 0.0135, 0.0071, 0.0112, 0.0122, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 10:49:45,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0712, 2.7543, 2.8108, 1.9219, 2.6054, 2.0092, 2.6821, 2.9040], device='cuda:7'), covar=tensor([0.0293, 0.0831, 0.0570, 0.1967, 0.0920, 0.1032, 0.0734, 0.0925], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0147, 0.0158, 0.0145, 0.0136, 0.0123, 0.0137, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 10:50:22,156 INFO [train.py:904] (7/8) Epoch 16, batch 9950, loss[loss=0.1572, simple_loss=0.2499, pruned_loss=0.03228, over 16561.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2687, pruned_loss=0.03881, over 3085601.28 frames. ], batch size: 68, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:51:26,526 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.161e+02 2.630e+02 3.300e+02 8.357e+02, threshold=5.260e+02, percent-clipped=1.0 2023-04-30 10:51:36,431 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162231.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:52:06,247 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2023-04-30 10:52:23,909 INFO [train.py:904] (7/8) Epoch 16, batch 10000, loss[loss=0.1596, simple_loss=0.2575, pruned_loss=0.03085, over 16501.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2677, pruned_loss=0.03881, over 3086893.52 frames. ], batch size: 68, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:53:28,671 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4144, 2.9472, 2.6817, 2.2159, 2.1114, 2.2474, 2.9448, 2.8163], device='cuda:7'), covar=tensor([0.2730, 0.0755, 0.1636, 0.2773, 0.2776, 0.2076, 0.0499, 0.1360], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0250, 0.0284, 0.0286, 0.0269, 0.0231, 0.0271, 0.0304], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:53:46,745 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162292.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:54:04,090 INFO [train.py:904] (7/8) Epoch 16, batch 10050, loss[loss=0.1824, simple_loss=0.2774, pruned_loss=0.04375, over 16912.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2678, pruned_loss=0.03895, over 3058331.75 frames. ], batch size: 116, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:54:54,660 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.225e+02 2.711e+02 3.158e+02 7.314e+02, threshold=5.421e+02, percent-clipped=6.0 2023-04-30 10:54:55,689 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 10:55:36,389 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4929, 4.4903, 4.3219, 3.7939, 4.3580, 1.6561, 4.2059, 4.2039], device='cuda:7'), covar=tensor([0.0161, 0.0193, 0.0222, 0.0362, 0.0171, 0.2666, 0.0177, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0129, 0.0173, 0.0156, 0.0148, 0.0189, 0.0161, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:55:38,464 INFO [train.py:904] (7/8) Epoch 16, batch 10100, loss[loss=0.1714, simple_loss=0.2609, pruned_loss=0.04094, over 16194.00 frames. ], tot_loss[loss=0.173, simple_loss=0.268, pruned_loss=0.03905, over 3059969.58 frames. ], batch size: 165, lr: 4.18e-03, grad_scale: 8.0 2023-04-30 10:56:17,446 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.53 vs. limit=5.0 2023-04-30 10:56:42,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5206, 3.5540, 3.3754, 3.0636, 3.2249, 3.4835, 3.3060, 3.3165], device='cuda:7'), covar=tensor([0.0596, 0.0676, 0.0297, 0.0273, 0.0596, 0.0599, 0.1016, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0251, 0.0346, 0.0293, 0.0281, 0.0300, 0.0326, 0.0200, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-30 10:57:23,036 INFO [train.py:904] (7/8) Epoch 17, batch 0, loss[loss=0.2661, simple_loss=0.3402, pruned_loss=0.09606, over 12129.00 frames. ], tot_loss[loss=0.2661, simple_loss=0.3402, pruned_loss=0.09606, over 12129.00 frames. ], batch size: 246, lr: 4.05e-03, grad_scale: 8.0 2023-04-30 10:57:23,036 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 10:57:30,750 INFO [train.py:938] (7/8) Epoch 17, validation: loss=0.1481, simple_loss=0.2518, pruned_loss=0.02217, over 944034.00 frames. 2023-04-30 10:57:30,751 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 10:57:39,311 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5260, 5.9427, 5.6480, 5.6886, 5.3339, 5.2279, 5.3153, 5.9606], device='cuda:7'), covar=tensor([0.1321, 0.0853, 0.1131, 0.0804, 0.0867, 0.0679, 0.1119, 0.0942], device='cuda:7'), in_proj_covar=tensor([0.0582, 0.0715, 0.0578, 0.0526, 0.0452, 0.0464, 0.0595, 0.0553], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 10:58:09,507 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.232e+02 2.721e+02 3.359e+02 8.921e+02, threshold=5.442e+02, percent-clipped=3.0 2023-04-30 10:58:14,193 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:58:39,859 INFO [train.py:904] (7/8) Epoch 17, batch 50, loss[loss=0.2176, simple_loss=0.2906, pruned_loss=0.07231, over 16482.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.278, pruned_loss=0.05647, over 746403.58 frames. ], batch size: 146, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 10:59:06,512 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162471.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 10:59:23,695 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 10:59:26,540 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-30 10:59:47,959 INFO [train.py:904] (7/8) Epoch 17, batch 100, loss[loss=0.1606, simple_loss=0.2542, pruned_loss=0.03345, over 17214.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2722, pruned_loss=0.05144, over 1321850.84 frames. ], batch size: 45, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:00:00,640 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162511.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:00:12,327 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162519.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:00:26,388 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.355e+02 2.682e+02 3.198e+02 6.473e+02, threshold=5.364e+02, percent-clipped=1.0 2023-04-30 11:00:30,357 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-04-30 11:00:56,553 INFO [train.py:904] (7/8) Epoch 17, batch 150, loss[loss=0.1559, simple_loss=0.2385, pruned_loss=0.03666, over 16832.00 frames. ], tot_loss[loss=0.1835, simple_loss=0.2684, pruned_loss=0.04931, over 1769935.22 frames. ], batch size: 42, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:01:23,616 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162572.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 11:01:44,502 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162587.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:02:05,945 INFO [train.py:904] (7/8) Epoch 17, batch 200, loss[loss=0.1567, simple_loss=0.2427, pruned_loss=0.03535, over 16823.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.269, pruned_loss=0.04908, over 2105174.77 frames. ], batch size: 42, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:02:43,592 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.407e+02 2.768e+02 3.167e+02 5.394e+02, threshold=5.535e+02, percent-clipped=0.0 2023-04-30 11:03:12,320 INFO [train.py:904] (7/8) Epoch 17, batch 250, loss[loss=0.1432, simple_loss=0.2365, pruned_loss=0.02501, over 17216.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2668, pruned_loss=0.04892, over 2368784.23 frames. ], batch size: 45, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:03:22,923 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0164, 5.4175, 5.1636, 5.1616, 4.9117, 4.8498, 4.9170, 5.5004], device='cuda:7'), covar=tensor([0.1415, 0.1006, 0.1213, 0.0892, 0.1010, 0.0940, 0.1153, 0.1086], device='cuda:7'), in_proj_covar=tensor([0.0608, 0.0750, 0.0609, 0.0551, 0.0473, 0.0482, 0.0626, 0.0577], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:03:39,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8595, 2.3593, 2.5021, 4.7757, 2.4475, 2.7858, 2.4782, 2.6453], device='cuda:7'), covar=tensor([0.0952, 0.3644, 0.2723, 0.0353, 0.3771, 0.2507, 0.3392, 0.3399], device='cuda:7'), in_proj_covar=tensor([0.0376, 0.0415, 0.0350, 0.0316, 0.0423, 0.0477, 0.0387, 0.0484], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:04:20,332 INFO [train.py:904] (7/8) Epoch 17, batch 300, loss[loss=0.1515, simple_loss=0.2474, pruned_loss=0.0278, over 17182.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2642, pruned_loss=0.04816, over 2578683.35 frames. ], batch size: 46, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:04:59,737 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.237e+02 2.646e+02 3.133e+02 7.879e+02, threshold=5.293e+02, percent-clipped=2.0 2023-04-30 11:05:03,791 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162733.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:05:29,583 INFO [train.py:904] (7/8) Epoch 17, batch 350, loss[loss=0.1541, simple_loss=0.243, pruned_loss=0.03263, over 17227.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2612, pruned_loss=0.04689, over 2736168.18 frames. ], batch size: 45, lr: 4.05e-03, grad_scale: 2.0 2023-04-30 11:06:07,974 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162781.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:06:36,769 INFO [train.py:904] (7/8) Epoch 17, batch 400, loss[loss=0.1739, simple_loss=0.2638, pruned_loss=0.04199, over 17046.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2595, pruned_loss=0.04651, over 2863264.35 frames. ], batch size: 50, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:07:06,315 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6607, 2.6319, 2.3381, 2.4140, 3.0038, 2.7883, 3.2872, 3.2192], device='cuda:7'), covar=tensor([0.0126, 0.0366, 0.0423, 0.0410, 0.0251, 0.0330, 0.0255, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0227, 0.0218, 0.0218, 0.0225, 0.0226, 0.0228, 0.0218], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:07:11,744 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=162827.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:07:15,903 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.145e+02 2.685e+02 3.328e+02 1.079e+03, threshold=5.370e+02, percent-clipped=6.0 2023-04-30 11:07:46,590 INFO [train.py:904] (7/8) Epoch 17, batch 450, loss[loss=0.1651, simple_loss=0.2567, pruned_loss=0.03673, over 17169.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2578, pruned_loss=0.04519, over 2974174.80 frames. ], batch size: 46, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:07:54,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8031, 2.7379, 2.8235, 4.1830, 3.7212, 4.1626, 1.4204, 3.1912], device='cuda:7'), covar=tensor([0.1359, 0.0647, 0.0983, 0.0153, 0.0135, 0.0344, 0.1594, 0.0682], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0164, 0.0186, 0.0168, 0.0195, 0.0209, 0.0192, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:08:06,782 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=162867.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 11:08:34,331 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=162887.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:08:35,645 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=162888.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:08:55,334 INFO [train.py:904] (7/8) Epoch 17, batch 500, loss[loss=0.1673, simple_loss=0.2602, pruned_loss=0.03721, over 17227.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2561, pruned_loss=0.04439, over 3051557.40 frames. ], batch size: 45, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:08:57,375 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 11:09:32,595 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.133e+02 2.499e+02 3.161e+02 5.064e+02, threshold=4.998e+02, percent-clipped=0.0 2023-04-30 11:09:39,964 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=162935.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:10:01,803 INFO [train.py:904] (7/8) Epoch 17, batch 550, loss[loss=0.1569, simple_loss=0.2437, pruned_loss=0.03509, over 17215.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.256, pruned_loss=0.04458, over 3121354.11 frames. ], batch size: 44, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:10:29,236 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.44 vs. limit=5.0 2023-04-30 11:11:10,218 INFO [train.py:904] (7/8) Epoch 17, batch 600, loss[loss=0.168, simple_loss=0.2414, pruned_loss=0.04726, over 16863.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2557, pruned_loss=0.04493, over 3152115.24 frames. ], batch size: 90, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:11:47,299 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.692e+02 2.256e+02 2.602e+02 3.226e+02 5.329e+02, threshold=5.204e+02, percent-clipped=2.0 2023-04-30 11:12:16,989 INFO [train.py:904] (7/8) Epoch 17, batch 650, loss[loss=0.1497, simple_loss=0.2379, pruned_loss=0.03077, over 16863.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2544, pruned_loss=0.04426, over 3197314.48 frames. ], batch size: 42, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:13:09,338 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6730, 6.1069, 5.8060, 5.9288, 5.4820, 5.4899, 5.5330, 6.1995], device='cuda:7'), covar=tensor([0.1376, 0.0945, 0.1096, 0.0761, 0.0910, 0.0604, 0.1265, 0.0938], device='cuda:7'), in_proj_covar=tensor([0.0629, 0.0777, 0.0631, 0.0570, 0.0489, 0.0496, 0.0647, 0.0599], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:13:09,463 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=163090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:13:25,542 INFO [train.py:904] (7/8) Epoch 17, batch 700, loss[loss=0.1828, simple_loss=0.256, pruned_loss=0.05482, over 16844.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2546, pruned_loss=0.04369, over 3220389.38 frames. ], batch size: 116, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:14:04,490 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.287e+02 2.752e+02 3.270e+02 5.446e+02, threshold=5.503e+02, percent-clipped=2.0 2023-04-30 11:14:34,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7526, 3.7349, 4.0987, 1.8728, 4.2282, 4.2486, 3.2193, 3.1700], device='cuda:7'), covar=tensor([0.0702, 0.0234, 0.0213, 0.1282, 0.0083, 0.0188, 0.0409, 0.0422], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0139, 0.0074, 0.0118, 0.0126, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:14:34,412 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=163151.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:14:35,155 INFO [train.py:904] (7/8) Epoch 17, batch 750, loss[loss=0.1612, simple_loss=0.2429, pruned_loss=0.03977, over 16447.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2553, pruned_loss=0.04425, over 3242963.27 frames. ], batch size: 68, lr: 4.05e-03, grad_scale: 4.0 2023-04-30 11:14:57,157 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163167.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 11:15:18,769 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=163183.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:15:44,403 INFO [train.py:904] (7/8) Epoch 17, batch 800, loss[loss=0.1928, simple_loss=0.271, pruned_loss=0.05727, over 11714.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2553, pruned_loss=0.04459, over 3249472.74 frames. ], batch size: 246, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:16:03,278 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163215.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:16:23,807 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.209e+02 2.648e+02 3.284e+02 7.557e+02, threshold=5.297e+02, percent-clipped=2.0 2023-04-30 11:16:36,603 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=163240.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:16:53,814 INFO [train.py:904] (7/8) Epoch 17, batch 850, loss[loss=0.1796, simple_loss=0.2503, pruned_loss=0.05445, over 16727.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2541, pruned_loss=0.04428, over 3264105.12 frames. ], batch size: 134, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:18:01,153 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=163301.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:18:01,856 INFO [train.py:904] (7/8) Epoch 17, batch 900, loss[loss=0.1405, simple_loss=0.2274, pruned_loss=0.02673, over 16766.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2536, pruned_loss=0.04384, over 3274720.42 frames. ], batch size: 39, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:18:26,590 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2418, 4.2977, 4.6403, 4.6175, 4.6447, 4.3293, 4.3884, 4.2347], device='cuda:7'), covar=tensor([0.0360, 0.0593, 0.0365, 0.0394, 0.0444, 0.0398, 0.0751, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0418, 0.0408, 0.0383, 0.0456, 0.0431, 0.0525, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:18:40,400 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.201e+02 2.666e+02 3.227e+02 4.456e+02, threshold=5.332e+02, percent-clipped=0.0 2023-04-30 11:18:42,654 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6802, 2.5952, 2.2458, 2.7093, 2.9043, 2.8020, 3.3887, 3.1577], device='cuda:7'), covar=tensor([0.0122, 0.0395, 0.0483, 0.0367, 0.0291, 0.0381, 0.0207, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0190, 0.0230, 0.0221, 0.0220, 0.0228, 0.0230, 0.0234, 0.0222], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:18:52,607 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-04-30 11:19:09,582 INFO [train.py:904] (7/8) Epoch 17, batch 950, loss[loss=0.169, simple_loss=0.2489, pruned_loss=0.04455, over 15496.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2539, pruned_loss=0.04398, over 3281392.66 frames. ], batch size: 190, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:19:22,281 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-30 11:20:17,970 INFO [train.py:904] (7/8) Epoch 17, batch 1000, loss[loss=0.1342, simple_loss=0.2202, pruned_loss=0.02412, over 16988.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2524, pruned_loss=0.04318, over 3279241.96 frames. ], batch size: 41, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:20:18,579 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2023-04-30 11:20:35,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8951, 4.9453, 5.4028, 5.3832, 5.3633, 5.0511, 4.9988, 4.7862], device='cuda:7'), covar=tensor([0.0353, 0.0575, 0.0428, 0.0424, 0.0531, 0.0397, 0.0873, 0.0433], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0418, 0.0410, 0.0385, 0.0458, 0.0431, 0.0527, 0.0343], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:20:54,829 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.252e+02 2.741e+02 3.139e+02 5.661e+02, threshold=5.483e+02, percent-clipped=2.0 2023-04-30 11:21:18,983 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=163446.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:21:20,540 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 11:21:26,457 INFO [train.py:904] (7/8) Epoch 17, batch 1050, loss[loss=0.1554, simple_loss=0.234, pruned_loss=0.0384, over 16218.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2515, pruned_loss=0.04257, over 3289226.73 frames. ], batch size: 165, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:22:10,663 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163483.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:22:19,714 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.08 vs. limit=5.0 2023-04-30 11:22:36,974 INFO [train.py:904] (7/8) Epoch 17, batch 1100, loss[loss=0.1756, simple_loss=0.2739, pruned_loss=0.03869, over 17088.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2517, pruned_loss=0.04209, over 3302290.02 frames. ], batch size: 49, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:23:14,825 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.061e+02 2.430e+02 2.825e+02 4.071e+02, threshold=4.861e+02, percent-clipped=0.0 2023-04-30 11:23:16,261 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163531.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:23:43,931 INFO [train.py:904] (7/8) Epoch 17, batch 1150, loss[loss=0.1565, simple_loss=0.2377, pruned_loss=0.03768, over 16620.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2509, pruned_loss=0.04158, over 3300960.33 frames. ], batch size: 76, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:23:53,440 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 11:24:43,085 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=163596.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:24:52,260 INFO [train.py:904] (7/8) Epoch 17, batch 1200, loss[loss=0.1591, simple_loss=0.2617, pruned_loss=0.02828, over 17248.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.25, pruned_loss=0.04105, over 3306386.52 frames. ], batch size: 52, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:25:02,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8182, 2.9458, 3.1408, 2.1089, 2.7547, 2.1756, 3.3369, 3.2442], device='cuda:7'), covar=tensor([0.0259, 0.0952, 0.0579, 0.1830, 0.0845, 0.0992, 0.0589, 0.1025], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0156, 0.0162, 0.0150, 0.0141, 0.0127, 0.0142, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:25:28,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6446, 2.3145, 2.3568, 4.6034, 2.3202, 2.7957, 2.4363, 2.5359], device='cuda:7'), covar=tensor([0.1075, 0.3800, 0.2946, 0.0394, 0.3968, 0.2554, 0.3357, 0.3531], device='cuda:7'), in_proj_covar=tensor([0.0385, 0.0424, 0.0356, 0.0325, 0.0429, 0.0488, 0.0395, 0.0496], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:25:29,956 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.305e+02 2.671e+02 3.522e+02 8.114e+02, threshold=5.342e+02, percent-clipped=7.0 2023-04-30 11:25:58,596 INFO [train.py:904] (7/8) Epoch 17, batch 1250, loss[loss=0.1404, simple_loss=0.226, pruned_loss=0.02737, over 16756.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2509, pruned_loss=0.0414, over 3315027.39 frames. ], batch size: 39, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:27:06,157 INFO [train.py:904] (7/8) Epoch 17, batch 1300, loss[loss=0.167, simple_loss=0.2531, pruned_loss=0.04043, over 17171.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.251, pruned_loss=0.04136, over 3319265.24 frames. ], batch size: 45, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:27:45,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.213e+02 2.483e+02 3.083e+02 7.276e+02, threshold=4.967e+02, percent-clipped=1.0 2023-04-30 11:28:08,421 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163746.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:28:16,452 INFO [train.py:904] (7/8) Epoch 17, batch 1350, loss[loss=0.1592, simple_loss=0.2462, pruned_loss=0.03606, over 16877.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2517, pruned_loss=0.04163, over 3318265.98 frames. ], batch size: 42, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:28:36,315 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4257, 5.3855, 5.2950, 4.7766, 4.8508, 5.3349, 5.2968, 4.9376], device='cuda:7'), covar=tensor([0.0571, 0.0527, 0.0279, 0.0320, 0.1093, 0.0463, 0.0277, 0.0730], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0393, 0.0329, 0.0320, 0.0341, 0.0370, 0.0226, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:29:03,825 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6566, 4.7430, 4.9212, 4.7572, 4.7287, 5.3617, 4.8429, 4.5432], device='cuda:7'), covar=tensor([0.1555, 0.2198, 0.2527, 0.2285, 0.2828, 0.1204, 0.1798, 0.2804], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0569, 0.0627, 0.0479, 0.0642, 0.0664, 0.0494, 0.0640], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:29:15,217 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163794.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:29:25,105 INFO [train.py:904] (7/8) Epoch 17, batch 1400, loss[loss=0.1443, simple_loss=0.2213, pruned_loss=0.03364, over 16771.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.252, pruned_loss=0.04163, over 3322068.96 frames. ], batch size: 39, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:29:52,495 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6714, 2.7267, 2.4181, 2.6214, 2.9907, 2.8362, 3.3472, 3.2123], device='cuda:7'), covar=tensor([0.0141, 0.0352, 0.0405, 0.0381, 0.0258, 0.0322, 0.0234, 0.0249], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0229, 0.0220, 0.0219, 0.0228, 0.0229, 0.0233, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:30:05,131 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.111e+02 2.569e+02 3.362e+02 6.310e+02, threshold=5.138e+02, percent-clipped=5.0 2023-04-30 11:30:06,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1020, 3.9298, 4.2830, 2.1465, 4.4177, 4.5455, 3.2258, 3.5546], device='cuda:7'), covar=tensor([0.0631, 0.0226, 0.0199, 0.1179, 0.0091, 0.0164, 0.0421, 0.0361], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0108, 0.0095, 0.0140, 0.0076, 0.0121, 0.0128, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:30:36,599 INFO [train.py:904] (7/8) Epoch 17, batch 1450, loss[loss=0.1483, simple_loss=0.2461, pruned_loss=0.0252, over 17049.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2516, pruned_loss=0.0414, over 3316623.43 frames. ], batch size: 50, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:31:38,066 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=163896.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:31:45,959 INFO [train.py:904] (7/8) Epoch 17, batch 1500, loss[loss=0.1742, simple_loss=0.2535, pruned_loss=0.04745, over 16918.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2514, pruned_loss=0.0422, over 3321734.15 frames. ], batch size: 90, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:32:24,623 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 2.188e+02 2.579e+02 3.200e+02 4.820e+02, threshold=5.157e+02, percent-clipped=0.0 2023-04-30 11:32:45,103 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=163944.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:32:56,217 INFO [train.py:904] (7/8) Epoch 17, batch 1550, loss[loss=0.1568, simple_loss=0.2473, pruned_loss=0.03322, over 17226.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2529, pruned_loss=0.04321, over 3316212.82 frames. ], batch size: 45, lr: 4.04e-03, grad_scale: 8.0 2023-04-30 11:33:04,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2128, 4.2324, 4.6235, 4.6074, 4.6331, 4.3197, 4.3760, 4.2019], device='cuda:7'), covar=tensor([0.0410, 0.0755, 0.0449, 0.0449, 0.0441, 0.0436, 0.0798, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0392, 0.0425, 0.0415, 0.0388, 0.0460, 0.0436, 0.0532, 0.0345], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:33:42,416 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-04-30 11:34:07,075 INFO [train.py:904] (7/8) Epoch 17, batch 1600, loss[loss=0.1761, simple_loss=0.2511, pruned_loss=0.05059, over 16850.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2546, pruned_loss=0.04472, over 3309639.11 frames. ], batch size: 96, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:34:45,112 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.274e+02 2.815e+02 3.289e+02 5.501e+02, threshold=5.629e+02, percent-clipped=1.0 2023-04-30 11:35:10,859 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1085, 1.9412, 1.5087, 1.6190, 2.1576, 1.9256, 2.1381, 2.2959], device='cuda:7'), covar=tensor([0.0266, 0.0445, 0.0591, 0.0510, 0.0274, 0.0397, 0.0257, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0232, 0.0222, 0.0222, 0.0231, 0.0232, 0.0237, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:35:15,612 INFO [train.py:904] (7/8) Epoch 17, batch 1650, loss[loss=0.1831, simple_loss=0.2562, pruned_loss=0.05502, over 16867.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2569, pruned_loss=0.04596, over 3310947.96 frames. ], batch size: 109, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:35:39,320 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0993, 4.6218, 4.6776, 3.3913, 3.9129, 4.5971, 4.1243, 2.7519], device='cuda:7'), covar=tensor([0.0397, 0.0063, 0.0032, 0.0297, 0.0107, 0.0082, 0.0078, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0079, 0.0078, 0.0133, 0.0091, 0.0102, 0.0090, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:36:08,354 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:36:24,912 INFO [train.py:904] (7/8) Epoch 17, batch 1700, loss[loss=0.1816, simple_loss=0.2667, pruned_loss=0.04825, over 16695.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2588, pruned_loss=0.04631, over 3310996.41 frames. ], batch size: 89, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:36:40,703 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2028, 4.2234, 4.6305, 4.5684, 4.6090, 4.2842, 4.3164, 4.1639], device='cuda:7'), covar=tensor([0.0393, 0.0623, 0.0372, 0.0457, 0.0515, 0.0469, 0.0848, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0425, 0.0415, 0.0390, 0.0462, 0.0438, 0.0535, 0.0346], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:36:50,151 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4845, 5.3280, 5.3628, 4.8877, 4.9651, 5.3222, 5.3528, 5.0132], device='cuda:7'), covar=tensor([0.0558, 0.0526, 0.0276, 0.0337, 0.1089, 0.0468, 0.0304, 0.0715], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0400, 0.0335, 0.0328, 0.0350, 0.0378, 0.0230, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:37:01,951 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.646e+02 2.209e+02 2.796e+02 3.472e+02 6.333e+02, threshold=5.591e+02, percent-clipped=2.0 2023-04-30 11:37:02,655 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8270, 4.2900, 3.2138, 2.2818, 2.7373, 2.7182, 4.6393, 3.7173], device='cuda:7'), covar=tensor([0.2769, 0.0546, 0.1688, 0.2798, 0.2818, 0.1821, 0.0372, 0.1168], device='cuda:7'), in_proj_covar=tensor([0.0317, 0.0262, 0.0296, 0.0296, 0.0287, 0.0241, 0.0281, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:37:06,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2580, 3.1545, 3.3279, 1.8507, 3.4731, 3.4623, 2.8184, 2.6571], device='cuda:7'), covar=tensor([0.0770, 0.0216, 0.0188, 0.1131, 0.0094, 0.0199, 0.0436, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0107, 0.0094, 0.0139, 0.0075, 0.0121, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:37:13,473 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8323, 4.0354, 4.2978, 2.0970, 4.5111, 4.5784, 3.2125, 3.5025], device='cuda:7'), covar=tensor([0.0801, 0.0218, 0.0204, 0.1209, 0.0070, 0.0167, 0.0457, 0.0385], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0107, 0.0094, 0.0139, 0.0075, 0.0121, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:37:32,302 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164151.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:37:32,929 INFO [train.py:904] (7/8) Epoch 17, batch 1750, loss[loss=0.173, simple_loss=0.2601, pruned_loss=0.04296, over 16836.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2589, pruned_loss=0.04576, over 3308194.26 frames. ], batch size: 83, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:37:36,751 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164155.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:38:05,317 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 11:38:38,293 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8457, 5.1991, 4.9191, 4.9788, 4.7140, 4.6620, 4.6846, 5.2492], device='cuda:7'), covar=tensor([0.1197, 0.0897, 0.1092, 0.0799, 0.0814, 0.1012, 0.1207, 0.0948], device='cuda:7'), in_proj_covar=tensor([0.0646, 0.0792, 0.0647, 0.0588, 0.0503, 0.0506, 0.0660, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:38:41,941 INFO [train.py:904] (7/8) Epoch 17, batch 1800, loss[loss=0.1578, simple_loss=0.2528, pruned_loss=0.03139, over 17143.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2605, pruned_loss=0.04558, over 3309737.02 frames. ], batch size: 48, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:38:58,201 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2040, 3.2401, 3.5039, 2.3263, 3.1498, 3.5311, 3.2706, 1.9979], device='cuda:7'), covar=tensor([0.0472, 0.0141, 0.0051, 0.0371, 0.0104, 0.0083, 0.0093, 0.0445], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0078, 0.0077, 0.0132, 0.0091, 0.0101, 0.0089, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:39:00,533 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6024, 3.7271, 2.2264, 3.9562, 2.7540, 3.9181, 2.2978, 2.9839], device='cuda:7'), covar=tensor([0.0264, 0.0344, 0.1433, 0.0320, 0.0789, 0.0754, 0.1291, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0175, 0.0195, 0.0157, 0.0175, 0.0218, 0.0204, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:39:01,684 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:39:19,769 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.179e+02 2.567e+02 3.142e+02 4.810e+02, threshold=5.135e+02, percent-clipped=0.0 2023-04-30 11:39:49,969 INFO [train.py:904] (7/8) Epoch 17, batch 1850, loss[loss=0.1822, simple_loss=0.2752, pruned_loss=0.04456, over 16550.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2618, pruned_loss=0.04558, over 3311475.99 frames. ], batch size: 62, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:41:01,376 INFO [train.py:904] (7/8) Epoch 17, batch 1900, loss[loss=0.1588, simple_loss=0.2404, pruned_loss=0.03864, over 16838.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2612, pruned_loss=0.04515, over 3307703.67 frames. ], batch size: 102, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:41:26,312 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7467, 2.6680, 2.2384, 2.6520, 3.0285, 2.8113, 3.4324, 3.2074], device='cuda:7'), covar=tensor([0.0126, 0.0394, 0.0487, 0.0407, 0.0277, 0.0361, 0.0223, 0.0261], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0231, 0.0222, 0.0221, 0.0230, 0.0232, 0.0236, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:41:40,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7135, 2.3199, 1.7168, 2.0498, 2.7701, 2.4859, 2.8082, 2.8651], device='cuda:7'), covar=tensor([0.0191, 0.0386, 0.0569, 0.0449, 0.0209, 0.0336, 0.0217, 0.0274], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0230, 0.0221, 0.0221, 0.0230, 0.0231, 0.0236, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:41:41,206 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.257e+02 2.647e+02 3.150e+02 4.785e+02, threshold=5.294e+02, percent-clipped=0.0 2023-04-30 11:42:09,129 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0672, 2.4502, 1.9144, 2.2689, 2.9427, 2.6492, 2.9704, 3.0264], device='cuda:7'), covar=tensor([0.0207, 0.0399, 0.0535, 0.0430, 0.0217, 0.0342, 0.0228, 0.0249], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0230, 0.0221, 0.0221, 0.0230, 0.0231, 0.0236, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:42:12,326 INFO [train.py:904] (7/8) Epoch 17, batch 1950, loss[loss=0.1447, simple_loss=0.2327, pruned_loss=0.02837, over 17220.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2614, pruned_loss=0.04438, over 3307201.09 frames. ], batch size: 44, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:42:15,542 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2493, 5.9126, 5.9878, 5.7199, 5.7476, 6.2972, 5.9105, 5.6298], device='cuda:7'), covar=tensor([0.0876, 0.1989, 0.2288, 0.2085, 0.2721, 0.1091, 0.1384, 0.2449], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0566, 0.0626, 0.0478, 0.0646, 0.0661, 0.0494, 0.0640], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:42:24,833 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164360.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:43:00,396 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5215, 2.1988, 2.2503, 4.3517, 2.1734, 2.6663, 2.3000, 2.4141], device='cuda:7'), covar=tensor([0.1126, 0.3810, 0.3008, 0.0469, 0.4278, 0.2627, 0.3472, 0.3720], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0427, 0.0359, 0.0327, 0.0432, 0.0494, 0.0397, 0.0501], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:43:23,961 INFO [train.py:904] (7/8) Epoch 17, batch 2000, loss[loss=0.1629, simple_loss=0.2462, pruned_loss=0.03975, over 16699.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2604, pruned_loss=0.04367, over 3308005.81 frames. ], batch size: 89, lr: 4.03e-03, grad_scale: 8.0 2023-04-30 11:43:26,746 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 11:43:51,080 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164421.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:44:02,063 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.498e+02 2.950e+02 3.427e+02 6.648e+02, threshold=5.900e+02, percent-clipped=3.0 2023-04-30 11:44:25,541 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164446.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:44:32,481 INFO [train.py:904] (7/8) Epoch 17, batch 2050, loss[loss=0.1718, simple_loss=0.2653, pruned_loss=0.03919, over 17055.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2602, pruned_loss=0.04371, over 3304320.27 frames. ], batch size: 50, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:44:47,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8463, 4.0383, 2.6440, 4.7296, 3.1490, 4.6746, 2.7374, 3.3974], device='cuda:7'), covar=tensor([0.0294, 0.0357, 0.1439, 0.0227, 0.0757, 0.0411, 0.1328, 0.0609], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0175, 0.0195, 0.0157, 0.0176, 0.0218, 0.0204, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:44:53,412 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5364, 2.3151, 2.3115, 4.5105, 2.2239, 2.7364, 2.3731, 2.5061], device='cuda:7'), covar=tensor([0.1170, 0.3671, 0.2935, 0.0413, 0.4126, 0.2583, 0.3539, 0.3472], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0428, 0.0358, 0.0326, 0.0431, 0.0493, 0.0397, 0.0500], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:45:07,640 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.39 vs. limit=5.0 2023-04-30 11:45:41,566 INFO [train.py:904] (7/8) Epoch 17, batch 2100, loss[loss=0.1806, simple_loss=0.2738, pruned_loss=0.04368, over 16694.00 frames. ], tot_loss[loss=0.174, simple_loss=0.26, pruned_loss=0.04399, over 3307783.42 frames. ], batch size: 76, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:45:55,658 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164511.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:46:13,735 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0642, 5.5223, 5.6815, 5.4277, 5.4617, 6.0908, 5.5614, 5.3093], device='cuda:7'), covar=tensor([0.0948, 0.1842, 0.2447, 0.1936, 0.2825, 0.0968, 0.1398, 0.2231], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0570, 0.0630, 0.0483, 0.0651, 0.0665, 0.0497, 0.0646], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:46:20,649 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.320e+02 2.783e+02 3.356e+02 6.919e+02, threshold=5.567e+02, percent-clipped=1.0 2023-04-30 11:46:50,967 INFO [train.py:904] (7/8) Epoch 17, batch 2150, loss[loss=0.161, simple_loss=0.252, pruned_loss=0.035, over 17046.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2601, pruned_loss=0.04391, over 3317114.67 frames. ], batch size: 50, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:47:13,634 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164568.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:47:27,048 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164578.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:47:58,411 INFO [train.py:904] (7/8) Epoch 17, batch 2200, loss[loss=0.1878, simple_loss=0.2813, pruned_loss=0.04717, over 16665.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2609, pruned_loss=0.04429, over 3317656.62 frames. ], batch size: 62, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:48:36,618 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164629.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:48:37,258 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.246e+02 2.692e+02 3.358e+02 7.856e+02, threshold=5.383e+02, percent-clipped=4.0 2023-04-30 11:48:44,248 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9760, 4.4660, 4.5264, 3.3242, 3.6974, 4.4959, 4.0612, 2.7400], device='cuda:7'), covar=tensor([0.0402, 0.0061, 0.0031, 0.0292, 0.0112, 0.0076, 0.0077, 0.0379], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0077, 0.0077, 0.0132, 0.0091, 0.0101, 0.0089, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:48:44,525 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-04-30 11:48:50,046 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164639.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 11:48:51,363 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.49 vs. limit=2.0 2023-04-30 11:48:58,173 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1348, 2.9062, 3.1223, 1.8153, 3.2870, 3.1876, 2.6230, 2.5200], device='cuda:7'), covar=tensor([0.0812, 0.0254, 0.0234, 0.1097, 0.0108, 0.0326, 0.0524, 0.0448], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0108, 0.0095, 0.0139, 0.0075, 0.0122, 0.0127, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:49:02,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 11:49:06,809 INFO [train.py:904] (7/8) Epoch 17, batch 2250, loss[loss=0.1751, simple_loss=0.2677, pruned_loss=0.04122, over 17045.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2623, pruned_loss=0.04513, over 3317372.81 frames. ], batch size: 55, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:49:15,683 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164658.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:50:15,843 INFO [train.py:904] (7/8) Epoch 17, batch 2300, loss[loss=0.1714, simple_loss=0.2671, pruned_loss=0.03782, over 17103.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2631, pruned_loss=0.04548, over 3320193.64 frames. ], batch size: 49, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:50:35,313 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164716.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:50:38,830 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164719.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:50:53,239 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.267e+02 2.732e+02 3.248e+02 6.270e+02, threshold=5.465e+02, percent-clipped=1.0 2023-04-30 11:51:16,719 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=164746.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:51:16,844 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=164746.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:51:24,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8019, 2.0554, 2.2794, 3.1334, 2.0945, 2.1872, 2.2142, 2.1079], device='cuda:7'), covar=tensor([0.1316, 0.3328, 0.2530, 0.0715, 0.3993, 0.2578, 0.3172, 0.3422], device='cuda:7'), in_proj_covar=tensor([0.0390, 0.0429, 0.0359, 0.0327, 0.0431, 0.0496, 0.0398, 0.0501], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:51:24,576 INFO [train.py:904] (7/8) Epoch 17, batch 2350, loss[loss=0.1972, simple_loss=0.2884, pruned_loss=0.05296, over 11987.00 frames. ], tot_loss[loss=0.178, simple_loss=0.264, pruned_loss=0.04597, over 3318822.15 frames. ], batch size: 246, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:52:23,527 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=164794.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:52:24,325 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 11:52:34,384 INFO [train.py:904] (7/8) Epoch 17, batch 2400, loss[loss=0.1707, simple_loss=0.2578, pruned_loss=0.04177, over 16285.00 frames. ], tot_loss[loss=0.1788, simple_loss=0.2651, pruned_loss=0.04628, over 3319546.00 frames. ], batch size: 36, lr: 4.03e-03, grad_scale: 16.0 2023-04-30 11:52:41,615 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=164807.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:52:46,695 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=164811.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:53:12,673 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.216e+02 2.538e+02 3.112e+02 6.074e+02, threshold=5.077e+02, percent-clipped=1.0 2023-04-30 11:53:15,565 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9617, 3.2310, 2.9610, 5.1643, 4.3502, 4.5084, 1.7410, 3.3875], device='cuda:7'), covar=tensor([0.1234, 0.0676, 0.1014, 0.0159, 0.0188, 0.0358, 0.1541, 0.0685], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0164, 0.0185, 0.0172, 0.0198, 0.0210, 0.0190, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:53:19,620 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7552, 3.8866, 2.3140, 4.4998, 3.0217, 4.4727, 2.4531, 3.1539], device='cuda:7'), covar=tensor([0.0292, 0.0351, 0.1644, 0.0207, 0.0801, 0.0455, 0.1498, 0.0756], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0174, 0.0193, 0.0157, 0.0173, 0.0217, 0.0202, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:53:41,596 INFO [train.py:904] (7/8) Epoch 17, batch 2450, loss[loss=0.1814, simple_loss=0.2801, pruned_loss=0.04136, over 17072.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2648, pruned_loss=0.04547, over 3331653.19 frames. ], batch size: 53, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:53:47,280 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0517, 5.3999, 5.1286, 5.1558, 4.8722, 4.7955, 4.8439, 5.4965], device='cuda:7'), covar=tensor([0.1117, 0.0794, 0.0937, 0.0786, 0.0843, 0.0985, 0.1057, 0.0899], device='cuda:7'), in_proj_covar=tensor([0.0647, 0.0797, 0.0649, 0.0593, 0.0506, 0.0509, 0.0662, 0.0618], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:53:51,227 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=164859.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:54:46,719 INFO [train.py:904] (7/8) Epoch 17, batch 2500, loss[loss=0.1612, simple_loss=0.2632, pruned_loss=0.0296, over 17125.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2643, pruned_loss=0.04547, over 3330559.07 frames. ], batch size: 49, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:55:12,212 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1239, 2.1106, 2.2280, 3.8250, 2.0763, 2.4025, 2.1548, 2.2548], device='cuda:7'), covar=tensor([0.1321, 0.3561, 0.2695, 0.0575, 0.3790, 0.2606, 0.3683, 0.3035], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0427, 0.0357, 0.0326, 0.0429, 0.0494, 0.0396, 0.0500], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:55:17,578 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164924.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:55:26,644 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.192e+02 2.472e+02 2.956e+02 6.595e+02, threshold=4.945e+02, percent-clipped=3.0 2023-04-30 11:55:29,666 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 11:55:30,903 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=164934.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 11:55:55,242 INFO [train.py:904] (7/8) Epoch 17, batch 2550, loss[loss=0.1779, simple_loss=0.2616, pruned_loss=0.04716, over 16810.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2641, pruned_loss=0.04555, over 3324025.27 frames. ], batch size: 83, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:57:02,087 INFO [train.py:904] (7/8) Epoch 17, batch 2600, loss[loss=0.1708, simple_loss=0.252, pruned_loss=0.04482, over 16861.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2639, pruned_loss=0.04536, over 3328519.02 frames. ], batch size: 96, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:57:17,931 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165014.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:57:20,617 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165016.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:57:32,425 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7847, 3.0811, 2.7785, 5.0573, 4.1002, 4.5462, 1.6771, 3.2588], device='cuda:7'), covar=tensor([0.1345, 0.0698, 0.1180, 0.0171, 0.0237, 0.0365, 0.1534, 0.0725], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0166, 0.0187, 0.0174, 0.0199, 0.0212, 0.0192, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 11:57:41,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.187e+02 2.619e+02 3.229e+02 7.768e+02, threshold=5.237e+02, percent-clipped=3.0 2023-04-30 11:57:43,706 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4008, 5.7663, 5.4364, 5.5926, 5.1902, 5.1701, 5.1385, 5.8904], device='cuda:7'), covar=tensor([0.1304, 0.0974, 0.1133, 0.0852, 0.0981, 0.0756, 0.1201, 0.0907], device='cuda:7'), in_proj_covar=tensor([0.0647, 0.0800, 0.0651, 0.0594, 0.0508, 0.0509, 0.0663, 0.0618], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 11:58:08,463 INFO [train.py:904] (7/8) Epoch 17, batch 2650, loss[loss=0.1867, simple_loss=0.271, pruned_loss=0.05117, over 16493.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2639, pruned_loss=0.04485, over 3327498.76 frames. ], batch size: 68, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:58:20,507 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0286, 3.8444, 4.3596, 1.9911, 4.6084, 4.6233, 3.2454, 3.5835], device='cuda:7'), covar=tensor([0.0647, 0.0236, 0.0201, 0.1172, 0.0053, 0.0156, 0.0413, 0.0339], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0095, 0.0138, 0.0075, 0.0121, 0.0127, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:58:26,319 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165064.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:58:27,833 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9933, 4.4633, 3.1819, 2.3860, 2.9199, 2.6971, 4.8407, 3.7590], device='cuda:7'), covar=tensor([0.2585, 0.0510, 0.1626, 0.2764, 0.2609, 0.1903, 0.0300, 0.1231], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0264, 0.0298, 0.0300, 0.0290, 0.0244, 0.0283, 0.0325], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:58:57,878 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4229, 3.5494, 3.8163, 2.6882, 3.4668, 3.8984, 3.6785, 2.2934], device='cuda:7'), covar=tensor([0.0484, 0.0202, 0.0054, 0.0363, 0.0095, 0.0084, 0.0075, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0076, 0.0076, 0.0129, 0.0090, 0.0099, 0.0088, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 11:59:18,037 INFO [train.py:904] (7/8) Epoch 17, batch 2700, loss[loss=0.1806, simple_loss=0.2793, pruned_loss=0.04091, over 16622.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2647, pruned_loss=0.04507, over 3327599.20 frames. ], batch size: 57, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 11:59:18,309 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165102.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:59:21,124 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 11:59:48,037 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165123.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 11:59:55,434 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0752, 3.1551, 3.4448, 1.6550, 3.5706, 3.6451, 2.8656, 2.7292], device='cuda:7'), covar=tensor([0.1097, 0.0228, 0.0183, 0.1335, 0.0103, 0.0192, 0.0439, 0.0509], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0108, 0.0095, 0.0139, 0.0075, 0.0122, 0.0127, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 11:59:59,034 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.045e+02 2.516e+02 2.952e+02 6.082e+02, threshold=5.032e+02, percent-clipped=3.0 2023-04-30 12:00:28,483 INFO [train.py:904] (7/8) Epoch 17, batch 2750, loss[loss=0.1674, simple_loss=0.2521, pruned_loss=0.04141, over 16839.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2647, pruned_loss=0.0447, over 3326983.19 frames. ], batch size: 42, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:01:13,232 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165184.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:01:38,396 INFO [train.py:904] (7/8) Epoch 17, batch 2800, loss[loss=0.1876, simple_loss=0.2635, pruned_loss=0.05584, over 16743.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2645, pruned_loss=0.04443, over 3332828.46 frames. ], batch size: 124, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:02:10,325 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165224.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:02:19,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.251e+02 2.601e+02 3.544e+02 6.990e+02, threshold=5.203e+02, percent-clipped=6.0 2023-04-30 12:02:24,231 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165234.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 12:02:48,658 INFO [train.py:904] (7/8) Epoch 17, batch 2850, loss[loss=0.1498, simple_loss=0.2318, pruned_loss=0.03391, over 16778.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2637, pruned_loss=0.04397, over 3334911.10 frames. ], batch size: 102, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:03:18,271 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165272.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:03:18,387 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1729, 4.0263, 4.2347, 4.3545, 4.4350, 4.0461, 4.1767, 4.4379], device='cuda:7'), covar=tensor([0.1424, 0.1072, 0.1222, 0.0661, 0.0610, 0.1271, 0.2184, 0.0689], device='cuda:7'), in_proj_covar=tensor([0.0633, 0.0779, 0.0930, 0.0789, 0.0593, 0.0629, 0.0633, 0.0738], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:03:31,475 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:03:34,713 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2830, 5.8887, 6.0235, 5.6342, 5.9007, 6.3476, 5.8020, 5.5753], device='cuda:7'), covar=tensor([0.0839, 0.1973, 0.2185, 0.2103, 0.2343, 0.0969, 0.1490, 0.2297], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0572, 0.0631, 0.0485, 0.0650, 0.0666, 0.0500, 0.0650], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:03:47,585 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9943, 5.3919, 5.5287, 5.2074, 5.3230, 5.9203, 5.3176, 5.1361], device='cuda:7'), covar=tensor([0.1018, 0.1978, 0.2217, 0.2101, 0.2588, 0.0995, 0.1420, 0.2331], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0571, 0.0631, 0.0484, 0.0649, 0.0665, 0.0499, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:03:58,725 INFO [train.py:904] (7/8) Epoch 17, batch 2900, loss[loss=0.1611, simple_loss=0.2605, pruned_loss=0.03086, over 17259.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2626, pruned_loss=0.04472, over 3326404.99 frames. ], batch size: 52, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:04:17,085 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165314.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:04:39,536 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.311e+02 2.666e+02 3.214e+02 5.469e+02, threshold=5.331e+02, percent-clipped=1.0 2023-04-30 12:05:09,252 INFO [train.py:904] (7/8) Epoch 17, batch 2950, loss[loss=0.1929, simple_loss=0.266, pruned_loss=0.05986, over 16886.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2621, pruned_loss=0.0447, over 3332201.62 frames. ], batch size: 116, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:05:19,747 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1768, 5.6666, 5.8051, 5.5199, 5.5769, 6.1720, 5.6898, 5.4073], device='cuda:7'), covar=tensor([0.0968, 0.1957, 0.2453, 0.2143, 0.2829, 0.1027, 0.1396, 0.2555], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0575, 0.0636, 0.0489, 0.0654, 0.0671, 0.0502, 0.0654], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:05:23,902 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165362.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:06:20,624 INFO [train.py:904] (7/8) Epoch 17, batch 3000, loss[loss=0.1738, simple_loss=0.2716, pruned_loss=0.03799, over 17114.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2623, pruned_loss=0.04499, over 3325707.93 frames. ], batch size: 49, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:06:20,625 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 12:06:29,129 INFO [train.py:938] (7/8) Epoch 17, validation: loss=0.1364, simple_loss=0.2422, pruned_loss=0.0153, over 944034.00 frames. 2023-04-30 12:06:29,129 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 12:06:29,562 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165402.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:07:09,777 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.715e+02 2.395e+02 2.828e+02 3.269e+02 8.319e+02, threshold=5.656e+02, percent-clipped=3.0 2023-04-30 12:07:23,776 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5446, 3.6764, 3.8916, 2.3285, 3.2112, 2.5085, 4.0894, 3.7715], device='cuda:7'), covar=tensor([0.0268, 0.0916, 0.0539, 0.1846, 0.0809, 0.0985, 0.0584, 0.1100], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0158, 0.0163, 0.0149, 0.0140, 0.0126, 0.0140, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:07:36,652 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165450.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:07:38,658 INFO [train.py:904] (7/8) Epoch 17, batch 3050, loss[loss=0.171, simple_loss=0.2567, pruned_loss=0.04271, over 16055.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2613, pruned_loss=0.04489, over 3322311.87 frames. ], batch size: 35, lr: 4.02e-03, grad_scale: 4.0 2023-04-30 12:08:12,722 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 12:08:15,875 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165479.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:08:45,841 INFO [train.py:904] (7/8) Epoch 17, batch 3100, loss[loss=0.2005, simple_loss=0.2909, pruned_loss=0.055, over 17068.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2621, pruned_loss=0.04568, over 3324187.45 frames. ], batch size: 53, lr: 4.02e-03, grad_scale: 4.0 2023-04-30 12:09:27,662 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.216e+02 2.589e+02 3.162e+02 6.656e+02, threshold=5.179e+02, percent-clipped=4.0 2023-04-30 12:09:54,971 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165551.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:09:55,719 INFO [train.py:904] (7/8) Epoch 17, batch 3150, loss[loss=0.194, simple_loss=0.2669, pruned_loss=0.0605, over 16936.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2612, pruned_loss=0.04532, over 3326384.63 frames. ], batch size: 109, lr: 4.02e-03, grad_scale: 4.0 2023-04-30 12:10:17,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-04-30 12:11:06,283 INFO [train.py:904] (7/8) Epoch 17, batch 3200, loss[loss=0.1625, simple_loss=0.2541, pruned_loss=0.03542, over 17222.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2606, pruned_loss=0.04481, over 3327801.32 frames. ], batch size: 45, lr: 4.02e-03, grad_scale: 8.0 2023-04-30 12:11:21,049 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165612.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:11:23,511 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6627, 2.6350, 2.1987, 2.3985, 2.9277, 2.7322, 3.3768, 3.2123], device='cuda:7'), covar=tensor([0.0126, 0.0361, 0.0468, 0.0425, 0.0282, 0.0354, 0.0212, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0231, 0.0221, 0.0222, 0.0233, 0.0232, 0.0238, 0.0228], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:11:49,641 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.164e+02 2.494e+02 3.096e+02 4.677e+02, threshold=4.988e+02, percent-clipped=0.0 2023-04-30 12:12:15,443 INFO [train.py:904] (7/8) Epoch 17, batch 3250, loss[loss=0.1484, simple_loss=0.2338, pruned_loss=0.03155, over 16985.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2605, pruned_loss=0.04537, over 3330855.66 frames. ], batch size: 41, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:12:49,721 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-30 12:13:23,330 INFO [train.py:904] (7/8) Epoch 17, batch 3300, loss[loss=0.1906, simple_loss=0.2698, pruned_loss=0.05572, over 16258.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2604, pruned_loss=0.04481, over 3329717.10 frames. ], batch size: 165, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:13:36,042 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7845, 5.0936, 4.8422, 4.8728, 4.6142, 4.6273, 4.6278, 5.1805], device='cuda:7'), covar=tensor([0.1123, 0.0814, 0.0948, 0.0811, 0.0887, 0.1062, 0.1153, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0655, 0.0811, 0.0658, 0.0599, 0.0513, 0.0512, 0.0672, 0.0623], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:14:06,780 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.306e+02 2.673e+02 3.210e+02 6.560e+02, threshold=5.346e+02, percent-clipped=3.0 2023-04-30 12:14:30,239 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165750.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:14:32,753 INFO [train.py:904] (7/8) Epoch 17, batch 3350, loss[loss=0.1541, simple_loss=0.2512, pruned_loss=0.02852, over 17101.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2616, pruned_loss=0.04505, over 3332589.51 frames. ], batch size: 47, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:15:11,116 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=165779.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:15:42,359 INFO [train.py:904] (7/8) Epoch 17, batch 3400, loss[loss=0.1802, simple_loss=0.2605, pruned_loss=0.05, over 16677.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2617, pruned_loss=0.04453, over 3330643.14 frames. ], batch size: 83, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:15:44,657 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=165803.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:15:56,264 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165811.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:16:19,114 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=165827.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:16:27,738 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.228e+02 2.640e+02 3.185e+02 6.119e+02, threshold=5.280e+02, percent-clipped=1.0 2023-04-30 12:16:54,382 INFO [train.py:904] (7/8) Epoch 17, batch 3450, loss[loss=0.1652, simple_loss=0.2547, pruned_loss=0.03779, over 17213.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2603, pruned_loss=0.04448, over 3328356.66 frames. ], batch size: 45, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:17:11,569 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=165864.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:17:29,658 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7848, 2.8165, 2.6624, 4.7546, 3.8985, 4.2365, 1.5812, 3.1271], device='cuda:7'), covar=tensor([0.1294, 0.0763, 0.1157, 0.0176, 0.0251, 0.0406, 0.1638, 0.0737], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0168, 0.0189, 0.0179, 0.0204, 0.0215, 0.0195, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:17:42,795 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 12:18:05,614 INFO [train.py:904] (7/8) Epoch 17, batch 3500, loss[loss=0.1419, simple_loss=0.2217, pruned_loss=0.03111, over 16849.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2598, pruned_loss=0.04435, over 3330824.40 frames. ], batch size: 96, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:18:13,075 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=165907.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:18:33,859 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5932, 4.6628, 4.9886, 4.9957, 5.0059, 4.7183, 4.6969, 4.5006], device='cuda:7'), covar=tensor([0.0329, 0.0660, 0.0403, 0.0402, 0.0508, 0.0400, 0.0861, 0.0566], device='cuda:7'), in_proj_covar=tensor([0.0392, 0.0429, 0.0418, 0.0394, 0.0466, 0.0443, 0.0540, 0.0350], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:18:49,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.135e+02 2.523e+02 3.067e+02 7.185e+02, threshold=5.046e+02, percent-clipped=1.0 2023-04-30 12:19:07,877 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9909, 3.8234, 4.1738, 2.1339, 4.4384, 4.4390, 3.3150, 3.3735], device='cuda:7'), covar=tensor([0.0653, 0.0211, 0.0209, 0.1080, 0.0062, 0.0168, 0.0368, 0.0372], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0095, 0.0138, 0.0076, 0.0122, 0.0127, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:19:16,047 INFO [train.py:904] (7/8) Epoch 17, batch 3550, loss[loss=0.1882, simple_loss=0.2702, pruned_loss=0.05311, over 11673.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2591, pruned_loss=0.04432, over 3329068.23 frames. ], batch size: 246, lr: 4.01e-03, grad_scale: 4.0 2023-04-30 12:20:27,934 INFO [train.py:904] (7/8) Epoch 17, batch 3600, loss[loss=0.1604, simple_loss=0.2321, pruned_loss=0.04434, over 16876.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2584, pruned_loss=0.04415, over 3327670.82 frames. ], batch size: 90, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:21:12,109 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.087e+02 2.490e+02 2.982e+02 5.501e+02, threshold=4.979e+02, percent-clipped=1.0 2023-04-30 12:21:40,374 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8084, 2.5347, 2.3997, 1.8513, 2.6203, 2.6437, 2.6142, 1.6987], device='cuda:7'), covar=tensor([0.0475, 0.0100, 0.0078, 0.0432, 0.0108, 0.0147, 0.0111, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0077, 0.0077, 0.0131, 0.0091, 0.0101, 0.0088, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:21:40,968 INFO [train.py:904] (7/8) Epoch 17, batch 3650, loss[loss=0.1898, simple_loss=0.2668, pruned_loss=0.05639, over 11499.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2569, pruned_loss=0.04428, over 3311128.00 frames. ], batch size: 248, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:21:44,579 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166054.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:22:55,847 INFO [train.py:904] (7/8) Epoch 17, batch 3700, loss[loss=0.1623, simple_loss=0.2371, pruned_loss=0.04376, over 16873.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2558, pruned_loss=0.04551, over 3282243.46 frames. ], batch size: 102, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:23:02,851 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166106.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:23:16,058 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166115.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:23:42,391 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.852e+02 2.227e+02 2.592e+02 3.005e+02 5.986e+02, threshold=5.184e+02, percent-clipped=3.0 2023-04-30 12:23:43,005 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3737, 2.3337, 2.4417, 4.2321, 2.2603, 2.7240, 2.3908, 2.5609], device='cuda:7'), covar=tensor([0.1277, 0.3369, 0.2512, 0.0476, 0.3624, 0.2273, 0.3366, 0.2834], device='cuda:7'), in_proj_covar=tensor([0.0390, 0.0428, 0.0357, 0.0327, 0.0429, 0.0496, 0.0397, 0.0503], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:24:04,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166147.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:24:10,350 INFO [train.py:904] (7/8) Epoch 17, batch 3750, loss[loss=0.176, simple_loss=0.2509, pruned_loss=0.05056, over 16831.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2563, pruned_loss=0.04699, over 3274015.24 frames. ], batch size: 102, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:24:21,468 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166159.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:25:01,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3164, 3.3799, 3.4430, 2.2537, 3.0143, 2.4252, 3.8534, 3.8638], device='cuda:7'), covar=tensor([0.0200, 0.0829, 0.0583, 0.1708, 0.0775, 0.0931, 0.0392, 0.0625], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0159, 0.0163, 0.0149, 0.0140, 0.0127, 0.0140, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:25:24,423 INFO [train.py:904] (7/8) Epoch 17, batch 3800, loss[loss=0.1686, simple_loss=0.2391, pruned_loss=0.04904, over 16772.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2568, pruned_loss=0.04799, over 3273724.65 frames. ], batch size: 83, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:25:24,997 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6939, 3.8164, 2.3788, 4.1103, 2.9407, 4.1649, 2.4499, 2.9875], device='cuda:7'), covar=tensor([0.0291, 0.0376, 0.1574, 0.0303, 0.0753, 0.0669, 0.1410, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0176, 0.0194, 0.0160, 0.0174, 0.0219, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:25:32,339 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166207.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:25:33,732 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166208.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:25:58,953 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-04-30 12:26:10,621 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.250e+02 2.656e+02 3.137e+02 5.830e+02, threshold=5.312e+02, percent-clipped=1.0 2023-04-30 12:26:11,220 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3972, 3.2743, 3.6063, 1.7684, 3.7284, 3.7022, 2.9423, 2.7167], device='cuda:7'), covar=tensor([0.0747, 0.0233, 0.0165, 0.1189, 0.0092, 0.0179, 0.0411, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0096, 0.0140, 0.0077, 0.0124, 0.0128, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:26:38,516 INFO [train.py:904] (7/8) Epoch 17, batch 3850, loss[loss=0.1896, simple_loss=0.268, pruned_loss=0.05563, over 16910.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2575, pruned_loss=0.04872, over 3273933.66 frames. ], batch size: 109, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:26:44,003 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166255.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:27:52,943 INFO [train.py:904] (7/8) Epoch 17, batch 3900, loss[loss=0.1835, simple_loss=0.255, pruned_loss=0.05595, over 16718.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2567, pruned_loss=0.04946, over 3274538.51 frames. ], batch size: 134, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:28:01,057 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166307.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:28:33,264 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8593, 4.1187, 2.9684, 2.3783, 2.7436, 2.4713, 4.3636, 3.8148], device='cuda:7'), covar=tensor([0.2519, 0.0530, 0.1663, 0.2274, 0.2589, 0.1874, 0.0363, 0.0832], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0264, 0.0297, 0.0300, 0.0292, 0.0244, 0.0283, 0.0325], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:28:37,862 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.433e+02 2.279e+02 2.722e+02 3.230e+02 5.998e+02, threshold=5.443e+02, percent-clipped=1.0 2023-04-30 12:28:49,828 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0273, 2.4816, 2.7054, 1.8224, 2.8006, 2.7942, 2.4124, 2.3340], device='cuda:7'), covar=tensor([0.0732, 0.0281, 0.0210, 0.1025, 0.0109, 0.0261, 0.0459, 0.0447], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0108, 0.0095, 0.0140, 0.0077, 0.0123, 0.0128, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:29:07,067 INFO [train.py:904] (7/8) Epoch 17, batch 3950, loss[loss=0.1578, simple_loss=0.2353, pruned_loss=0.04014, over 16782.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2563, pruned_loss=0.0498, over 3278181.29 frames. ], batch size: 102, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:29:30,726 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166368.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:30:18,535 INFO [train.py:904] (7/8) Epoch 17, batch 4000, loss[loss=0.1613, simple_loss=0.2423, pruned_loss=0.04017, over 16688.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2559, pruned_loss=0.05022, over 3278219.85 frames. ], batch size: 124, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:30:25,201 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166406.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:30:31,179 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166410.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 12:31:03,721 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.351e+02 2.185e+02 2.579e+02 2.981e+02 5.548e+02, threshold=5.157e+02, percent-clipped=1.0 2023-04-30 12:31:31,410 INFO [train.py:904] (7/8) Epoch 17, batch 4050, loss[loss=0.1875, simple_loss=0.2669, pruned_loss=0.0541, over 12415.00 frames. ], tot_loss[loss=0.1773, simple_loss=0.2562, pruned_loss=0.04926, over 3283355.65 frames. ], batch size: 246, lr: 4.01e-03, grad_scale: 8.0 2023-04-30 12:31:34,117 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166454.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:31:42,509 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166459.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:14,639 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:44,115 INFO [train.py:904] (7/8) Epoch 17, batch 4100, loss[loss=0.1993, simple_loss=0.2777, pruned_loss=0.06046, over 12117.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2575, pruned_loss=0.04847, over 3279624.74 frames. ], batch size: 248, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:32:46,315 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166503.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:51,159 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166507.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:32:55,435 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166509.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:33:29,302 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166531.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:33:31,680 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.012e+02 2.257e+02 2.846e+02 6.984e+02, threshold=4.513e+02, percent-clipped=1.0 2023-04-30 12:33:45,793 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166542.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:33:59,695 INFO [train.py:904] (7/8) Epoch 17, batch 4150, loss[loss=0.2006, simple_loss=0.2857, pruned_loss=0.05773, over 16832.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2655, pruned_loss=0.05157, over 3230444.91 frames. ], batch size: 116, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:34:10,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8311, 1.3739, 1.7145, 1.6803, 1.8423, 1.9591, 1.5714, 1.8156], device='cuda:7'), covar=tensor([0.0194, 0.0306, 0.0170, 0.0229, 0.0204, 0.0163, 0.0333, 0.0101], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0187, 0.0174, 0.0178, 0.0187, 0.0144, 0.0188, 0.0139], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:34:28,445 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:35:01,108 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166592.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:35:14,017 INFO [train.py:904] (7/8) Epoch 17, batch 4200, loss[loss=0.2239, simple_loss=0.3156, pruned_loss=0.0661, over 16657.00 frames. ], tot_loss[loss=0.1886, simple_loss=0.2723, pruned_loss=0.05243, over 3225714.54 frames. ], batch size: 57, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:36:00,018 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.403e+02 2.795e+02 3.348e+02 5.640e+02, threshold=5.591e+02, percent-clipped=5.0 2023-04-30 12:36:11,853 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.43 vs. limit=5.0 2023-04-30 12:36:27,723 INFO [train.py:904] (7/8) Epoch 17, batch 4250, loss[loss=0.1626, simple_loss=0.2646, pruned_loss=0.03033, over 16827.00 frames. ], tot_loss[loss=0.1904, simple_loss=0.2758, pruned_loss=0.05248, over 3203329.00 frames. ], batch size: 96, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:36:43,552 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166663.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:37:39,004 INFO [train.py:904] (7/8) Epoch 17, batch 4300, loss[loss=0.2134, simple_loss=0.3002, pruned_loss=0.06327, over 16404.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2765, pruned_loss=0.05113, over 3194682.16 frames. ], batch size: 146, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:37:52,261 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166710.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:38:24,694 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.116e+02 2.589e+02 3.113e+02 5.603e+02, threshold=5.178e+02, percent-clipped=2.0 2023-04-30 12:38:28,934 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166735.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:38:47,764 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.72 vs. limit=5.0 2023-04-30 12:38:52,948 INFO [train.py:904] (7/8) Epoch 17, batch 4350, loss[loss=0.2147, simple_loss=0.3046, pruned_loss=0.0624, over 16775.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2795, pruned_loss=0.05227, over 3188081.73 frames. ], batch size: 124, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:39:02,688 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166758.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 12:39:04,584 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9314, 3.2766, 3.2384, 2.1403, 2.9901, 3.2520, 3.0902, 1.8934], device='cuda:7'), covar=tensor([0.0466, 0.0041, 0.0046, 0.0380, 0.0085, 0.0081, 0.0081, 0.0407], device='cuda:7'), in_proj_covar=tensor([0.0129, 0.0075, 0.0075, 0.0127, 0.0088, 0.0098, 0.0087, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:39:10,379 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166764.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:39:57,562 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166796.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:40:05,345 INFO [train.py:904] (7/8) Epoch 17, batch 4400, loss[loss=0.1893, simple_loss=0.2812, pruned_loss=0.04869, over 17202.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2819, pruned_loss=0.05348, over 3192472.73 frames. ], batch size: 46, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:40:06,981 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166803.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:40:38,778 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166825.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:40:49,148 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.232e+02 2.626e+02 3.012e+02 5.313e+02, threshold=5.252e+02, percent-clipped=1.0 2023-04-30 12:40:53,905 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9845, 5.2764, 5.0827, 5.0914, 4.8385, 4.6054, 4.7058, 5.3806], device='cuda:7'), covar=tensor([0.1103, 0.0757, 0.0835, 0.0828, 0.0718, 0.1005, 0.1072, 0.0790], device='cuda:7'), in_proj_covar=tensor([0.0634, 0.0782, 0.0635, 0.0581, 0.0495, 0.0501, 0.0649, 0.0603], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:40:55,210 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166837.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:41:15,724 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=166851.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:41:16,602 INFO [train.py:904] (7/8) Epoch 17, batch 4450, loss[loss=0.2261, simple_loss=0.2958, pruned_loss=0.07819, over 11898.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2852, pruned_loss=0.05454, over 3206017.77 frames. ], batch size: 247, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:41:22,937 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0549, 2.4649, 2.5959, 1.8546, 2.7333, 2.7971, 2.4191, 2.3698], device='cuda:7'), covar=tensor([0.0731, 0.0248, 0.0225, 0.0928, 0.0093, 0.0216, 0.0461, 0.0423], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0137, 0.0075, 0.0120, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:41:26,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9601, 4.0595, 4.3047, 4.2529, 4.2736, 4.0191, 4.0227, 3.9710], device='cuda:7'), covar=tensor([0.0314, 0.0555, 0.0330, 0.0409, 0.0431, 0.0420, 0.0919, 0.0515], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0410, 0.0398, 0.0377, 0.0447, 0.0422, 0.0518, 0.0334], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:41:36,475 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166865.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:42:07,452 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=166887.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:42:19,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5974, 4.7976, 5.0136, 4.8111, 4.8552, 5.4395, 4.8831, 4.6132], device='cuda:7'), covar=tensor([0.1212, 0.1779, 0.1925, 0.1765, 0.2461, 0.0924, 0.1345, 0.2269], device='cuda:7'), in_proj_covar=tensor([0.0389, 0.0559, 0.0610, 0.0467, 0.0627, 0.0646, 0.0481, 0.0627], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:42:28,864 INFO [train.py:904] (7/8) Epoch 17, batch 4500, loss[loss=0.1979, simple_loss=0.2795, pruned_loss=0.05816, over 12177.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2856, pruned_loss=0.05512, over 3205237.27 frames. ], batch size: 247, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:43:07,250 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=166929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:43:12,782 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 1.901e+02 2.204e+02 2.572e+02 5.344e+02, threshold=4.409e+02, percent-clipped=1.0 2023-04-30 12:43:40,965 INFO [train.py:904] (7/8) Epoch 17, batch 4550, loss[loss=0.1756, simple_loss=0.271, pruned_loss=0.04013, over 16729.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2863, pruned_loss=0.05625, over 3190230.28 frames. ], batch size: 83, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:43:57,358 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=166963.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:44:19,081 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9288, 4.8054, 5.0023, 5.1579, 5.2902, 4.7154, 5.3142, 5.3179], device='cuda:7'), covar=tensor([0.1636, 0.0926, 0.1291, 0.0583, 0.0391, 0.0778, 0.0429, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0599, 0.0738, 0.0875, 0.0751, 0.0565, 0.0601, 0.0595, 0.0699], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:44:27,880 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8666, 3.7035, 4.1021, 1.8258, 4.4194, 4.4079, 3.1907, 3.4002], device='cuda:7'), covar=tensor([0.0747, 0.0225, 0.0191, 0.1349, 0.0062, 0.0097, 0.0401, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0105, 0.0093, 0.0136, 0.0074, 0.0119, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 12:44:35,648 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=166990.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:44:53,180 INFO [train.py:904] (7/8) Epoch 17, batch 4600, loss[loss=0.2062, simple_loss=0.2952, pruned_loss=0.05857, over 16837.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2875, pruned_loss=0.05648, over 3197875.15 frames. ], batch size: 116, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:45:07,124 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167011.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:45:34,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7214, 4.7701, 4.5667, 4.2546, 4.2800, 4.6882, 4.4471, 4.4139], device='cuda:7'), covar=tensor([0.0449, 0.0329, 0.0224, 0.0215, 0.0756, 0.0294, 0.0412, 0.0534], device='cuda:7'), in_proj_covar=tensor([0.0273, 0.0379, 0.0324, 0.0311, 0.0333, 0.0360, 0.0221, 0.0387], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:45:38,258 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.923e+02 2.253e+02 2.642e+02 5.543e+02, threshold=4.506e+02, percent-clipped=2.0 2023-04-30 12:46:05,383 INFO [train.py:904] (7/8) Epoch 17, batch 4650, loss[loss=0.1853, simple_loss=0.2648, pruned_loss=0.05293, over 16972.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2866, pruned_loss=0.05643, over 3210894.30 frames. ], batch size: 55, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:46:33,958 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3901, 2.3608, 2.3541, 4.2130, 2.3083, 2.6447, 2.3987, 2.4625], device='cuda:7'), covar=tensor([0.1134, 0.3063, 0.2516, 0.0413, 0.3623, 0.2295, 0.2965, 0.3123], device='cuda:7'), in_proj_covar=tensor([0.0389, 0.0429, 0.0354, 0.0323, 0.0430, 0.0496, 0.0397, 0.0501], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:47:00,704 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167091.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:13,879 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167100.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:16,423 INFO [train.py:904] (7/8) Epoch 17, batch 4700, loss[loss=0.1808, simple_loss=0.2652, pruned_loss=0.04818, over 16292.00 frames. ], tot_loss[loss=0.1971, simple_loss=0.2837, pruned_loss=0.05527, over 3206092.43 frames. ], batch size: 35, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:47:17,137 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 12:47:18,958 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167104.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:27,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9593, 2.0675, 2.2384, 3.5420, 2.0587, 2.4015, 2.2275, 2.2251], device='cuda:7'), covar=tensor([0.1317, 0.3378, 0.2582, 0.0542, 0.3960, 0.2416, 0.3251, 0.3268], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0429, 0.0354, 0.0323, 0.0429, 0.0496, 0.0396, 0.0500], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:47:42,170 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167120.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:47:57,072 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1104, 5.1530, 4.9901, 4.5849, 4.5419, 5.0438, 4.9039, 4.7593], device='cuda:7'), covar=tensor([0.0516, 0.0436, 0.0234, 0.0240, 0.0984, 0.0456, 0.0268, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0273, 0.0382, 0.0325, 0.0312, 0.0334, 0.0362, 0.0222, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:48:00,932 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 1.982e+02 2.244e+02 2.685e+02 3.929e+02, threshold=4.489e+02, percent-clipped=0.0 2023-04-30 12:48:01,562 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6394, 2.6026, 1.8075, 2.7187, 2.0667, 2.7880, 2.0225, 2.3274], device='cuda:7'), covar=tensor([0.0317, 0.0349, 0.1235, 0.0187, 0.0683, 0.0437, 0.1144, 0.0628], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0170, 0.0190, 0.0150, 0.0169, 0.0210, 0.0197, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 12:48:06,774 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167137.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:28,411 INFO [train.py:904] (7/8) Epoch 17, batch 4750, loss[loss=0.1655, simple_loss=0.2633, pruned_loss=0.03387, over 16869.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2787, pruned_loss=0.05247, over 3220468.12 frames. ], batch size: 102, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:48:42,009 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167161.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:46,248 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0619, 2.9484, 3.1292, 1.6512, 3.2736, 3.3465, 2.6560, 2.5320], device='cuda:7'), covar=tensor([0.0843, 0.0246, 0.0201, 0.1236, 0.0080, 0.0165, 0.0454, 0.0473], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0106, 0.0093, 0.0138, 0.0075, 0.0120, 0.0126, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:48:48,093 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:48:48,143 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:49:16,409 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:49:18,982 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167187.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:49:40,740 INFO [train.py:904] (7/8) Epoch 17, batch 4800, loss[loss=0.2056, simple_loss=0.2844, pruned_loss=0.0634, over 11989.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2757, pruned_loss=0.05082, over 3205849.63 frames. ], batch size: 248, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:49:59,003 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167213.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:50:01,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0017, 5.2736, 5.0667, 5.0535, 4.8490, 4.7207, 4.6918, 5.3679], device='cuda:7'), covar=tensor([0.1129, 0.0772, 0.0873, 0.0755, 0.0732, 0.0891, 0.1079, 0.0887], device='cuda:7'), in_proj_covar=tensor([0.0621, 0.0761, 0.0623, 0.0565, 0.0484, 0.0489, 0.0635, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:50:24,551 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5811, 3.6661, 3.3774, 3.0739, 3.2346, 3.5236, 3.3191, 3.3679], device='cuda:7'), covar=tensor([0.0570, 0.0463, 0.0269, 0.0231, 0.0603, 0.0389, 0.1294, 0.0469], device='cuda:7'), in_proj_covar=tensor([0.0273, 0.0382, 0.0325, 0.0312, 0.0334, 0.0362, 0.0223, 0.0387], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:50:27,972 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.008e+02 2.170e+02 2.594e+02 4.937e+02, threshold=4.340e+02, percent-clipped=1.0 2023-04-30 12:50:32,465 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167235.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:50:58,077 INFO [train.py:904] (7/8) Epoch 17, batch 4850, loss[loss=0.1793, simple_loss=0.2759, pruned_loss=0.04139, over 16191.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.277, pruned_loss=0.05063, over 3167967.44 frames. ], batch size: 165, lr: 4.00e-03, grad_scale: 8.0 2023-04-30 12:51:28,417 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5748, 4.6160, 4.9235, 4.8987, 4.8621, 4.6438, 4.5729, 4.4245], device='cuda:7'), covar=tensor([0.0246, 0.0474, 0.0312, 0.0306, 0.0405, 0.0284, 0.0803, 0.0397], device='cuda:7'), in_proj_covar=tensor([0.0371, 0.0405, 0.0396, 0.0372, 0.0445, 0.0416, 0.0514, 0.0332], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:51:46,726 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167285.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:52:12,190 INFO [train.py:904] (7/8) Epoch 17, batch 4900, loss[loss=0.1881, simple_loss=0.2765, pruned_loss=0.04987, over 16722.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2759, pruned_loss=0.04955, over 3181152.16 frames. ], batch size: 124, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:52:16,008 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4076, 3.2945, 3.7696, 1.7438, 3.8930, 3.9397, 2.9318, 2.6945], device='cuda:7'), covar=tensor([0.0864, 0.0290, 0.0136, 0.1347, 0.0059, 0.0124, 0.0395, 0.0562], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0138, 0.0075, 0.0120, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 12:52:55,966 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 1.938e+02 2.305e+02 2.710e+02 4.454e+02, threshold=4.611e+02, percent-clipped=1.0 2023-04-30 12:53:24,987 INFO [train.py:904] (7/8) Epoch 17, batch 4950, loss[loss=0.1789, simple_loss=0.2704, pruned_loss=0.04364, over 17219.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2754, pruned_loss=0.04902, over 3182803.24 frames. ], batch size: 44, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:54:22,644 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167391.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:54:37,753 INFO [train.py:904] (7/8) Epoch 17, batch 5000, loss[loss=0.1981, simple_loss=0.2869, pruned_loss=0.05466, over 16875.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2768, pruned_loss=0.04905, over 3179375.49 frames. ], batch size: 42, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:55:02,828 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:19,921 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167432.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:20,698 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.024e+02 2.500e+02 2.980e+02 7.048e+02, threshold=5.000e+02, percent-clipped=1.0 2023-04-30 12:55:29,658 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167439.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:46,699 INFO [train.py:904] (7/8) Epoch 17, batch 5050, loss[loss=0.2158, simple_loss=0.2952, pruned_loss=0.06819, over 12107.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2766, pruned_loss=0.04831, over 3194981.25 frames. ], batch size: 247, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:55:51,684 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167456.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:55:57,216 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:56:08,481 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167468.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:56:44,067 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167493.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:56:56,788 INFO [train.py:904] (7/8) Epoch 17, batch 5100, loss[loss=0.1642, simple_loss=0.2492, pruned_loss=0.03963, over 17133.00 frames. ], tot_loss[loss=0.1857, simple_loss=0.2752, pruned_loss=0.04808, over 3189019.78 frames. ], batch size: 49, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:57:32,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9421, 3.2234, 3.2132, 2.0937, 2.9493, 3.1809, 3.0537, 2.0291], device='cuda:7'), covar=tensor([0.0495, 0.0045, 0.0051, 0.0393, 0.0093, 0.0097, 0.0089, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0076, 0.0077, 0.0131, 0.0091, 0.0100, 0.0089, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 12:57:39,789 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.010e+02 2.231e+02 2.542e+02 5.876e+02, threshold=4.463e+02, percent-clipped=1.0 2023-04-30 12:58:08,926 INFO [train.py:904] (7/8) Epoch 17, batch 5150, loss[loss=0.2008, simple_loss=0.2911, pruned_loss=0.05521, over 16892.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2751, pruned_loss=0.04728, over 3181359.37 frames. ], batch size: 109, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:58:31,651 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2023-04-30 12:58:56,779 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167585.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 12:59:03,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2108, 5.1438, 5.0624, 4.3129, 5.0615, 1.8288, 4.8449, 4.9843], device='cuda:7'), covar=tensor([0.0091, 0.0079, 0.0160, 0.0432, 0.0105, 0.2568, 0.0141, 0.0190], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0139, 0.0184, 0.0169, 0.0159, 0.0195, 0.0173, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 12:59:21,214 INFO [train.py:904] (7/8) Epoch 17, batch 5200, loss[loss=0.2147, simple_loss=0.2905, pruned_loss=0.0694, over 12365.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2738, pruned_loss=0.04684, over 3185840.79 frames. ], batch size: 246, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 12:59:38,575 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5015, 2.4695, 2.4132, 4.2689, 2.2747, 2.7510, 2.5108, 2.6552], device='cuda:7'), covar=tensor([0.1179, 0.3233, 0.2508, 0.0445, 0.3661, 0.2433, 0.3118, 0.2674], device='cuda:7'), in_proj_covar=tensor([0.0385, 0.0425, 0.0350, 0.0320, 0.0424, 0.0491, 0.0394, 0.0496], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:00:07,261 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.038e+02 2.260e+02 2.754e+02 5.460e+02, threshold=4.520e+02, percent-clipped=3.0 2023-04-30 13:00:08,172 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167633.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:00:36,329 INFO [train.py:904] (7/8) Epoch 17, batch 5250, loss[loss=0.1835, simple_loss=0.2811, pruned_loss=0.04295, over 16222.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2711, pruned_loss=0.04677, over 3201960.71 frames. ], batch size: 165, lr: 3.99e-03, grad_scale: 16.0 2023-04-30 13:01:48,354 INFO [train.py:904] (7/8) Epoch 17, batch 5300, loss[loss=0.1637, simple_loss=0.2621, pruned_loss=0.03265, over 16731.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2682, pruned_loss=0.04564, over 3215952.24 frames. ], batch size: 89, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:02:09,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1850, 3.8768, 3.8852, 2.5420, 3.3952, 3.8050, 3.5433, 2.0701], device='cuda:7'), covar=tensor([0.0536, 0.0039, 0.0041, 0.0358, 0.0096, 0.0093, 0.0084, 0.0456], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0076, 0.0077, 0.0130, 0.0090, 0.0100, 0.0089, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:02:12,884 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167719.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:02:35,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.142e+02 2.405e+02 3.044e+02 5.207e+02, threshold=4.809e+02, percent-clipped=2.0 2023-04-30 13:03:01,794 INFO [train.py:904] (7/8) Epoch 17, batch 5350, loss[loss=0.1782, simple_loss=0.2682, pruned_loss=0.04408, over 16765.00 frames. ], tot_loss[loss=0.1778, simple_loss=0.2663, pruned_loss=0.04466, over 3209885.71 frames. ], batch size: 83, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:03:08,808 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167756.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:13,829 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=167760.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:23,340 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=167767.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:40,456 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4912, 3.4091, 3.9882, 1.6781, 4.1495, 4.1557, 2.9386, 2.9652], device='cuda:7'), covar=tensor([0.0833, 0.0289, 0.0138, 0.1354, 0.0050, 0.0101, 0.0403, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0093, 0.0138, 0.0075, 0.0118, 0.0125, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 13:03:41,775 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167780.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:03:44,889 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.02 vs. limit=5.0 2023-04-30 13:03:53,661 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=167788.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:04:14,439 INFO [train.py:904] (7/8) Epoch 17, batch 5400, loss[loss=0.2005, simple_loss=0.2871, pruned_loss=0.05695, over 16615.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2685, pruned_loss=0.04528, over 3202917.95 frames. ], batch size: 62, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:04:17,679 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167804.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:04:24,953 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=167808.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:04:54,068 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=167828.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:05:02,697 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.037e+02 2.360e+02 2.749e+02 5.264e+02, threshold=4.720e+02, percent-clipped=1.0 2023-04-30 13:05:31,692 INFO [train.py:904] (7/8) Epoch 17, batch 5450, loss[loss=0.2087, simple_loss=0.2953, pruned_loss=0.06106, over 16705.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.272, pruned_loss=0.04678, over 3192567.39 frames. ], batch size: 134, lr: 3.99e-03, grad_scale: 4.0 2023-04-30 13:06:48,605 INFO [train.py:904] (7/8) Epoch 17, batch 5500, loss[loss=0.2014, simple_loss=0.2952, pruned_loss=0.05385, over 16488.00 frames. ], tot_loss[loss=0.1921, simple_loss=0.2801, pruned_loss=0.05202, over 3159651.45 frames. ], batch size: 75, lr: 3.99e-03, grad_scale: 4.0 2023-04-30 13:07:21,628 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1778, 4.2395, 4.0236, 3.7752, 3.7451, 4.1584, 3.8279, 3.8982], device='cuda:7'), covar=tensor([0.0639, 0.0520, 0.0303, 0.0283, 0.0821, 0.0480, 0.0895, 0.0618], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0388, 0.0328, 0.0316, 0.0337, 0.0370, 0.0223, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:07:39,670 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.878e+02 2.929e+02 3.465e+02 4.184e+02 8.083e+02, threshold=6.929e+02, percent-clipped=17.0 2023-04-30 13:08:05,273 INFO [train.py:904] (7/8) Epoch 17, batch 5550, loss[loss=0.2126, simple_loss=0.3036, pruned_loss=0.06077, over 16873.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2878, pruned_loss=0.05717, over 3143412.70 frames. ], batch size: 96, lr: 3.99e-03, grad_scale: 4.0 2023-04-30 13:08:58,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-30 13:09:31,464 INFO [train.py:904] (7/8) Epoch 17, batch 5600, loss[loss=0.2269, simple_loss=0.311, pruned_loss=0.07146, over 16751.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2934, pruned_loss=0.06186, over 3102089.84 frames. ], batch size: 124, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:09:37,996 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1683, 4.1821, 4.0940, 3.3811, 4.1312, 1.7565, 3.9285, 3.7886], device='cuda:7'), covar=tensor([0.0126, 0.0088, 0.0179, 0.0317, 0.0100, 0.2648, 0.0139, 0.0237], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0139, 0.0185, 0.0170, 0.0160, 0.0196, 0.0174, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:10:05,804 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1762, 5.1772, 5.5854, 5.5847, 5.6320, 5.2724, 5.1754, 4.8993], device='cuda:7'), covar=tensor([0.0317, 0.0476, 0.0299, 0.0376, 0.0464, 0.0358, 0.0968, 0.0501], device='cuda:7'), in_proj_covar=tensor([0.0376, 0.0410, 0.0401, 0.0378, 0.0447, 0.0421, 0.0520, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 13:10:27,366 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.908e+02 3.389e+02 4.058e+02 4.967e+02 1.149e+03, threshold=8.116e+02, percent-clipped=3.0 2023-04-30 13:10:27,969 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168035.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:10:54,426 INFO [train.py:904] (7/8) Epoch 17, batch 5650, loss[loss=0.2669, simple_loss=0.3271, pruned_loss=0.1033, over 11319.00 frames. ], tot_loss[loss=0.2171, simple_loss=0.2994, pruned_loss=0.06738, over 3049392.66 frames. ], batch size: 248, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:11:30,931 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168075.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:11:50,878 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:12:04,710 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=168096.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 13:12:12,314 INFO [train.py:904] (7/8) Epoch 17, batch 5700, loss[loss=0.2093, simple_loss=0.3037, pruned_loss=0.0574, over 16398.00 frames. ], tot_loss[loss=0.2182, simple_loss=0.3003, pruned_loss=0.06807, over 3063994.24 frames. ], batch size: 146, lr: 3.99e-03, grad_scale: 8.0 2023-04-30 13:12:46,686 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168123.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:12:57,782 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-30 13:13:04,295 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.185e+02 3.066e+02 3.948e+02 4.929e+02 1.585e+03, threshold=7.895e+02, percent-clipped=5.0 2023-04-30 13:13:06,079 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168136.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:13:31,691 INFO [train.py:904] (7/8) Epoch 17, batch 5750, loss[loss=0.2157, simple_loss=0.2983, pruned_loss=0.06651, over 16698.00 frames. ], tot_loss[loss=0.2203, simple_loss=0.3024, pruned_loss=0.06911, over 3049700.72 frames. ], batch size: 124, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:14:32,880 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4721, 2.1099, 1.7239, 1.8754, 2.3992, 2.1214, 2.1432, 2.5331], device='cuda:7'), covar=tensor([0.0204, 0.0397, 0.0533, 0.0469, 0.0242, 0.0371, 0.0221, 0.0259], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0222, 0.0215, 0.0216, 0.0224, 0.0223, 0.0224, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:14:50,321 INFO [train.py:904] (7/8) Epoch 17, batch 5800, loss[loss=0.2097, simple_loss=0.2985, pruned_loss=0.06042, over 15232.00 frames. ], tot_loss[loss=0.2193, simple_loss=0.3021, pruned_loss=0.06826, over 3046114.22 frames. ], batch size: 191, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:15:41,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168233.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:15:43,218 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.888e+02 3.447e+02 4.230e+02 9.306e+02, threshold=6.893e+02, percent-clipped=1.0 2023-04-30 13:16:05,065 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-04-30 13:16:09,630 INFO [train.py:904] (7/8) Epoch 17, batch 5850, loss[loss=0.1974, simple_loss=0.2931, pruned_loss=0.0509, over 16491.00 frames. ], tot_loss[loss=0.2162, simple_loss=0.2996, pruned_loss=0.06634, over 3051923.92 frames. ], batch size: 68, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:17:19,318 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=168294.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:17:32,251 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9794, 3.8914, 4.4809, 1.7593, 4.7467, 4.6653, 3.3176, 3.3741], device='cuda:7'), covar=tensor([0.0693, 0.0227, 0.0148, 0.1352, 0.0042, 0.0115, 0.0332, 0.0444], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0104, 0.0092, 0.0136, 0.0074, 0.0118, 0.0124, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 13:17:32,965 INFO [train.py:904] (7/8) Epoch 17, batch 5900, loss[loss=0.2182, simple_loss=0.2998, pruned_loss=0.06826, over 17186.00 frames. ], tot_loss[loss=0.2154, simple_loss=0.2991, pruned_loss=0.06589, over 3063987.00 frames. ], batch size: 46, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:17:52,245 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2841, 2.0050, 1.6934, 1.7450, 2.2337, 1.9398, 2.0329, 2.3466], device='cuda:7'), covar=tensor([0.0177, 0.0347, 0.0438, 0.0396, 0.0223, 0.0329, 0.0180, 0.0232], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0222, 0.0215, 0.0215, 0.0224, 0.0222, 0.0224, 0.0218], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:18:28,194 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.786e+02 3.302e+02 4.334e+02 9.819e+02, threshold=6.604e+02, percent-clipped=5.0 2023-04-30 13:18:52,640 INFO [train.py:904] (7/8) Epoch 17, batch 5950, loss[loss=0.1888, simple_loss=0.2885, pruned_loss=0.04455, over 16903.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.3001, pruned_loss=0.06452, over 3073687.31 frames. ], batch size: 96, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:19:29,643 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168375.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:19:53,448 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168391.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 13:20:12,042 INFO [train.py:904] (7/8) Epoch 17, batch 6000, loss[loss=0.187, simple_loss=0.2797, pruned_loss=0.04717, over 16870.00 frames. ], tot_loss[loss=0.2134, simple_loss=0.2989, pruned_loss=0.06389, over 3099727.19 frames. ], batch size: 96, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:20:12,042 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 13:20:21,978 INFO [train.py:938] (7/8) Epoch 17, validation: loss=0.1535, simple_loss=0.2667, pruned_loss=0.0202, over 944034.00 frames. 2023-04-30 13:20:21,979 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 13:20:44,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2355, 3.1681, 3.4780, 1.6517, 3.6044, 3.6440, 2.7992, 2.7038], device='cuda:7'), covar=tensor([0.0907, 0.0261, 0.0215, 0.1330, 0.0076, 0.0180, 0.0447, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0104, 0.0092, 0.0136, 0.0074, 0.0117, 0.0123, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 13:20:53,937 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168423.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:20:53,977 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168423.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:21:15,437 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.147e+02 2.768e+02 3.388e+02 4.185e+02 7.743e+02, threshold=6.777e+02, percent-clipped=1.0 2023-04-30 13:21:29,113 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2797, 2.2674, 2.2450, 4.1107, 2.1728, 2.6484, 2.2792, 2.4275], device='cuda:7'), covar=tensor([0.1240, 0.3472, 0.2813, 0.0448, 0.3859, 0.2371, 0.3474, 0.3070], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0423, 0.0351, 0.0318, 0.0425, 0.0490, 0.0393, 0.0495], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:21:39,931 INFO [train.py:904] (7/8) Epoch 17, batch 6050, loss[loss=0.2625, simple_loss=0.3176, pruned_loss=0.1037, over 11891.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2979, pruned_loss=0.06383, over 3091434.91 frames. ], batch size: 250, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:22:09,545 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168471.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:22:38,616 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4598, 2.5975, 2.1641, 2.3250, 2.9592, 2.5691, 3.1151, 3.1510], device='cuda:7'), covar=tensor([0.0102, 0.0360, 0.0468, 0.0434, 0.0233, 0.0374, 0.0209, 0.0243], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0223, 0.0216, 0.0217, 0.0226, 0.0224, 0.0225, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:22:59,850 INFO [train.py:904] (7/8) Epoch 17, batch 6100, loss[loss=0.2111, simple_loss=0.2955, pruned_loss=0.0634, over 16854.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2967, pruned_loss=0.06256, over 3084655.40 frames. ], batch size: 116, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:23:05,046 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5261, 3.5318, 4.0037, 1.8938, 4.1191, 4.1360, 3.0232, 3.1365], device='cuda:7'), covar=tensor([0.0821, 0.0220, 0.0156, 0.1215, 0.0055, 0.0128, 0.0377, 0.0448], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0104, 0.0092, 0.0136, 0.0074, 0.0118, 0.0124, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 13:23:55,309 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.024e+02 2.750e+02 3.352e+02 4.017e+02 8.288e+02, threshold=6.704e+02, percent-clipped=2.0 2023-04-30 13:24:18,369 INFO [train.py:904] (7/8) Epoch 17, batch 6150, loss[loss=0.187, simple_loss=0.2704, pruned_loss=0.05181, over 16547.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.294, pruned_loss=0.06149, over 3085179.46 frames. ], batch size: 62, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:25:17,284 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=168589.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:25:36,303 INFO [train.py:904] (7/8) Epoch 17, batch 6200, loss[loss=0.1725, simple_loss=0.2623, pruned_loss=0.04141, over 17251.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.292, pruned_loss=0.06056, over 3106497.48 frames. ], batch size: 52, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:26:31,028 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.813e+02 2.748e+02 3.467e+02 4.001e+02 6.501e+02, threshold=6.934e+02, percent-clipped=0.0 2023-04-30 13:26:53,721 INFO [train.py:904] (7/8) Epoch 17, batch 6250, loss[loss=0.1981, simple_loss=0.2865, pruned_loss=0.05487, over 16897.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.291, pruned_loss=0.05988, over 3119526.30 frames. ], batch size: 42, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:27:53,895 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168691.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:28:10,637 INFO [train.py:904] (7/8) Epoch 17, batch 6300, loss[loss=0.1916, simple_loss=0.2799, pruned_loss=0.05162, over 16725.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2908, pruned_loss=0.05919, over 3130002.04 frames. ], batch size: 134, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:28:26,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7293, 3.7073, 3.8737, 3.6595, 3.7893, 4.1856, 3.8329, 3.6597], device='cuda:7'), covar=tensor([0.2223, 0.2347, 0.2416, 0.2346, 0.2638, 0.1847, 0.1763, 0.2434], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0557, 0.0610, 0.0469, 0.0627, 0.0642, 0.0484, 0.0629], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:28:38,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2124, 3.3856, 3.5357, 3.5146, 3.5371, 3.3448, 3.3627, 3.4414], device='cuda:7'), covar=tensor([0.0450, 0.0684, 0.0497, 0.0506, 0.0535, 0.0586, 0.0914, 0.0561], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0417, 0.0407, 0.0384, 0.0455, 0.0430, 0.0529, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 13:28:54,841 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7842, 3.8256, 4.1017, 4.0839, 4.1139, 3.8501, 3.8696, 3.8875], device='cuda:7'), covar=tensor([0.0396, 0.0707, 0.0471, 0.0504, 0.0501, 0.0506, 0.0977, 0.0599], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0417, 0.0407, 0.0384, 0.0455, 0.0429, 0.0528, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 13:29:06,432 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.898e+02 2.739e+02 3.263e+02 4.043e+02 7.710e+02, threshold=6.525e+02, percent-clipped=1.0 2023-04-30 13:29:09,987 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168739.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:29:29,591 INFO [train.py:904] (7/8) Epoch 17, batch 6350, loss[loss=0.2366, simple_loss=0.3059, pruned_loss=0.0836, over 12000.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2919, pruned_loss=0.06055, over 3115698.73 frames. ], batch size: 248, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:29:43,917 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2017, 5.1563, 5.0600, 4.6951, 4.6879, 5.0907, 5.0193, 4.7958], device='cuda:7'), covar=tensor([0.0556, 0.0372, 0.0283, 0.0287, 0.1009, 0.0390, 0.0304, 0.0683], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0385, 0.0324, 0.0312, 0.0332, 0.0364, 0.0221, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:29:49,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8855, 5.1884, 4.8755, 4.9273, 4.6796, 4.6632, 4.6226, 5.2515], device='cuda:7'), covar=tensor([0.1150, 0.0864, 0.1049, 0.0841, 0.0800, 0.0947, 0.1160, 0.0950], device='cuda:7'), in_proj_covar=tensor([0.0629, 0.0772, 0.0630, 0.0571, 0.0484, 0.0496, 0.0640, 0.0594], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:30:46,696 INFO [train.py:904] (7/8) Epoch 17, batch 6400, loss[loss=0.1995, simple_loss=0.283, pruned_loss=0.05799, over 16650.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2919, pruned_loss=0.06133, over 3114109.35 frames. ], batch size: 62, lr: 3.98e-03, grad_scale: 8.0 2023-04-30 13:31:34,102 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6240, 4.8936, 5.0480, 4.8538, 4.8494, 5.4426, 4.9032, 4.6986], device='cuda:7'), covar=tensor([0.1223, 0.1932, 0.2335, 0.1875, 0.2650, 0.0948, 0.1672, 0.2387], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0562, 0.0615, 0.0471, 0.0633, 0.0648, 0.0488, 0.0633], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:31:42,107 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.043e+02 2.957e+02 3.621e+02 4.188e+02 7.946e+02, threshold=7.242e+02, percent-clipped=4.0 2023-04-30 13:32:03,557 INFO [train.py:904] (7/8) Epoch 17, batch 6450, loss[loss=0.2006, simple_loss=0.2846, pruned_loss=0.05828, over 15291.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2921, pruned_loss=0.06105, over 3096710.60 frames. ], batch size: 190, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:32:16,618 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-30 13:32:55,663 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 13:33:02,352 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=168889.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:33:20,855 INFO [train.py:904] (7/8) Epoch 17, batch 6500, loss[loss=0.2057, simple_loss=0.2869, pruned_loss=0.06227, over 15422.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2906, pruned_loss=0.06047, over 3112894.94 frames. ], batch size: 191, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:33:48,548 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2691, 3.4074, 3.5685, 3.5363, 3.5727, 3.3757, 3.4127, 3.4846], device='cuda:7'), covar=tensor([0.0390, 0.0693, 0.0526, 0.0521, 0.0534, 0.0549, 0.0853, 0.0544], device='cuda:7'), in_proj_covar=tensor([0.0379, 0.0413, 0.0405, 0.0381, 0.0451, 0.0426, 0.0524, 0.0338], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 13:34:15,422 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=168937.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:34:16,253 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.781e+02 2.744e+02 3.257e+02 4.082e+02 7.063e+02, threshold=6.513e+02, percent-clipped=0.0 2023-04-30 13:34:40,492 INFO [train.py:904] (7/8) Epoch 17, batch 6550, loss[loss=0.2067, simple_loss=0.3122, pruned_loss=0.05059, over 16702.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2941, pruned_loss=0.06181, over 3108409.45 frames. ], batch size: 89, lr: 3.98e-03, grad_scale: 4.0 2023-04-30 13:34:42,601 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 13:35:03,787 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168967.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:35:06,392 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=168969.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 13:35:58,170 INFO [train.py:904] (7/8) Epoch 17, batch 6600, loss[loss=0.2162, simple_loss=0.2994, pruned_loss=0.06648, over 16635.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2957, pruned_loss=0.06198, over 3115839.51 frames. ], batch size: 57, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:36:07,677 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-04-30 13:36:37,611 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169028.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:36:40,768 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169030.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 13:36:50,873 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.775e+02 2.872e+02 3.571e+02 4.497e+02 9.620e+02, threshold=7.142e+02, percent-clipped=5.0 2023-04-30 13:37:13,184 INFO [train.py:904] (7/8) Epoch 17, batch 6650, loss[loss=0.2386, simple_loss=0.3064, pruned_loss=0.08535, over 11495.00 frames. ], tot_loss[loss=0.2118, simple_loss=0.2966, pruned_loss=0.06346, over 3099362.47 frames. ], batch size: 247, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:37:40,557 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-04-30 13:37:59,856 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169082.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:38:27,269 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 13:38:28,131 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4928, 4.4590, 4.3425, 3.4338, 4.4098, 1.5429, 4.1248, 4.0609], device='cuda:7'), covar=tensor([0.0105, 0.0094, 0.0206, 0.0461, 0.0108, 0.2952, 0.0146, 0.0262], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0138, 0.0185, 0.0169, 0.0158, 0.0195, 0.0172, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:38:30,530 INFO [train.py:904] (7/8) Epoch 17, batch 6700, loss[loss=0.1885, simple_loss=0.2745, pruned_loss=0.05126, over 16472.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2957, pruned_loss=0.06359, over 3086103.85 frames. ], batch size: 75, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:38:53,646 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169116.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:39:01,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3505, 2.9396, 2.6987, 2.2976, 2.2419, 2.2677, 2.8633, 2.8521], device='cuda:7'), covar=tensor([0.2372, 0.0691, 0.1418, 0.2334, 0.2208, 0.1948, 0.0460, 0.1172], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0260, 0.0295, 0.0299, 0.0289, 0.0242, 0.0283, 0.0319], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:39:02,420 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0143, 3.9173, 4.1148, 4.2368, 4.3594, 3.9512, 4.3214, 4.3733], device='cuda:7'), covar=tensor([0.1658, 0.1253, 0.1403, 0.0702, 0.0552, 0.1573, 0.0726, 0.0664], device='cuda:7'), in_proj_covar=tensor([0.0586, 0.0722, 0.0858, 0.0740, 0.0557, 0.0588, 0.0593, 0.0691], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:39:26,520 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.893e+02 2.876e+02 3.506e+02 4.097e+02 7.829e+02, threshold=7.011e+02, percent-clipped=1.0 2023-04-30 13:39:35,023 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169143.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:39:47,632 INFO [train.py:904] (7/8) Epoch 17, batch 6750, loss[loss=0.1839, simple_loss=0.2693, pruned_loss=0.04924, over 16457.00 frames. ], tot_loss[loss=0.2116, simple_loss=0.295, pruned_loss=0.06405, over 3072140.26 frames. ], batch size: 75, lr: 3.97e-03, grad_scale: 4.0 2023-04-30 13:39:58,733 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169159.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:40:26,607 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9488, 5.1981, 4.9609, 4.9458, 4.7409, 4.7051, 4.6287, 5.3242], device='cuda:7'), covar=tensor([0.1155, 0.0786, 0.0950, 0.0880, 0.0811, 0.0878, 0.1116, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0631, 0.0772, 0.0630, 0.0573, 0.0484, 0.0495, 0.0639, 0.0596], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:40:26,761 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169177.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:41:04,391 INFO [train.py:904] (7/8) Epoch 17, batch 6800, loss[loss=0.2143, simple_loss=0.2962, pruned_loss=0.06613, over 16727.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2945, pruned_loss=0.06366, over 3070695.17 frames. ], batch size: 134, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:41:33,772 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169220.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:42:02,689 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.174e+02 2.920e+02 3.483e+02 4.404e+02 7.991e+02, threshold=6.967e+02, percent-clipped=1.0 2023-04-30 13:42:23,105 INFO [train.py:904] (7/8) Epoch 17, batch 6850, loss[loss=0.2144, simple_loss=0.3115, pruned_loss=0.05865, over 16915.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2957, pruned_loss=0.06358, over 3066782.41 frames. ], batch size: 109, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:42:43,829 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2441, 3.4737, 3.5091, 2.4212, 3.3098, 3.5850, 3.3830, 1.9819], device='cuda:7'), covar=tensor([0.0467, 0.0053, 0.0052, 0.0355, 0.0083, 0.0087, 0.0073, 0.0445], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0077, 0.0078, 0.0133, 0.0091, 0.0102, 0.0090, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:43:37,895 INFO [train.py:904] (7/8) Epoch 17, batch 6900, loss[loss=0.2064, simple_loss=0.2905, pruned_loss=0.06109, over 16678.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2977, pruned_loss=0.06269, over 3097908.75 frames. ], batch size: 57, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:43:58,626 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8388, 5.1889, 5.3845, 5.1363, 5.2037, 5.7845, 5.2309, 5.0601], device='cuda:7'), covar=tensor([0.1093, 0.1889, 0.2170, 0.1848, 0.2428, 0.0953, 0.1575, 0.2171], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0558, 0.0612, 0.0467, 0.0631, 0.0645, 0.0487, 0.0632], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:44:11,083 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169323.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:44:14,334 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169325.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 13:44:33,267 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.634e+02 3.268e+02 4.133e+02 7.362e+02, threshold=6.535e+02, percent-clipped=1.0 2023-04-30 13:44:52,823 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5291, 4.0858, 4.0131, 2.7710, 3.6003, 4.0468, 3.6975, 2.3228], device='cuda:7'), covar=tensor([0.0450, 0.0033, 0.0041, 0.0334, 0.0089, 0.0095, 0.0076, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0076, 0.0077, 0.0132, 0.0090, 0.0102, 0.0089, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:44:55,631 INFO [train.py:904] (7/8) Epoch 17, batch 6950, loss[loss=0.2246, simple_loss=0.3034, pruned_loss=0.07289, over 16447.00 frames. ], tot_loss[loss=0.2146, simple_loss=0.2996, pruned_loss=0.06481, over 3097384.59 frames. ], batch size: 35, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:45:00,685 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 13:46:06,135 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-30 13:46:11,853 INFO [train.py:904] (7/8) Epoch 17, batch 7000, loss[loss=0.1928, simple_loss=0.2909, pruned_loss=0.04729, over 17026.00 frames. ], tot_loss[loss=0.2137, simple_loss=0.2994, pruned_loss=0.064, over 3091572.40 frames. ], batch size: 53, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:47:07,116 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 3.013e+02 3.558e+02 4.321e+02 1.050e+03, threshold=7.116e+02, percent-clipped=3.0 2023-04-30 13:47:07,492 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169438.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:47:28,668 INFO [train.py:904] (7/8) Epoch 17, batch 7050, loss[loss=0.2034, simple_loss=0.288, pruned_loss=0.0594, over 16593.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.299, pruned_loss=0.06281, over 3113987.56 frames. ], batch size: 62, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:48:00,529 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169472.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:48:47,191 INFO [train.py:904] (7/8) Epoch 17, batch 7100, loss[loss=0.2343, simple_loss=0.3121, pruned_loss=0.07822, over 16476.00 frames. ], tot_loss[loss=0.2125, simple_loss=0.2982, pruned_loss=0.06339, over 3094586.62 frames. ], batch size: 68, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:48:51,000 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 13:48:52,284 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0238, 3.3271, 3.4581, 2.0247, 2.9934, 2.2772, 3.5081, 3.5698], device='cuda:7'), covar=tensor([0.0270, 0.0735, 0.0565, 0.1979, 0.0803, 0.0898, 0.0635, 0.0941], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0158, 0.0163, 0.0150, 0.0142, 0.0126, 0.0141, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 13:49:08,167 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169515.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:49:21,942 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.81 vs. limit=5.0 2023-04-30 13:49:42,153 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.876e+02 3.484e+02 4.198e+02 1.044e+03, threshold=6.968e+02, percent-clipped=3.0 2023-04-30 13:50:02,983 INFO [train.py:904] (7/8) Epoch 17, batch 7150, loss[loss=0.2207, simple_loss=0.2971, pruned_loss=0.07213, over 15307.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2963, pruned_loss=0.0632, over 3086311.17 frames. ], batch size: 190, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:50:46,513 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-04-30 13:51:10,886 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0838, 4.0365, 4.4793, 2.0691, 4.7801, 4.6806, 3.2165, 3.4490], device='cuda:7'), covar=tensor([0.0721, 0.0216, 0.0164, 0.1322, 0.0045, 0.0117, 0.0435, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0105, 0.0094, 0.0139, 0.0075, 0.0119, 0.0126, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 13:51:19,564 INFO [train.py:904] (7/8) Epoch 17, batch 7200, loss[loss=0.2068, simple_loss=0.2831, pruned_loss=0.06529, over 11789.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2942, pruned_loss=0.06185, over 3065661.75 frames. ], batch size: 247, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:51:48,704 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169621.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:51:51,971 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169623.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:51:55,941 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169625.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 13:52:09,259 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 13:52:16,449 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.818e+02 3.396e+02 4.262e+02 6.944e+02, threshold=6.791e+02, percent-clipped=0.0 2023-04-30 13:52:38,963 INFO [train.py:904] (7/8) Epoch 17, batch 7250, loss[loss=0.189, simple_loss=0.2715, pruned_loss=0.05328, over 15294.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2919, pruned_loss=0.06126, over 3050590.40 frames. ], batch size: 190, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:53:08,593 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169671.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:53:12,519 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169673.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 13:53:25,566 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169682.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:53:32,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6210, 3.0906, 3.1962, 1.7722, 2.7603, 2.0726, 3.1881, 3.3657], device='cuda:7'), covar=tensor([0.0360, 0.0889, 0.0645, 0.2313, 0.0945, 0.1101, 0.0797, 0.1006], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0160, 0.0166, 0.0152, 0.0144, 0.0128, 0.0143, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 13:53:32,215 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-30 13:53:57,123 INFO [train.py:904] (7/8) Epoch 17, batch 7300, loss[loss=0.199, simple_loss=0.2933, pruned_loss=0.05238, over 16910.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2912, pruned_loss=0.06078, over 3053107.48 frames. ], batch size: 96, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:54:41,445 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169729.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:54:55,064 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 2.965e+02 3.498e+02 4.419e+02 8.107e+02, threshold=6.996e+02, percent-clipped=3.0 2023-04-30 13:54:55,502 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169738.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:55:10,730 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169747.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:55:17,629 INFO [train.py:904] (7/8) Epoch 17, batch 7350, loss[loss=0.2065, simple_loss=0.3005, pruned_loss=0.05624, over 16569.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.292, pruned_loss=0.06143, over 3036130.31 frames. ], batch size: 62, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:55:50,570 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169772.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:12,760 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169786.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:12,965 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=169786.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:19,769 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169790.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:20,990 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8887, 4.1193, 3.8899, 4.0028, 3.6551, 3.7687, 3.8148, 4.0797], device='cuda:7'), covar=tensor([0.0981, 0.0883, 0.1065, 0.0806, 0.0797, 0.1537, 0.0884, 0.0992], device='cuda:7'), in_proj_covar=tensor([0.0620, 0.0757, 0.0618, 0.0561, 0.0477, 0.0488, 0.0625, 0.0586], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:56:37,958 INFO [train.py:904] (7/8) Epoch 17, batch 7400, loss[loss=0.2308, simple_loss=0.3011, pruned_loss=0.0802, over 11055.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2932, pruned_loss=0.06181, over 3036956.71 frames. ], batch size: 247, lr: 3.97e-03, grad_scale: 8.0 2023-04-30 13:56:49,178 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169808.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:56:59,914 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=169815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:57:06,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9037, 4.1806, 3.1192, 2.4733, 2.9948, 2.6466, 4.5780, 3.7399], device='cuda:7'), covar=tensor([0.2829, 0.0657, 0.1746, 0.2385, 0.2468, 0.1892, 0.0438, 0.1143], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0264, 0.0301, 0.0303, 0.0293, 0.0246, 0.0288, 0.0324], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 13:57:07,986 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169820.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:57:36,687 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.775e+02 3.366e+02 3.998e+02 8.087e+02, threshold=6.732e+02, percent-clipped=3.0 2023-04-30 13:57:52,257 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=169847.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:57:55,515 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8067, 2.3175, 1.8707, 2.1645, 2.6969, 2.3398, 2.6836, 2.8599], device='cuda:7'), covar=tensor([0.0171, 0.0392, 0.0506, 0.0406, 0.0230, 0.0384, 0.0222, 0.0255], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0221, 0.0214, 0.0215, 0.0220, 0.0220, 0.0221, 0.0215], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 13:57:59,471 INFO [train.py:904] (7/8) Epoch 17, batch 7450, loss[loss=0.2471, simple_loss=0.3114, pruned_loss=0.09139, over 11583.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2944, pruned_loss=0.06269, over 3047368.10 frames. ], batch size: 247, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 13:58:19,043 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=169863.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 13:59:20,323 INFO [train.py:904] (7/8) Epoch 17, batch 7500, loss[loss=0.1931, simple_loss=0.2783, pruned_loss=0.05391, over 16679.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2948, pruned_loss=0.06191, over 3043702.03 frames. ], batch size: 57, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:00:17,629 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.669e+02 3.210e+02 3.850e+02 6.683e+02, threshold=6.420e+02, percent-clipped=0.0 2023-04-30 14:00:39,252 INFO [train.py:904] (7/8) Epoch 17, batch 7550, loss[loss=0.1767, simple_loss=0.2648, pruned_loss=0.04431, over 16755.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2941, pruned_loss=0.06229, over 3043127.68 frames. ], batch size: 89, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:00:43,083 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4757, 4.6719, 4.7865, 4.6191, 4.6646, 5.1844, 4.7009, 4.5231], device='cuda:7'), covar=tensor([0.1313, 0.1832, 0.2371, 0.2066, 0.2595, 0.1035, 0.1608, 0.2363], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0560, 0.0616, 0.0468, 0.0633, 0.0647, 0.0488, 0.0630], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 14:00:54,443 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7467, 1.6973, 2.2597, 2.6594, 2.5608, 3.0649, 1.8619, 2.9999], device='cuda:7'), covar=tensor([0.0186, 0.0510, 0.0324, 0.0277, 0.0292, 0.0155, 0.0508, 0.0122], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0185, 0.0171, 0.0174, 0.0184, 0.0142, 0.0187, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:01:09,149 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2742, 3.4054, 3.6053, 3.5816, 3.5994, 3.3739, 3.4487, 3.4948], device='cuda:7'), covar=tensor([0.0416, 0.0768, 0.0458, 0.0472, 0.0494, 0.0578, 0.0863, 0.0524], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0408, 0.0401, 0.0379, 0.0449, 0.0423, 0.0518, 0.0334], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 14:01:19,602 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=169977.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:01:45,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4080, 2.9624, 3.0155, 1.9955, 2.6999, 2.1404, 3.0723, 3.1240], device='cuda:7'), covar=tensor([0.0293, 0.0744, 0.0563, 0.1957, 0.0873, 0.0987, 0.0662, 0.0877], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0159, 0.0165, 0.0151, 0.0144, 0.0128, 0.0142, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 14:02:00,842 INFO [train.py:904] (7/8) Epoch 17, batch 7600, loss[loss=0.1905, simple_loss=0.2806, pruned_loss=0.05023, over 16464.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2933, pruned_loss=0.06209, over 3054364.31 frames. ], batch size: 75, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:02:09,831 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4998, 4.5031, 4.3677, 3.6246, 4.4155, 1.7052, 4.1488, 4.0911], device='cuda:7'), covar=tensor([0.0095, 0.0079, 0.0160, 0.0330, 0.0089, 0.2565, 0.0128, 0.0201], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0136, 0.0183, 0.0167, 0.0155, 0.0193, 0.0169, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:02:58,192 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 2.819e+02 3.511e+02 4.461e+02 7.625e+02, threshold=7.022e+02, percent-clipped=5.0 2023-04-30 14:03:08,011 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170045.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:03:17,372 INFO [train.py:904] (7/8) Epoch 17, batch 7650, loss[loss=0.1842, simple_loss=0.2795, pruned_loss=0.0444, over 16884.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2941, pruned_loss=0.06263, over 3070002.88 frames. ], batch size: 96, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:04:07,618 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170085.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:04:33,143 INFO [train.py:904] (7/8) Epoch 17, batch 7700, loss[loss=0.1963, simple_loss=0.2811, pruned_loss=0.05581, over 16404.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2944, pruned_loss=0.06356, over 3064403.38 frames. ], batch size: 68, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:04:35,299 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170103.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:04:39,608 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170106.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:05:29,557 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.959e+02 3.624e+02 4.617e+02 9.999e+02, threshold=7.248e+02, percent-clipped=4.0 2023-04-30 14:05:34,961 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170142.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:05:50,033 INFO [train.py:904] (7/8) Epoch 17, batch 7750, loss[loss=0.2433, simple_loss=0.3109, pruned_loss=0.0878, over 11349.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.294, pruned_loss=0.06269, over 3089264.20 frames. ], batch size: 247, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:06:07,970 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9559, 3.8948, 4.0138, 4.1364, 4.2389, 3.8812, 4.1844, 4.2545], device='cuda:7'), covar=tensor([0.1635, 0.1105, 0.1327, 0.0706, 0.0583, 0.1634, 0.0797, 0.0684], device='cuda:7'), in_proj_covar=tensor([0.0586, 0.0724, 0.0861, 0.0736, 0.0559, 0.0587, 0.0598, 0.0691], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:06:08,343 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-30 14:06:39,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1610, 3.7597, 3.7556, 2.4708, 3.4371, 3.7425, 3.5040, 2.0419], device='cuda:7'), covar=tensor([0.0513, 0.0047, 0.0053, 0.0375, 0.0100, 0.0128, 0.0090, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0076, 0.0078, 0.0132, 0.0091, 0.0102, 0.0090, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 14:06:51,330 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 14:07:07,234 INFO [train.py:904] (7/8) Epoch 17, batch 7800, loss[loss=0.1969, simple_loss=0.285, pruned_loss=0.05442, over 16882.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2946, pruned_loss=0.0637, over 3069059.57 frames. ], batch size: 116, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:07:25,362 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-04-30 14:07:29,737 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0134, 3.3755, 3.3514, 2.1559, 3.1333, 3.3580, 3.2067, 1.9046], device='cuda:7'), covar=tensor([0.0549, 0.0054, 0.0064, 0.0410, 0.0097, 0.0120, 0.0095, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0076, 0.0077, 0.0132, 0.0091, 0.0102, 0.0089, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 14:07:48,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8253, 2.0048, 2.4117, 1.7419, 2.4176, 2.7066, 2.3739, 2.1889], device='cuda:7'), covar=tensor([0.0981, 0.0248, 0.0221, 0.1138, 0.0125, 0.0255, 0.0446, 0.0537], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0103, 0.0092, 0.0135, 0.0073, 0.0117, 0.0124, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 14:08:03,351 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.635e+02 2.884e+02 3.642e+02 4.850e+02 8.293e+02, threshold=7.285e+02, percent-clipped=4.0 2023-04-30 14:08:16,493 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-30 14:08:23,357 INFO [train.py:904] (7/8) Epoch 17, batch 7850, loss[loss=0.1916, simple_loss=0.2826, pruned_loss=0.0503, over 16939.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2948, pruned_loss=0.06266, over 3083648.77 frames. ], batch size: 109, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:09:00,979 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170277.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:09:38,285 INFO [train.py:904] (7/8) Epoch 17, batch 7900, loss[loss=0.2289, simple_loss=0.2973, pruned_loss=0.08021, over 11571.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2944, pruned_loss=0.06233, over 3084977.27 frames. ], batch size: 249, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:10:13,069 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170325.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:10:22,511 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9846, 4.0761, 3.8804, 3.6476, 3.5817, 4.0026, 3.7325, 3.7235], device='cuda:7'), covar=tensor([0.0666, 0.0610, 0.0325, 0.0307, 0.0779, 0.0516, 0.0939, 0.0657], device='cuda:7'), in_proj_covar=tensor([0.0266, 0.0378, 0.0316, 0.0302, 0.0324, 0.0352, 0.0218, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:10:36,114 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.088e+02 2.800e+02 3.465e+02 4.175e+02 8.586e+02, threshold=6.929e+02, percent-clipped=4.0 2023-04-30 14:10:55,573 INFO [train.py:904] (7/8) Epoch 17, batch 7950, loss[loss=0.2584, simple_loss=0.3134, pruned_loss=0.1017, over 11604.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2952, pruned_loss=0.06314, over 3086375.81 frames. ], batch size: 248, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:11:07,147 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-04-30 14:11:31,454 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6030, 2.7330, 2.4635, 4.1262, 2.8745, 3.9361, 1.5033, 2.8484], device='cuda:7'), covar=tensor([0.1429, 0.0742, 0.1279, 0.0234, 0.0300, 0.0439, 0.1746, 0.0863], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0169, 0.0190, 0.0176, 0.0204, 0.0213, 0.0195, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 14:11:49,337 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170385.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:12:13,328 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170401.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:12:14,153 INFO [train.py:904] (7/8) Epoch 17, batch 8000, loss[loss=0.2137, simple_loss=0.2982, pruned_loss=0.06454, over 16607.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2957, pruned_loss=0.06347, over 3097185.29 frames. ], batch size: 62, lr: 3.96e-03, grad_scale: 8.0 2023-04-30 14:12:15,836 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170403.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:12:31,801 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9098, 4.1673, 3.9694, 4.0164, 3.7294, 3.8067, 3.8452, 4.1406], device='cuda:7'), covar=tensor([0.1116, 0.1033, 0.1038, 0.0926, 0.0837, 0.1591, 0.0962, 0.1118], device='cuda:7'), in_proj_covar=tensor([0.0625, 0.0763, 0.0625, 0.0567, 0.0480, 0.0492, 0.0632, 0.0592], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:13:00,995 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:11,713 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 2.905e+02 3.477e+02 4.096e+02 7.787e+02, threshold=6.954e+02, percent-clipped=1.0 2023-04-30 14:13:15,190 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170442.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:21,444 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4561, 1.5891, 2.1629, 2.4069, 2.4778, 2.7590, 1.7693, 2.6788], device='cuda:7'), covar=tensor([0.0188, 0.0456, 0.0264, 0.0298, 0.0260, 0.0148, 0.0446, 0.0132], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0184, 0.0170, 0.0173, 0.0183, 0.0142, 0.0187, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:13:29,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170451.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:13:29,849 INFO [train.py:904] (7/8) Epoch 17, batch 8050, loss[loss=0.2072, simple_loss=0.2963, pruned_loss=0.05907, over 16239.00 frames. ], tot_loss[loss=0.2107, simple_loss=0.2954, pruned_loss=0.06303, over 3111003.33 frames. ], batch size: 165, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:13:44,264 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170461.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:14:28,181 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170490.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:14:29,891 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.53 vs. limit=5.0 2023-04-30 14:14:37,821 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-04-30 14:14:46,144 INFO [train.py:904] (7/8) Epoch 17, batch 8100, loss[loss=0.1978, simple_loss=0.2852, pruned_loss=0.05519, over 16426.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.294, pruned_loss=0.06174, over 3111098.01 frames. ], batch size: 75, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:15:16,562 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170522.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 14:15:42,388 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.803e+02 2.570e+02 3.195e+02 3.934e+02 7.205e+02, threshold=6.390e+02, percent-clipped=3.0 2023-04-30 14:16:00,590 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-30 14:16:00,988 INFO [train.py:904] (7/8) Epoch 17, batch 8150, loss[loss=0.2173, simple_loss=0.2918, pruned_loss=0.07141, over 16758.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2917, pruned_loss=0.06114, over 3109778.12 frames. ], batch size: 124, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:16:27,632 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170569.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:16:36,898 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-04-30 14:17:14,951 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2023-04-30 14:17:18,500 INFO [train.py:904] (7/8) Epoch 17, batch 8200, loss[loss=0.2177, simple_loss=0.2906, pruned_loss=0.07234, over 11674.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2889, pruned_loss=0.06044, over 3097608.99 frames. ], batch size: 248, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:18:05,936 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170630.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:18:21,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.868e+02 2.599e+02 3.232e+02 3.801e+02 6.958e+02, threshold=6.463e+02, percent-clipped=1.0 2023-04-30 14:18:41,078 INFO [train.py:904] (7/8) Epoch 17, batch 8250, loss[loss=0.1933, simple_loss=0.2828, pruned_loss=0.0519, over 16678.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2879, pruned_loss=0.05803, over 3077654.65 frames. ], batch size: 134, lr: 3.96e-03, grad_scale: 4.0 2023-04-30 14:20:04,659 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=170701.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:20:05,307 INFO [train.py:904] (7/8) Epoch 17, batch 8300, loss[loss=0.1837, simple_loss=0.2797, pruned_loss=0.04388, over 16377.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2852, pruned_loss=0.05532, over 3060320.50 frames. ], batch size: 146, lr: 3.95e-03, grad_scale: 4.0 2023-04-30 14:20:45,690 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-04-30 14:21:09,573 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.327e+02 2.774e+02 3.132e+02 7.452e+02, threshold=5.547e+02, percent-clipped=1.0 2023-04-30 14:21:26,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=170749.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:21:29,860 INFO [train.py:904] (7/8) Epoch 17, batch 8350, loss[loss=0.1868, simple_loss=0.2816, pruned_loss=0.04594, over 16077.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2841, pruned_loss=0.05353, over 3042198.89 frames. ], batch size: 165, lr: 3.95e-03, grad_scale: 4.0 2023-04-30 14:22:51,867 INFO [train.py:904] (7/8) Epoch 17, batch 8400, loss[loss=0.1935, simple_loss=0.2931, pruned_loss=0.04696, over 16691.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.282, pruned_loss=0.05174, over 3031859.31 frames. ], batch size: 134, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:23:13,356 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 14:23:16,982 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170817.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 14:23:17,394 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 14:23:52,184 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.262e+02 2.769e+02 3.288e+02 8.219e+02, threshold=5.539e+02, percent-clipped=2.0 2023-04-30 14:24:10,764 INFO [train.py:904] (7/8) Epoch 17, batch 8450, loss[loss=0.1709, simple_loss=0.2698, pruned_loss=0.03607, over 16576.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2802, pruned_loss=0.04992, over 3034637.88 frames. ], batch size: 62, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:24:16,265 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6186, 4.5758, 4.4105, 3.8268, 4.4812, 1.7760, 4.2503, 4.2314], device='cuda:7'), covar=tensor([0.0073, 0.0074, 0.0150, 0.0255, 0.0082, 0.2379, 0.0114, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0134, 0.0180, 0.0163, 0.0154, 0.0190, 0.0167, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:24:42,158 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 14:25:32,475 INFO [train.py:904] (7/8) Epoch 17, batch 8500, loss[loss=0.174, simple_loss=0.2645, pruned_loss=0.04177, over 16830.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.2763, pruned_loss=0.04747, over 3035991.81 frames. ], batch size: 116, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:25:51,525 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2023-04-30 14:26:10,261 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=170925.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:26:15,387 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=170928.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:26:34,261 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.118e+02 2.662e+02 3.144e+02 4.893e+02, threshold=5.323e+02, percent-clipped=0.0 2023-04-30 14:26:56,014 INFO [train.py:904] (7/8) Epoch 17, batch 8550, loss[loss=0.1869, simple_loss=0.2812, pruned_loss=0.04631, over 15418.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2741, pruned_loss=0.04639, over 3033278.24 frames. ], batch size: 191, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:28:08,520 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=170989.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:28:33,702 INFO [train.py:904] (7/8) Epoch 17, batch 8600, loss[loss=0.1721, simple_loss=0.2699, pruned_loss=0.03718, over 15368.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2741, pruned_loss=0.04521, over 3041727.82 frames. ], batch size: 191, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:29:36,357 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 14:29:52,155 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.454e+02 2.845e+02 3.449e+02 5.171e+02, threshold=5.691e+02, percent-clipped=0.0 2023-04-30 14:30:15,685 INFO [train.py:904] (7/8) Epoch 17, batch 8650, loss[loss=0.1638, simple_loss=0.2627, pruned_loss=0.03243, over 16660.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2722, pruned_loss=0.04378, over 3032974.87 frames. ], batch size: 89, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:31:44,048 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171092.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:32:02,422 INFO [train.py:904] (7/8) Epoch 17, batch 8700, loss[loss=0.1583, simple_loss=0.2525, pruned_loss=0.03209, over 16766.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2693, pruned_loss=0.0427, over 3038284.01 frames. ], batch size: 83, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:32:20,398 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9505, 2.2419, 2.3542, 3.0904, 1.9432, 3.2463, 1.7196, 2.7911], device='cuda:7'), covar=tensor([0.1266, 0.0723, 0.1095, 0.0190, 0.0092, 0.0392, 0.1605, 0.0688], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0166, 0.0188, 0.0173, 0.0201, 0.0211, 0.0194, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 14:32:31,833 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171117.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 14:33:13,279 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.121e+02 2.631e+02 3.199e+02 5.449e+02, threshold=5.263e+02, percent-clipped=0.0 2023-04-30 14:33:24,251 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6370, 4.8858, 4.9603, 4.7727, 4.7657, 5.3477, 4.8453, 4.5707], device='cuda:7'), covar=tensor([0.1118, 0.1781, 0.1968, 0.1913, 0.2525, 0.0930, 0.1569, 0.2363], device='cuda:7'), in_proj_covar=tensor([0.0370, 0.0538, 0.0590, 0.0451, 0.0602, 0.0626, 0.0469, 0.0604], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 14:33:38,788 INFO [train.py:904] (7/8) Epoch 17, batch 8750, loss[loss=0.1922, simple_loss=0.2868, pruned_loss=0.04882, over 16845.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.269, pruned_loss=0.04217, over 3040852.47 frames. ], batch size: 116, lr: 3.95e-03, grad_scale: 4.0 2023-04-30 14:33:43,347 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171153.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:34:13,585 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171165.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:34:23,441 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-04-30 14:34:39,150 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-04-30 14:35:21,538 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171196.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:35:33,201 INFO [train.py:904] (7/8) Epoch 17, batch 8800, loss[loss=0.1592, simple_loss=0.2564, pruned_loss=0.031, over 16137.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2673, pruned_loss=0.04103, over 3035577.41 frames. ], batch size: 165, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:36:04,220 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3489, 1.5094, 1.9991, 2.2569, 2.2634, 2.6148, 1.7307, 2.5320], device='cuda:7'), covar=tensor([0.0245, 0.0537, 0.0315, 0.0309, 0.0342, 0.0204, 0.0507, 0.0147], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0183, 0.0170, 0.0172, 0.0183, 0.0139, 0.0185, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:36:22,804 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171225.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:36:54,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.375e+02 2.916e+02 3.471e+02 5.213e+02, threshold=5.833e+02, percent-clipped=0.0 2023-04-30 14:37:17,949 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.11 vs. limit=2.0 2023-04-30 14:37:18,425 INFO [train.py:904] (7/8) Epoch 17, batch 8850, loss[loss=0.1587, simple_loss=0.2485, pruned_loss=0.03451, over 12402.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2699, pruned_loss=0.04085, over 3010212.45 frames. ], batch size: 249, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:37:22,689 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 14:37:29,144 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:38:03,311 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171273.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:38:12,952 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2745, 2.1021, 2.1720, 3.8973, 2.1029, 2.4773, 2.2559, 2.2563], device='cuda:7'), covar=tensor([0.1074, 0.3673, 0.2967, 0.0469, 0.4392, 0.2620, 0.3568, 0.3631], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0418, 0.0348, 0.0311, 0.0422, 0.0479, 0.0389, 0.0484], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:38:27,516 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171284.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:39:01,956 INFO [train.py:904] (7/8) Epoch 17, batch 8900, loss[loss=0.1776, simple_loss=0.2732, pruned_loss=0.04104, over 16699.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2701, pruned_loss=0.03981, over 3024917.68 frames. ], batch size: 89, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:40:38,368 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.247e+02 2.656e+02 3.091e+02 5.659e+02, threshold=5.312e+02, percent-clipped=0.0 2023-04-30 14:40:51,349 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171345.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:41:04,770 INFO [train.py:904] (7/8) Epoch 17, batch 8950, loss[loss=0.1533, simple_loss=0.2436, pruned_loss=0.03152, over 15271.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2694, pruned_loss=0.04019, over 3030404.04 frames. ], batch size: 191, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:41:06,759 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7661, 4.9751, 5.1259, 4.9688, 5.0587, 5.5509, 5.0715, 4.7576], device='cuda:7'), covar=tensor([0.0912, 0.1744, 0.1817, 0.1799, 0.2313, 0.0895, 0.1568, 0.2278], device='cuda:7'), in_proj_covar=tensor([0.0369, 0.0535, 0.0587, 0.0450, 0.0600, 0.0626, 0.0469, 0.0601], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 14:42:53,011 INFO [train.py:904] (7/8) Epoch 17, batch 9000, loss[loss=0.1663, simple_loss=0.2557, pruned_loss=0.03846, over 11993.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2663, pruned_loss=0.03905, over 3039345.62 frames. ], batch size: 250, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:42:53,012 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 14:43:02,952 INFO [train.py:938] (7/8) Epoch 17, validation: loss=0.1478, simple_loss=0.2519, pruned_loss=0.02187, over 944034.00 frames. 2023-04-30 14:43:02,952 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 14:43:11,084 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171406.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:44:23,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 2.012e+02 2.627e+02 3.134e+02 6.797e+02, threshold=5.255e+02, percent-clipped=3.0 2023-04-30 14:44:41,575 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171448.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:44:42,277 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-04-30 14:44:47,387 INFO [train.py:904] (7/8) Epoch 17, batch 9050, loss[loss=0.1741, simple_loss=0.2549, pruned_loss=0.0467, over 16932.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2672, pruned_loss=0.0398, over 3045881.26 frames. ], batch size: 109, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:46:24,757 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:46:30,498 INFO [train.py:904] (7/8) Epoch 17, batch 9100, loss[loss=0.1754, simple_loss=0.2581, pruned_loss=0.0464, over 12452.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2675, pruned_loss=0.04025, over 3055523.20 frames. ], batch size: 247, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:46:50,765 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2955, 2.3448, 1.9390, 2.1199, 2.7605, 2.3925, 2.8813, 2.8981], device='cuda:7'), covar=tensor([0.0107, 0.0412, 0.0507, 0.0454, 0.0259, 0.0367, 0.0206, 0.0260], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0218, 0.0211, 0.0212, 0.0217, 0.0216, 0.0215, 0.0208], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:47:32,531 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-04-30 14:48:02,856 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.178e+02 2.662e+02 3.383e+02 6.301e+02, threshold=5.324e+02, percent-clipped=5.0 2023-04-30 14:48:28,390 INFO [train.py:904] (7/8) Epoch 17, batch 9150, loss[loss=0.1645, simple_loss=0.2611, pruned_loss=0.03397, over 16728.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2677, pruned_loss=0.03959, over 3059550.25 frames. ], batch size: 134, lr: 3.95e-03, grad_scale: 8.0 2023-04-30 14:48:29,517 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171552.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:48:44,440 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171559.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:49:37,081 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171584.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:50:13,343 INFO [train.py:904] (7/8) Epoch 17, batch 9200, loss[loss=0.1645, simple_loss=0.258, pruned_loss=0.03551, over 16721.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2635, pruned_loss=0.0387, over 3079300.51 frames. ], batch size: 62, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 14:51:10,344 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171632.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:51:27,842 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.614e+02 2.277e+02 2.610e+02 3.229e+02 7.994e+02, threshold=5.220e+02, percent-clipped=2.0 2023-04-30 14:51:50,313 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4269, 3.0849, 2.7054, 2.2136, 2.1505, 2.2050, 2.9662, 2.8078], device='cuda:7'), covar=tensor([0.2418, 0.0671, 0.1430, 0.2604, 0.2784, 0.2156, 0.0425, 0.1362], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0251, 0.0288, 0.0291, 0.0275, 0.0235, 0.0274, 0.0309], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:51:50,827 INFO [train.py:904] (7/8) Epoch 17, batch 9250, loss[loss=0.175, simple_loss=0.2525, pruned_loss=0.04878, over 12398.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2625, pruned_loss=0.03827, over 3070909.11 frames. ], batch size: 248, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 14:53:42,080 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171701.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:53:42,881 INFO [train.py:904] (7/8) Epoch 17, batch 9300, loss[loss=0.1445, simple_loss=0.2326, pruned_loss=0.02817, over 12293.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2607, pruned_loss=0.03748, over 3057946.56 frames. ], batch size: 250, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 14:53:58,590 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5607, 3.8098, 2.7564, 2.1089, 2.2325, 2.2901, 3.9805, 3.2539], device='cuda:7'), covar=tensor([0.3030, 0.0572, 0.1870, 0.2933, 0.2975, 0.2111, 0.0404, 0.1201], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0253, 0.0291, 0.0294, 0.0277, 0.0237, 0.0277, 0.0312], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:55:09,821 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.061e+02 2.530e+02 2.950e+02 5.223e+02, threshold=5.060e+02, percent-clipped=1.0 2023-04-30 14:55:21,265 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171748.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:55:27,777 INFO [train.py:904] (7/8) Epoch 17, batch 9350, loss[loss=0.1823, simple_loss=0.2817, pruned_loss=0.04145, over 16805.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.261, pruned_loss=0.03764, over 3056974.21 frames. ], batch size: 124, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 14:56:05,621 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8195, 3.8267, 4.0338, 3.7752, 4.0036, 4.3422, 3.9299, 3.7079], device='cuda:7'), covar=tensor([0.2124, 0.2275, 0.2029, 0.2362, 0.2261, 0.1336, 0.1643, 0.2508], device='cuda:7'), in_proj_covar=tensor([0.0361, 0.0529, 0.0579, 0.0444, 0.0594, 0.0614, 0.0462, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 14:56:58,065 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171796.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:57:07,774 INFO [train.py:904] (7/8) Epoch 17, batch 9400, loss[loss=0.1434, simple_loss=0.2329, pruned_loss=0.02699, over 12332.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2609, pruned_loss=0.03745, over 3050239.35 frames. ], batch size: 246, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 14:58:21,326 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9769, 3.8039, 3.8972, 4.1720, 4.2241, 3.9000, 4.2868, 4.2948], device='cuda:7'), covar=tensor([0.1693, 0.1282, 0.1893, 0.0862, 0.0842, 0.1517, 0.0811, 0.0859], device='cuda:7'), in_proj_covar=tensor([0.0558, 0.0685, 0.0808, 0.0699, 0.0530, 0.0551, 0.0566, 0.0660], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 14:58:29,674 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.242e+02 2.701e+02 3.369e+02 7.708e+02, threshold=5.403e+02, percent-clipped=2.0 2023-04-30 14:58:48,379 INFO [train.py:904] (7/8) Epoch 17, batch 9450, loss[loss=0.1882, simple_loss=0.2776, pruned_loss=0.0494, over 16653.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2631, pruned_loss=0.03791, over 3043301.43 frames. ], batch size: 134, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 14:58:49,685 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=171852.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 14:58:53,018 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=171854.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:00:27,219 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=171900.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:00:31,023 INFO [train.py:904] (7/8) Epoch 17, batch 9500, loss[loss=0.1479, simple_loss=0.2378, pruned_loss=0.02894, over 12666.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2624, pruned_loss=0.03767, over 3048781.28 frames. ], batch size: 248, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:01:41,975 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=171937.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:01:42,587 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-04-30 15:01:51,142 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.125e+02 2.584e+02 3.095e+02 6.778e+02, threshold=5.168e+02, percent-clipped=1.0 2023-04-30 15:01:55,966 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-30 15:02:14,948 INFO [train.py:904] (7/8) Epoch 17, batch 9550, loss[loss=0.1707, simple_loss=0.2702, pruned_loss=0.03557, over 16366.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2619, pruned_loss=0.03749, over 3052312.58 frames. ], batch size: 146, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:03:49,317 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=171998.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:03:57,563 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172001.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:03:58,214 INFO [train.py:904] (7/8) Epoch 17, batch 9600, loss[loss=0.1829, simple_loss=0.266, pruned_loss=0.04993, over 12314.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2633, pruned_loss=0.03857, over 3022144.63 frames. ], batch size: 248, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 15:04:36,293 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5504, 3.8602, 4.0192, 3.9836, 3.9912, 3.7905, 3.5594, 3.8137], device='cuda:7'), covar=tensor([0.0658, 0.0742, 0.0638, 0.0676, 0.0743, 0.0655, 0.1372, 0.0589], device='cuda:7'), in_proj_covar=tensor([0.0364, 0.0393, 0.0388, 0.0367, 0.0435, 0.0407, 0.0495, 0.0323], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 15:05:20,545 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.253e+02 2.956e+02 3.661e+02 6.195e+02, threshold=5.912e+02, percent-clipped=3.0 2023-04-30 15:05:38,438 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172049.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:05:44,206 INFO [train.py:904] (7/8) Epoch 17, batch 9650, loss[loss=0.1694, simple_loss=0.2638, pruned_loss=0.03755, over 16903.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2653, pruned_loss=0.03853, over 3053608.79 frames. ], batch size: 116, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:07:32,492 INFO [train.py:904] (7/8) Epoch 17, batch 9700, loss[loss=0.182, simple_loss=0.2795, pruned_loss=0.04229, over 16350.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2651, pruned_loss=0.03866, over 3054667.50 frames. ], batch size: 146, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:08:06,091 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172118.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:08:57,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.330e+02 2.721e+02 3.165e+02 5.976e+02, threshold=5.442e+02, percent-clipped=1.0 2023-04-30 15:09:14,204 INFO [train.py:904] (7/8) Epoch 17, batch 9750, loss[loss=0.1605, simple_loss=0.2605, pruned_loss=0.03027, over 16252.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2643, pruned_loss=0.03898, over 3072992.43 frames. ], batch size: 165, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:09:18,922 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172154.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:10:10,078 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172179.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:10:54,102 INFO [train.py:904] (7/8) Epoch 17, batch 9800, loss[loss=0.1674, simple_loss=0.2719, pruned_loss=0.0315, over 16534.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2644, pruned_loss=0.03811, over 3076970.84 frames. ], batch size: 75, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:10:54,809 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172202.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:10:57,483 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 15:10:59,419 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0986, 2.5614, 2.6668, 1.8815, 2.7864, 2.8901, 2.4993, 2.4692], device='cuda:7'), covar=tensor([0.0632, 0.0234, 0.0190, 0.0958, 0.0093, 0.0193, 0.0413, 0.0375], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0100, 0.0086, 0.0132, 0.0071, 0.0111, 0.0119, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 15:11:34,503 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2111, 3.8871, 4.3042, 2.2459, 4.4400, 4.5110, 3.3663, 3.5328], device='cuda:7'), covar=tensor([0.0518, 0.0192, 0.0142, 0.1070, 0.0052, 0.0101, 0.0311, 0.0317], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0100, 0.0086, 0.0132, 0.0071, 0.0111, 0.0119, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 15:12:17,142 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.047e+02 2.504e+02 3.250e+02 5.993e+02, threshold=5.009e+02, percent-clipped=1.0 2023-04-30 15:12:39,002 INFO [train.py:904] (7/8) Epoch 17, batch 9850, loss[loss=0.16, simple_loss=0.2562, pruned_loss=0.03189, over 15412.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2648, pruned_loss=0.03799, over 3064193.53 frames. ], batch size: 191, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:13:51,832 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9844, 4.0109, 4.3212, 4.2864, 4.3012, 4.0761, 4.0521, 4.0588], device='cuda:7'), covar=tensor([0.0377, 0.1105, 0.0502, 0.0469, 0.0455, 0.0527, 0.0853, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0356, 0.0386, 0.0379, 0.0359, 0.0424, 0.0397, 0.0483, 0.0316], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 15:14:14,443 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172293.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:14:32,014 INFO [train.py:904] (7/8) Epoch 17, batch 9900, loss[loss=0.1884, simple_loss=0.2722, pruned_loss=0.05234, over 12661.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2651, pruned_loss=0.03807, over 3048504.14 frames. ], batch size: 247, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:15:39,334 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4691, 1.6041, 2.0566, 2.4860, 2.3335, 2.7465, 1.8218, 2.6998], device='cuda:7'), covar=tensor([0.0190, 0.0490, 0.0314, 0.0270, 0.0307, 0.0181, 0.0464, 0.0130], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0180, 0.0166, 0.0168, 0.0179, 0.0137, 0.0182, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:16:10,692 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.206e+02 2.589e+02 2.968e+02 8.308e+02, threshold=5.178e+02, percent-clipped=1.0 2023-04-30 15:16:31,673 INFO [train.py:904] (7/8) Epoch 17, batch 9950, loss[loss=0.172, simple_loss=0.2689, pruned_loss=0.03752, over 16234.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2671, pruned_loss=0.03813, over 3054891.86 frames. ], batch size: 165, lr: 3.94e-03, grad_scale: 4.0 2023-04-30 15:16:33,760 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-04-30 15:16:58,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3431, 3.4376, 3.6651, 3.6469, 3.6522, 3.4720, 3.5042, 3.5305], device='cuda:7'), covar=tensor([0.0397, 0.0767, 0.0506, 0.0495, 0.0492, 0.0515, 0.0777, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0356, 0.0386, 0.0380, 0.0360, 0.0425, 0.0397, 0.0484, 0.0317], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 15:18:01,688 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2023-04-30 15:18:33,057 INFO [train.py:904] (7/8) Epoch 17, batch 10000, loss[loss=0.1776, simple_loss=0.2759, pruned_loss=0.03968, over 16188.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2659, pruned_loss=0.03773, over 3077967.19 frames. ], batch size: 165, lr: 3.94e-03, grad_scale: 8.0 2023-04-30 15:19:56,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 2.112e+02 2.435e+02 3.073e+02 5.359e+02, threshold=4.869e+02, percent-clipped=1.0 2023-04-30 15:20:13,529 INFO [train.py:904] (7/8) Epoch 17, batch 10050, loss[loss=0.168, simple_loss=0.2601, pruned_loss=0.03795, over 16606.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2657, pruned_loss=0.03782, over 3067510.87 frames. ], batch size: 62, lr: 3.93e-03, grad_scale: 8.0 2023-04-30 15:20:36,186 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8801, 4.6857, 5.0033, 5.1053, 5.2686, 4.6463, 5.3298, 5.2529], device='cuda:7'), covar=tensor([0.1856, 0.1286, 0.1356, 0.0584, 0.0501, 0.0854, 0.0446, 0.0491], device='cuda:7'), in_proj_covar=tensor([0.0557, 0.0684, 0.0805, 0.0701, 0.0528, 0.0553, 0.0566, 0.0658], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:20:58,578 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172474.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:21:11,178 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5946, 3.0069, 3.3045, 2.0444, 2.7967, 2.1578, 3.1932, 3.2340], device='cuda:7'), covar=tensor([0.0275, 0.0778, 0.0502, 0.1863, 0.0779, 0.0996, 0.0625, 0.0902], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0147, 0.0158, 0.0145, 0.0136, 0.0123, 0.0136, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:21:47,148 INFO [train.py:904] (7/8) Epoch 17, batch 10100, loss[loss=0.1784, simple_loss=0.2652, pruned_loss=0.04586, over 16676.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2659, pruned_loss=0.03797, over 3061898.98 frames. ], batch size: 134, lr: 3.93e-03, grad_scale: 8.0 2023-04-30 15:22:57,879 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.137e+02 2.474e+02 3.171e+02 5.916e+02, threshold=4.949e+02, percent-clipped=1.0 2023-04-30 15:23:33,094 INFO [train.py:904] (7/8) Epoch 18, batch 0, loss[loss=0.1882, simple_loss=0.275, pruned_loss=0.0507, over 17228.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.275, pruned_loss=0.0507, over 17228.00 frames. ], batch size: 45, lr: 3.82e-03, grad_scale: 8.0 2023-04-30 15:23:33,094 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 15:23:40,343 INFO [train.py:938] (7/8) Epoch 18, validation: loss=0.148, simple_loss=0.2515, pruned_loss=0.02223, over 944034.00 frames. 2023-04-30 15:23:40,344 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 15:24:36,457 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172593.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:24:44,254 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172598.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:24:50,180 INFO [train.py:904] (7/8) Epoch 18, batch 50, loss[loss=0.204, simple_loss=0.2771, pruned_loss=0.06543, over 16892.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2745, pruned_loss=0.0526, over 753541.82 frames. ], batch size: 109, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:25:21,795 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172625.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:25:42,963 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172641.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:25:48,588 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.462e+02 2.938e+02 3.654e+02 9.185e+02, threshold=5.877e+02, percent-clipped=7.0 2023-04-30 15:25:50,869 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1181, 3.9225, 4.2050, 4.3176, 4.4201, 4.0525, 4.2408, 4.4068], device='cuda:7'), covar=tensor([0.1598, 0.1188, 0.1222, 0.0687, 0.0612, 0.1162, 0.2073, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0561, 0.0694, 0.0813, 0.0708, 0.0530, 0.0558, 0.0573, 0.0665], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:25:56,043 INFO [train.py:904] (7/8) Epoch 18, batch 100, loss[loss=0.2164, simple_loss=0.2907, pruned_loss=0.07104, over 16481.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2677, pruned_loss=0.04778, over 1330113.75 frames. ], batch size: 68, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:25:59,588 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8510, 2.1083, 2.3705, 3.1777, 2.1957, 2.2536, 2.2610, 2.2171], device='cuda:7'), covar=tensor([0.1184, 0.3207, 0.2316, 0.0637, 0.3732, 0.2379, 0.3109, 0.2989], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0415, 0.0347, 0.0311, 0.0418, 0.0474, 0.0385, 0.0481], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:26:05,974 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172659.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:26:44,111 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:27:02,728 INFO [train.py:904] (7/8) Epoch 18, batch 150, loss[loss=0.1988, simple_loss=0.2863, pruned_loss=0.05563, over 17058.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2668, pruned_loss=0.04782, over 1769320.40 frames. ], batch size: 55, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:27:46,377 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172734.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:27:52,988 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-04-30 15:28:01,450 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.271e+02 2.721e+02 3.424e+02 7.505e+02, threshold=5.441e+02, percent-clipped=3.0 2023-04-30 15:28:09,987 INFO [train.py:904] (7/8) Epoch 18, batch 200, loss[loss=0.2033, simple_loss=0.286, pruned_loss=0.06027, over 16517.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2688, pruned_loss=0.0489, over 2105014.17 frames. ], batch size: 68, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:28:36,209 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5486, 3.3924, 3.8347, 2.8237, 3.5343, 3.9493, 3.6343, 2.4850], device='cuda:7'), covar=tensor([0.0486, 0.0300, 0.0053, 0.0337, 0.0110, 0.0092, 0.0091, 0.0416], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0077, 0.0077, 0.0132, 0.0091, 0.0100, 0.0088, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 15:28:41,387 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=172774.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:29:10,036 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172795.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:29:10,435 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-04-30 15:29:18,641 INFO [train.py:904] (7/8) Epoch 18, batch 250, loss[loss=0.1991, simple_loss=0.2719, pruned_loss=0.06317, over 16880.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2674, pruned_loss=0.0487, over 2373018.43 frames. ], batch size: 90, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:29:29,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3913, 5.3177, 5.0747, 4.6535, 5.1386, 2.0202, 4.9499, 5.1153], device='cuda:7'), covar=tensor([0.0070, 0.0070, 0.0191, 0.0342, 0.0099, 0.2439, 0.0127, 0.0186], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0138, 0.0181, 0.0164, 0.0157, 0.0196, 0.0170, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:29:47,793 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=172822.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:30:18,870 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=172845.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:30:19,551 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.240e+02 2.726e+02 3.152e+02 7.045e+02, threshold=5.451e+02, percent-clipped=2.0 2023-04-30 15:30:23,649 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2266, 3.2699, 3.4446, 2.3475, 3.0460, 2.3709, 3.6980, 3.6370], device='cuda:7'), covar=tensor([0.0253, 0.0898, 0.0591, 0.1744, 0.0807, 0.0953, 0.0505, 0.0920], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0154, 0.0163, 0.0150, 0.0141, 0.0127, 0.0140, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:30:29,003 INFO [train.py:904] (7/8) Epoch 18, batch 300, loss[loss=0.1756, simple_loss=0.2745, pruned_loss=0.03841, over 17108.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2639, pruned_loss=0.04726, over 2584291.31 frames. ], batch size: 49, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:31:32,731 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9092, 1.9661, 2.4604, 2.8242, 2.7923, 3.1328, 2.1401, 3.1041], device='cuda:7'), covar=tensor([0.0216, 0.0470, 0.0304, 0.0268, 0.0261, 0.0217, 0.0474, 0.0171], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0185, 0.0170, 0.0172, 0.0182, 0.0141, 0.0186, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:31:39,807 INFO [train.py:904] (7/8) Epoch 18, batch 350, loss[loss=0.1699, simple_loss=0.2743, pruned_loss=0.03278, over 17105.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2613, pruned_loss=0.04645, over 2749789.29 frames. ], batch size: 49, lr: 3.82e-03, grad_scale: 1.0 2023-04-30 15:31:45,352 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=172906.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:31:50,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7522, 2.5360, 2.0302, 2.2978, 2.9205, 2.6209, 2.9344, 2.9930], device='cuda:7'), covar=tensor([0.0228, 0.0368, 0.0513, 0.0433, 0.0217, 0.0317, 0.0201, 0.0247], device='cuda:7'), in_proj_covar=tensor([0.0189, 0.0229, 0.0220, 0.0222, 0.0228, 0.0227, 0.0228, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:31:54,987 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4383, 1.6606, 2.0894, 2.2314, 2.4349, 2.3525, 1.8279, 2.4373], device='cuda:7'), covar=tensor([0.0220, 0.0497, 0.0276, 0.0282, 0.0290, 0.0292, 0.0466, 0.0176], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0185, 0.0170, 0.0173, 0.0183, 0.0141, 0.0186, 0.0135], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:32:40,486 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.138e+02 2.555e+02 2.961e+02 4.551e+02, threshold=5.109e+02, percent-clipped=0.0 2023-04-30 15:32:47,906 INFO [train.py:904] (7/8) Epoch 18, batch 400, loss[loss=0.1742, simple_loss=0.2601, pruned_loss=0.04411, over 16475.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2598, pruned_loss=0.04606, over 2874900.29 frames. ], batch size: 68, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:32:51,730 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172954.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:33:11,059 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3604, 4.2477, 4.2551, 3.9653, 4.0017, 4.2904, 4.0113, 4.0668], device='cuda:7'), covar=tensor([0.0581, 0.0670, 0.0318, 0.0269, 0.0728, 0.0497, 0.0712, 0.0664], device='cuda:7'), in_proj_covar=tensor([0.0273, 0.0384, 0.0321, 0.0308, 0.0330, 0.0358, 0.0219, 0.0381], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:33:30,476 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=172981.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:33:44,188 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-04-30 15:33:59,541 INFO [train.py:904] (7/8) Epoch 18, batch 450, loss[loss=0.1581, simple_loss=0.2342, pruned_loss=0.04105, over 15797.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2577, pruned_loss=0.04478, over 2973494.13 frames. ], batch size: 35, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:35:00,084 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.073e+02 2.463e+02 2.995e+02 6.633e+02, threshold=4.925e+02, percent-clipped=1.0 2023-04-30 15:35:08,313 INFO [train.py:904] (7/8) Epoch 18, batch 500, loss[loss=0.1734, simple_loss=0.2587, pruned_loss=0.04403, over 16680.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2562, pruned_loss=0.04418, over 3048995.44 frames. ], batch size: 62, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:35:22,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0351, 3.9657, 4.4778, 2.1033, 4.6577, 4.6457, 3.3059, 3.5793], device='cuda:7'), covar=tensor([0.0715, 0.0216, 0.0182, 0.1213, 0.0059, 0.0164, 0.0415, 0.0389], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0107, 0.0093, 0.0141, 0.0076, 0.0121, 0.0127, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:35:28,192 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4295, 3.7113, 4.1068, 2.3355, 3.3014, 2.5813, 3.8927, 3.8537], device='cuda:7'), covar=tensor([0.0280, 0.0880, 0.0456, 0.1909, 0.0774, 0.0941, 0.0607, 0.1003], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0155, 0.0163, 0.0150, 0.0141, 0.0126, 0.0140, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:35:59,657 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173090.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:36:14,716 INFO [train.py:904] (7/8) Epoch 18, batch 550, loss[loss=0.1566, simple_loss=0.2418, pruned_loss=0.03565, over 16768.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2557, pruned_loss=0.04395, over 3113498.68 frames. ], batch size: 83, lr: 3.82e-03, grad_scale: 2.0 2023-04-30 15:36:34,766 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-04-30 15:37:14,572 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.114e+02 2.403e+02 3.161e+02 6.470e+02, threshold=4.805e+02, percent-clipped=1.0 2023-04-30 15:37:20,900 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173150.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:37:22,691 INFO [train.py:904] (7/8) Epoch 18, batch 600, loss[loss=0.1863, simple_loss=0.2526, pruned_loss=0.06003, over 16877.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.255, pruned_loss=0.04393, over 3160395.66 frames. ], batch size: 116, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:37:45,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5660, 2.0362, 2.1607, 4.1988, 2.0844, 2.3649, 2.1451, 2.1630], device='cuda:7'), covar=tensor([0.1357, 0.4577, 0.3301, 0.0606, 0.5185, 0.3384, 0.3961, 0.4618], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0425, 0.0354, 0.0320, 0.0425, 0.0488, 0.0394, 0.0494], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:37:48,697 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5859, 4.7032, 4.8109, 4.6669, 4.6676, 5.2828, 4.7814, 4.4506], device='cuda:7'), covar=tensor([0.1680, 0.2271, 0.2710, 0.2565, 0.3044, 0.1273, 0.1933, 0.2863], device='cuda:7'), in_proj_covar=tensor([0.0386, 0.0564, 0.0619, 0.0468, 0.0634, 0.0653, 0.0492, 0.0631], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 15:38:08,467 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.73 vs. limit=5.0 2023-04-30 15:38:30,590 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173201.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:38:31,387 INFO [train.py:904] (7/8) Epoch 18, batch 650, loss[loss=0.1758, simple_loss=0.2481, pruned_loss=0.05174, over 16863.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2535, pruned_loss=0.04385, over 3193645.53 frames. ], batch size: 96, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:38:45,533 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173211.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:39:32,448 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.711e+02 2.266e+02 2.757e+02 3.739e+02 8.562e+02, threshold=5.513e+02, percent-clipped=7.0 2023-04-30 15:39:32,837 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6332, 4.5212, 4.4850, 4.1675, 4.1656, 4.5405, 4.4045, 4.2508], device='cuda:7'), covar=tensor([0.0632, 0.0730, 0.0316, 0.0300, 0.0926, 0.0495, 0.0499, 0.0702], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0396, 0.0329, 0.0317, 0.0339, 0.0368, 0.0225, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:39:40,842 INFO [train.py:904] (7/8) Epoch 18, batch 700, loss[loss=0.1956, simple_loss=0.2659, pruned_loss=0.06262, over 16871.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2535, pruned_loss=0.04375, over 3217190.08 frames. ], batch size: 109, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:39:44,112 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173254.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:39:58,140 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9748, 5.0292, 5.4939, 5.4865, 5.4874, 5.1144, 5.0574, 4.9199], device='cuda:7'), covar=tensor([0.0329, 0.0599, 0.0382, 0.0443, 0.0451, 0.0369, 0.0933, 0.0442], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0422, 0.0413, 0.0390, 0.0456, 0.0434, 0.0528, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 15:40:22,683 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173281.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:40:50,346 INFO [train.py:904] (7/8) Epoch 18, batch 750, loss[loss=0.176, simple_loss=0.2518, pruned_loss=0.05009, over 16250.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2537, pruned_loss=0.04374, over 3226456.19 frames. ], batch size: 164, lr: 3.81e-03, grad_scale: 2.0 2023-04-30 15:40:51,631 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173302.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:41:00,075 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173309.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:41:26,735 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173329.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:41:50,588 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.162e+02 2.529e+02 2.992e+02 5.194e+02, threshold=5.057e+02, percent-clipped=0.0 2023-04-30 15:41:58,289 INFO [train.py:904] (7/8) Epoch 18, batch 800, loss[loss=0.1488, simple_loss=0.2352, pruned_loss=0.03119, over 16829.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2534, pruned_loss=0.04348, over 3251216.42 frames. ], batch size: 42, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:42:22,701 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173370.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:42:49,215 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173390.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:43:05,174 INFO [train.py:904] (7/8) Epoch 18, batch 850, loss[loss=0.1651, simple_loss=0.2458, pruned_loss=0.04218, over 16750.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2534, pruned_loss=0.04364, over 3267367.33 frames. ], batch size: 102, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:43:19,661 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3892, 5.7513, 5.4716, 5.5311, 5.0407, 5.1682, 5.1925, 5.8680], device='cuda:7'), covar=tensor([0.1282, 0.0986, 0.1169, 0.0896, 0.0958, 0.0719, 0.1218, 0.0930], device='cuda:7'), in_proj_covar=tensor([0.0653, 0.0798, 0.0645, 0.0591, 0.0503, 0.0506, 0.0662, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:43:55,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173438.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:44:07,465 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.076e+02 2.484e+02 3.047e+02 7.232e+02, threshold=4.969e+02, percent-clipped=3.0 2023-04-30 15:44:15,666 INFO [train.py:904] (7/8) Epoch 18, batch 900, loss[loss=0.1504, simple_loss=0.2385, pruned_loss=0.03113, over 17169.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2525, pruned_loss=0.04301, over 3278378.54 frames. ], batch size: 46, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:45:11,327 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173491.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 15:45:24,867 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173501.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:45:25,658 INFO [train.py:904] (7/8) Epoch 18, batch 950, loss[loss=0.1678, simple_loss=0.2659, pruned_loss=0.03488, over 17130.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.252, pruned_loss=0.04292, over 3269937.67 frames. ], batch size: 49, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:45:27,906 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3340, 3.3851, 2.1643, 3.5466, 2.6175, 3.5480, 2.1042, 2.6801], device='cuda:7'), covar=tensor([0.0267, 0.0389, 0.1481, 0.0317, 0.0755, 0.0695, 0.1419, 0.0740], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0174, 0.0197, 0.0156, 0.0175, 0.0215, 0.0204, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:45:31,804 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173506.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:46:26,060 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.297e+02 2.589e+02 3.111e+02 6.149e+02, threshold=5.178e+02, percent-clipped=2.0 2023-04-30 15:46:31,946 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173549.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:46:35,249 INFO [train.py:904] (7/8) Epoch 18, batch 1000, loss[loss=0.1673, simple_loss=0.2601, pruned_loss=0.03725, over 16681.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2514, pruned_loss=0.04273, over 3271369.18 frames. ], batch size: 57, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:46:35,737 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173552.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 15:47:12,574 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7308, 2.4390, 2.3294, 3.7388, 2.9718, 3.8224, 1.5122, 2.8443], device='cuda:7'), covar=tensor([0.1405, 0.0732, 0.1197, 0.0224, 0.0162, 0.0398, 0.1632, 0.0818], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0167, 0.0188, 0.0175, 0.0198, 0.0213, 0.0193, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:47:37,041 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4549, 5.4010, 5.0987, 4.6143, 5.2406, 2.2260, 5.0077, 5.1024], device='cuda:7'), covar=tensor([0.0070, 0.0078, 0.0191, 0.0357, 0.0092, 0.2377, 0.0117, 0.0168], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0143, 0.0188, 0.0170, 0.0163, 0.0201, 0.0177, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:47:44,522 INFO [train.py:904] (7/8) Epoch 18, batch 1050, loss[loss=0.1616, simple_loss=0.2407, pruned_loss=0.04125, over 16859.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2511, pruned_loss=0.04249, over 3285001.81 frames. ], batch size: 96, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:47:53,775 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0512, 2.1433, 2.6349, 3.0504, 2.8168, 3.4959, 2.4202, 3.3972], device='cuda:7'), covar=tensor([0.0223, 0.0432, 0.0315, 0.0262, 0.0297, 0.0158, 0.0436, 0.0150], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0187, 0.0173, 0.0176, 0.0184, 0.0144, 0.0190, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:48:02,534 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=173614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:48:46,278 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.302e+02 2.654e+02 3.076e+02 5.985e+02, threshold=5.308e+02, percent-clipped=1.0 2023-04-30 15:48:54,961 INFO [train.py:904] (7/8) Epoch 18, batch 1100, loss[loss=0.1886, simple_loss=0.2588, pruned_loss=0.05924, over 17211.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2504, pruned_loss=0.0421, over 3289707.39 frames. ], batch size: 46, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:49:12,458 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173665.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:49:19,244 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4215, 3.1764, 3.4987, 1.9277, 3.6281, 3.5826, 2.9533, 2.7258], device='cuda:7'), covar=tensor([0.0777, 0.0233, 0.0199, 0.1133, 0.0096, 0.0186, 0.0410, 0.0436], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0107, 0.0094, 0.0140, 0.0076, 0.0121, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 15:49:26,677 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=173675.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 15:50:01,552 INFO [train.py:904] (7/8) Epoch 18, batch 1150, loss[loss=0.1666, simple_loss=0.2604, pruned_loss=0.03637, over 17045.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.25, pruned_loss=0.04179, over 3298326.56 frames. ], batch size: 50, lr: 3.81e-03, grad_scale: 4.0 2023-04-30 15:51:02,216 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.041e+02 2.391e+02 2.834e+02 4.820e+02, threshold=4.781e+02, percent-clipped=0.0 2023-04-30 15:51:11,442 INFO [train.py:904] (7/8) Epoch 18, batch 1200, loss[loss=0.163, simple_loss=0.24, pruned_loss=0.04299, over 16863.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2491, pruned_loss=0.04146, over 3312125.83 frames. ], batch size: 96, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:51:54,511 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5161, 2.3188, 2.3588, 4.4345, 2.2579, 2.7589, 2.4116, 2.4684], device='cuda:7'), covar=tensor([0.1221, 0.3752, 0.2966, 0.0483, 0.4201, 0.2599, 0.3358, 0.3784], device='cuda:7'), in_proj_covar=tensor([0.0389, 0.0429, 0.0358, 0.0325, 0.0429, 0.0493, 0.0399, 0.0501], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:52:19,921 INFO [train.py:904] (7/8) Epoch 18, batch 1250, loss[loss=0.1949, simple_loss=0.2742, pruned_loss=0.05778, over 16537.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2497, pruned_loss=0.04172, over 3314650.38 frames. ], batch size: 146, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:52:24,737 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173806.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:53:21,078 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.313e+02 2.602e+02 3.076e+02 5.284e+02, threshold=5.204e+02, percent-clipped=3.0 2023-04-30 15:53:22,683 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173847.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 15:53:28,951 INFO [train.py:904] (7/8) Epoch 18, batch 1300, loss[loss=0.1641, simple_loss=0.2587, pruned_loss=0.03474, over 17044.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2494, pruned_loss=0.04174, over 3307562.52 frames. ], batch size: 55, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:53:32,736 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=173854.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:53:42,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6568, 4.6473, 5.0808, 5.0700, 5.1135, 4.7718, 4.7541, 4.5393], device='cuda:7'), covar=tensor([0.0401, 0.0762, 0.0474, 0.0441, 0.0468, 0.0401, 0.0853, 0.0602], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0430, 0.0419, 0.0394, 0.0464, 0.0441, 0.0535, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 15:54:09,989 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9157, 5.2548, 5.0566, 5.0437, 4.7798, 4.7617, 4.7025, 5.3531], device='cuda:7'), covar=tensor([0.1183, 0.0876, 0.0979, 0.0863, 0.0832, 0.0916, 0.1105, 0.0905], device='cuda:7'), in_proj_covar=tensor([0.0651, 0.0799, 0.0644, 0.0593, 0.0503, 0.0505, 0.0663, 0.0617], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 15:54:37,711 INFO [train.py:904] (7/8) Epoch 18, batch 1350, loss[loss=0.1823, simple_loss=0.2761, pruned_loss=0.04427, over 16737.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2505, pruned_loss=0.04145, over 3312436.34 frames. ], batch size: 57, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:55:06,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4846, 3.6703, 4.1814, 2.3319, 3.3924, 2.5711, 4.0000, 4.0262], device='cuda:7'), covar=tensor([0.0318, 0.0930, 0.0469, 0.1939, 0.0727, 0.0934, 0.0597, 0.1073], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0158, 0.0164, 0.0151, 0.0142, 0.0127, 0.0142, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:55:37,173 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.229e+02 2.739e+02 3.130e+02 5.090e+02, threshold=5.477e+02, percent-clipped=0.0 2023-04-30 15:55:44,247 INFO [train.py:904] (7/8) Epoch 18, batch 1400, loss[loss=0.1519, simple_loss=0.244, pruned_loss=0.02989, over 17168.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2507, pruned_loss=0.04137, over 3318604.38 frames. ], batch size: 46, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:56:04,280 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=173965.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:56:09,770 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=173970.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 15:56:27,627 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 15:56:56,570 INFO [train.py:904] (7/8) Epoch 18, batch 1450, loss[loss=0.1841, simple_loss=0.2611, pruned_loss=0.0536, over 16485.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2499, pruned_loss=0.04105, over 3319662.98 frames. ], batch size: 68, lr: 3.81e-03, grad_scale: 8.0 2023-04-30 15:57:11,471 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174013.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:57:14,618 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174015.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:57:16,922 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174017.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:57:54,756 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.047e+02 2.425e+02 3.116e+02 5.904e+02, threshold=4.850e+02, percent-clipped=1.0 2023-04-30 15:57:55,681 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.59 vs. limit=5.0 2023-04-30 15:58:03,190 INFO [train.py:904] (7/8) Epoch 18, batch 1500, loss[loss=0.1599, simple_loss=0.2403, pruned_loss=0.03976, over 16862.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2496, pruned_loss=0.04128, over 3328848.30 frames. ], batch size: 96, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 15:58:36,095 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174076.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 15:58:38,853 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174078.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 15:59:05,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4661, 3.9814, 3.9429, 2.2872, 3.2670, 2.7174, 3.9708, 4.0412], device='cuda:7'), covar=tensor([0.0322, 0.0731, 0.0553, 0.1910, 0.0779, 0.0890, 0.0630, 0.1084], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0158, 0.0164, 0.0151, 0.0141, 0.0126, 0.0141, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:59:10,943 INFO [train.py:904] (7/8) Epoch 18, batch 1550, loss[loss=0.155, simple_loss=0.2428, pruned_loss=0.03366, over 17138.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.251, pruned_loss=0.04219, over 3334233.41 frames. ], batch size: 47, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 15:59:25,680 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8350, 4.0091, 2.5169, 4.4852, 3.0467, 4.4707, 2.6447, 3.3051], device='cuda:7'), covar=tensor([0.0275, 0.0339, 0.1421, 0.0255, 0.0767, 0.0479, 0.1337, 0.0647], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0175, 0.0196, 0.0159, 0.0175, 0.0217, 0.0204, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 15:59:27,232 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.60 vs. limit=5.0 2023-04-30 15:59:45,491 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 16:00:00,498 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174139.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:00:09,839 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.219e+02 2.746e+02 3.375e+02 7.585e+02, threshold=5.492e+02, percent-clipped=5.0 2023-04-30 16:00:11,878 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174147.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 16:00:17,420 INFO [train.py:904] (7/8) Epoch 18, batch 1600, loss[loss=0.1864, simple_loss=0.2739, pruned_loss=0.04941, over 17125.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2532, pruned_loss=0.04275, over 3336062.14 frames. ], batch size: 55, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:01:16,803 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174195.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 16:01:23,755 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174200.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:01:25,629 INFO [train.py:904] (7/8) Epoch 18, batch 1650, loss[loss=0.2135, simple_loss=0.2928, pruned_loss=0.06708, over 12258.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2551, pruned_loss=0.04352, over 3327085.02 frames. ], batch size: 248, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:01:46,578 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4598, 2.0744, 2.2484, 4.2197, 2.0743, 2.4445, 2.1999, 2.2767], device='cuda:7'), covar=tensor([0.1369, 0.4610, 0.3313, 0.0567, 0.5291, 0.3228, 0.4056, 0.4543], device='cuda:7'), in_proj_covar=tensor([0.0390, 0.0431, 0.0360, 0.0327, 0.0431, 0.0496, 0.0401, 0.0504], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:02:23,842 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.550e+02 2.934e+02 3.695e+02 6.371e+02, threshold=5.868e+02, percent-clipped=1.0 2023-04-30 16:02:32,766 INFO [train.py:904] (7/8) Epoch 18, batch 1700, loss[loss=0.1922, simple_loss=0.2788, pruned_loss=0.05275, over 16534.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2569, pruned_loss=0.04437, over 3314860.20 frames. ], batch size: 75, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:02:57,280 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174270.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 16:03:15,681 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174283.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:03:22,128 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8152, 3.7122, 3.8481, 3.5965, 3.7330, 4.2462, 3.9022, 3.5486], device='cuda:7'), covar=tensor([0.2144, 0.2362, 0.2428, 0.2668, 0.3177, 0.1828, 0.1563, 0.2512], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0576, 0.0631, 0.0475, 0.0646, 0.0661, 0.0496, 0.0646], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:03:25,144 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174291.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:03:40,508 INFO [train.py:904] (7/8) Epoch 18, batch 1750, loss[loss=0.1599, simple_loss=0.2568, pruned_loss=0.0315, over 17129.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2577, pruned_loss=0.04445, over 3313710.09 frames. ], batch size: 48, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:04:01,307 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174318.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:04:36,740 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:04:38,536 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.659e+02 2.402e+02 2.710e+02 3.242e+02 7.254e+02, threshold=5.419e+02, percent-clipped=1.0 2023-04-30 16:04:46,165 INFO [train.py:904] (7/8) Epoch 18, batch 1800, loss[loss=0.1737, simple_loss=0.2676, pruned_loss=0.03992, over 16745.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2587, pruned_loss=0.04419, over 3315173.43 frames. ], batch size: 57, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:04:46,552 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174352.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 16:05:10,976 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174371.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 16:05:13,935 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174373.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:05:33,494 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174388.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:05:51,529 INFO [train.py:904] (7/8) Epoch 18, batch 1850, loss[loss=0.1796, simple_loss=0.2562, pruned_loss=0.05149, over 16727.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2589, pruned_loss=0.04398, over 3314485.59 frames. ], batch size: 89, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:06:06,122 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-04-30 16:06:40,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9917, 5.0026, 4.8678, 4.4435, 4.3243, 4.9971, 4.8928, 4.4830], device='cuda:7'), covar=tensor([0.0753, 0.0626, 0.0412, 0.0430, 0.1489, 0.0559, 0.0447, 0.0908], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0410, 0.0339, 0.0329, 0.0351, 0.0383, 0.0234, 0.0408], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:06:51,032 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.281e+02 2.184e+02 2.475e+02 2.961e+02 4.707e+02, threshold=4.951e+02, percent-clipped=0.0 2023-04-30 16:06:55,557 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174449.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:06:58,378 INFO [train.py:904] (7/8) Epoch 18, batch 1900, loss[loss=0.1511, simple_loss=0.242, pruned_loss=0.03012, over 17139.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2579, pruned_loss=0.04289, over 3319661.35 frames. ], batch size: 48, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:07:33,386 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 16:07:49,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0070, 3.1263, 3.1531, 2.1125, 2.7673, 2.2173, 3.4828, 3.4932], device='cuda:7'), covar=tensor([0.0218, 0.0862, 0.0631, 0.1743, 0.0897, 0.0974, 0.0504, 0.0821], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0157, 0.0163, 0.0150, 0.0141, 0.0126, 0.0141, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 16:07:53,500 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8857, 5.2077, 4.9881, 4.9859, 4.7453, 4.6753, 4.6365, 5.3149], device='cuda:7'), covar=tensor([0.1147, 0.0902, 0.0965, 0.0829, 0.0814, 0.1003, 0.1185, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0656, 0.0807, 0.0648, 0.0598, 0.0508, 0.0508, 0.0670, 0.0621], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:07:57,455 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174495.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:08:06,714 INFO [train.py:904] (7/8) Epoch 18, batch 1950, loss[loss=0.1423, simple_loss=0.2385, pruned_loss=0.02309, over 16815.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2578, pruned_loss=0.04234, over 3322711.84 frames. ], batch size: 42, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:08:17,057 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-04-30 16:09:05,471 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.251e+02 2.513e+02 3.165e+02 5.916e+02, threshold=5.027e+02, percent-clipped=3.0 2023-04-30 16:09:14,580 INFO [train.py:904] (7/8) Epoch 18, batch 2000, loss[loss=0.1448, simple_loss=0.2331, pruned_loss=0.02825, over 17224.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2575, pruned_loss=0.04242, over 3316797.15 frames. ], batch size: 44, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:09:33,732 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9182, 5.3236, 5.0791, 5.0634, 4.7965, 4.6920, 4.7583, 5.4277], device='cuda:7'), covar=tensor([0.1353, 0.0915, 0.1105, 0.0850, 0.0866, 0.1085, 0.1209, 0.0935], device='cuda:7'), in_proj_covar=tensor([0.0657, 0.0810, 0.0651, 0.0601, 0.0511, 0.0511, 0.0672, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:10:23,411 INFO [train.py:904] (7/8) Epoch 18, batch 2050, loss[loss=0.1955, simple_loss=0.2703, pruned_loss=0.06034, over 16710.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2579, pruned_loss=0.04329, over 3305796.05 frames. ], batch size: 134, lr: 3.80e-03, grad_scale: 16.0 2023-04-30 16:10:35,629 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9109, 1.9033, 2.4792, 2.8584, 2.6677, 3.3210, 2.3004, 3.3145], device='cuda:7'), covar=tensor([0.0218, 0.0499, 0.0291, 0.0305, 0.0302, 0.0165, 0.0452, 0.0173], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0190, 0.0176, 0.0181, 0.0188, 0.0148, 0.0193, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:10:39,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1983, 2.9044, 3.1650, 1.7316, 3.3122, 3.2439, 2.6566, 2.5503], device='cuda:7'), covar=tensor([0.0816, 0.0255, 0.0231, 0.1114, 0.0099, 0.0228, 0.0462, 0.0432], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0106, 0.0094, 0.0139, 0.0076, 0.0121, 0.0125, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 16:11:17,548 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174639.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:11:26,653 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.708e+02 2.224e+02 2.527e+02 3.035e+02 6.825e+02, threshold=5.053e+02, percent-clipped=2.0 2023-04-30 16:11:28,705 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174647.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 16:11:35,066 INFO [train.py:904] (7/8) Epoch 18, batch 2100, loss[loss=0.1797, simple_loss=0.2765, pruned_loss=0.04143, over 17080.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2583, pruned_loss=0.04339, over 3303881.55 frames. ], batch size: 53, lr: 3.80e-03, grad_scale: 16.0 2023-04-30 16:12:02,043 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174671.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 16:12:06,238 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174673.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:12:29,973 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 16:12:32,491 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1888, 5.5268, 5.3043, 5.3100, 5.0117, 4.9143, 5.0143, 5.6439], device='cuda:7'), covar=tensor([0.1063, 0.0832, 0.0870, 0.0756, 0.0776, 0.0841, 0.1050, 0.0777], device='cuda:7'), in_proj_covar=tensor([0.0656, 0.0809, 0.0650, 0.0600, 0.0508, 0.0511, 0.0669, 0.0623], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:12:45,818 INFO [train.py:904] (7/8) Epoch 18, batch 2150, loss[loss=0.156, simple_loss=0.2497, pruned_loss=0.03114, over 17211.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2591, pruned_loss=0.04359, over 3316036.02 frames. ], batch size: 45, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:12:50,398 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=174705.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:10,644 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174719.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:13,524 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174721.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:45,455 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=174744.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:13:49,508 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.235e+02 2.673e+02 3.216e+02 6.695e+02, threshold=5.347e+02, percent-clipped=2.0 2023-04-30 16:13:56,941 INFO [train.py:904] (7/8) Epoch 18, batch 2200, loss[loss=0.1864, simple_loss=0.2658, pruned_loss=0.05348, over 16538.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2593, pruned_loss=0.04395, over 3309511.77 frames. ], batch size: 68, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:14:16,338 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=174766.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:14:56,080 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174795.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:15:01,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5559, 2.2579, 2.2132, 4.4401, 2.2979, 2.6884, 2.3415, 2.4945], device='cuda:7'), covar=tensor([0.1163, 0.3673, 0.3042, 0.0460, 0.3990, 0.2611, 0.3413, 0.3468], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0432, 0.0360, 0.0327, 0.0430, 0.0498, 0.0401, 0.0505], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:15:04,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1559, 5.2383, 5.6343, 5.6358, 5.6657, 5.2572, 5.1903, 4.9884], device='cuda:7'), covar=tensor([0.0350, 0.0544, 0.0429, 0.0383, 0.0494, 0.0382, 0.1002, 0.0478], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0431, 0.0419, 0.0393, 0.0467, 0.0442, 0.0535, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 16:15:06,366 INFO [train.py:904] (7/8) Epoch 18, batch 2250, loss[loss=0.1751, simple_loss=0.2714, pruned_loss=0.03933, over 17129.00 frames. ], tot_loss[loss=0.174, simple_loss=0.26, pruned_loss=0.044, over 3315387.43 frames. ], batch size: 49, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:15:25,000 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6639, 6.0528, 5.8008, 5.9119, 5.4077, 5.4138, 5.4360, 6.1956], device='cuda:7'), covar=tensor([0.1298, 0.0983, 0.1120, 0.0785, 0.0950, 0.0625, 0.1204, 0.0869], device='cuda:7'), in_proj_covar=tensor([0.0657, 0.0813, 0.0652, 0.0602, 0.0509, 0.0513, 0.0671, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:16:02,782 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174843.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:16:07,688 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.164e+02 2.481e+02 3.118e+02 6.907e+02, threshold=4.963e+02, percent-clipped=2.0 2023-04-30 16:16:14,126 INFO [train.py:904] (7/8) Epoch 18, batch 2300, loss[loss=0.1747, simple_loss=0.2475, pruned_loss=0.05095, over 16922.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2593, pruned_loss=0.04358, over 3323845.33 frames. ], batch size: 109, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:16:31,373 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2924, 5.2985, 5.1839, 4.7133, 4.7764, 5.2255, 5.1569, 4.8410], device='cuda:7'), covar=tensor([0.0637, 0.0457, 0.0260, 0.0313, 0.1017, 0.0455, 0.0265, 0.0691], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0410, 0.0340, 0.0331, 0.0354, 0.0383, 0.0235, 0.0409], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:17:22,142 INFO [train.py:904] (7/8) Epoch 18, batch 2350, loss[loss=0.171, simple_loss=0.2678, pruned_loss=0.03708, over 17146.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2593, pruned_loss=0.04414, over 3327426.20 frames. ], batch size: 48, lr: 3.80e-03, grad_scale: 8.0 2023-04-30 16:18:14,658 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174939.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:18:25,287 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.298e+02 2.659e+02 3.331e+02 6.396e+02, threshold=5.319e+02, percent-clipped=2.0 2023-04-30 16:18:25,689 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=174947.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 16:18:27,392 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-30 16:18:32,773 INFO [train.py:904] (7/8) Epoch 18, batch 2400, loss[loss=0.1798, simple_loss=0.2702, pruned_loss=0.04471, over 15884.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2606, pruned_loss=0.04436, over 3328286.75 frames. ], batch size: 35, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:19:21,462 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174987.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:19:33,420 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=174995.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:19:42,218 INFO [train.py:904] (7/8) Epoch 18, batch 2450, loss[loss=0.2051, simple_loss=0.2873, pruned_loss=0.06149, over 16778.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2614, pruned_loss=0.0443, over 3326609.59 frames. ], batch size: 124, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:20:04,114 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7921, 4.3890, 4.3046, 3.0836, 3.6646, 4.3617, 3.9094, 2.3663], device='cuda:7'), covar=tensor([0.0446, 0.0055, 0.0046, 0.0322, 0.0117, 0.0067, 0.0076, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0080, 0.0080, 0.0133, 0.0093, 0.0104, 0.0091, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:20:08,219 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8652, 4.2937, 3.2747, 2.3470, 2.7688, 2.5547, 4.7066, 3.7420], device='cuda:7'), covar=tensor([0.2677, 0.0631, 0.1626, 0.2706, 0.2729, 0.1988, 0.0342, 0.1282], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0266, 0.0302, 0.0303, 0.0292, 0.0248, 0.0288, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:20:34,637 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-04-30 16:20:40,467 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175044.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:20:43,746 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.297e+02 2.808e+02 3.204e+02 6.791e+02, threshold=5.617e+02, percent-clipped=2.0 2023-04-30 16:20:51,501 INFO [train.py:904] (7/8) Epoch 18, batch 2500, loss[loss=0.1897, simple_loss=0.2739, pruned_loss=0.05273, over 12387.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.2618, pruned_loss=0.04437, over 3318560.77 frames. ], batch size: 247, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:20:54,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8747, 2.1297, 2.3875, 3.1590, 2.1883, 2.3056, 2.3216, 2.2389], device='cuda:7'), covar=tensor([0.1340, 0.3221, 0.2408, 0.0698, 0.3739, 0.2418, 0.3087, 0.3499], device='cuda:7'), in_proj_covar=tensor([0.0392, 0.0431, 0.0360, 0.0327, 0.0431, 0.0499, 0.0401, 0.0504], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:21:04,692 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175061.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:21:46,596 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175092.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:22:00,369 INFO [train.py:904] (7/8) Epoch 18, batch 2550, loss[loss=0.1927, simple_loss=0.2658, pruned_loss=0.05979, over 16752.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2624, pruned_loss=0.04481, over 3312521.09 frames. ], batch size: 134, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:22:22,083 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.06 vs. limit=5.0 2023-04-30 16:22:31,053 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6454, 4.4445, 4.7291, 4.8638, 4.9862, 4.4318, 4.9606, 4.9731], device='cuda:7'), covar=tensor([0.1794, 0.1219, 0.1353, 0.0692, 0.0623, 0.1226, 0.1001, 0.0667], device='cuda:7'), in_proj_covar=tensor([0.0624, 0.0772, 0.0908, 0.0790, 0.0583, 0.0623, 0.0637, 0.0740], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:22:52,063 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175139.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:23:03,053 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.483e+02 2.226e+02 2.530e+02 3.220e+02 6.959e+02, threshold=5.059e+02, percent-clipped=1.0 2023-04-30 16:23:09,313 INFO [train.py:904] (7/8) Epoch 18, batch 2600, loss[loss=0.1864, simple_loss=0.2783, pruned_loss=0.04721, over 16399.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2621, pruned_loss=0.04439, over 3313509.28 frames. ], batch size: 68, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:23:49,544 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7522, 4.6100, 4.6307, 4.2791, 4.2879, 4.6976, 4.5627, 4.4069], device='cuda:7'), covar=tensor([0.0658, 0.0876, 0.0414, 0.0378, 0.1015, 0.0533, 0.0440, 0.0705], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0412, 0.0341, 0.0333, 0.0355, 0.0384, 0.0235, 0.0410], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:24:17,668 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175200.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:24:19,537 INFO [train.py:904] (7/8) Epoch 18, batch 2650, loss[loss=0.1634, simple_loss=0.2578, pruned_loss=0.03456, over 17203.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2624, pruned_loss=0.04401, over 3308542.64 frames. ], batch size: 44, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:24:27,531 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175207.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:25:00,223 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 16:25:21,568 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8781, 1.9226, 2.3912, 2.7565, 2.7532, 2.9678, 1.9689, 3.0904], device='cuda:7'), covar=tensor([0.0157, 0.0456, 0.0305, 0.0286, 0.0269, 0.0215, 0.0512, 0.0130], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0191, 0.0177, 0.0183, 0.0190, 0.0149, 0.0194, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:25:22,166 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.085e+02 2.405e+02 2.864e+02 4.837e+02, threshold=4.811e+02, percent-clipped=0.0 2023-04-30 16:25:27,547 INFO [train.py:904] (7/8) Epoch 18, batch 2700, loss[loss=0.1833, simple_loss=0.2651, pruned_loss=0.05082, over 16660.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2629, pruned_loss=0.044, over 3316488.71 frames. ], batch size: 134, lr: 3.79e-03, grad_scale: 4.0 2023-04-30 16:25:50,173 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175268.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:26:36,868 INFO [train.py:904] (7/8) Epoch 18, batch 2750, loss[loss=0.1667, simple_loss=0.2495, pruned_loss=0.04194, over 16809.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.262, pruned_loss=0.04293, over 3330814.51 frames. ], batch size: 102, lr: 3.79e-03, grad_scale: 4.0 2023-04-30 16:26:59,236 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9627, 4.2335, 3.2441, 2.3691, 2.9276, 2.7213, 4.6696, 3.8255], device='cuda:7'), covar=tensor([0.2634, 0.0707, 0.1692, 0.2779, 0.2718, 0.1919, 0.0382, 0.1299], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0266, 0.0301, 0.0303, 0.0292, 0.0247, 0.0287, 0.0329], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:27:40,030 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 2.063e+02 2.386e+02 2.912e+02 4.096e+02, threshold=4.773e+02, percent-clipped=0.0 2023-04-30 16:27:45,212 INFO [train.py:904] (7/8) Epoch 18, batch 2800, loss[loss=0.1509, simple_loss=0.2465, pruned_loss=0.02763, over 17144.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2619, pruned_loss=0.04286, over 3327709.29 frames. ], batch size: 47, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:27:58,783 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175361.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:28:11,750 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175370.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:28:45,612 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175394.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:28:56,406 INFO [train.py:904] (7/8) Epoch 18, batch 2850, loss[loss=0.159, simple_loss=0.2392, pruned_loss=0.03935, over 16836.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2608, pruned_loss=0.0425, over 3328896.69 frames. ], batch size: 96, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:29:05,422 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175409.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:29:35,325 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175431.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:29:45,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1179, 5.0867, 5.5713, 5.5319, 5.5569, 5.2061, 5.1270, 4.9479], device='cuda:7'), covar=tensor([0.0339, 0.0683, 0.0349, 0.0447, 0.0511, 0.0393, 0.1055, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0438, 0.0427, 0.0402, 0.0474, 0.0451, 0.0546, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 16:29:54,006 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2167, 4.0280, 4.3396, 2.2249, 4.5855, 4.6551, 3.4186, 3.5768], device='cuda:7'), covar=tensor([0.0615, 0.0212, 0.0215, 0.1096, 0.0062, 0.0129, 0.0387, 0.0351], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0138, 0.0077, 0.0122, 0.0125, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 16:29:58,778 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.293e+02 2.737e+02 3.416e+02 1.541e+03, threshold=5.475e+02, percent-clipped=7.0 2023-04-30 16:30:04,920 INFO [train.py:904] (7/8) Epoch 18, batch 2900, loss[loss=0.1867, simple_loss=0.2614, pruned_loss=0.05596, over 16517.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2602, pruned_loss=0.04329, over 3316580.55 frames. ], batch size: 146, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:30:09,544 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175455.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:30:45,664 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:31:04,441 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175495.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:31:13,723 INFO [train.py:904] (7/8) Epoch 18, batch 2950, loss[loss=0.165, simple_loss=0.2573, pruned_loss=0.03636, over 17168.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2595, pruned_loss=0.04376, over 3318561.37 frames. ], batch size: 44, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:32:09,644 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175542.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:32:17,356 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.280e+02 2.587e+02 3.034e+02 5.964e+02, threshold=5.174e+02, percent-clipped=1.0 2023-04-30 16:32:23,321 INFO [train.py:904] (7/8) Epoch 18, batch 3000, loss[loss=0.2003, simple_loss=0.276, pruned_loss=0.06233, over 15477.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2605, pruned_loss=0.04475, over 3318278.27 frames. ], batch size: 190, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:32:23,321 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 16:32:32,127 INFO [train.py:938] (7/8) Epoch 18, validation: loss=0.1359, simple_loss=0.2415, pruned_loss=0.01516, over 944034.00 frames. 2023-04-30 16:32:32,127 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 16:32:48,761 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175563.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:33:42,222 INFO [train.py:904] (7/8) Epoch 18, batch 3050, loss[loss=0.1929, simple_loss=0.2701, pruned_loss=0.05787, over 15523.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2601, pruned_loss=0.04465, over 3319105.58 frames. ], batch size: 191, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:34:08,956 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-04-30 16:34:32,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0042, 1.9006, 2.4994, 2.9500, 2.8432, 3.1940, 1.8612, 3.2210], device='cuda:7'), covar=tensor([0.0177, 0.0507, 0.0298, 0.0241, 0.0257, 0.0175, 0.0612, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0192, 0.0179, 0.0184, 0.0191, 0.0150, 0.0195, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:34:46,187 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.265e+02 2.697e+02 3.394e+02 4.892e+02, threshold=5.395e+02, percent-clipped=0.0 2023-04-30 16:34:51,901 INFO [train.py:904] (7/8) Epoch 18, batch 3100, loss[loss=0.1722, simple_loss=0.2632, pruned_loss=0.04066, over 17135.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2601, pruned_loss=0.04512, over 3306697.40 frames. ], batch size: 47, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:36:01,371 INFO [train.py:904] (7/8) Epoch 18, batch 3150, loss[loss=0.1873, simple_loss=0.2623, pruned_loss=0.05609, over 16890.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2599, pruned_loss=0.04492, over 3302360.30 frames. ], batch size: 116, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:36:13,053 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175710.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:36:36,365 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175726.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:37:01,013 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0493, 5.1887, 5.6466, 5.6048, 5.5966, 5.2422, 5.1765, 4.9659], device='cuda:7'), covar=tensor([0.0349, 0.0597, 0.0371, 0.0390, 0.0499, 0.0385, 0.1014, 0.0422], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0440, 0.0429, 0.0402, 0.0476, 0.0452, 0.0549, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 16:37:05,964 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.210e+02 2.531e+02 3.035e+02 7.463e+02, threshold=5.061e+02, percent-clipped=4.0 2023-04-30 16:37:09,223 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175750.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:37:11,228 INFO [train.py:904] (7/8) Epoch 18, batch 3200, loss[loss=0.1749, simple_loss=0.272, pruned_loss=0.03894, over 16708.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2602, pruned_loss=0.04437, over 3295452.27 frames. ], batch size: 57, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:37:20,658 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 16:37:37,926 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175771.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:37:58,642 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1307, 4.2531, 2.9738, 4.8592, 3.4258, 4.7688, 3.1171, 3.5112], device='cuda:7'), covar=tensor([0.0240, 0.0325, 0.1203, 0.0233, 0.0651, 0.0466, 0.1126, 0.0591], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0178, 0.0196, 0.0163, 0.0177, 0.0221, 0.0205, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 16:37:59,161 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 16:38:11,838 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175795.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:38:20,679 INFO [train.py:904] (7/8) Epoch 18, batch 3250, loss[loss=0.1726, simple_loss=0.2678, pruned_loss=0.03869, over 17118.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2601, pruned_loss=0.04426, over 3288036.38 frames. ], batch size: 49, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:38:28,537 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2062, 2.4184, 2.8215, 3.1700, 2.9338, 3.7317, 2.7285, 3.6612], device='cuda:7'), covar=tensor([0.0209, 0.0384, 0.0283, 0.0258, 0.0283, 0.0131, 0.0367, 0.0137], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0191, 0.0178, 0.0183, 0.0190, 0.0149, 0.0194, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:39:01,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0631, 3.9140, 4.3472, 2.1343, 4.5981, 4.6159, 3.3421, 3.5244], device='cuda:7'), covar=tensor([0.0690, 0.0241, 0.0203, 0.1191, 0.0073, 0.0176, 0.0390, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0137, 0.0076, 0.0122, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 16:39:10,118 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=175837.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:39:15,471 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-30 16:39:17,792 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175843.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:39:23,831 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.233e+02 2.626e+02 2.962e+02 5.772e+02, threshold=5.253e+02, percent-clipped=1.0 2023-04-30 16:39:29,880 INFO [train.py:904] (7/8) Epoch 18, batch 3300, loss[loss=0.1898, simple_loss=0.2762, pruned_loss=0.05166, over 16552.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.261, pruned_loss=0.04458, over 3294539.28 frames. ], batch size: 68, lr: 3.79e-03, grad_scale: 8.0 2023-04-30 16:39:45,175 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=175863.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:40:09,818 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6941, 4.5883, 4.5842, 4.3207, 4.2903, 4.6365, 4.4137, 4.3802], device='cuda:7'), covar=tensor([0.0635, 0.0705, 0.0294, 0.0285, 0.0864, 0.0507, 0.0537, 0.0674], device='cuda:7'), in_proj_covar=tensor([0.0297, 0.0421, 0.0348, 0.0339, 0.0364, 0.0394, 0.0240, 0.0419], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:40:10,964 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175881.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:40:38,828 INFO [train.py:904] (7/8) Epoch 18, batch 3350, loss[loss=0.1752, simple_loss=0.2671, pruned_loss=0.04164, over 17023.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2611, pruned_loss=0.04437, over 3296417.86 frames. ], batch size: 55, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:40:51,561 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=175911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:41:34,592 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=175942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:41:38,255 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=175945.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:41:41,872 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.233e+02 2.642e+02 3.041e+02 9.151e+02, threshold=5.284e+02, percent-clipped=6.0 2023-04-30 16:41:47,214 INFO [train.py:904] (7/8) Epoch 18, batch 3400, loss[loss=0.1795, simple_loss=0.2604, pruned_loss=0.04933, over 16536.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.26, pruned_loss=0.04369, over 3306149.73 frames. ], batch size: 146, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:42:51,037 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-04-30 16:43:01,869 INFO [train.py:904] (7/8) Epoch 18, batch 3450, loss[loss=0.1517, simple_loss=0.2524, pruned_loss=0.02545, over 17220.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2579, pruned_loss=0.04245, over 3321669.14 frames. ], batch size: 45, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:43:04,379 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8169, 1.9781, 2.4257, 2.8648, 2.5989, 3.3198, 1.9846, 3.3991], device='cuda:7'), covar=tensor([0.0287, 0.0524, 0.0345, 0.0306, 0.0338, 0.0168, 0.0554, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0191, 0.0178, 0.0182, 0.0190, 0.0150, 0.0194, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:43:07,954 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176006.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:43:35,517 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:44:06,116 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.158e+02 2.512e+02 2.983e+02 6.813e+02, threshold=5.023e+02, percent-clipped=2.0 2023-04-30 16:44:09,501 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176050.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:44:11,311 INFO [train.py:904] (7/8) Epoch 18, batch 3500, loss[loss=0.1918, simple_loss=0.2693, pruned_loss=0.05715, over 16829.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2568, pruned_loss=0.04159, over 3327267.42 frames. ], batch size: 109, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:44:31,794 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176066.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:44:41,808 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176074.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:45:15,801 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176098.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:45:20,521 INFO [train.py:904] (7/8) Epoch 18, batch 3550, loss[loss=0.1685, simple_loss=0.2426, pruned_loss=0.04719, over 16917.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2561, pruned_loss=0.0415, over 3328274.88 frames. ], batch size: 96, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:46:06,512 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7554, 2.7872, 2.6102, 4.7749, 3.7030, 4.3622, 1.6663, 3.1767], device='cuda:7'), covar=tensor([0.1403, 0.0828, 0.1207, 0.0228, 0.0223, 0.0376, 0.1645, 0.0724], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0168, 0.0188, 0.0181, 0.0202, 0.0214, 0.0193, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 16:46:10,563 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176137.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:46:26,483 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.026e+02 2.447e+02 2.938e+02 5.668e+02, threshold=4.895e+02, percent-clipped=1.0 2023-04-30 16:46:32,328 INFO [train.py:904] (7/8) Epoch 18, batch 3600, loss[loss=0.1731, simple_loss=0.2531, pruned_loss=0.04659, over 16294.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2553, pruned_loss=0.0415, over 3319928.05 frames. ], batch size: 165, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:47:18,726 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:47:43,331 INFO [train.py:904] (7/8) Epoch 18, batch 3650, loss[loss=0.1636, simple_loss=0.2373, pruned_loss=0.045, over 16773.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2542, pruned_loss=0.04194, over 3327097.58 frames. ], batch size: 102, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:48:35,889 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176237.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:48:43,919 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176242.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:48:53,037 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.298e+02 2.711e+02 3.292e+02 6.434e+02, threshold=5.422e+02, percent-clipped=3.0 2023-04-30 16:48:58,536 INFO [train.py:904] (7/8) Epoch 18, batch 3700, loss[loss=0.1734, simple_loss=0.2475, pruned_loss=0.04967, over 16867.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2532, pruned_loss=0.04363, over 3289782.83 frames. ], batch size: 102, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:49:02,445 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.78 vs. limit=5.0 2023-04-30 16:49:06,989 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:49:24,111 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7245, 3.8120, 3.0741, 2.2753, 2.4863, 2.4369, 3.8295, 3.4033], device='cuda:7'), covar=tensor([0.2383, 0.0515, 0.1435, 0.2968, 0.2598, 0.1923, 0.0470, 0.1303], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0264, 0.0300, 0.0303, 0.0292, 0.0248, 0.0288, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:49:57,286 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8434, 5.0052, 5.2004, 4.9884, 5.0275, 5.6372, 5.1607, 4.8536], device='cuda:7'), covar=tensor([0.1220, 0.2029, 0.1958, 0.2011, 0.2594, 0.1058, 0.1645, 0.2634], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0583, 0.0638, 0.0492, 0.0659, 0.0670, 0.0506, 0.0656], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:50:10,768 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176301.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:50:11,698 INFO [train.py:904] (7/8) Epoch 18, batch 3750, loss[loss=0.1778, simple_loss=0.2496, pruned_loss=0.05298, over 16418.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2543, pruned_loss=0.04519, over 3278536.44 frames. ], batch size: 146, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:50:13,785 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176303.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:50:35,146 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176318.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:51:17,336 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.173e+02 2.603e+02 3.284e+02 5.784e+02, threshold=5.206e+02, percent-clipped=1.0 2023-04-30 16:51:17,796 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7997, 4.7816, 4.6712, 4.0061, 4.7735, 1.9093, 4.5033, 4.4407], device='cuda:7'), covar=tensor([0.0121, 0.0091, 0.0212, 0.0393, 0.0093, 0.2561, 0.0167, 0.0210], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0148, 0.0196, 0.0178, 0.0170, 0.0205, 0.0186, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 16:51:23,251 INFO [train.py:904] (7/8) Epoch 18, batch 3800, loss[loss=0.1844, simple_loss=0.2669, pruned_loss=0.05088, over 12433.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2561, pruned_loss=0.04697, over 3274032.32 frames. ], batch size: 246, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:51:44,105 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176366.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:52:35,512 INFO [train.py:904] (7/8) Epoch 18, batch 3850, loss[loss=0.175, simple_loss=0.2562, pruned_loss=0.04688, over 16641.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2567, pruned_loss=0.04816, over 3283938.19 frames. ], batch size: 62, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:52:53,512 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176414.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:53:42,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.342e+02 2.782e+02 3.184e+02 4.336e+02, threshold=5.564e+02, percent-clipped=0.0 2023-04-30 16:53:49,454 INFO [train.py:904] (7/8) Epoch 18, batch 3900, loss[loss=0.1906, simple_loss=0.2574, pruned_loss=0.06188, over 16715.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.256, pruned_loss=0.0486, over 3281261.71 frames. ], batch size: 134, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:54:11,210 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176466.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:55:00,911 INFO [train.py:904] (7/8) Epoch 18, batch 3950, loss[loss=0.1629, simple_loss=0.2452, pruned_loss=0.04029, over 16740.00 frames. ], tot_loss[loss=0.1776, simple_loss=0.2562, pruned_loss=0.04955, over 3274270.94 frames. ], batch size: 124, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:55:35,356 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176527.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:55:48,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5275, 3.4908, 3.9325, 1.9499, 4.1806, 4.1285, 3.0780, 2.9099], device='cuda:7'), covar=tensor([0.0860, 0.0282, 0.0207, 0.1280, 0.0071, 0.0186, 0.0403, 0.0505], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0094, 0.0138, 0.0077, 0.0123, 0.0126, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 16:55:50,496 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176537.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:56:05,770 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.227e+02 2.521e+02 3.093e+02 7.043e+02, threshold=5.041e+02, percent-clipped=2.0 2023-04-30 16:56:12,324 INFO [train.py:904] (7/8) Epoch 18, batch 4000, loss[loss=0.1718, simple_loss=0.2428, pruned_loss=0.05043, over 16820.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2556, pruned_loss=0.04957, over 3277591.80 frames. ], batch size: 116, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:57:00,439 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176585.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:18,460 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176598.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:23,638 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176601.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:24,373 INFO [train.py:904] (7/8) Epoch 18, batch 4050, loss[loss=0.1746, simple_loss=0.2613, pruned_loss=0.04394, over 16669.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2567, pruned_loss=0.04917, over 3271307.81 frames. ], batch size: 134, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:57:39,764 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176613.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:57:58,588 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176625.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:58:24,939 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3553, 4.1099, 4.0157, 2.6128, 3.5997, 3.9932, 3.5904, 2.2848], device='cuda:7'), covar=tensor([0.0498, 0.0032, 0.0041, 0.0387, 0.0085, 0.0093, 0.0088, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0079, 0.0079, 0.0132, 0.0094, 0.0104, 0.0092, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 16:58:30,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 1.835e+02 2.096e+02 2.511e+02 3.496e+02, threshold=4.192e+02, percent-clipped=0.0 2023-04-30 16:58:33,596 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176649.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:58:35,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 16:58:36,778 INFO [train.py:904] (7/8) Epoch 18, batch 4100, loss[loss=0.2025, simple_loss=0.2994, pruned_loss=0.05277, over 15520.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2587, pruned_loss=0.04872, over 3267226.41 frames. ], batch size: 190, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:58:57,216 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176665.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:59:28,859 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 16:59:51,172 INFO [train.py:904] (7/8) Epoch 18, batch 4150, loss[loss=0.2205, simple_loss=0.3063, pruned_loss=0.06734, over 15388.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2659, pruned_loss=0.05084, over 3251036.53 frames. ], batch size: 191, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 16:59:56,466 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.62 vs. limit=5.0 2023-04-30 17:00:27,811 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176726.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:01:00,610 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.334e+02 2.630e+02 3.373e+02 5.770e+02, threshold=5.259e+02, percent-clipped=9.0 2023-04-30 17:01:06,259 INFO [train.py:904] (7/8) Epoch 18, batch 4200, loss[loss=0.2429, simple_loss=0.3323, pruned_loss=0.07672, over 16455.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.2726, pruned_loss=0.05238, over 3219873.30 frames. ], batch size: 146, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 17:01:44,910 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3965, 4.3457, 4.1437, 3.4479, 4.2531, 1.8291, 4.0036, 3.8651], device='cuda:7'), covar=tensor([0.0122, 0.0134, 0.0241, 0.0365, 0.0118, 0.2553, 0.0158, 0.0199], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0147, 0.0194, 0.0177, 0.0169, 0.0203, 0.0185, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:02:21,471 INFO [train.py:904] (7/8) Epoch 18, batch 4250, loss[loss=0.1863, simple_loss=0.284, pruned_loss=0.04431, over 17194.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2756, pruned_loss=0.05218, over 3210202.76 frames. ], batch size: 46, lr: 3.78e-03, grad_scale: 8.0 2023-04-30 17:02:42,733 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1588, 2.4945, 1.9514, 2.1478, 2.8469, 2.4075, 2.9003, 2.9923], device='cuda:7'), covar=tensor([0.0138, 0.0379, 0.0534, 0.0480, 0.0220, 0.0412, 0.0193, 0.0239], device='cuda:7'), in_proj_covar=tensor([0.0200, 0.0233, 0.0223, 0.0223, 0.0233, 0.0231, 0.0236, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:02:51,832 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176822.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:03:12,786 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4294, 4.5145, 4.2915, 4.0403, 4.0286, 4.4132, 4.0766, 4.1206], device='cuda:7'), covar=tensor([0.0557, 0.0408, 0.0279, 0.0255, 0.0734, 0.0339, 0.0631, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0400, 0.0334, 0.0323, 0.0344, 0.0374, 0.0227, 0.0396], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:03:30,838 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.197e+02 2.694e+02 3.150e+02 7.622e+02, threshold=5.388e+02, percent-clipped=1.0 2023-04-30 17:03:35,716 INFO [train.py:904] (7/8) Epoch 18, batch 4300, loss[loss=0.2014, simple_loss=0.2963, pruned_loss=0.05328, over 15240.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2772, pruned_loss=0.05135, over 3206410.29 frames. ], batch size: 190, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:03:38,837 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=176853.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:04:18,054 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-04-30 17:04:44,709 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176898.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:04:49,716 INFO [train.py:904] (7/8) Epoch 18, batch 4350, loss[loss=0.2041, simple_loss=0.2945, pruned_loss=0.0569, over 17253.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2806, pruned_loss=0.05288, over 3187564.33 frames. ], batch size: 43, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:05:06,476 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=176913.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:05:07,771 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=176914.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 17:05:26,666 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.12 vs. limit=5.0 2023-04-30 17:05:38,201 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1267, 3.7186, 3.6178, 2.3909, 3.3884, 3.5983, 3.3064, 2.0536], device='cuda:7'), covar=tensor([0.0519, 0.0039, 0.0053, 0.0393, 0.0087, 0.0095, 0.0102, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0078, 0.0079, 0.0132, 0.0094, 0.0104, 0.0091, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 17:05:55,204 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176946.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:05:57,200 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 2.251e+02 2.629e+02 3.304e+02 5.605e+02, threshold=5.258e+02, percent-clipped=2.0 2023-04-30 17:06:03,309 INFO [train.py:904] (7/8) Epoch 18, batch 4400, loss[loss=0.1869, simple_loss=0.2732, pruned_loss=0.05034, over 16830.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2823, pruned_loss=0.05393, over 3181016.95 frames. ], batch size: 42, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:06:17,424 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=176961.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:06:45,572 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=176981.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:07:00,586 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4018, 2.9500, 3.1142, 1.9400, 2.7691, 2.0497, 2.9932, 3.0937], device='cuda:7'), covar=tensor([0.0225, 0.0696, 0.0522, 0.1818, 0.0797, 0.0960, 0.0606, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0160, 0.0164, 0.0149, 0.0141, 0.0126, 0.0141, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 17:07:08,004 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 17:07:16,204 INFO [train.py:904] (7/8) Epoch 18, batch 4450, loss[loss=0.2105, simple_loss=0.3013, pruned_loss=0.05991, over 16436.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2853, pruned_loss=0.05475, over 3191596.00 frames. ], batch size: 75, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:07:42,932 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8650, 3.7166, 3.6888, 4.0287, 4.1111, 3.7954, 4.0681, 4.1352], device='cuda:7'), covar=tensor([0.1488, 0.1188, 0.1906, 0.0900, 0.0764, 0.2040, 0.1163, 0.0961], device='cuda:7'), in_proj_covar=tensor([0.0609, 0.0755, 0.0885, 0.0769, 0.0566, 0.0609, 0.0619, 0.0718], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:07:44,712 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177021.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:08:14,925 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3388, 5.2230, 5.3757, 5.5643, 5.6969, 5.0190, 5.6930, 5.6945], device='cuda:7'), covar=tensor([0.1462, 0.0910, 0.1192, 0.0469, 0.0372, 0.0627, 0.0345, 0.0404], device='cuda:7'), in_proj_covar=tensor([0.0609, 0.0755, 0.0885, 0.0770, 0.0566, 0.0609, 0.0619, 0.0718], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:08:24,610 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 1.969e+02 2.287e+02 2.678e+02 4.361e+02, threshold=4.575e+02, percent-clipped=0.0 2023-04-30 17:08:30,162 INFO [train.py:904] (7/8) Epoch 18, batch 4500, loss[loss=0.2011, simple_loss=0.2811, pruned_loss=0.06057, over 16654.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.285, pruned_loss=0.05531, over 3210015.63 frames. ], batch size: 62, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:08:40,399 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2251, 2.1444, 2.7243, 3.2239, 3.0557, 3.6537, 2.0999, 3.5194], device='cuda:7'), covar=tensor([0.0175, 0.0460, 0.0272, 0.0234, 0.0230, 0.0123, 0.0526, 0.0124], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0188, 0.0176, 0.0179, 0.0187, 0.0147, 0.0191, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:09:10,368 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1506, 5.1925, 5.5669, 5.5378, 5.6216, 5.1386, 5.0891, 4.7915], device='cuda:7'), covar=tensor([0.0283, 0.0357, 0.0290, 0.0336, 0.0388, 0.0332, 0.0939, 0.0467], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0416, 0.0408, 0.0382, 0.0452, 0.0429, 0.0522, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 17:09:43,620 INFO [train.py:904] (7/8) Epoch 18, batch 4550, loss[loss=0.2134, simple_loss=0.2926, pruned_loss=0.06707, over 16735.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2862, pruned_loss=0.05647, over 3210747.33 frames. ], batch size: 62, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:09:53,008 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177109.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:09:55,566 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177111.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:10:11,724 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177122.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:10:13,429 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2023-04-30 17:10:48,537 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 1.819e+02 2.100e+02 2.524e+02 4.531e+02, threshold=4.200e+02, percent-clipped=0.0 2023-04-30 17:10:53,841 INFO [train.py:904] (7/8) Epoch 18, batch 4600, loss[loss=0.1909, simple_loss=0.2765, pruned_loss=0.05259, over 17032.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2871, pruned_loss=0.05652, over 3217765.16 frames. ], batch size: 55, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:11:20,813 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:11:20,993 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:11:24,041 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177172.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:11:40,978 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7684, 2.9040, 2.7254, 4.4419, 3.5123, 4.0546, 1.7024, 2.9270], device='cuda:7'), covar=tensor([0.1309, 0.0733, 0.1110, 0.0131, 0.0325, 0.0388, 0.1513, 0.0828], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0171, 0.0192, 0.0182, 0.0206, 0.0215, 0.0196, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 17:12:05,786 INFO [train.py:904] (7/8) Epoch 18, batch 4650, loss[loss=0.1945, simple_loss=0.2854, pruned_loss=0.05181, over 15457.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2863, pruned_loss=0.05642, over 3219966.73 frames. ], batch size: 191, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:12:16,081 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177209.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 17:13:10,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 1.973e+02 2.204e+02 2.634e+02 3.745e+02, threshold=4.408e+02, percent-clipped=0.0 2023-04-30 17:13:16,223 INFO [train.py:904] (7/8) Epoch 18, batch 4700, loss[loss=0.2086, simple_loss=0.2901, pruned_loss=0.06351, over 16934.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2834, pruned_loss=0.05506, over 3221584.24 frames. ], batch size: 109, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:13:59,525 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177281.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:14:28,061 INFO [train.py:904] (7/8) Epoch 18, batch 4750, loss[loss=0.1745, simple_loss=0.2631, pruned_loss=0.04297, over 16706.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2794, pruned_loss=0.05314, over 3213575.39 frames. ], batch size: 83, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:14:56,884 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177321.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:15:07,432 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177329.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:15:27,449 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8698, 2.3573, 1.9432, 2.1115, 2.7648, 2.3483, 2.5073, 2.8890], device='cuda:7'), covar=tensor([0.0146, 0.0429, 0.0534, 0.0461, 0.0245, 0.0401, 0.0172, 0.0265], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0229, 0.0220, 0.0219, 0.0229, 0.0228, 0.0231, 0.0224], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:15:34,637 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.255e+02 1.837e+02 2.097e+02 2.577e+02 4.933e+02, threshold=4.195e+02, percent-clipped=1.0 2023-04-30 17:15:40,562 INFO [train.py:904] (7/8) Epoch 18, batch 4800, loss[loss=0.208, simple_loss=0.2991, pruned_loss=0.05845, over 15442.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2747, pruned_loss=0.05079, over 3210660.65 frames. ], batch size: 190, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:16:07,685 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177369.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:16:57,034 INFO [train.py:904] (7/8) Epoch 18, batch 4850, loss[loss=0.176, simple_loss=0.2728, pruned_loss=0.03957, over 16398.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2755, pruned_loss=0.05019, over 3178301.07 frames. ], batch size: 146, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:17:36,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6505, 3.8594, 2.8646, 2.2079, 2.4982, 2.4280, 4.0443, 3.3354], device='cuda:7'), covar=tensor([0.2738, 0.0589, 0.1790, 0.2696, 0.2518, 0.1846, 0.0504, 0.1160], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0263, 0.0298, 0.0301, 0.0291, 0.0245, 0.0286, 0.0325], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 17:17:49,733 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7014, 4.7254, 4.5647, 4.2003, 4.2489, 4.6161, 4.4706, 4.3519], device='cuda:7'), covar=tensor([0.0561, 0.0347, 0.0299, 0.0321, 0.0964, 0.0430, 0.0421, 0.0623], device='cuda:7'), in_proj_covar=tensor([0.0275, 0.0390, 0.0326, 0.0316, 0.0335, 0.0365, 0.0222, 0.0386], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:18:05,881 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 1.952e+02 2.254e+02 2.679e+02 4.935e+02, threshold=4.508e+02, percent-clipped=3.0 2023-04-30 17:18:11,219 INFO [train.py:904] (7/8) Epoch 18, batch 4900, loss[loss=0.1669, simple_loss=0.2652, pruned_loss=0.03432, over 16818.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2745, pruned_loss=0.04891, over 3175882.42 frames. ], batch size: 96, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:18:30,805 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177465.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:18:33,658 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177467.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:19:03,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8340, 4.8254, 4.6597, 3.8643, 4.8208, 1.7431, 4.4788, 4.4209], device='cuda:7'), covar=tensor([0.0091, 0.0090, 0.0177, 0.0505, 0.0092, 0.2705, 0.0142, 0.0251], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0144, 0.0189, 0.0173, 0.0164, 0.0199, 0.0179, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:19:25,468 INFO [train.py:904] (7/8) Epoch 18, batch 4950, loss[loss=0.2153, simple_loss=0.3133, pruned_loss=0.05868, over 16556.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2747, pruned_loss=0.04853, over 3168601.04 frames. ], batch size: 75, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:19:34,714 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177509.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 17:19:54,137 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-04-30 17:20:30,520 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.064e+02 2.400e+02 2.988e+02 6.295e+02, threshold=4.800e+02, percent-clipped=2.0 2023-04-30 17:20:35,829 INFO [train.py:904] (7/8) Epoch 18, batch 5000, loss[loss=0.2033, simple_loss=0.3052, pruned_loss=0.05066, over 16684.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2765, pruned_loss=0.04883, over 3176902.91 frames. ], batch size: 134, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:20:37,501 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177553.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:20:41,868 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177557.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:21:47,177 INFO [train.py:904] (7/8) Epoch 18, batch 5050, loss[loss=0.1703, simple_loss=0.2609, pruned_loss=0.03981, over 16657.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2773, pruned_loss=0.04896, over 3181502.54 frames. ], batch size: 57, lr: 3.77e-03, grad_scale: 16.0 2023-04-30 17:22:04,587 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:22:54,211 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.030e+02 2.394e+02 2.756e+02 4.573e+02, threshold=4.789e+02, percent-clipped=0.0 2023-04-30 17:22:55,063 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 17:22:58,443 INFO [train.py:904] (7/8) Epoch 18, batch 5100, loss[loss=0.1886, simple_loss=0.2774, pruned_loss=0.04989, over 15445.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2759, pruned_loss=0.04833, over 3186450.35 frames. ], batch size: 190, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:23:16,167 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177664.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:24:10,478 INFO [train.py:904] (7/8) Epoch 18, batch 5150, loss[loss=0.1753, simple_loss=0.2701, pruned_loss=0.04025, over 16863.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2755, pruned_loss=0.04745, over 3198967.70 frames. ], batch size: 116, lr: 3.77e-03, grad_scale: 8.0 2023-04-30 17:24:46,124 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177725.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:25:20,505 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 1.888e+02 2.241e+02 2.680e+02 5.710e+02, threshold=4.482e+02, percent-clipped=2.0 2023-04-30 17:25:24,688 INFO [train.py:904] (7/8) Epoch 18, batch 5200, loss[loss=0.1804, simple_loss=0.2638, pruned_loss=0.04853, over 16733.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2733, pruned_loss=0.04658, over 3221144.88 frames. ], batch size: 89, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:25:43,086 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2955, 4.3746, 4.1747, 3.8778, 3.8661, 4.2772, 4.0068, 4.0294], device='cuda:7'), covar=tensor([0.0625, 0.0485, 0.0317, 0.0309, 0.0869, 0.0501, 0.0672, 0.0602], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0395, 0.0328, 0.0319, 0.0340, 0.0370, 0.0224, 0.0390], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:25:44,856 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177765.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:25:47,902 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=177767.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:25:53,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4152, 3.3589, 3.4542, 3.5406, 3.5839, 3.3086, 3.5519, 3.6305], device='cuda:7'), covar=tensor([0.1231, 0.0889, 0.1057, 0.0637, 0.0597, 0.2230, 0.0892, 0.0676], device='cuda:7'), in_proj_covar=tensor([0.0601, 0.0741, 0.0873, 0.0763, 0.0560, 0.0600, 0.0610, 0.0707], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:26:06,854 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177780.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:32,594 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=177797.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:39,166 INFO [train.py:904] (7/8) Epoch 18, batch 5250, loss[loss=0.184, simple_loss=0.267, pruned_loss=0.0505, over 12313.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2712, pruned_loss=0.04638, over 3210614.60 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:26:42,477 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2012, 4.2665, 4.0666, 3.7824, 3.7763, 4.1628, 3.8238, 3.9406], device='cuda:7'), covar=tensor([0.0566, 0.0505, 0.0305, 0.0279, 0.0827, 0.0506, 0.0851, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0395, 0.0329, 0.0320, 0.0340, 0.0370, 0.0224, 0.0390], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:26:56,143 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177813.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:26:58,861 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=177815.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:27:37,232 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177841.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:27:48,926 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 1.963e+02 2.196e+02 2.536e+02 4.359e+02, threshold=4.392e+02, percent-clipped=1.0 2023-04-30 17:27:52,851 INFO [train.py:904] (7/8) Epoch 18, batch 5300, loss[loss=0.1652, simple_loss=0.2528, pruned_loss=0.03887, over 16879.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2672, pruned_loss=0.04495, over 3217562.74 frames. ], batch size: 96, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:27:55,193 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1795, 5.3082, 5.6148, 5.5787, 5.6434, 5.3056, 5.0362, 4.9007], device='cuda:7'), covar=tensor([0.0366, 0.0565, 0.0318, 0.0372, 0.0464, 0.0345, 0.1172, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0381, 0.0416, 0.0408, 0.0383, 0.0455, 0.0428, 0.0524, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 17:28:02,175 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=177858.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:28:17,471 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9006, 2.8228, 2.3725, 2.7346, 3.2544, 2.8935, 3.4219, 3.3775], device='cuda:7'), covar=tensor([0.0068, 0.0350, 0.0474, 0.0367, 0.0221, 0.0355, 0.0198, 0.0246], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0227, 0.0219, 0.0217, 0.0228, 0.0226, 0.0227, 0.0221], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:29:05,788 INFO [train.py:904] (7/8) Epoch 18, batch 5350, loss[loss=0.1804, simple_loss=0.2791, pruned_loss=0.04085, over 16233.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2663, pruned_loss=0.04424, over 3221384.59 frames. ], batch size: 165, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:29:17,498 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=177909.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:29:54,365 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5594, 3.6023, 3.3899, 3.0689, 3.2164, 3.4931, 3.3414, 3.3666], device='cuda:7'), covar=tensor([0.0566, 0.0531, 0.0291, 0.0261, 0.0569, 0.0425, 0.1121, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0284, 0.0400, 0.0333, 0.0323, 0.0343, 0.0375, 0.0226, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:30:15,239 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 1.886e+02 2.232e+02 2.628e+02 4.605e+02, threshold=4.463e+02, percent-clipped=1.0 2023-04-30 17:30:19,777 INFO [train.py:904] (7/8) Epoch 18, batch 5400, loss[loss=0.2021, simple_loss=0.289, pruned_loss=0.05756, over 15380.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2692, pruned_loss=0.04535, over 3205283.58 frames. ], batch size: 191, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:30:33,256 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-30 17:31:39,779 INFO [train.py:904] (7/8) Epoch 18, batch 5450, loss[loss=0.2083, simple_loss=0.2851, pruned_loss=0.06581, over 12034.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2722, pruned_loss=0.04674, over 3185756.94 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:32:08,848 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178020.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:32:36,341 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 17:32:52,434 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.664e+02 2.290e+02 2.936e+02 3.716e+02 7.248e+02, threshold=5.872e+02, percent-clipped=9.0 2023-04-30 17:32:56,928 INFO [train.py:904] (7/8) Epoch 18, batch 5500, loss[loss=0.2333, simple_loss=0.3148, pruned_loss=0.07592, over 16184.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2802, pruned_loss=0.05176, over 3157879.26 frames. ], batch size: 165, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:34:14,395 INFO [train.py:904] (7/8) Epoch 18, batch 5550, loss[loss=0.2827, simple_loss=0.3402, pruned_loss=0.1126, over 11322.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.287, pruned_loss=0.05627, over 3148965.22 frames. ], batch size: 248, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:34:52,598 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-04-30 17:34:53,395 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3104, 2.1945, 2.2582, 4.0440, 2.1209, 2.5385, 2.3281, 2.3220], device='cuda:7'), covar=tensor([0.1232, 0.3468, 0.2616, 0.0481, 0.3876, 0.2371, 0.3182, 0.3181], device='cuda:7'), in_proj_covar=tensor([0.0386, 0.0427, 0.0351, 0.0319, 0.0423, 0.0492, 0.0396, 0.0497], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:35:09,900 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178136.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:35:31,153 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.108e+02 3.250e+02 3.857e+02 5.113e+02 1.112e+03, threshold=7.715e+02, percent-clipped=10.0 2023-04-30 17:35:34,168 INFO [train.py:904] (7/8) Epoch 18, batch 5600, loss[loss=0.3246, simple_loss=0.3644, pruned_loss=0.1424, over 11378.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2918, pruned_loss=0.06046, over 3118927.67 frames. ], batch size: 247, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:35:36,604 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178153.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:36:24,311 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-04-30 17:36:26,277 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9171, 2.7229, 2.6322, 1.9315, 2.5466, 2.6880, 2.5350, 1.8935], device='cuda:7'), covar=tensor([0.0411, 0.0073, 0.0085, 0.0357, 0.0131, 0.0121, 0.0122, 0.0384], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0078, 0.0079, 0.0131, 0.0093, 0.0103, 0.0090, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 17:36:43,424 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.76 vs. limit=5.0 2023-04-30 17:36:58,600 INFO [train.py:904] (7/8) Epoch 18, batch 5650, loss[loss=0.2083, simple_loss=0.2946, pruned_loss=0.06099, over 16697.00 frames. ], tot_loss[loss=0.2132, simple_loss=0.2968, pruned_loss=0.06482, over 3090235.96 frames. ], batch size: 134, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:37:09,804 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178209.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:37:52,369 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5242, 4.1652, 4.0760, 2.6590, 3.7243, 4.1312, 3.6736, 2.2542], device='cuda:7'), covar=tensor([0.0479, 0.0037, 0.0046, 0.0397, 0.0090, 0.0094, 0.0087, 0.0434], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0078, 0.0079, 0.0130, 0.0093, 0.0103, 0.0090, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 17:37:59,605 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.09 vs. limit=5.0 2023-04-30 17:38:16,466 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 3.197e+02 3.818e+02 4.692e+02 7.543e+02, threshold=7.636e+02, percent-clipped=0.0 2023-04-30 17:38:17,800 INFO [train.py:904] (7/8) Epoch 18, batch 5700, loss[loss=0.224, simple_loss=0.309, pruned_loss=0.06956, over 16767.00 frames. ], tot_loss[loss=0.216, simple_loss=0.2987, pruned_loss=0.06663, over 3068891.79 frames. ], batch size: 124, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:38:22,339 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 17:38:25,291 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178257.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:38:36,070 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0759, 2.0814, 2.2092, 3.6567, 1.9964, 2.5047, 2.2258, 2.2752], device='cuda:7'), covar=tensor([0.1290, 0.3480, 0.2719, 0.0537, 0.4103, 0.2307, 0.3294, 0.3377], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0428, 0.0352, 0.0320, 0.0425, 0.0493, 0.0397, 0.0499], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:39:39,098 INFO [train.py:904] (7/8) Epoch 18, batch 5750, loss[loss=0.2288, simple_loss=0.31, pruned_loss=0.07383, over 16904.00 frames. ], tot_loss[loss=0.218, simple_loss=0.3009, pruned_loss=0.06752, over 3071701.18 frames. ], batch size: 109, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:39:47,696 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0954, 2.4216, 2.3600, 2.6347, 2.0090, 3.1841, 1.8403, 2.7252], device='cuda:7'), covar=tensor([0.1189, 0.0600, 0.1101, 0.0180, 0.0155, 0.0405, 0.1524, 0.0725], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0169, 0.0190, 0.0178, 0.0204, 0.0212, 0.0195, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 17:40:08,790 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178320.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:40:59,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 3.189e+02 3.915e+02 4.745e+02 1.114e+03, threshold=7.831e+02, percent-clipped=3.0 2023-04-30 17:41:00,713 INFO [train.py:904] (7/8) Epoch 18, batch 5800, loss[loss=0.2533, simple_loss=0.3159, pruned_loss=0.09536, over 12059.00 frames. ], tot_loss[loss=0.2168, simple_loss=0.3006, pruned_loss=0.06656, over 3055207.50 frames. ], batch size: 248, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:41:28,155 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178368.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:42:19,752 INFO [train.py:904] (7/8) Epoch 18, batch 5850, loss[loss=0.222, simple_loss=0.2944, pruned_loss=0.07475, over 11570.00 frames. ], tot_loss[loss=0.2135, simple_loss=0.2979, pruned_loss=0.06459, over 3040661.05 frames. ], batch size: 247, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:42:44,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5945, 3.0468, 3.0988, 2.0544, 2.7922, 2.0616, 3.1829, 3.2568], device='cuda:7'), covar=tensor([0.0305, 0.0803, 0.0600, 0.1884, 0.0823, 0.1059, 0.0641, 0.0942], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0160, 0.0166, 0.0150, 0.0143, 0.0128, 0.0143, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 17:42:50,307 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2625, 4.2955, 4.6761, 4.6198, 4.6315, 4.2817, 4.3124, 4.1901], device='cuda:7'), covar=tensor([0.0329, 0.0587, 0.0383, 0.0393, 0.0434, 0.0435, 0.0953, 0.0518], device='cuda:7'), in_proj_covar=tensor([0.0384, 0.0420, 0.0409, 0.0385, 0.0457, 0.0431, 0.0527, 0.0345], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 17:43:16,262 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178436.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:43:39,889 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.654e+02 3.099e+02 3.804e+02 9.220e+02, threshold=6.197e+02, percent-clipped=1.0 2023-04-30 17:43:41,468 INFO [train.py:904] (7/8) Epoch 18, batch 5900, loss[loss=0.2078, simple_loss=0.2939, pruned_loss=0.06082, over 16232.00 frames. ], tot_loss[loss=0.2128, simple_loss=0.2973, pruned_loss=0.06413, over 3066356.36 frames. ], batch size: 165, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:43:42,307 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178452.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:43:44,090 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=178453.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:43:51,701 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-04-30 17:44:32,741 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178484.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:44:46,065 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4325, 4.4164, 4.3148, 3.6057, 4.3904, 1.5445, 4.1287, 4.0158], device='cuda:7'), covar=tensor([0.0105, 0.0094, 0.0172, 0.0345, 0.0093, 0.2953, 0.0156, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0142, 0.0187, 0.0172, 0.0162, 0.0198, 0.0178, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:44:58,759 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=178501.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:44:59,558 INFO [train.py:904] (7/8) Epoch 18, batch 5950, loss[loss=0.2171, simple_loss=0.2938, pruned_loss=0.07015, over 11780.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.298, pruned_loss=0.06253, over 3084188.12 frames. ], batch size: 247, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:45:14,217 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-04-30 17:45:18,589 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178513.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:45:35,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1950, 4.3170, 4.6136, 4.5574, 4.5750, 4.2662, 4.2872, 4.2043], device='cuda:7'), covar=tensor([0.0319, 0.0476, 0.0343, 0.0399, 0.0386, 0.0410, 0.0857, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0383, 0.0418, 0.0409, 0.0385, 0.0457, 0.0430, 0.0527, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 17:45:40,343 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2023-04-30 17:46:01,440 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 17:46:17,848 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.727e+02 3.457e+02 4.156e+02 1.353e+03, threshold=6.914e+02, percent-clipped=3.0 2023-04-30 17:46:19,087 INFO [train.py:904] (7/8) Epoch 18, batch 6000, loss[loss=0.2416, simple_loss=0.3055, pruned_loss=0.08889, over 11549.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2972, pruned_loss=0.06282, over 3071203.99 frames. ], batch size: 246, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:46:19,088 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 17:46:29,950 INFO [train.py:938] (7/8) Epoch 18, validation: loss=0.1531, simple_loss=0.2661, pruned_loss=0.02007, over 944034.00 frames. 2023-04-30 17:46:29,951 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 17:46:39,767 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6486, 2.5203, 2.2831, 3.6371, 2.6835, 3.8323, 1.4212, 2.6880], device='cuda:7'), covar=tensor([0.1321, 0.0764, 0.1241, 0.0178, 0.0244, 0.0396, 0.1676, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0169, 0.0190, 0.0179, 0.0204, 0.0213, 0.0195, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 17:47:48,044 INFO [train.py:904] (7/8) Epoch 18, batch 6050, loss[loss=0.2051, simple_loss=0.2994, pruned_loss=0.05543, over 16917.00 frames. ], tot_loss[loss=0.2101, simple_loss=0.2957, pruned_loss=0.06225, over 3064698.73 frames. ], batch size: 109, lr: 3.76e-03, grad_scale: 8.0 2023-04-30 17:47:52,854 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5475, 2.6560, 2.2958, 2.4763, 3.1082, 2.7194, 3.2073, 3.2826], device='cuda:7'), covar=tensor([0.0114, 0.0368, 0.0458, 0.0382, 0.0213, 0.0335, 0.0225, 0.0217], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0224, 0.0218, 0.0216, 0.0225, 0.0224, 0.0226, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:48:14,412 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178619.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:49:06,487 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.761e+02 2.787e+02 3.191e+02 3.956e+02 7.476e+02, threshold=6.381e+02, percent-clipped=1.0 2023-04-30 17:49:06,508 INFO [train.py:904] (7/8) Epoch 18, batch 6100, loss[loss=0.2462, simple_loss=0.3122, pruned_loss=0.09011, over 11656.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2946, pruned_loss=0.061, over 3086437.80 frames. ], batch size: 248, lr: 3.76e-03, grad_scale: 4.0 2023-04-30 17:49:25,456 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1448, 2.1336, 2.1327, 3.7507, 2.0798, 2.4849, 2.2155, 2.3288], device='cuda:7'), covar=tensor([0.1241, 0.3463, 0.2860, 0.0518, 0.3972, 0.2452, 0.3411, 0.3126], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0429, 0.0353, 0.0320, 0.0426, 0.0495, 0.0398, 0.0499], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:49:51,591 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178680.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:50:02,820 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7380, 2.7363, 2.2915, 2.5245, 3.2182, 2.8125, 3.3537, 3.4077], device='cuda:7'), covar=tensor([0.0083, 0.0383, 0.0504, 0.0450, 0.0227, 0.0351, 0.0201, 0.0202], device='cuda:7'), in_proj_covar=tensor([0.0190, 0.0224, 0.0217, 0.0217, 0.0226, 0.0224, 0.0226, 0.0220], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:50:13,195 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178694.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:50:16,799 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178697.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 17:50:23,996 INFO [train.py:904] (7/8) Epoch 18, batch 6150, loss[loss=0.2258, simple_loss=0.3006, pruned_loss=0.07552, over 15403.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2933, pruned_loss=0.06092, over 3076446.05 frames. ], batch size: 190, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:51:39,661 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.052e+02 2.816e+02 3.285e+02 4.160e+02 8.412e+02, threshold=6.570e+02, percent-clipped=4.0 2023-04-30 17:51:39,683 INFO [train.py:904] (7/8) Epoch 18, batch 6200, loss[loss=0.1957, simple_loss=0.2787, pruned_loss=0.05631, over 17034.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2906, pruned_loss=0.05958, over 3102382.52 frames. ], batch size: 55, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:51:45,806 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178755.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 17:51:49,718 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178758.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 17:52:09,039 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178770.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:52:56,579 INFO [train.py:904] (7/8) Epoch 18, batch 6250, loss[loss=0.204, simple_loss=0.2963, pruned_loss=0.05588, over 17129.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2904, pruned_loss=0.05954, over 3095426.54 frames. ], batch size: 48, lr: 3.75e-03, grad_scale: 2.0 2023-04-30 17:53:07,998 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178808.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:53:42,326 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178831.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:54:15,698 INFO [train.py:904] (7/8) Epoch 18, batch 6300, loss[loss=0.2248, simple_loss=0.3178, pruned_loss=0.06591, over 16911.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2905, pruned_loss=0.05908, over 3100016.95 frames. ], batch size: 109, lr: 3.75e-03, grad_scale: 2.0 2023-04-30 17:54:17,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.018e+02 2.791e+02 3.273e+02 3.863e+02 9.337e+02, threshold=6.545e+02, percent-clipped=2.0 2023-04-30 17:54:38,976 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7507, 1.8329, 1.5666, 1.5424, 1.9506, 1.6368, 1.6099, 1.9147], device='cuda:7'), covar=tensor([0.0169, 0.0238, 0.0371, 0.0306, 0.0185, 0.0206, 0.0172, 0.0186], device='cuda:7'), in_proj_covar=tensor([0.0189, 0.0223, 0.0216, 0.0216, 0.0225, 0.0222, 0.0225, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:55:22,846 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=178894.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 17:55:25,919 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1953, 3.6630, 3.6288, 2.3890, 3.3396, 3.6464, 3.4201, 2.0881], device='cuda:7'), covar=tensor([0.0502, 0.0044, 0.0059, 0.0394, 0.0103, 0.0113, 0.0089, 0.0426], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0078, 0.0079, 0.0131, 0.0093, 0.0104, 0.0091, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 17:55:31,658 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7528, 1.9178, 2.3538, 2.7013, 2.6678, 3.0226, 1.9473, 3.0637], device='cuda:7'), covar=tensor([0.0187, 0.0453, 0.0295, 0.0289, 0.0257, 0.0164, 0.0489, 0.0118], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0188, 0.0173, 0.0177, 0.0185, 0.0145, 0.0190, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:55:34,233 INFO [train.py:904] (7/8) Epoch 18, batch 6350, loss[loss=0.1963, simple_loss=0.2793, pruned_loss=0.05671, over 16612.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2912, pruned_loss=0.05999, over 3106579.58 frames. ], batch size: 62, lr: 3.75e-03, grad_scale: 2.0 2023-04-30 17:56:00,906 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1634, 3.1240, 3.4299, 1.7622, 3.6148, 3.6368, 2.8398, 2.7478], device='cuda:7'), covar=tensor([0.0903, 0.0263, 0.0185, 0.1217, 0.0070, 0.0174, 0.0411, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0139, 0.0076, 0.0121, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 17:56:52,052 INFO [train.py:904] (7/8) Epoch 18, batch 6400, loss[loss=0.2293, simple_loss=0.309, pruned_loss=0.07483, over 17048.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2918, pruned_loss=0.06128, over 3094434.34 frames. ], batch size: 53, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:56:53,839 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.717e+02 3.036e+02 3.548e+02 4.325e+02 9.483e+02, threshold=7.097e+02, percent-clipped=4.0 2023-04-30 17:56:58,158 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=178955.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 17:57:28,087 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=178975.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 17:57:34,064 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 17:58:07,515 INFO [train.py:904] (7/8) Epoch 18, batch 6450, loss[loss=0.1712, simple_loss=0.2604, pruned_loss=0.04102, over 16438.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2924, pruned_loss=0.06091, over 3100276.06 frames. ], batch size: 75, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:59:12,195 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5755, 4.5806, 4.4406, 4.1610, 4.1197, 4.5387, 4.3301, 4.2231], device='cuda:7'), covar=tensor([0.0649, 0.0601, 0.0314, 0.0322, 0.1007, 0.0499, 0.0511, 0.0698], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0394, 0.0324, 0.0316, 0.0337, 0.0369, 0.0221, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 17:59:24,107 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179050.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 17:59:26,164 INFO [train.py:904] (7/8) Epoch 18, batch 6500, loss[loss=0.1813, simple_loss=0.2661, pruned_loss=0.0482, over 17013.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.29, pruned_loss=0.05977, over 3124431.87 frames. ], batch size: 50, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 17:59:27,322 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.918e+02 3.772e+02 4.451e+02 8.133e+02, threshold=7.544e+02, percent-clipped=2.0 2023-04-30 17:59:28,317 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179053.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:00:05,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6303, 3.6910, 2.1246, 4.1435, 2.7836, 4.1052, 2.3079, 2.9470], device='cuda:7'), covar=tensor([0.0282, 0.0400, 0.1769, 0.0229, 0.0861, 0.0571, 0.1639, 0.0769], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0172, 0.0192, 0.0152, 0.0174, 0.0212, 0.0199, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 18:00:44,328 INFO [train.py:904] (7/8) Epoch 18, batch 6550, loss[loss=0.198, simple_loss=0.2956, pruned_loss=0.05017, over 15362.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2927, pruned_loss=0.06093, over 3102491.28 frames. ], batch size: 190, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:00:54,324 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179108.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:01:20,235 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179126.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:01:58,909 INFO [train.py:904] (7/8) Epoch 18, batch 6600, loss[loss=0.1927, simple_loss=0.2878, pruned_loss=0.0488, over 16876.00 frames. ], tot_loss[loss=0.2091, simple_loss=0.2948, pruned_loss=0.06168, over 3090769.58 frames. ], batch size: 96, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:02:00,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.896e+02 2.804e+02 3.237e+02 3.890e+02 6.888e+02, threshold=6.473e+02, percent-clipped=0.0 2023-04-30 18:02:05,162 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179156.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:02:56,391 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4828, 4.3722, 4.2287, 2.8011, 3.7213, 4.2962, 3.7549, 2.4519], device='cuda:7'), covar=tensor([0.0508, 0.0031, 0.0042, 0.0385, 0.0097, 0.0093, 0.0095, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0077, 0.0078, 0.0130, 0.0092, 0.0103, 0.0090, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:03:17,507 INFO [train.py:904] (7/8) Epoch 18, batch 6650, loss[loss=0.2183, simple_loss=0.2989, pruned_loss=0.06887, over 15204.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2954, pruned_loss=0.06269, over 3080591.88 frames. ], batch size: 191, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:04:30,599 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179250.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 18:04:32,478 INFO [train.py:904] (7/8) Epoch 18, batch 6700, loss[loss=0.1986, simple_loss=0.2908, pruned_loss=0.05321, over 16461.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2943, pruned_loss=0.0624, over 3099070.38 frames. ], batch size: 146, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:04:34,184 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 2.690e+02 3.432e+02 4.163e+02 9.246e+02, threshold=6.864e+02, percent-clipped=3.0 2023-04-30 18:04:46,050 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0770, 5.1323, 5.5546, 5.5297, 5.5354, 5.1278, 5.0597, 4.7552], device='cuda:7'), covar=tensor([0.0308, 0.0538, 0.0360, 0.0386, 0.0506, 0.0386, 0.1086, 0.0514], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0422, 0.0412, 0.0390, 0.0463, 0.0434, 0.0532, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 18:05:09,303 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179275.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:05:19,301 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6185, 2.5567, 1.8950, 2.6842, 2.1116, 2.7533, 2.1249, 2.3737], device='cuda:7'), covar=tensor([0.0310, 0.0380, 0.1285, 0.0223, 0.0671, 0.0411, 0.1191, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0171, 0.0191, 0.0152, 0.0173, 0.0212, 0.0199, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 18:05:48,870 INFO [train.py:904] (7/8) Epoch 18, batch 6750, loss[loss=0.2008, simple_loss=0.282, pruned_loss=0.05978, over 16394.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2928, pruned_loss=0.06224, over 3088302.13 frames. ], batch size: 146, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:06:20,235 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179323.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:07:00,954 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179350.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:07:03,110 INFO [train.py:904] (7/8) Epoch 18, batch 6800, loss[loss=0.2321, simple_loss=0.3049, pruned_loss=0.07969, over 11718.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2932, pruned_loss=0.06244, over 3085264.80 frames. ], batch size: 247, lr: 3.75e-03, grad_scale: 8.0 2023-04-30 18:07:04,936 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 2.962e+02 3.412e+02 4.074e+02 1.001e+03, threshold=6.824e+02, percent-clipped=4.0 2023-04-30 18:07:06,055 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179353.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:07:18,277 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179361.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:07:18,706 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-04-30 18:08:16,242 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179398.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:08:20,459 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179401.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:08:21,234 INFO [train.py:904] (7/8) Epoch 18, batch 6850, loss[loss=0.1786, simple_loss=0.2861, pruned_loss=0.03559, over 16928.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2941, pruned_loss=0.06276, over 3089359.37 frames. ], batch size: 96, lr: 3.75e-03, grad_scale: 8.0 2023-04-30 18:08:27,420 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0297, 4.0073, 3.9870, 3.1760, 3.9724, 1.7804, 3.7222, 3.5225], device='cuda:7'), covar=tensor([0.0140, 0.0127, 0.0208, 0.0304, 0.0122, 0.2816, 0.0170, 0.0283], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0141, 0.0187, 0.0171, 0.0162, 0.0198, 0.0176, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:08:50,209 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179422.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:08:56,242 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179426.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:09:24,219 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179444.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:09:34,758 INFO [train.py:904] (7/8) Epoch 18, batch 6900, loss[loss=0.2242, simple_loss=0.3161, pruned_loss=0.06613, over 17056.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2962, pruned_loss=0.06195, over 3102436.31 frames. ], batch size: 53, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:09:38,471 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.677e+02 2.748e+02 3.355e+02 3.841e+02 5.597e+02, threshold=6.710e+02, percent-clipped=0.0 2023-04-30 18:10:10,838 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179474.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:10:53,466 INFO [train.py:904] (7/8) Epoch 18, batch 6950, loss[loss=0.2183, simple_loss=0.2997, pruned_loss=0.06848, over 16891.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2976, pruned_loss=0.0631, over 3108532.10 frames. ], batch size: 109, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:10:58,868 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179505.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 18:12:07,114 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=179550.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 18:12:09,850 INFO [train.py:904] (7/8) Epoch 18, batch 7000, loss[loss=0.1993, simple_loss=0.3014, pruned_loss=0.04861, over 16763.00 frames. ], tot_loss[loss=0.2115, simple_loss=0.2977, pruned_loss=0.06266, over 3102794.29 frames. ], batch size: 124, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:12:12,181 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.911e+02 2.791e+02 3.383e+02 4.252e+02 7.699e+02, threshold=6.767e+02, percent-clipped=2.0 2023-04-30 18:13:17,064 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=179598.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:13:21,684 INFO [train.py:904] (7/8) Epoch 18, batch 7050, loss[loss=0.2036, simple_loss=0.2952, pruned_loss=0.05602, over 16698.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2985, pruned_loss=0.06209, over 3108142.11 frames. ], batch size: 62, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:14:13,007 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179636.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:14:37,725 INFO [train.py:904] (7/8) Epoch 18, batch 7100, loss[loss=0.2147, simple_loss=0.2983, pruned_loss=0.06561, over 15468.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2976, pruned_loss=0.0625, over 3088241.11 frames. ], batch size: 191, lr: 3.75e-03, grad_scale: 4.0 2023-04-30 18:14:40,300 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.117e+02 2.863e+02 3.448e+02 4.199e+02 1.001e+03, threshold=6.895e+02, percent-clipped=2.0 2023-04-30 18:15:46,850 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179697.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:15:53,688 INFO [train.py:904] (7/8) Epoch 18, batch 7150, loss[loss=0.2077, simple_loss=0.2902, pruned_loss=0.06258, over 16696.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2957, pruned_loss=0.06252, over 3080123.51 frames. ], batch size: 89, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:16:16,358 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179717.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:16:42,879 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2784, 3.7048, 3.6999, 2.5805, 3.3648, 3.7010, 3.4555, 2.1526], device='cuda:7'), covar=tensor([0.0487, 0.0043, 0.0050, 0.0343, 0.0099, 0.0092, 0.0081, 0.0419], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0078, 0.0078, 0.0132, 0.0092, 0.0104, 0.0091, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:17:08,213 INFO [train.py:904] (7/8) Epoch 18, batch 7200, loss[loss=0.1902, simple_loss=0.2732, pruned_loss=0.05359, over 11554.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2938, pruned_loss=0.06094, over 3088381.04 frames. ], batch size: 248, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:17:10,640 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.743e+02 2.765e+02 3.366e+02 4.196e+02 7.871e+02, threshold=6.733e+02, percent-clipped=4.0 2023-04-30 18:18:17,998 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179796.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:18:25,229 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179800.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:18:27,055 INFO [train.py:904] (7/8) Epoch 18, batch 7250, loss[loss=0.1643, simple_loss=0.2523, pruned_loss=0.03812, over 16735.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2908, pruned_loss=0.05928, over 3090142.69 frames. ], batch size: 76, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:18:52,137 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3278, 5.3438, 5.1464, 3.7243, 5.2158, 1.8529, 4.8229, 4.8735], device='cuda:7'), covar=tensor([0.0143, 0.0099, 0.0257, 0.0696, 0.0132, 0.3182, 0.0178, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0140, 0.0185, 0.0170, 0.0161, 0.0196, 0.0174, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:18:56,711 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0822, 2.5215, 2.5388, 1.9161, 2.7340, 2.7902, 2.4418, 2.3663], device='cuda:7'), covar=tensor([0.0717, 0.0231, 0.0238, 0.0957, 0.0113, 0.0299, 0.0463, 0.0465], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0105, 0.0094, 0.0138, 0.0075, 0.0121, 0.0125, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 18:19:00,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2023-04-30 18:19:42,152 INFO [train.py:904] (7/8) Epoch 18, batch 7300, loss[loss=0.1984, simple_loss=0.2954, pruned_loss=0.05064, over 16914.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2908, pruned_loss=0.05976, over 3076252.28 frames. ], batch size: 96, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:19:45,261 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.793e+02 3.453e+02 4.292e+02 8.148e+02, threshold=6.907e+02, percent-clipped=1.0 2023-04-30 18:19:51,014 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179857.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:20:58,402 INFO [train.py:904] (7/8) Epoch 18, batch 7350, loss[loss=0.1869, simple_loss=0.2722, pruned_loss=0.05083, over 16644.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2917, pruned_loss=0.06031, over 3083633.80 frames. ], batch size: 62, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:21:05,714 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179906.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:21:32,229 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=179924.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:22:16,607 INFO [train.py:904] (7/8) Epoch 18, batch 7400, loss[loss=0.2064, simple_loss=0.298, pruned_loss=0.05743, over 15394.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2926, pruned_loss=0.06047, over 3091625.12 frames. ], batch size: 191, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:22:19,967 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.864e+02 3.444e+02 4.186e+02 8.589e+02, threshold=6.889e+02, percent-clipped=1.0 2023-04-30 18:22:41,115 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179967.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:23:09,740 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=179985.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:23:21,129 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=179992.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:23:39,787 INFO [train.py:904] (7/8) Epoch 18, batch 7450, loss[loss=0.1849, simple_loss=0.2711, pruned_loss=0.04934, over 17050.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2928, pruned_loss=0.06116, over 3083960.11 frames. ], batch size: 55, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:23:47,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0513, 2.3375, 2.2821, 2.7575, 1.7575, 3.1724, 1.8355, 2.6054], device='cuda:7'), covar=tensor([0.1149, 0.0663, 0.1134, 0.0219, 0.0137, 0.0384, 0.1430, 0.0755], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0169, 0.0191, 0.0178, 0.0205, 0.0213, 0.0196, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 18:23:56,274 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1329, 1.4580, 1.8902, 2.0878, 2.1957, 2.2805, 1.6735, 2.2263], device='cuda:7'), covar=tensor([0.0219, 0.0473, 0.0264, 0.0307, 0.0271, 0.0201, 0.0495, 0.0136], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0186, 0.0172, 0.0175, 0.0184, 0.0143, 0.0188, 0.0139], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:24:05,911 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180017.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:24:19,917 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180026.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:25:01,499 INFO [train.py:904] (7/8) Epoch 18, batch 7500, loss[loss=0.1902, simple_loss=0.2788, pruned_loss=0.05078, over 16910.00 frames. ], tot_loss[loss=0.2078, simple_loss=0.2935, pruned_loss=0.06101, over 3082918.84 frames. ], batch size: 109, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:25:04,523 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 3.002e+02 3.628e+02 4.895e+02 8.341e+02, threshold=7.256e+02, percent-clipped=3.0 2023-04-30 18:25:08,261 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2155, 3.2656, 1.9612, 3.5867, 2.4691, 3.5705, 2.0287, 2.6244], device='cuda:7'), covar=tensor([0.0291, 0.0399, 0.1657, 0.0206, 0.0819, 0.0685, 0.1556, 0.0790], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0172, 0.0190, 0.0150, 0.0175, 0.0212, 0.0199, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 18:25:22,815 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180065.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:25:56,700 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180087.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:25:58,152 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-04-30 18:26:16,501 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180100.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 18:26:19,022 INFO [train.py:904] (7/8) Epoch 18, batch 7550, loss[loss=0.2266, simple_loss=0.2971, pruned_loss=0.07807, over 11587.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2934, pruned_loss=0.06146, over 3089414.90 frames. ], batch size: 247, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:27:30,329 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180148.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:27:36,553 INFO [train.py:904] (7/8) Epoch 18, batch 7600, loss[loss=0.2368, simple_loss=0.3039, pruned_loss=0.08489, over 10878.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2924, pruned_loss=0.06141, over 3078741.26 frames. ], batch size: 248, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:27:37,487 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180152.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:27:39,408 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.698e+02 3.315e+02 4.368e+02 9.372e+02, threshold=6.630e+02, percent-clipped=3.0 2023-04-30 18:28:55,657 INFO [train.py:904] (7/8) Epoch 18, batch 7650, loss[loss=0.2407, simple_loss=0.3128, pruned_loss=0.08432, over 11360.00 frames. ], tot_loss[loss=0.2102, simple_loss=0.2941, pruned_loss=0.06316, over 3049700.27 frames. ], batch size: 247, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:29:00,778 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180205.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:29:08,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4697, 3.4880, 3.4478, 2.7089, 3.3560, 2.1092, 3.1058, 2.8211], device='cuda:7'), covar=tensor([0.0159, 0.0118, 0.0184, 0.0214, 0.0105, 0.2089, 0.0132, 0.0206], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0138, 0.0184, 0.0168, 0.0159, 0.0195, 0.0173, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:29:13,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3854, 4.4302, 4.2903, 3.5733, 4.3351, 1.7040, 4.1104, 4.0345], device='cuda:7'), covar=tensor([0.0101, 0.0088, 0.0176, 0.0323, 0.0089, 0.2607, 0.0129, 0.0210], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0138, 0.0184, 0.0168, 0.0159, 0.0195, 0.0173, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:29:49,376 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6275, 3.5820, 2.7880, 2.2673, 2.3447, 2.2917, 3.7945, 3.3234], device='cuda:7'), covar=tensor([0.2813, 0.0707, 0.1850, 0.2690, 0.2749, 0.2186, 0.0496, 0.1277], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0264, 0.0301, 0.0306, 0.0292, 0.0249, 0.0287, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:30:13,299 INFO [train.py:904] (7/8) Epoch 18, batch 7700, loss[loss=0.2067, simple_loss=0.286, pruned_loss=0.06368, over 16688.00 frames. ], tot_loss[loss=0.2099, simple_loss=0.2936, pruned_loss=0.06305, over 3073075.29 frames. ], batch size: 134, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:30:18,208 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.849e+02 3.108e+02 3.616e+02 4.495e+02 6.527e+02, threshold=7.232e+02, percent-clipped=0.0 2023-04-30 18:30:26,345 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-04-30 18:30:29,252 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180262.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 18:30:29,485 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5074, 3.3841, 2.6084, 2.1456, 2.2507, 2.2072, 3.5861, 3.1635], device='cuda:7'), covar=tensor([0.2875, 0.0803, 0.1994, 0.2718, 0.2625, 0.2146, 0.0567, 0.1337], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0264, 0.0301, 0.0306, 0.0292, 0.0249, 0.0287, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:30:35,671 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180266.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:30:55,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9058, 2.0923, 2.3153, 3.4579, 2.0646, 2.3618, 2.2064, 2.2166], device='cuda:7'), covar=tensor([0.1360, 0.3362, 0.2595, 0.0597, 0.4014, 0.2398, 0.3294, 0.3134], device='cuda:7'), in_proj_covar=tensor([0.0385, 0.0428, 0.0353, 0.0320, 0.0428, 0.0492, 0.0398, 0.0499], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:30:57,474 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180280.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:31:10,599 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.77 vs. limit=5.0 2023-04-30 18:31:15,999 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180292.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:31:30,824 INFO [train.py:904] (7/8) Epoch 18, batch 7750, loss[loss=0.249, simple_loss=0.3101, pruned_loss=0.09395, over 11238.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2936, pruned_loss=0.06303, over 3071800.57 frames. ], batch size: 248, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:32:23,886 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180336.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:32:29,342 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180340.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:32:46,614 INFO [train.py:904] (7/8) Epoch 18, batch 7800, loss[loss=0.2474, simple_loss=0.3139, pruned_loss=0.09051, over 11725.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2939, pruned_loss=0.06298, over 3081454.17 frames. ], batch size: 247, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:32:51,026 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.798e+02 2.996e+02 3.455e+02 4.260e+02 9.369e+02, threshold=6.911e+02, percent-clipped=1.0 2023-04-30 18:33:33,639 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180382.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:33:55,568 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180397.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:34:02,040 INFO [train.py:904] (7/8) Epoch 18, batch 7850, loss[loss=0.2252, simple_loss=0.313, pruned_loss=0.06873, over 16747.00 frames. ], tot_loss[loss=0.21, simple_loss=0.2948, pruned_loss=0.06264, over 3084975.03 frames. ], batch size: 124, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:35:16,227 INFO [train.py:904] (7/8) Epoch 18, batch 7900, loss[loss=0.2679, simple_loss=0.3263, pruned_loss=0.1048, over 11327.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2941, pruned_loss=0.06273, over 3059891.43 frames. ], batch size: 247, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:35:17,208 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180452.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:35:20,364 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.825e+02 3.445e+02 4.253e+02 7.547e+02, threshold=6.890e+02, percent-clipped=0.0 2023-04-30 18:36:32,645 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180500.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:36:34,744 INFO [train.py:904] (7/8) Epoch 18, batch 7950, loss[loss=0.1946, simple_loss=0.2799, pruned_loss=0.05463, over 16429.00 frames. ], tot_loss[loss=0.2104, simple_loss=0.2949, pruned_loss=0.06298, over 3072986.55 frames. ], batch size: 35, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:37:52,811 INFO [train.py:904] (7/8) Epoch 18, batch 8000, loss[loss=0.2122, simple_loss=0.2965, pruned_loss=0.0639, over 16856.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2951, pruned_loss=0.06303, over 3079346.44 frames. ], batch size: 116, lr: 3.74e-03, grad_scale: 8.0 2023-04-30 18:37:57,088 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.803e+02 2.710e+02 3.198e+02 3.720e+02 8.207e+02, threshold=6.397e+02, percent-clipped=2.0 2023-04-30 18:38:07,700 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180561.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:38:08,920 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180562.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 18:38:21,409 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:38:36,692 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180580.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:39:10,314 INFO [train.py:904] (7/8) Epoch 18, batch 8050, loss[loss=0.2023, simple_loss=0.2993, pruned_loss=0.05265, over 16722.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2941, pruned_loss=0.062, over 3091371.69 frames. ], batch size: 83, lr: 3.74e-03, grad_scale: 4.0 2023-04-30 18:39:23,843 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180610.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 18:39:32,891 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-04-30 18:39:50,351 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180628.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:39:54,567 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180631.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:40:26,589 INFO [train.py:904] (7/8) Epoch 18, batch 8100, loss[loss=0.2084, simple_loss=0.292, pruned_loss=0.06241, over 16284.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.294, pruned_loss=0.06186, over 3079066.76 frames. ], batch size: 165, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:40:32,030 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.839e+02 3.233e+02 3.954e+02 8.532e+02, threshold=6.466e+02, percent-clipped=3.0 2023-04-30 18:41:12,089 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180682.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:41:27,699 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180692.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:41:41,227 INFO [train.py:904] (7/8) Epoch 18, batch 8150, loss[loss=0.2061, simple_loss=0.2926, pruned_loss=0.05979, over 15362.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2923, pruned_loss=0.06155, over 3064372.10 frames. ], batch size: 190, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:42:24,162 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180730.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:42:26,767 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180732.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:42:56,056 INFO [train.py:904] (7/8) Epoch 18, batch 8200, loss[loss=0.2174, simple_loss=0.2884, pruned_loss=0.07314, over 11495.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2897, pruned_loss=0.061, over 3067506.30 frames. ], batch size: 248, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:43:02,076 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.834e+02 3.334e+02 4.232e+02 1.685e+03, threshold=6.669e+02, percent-clipped=1.0 2023-04-30 18:43:24,160 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7909, 4.5774, 4.8283, 5.0044, 5.1629, 4.5405, 5.1357, 5.1504], device='cuda:7'), covar=tensor([0.1984, 0.1379, 0.1651, 0.0754, 0.0604, 0.1120, 0.0557, 0.0765], device='cuda:7'), in_proj_covar=tensor([0.0597, 0.0735, 0.0863, 0.0750, 0.0561, 0.0591, 0.0613, 0.0708], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:43:33,662 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0734, 3.9014, 4.1498, 4.2704, 4.3755, 3.9261, 4.2997, 4.3894], device='cuda:7'), covar=tensor([0.1797, 0.1265, 0.1461, 0.0711, 0.0614, 0.1527, 0.0809, 0.0757], device='cuda:7'), in_proj_covar=tensor([0.0597, 0.0736, 0.0863, 0.0750, 0.0561, 0.0592, 0.0613, 0.0709], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:44:01,916 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180793.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:44:05,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2049, 1.5857, 2.0653, 2.2603, 2.3865, 2.5870, 1.8959, 2.5252], device='cuda:7'), covar=tensor([0.0192, 0.0475, 0.0280, 0.0312, 0.0269, 0.0176, 0.0430, 0.0146], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0186, 0.0172, 0.0175, 0.0183, 0.0142, 0.0188, 0.0138], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:44:14,892 INFO [train.py:904] (7/8) Epoch 18, batch 8250, loss[loss=0.1789, simple_loss=0.2671, pruned_loss=0.04541, over 11996.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2886, pruned_loss=0.05879, over 3042542.21 frames. ], batch size: 247, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:44:53,490 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=180825.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:45:37,343 INFO [train.py:904] (7/8) Epoch 18, batch 8300, loss[loss=0.1835, simple_loss=0.2586, pruned_loss=0.05416, over 11987.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2861, pruned_loss=0.05592, over 3029129.78 frames. ], batch size: 246, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:45:43,844 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.692e+02 2.270e+02 2.749e+02 3.299e+02 7.574e+02, threshold=5.499e+02, percent-clipped=1.0 2023-04-30 18:45:52,437 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180861.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:46:32,642 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=180886.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:46:58,365 INFO [train.py:904] (7/8) Epoch 18, batch 8350, loss[loss=0.1921, simple_loss=0.2901, pruned_loss=0.04703, over 15362.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2854, pruned_loss=0.05381, over 3034546.57 frames. ], batch size: 191, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:47:10,065 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=180909.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:47:36,518 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=180926.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:48:00,818 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3940, 4.4854, 4.6040, 4.4005, 4.5131, 5.0290, 4.5502, 4.2243], device='cuda:7'), covar=tensor([0.1324, 0.1705, 0.1895, 0.2080, 0.2347, 0.0922, 0.1508, 0.2401], device='cuda:7'), in_proj_covar=tensor([0.0380, 0.0548, 0.0610, 0.0459, 0.0613, 0.0633, 0.0482, 0.0620], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:48:16,302 INFO [train.py:904] (7/8) Epoch 18, batch 8400, loss[loss=0.1857, simple_loss=0.2792, pruned_loss=0.04611, over 16845.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2824, pruned_loss=0.05164, over 3031849.72 frames. ], batch size: 116, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 18:48:22,152 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.789e+02 2.220e+02 2.634e+02 3.245e+02 6.969e+02, threshold=5.268e+02, percent-clipped=3.0 2023-04-30 18:49:06,324 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1009, 5.4579, 4.9907, 5.3522, 5.0426, 4.7656, 4.9673, 5.5108], device='cuda:7'), covar=tensor([0.1973, 0.1511, 0.2673, 0.1316, 0.1498, 0.1388, 0.2200, 0.1907], device='cuda:7'), in_proj_covar=tensor([0.0621, 0.0760, 0.0627, 0.0567, 0.0474, 0.0491, 0.0635, 0.0591], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:49:16,588 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=180992.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:49:31,289 INFO [train.py:904] (7/8) Epoch 18, batch 8450, loss[loss=0.1772, simple_loss=0.2737, pruned_loss=0.04032, over 16605.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2806, pruned_loss=0.04976, over 3052087.48 frames. ], batch size: 62, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 18:49:57,242 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 18:50:13,052 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1195, 1.4692, 1.9388, 2.1140, 2.1667, 2.3444, 1.8462, 2.2121], device='cuda:7'), covar=tensor([0.0209, 0.0460, 0.0247, 0.0257, 0.0270, 0.0165, 0.0400, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0184, 0.0170, 0.0172, 0.0181, 0.0140, 0.0185, 0.0136], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:50:13,118 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5182, 3.4226, 2.8049, 2.1949, 2.1678, 2.3190, 3.4975, 3.1258], device='cuda:7'), covar=tensor([0.2639, 0.0656, 0.1570, 0.2746, 0.2817, 0.2090, 0.0452, 0.1197], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0257, 0.0292, 0.0298, 0.0286, 0.0243, 0.0281, 0.0319], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:50:31,761 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181040.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:50:33,657 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-04-30 18:50:50,508 INFO [train.py:904] (7/8) Epoch 18, batch 8500, loss[loss=0.1624, simple_loss=0.2453, pruned_loss=0.03973, over 12325.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2763, pruned_loss=0.04726, over 3048381.20 frames. ], batch size: 246, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:50:58,705 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.304e+02 3.083e+02 3.822e+02 7.945e+02, threshold=6.166e+02, percent-clipped=7.0 2023-04-30 18:51:49,232 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181088.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:52:13,300 INFO [train.py:904] (7/8) Epoch 18, batch 8550, loss[loss=0.1793, simple_loss=0.2758, pruned_loss=0.04139, over 15480.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2741, pruned_loss=0.04601, over 3030635.91 frames. ], batch size: 190, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:52:16,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3246, 3.5539, 3.6086, 2.5296, 3.2230, 3.5742, 3.3794, 1.9446], device='cuda:7'), covar=tensor([0.0489, 0.0059, 0.0047, 0.0344, 0.0109, 0.0077, 0.0079, 0.0513], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0076, 0.0076, 0.0128, 0.0090, 0.0101, 0.0089, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 18:53:50,712 INFO [train.py:904] (7/8) Epoch 18, batch 8600, loss[loss=0.1787, simple_loss=0.2763, pruned_loss=0.04058, over 16693.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2742, pruned_loss=0.04531, over 3020448.20 frames. ], batch size: 134, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:54:01,222 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.216e+02 2.624e+02 3.286e+02 8.410e+02, threshold=5.248e+02, percent-clipped=1.0 2023-04-30 18:54:50,077 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181181.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:55:10,729 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1016, 2.5054, 2.6376, 1.8261, 2.7561, 2.8514, 2.5712, 2.4444], device='cuda:7'), covar=tensor([0.0674, 0.0256, 0.0228, 0.1085, 0.0106, 0.0241, 0.0390, 0.0434], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0102, 0.0090, 0.0134, 0.0072, 0.0115, 0.0120, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 18:55:29,862 INFO [train.py:904] (7/8) Epoch 18, batch 8650, loss[loss=0.1844, simple_loss=0.281, pruned_loss=0.04384, over 16879.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2723, pruned_loss=0.04397, over 3019008.63 frames. ], batch size: 116, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:55:48,204 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4185, 4.2148, 4.4881, 4.6163, 4.7550, 4.3338, 4.7326, 4.7313], device='cuda:7'), covar=tensor([0.1573, 0.1141, 0.1286, 0.0621, 0.0447, 0.1002, 0.0482, 0.0615], device='cuda:7'), in_proj_covar=tensor([0.0578, 0.0714, 0.0835, 0.0730, 0.0544, 0.0575, 0.0591, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 18:56:25,799 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181226.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:57:15,665 INFO [train.py:904] (7/8) Epoch 18, batch 8700, loss[loss=0.1828, simple_loss=0.2793, pruned_loss=0.04318, over 16842.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2697, pruned_loss=0.04233, over 3038429.34 frames. ], batch size: 116, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 18:57:17,313 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7309, 3.1000, 3.4294, 1.8977, 2.9172, 2.2663, 3.3270, 3.3376], device='cuda:7'), covar=tensor([0.0265, 0.0831, 0.0465, 0.1990, 0.0775, 0.0945, 0.0581, 0.0772], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0154, 0.0160, 0.0146, 0.0138, 0.0125, 0.0138, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 18:57:25,080 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.665e+02 2.245e+02 2.629e+02 3.109e+02 5.522e+02, threshold=5.258e+02, percent-clipped=1.0 2023-04-30 18:57:56,766 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181274.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:58:39,715 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=181297.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 18:58:48,997 INFO [train.py:904] (7/8) Epoch 18, batch 8750, loss[loss=0.1581, simple_loss=0.2467, pruned_loss=0.03474, over 12507.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2695, pruned_loss=0.04184, over 3039676.33 frames. ], batch size: 246, lr: 3.73e-03, grad_scale: 4.0 2023-04-30 19:00:41,596 INFO [train.py:904] (7/8) Epoch 18, batch 8800, loss[loss=0.1822, simple_loss=0.2764, pruned_loss=0.04403, over 16875.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2675, pruned_loss=0.04064, over 3043494.78 frames. ], batch size: 124, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:00:51,214 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.344e+02 2.185e+02 2.552e+02 3.082e+02 8.545e+02, threshold=5.104e+02, percent-clipped=2.0 2023-04-30 19:00:54,115 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=181358.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:01:57,941 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181388.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:02:27,194 INFO [train.py:904] (7/8) Epoch 18, batch 8850, loss[loss=0.1674, simple_loss=0.2691, pruned_loss=0.03286, over 15299.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2702, pruned_loss=0.04014, over 3040366.33 frames. ], batch size: 190, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:03:23,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5767, 3.5484, 3.5413, 2.7528, 3.4658, 1.9626, 3.2502, 2.9324], device='cuda:7'), covar=tensor([0.0109, 0.0088, 0.0151, 0.0164, 0.0087, 0.2325, 0.0108, 0.0211], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0136, 0.0179, 0.0164, 0.0156, 0.0193, 0.0170, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:03:44,880 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181436.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:04:15,175 INFO [train.py:904] (7/8) Epoch 18, batch 8900, loss[loss=0.1795, simple_loss=0.2778, pruned_loss=0.04066, over 15327.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2702, pruned_loss=0.03954, over 3032367.63 frames. ], batch size: 190, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:04:25,760 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.184e+02 2.557e+02 3.261e+02 8.427e+02, threshold=5.113e+02, percent-clipped=4.0 2023-04-30 19:04:45,029 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-04-30 19:05:23,276 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:05:36,507 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-04-30 19:06:18,296 INFO [train.py:904] (7/8) Epoch 18, batch 8950, loss[loss=0.168, simple_loss=0.2551, pruned_loss=0.04052, over 12604.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2693, pruned_loss=0.03963, over 3045722.97 frames. ], batch size: 246, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:07:17,087 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=181529.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:07:19,927 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=181530.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:08:08,295 INFO [train.py:904] (7/8) Epoch 18, batch 9000, loss[loss=0.1696, simple_loss=0.2603, pruned_loss=0.03941, over 12447.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2658, pruned_loss=0.0386, over 3031394.54 frames. ], batch size: 248, lr: 3.73e-03, grad_scale: 8.0 2023-04-30 19:08:08,297 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 19:08:17,837 INFO [train.py:938] (7/8) Epoch 18, validation: loss=0.1475, simple_loss=0.2516, pruned_loss=0.02169, over 944034.00 frames. 2023-04-30 19:08:17,838 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 19:08:19,031 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9575, 1.7991, 1.6123, 1.4543, 1.9799, 1.6581, 1.6268, 1.9509], device='cuda:7'), covar=tensor([0.0163, 0.0307, 0.0429, 0.0403, 0.0228, 0.0302, 0.0173, 0.0236], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0221, 0.0215, 0.0216, 0.0223, 0.0220, 0.0222, 0.0214], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:08:27,945 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 2.270e+02 2.606e+02 3.112e+02 6.247e+02, threshold=5.212e+02, percent-clipped=1.0 2023-04-30 19:09:34,175 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5679, 3.8789, 3.8851, 2.5967, 3.4242, 3.8563, 3.5708, 2.1115], device='cuda:7'), covar=tensor([0.0465, 0.0042, 0.0039, 0.0359, 0.0099, 0.0075, 0.0072, 0.0460], device='cuda:7'), in_proj_covar=tensor([0.0130, 0.0075, 0.0075, 0.0127, 0.0090, 0.0099, 0.0087, 0.0121], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 19:09:40,028 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=181591.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:10:01,204 INFO [train.py:904] (7/8) Epoch 18, batch 9050, loss[loss=0.1722, simple_loss=0.2618, pruned_loss=0.04135, over 16341.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2676, pruned_loss=0.03954, over 3038368.14 frames. ], batch size: 146, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:11:46,781 INFO [train.py:904] (7/8) Epoch 18, batch 9100, loss[loss=0.1796, simple_loss=0.2788, pruned_loss=0.04015, over 16664.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2673, pruned_loss=0.04014, over 3023706.41 frames. ], batch size: 134, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:11:50,081 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181653.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:11:55,854 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.401e+02 2.871e+02 3.524e+02 6.480e+02, threshold=5.743e+02, percent-clipped=6.0 2023-04-30 19:12:13,424 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3735, 3.0580, 2.6587, 2.2515, 2.1520, 2.2441, 3.0228, 2.8441], device='cuda:7'), covar=tensor([0.2580, 0.0714, 0.1607, 0.2605, 0.2559, 0.2140, 0.0470, 0.1430], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0254, 0.0289, 0.0294, 0.0278, 0.0240, 0.0277, 0.0313], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:13:16,910 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4250, 3.3579, 3.4649, 3.5556, 3.5915, 3.3200, 3.5552, 3.6146], device='cuda:7'), covar=tensor([0.1205, 0.0883, 0.1061, 0.0586, 0.0545, 0.1953, 0.0812, 0.0710], device='cuda:7'), in_proj_covar=tensor([0.0572, 0.0702, 0.0823, 0.0722, 0.0537, 0.0570, 0.0585, 0.0675], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:13:43,595 INFO [train.py:904] (7/8) Epoch 18, batch 9150, loss[loss=0.1707, simple_loss=0.2734, pruned_loss=0.03396, over 17117.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2674, pruned_loss=0.0395, over 3037404.42 frames. ], batch size: 49, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:15:27,787 INFO [train.py:904] (7/8) Epoch 18, batch 9200, loss[loss=0.1626, simple_loss=0.2444, pruned_loss=0.04039, over 11932.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2631, pruned_loss=0.03875, over 3020727.16 frames. ], batch size: 250, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:15:36,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.118e+02 2.449e+02 2.968e+02 5.097e+02, threshold=4.898e+02, percent-clipped=0.0 2023-04-30 19:17:05,949 INFO [train.py:904] (7/8) Epoch 18, batch 9250, loss[loss=0.1505, simple_loss=0.2445, pruned_loss=0.02826, over 16538.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2625, pruned_loss=0.03859, over 3010291.79 frames. ], batch size: 68, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:18:48,249 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9881, 3.0239, 2.7372, 4.7450, 3.4623, 4.3386, 1.6605, 3.4217], device='cuda:7'), covar=tensor([0.1280, 0.0717, 0.1116, 0.0188, 0.0199, 0.0355, 0.1612, 0.0601], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0165, 0.0187, 0.0173, 0.0197, 0.0208, 0.0192, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 19:18:57,438 INFO [train.py:904] (7/8) Epoch 18, batch 9300, loss[loss=0.1757, simple_loss=0.2653, pruned_loss=0.04307, over 15260.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2611, pruned_loss=0.03813, over 3010799.17 frames. ], batch size: 191, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:19:06,081 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.346e+02 2.677e+02 3.465e+02 6.012e+02, threshold=5.355e+02, percent-clipped=4.0 2023-04-30 19:19:10,712 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2117, 5.2631, 5.0865, 4.6446, 4.7794, 5.1670, 4.9993, 4.7778], device='cuda:7'), covar=tensor([0.0549, 0.0473, 0.0237, 0.0284, 0.0828, 0.0486, 0.0297, 0.0636], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0379, 0.0313, 0.0301, 0.0319, 0.0352, 0.0216, 0.0372], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:20:12,922 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=181886.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:20:15,657 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-30 19:20:41,436 INFO [train.py:904] (7/8) Epoch 18, batch 9350, loss[loss=0.1776, simple_loss=0.2688, pruned_loss=0.04315, over 16806.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2603, pruned_loss=0.03782, over 3025340.30 frames. ], batch size: 124, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:22:24,027 INFO [train.py:904] (7/8) Epoch 18, batch 9400, loss[loss=0.1716, simple_loss=0.2701, pruned_loss=0.03661, over 16836.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2603, pruned_loss=0.03727, over 3037697.91 frames. ], batch size: 124, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:22:28,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=181953.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:22:33,106 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.111e+02 2.455e+02 3.021e+02 7.291e+02, threshold=4.911e+02, percent-clipped=2.0 2023-04-30 19:23:16,345 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=181978.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:23:59,203 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7960, 3.7541, 3.9298, 3.6948, 3.9051, 4.2847, 3.9549, 3.6868], device='cuda:7'), covar=tensor([0.2131, 0.2677, 0.2319, 0.2895, 0.2922, 0.1577, 0.1587, 0.2733], device='cuda:7'), in_proj_covar=tensor([0.0367, 0.0532, 0.0591, 0.0450, 0.0596, 0.0621, 0.0465, 0.0598], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 19:24:05,243 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182001.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:24:06,048 INFO [train.py:904] (7/8) Epoch 18, batch 9450, loss[loss=0.1649, simple_loss=0.2613, pruned_loss=0.03427, over 15246.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2629, pruned_loss=0.03769, over 3054038.22 frames. ], batch size: 191, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:25:19,137 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182039.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:25:41,960 INFO [train.py:904] (7/8) Epoch 18, batch 9500, loss[loss=0.1826, simple_loss=0.2744, pruned_loss=0.04539, over 16334.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2622, pruned_loss=0.03749, over 3046031.26 frames. ], batch size: 146, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:25:51,890 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 19:25:55,089 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.182e+02 2.603e+02 3.134e+02 5.554e+02, threshold=5.207e+02, percent-clipped=1.0 2023-04-30 19:27:27,078 INFO [train.py:904] (7/8) Epoch 18, batch 9550, loss[loss=0.1782, simple_loss=0.2734, pruned_loss=0.04151, over 16930.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2622, pruned_loss=0.03762, over 3056285.79 frames. ], batch size: 125, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:29:06,830 INFO [train.py:904] (7/8) Epoch 18, batch 9600, loss[loss=0.1932, simple_loss=0.2917, pruned_loss=0.04735, over 16205.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2635, pruned_loss=0.03837, over 3042034.55 frames. ], batch size: 165, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:29:15,385 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 2.461e+02 2.915e+02 3.491e+02 6.374e+02, threshold=5.830e+02, percent-clipped=5.0 2023-04-30 19:30:14,150 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=182186.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:30:33,423 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.00 vs. limit=5.0 2023-04-30 19:30:52,031 INFO [train.py:904] (7/8) Epoch 18, batch 9650, loss[loss=0.1872, simple_loss=0.2742, pruned_loss=0.05013, over 16998.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.266, pruned_loss=0.03889, over 3042963.79 frames. ], batch size: 109, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:32:00,719 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182234.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:32:37,732 INFO [train.py:904] (7/8) Epoch 18, batch 9700, loss[loss=0.1672, simple_loss=0.2563, pruned_loss=0.03903, over 12423.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2648, pruned_loss=0.03877, over 3034975.49 frames. ], batch size: 248, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:32:45,795 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.275e+02 2.741e+02 3.246e+02 5.302e+02, threshold=5.483e+02, percent-clipped=0.0 2023-04-30 19:33:37,495 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182281.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:34:17,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9619, 3.1706, 2.7511, 4.9076, 3.6966, 4.3821, 1.6410, 3.4604], device='cuda:7'), covar=tensor([0.1303, 0.0696, 0.1099, 0.0155, 0.0177, 0.0351, 0.1616, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0165, 0.0187, 0.0171, 0.0194, 0.0207, 0.0191, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:7') 2023-04-30 19:34:17,897 INFO [train.py:904] (7/8) Epoch 18, batch 9750, loss[loss=0.1553, simple_loss=0.247, pruned_loss=0.03175, over 16643.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2638, pruned_loss=0.03893, over 3051619.71 frames. ], batch size: 62, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:34:40,478 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0827, 4.1512, 4.4889, 4.4174, 4.4826, 4.2691, 4.1185, 4.1902], device='cuda:7'), covar=tensor([0.0484, 0.0947, 0.0571, 0.0831, 0.0701, 0.0634, 0.1253, 0.0524], device='cuda:7'), in_proj_covar=tensor([0.0365, 0.0397, 0.0388, 0.0369, 0.0432, 0.0409, 0.0495, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 19:35:09,044 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5076, 3.6191, 2.7011, 2.1437, 2.2407, 2.4190, 3.8007, 3.1133], device='cuda:7'), covar=tensor([0.2874, 0.0635, 0.1835, 0.2976, 0.2968, 0.2009, 0.0399, 0.1429], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0253, 0.0288, 0.0292, 0.0275, 0.0239, 0.0277, 0.0311], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:35:25,187 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=182334.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:35:39,148 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182342.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:35:56,067 INFO [train.py:904] (7/8) Epoch 18, batch 9800, loss[loss=0.1617, simple_loss=0.2492, pruned_loss=0.03708, over 12231.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2638, pruned_loss=0.03776, over 3063943.53 frames. ], batch size: 248, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:36:05,432 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.961e+02 2.421e+02 2.847e+02 6.201e+02, threshold=4.843e+02, percent-clipped=1.0 2023-04-30 19:37:39,093 INFO [train.py:904] (7/8) Epoch 18, batch 9850, loss[loss=0.1723, simple_loss=0.2696, pruned_loss=0.03746, over 16801.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2645, pruned_loss=0.03719, over 3064132.35 frames. ], batch size: 83, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:39:29,923 INFO [train.py:904] (7/8) Epoch 18, batch 9900, loss[loss=0.1572, simple_loss=0.2651, pruned_loss=0.02467, over 16718.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2652, pruned_loss=0.03717, over 3066575.68 frames. ], batch size: 89, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:39:40,689 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.082e+02 2.451e+02 2.864e+02 7.377e+02, threshold=4.903e+02, percent-clipped=2.0 2023-04-30 19:41:28,680 INFO [train.py:904] (7/8) Epoch 18, batch 9950, loss[loss=0.1647, simple_loss=0.2583, pruned_loss=0.0355, over 12529.00 frames. ], tot_loss[loss=0.171, simple_loss=0.267, pruned_loss=0.03753, over 3064561.81 frames. ], batch size: 248, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:41:58,934 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182514.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:43:31,285 INFO [train.py:904] (7/8) Epoch 18, batch 10000, loss[loss=0.1463, simple_loss=0.2473, pruned_loss=0.02265, over 16919.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2656, pruned_loss=0.03712, over 3074621.57 frames. ], batch size: 96, lr: 3.72e-03, grad_scale: 8.0 2023-04-30 19:43:32,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2236, 4.2396, 4.1048, 3.5004, 4.1285, 1.6137, 3.9176, 3.8038], device='cuda:7'), covar=tensor([0.0099, 0.0087, 0.0163, 0.0214, 0.0092, 0.2749, 0.0123, 0.0230], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0134, 0.0175, 0.0158, 0.0154, 0.0191, 0.0166, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:43:42,214 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.117e+02 2.346e+02 2.932e+02 5.510e+02, threshold=4.691e+02, percent-clipped=3.0 2023-04-30 19:44:17,883 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182575.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:45:14,031 INFO [train.py:904] (7/8) Epoch 18, batch 10050, loss[loss=0.1707, simple_loss=0.2598, pruned_loss=0.04083, over 12071.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2659, pruned_loss=0.03716, over 3077910.06 frames. ], batch size: 248, lr: 3.71e-03, grad_scale: 8.0 2023-04-30 19:46:11,623 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8234, 1.2509, 1.7140, 1.6698, 1.8041, 1.9622, 1.5762, 1.8515], device='cuda:7'), covar=tensor([0.0235, 0.0411, 0.0224, 0.0285, 0.0294, 0.0189, 0.0408, 0.0118], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0180, 0.0169, 0.0168, 0.0179, 0.0137, 0.0183, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:46:14,968 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=182634.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:46:20,794 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=182637.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:46:46,725 INFO [train.py:904] (7/8) Epoch 18, batch 10100, loss[loss=0.1662, simple_loss=0.2578, pruned_loss=0.03727, over 16810.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2661, pruned_loss=0.0374, over 3070666.72 frames. ], batch size: 90, lr: 3.71e-03, grad_scale: 8.0 2023-04-30 19:46:55,749 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.131e+02 2.554e+02 3.108e+02 6.507e+02, threshold=5.108e+02, percent-clipped=7.0 2023-04-30 19:47:44,866 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182682.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:48:00,876 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-04-30 19:48:32,151 INFO [train.py:904] (7/8) Epoch 19, batch 0, loss[loss=0.1933, simple_loss=0.2806, pruned_loss=0.05299, over 16777.00 frames. ], tot_loss[loss=0.1933, simple_loss=0.2806, pruned_loss=0.05299, over 16777.00 frames. ], batch size: 57, lr: 3.61e-03, grad_scale: 8.0 2023-04-30 19:48:32,151 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 19:48:39,776 INFO [train.py:938] (7/8) Epoch 19, validation: loss=0.1468, simple_loss=0.2504, pruned_loss=0.0216, over 944034.00 frames. 2023-04-30 19:48:39,777 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 19:49:37,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9655, 5.3401, 5.0482, 5.0681, 4.7874, 4.7198, 4.7647, 5.4433], device='cuda:7'), covar=tensor([0.1286, 0.0910, 0.1153, 0.0935, 0.0861, 0.1119, 0.1283, 0.0878], device='cuda:7'), in_proj_covar=tensor([0.0612, 0.0750, 0.0615, 0.0557, 0.0471, 0.0483, 0.0626, 0.0580], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:49:50,022 INFO [train.py:904] (7/8) Epoch 19, batch 50, loss[loss=0.1856, simple_loss=0.2627, pruned_loss=0.05423, over 16863.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2697, pruned_loss=0.04996, over 759743.45 frames. ], batch size: 96, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:49:59,029 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.571e+02 3.042e+02 3.790e+02 6.505e+02, threshold=6.085e+02, percent-clipped=6.0 2023-04-30 19:50:55,407 INFO [train.py:904] (7/8) Epoch 19, batch 100, loss[loss=0.176, simple_loss=0.2562, pruned_loss=0.04791, over 16774.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2673, pruned_loss=0.04925, over 1324154.03 frames. ], batch size: 124, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:52:03,148 INFO [train.py:904] (7/8) Epoch 19, batch 150, loss[loss=0.1878, simple_loss=0.2612, pruned_loss=0.05726, over 16874.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2656, pruned_loss=0.0471, over 1766798.40 frames. ], batch size: 109, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:52:10,227 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182856.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:52:15,180 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.297e+02 2.720e+02 3.223e+02 8.471e+02, threshold=5.441e+02, percent-clipped=1.0 2023-04-30 19:52:18,099 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=182862.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:52:28,145 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=182870.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:53:13,457 INFO [train.py:904] (7/8) Epoch 19, batch 200, loss[loss=0.1978, simple_loss=0.2718, pruned_loss=0.06186, over 16791.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2657, pruned_loss=0.04777, over 2106631.74 frames. ], batch size: 83, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:53:31,905 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1450, 4.9038, 5.1091, 5.3192, 5.5257, 4.8301, 5.4766, 5.5052], device='cuda:7'), covar=tensor([0.1763, 0.1126, 0.1747, 0.0810, 0.0596, 0.0761, 0.0567, 0.0632], device='cuda:7'), in_proj_covar=tensor([0.0583, 0.0719, 0.0839, 0.0737, 0.0549, 0.0576, 0.0599, 0.0688], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:53:34,337 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182917.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:53:42,591 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=182923.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:54:01,089 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=182937.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:54:21,120 INFO [train.py:904] (7/8) Epoch 19, batch 250, loss[loss=0.1892, simple_loss=0.2671, pruned_loss=0.05562, over 12093.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.2616, pruned_loss=0.04588, over 2379643.58 frames. ], batch size: 246, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:54:32,989 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.306e+02 2.752e+02 3.234e+02 1.372e+03, threshold=5.503e+02, percent-clipped=2.0 2023-04-30 19:55:08,834 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=182985.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 19:55:30,357 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-04-30 19:55:30,557 INFO [train.py:904] (7/8) Epoch 19, batch 300, loss[loss=0.1815, simple_loss=0.2557, pruned_loss=0.05366, over 16879.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2609, pruned_loss=0.04574, over 2596036.03 frames. ], batch size: 116, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:56:07,227 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6880, 3.7171, 2.2561, 4.2503, 2.9326, 4.2284, 2.4432, 3.0945], device='cuda:7'), covar=tensor([0.0285, 0.0418, 0.1610, 0.0341, 0.0849, 0.0478, 0.1535, 0.0732], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0172, 0.0192, 0.0152, 0.0174, 0.0209, 0.0200, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 19:56:41,131 INFO [train.py:904] (7/8) Epoch 19, batch 350, loss[loss=0.1877, simple_loss=0.261, pruned_loss=0.05713, over 16855.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2588, pruned_loss=0.04516, over 2764913.58 frames. ], batch size: 90, lr: 3.61e-03, grad_scale: 1.0 2023-04-30 19:56:52,345 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.344e+02 2.193e+02 2.670e+02 3.333e+02 1.052e+03, threshold=5.340e+02, percent-clipped=4.0 2023-04-30 19:57:51,267 INFO [train.py:904] (7/8) Epoch 19, batch 400, loss[loss=0.1732, simple_loss=0.2633, pruned_loss=0.04157, over 16659.00 frames. ], tot_loss[loss=0.174, simple_loss=0.258, pruned_loss=0.04498, over 2884667.13 frames. ], batch size: 62, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 19:58:33,800 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0606, 4.8611, 5.0794, 5.2878, 5.4989, 4.7362, 5.4860, 5.4796], device='cuda:7'), covar=tensor([0.1912, 0.1275, 0.1648, 0.0799, 0.0566, 0.0892, 0.0529, 0.0652], device='cuda:7'), in_proj_covar=tensor([0.0601, 0.0739, 0.0862, 0.0756, 0.0562, 0.0592, 0.0615, 0.0708], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 19:59:03,157 INFO [train.py:904] (7/8) Epoch 19, batch 450, loss[loss=0.1606, simple_loss=0.2404, pruned_loss=0.04037, over 16475.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2566, pruned_loss=0.04456, over 2983384.36 frames. ], batch size: 75, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 19:59:14,100 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.162e+02 2.492e+02 3.004e+02 8.634e+02, threshold=4.984e+02, percent-clipped=1.0 2023-04-30 19:59:18,872 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-04-30 19:59:28,186 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=183170.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:12,295 INFO [train.py:904] (7/8) Epoch 19, batch 500, loss[loss=0.1445, simple_loss=0.2347, pruned_loss=0.0271, over 17224.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2558, pruned_loss=0.04368, over 3052515.53 frames. ], batch size: 44, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:00:28,418 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183212.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:35,919 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=183218.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:35,938 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183218.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:00:44,638 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-04-30 20:01:23,256 INFO [train.py:904] (7/8) Epoch 19, batch 550, loss[loss=0.1459, simple_loss=0.2376, pruned_loss=0.02705, over 17148.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2547, pruned_loss=0.04298, over 3118670.33 frames. ], batch size: 47, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:01:34,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.167e+02 2.560e+02 2.885e+02 5.610e+02, threshold=5.120e+02, percent-clipped=1.0 2023-04-30 20:02:32,735 INFO [train.py:904] (7/8) Epoch 19, batch 600, loss[loss=0.1882, simple_loss=0.2816, pruned_loss=0.04739, over 17039.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2549, pruned_loss=0.04272, over 3166540.69 frames. ], batch size: 55, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:03:13,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1047, 2.4225, 2.6289, 1.9047, 2.7635, 2.7490, 2.4352, 2.4246], device='cuda:7'), covar=tensor([0.0713, 0.0271, 0.0237, 0.0947, 0.0135, 0.0301, 0.0453, 0.0414], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0094, 0.0140, 0.0077, 0.0121, 0.0126, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 20:03:40,905 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3562, 4.3799, 4.7074, 4.6836, 4.6921, 4.4314, 4.4294, 4.2987], device='cuda:7'), covar=tensor([0.0358, 0.0687, 0.0436, 0.0436, 0.0501, 0.0426, 0.0821, 0.0644], device='cuda:7'), in_proj_covar=tensor([0.0392, 0.0429, 0.0418, 0.0394, 0.0464, 0.0441, 0.0530, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 20:03:42,747 INFO [train.py:904] (7/8) Epoch 19, batch 650, loss[loss=0.174, simple_loss=0.2565, pruned_loss=0.04571, over 16503.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2527, pruned_loss=0.04231, over 3198089.00 frames. ], batch size: 75, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:03:54,611 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.241e+02 2.597e+02 3.196e+02 6.464e+02, threshold=5.194e+02, percent-clipped=2.0 2023-04-30 20:04:53,085 INFO [train.py:904] (7/8) Epoch 19, batch 700, loss[loss=0.1856, simple_loss=0.2781, pruned_loss=0.04655, over 16701.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2532, pruned_loss=0.04235, over 3234154.65 frames. ], batch size: 62, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:05:01,077 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=183408.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:05:12,603 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 20:05:59,960 INFO [train.py:904] (7/8) Epoch 19, batch 750, loss[loss=0.1618, simple_loss=0.2405, pruned_loss=0.04161, over 16752.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2535, pruned_loss=0.04283, over 3255315.47 frames. ], batch size: 83, lr: 3.61e-03, grad_scale: 2.0 2023-04-30 20:06:11,900 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.150e+02 2.519e+02 3.057e+02 7.230e+02, threshold=5.039e+02, percent-clipped=2.0 2023-04-30 20:06:25,263 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=183469.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:07:10,068 INFO [train.py:904] (7/8) Epoch 19, batch 800, loss[loss=0.1478, simple_loss=0.2321, pruned_loss=0.03171, over 16993.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2528, pruned_loss=0.04225, over 3276092.10 frames. ], batch size: 41, lr: 3.61e-03, grad_scale: 4.0 2023-04-30 20:07:24,234 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=183512.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:07:32,509 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=183518.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:07:32,598 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1810, 3.1795, 3.3298, 2.2763, 3.1015, 3.4155, 3.1750, 1.8463], device='cuda:7'), covar=tensor([0.0490, 0.0115, 0.0064, 0.0403, 0.0119, 0.0105, 0.0094, 0.0522], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0080, 0.0080, 0.0131, 0.0095, 0.0104, 0.0091, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:08:07,037 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7451, 2.5965, 2.4580, 3.8299, 3.1461, 3.9426, 1.4711, 2.8099], device='cuda:7'), covar=tensor([0.1398, 0.0745, 0.1197, 0.0198, 0.0168, 0.0380, 0.1625, 0.0861], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0170, 0.0192, 0.0180, 0.0200, 0.0213, 0.0196, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 20:08:19,002 INFO [train.py:904] (7/8) Epoch 19, batch 850, loss[loss=0.1581, simple_loss=0.2395, pruned_loss=0.03838, over 16471.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2519, pruned_loss=0.04194, over 3291219.08 frames. ], batch size: 146, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:08:29,788 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.149e+02 2.547e+02 3.022e+02 7.911e+02, threshold=5.094e+02, percent-clipped=1.0 2023-04-30 20:08:30,866 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=183560.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:08:38,502 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=183566.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:09:28,461 INFO [train.py:904] (7/8) Epoch 19, batch 900, loss[loss=0.1435, simple_loss=0.2246, pruned_loss=0.0312, over 16777.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2514, pruned_loss=0.04167, over 3303040.86 frames. ], batch size: 39, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:09:38,680 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9679, 4.8902, 4.6791, 3.4666, 4.8113, 1.6673, 4.4186, 4.4896], device='cuda:7'), covar=tensor([0.0190, 0.0155, 0.0301, 0.0879, 0.0175, 0.3704, 0.0246, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0144, 0.0188, 0.0172, 0.0165, 0.0201, 0.0179, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:09:44,279 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=183614.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:10:15,776 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-04-30 20:10:35,724 INFO [train.py:904] (7/8) Epoch 19, batch 950, loss[loss=0.1587, simple_loss=0.2474, pruned_loss=0.03506, over 17211.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2519, pruned_loss=0.04175, over 3316350.53 frames. ], batch size: 44, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:10:45,890 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 1.990e+02 2.288e+02 2.656e+02 5.552e+02, threshold=4.575e+02, percent-clipped=1.0 2023-04-30 20:11:08,308 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=183675.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:11:42,846 INFO [train.py:904] (7/8) Epoch 19, batch 1000, loss[loss=0.1675, simple_loss=0.259, pruned_loss=0.03797, over 17090.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2498, pruned_loss=0.0416, over 3321378.25 frames. ], batch size: 53, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:12:09,092 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0384, 3.0133, 3.2327, 2.1012, 2.8748, 2.2327, 3.5256, 3.3884], device='cuda:7'), covar=tensor([0.0215, 0.0932, 0.0626, 0.1803, 0.0828, 0.0979, 0.0501, 0.0910], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0158, 0.0165, 0.0151, 0.0142, 0.0127, 0.0142, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 20:12:50,602 INFO [train.py:904] (7/8) Epoch 19, batch 1050, loss[loss=0.1801, simple_loss=0.2532, pruned_loss=0.05356, over 16722.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2503, pruned_loss=0.04215, over 3328303.38 frames. ], batch size: 89, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:13:01,770 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.214e+02 2.595e+02 3.147e+02 6.979e+02, threshold=5.191e+02, percent-clipped=4.0 2023-04-30 20:13:06,527 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183764.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:13:41,940 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2590, 2.1203, 2.3497, 3.8987, 2.0497, 2.4121, 2.2773, 2.2451], device='cuda:7'), covar=tensor([0.1695, 0.4588, 0.3268, 0.0845, 0.5434, 0.3407, 0.4046, 0.4498], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0435, 0.0360, 0.0324, 0.0433, 0.0499, 0.0405, 0.0507], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:14:00,548 INFO [train.py:904] (7/8) Epoch 19, batch 1100, loss[loss=0.1481, simple_loss=0.2394, pruned_loss=0.02842, over 17196.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2493, pruned_loss=0.04142, over 3329945.16 frames. ], batch size: 46, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:14:03,334 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2371, 5.8036, 5.9117, 5.5763, 5.6450, 6.2453, 5.7495, 5.4470], device='cuda:7'), covar=tensor([0.0838, 0.1941, 0.2272, 0.2183, 0.2883, 0.1011, 0.1642, 0.2371], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0577, 0.0645, 0.0487, 0.0652, 0.0677, 0.0503, 0.0652], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:14:36,829 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3480, 5.3130, 5.2188, 4.6685, 4.7760, 5.2349, 5.2018, 4.8761], device='cuda:7'), covar=tensor([0.0586, 0.0466, 0.0289, 0.0342, 0.1190, 0.0457, 0.0343, 0.0738], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0414, 0.0341, 0.0329, 0.0351, 0.0385, 0.0233, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:14:56,594 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=183842.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:15:08,306 INFO [train.py:904] (7/8) Epoch 19, batch 1150, loss[loss=0.1535, simple_loss=0.2475, pruned_loss=0.02976, over 17133.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2496, pruned_loss=0.04096, over 3322251.14 frames. ], batch size: 49, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:15:20,305 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.057e+02 2.490e+02 3.144e+02 5.373e+02, threshold=4.980e+02, percent-clipped=1.0 2023-04-30 20:16:18,544 INFO [train.py:904] (7/8) Epoch 19, batch 1200, loss[loss=0.1419, simple_loss=0.2297, pruned_loss=0.02708, over 16943.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2491, pruned_loss=0.04066, over 3320690.99 frames. ], batch size: 41, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:16:20,060 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=183903.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:16:47,390 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-04-30 20:17:25,067 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-04-30 20:17:25,278 INFO [train.py:904] (7/8) Epoch 19, batch 1250, loss[loss=0.1673, simple_loss=0.2614, pruned_loss=0.03664, over 17126.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2502, pruned_loss=0.04166, over 3323056.48 frames. ], batch size: 49, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:17:35,900 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.248e+02 2.515e+02 3.055e+02 4.782e+02, threshold=5.031e+02, percent-clipped=0.0 2023-04-30 20:17:50,877 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=183970.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:18:15,912 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4895, 5.4560, 5.3042, 4.7963, 4.8797, 5.3536, 5.3197, 4.9366], device='cuda:7'), covar=tensor([0.0520, 0.0434, 0.0284, 0.0333, 0.1085, 0.0415, 0.0241, 0.0734], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0416, 0.0344, 0.0332, 0.0352, 0.0387, 0.0234, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:18:19,857 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 20:18:37,499 INFO [train.py:904] (7/8) Epoch 19, batch 1300, loss[loss=0.188, simple_loss=0.2678, pruned_loss=0.05412, over 16496.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2495, pruned_loss=0.0411, over 3325029.56 frames. ], batch size: 146, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:19:44,253 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184050.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:19:46,330 INFO [train.py:904] (7/8) Epoch 19, batch 1350, loss[loss=0.165, simple_loss=0.2582, pruned_loss=0.03591, over 17245.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2502, pruned_loss=0.04082, over 3331880.97 frames. ], batch size: 52, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:19:48,878 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4578, 3.4734, 2.2475, 3.6846, 2.7629, 3.6492, 2.1851, 2.8129], device='cuda:7'), covar=tensor([0.0249, 0.0391, 0.1374, 0.0345, 0.0688, 0.0713, 0.1342, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0176, 0.0195, 0.0159, 0.0175, 0.0215, 0.0203, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 20:19:55,441 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.201e+02 2.541e+02 3.098e+02 6.218e+02, threshold=5.081e+02, percent-clipped=6.0 2023-04-30 20:19:58,629 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2023-04-30 20:20:03,845 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184064.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:20:55,162 INFO [train.py:904] (7/8) Epoch 19, batch 1400, loss[loss=0.1494, simple_loss=0.2353, pruned_loss=0.03175, over 16515.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2497, pruned_loss=0.04075, over 3323508.85 frames. ], batch size: 68, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:21:09,640 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184111.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:21:10,496 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184112.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:22:05,085 INFO [train.py:904] (7/8) Epoch 19, batch 1450, loss[loss=0.183, simple_loss=0.2764, pruned_loss=0.04479, over 17122.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2493, pruned_loss=0.04087, over 3319578.62 frames. ], batch size: 47, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:22:15,460 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.293e+02 2.595e+02 3.219e+02 5.738e+02, threshold=5.191e+02, percent-clipped=1.0 2023-04-30 20:22:18,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7550, 4.6293, 4.6299, 4.3174, 4.3357, 4.6845, 4.5104, 4.4275], device='cuda:7'), covar=tensor([0.0606, 0.0734, 0.0324, 0.0309, 0.0874, 0.0444, 0.0477, 0.0676], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0418, 0.0344, 0.0333, 0.0353, 0.0388, 0.0235, 0.0408], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:23:08,681 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184198.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:23:13,294 INFO [train.py:904] (7/8) Epoch 19, batch 1500, loss[loss=0.1616, simple_loss=0.242, pruned_loss=0.04059, over 12099.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2494, pruned_loss=0.04116, over 3318102.27 frames. ], batch size: 246, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:24:22,396 INFO [train.py:904] (7/8) Epoch 19, batch 1550, loss[loss=0.1936, simple_loss=0.2728, pruned_loss=0.0572, over 15408.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2505, pruned_loss=0.04161, over 3320320.62 frames. ], batch size: 190, lr: 3.60e-03, grad_scale: 4.0 2023-04-30 20:24:34,829 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.353e+02 2.741e+02 3.128e+02 4.649e+02, threshold=5.482e+02, percent-clipped=0.0 2023-04-30 20:24:47,454 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184270.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:25:31,176 INFO [train.py:904] (7/8) Epoch 19, batch 1600, loss[loss=0.1585, simple_loss=0.2523, pruned_loss=0.03236, over 17143.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2523, pruned_loss=0.04223, over 3323401.04 frames. ], batch size: 46, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:25:41,071 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1153, 3.8841, 4.3373, 2.2162, 4.5203, 4.7126, 3.3803, 3.5897], device='cuda:7'), covar=tensor([0.0695, 0.0279, 0.0263, 0.1213, 0.0100, 0.0173, 0.0414, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0140, 0.0078, 0.0123, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 20:25:53,349 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184318.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:26:39,949 INFO [train.py:904] (7/8) Epoch 19, batch 1650, loss[loss=0.1941, simple_loss=0.2908, pruned_loss=0.04866, over 16777.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.255, pruned_loss=0.04303, over 3310226.94 frames. ], batch size: 57, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:26:50,283 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184359.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:26:52,313 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.220e+02 2.674e+02 3.187e+02 5.456e+02, threshold=5.348e+02, percent-clipped=0.0 2023-04-30 20:27:49,679 INFO [train.py:904] (7/8) Epoch 19, batch 1700, loss[loss=0.1988, simple_loss=0.2815, pruned_loss=0.05803, over 15572.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2567, pruned_loss=0.04326, over 3315504.62 frames. ], batch size: 191, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:27:50,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6299, 2.4305, 1.9632, 2.1214, 2.7823, 2.5284, 2.7518, 2.8627], device='cuda:7'), covar=tensor([0.0204, 0.0329, 0.0470, 0.0420, 0.0215, 0.0280, 0.0212, 0.0239], device='cuda:7'), in_proj_covar=tensor([0.0204, 0.0235, 0.0226, 0.0227, 0.0236, 0.0236, 0.0239, 0.0230], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:27:55,050 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184406.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:28:14,041 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:28:38,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8438, 3.5230, 3.9408, 2.1358, 4.0721, 4.0744, 3.2713, 3.0648], device='cuda:7'), covar=tensor([0.0691, 0.0249, 0.0153, 0.1118, 0.0077, 0.0172, 0.0357, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0140, 0.0078, 0.0123, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 20:28:58,464 INFO [train.py:904] (7/8) Epoch 19, batch 1750, loss[loss=0.18, simple_loss=0.2612, pruned_loss=0.0494, over 16681.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.257, pruned_loss=0.04367, over 3317090.58 frames. ], batch size: 89, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:29:06,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8591, 2.0006, 2.4766, 2.8089, 2.5978, 3.3487, 2.2408, 3.3455], device='cuda:7'), covar=tensor([0.0234, 0.0458, 0.0331, 0.0292, 0.0327, 0.0170, 0.0441, 0.0152], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0191, 0.0178, 0.0180, 0.0190, 0.0147, 0.0193, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:29:10,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.342e+02 2.772e+02 3.206e+02 7.183e+02, threshold=5.544e+02, percent-clipped=2.0 2023-04-30 20:29:45,807 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184486.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:30:02,715 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184498.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:30:07,532 INFO [train.py:904] (7/8) Epoch 19, batch 1800, loss[loss=0.171, simple_loss=0.2659, pruned_loss=0.03811, over 17050.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2577, pruned_loss=0.0433, over 3315568.62 frames. ], batch size: 53, lr: 3.60e-03, grad_scale: 8.0 2023-04-30 20:31:07,267 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184546.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:31:09,224 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184547.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:31:16,704 INFO [train.py:904] (7/8) Epoch 19, batch 1850, loss[loss=0.1878, simple_loss=0.2842, pruned_loss=0.04573, over 16679.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2587, pruned_loss=0.04334, over 3318457.31 frames. ], batch size: 57, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:31:29,856 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.167e+02 2.563e+02 3.002e+02 5.511e+02, threshold=5.126e+02, percent-clipped=0.0 2023-04-30 20:31:34,713 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8426, 5.1512, 4.9131, 4.9315, 4.6879, 4.6569, 4.5720, 5.2344], device='cuda:7'), covar=tensor([0.1187, 0.0873, 0.1061, 0.0797, 0.0839, 0.1129, 0.1247, 0.0893], device='cuda:7'), in_proj_covar=tensor([0.0660, 0.0813, 0.0666, 0.0603, 0.0508, 0.0516, 0.0674, 0.0626], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:32:26,189 INFO [train.py:904] (7/8) Epoch 19, batch 1900, loss[loss=0.1655, simple_loss=0.2645, pruned_loss=0.03325, over 17158.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2584, pruned_loss=0.04285, over 3321006.66 frames. ], batch size: 47, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:32:35,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7772, 4.6129, 4.8360, 5.0259, 5.2015, 4.5554, 5.1851, 5.1792], device='cuda:7'), covar=tensor([0.1729, 0.1305, 0.1703, 0.0775, 0.0501, 0.0965, 0.0607, 0.0613], device='cuda:7'), in_proj_covar=tensor([0.0638, 0.0786, 0.0920, 0.0800, 0.0596, 0.0628, 0.0644, 0.0748], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:33:07,410 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9804, 3.9213, 4.4205, 2.0799, 4.5756, 4.6967, 3.3784, 3.4739], device='cuda:7'), covar=tensor([0.0765, 0.0239, 0.0194, 0.1218, 0.0067, 0.0140, 0.0405, 0.0414], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0108, 0.0097, 0.0141, 0.0079, 0.0125, 0.0128, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 20:33:35,933 INFO [train.py:904] (7/8) Epoch 19, batch 1950, loss[loss=0.1982, simple_loss=0.2839, pruned_loss=0.05628, over 12564.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2574, pruned_loss=0.04239, over 3317363.13 frames. ], batch size: 248, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:33:48,871 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.129e+02 2.542e+02 2.951e+02 4.781e+02, threshold=5.085e+02, percent-clipped=0.0 2023-04-30 20:33:59,135 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8556, 4.1892, 3.0572, 2.3605, 2.7397, 2.5550, 4.5181, 3.5795], device='cuda:7'), covar=tensor([0.2708, 0.0616, 0.1758, 0.2682, 0.2709, 0.2026, 0.0389, 0.1337], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0266, 0.0300, 0.0303, 0.0292, 0.0251, 0.0287, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:34:47,053 INFO [train.py:904] (7/8) Epoch 19, batch 2000, loss[loss=0.1503, simple_loss=0.2334, pruned_loss=0.03357, over 16815.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2574, pruned_loss=0.04211, over 3311078.54 frames. ], batch size: 42, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:34:53,180 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=184706.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:35:06,421 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184715.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:35:56,927 INFO [train.py:904] (7/8) Epoch 19, batch 2050, loss[loss=0.1915, simple_loss=0.2762, pruned_loss=0.05336, over 15532.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2574, pruned_loss=0.04212, over 3315787.45 frames. ], batch size: 190, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:35:59,955 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=184754.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:36:10,014 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.173e+02 2.577e+02 2.933e+02 5.913e+02, threshold=5.153e+02, percent-clipped=1.0 2023-04-30 20:36:16,129 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9128, 5.1357, 5.3716, 5.0702, 5.1131, 5.7725, 5.2881, 4.9452], device='cuda:7'), covar=tensor([0.1196, 0.2005, 0.2163, 0.2035, 0.2769, 0.0965, 0.1575, 0.2392], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0583, 0.0647, 0.0489, 0.0658, 0.0685, 0.0507, 0.0656], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:37:10,306 INFO [train.py:904] (7/8) Epoch 19, batch 2100, loss[loss=0.1693, simple_loss=0.255, pruned_loss=0.04182, over 16397.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2585, pruned_loss=0.04282, over 3315110.79 frames. ], batch size: 68, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:38:05,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1630, 5.6849, 5.8405, 5.5189, 5.6458, 6.1791, 5.6478, 5.3683], device='cuda:7'), covar=tensor([0.0820, 0.2025, 0.2347, 0.1982, 0.2632, 0.0906, 0.1502, 0.2306], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0583, 0.0648, 0.0489, 0.0656, 0.0685, 0.0507, 0.0654], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:38:05,098 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=184842.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:38:18,977 INFO [train.py:904] (7/8) Epoch 19, batch 2150, loss[loss=0.1443, simple_loss=0.2349, pruned_loss=0.02689, over 17197.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2591, pruned_loss=0.04337, over 3313855.63 frames. ], batch size: 45, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:38:30,742 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.154e+02 2.741e+02 3.179e+02 7.758e+02, threshold=5.482e+02, percent-clipped=2.0 2023-04-30 20:38:38,921 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184866.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:38:58,159 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 20:39:27,110 INFO [train.py:904] (7/8) Epoch 19, batch 2200, loss[loss=0.1893, simple_loss=0.2668, pruned_loss=0.05588, over 16711.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2602, pruned_loss=0.04385, over 3310732.76 frames. ], batch size: 134, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:40:01,247 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=184927.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:40:17,960 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=184939.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:40:35,952 INFO [train.py:904] (7/8) Epoch 19, batch 2250, loss[loss=0.1853, simple_loss=0.2832, pruned_loss=0.04372, over 17060.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.261, pruned_loss=0.04431, over 3304986.53 frames. ], batch size: 53, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:40:48,429 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.330e+02 2.636e+02 3.046e+02 4.750e+02, threshold=5.271e+02, percent-clipped=0.0 2023-04-30 20:41:45,005 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185000.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:41:46,803 INFO [train.py:904] (7/8) Epoch 19, batch 2300, loss[loss=0.1822, simple_loss=0.2744, pruned_loss=0.04498, over 16468.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2617, pruned_loss=0.04488, over 3308361.69 frames. ], batch size: 68, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:42:05,923 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185015.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:42:26,696 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 20:42:44,861 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.28 vs. limit=5.0 2023-04-30 20:42:56,746 INFO [train.py:904] (7/8) Epoch 19, batch 2350, loss[loss=0.1867, simple_loss=0.2634, pruned_loss=0.05496, over 16842.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2617, pruned_loss=0.04512, over 3311201.16 frames. ], batch size: 102, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:43:08,919 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185060.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:43:09,638 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.245e+02 2.599e+02 3.315e+02 5.469e+02, threshold=5.198e+02, percent-clipped=1.0 2023-04-30 20:43:12,099 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185063.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:44:05,789 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9516, 4.9114, 4.6991, 4.1975, 4.8765, 1.8588, 4.6000, 4.4861], device='cuda:7'), covar=tensor([0.0094, 0.0081, 0.0188, 0.0313, 0.0079, 0.2699, 0.0127, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0152, 0.0195, 0.0179, 0.0173, 0.0207, 0.0187, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 20:44:06,550 INFO [train.py:904] (7/8) Epoch 19, batch 2400, loss[loss=0.1991, simple_loss=0.298, pruned_loss=0.0501, over 16464.00 frames. ], tot_loss[loss=0.1767, simple_loss=0.263, pruned_loss=0.04525, over 3304353.53 frames. ], batch size: 68, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:44:26,490 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-04-30 20:44:33,411 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185121.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 20:44:43,083 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185129.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:45:02,229 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185142.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:45:15,145 INFO [train.py:904] (7/8) Epoch 19, batch 2450, loss[loss=0.15, simple_loss=0.241, pruned_loss=0.02949, over 17198.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2634, pruned_loss=0.04494, over 3309064.56 frames. ], batch size: 46, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:45:27,051 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.259e+02 2.760e+02 3.178e+02 5.977e+02, threshold=5.520e+02, percent-clipped=3.0 2023-04-30 20:46:03,474 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6831, 3.8177, 2.8586, 2.2262, 2.4951, 2.3280, 3.9580, 3.3659], device='cuda:7'), covar=tensor([0.2654, 0.0545, 0.1792, 0.2876, 0.2652, 0.2043, 0.0498, 0.1298], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0266, 0.0300, 0.0303, 0.0294, 0.0251, 0.0288, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:46:08,271 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185190.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:46:08,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7917, 3.6489, 4.1855, 1.9854, 4.4344, 4.4981, 3.2322, 3.3568], device='cuda:7'), covar=tensor([0.0790, 0.0283, 0.0256, 0.1279, 0.0074, 0.0180, 0.0435, 0.0416], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0109, 0.0097, 0.0142, 0.0079, 0.0126, 0.0129, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 20:46:08,502 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185190.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:46:18,532 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-04-30 20:46:23,976 INFO [train.py:904] (7/8) Epoch 19, batch 2500, loss[loss=0.1819, simple_loss=0.2748, pruned_loss=0.04447, over 17070.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2627, pruned_loss=0.04425, over 3310675.36 frames. ], batch size: 53, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:46:42,915 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 20:46:51,988 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185222.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:47:32,254 INFO [train.py:904] (7/8) Epoch 19, batch 2550, loss[loss=0.1448, simple_loss=0.2365, pruned_loss=0.02655, over 16845.00 frames. ], tot_loss[loss=0.1751, simple_loss=0.2621, pruned_loss=0.04408, over 3313386.80 frames. ], batch size: 42, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:47:44,041 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.145e+02 2.534e+02 2.988e+02 8.366e+02, threshold=5.067e+02, percent-clipped=1.0 2023-04-30 20:48:31,970 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185295.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:48:40,161 INFO [train.py:904] (7/8) Epoch 19, batch 2600, loss[loss=0.1754, simple_loss=0.2667, pruned_loss=0.04199, over 16671.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2616, pruned_loss=0.04324, over 3316294.02 frames. ], batch size: 62, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:49:11,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6452, 3.7496, 2.8151, 2.2218, 2.4639, 2.3923, 3.8728, 3.3712], device='cuda:7'), covar=tensor([0.2684, 0.0600, 0.1775, 0.2999, 0.2682, 0.1906, 0.0530, 0.1284], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0265, 0.0300, 0.0303, 0.0294, 0.0251, 0.0288, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:49:22,806 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 20:49:50,078 INFO [train.py:904] (7/8) Epoch 19, batch 2650, loss[loss=0.1735, simple_loss=0.2702, pruned_loss=0.03836, over 16738.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2622, pruned_loss=0.04351, over 3316295.15 frames. ], batch size: 62, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:50:03,487 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.113e+02 2.462e+02 3.036e+02 8.000e+02, threshold=4.924e+02, percent-clipped=5.0 2023-04-30 20:50:45,895 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 20:50:59,036 INFO [train.py:904] (7/8) Epoch 19, batch 2700, loss[loss=0.1703, simple_loss=0.2486, pruned_loss=0.04602, over 16783.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2626, pruned_loss=0.04342, over 3322530.16 frames. ], batch size: 102, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:51:19,150 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185416.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 20:51:34,344 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-04-30 20:51:49,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1546, 5.6196, 5.8438, 5.4418, 5.5858, 6.1622, 5.6542, 5.4365], device='cuda:7'), covar=tensor([0.0894, 0.1876, 0.2248, 0.1888, 0.2499, 0.0866, 0.1456, 0.2211], device='cuda:7'), in_proj_covar=tensor([0.0409, 0.0590, 0.0653, 0.0493, 0.0661, 0.0691, 0.0510, 0.0660], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:52:08,645 INFO [train.py:904] (7/8) Epoch 19, batch 2750, loss[loss=0.1697, simple_loss=0.2641, pruned_loss=0.03761, over 16711.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2623, pruned_loss=0.04274, over 3328080.01 frames. ], batch size: 57, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:52:18,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2436, 5.7799, 5.9312, 5.5580, 5.6951, 6.2569, 5.7406, 5.4091], device='cuda:7'), covar=tensor([0.0875, 0.1848, 0.2268, 0.1984, 0.2532, 0.0983, 0.1529, 0.2427], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0588, 0.0651, 0.0492, 0.0659, 0.0689, 0.0508, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:52:20,531 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.075e+02 2.462e+02 3.061e+02 7.702e+02, threshold=4.923e+02, percent-clipped=3.0 2023-04-30 20:52:30,910 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185468.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:52:54,431 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185485.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:53:18,361 INFO [train.py:904] (7/8) Epoch 19, batch 2800, loss[loss=0.1832, simple_loss=0.2615, pruned_loss=0.05246, over 16281.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2619, pruned_loss=0.0427, over 3320426.19 frames. ], batch size: 165, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:53:46,700 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185522.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:53:56,314 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185529.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:54:28,306 INFO [train.py:904] (7/8) Epoch 19, batch 2850, loss[loss=0.1381, simple_loss=0.2279, pruned_loss=0.02413, over 16765.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2614, pruned_loss=0.04258, over 3326615.86 frames. ], batch size: 39, lr: 3.59e-03, grad_scale: 8.0 2023-04-30 20:54:41,484 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.166e+02 2.620e+02 3.152e+02 5.145e+02, threshold=5.240e+02, percent-clipped=1.0 2023-04-30 20:54:53,471 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185570.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:55:01,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9747, 4.1173, 2.3380, 4.7600, 3.0868, 4.6473, 2.3951, 3.2405], device='cuda:7'), covar=tensor([0.0277, 0.0388, 0.1798, 0.0213, 0.0863, 0.0390, 0.1743, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0179, 0.0195, 0.0164, 0.0178, 0.0220, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 20:55:28,951 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3902, 5.3542, 5.2193, 4.6941, 4.8529, 5.2948, 5.2939, 4.8790], device='cuda:7'), covar=tensor([0.0560, 0.0473, 0.0286, 0.0366, 0.1126, 0.0458, 0.0251, 0.0753], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0426, 0.0351, 0.0342, 0.0360, 0.0396, 0.0239, 0.0417], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 20:55:28,962 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185595.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:55:39,065 INFO [train.py:904] (7/8) Epoch 19, batch 2900, loss[loss=0.1796, simple_loss=0.2551, pruned_loss=0.05208, over 16759.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2607, pruned_loss=0.04327, over 3316406.28 frames. ], batch size: 124, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:55:48,530 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 20:56:36,425 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185643.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:56:47,754 INFO [train.py:904] (7/8) Epoch 19, batch 2950, loss[loss=0.2451, simple_loss=0.3182, pruned_loss=0.08605, over 11680.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2601, pruned_loss=0.04335, over 3310338.29 frames. ], batch size: 246, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:57:00,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.308e+02 2.753e+02 3.431e+02 9.141e+02, threshold=5.506e+02, percent-clipped=4.0 2023-04-30 20:57:58,864 INFO [train.py:904] (7/8) Epoch 19, batch 3000, loss[loss=0.1985, simple_loss=0.2763, pruned_loss=0.06031, over 16267.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2599, pruned_loss=0.04369, over 3311249.59 frames. ], batch size: 165, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:57:58,864 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 20:58:07,638 INFO [train.py:938] (7/8) Epoch 19, validation: loss=0.1362, simple_loss=0.2416, pruned_loss=0.01538, over 944034.00 frames. 2023-04-30 20:58:07,639 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 20:58:28,288 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185716.0, num_to_drop=1, layers_to_drop={2} 2023-04-30 20:59:16,279 INFO [train.py:904] (7/8) Epoch 19, batch 3050, loss[loss=0.1598, simple_loss=0.2382, pruned_loss=0.04071, over 15971.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2593, pruned_loss=0.0435, over 3315830.65 frames. ], batch size: 35, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 20:59:28,175 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185759.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 20:59:29,998 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 2.137e+02 2.455e+02 2.756e+02 4.328e+02, threshold=4.910e+02, percent-clipped=0.0 2023-04-30 20:59:35,051 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185764.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:00:04,074 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=185785.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:00:23,284 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=185799.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:00:26,872 INFO [train.py:904] (7/8) Epoch 19, batch 3100, loss[loss=0.1683, simple_loss=0.2583, pruned_loss=0.03915, over 17129.00 frames. ], tot_loss[loss=0.174, simple_loss=0.26, pruned_loss=0.04403, over 3322292.91 frames. ], batch size: 46, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:00:42,791 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1238, 4.9122, 5.1637, 5.3418, 5.5392, 4.7770, 5.4955, 5.5318], device='cuda:7'), covar=tensor([0.1905, 0.1292, 0.1712, 0.0808, 0.0508, 0.0913, 0.0521, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0647, 0.0800, 0.0940, 0.0817, 0.0609, 0.0640, 0.0657, 0.0765], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:00:51,375 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185820.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:00:57,048 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=185824.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:01:07,017 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.91 vs. limit=5.0 2023-04-30 21:01:09,059 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=185833.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:01:34,496 INFO [train.py:904] (7/8) Epoch 19, batch 3150, loss[loss=0.1717, simple_loss=0.2663, pruned_loss=0.03854, over 17058.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2594, pruned_loss=0.04362, over 3319462.57 frames. ], batch size: 50, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:01:36,843 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7100, 3.6075, 2.7442, 2.2467, 2.3529, 2.3247, 3.7066, 3.2042], device='cuda:7'), covar=tensor([0.2312, 0.0616, 0.1702, 0.2845, 0.2634, 0.2022, 0.0472, 0.1465], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0266, 0.0301, 0.0305, 0.0295, 0.0252, 0.0290, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:01:46,523 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=185860.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:01:47,103 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.254e+02 2.659e+02 3.271e+02 6.332e+02, threshold=5.317e+02, percent-clipped=3.0 2023-04-30 21:02:43,074 INFO [train.py:904] (7/8) Epoch 19, batch 3200, loss[loss=0.1708, simple_loss=0.2634, pruned_loss=0.03908, over 16720.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2587, pruned_loss=0.0435, over 3319326.48 frames. ], batch size: 57, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:03:23,263 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 21:03:53,545 INFO [train.py:904] (7/8) Epoch 19, batch 3250, loss[loss=0.1819, simple_loss=0.2701, pruned_loss=0.04689, over 16417.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2582, pruned_loss=0.04354, over 3324382.94 frames. ], batch size: 68, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:04:06,310 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.264e+02 2.586e+02 3.072e+02 8.632e+02, threshold=5.173e+02, percent-clipped=3.0 2023-04-30 21:05:07,479 INFO [train.py:904] (7/8) Epoch 19, batch 3300, loss[loss=0.1636, simple_loss=0.2543, pruned_loss=0.03646, over 17114.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2582, pruned_loss=0.04336, over 3319385.53 frames. ], batch size: 48, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:05:37,550 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-04-30 21:06:15,810 INFO [train.py:904] (7/8) Epoch 19, batch 3350, loss[loss=0.208, simple_loss=0.2789, pruned_loss=0.0685, over 16947.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2584, pruned_loss=0.04294, over 3325077.56 frames. ], batch size: 109, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:06:28,001 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.170e+02 2.521e+02 2.996e+02 5.638e+02, threshold=5.041e+02, percent-clipped=4.0 2023-04-30 21:07:23,743 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186100.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:07:25,699 INFO [train.py:904] (7/8) Epoch 19, batch 3400, loss[loss=0.1617, simple_loss=0.2497, pruned_loss=0.03687, over 16400.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2586, pruned_loss=0.04364, over 3313530.86 frames. ], batch size: 75, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:07:44,671 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186115.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:07:49,334 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-04-30 21:07:57,590 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186124.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:08:34,736 INFO [train.py:904] (7/8) Epoch 19, batch 3450, loss[loss=0.1681, simple_loss=0.2477, pruned_loss=0.04426, over 16509.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2569, pruned_loss=0.04301, over 3317403.26 frames. ], batch size: 68, lr: 3.58e-03, grad_scale: 8.0 2023-04-30 21:08:39,255 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186155.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 21:08:44,131 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186158.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:08:47,842 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.301e+02 2.793e+02 3.380e+02 7.495e+02, threshold=5.586e+02, percent-clipped=5.0 2023-04-30 21:08:48,278 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186161.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:09:03,001 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186172.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:09:32,087 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9835, 5.0137, 5.4542, 5.4827, 5.4725, 5.1371, 5.0840, 4.8884], device='cuda:7'), covar=tensor([0.0359, 0.0593, 0.0418, 0.0417, 0.0453, 0.0374, 0.0922, 0.0439], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0450, 0.0435, 0.0408, 0.0483, 0.0459, 0.0554, 0.0365], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 21:09:45,134 INFO [train.py:904] (7/8) Epoch 19, batch 3500, loss[loss=0.1671, simple_loss=0.2466, pruned_loss=0.04377, over 16488.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2563, pruned_loss=0.04287, over 3314647.86 frames. ], batch size: 146, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:09:58,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5875, 4.5633, 4.4731, 4.2099, 4.2349, 4.5525, 4.3218, 4.3025], device='cuda:7'), covar=tensor([0.0661, 0.0753, 0.0323, 0.0274, 0.0717, 0.0521, 0.0528, 0.0597], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0430, 0.0353, 0.0344, 0.0363, 0.0400, 0.0242, 0.0419], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:10:08,963 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8987, 4.0689, 2.3527, 4.6068, 3.1026, 4.4793, 2.2363, 3.1106], device='cuda:7'), covar=tensor([0.0283, 0.0346, 0.1722, 0.0250, 0.0801, 0.0447, 0.1839, 0.0744], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0178, 0.0195, 0.0164, 0.0176, 0.0220, 0.0202, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:10:08,981 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186219.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:10:22,858 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8639, 4.9161, 5.3315, 5.3460, 5.3206, 5.0042, 4.9646, 4.7645], device='cuda:7'), covar=tensor([0.0343, 0.0530, 0.0371, 0.0372, 0.0488, 0.0377, 0.0881, 0.0454], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0449, 0.0435, 0.0407, 0.0482, 0.0459, 0.0554, 0.0365], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 21:10:24,346 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.46 vs. limit=5.0 2023-04-30 21:10:55,187 INFO [train.py:904] (7/8) Epoch 19, batch 3550, loss[loss=0.1723, simple_loss=0.2524, pruned_loss=0.04615, over 16384.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2553, pruned_loss=0.04255, over 3313674.56 frames. ], batch size: 75, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:11:05,311 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186259.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:11:06,986 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.073e+02 2.370e+02 2.856e+02 5.323e+02, threshold=4.740e+02, percent-clipped=0.0 2023-04-30 21:11:29,917 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5731, 3.5795, 2.2383, 3.8380, 2.8211, 3.7922, 2.1941, 2.8183], device='cuda:7'), covar=tensor([0.0271, 0.0456, 0.1460, 0.0377, 0.0779, 0.0794, 0.1451, 0.0722], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0178, 0.0194, 0.0164, 0.0176, 0.0219, 0.0202, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:11:33,599 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4179, 3.4064, 3.6820, 2.5535, 3.4645, 3.7590, 3.5266, 2.2471], device='cuda:7'), covar=tensor([0.0440, 0.0166, 0.0054, 0.0365, 0.0095, 0.0096, 0.0082, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0081, 0.0081, 0.0133, 0.0095, 0.0106, 0.0092, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:12:03,408 INFO [train.py:904] (7/8) Epoch 19, batch 3600, loss[loss=0.1402, simple_loss=0.2216, pruned_loss=0.02938, over 16527.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2538, pruned_loss=0.04215, over 3323165.20 frames. ], batch size: 75, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:12:28,425 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186320.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:13:13,929 INFO [train.py:904] (7/8) Epoch 19, batch 3650, loss[loss=0.1814, simple_loss=0.2587, pruned_loss=0.05202, over 11476.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2532, pruned_loss=0.04271, over 3306186.59 frames. ], batch size: 247, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:13:27,664 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.244e+02 2.696e+02 3.128e+02 5.239e+02, threshold=5.393e+02, percent-clipped=3.0 2023-04-30 21:14:28,225 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-04-30 21:14:28,525 INFO [train.py:904] (7/8) Epoch 19, batch 3700, loss[loss=0.1724, simple_loss=0.2529, pruned_loss=0.04595, over 11251.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2528, pruned_loss=0.04396, over 3277810.12 frames. ], batch size: 246, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:14:48,798 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186415.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:14:57,131 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186421.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:15:41,479 INFO [train.py:904] (7/8) Epoch 19, batch 3750, loss[loss=0.1806, simple_loss=0.256, pruned_loss=0.05264, over 16327.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2537, pruned_loss=0.04506, over 3286166.67 frames. ], batch size: 165, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:15:46,609 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186455.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:15:47,865 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186456.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:15:55,026 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.254e+02 2.571e+02 3.173e+02 4.680e+02, threshold=5.142e+02, percent-clipped=0.0 2023-04-30 21:15:58,456 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186463.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:16:15,903 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186475.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:16:26,597 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186482.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:16:53,859 INFO [train.py:904] (7/8) Epoch 19, batch 3800, loss[loss=0.1732, simple_loss=0.264, pruned_loss=0.04123, over 17226.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2549, pruned_loss=0.04633, over 3281028.02 frames. ], batch size: 45, lr: 3.58e-03, grad_scale: 16.0 2023-04-30 21:16:54,974 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5088, 4.5039, 4.4818, 3.9816, 4.5083, 1.8110, 4.2836, 4.1334], device='cuda:7'), covar=tensor([0.0123, 0.0103, 0.0169, 0.0321, 0.0100, 0.2674, 0.0147, 0.0225], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0150, 0.0196, 0.0179, 0.0173, 0.0205, 0.0187, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:16:55,995 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186503.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:17:08,799 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8900, 4.8905, 4.8059, 4.2531, 4.8864, 1.9755, 4.6383, 4.4757], device='cuda:7'), covar=tensor([0.0105, 0.0096, 0.0177, 0.0340, 0.0088, 0.2619, 0.0136, 0.0225], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0150, 0.0196, 0.0179, 0.0173, 0.0205, 0.0187, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:17:11,657 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186514.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:15,670 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186517.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:43,806 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186536.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:17:47,977 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186539.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:18:05,825 INFO [train.py:904] (7/8) Epoch 19, batch 3850, loss[loss=0.1826, simple_loss=0.2536, pruned_loss=0.05577, over 16637.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2544, pruned_loss=0.04689, over 3284592.41 frames. ], batch size: 89, lr: 3.58e-03, grad_scale: 4.0 2023-04-30 21:18:22,162 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.100e+02 2.475e+02 3.097e+02 5.903e+02, threshold=4.949e+02, percent-clipped=1.0 2023-04-30 21:18:29,061 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186567.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:18:45,018 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186578.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:18:56,210 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186586.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:19:07,966 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.70 vs. limit=5.0 2023-04-30 21:19:16,911 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186600.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:19:18,740 INFO [train.py:904] (7/8) Epoch 19, batch 3900, loss[loss=0.178, simple_loss=0.2554, pruned_loss=0.0503, over 16519.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2541, pruned_loss=0.04732, over 3286953.35 frames. ], batch size: 68, lr: 3.58e-03, grad_scale: 4.0 2023-04-30 21:19:21,391 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-04-30 21:19:37,489 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:19:55,127 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186628.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:20:21,616 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186647.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:20:28,334 INFO [train.py:904] (7/8) Epoch 19, batch 3950, loss[loss=0.1745, simple_loss=0.2609, pruned_loss=0.04404, over 16244.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2545, pruned_loss=0.04773, over 3288452.22 frames. ], batch size: 35, lr: 3.57e-03, grad_scale: 4.0 2023-04-30 21:20:42,981 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.601e+02 2.399e+02 2.822e+02 3.355e+02 8.994e+02, threshold=5.644e+02, percent-clipped=5.0 2023-04-30 21:20:46,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7242, 1.9627, 2.3612, 2.5757, 2.6696, 2.6431, 1.8735, 2.8467], device='cuda:7'), covar=tensor([0.0167, 0.0431, 0.0289, 0.0294, 0.0288, 0.0323, 0.0512, 0.0154], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0191, 0.0179, 0.0183, 0.0192, 0.0151, 0.0195, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:21:23,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9532, 4.9575, 4.8209, 4.4844, 4.4926, 4.8923, 4.6843, 4.6028], device='cuda:7'), covar=tensor([0.0608, 0.0542, 0.0297, 0.0284, 0.0939, 0.0459, 0.0469, 0.0588], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0424, 0.0350, 0.0340, 0.0360, 0.0396, 0.0239, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:21:26,962 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-04-30 21:21:39,186 INFO [train.py:904] (7/8) Epoch 19, batch 4000, loss[loss=0.1626, simple_loss=0.2452, pruned_loss=0.04, over 17119.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2549, pruned_loss=0.04855, over 3284753.14 frames. ], batch size: 49, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:21:49,084 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186709.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:22:49,926 INFO [train.py:904] (7/8) Epoch 19, batch 4050, loss[loss=0.1755, simple_loss=0.2537, pruned_loss=0.04867, over 16186.00 frames. ], tot_loss[loss=0.1753, simple_loss=0.255, pruned_loss=0.04776, over 3275597.65 frames. ], batch size: 35, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:22:57,409 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186756.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:23:03,286 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 21:23:06,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 1.873e+02 2.216e+02 2.589e+02 4.495e+02, threshold=4.432e+02, percent-clipped=0.0 2023-04-30 21:23:16,289 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186770.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:23:27,100 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186777.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:03,626 INFO [train.py:904] (7/8) Epoch 19, batch 4100, loss[loss=0.1904, simple_loss=0.2765, pruned_loss=0.05215, over 16757.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2567, pruned_loss=0.04728, over 3279881.20 frames. ], batch size: 83, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:24:07,125 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186804.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:22,596 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186814.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:46,683 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186831.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:24:59,944 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186840.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:25:17,900 INFO [train.py:904] (7/8) Epoch 19, batch 4150, loss[loss=0.1912, simple_loss=0.2842, pruned_loss=0.04906, over 16707.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2634, pruned_loss=0.04949, over 3251883.83 frames. ], batch size: 134, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:25:34,162 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186862.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:25:34,926 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.347e+02 2.016e+02 2.433e+02 2.890e+02 5.187e+02, threshold=4.866e+02, percent-clipped=4.0 2023-04-30 21:25:49,777 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186873.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:26:23,103 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186895.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:26:31,972 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186901.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:26:32,686 INFO [train.py:904] (7/8) Epoch 19, batch 4200, loss[loss=0.2142, simple_loss=0.3074, pruned_loss=0.06047, over 16769.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2707, pruned_loss=0.05138, over 3216686.44 frames. ], batch size: 124, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:26:47,215 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=186911.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:26:53,457 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=186915.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:27:06,452 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186923.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:27:34,747 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=186942.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:27:49,661 INFO [train.py:904] (7/8) Epoch 19, batch 4250, loss[loss=0.192, simple_loss=0.2909, pruned_loss=0.04649, over 16633.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2738, pruned_loss=0.05128, over 3190811.11 frames. ], batch size: 62, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:27:50,241 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8275, 3.9691, 3.0695, 2.3384, 2.8353, 2.7756, 4.3438, 3.6247], device='cuda:7'), covar=tensor([0.2699, 0.0729, 0.1791, 0.2758, 0.2593, 0.1788, 0.0514, 0.1135], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0265, 0.0299, 0.0303, 0.0296, 0.0251, 0.0288, 0.0329], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:28:05,626 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.577e+02 2.275e+02 2.568e+02 3.131e+02 7.157e+02, threshold=5.135e+02, percent-clipped=7.0 2023-04-30 21:28:06,529 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=186963.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:28:06,610 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1316, 5.4091, 5.1881, 5.2028, 4.9487, 4.8117, 4.7980, 5.5002], device='cuda:7'), covar=tensor([0.0976, 0.0807, 0.0948, 0.0738, 0.0728, 0.0928, 0.1022, 0.0804], device='cuda:7'), in_proj_covar=tensor([0.0654, 0.0805, 0.0663, 0.0602, 0.0506, 0.0515, 0.0673, 0.0619], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:28:09,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1704, 5.3755, 5.0752, 4.7386, 4.4441, 5.2163, 5.1503, 4.7583], device='cuda:7'), covar=tensor([0.0609, 0.0317, 0.0355, 0.0346, 0.1272, 0.0386, 0.0274, 0.0721], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0413, 0.0341, 0.0331, 0.0351, 0.0383, 0.0232, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:28:19,169 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=186972.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:29:02,889 INFO [train.py:904] (7/8) Epoch 19, batch 4300, loss[loss=0.1918, simple_loss=0.2788, pruned_loss=0.05239, over 11796.00 frames. ], tot_loss[loss=0.188, simple_loss=0.2751, pruned_loss=0.05043, over 3176508.96 frames. ], batch size: 246, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:29:37,834 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187025.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:29:56,609 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187038.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:30:16,630 INFO [train.py:904] (7/8) Epoch 19, batch 4350, loss[loss=0.1902, simple_loss=0.2707, pruned_loss=0.05483, over 11708.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2781, pruned_loss=0.05115, over 3172976.26 frames. ], batch size: 248, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:30:32,934 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.200e+02 2.562e+02 2.922e+02 5.186e+02, threshold=5.123e+02, percent-clipped=1.0 2023-04-30 21:30:35,927 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187065.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:30:47,400 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2023-04-30 21:30:54,525 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187077.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:31:08,011 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187086.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:31:26,999 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187099.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:31:31,229 INFO [train.py:904] (7/8) Epoch 19, batch 4400, loss[loss=0.1912, simple_loss=0.285, pruned_loss=0.04869, over 16871.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2806, pruned_loss=0.0525, over 3160956.17 frames. ], batch size: 102, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:32:05,654 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:32:13,530 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187131.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:32:41,960 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-04-30 21:32:43,363 INFO [train.py:904] (7/8) Epoch 19, batch 4450, loss[loss=0.196, simple_loss=0.292, pruned_loss=0.04995, over 16690.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2841, pruned_loss=0.05379, over 3173709.83 frames. ], batch size: 76, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:32:54,771 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5765, 4.6745, 4.4907, 4.2084, 4.1538, 4.5851, 4.2894, 4.2618], device='cuda:7'), covar=tensor([0.0508, 0.0261, 0.0221, 0.0224, 0.0743, 0.0295, 0.0487, 0.0517], device='cuda:7'), in_proj_covar=tensor([0.0285, 0.0405, 0.0336, 0.0325, 0.0346, 0.0377, 0.0230, 0.0398], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:33:00,622 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.070e+02 2.320e+02 2.761e+02 4.804e+02, threshold=4.641e+02, percent-clipped=0.0 2023-04-30 21:33:15,103 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187173.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:33:24,885 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187179.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:33:48,157 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187195.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:33:50,011 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187196.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:33:53,739 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8102, 3.9328, 4.1240, 4.0972, 4.1270, 3.8658, 3.9045, 3.8841], device='cuda:7'), covar=tensor([0.0320, 0.0450, 0.0373, 0.0384, 0.0425, 0.0431, 0.0825, 0.0461], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0429, 0.0418, 0.0390, 0.0459, 0.0440, 0.0531, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 21:33:57,528 INFO [train.py:904] (7/8) Epoch 19, batch 4500, loss[loss=0.2056, simple_loss=0.2933, pruned_loss=0.05893, over 16915.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2847, pruned_loss=0.05411, over 3197722.61 frames. ], batch size: 116, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:34:20,948 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-04-30 21:34:24,643 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:34:27,290 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187223.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:34:38,729 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9337, 2.3213, 1.9188, 2.1152, 2.6767, 2.3415, 2.6869, 2.8525], device='cuda:7'), covar=tensor([0.0146, 0.0378, 0.0493, 0.0422, 0.0217, 0.0350, 0.0178, 0.0237], device='cuda:7'), in_proj_covar=tensor([0.0198, 0.0227, 0.0218, 0.0220, 0.0229, 0.0228, 0.0232, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:34:54,359 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187242.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:34:56,192 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187243.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:35:08,497 INFO [train.py:904] (7/8) Epoch 19, batch 4550, loss[loss=0.2246, simple_loss=0.2906, pruned_loss=0.0793, over 12002.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2856, pruned_loss=0.05549, over 3195078.14 frames. ], batch size: 246, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:35:24,438 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 1.917e+02 2.180e+02 2.712e+02 4.446e+02, threshold=4.359e+02, percent-clipped=0.0 2023-04-30 21:35:30,869 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187267.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:35:36,696 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187271.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:36:05,136 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187290.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:36:20,321 INFO [train.py:904] (7/8) Epoch 19, batch 4600, loss[loss=0.1864, simple_loss=0.2818, pruned_loss=0.04553, over 16679.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2865, pruned_loss=0.05554, over 3207932.10 frames. ], batch size: 134, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:37:33,052 INFO [train.py:904] (7/8) Epoch 19, batch 4650, loss[loss=0.176, simple_loss=0.2627, pruned_loss=0.04462, over 16921.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2851, pruned_loss=0.05545, over 3214865.74 frames. ], batch size: 96, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:37:49,312 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 1.930e+02 2.188e+02 2.546e+02 5.637e+02, threshold=4.377e+02, percent-clipped=1.0 2023-04-30 21:37:52,236 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187365.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:38:01,563 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187372.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:38:05,316 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6465, 2.5572, 1.8292, 2.7557, 2.1371, 2.7445, 2.1166, 2.3246], device='cuda:7'), covar=tensor([0.0316, 0.0361, 0.1329, 0.0212, 0.0744, 0.0374, 0.1164, 0.0630], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0175, 0.0193, 0.0156, 0.0175, 0.0215, 0.0198, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:38:15,562 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187381.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 21:38:33,241 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187394.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:38:44,856 INFO [train.py:904] (7/8) Epoch 19, batch 4700, loss[loss=0.1947, simple_loss=0.2902, pruned_loss=0.04962, over 15373.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.282, pruned_loss=0.05411, over 3215951.02 frames. ], batch size: 190, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:39:01,840 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187413.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:39:30,050 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187433.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:39:55,450 INFO [train.py:904] (7/8) Epoch 19, batch 4750, loss[loss=0.1582, simple_loss=0.2476, pruned_loss=0.03439, over 16937.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2782, pruned_loss=0.05214, over 3211962.51 frames. ], batch size: 109, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:40:10,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3724, 2.8478, 3.0519, 1.9461, 2.6608, 2.0667, 3.0362, 3.0826], device='cuda:7'), covar=tensor([0.0278, 0.0820, 0.0609, 0.1960, 0.0892, 0.0995, 0.0631, 0.0889], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0160, 0.0164, 0.0150, 0.0142, 0.0127, 0.0141, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:40:11,093 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 1.903e+02 2.161e+02 2.490e+02 3.592e+02, threshold=4.322e+02, percent-clipped=0.0 2023-04-30 21:40:57,633 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187496.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:40:59,614 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7058, 3.8450, 2.2596, 4.3830, 2.9373, 4.2867, 2.3796, 3.0786], device='cuda:7'), covar=tensor([0.0264, 0.0313, 0.1600, 0.0130, 0.0796, 0.0437, 0.1542, 0.0697], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0173, 0.0191, 0.0155, 0.0173, 0.0213, 0.0196, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:41:05,673 INFO [train.py:904] (7/8) Epoch 19, batch 4800, loss[loss=0.1969, simple_loss=0.2833, pruned_loss=0.0552, over 16839.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2747, pruned_loss=0.05015, over 3203974.75 frames. ], batch size: 116, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:41:21,430 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187512.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:41:24,844 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9569, 2.0870, 2.1954, 3.6372, 2.0417, 2.3984, 2.2099, 2.2865], device='cuda:7'), covar=tensor([0.1488, 0.3737, 0.2833, 0.0549, 0.4200, 0.2652, 0.3625, 0.3112], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0435, 0.0358, 0.0321, 0.0431, 0.0503, 0.0405, 0.0508], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:41:30,415 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187519.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:41:34,758 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0941, 3.9844, 4.1422, 4.2621, 4.3751, 3.9475, 4.3368, 4.4234], device='cuda:7'), covar=tensor([0.1474, 0.1108, 0.1265, 0.0600, 0.0472, 0.1301, 0.0618, 0.0523], device='cuda:7'), in_proj_covar=tensor([0.0605, 0.0746, 0.0879, 0.0764, 0.0571, 0.0596, 0.0613, 0.0714], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:41:37,680 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=187523.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:42:08,454 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187544.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:42:19,542 INFO [train.py:904] (7/8) Epoch 19, batch 4850, loss[loss=0.2179, simple_loss=0.2918, pruned_loss=0.07197, over 11992.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2751, pruned_loss=0.04894, over 3197613.50 frames. ], batch size: 248, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:42:36,209 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 1.857e+02 2.110e+02 2.587e+02 7.641e+02, threshold=4.221e+02, percent-clipped=2.0 2023-04-30 21:42:41,974 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187567.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:42:51,139 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187573.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:43:02,034 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187580.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:43:07,578 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=187584.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:43:27,760 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6720, 2.6672, 1.8188, 2.7974, 2.1584, 2.8139, 2.0998, 2.3518], device='cuda:7'), covar=tensor([0.0284, 0.0319, 0.1282, 0.0170, 0.0653, 0.0452, 0.1205, 0.0615], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0172, 0.0189, 0.0153, 0.0171, 0.0211, 0.0195, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:43:32,181 INFO [train.py:904] (7/8) Epoch 19, batch 4900, loss[loss=0.1607, simple_loss=0.2506, pruned_loss=0.03539, over 16454.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2744, pruned_loss=0.04753, over 3196532.15 frames. ], batch size: 68, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:43:39,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8169, 4.9545, 5.2980, 5.2576, 5.2686, 4.9124, 4.8364, 4.6579], device='cuda:7'), covar=tensor([0.0283, 0.0411, 0.0276, 0.0345, 0.0464, 0.0324, 0.0948, 0.0411], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0427, 0.0415, 0.0387, 0.0458, 0.0437, 0.0525, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 21:43:51,091 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187615.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:44:07,444 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.01 vs. limit=5.0 2023-04-30 21:44:36,277 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 21:44:43,114 INFO [train.py:904] (7/8) Epoch 19, batch 4950, loss[loss=0.1964, simple_loss=0.2894, pruned_loss=0.05167, over 16700.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2736, pruned_loss=0.04684, over 3204928.90 frames. ], batch size: 124, lr: 3.57e-03, grad_scale: 8.0 2023-04-30 21:44:56,269 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4036, 5.7391, 5.4218, 5.5497, 5.1656, 5.0758, 5.1503, 5.8350], device='cuda:7'), covar=tensor([0.1166, 0.0776, 0.1011, 0.0687, 0.0750, 0.0731, 0.1074, 0.0778], device='cuda:7'), in_proj_covar=tensor([0.0641, 0.0791, 0.0651, 0.0586, 0.0498, 0.0504, 0.0656, 0.0610], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:44:57,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1973, 4.1117, 4.0356, 2.6992, 3.5719, 4.0402, 3.5566, 2.0854], device='cuda:7'), covar=tensor([0.0527, 0.0036, 0.0038, 0.0360, 0.0096, 0.0080, 0.0095, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0078, 0.0079, 0.0130, 0.0093, 0.0104, 0.0090, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:44:58,357 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.032e+02 2.342e+02 2.691e+02 5.269e+02, threshold=4.685e+02, percent-clipped=1.0 2023-04-30 21:45:24,756 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187681.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 21:45:37,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2078, 4.1617, 4.1407, 2.7244, 3.6565, 4.1518, 3.6658, 2.2399], device='cuda:7'), covar=tensor([0.0545, 0.0034, 0.0032, 0.0366, 0.0082, 0.0074, 0.0074, 0.0433], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0078, 0.0079, 0.0130, 0.0093, 0.0103, 0.0090, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:45:44,023 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=187694.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:45:55,108 INFO [train.py:904] (7/8) Epoch 19, batch 5000, loss[loss=0.195, simple_loss=0.2827, pruned_loss=0.05362, over 12096.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2758, pruned_loss=0.04712, over 3205793.83 frames. ], batch size: 247, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:46:32,256 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187728.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:46:33,447 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187729.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:46:52,690 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=187742.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:47:07,086 INFO [train.py:904] (7/8) Epoch 19, batch 5050, loss[loss=0.1829, simple_loss=0.2701, pruned_loss=0.04784, over 17036.00 frames. ], tot_loss[loss=0.186, simple_loss=0.2769, pruned_loss=0.04753, over 3201378.12 frames. ], batch size: 53, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:47:21,707 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.150e+02 2.460e+02 2.787e+02 5.516e+02, threshold=4.919e+02, percent-clipped=2.0 2023-04-30 21:47:35,817 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-30 21:48:17,851 INFO [train.py:904] (7/8) Epoch 19, batch 5100, loss[loss=0.1801, simple_loss=0.2749, pruned_loss=0.04268, over 16927.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2758, pruned_loss=0.04732, over 3200601.07 frames. ], batch size: 109, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:49:30,697 INFO [train.py:904] (7/8) Epoch 19, batch 5150, loss[loss=0.1785, simple_loss=0.2673, pruned_loss=0.04481, over 16686.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2759, pruned_loss=0.04715, over 3191768.59 frames. ], batch size: 62, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:49:47,463 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.013e+02 2.291e+02 2.768e+02 7.535e+02, threshold=4.582e+02, percent-clipped=1.0 2023-04-30 21:49:54,432 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187868.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:50:04,859 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187875.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:50:11,363 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=187879.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:50:13,917 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5850, 4.4449, 4.6556, 4.8344, 4.9925, 4.5218, 4.9410, 4.9909], device='cuda:7'), covar=tensor([0.1760, 0.1286, 0.1563, 0.0694, 0.0501, 0.0888, 0.0574, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0610, 0.0750, 0.0883, 0.0771, 0.0574, 0.0601, 0.0617, 0.0720], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:50:33,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8330, 3.1583, 3.2311, 1.8358, 2.7175, 2.2008, 3.2498, 3.3704], device='cuda:7'), covar=tensor([0.0280, 0.0778, 0.0721, 0.2128, 0.0939, 0.1047, 0.0694, 0.0857], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0161, 0.0165, 0.0151, 0.0143, 0.0128, 0.0143, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:50:41,951 INFO [train.py:904] (7/8) Epoch 19, batch 5200, loss[loss=0.1706, simple_loss=0.2599, pruned_loss=0.04067, over 16635.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.274, pruned_loss=0.04623, over 3186367.28 frames. ], batch size: 57, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:51:28,520 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3922, 3.5964, 3.7281, 2.2001, 3.0730, 2.5806, 3.7984, 3.8982], device='cuda:7'), covar=tensor([0.0237, 0.0795, 0.0602, 0.1875, 0.0842, 0.0872, 0.0593, 0.0795], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0160, 0.0166, 0.0151, 0.0143, 0.0128, 0.0143, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 21:51:38,960 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-04-30 21:51:53,335 INFO [train.py:904] (7/8) Epoch 19, batch 5250, loss[loss=0.1766, simple_loss=0.2681, pruned_loss=0.04256, over 16915.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2718, pruned_loss=0.04624, over 3184480.57 frames. ], batch size: 109, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:52:07,583 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4169, 5.3866, 5.2629, 4.8773, 4.8528, 5.2635, 5.2556, 5.0503], device='cuda:7'), covar=tensor([0.0588, 0.0479, 0.0282, 0.0289, 0.1166, 0.0527, 0.0255, 0.0622], device='cuda:7'), in_proj_covar=tensor([0.0282, 0.0404, 0.0333, 0.0323, 0.0343, 0.0377, 0.0228, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:52:08,312 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.924e+02 2.321e+02 2.643e+02 4.137e+02, threshold=4.643e+02, percent-clipped=0.0 2023-04-30 21:52:31,918 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6418, 3.8002, 2.8970, 2.2261, 2.4659, 2.4784, 4.0487, 3.3511], device='cuda:7'), covar=tensor([0.2865, 0.0620, 0.1806, 0.2673, 0.2503, 0.1927, 0.0449, 0.1167], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0263, 0.0298, 0.0303, 0.0291, 0.0248, 0.0287, 0.0326], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:53:08,598 INFO [train.py:904] (7/8) Epoch 19, batch 5300, loss[loss=0.1881, simple_loss=0.2704, pruned_loss=0.05288, over 12256.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2672, pruned_loss=0.04449, over 3197744.15 frames. ], batch size: 250, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:53:10,808 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-04-30 21:53:12,493 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5870, 3.6522, 3.4209, 3.1587, 3.2480, 3.5375, 3.3002, 3.3553], device='cuda:7'), covar=tensor([0.0547, 0.0566, 0.0302, 0.0258, 0.0570, 0.0437, 0.1689, 0.0459], device='cuda:7'), in_proj_covar=tensor([0.0282, 0.0405, 0.0333, 0.0323, 0.0344, 0.0378, 0.0228, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:53:46,820 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188028.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:53:52,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-04-30 21:54:01,183 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4037, 1.6633, 2.0681, 2.3854, 2.4583, 2.7563, 1.6895, 2.6816], device='cuda:7'), covar=tensor([0.0204, 0.0545, 0.0323, 0.0335, 0.0302, 0.0187, 0.0590, 0.0141], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0189, 0.0177, 0.0181, 0.0190, 0.0149, 0.0192, 0.0144], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:54:21,272 INFO [train.py:904] (7/8) Epoch 19, batch 5350, loss[loss=0.1911, simple_loss=0.278, pruned_loss=0.05214, over 16571.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2658, pruned_loss=0.04413, over 3201880.81 frames. ], batch size: 57, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:54:32,043 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9556, 2.0958, 2.2542, 3.4524, 2.0442, 2.3772, 2.2512, 2.2613], device='cuda:7'), covar=tensor([0.1373, 0.3446, 0.2722, 0.0587, 0.4135, 0.2512, 0.3367, 0.3075], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0432, 0.0357, 0.0320, 0.0428, 0.0499, 0.0402, 0.0504], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:54:35,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5392, 3.5058, 3.4825, 2.7761, 3.3686, 2.0124, 3.1726, 2.8182], device='cuda:7'), covar=tensor([0.0145, 0.0127, 0.0151, 0.0293, 0.0107, 0.2401, 0.0137, 0.0253], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0144, 0.0188, 0.0173, 0.0166, 0.0199, 0.0180, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:54:38,015 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 1.961e+02 2.159e+02 2.589e+02 4.705e+02, threshold=4.317e+02, percent-clipped=1.0 2023-04-30 21:54:57,114 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188076.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:55:35,761 INFO [train.py:904] (7/8) Epoch 19, batch 5400, loss[loss=0.1538, simple_loss=0.2571, pruned_loss=0.02526, over 16801.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.268, pruned_loss=0.04461, over 3184078.17 frames. ], batch size: 102, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:55:45,837 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 21:56:22,202 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4909, 4.6228, 4.7733, 4.5670, 4.6016, 5.1558, 4.6498, 4.3120], device='cuda:7'), covar=tensor([0.1327, 0.1649, 0.1872, 0.2059, 0.2541, 0.0961, 0.1475, 0.2506], device='cuda:7'), in_proj_covar=tensor([0.0395, 0.0565, 0.0619, 0.0474, 0.0633, 0.0657, 0.0486, 0.0637], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 21:56:54,056 INFO [train.py:904] (7/8) Epoch 19, batch 5450, loss[loss=0.2351, simple_loss=0.3241, pruned_loss=0.07304, over 15130.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2713, pruned_loss=0.04606, over 3175253.70 frames. ], batch size: 190, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:57:11,920 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 1.998e+02 2.677e+02 3.507e+02 6.066e+02, threshold=5.353e+02, percent-clipped=8.0 2023-04-30 21:57:20,385 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188168.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:57:32,223 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188175.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:57:38,481 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188179.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:58:14,503 INFO [train.py:904] (7/8) Epoch 19, batch 5500, loss[loss=0.2781, simple_loss=0.3434, pruned_loss=0.1064, over 11589.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2784, pruned_loss=0.05028, over 3157846.17 frames. ], batch size: 247, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:58:38,577 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:58:49,018 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188223.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:58:55,305 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188227.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 21:59:10,310 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.43 vs. limit=5.0 2023-04-30 21:59:35,425 INFO [train.py:904] (7/8) Epoch 19, batch 5550, loss[loss=0.2199, simple_loss=0.3042, pruned_loss=0.06776, over 15249.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2855, pruned_loss=0.05512, over 3143654.54 frames. ], batch size: 190, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 21:59:46,863 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9248, 4.7049, 4.7912, 5.1400, 5.2468, 4.7511, 5.3295, 5.2647], device='cuda:7'), covar=tensor([0.1839, 0.1369, 0.2002, 0.0758, 0.0794, 0.0975, 0.0762, 0.0851], device='cuda:7'), in_proj_covar=tensor([0.0617, 0.0757, 0.0891, 0.0777, 0.0581, 0.0607, 0.0622, 0.0727], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 21:59:53,557 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.484e+02 3.263e+02 3.931e+02 4.573e+02 1.048e+03, threshold=7.861e+02, percent-clipped=6.0 2023-04-30 22:00:25,999 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188283.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:00:55,099 INFO [train.py:904] (7/8) Epoch 19, batch 5600, loss[loss=0.3064, simple_loss=0.3584, pruned_loss=0.1272, over 11468.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2909, pruned_loss=0.05946, over 3117339.86 frames. ], batch size: 246, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:01:31,813 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0311, 4.0140, 3.9435, 3.1686, 3.9722, 1.7980, 3.7788, 3.5599], device='cuda:7'), covar=tensor([0.0120, 0.0111, 0.0196, 0.0329, 0.0106, 0.2715, 0.0141, 0.0256], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0144, 0.0190, 0.0174, 0.0166, 0.0199, 0.0180, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:02:07,250 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:02:15,413 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 22:02:19,466 INFO [train.py:904] (7/8) Epoch 19, batch 5650, loss[loss=0.207, simple_loss=0.2931, pruned_loss=0.06045, over 16787.00 frames. ], tot_loss[loss=0.2119, simple_loss=0.2961, pruned_loss=0.06384, over 3072741.82 frames. ], batch size: 124, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:02:36,992 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.386e+02 3.211e+02 3.758e+02 5.159e+02 1.051e+03, threshold=7.516e+02, percent-clipped=3.0 2023-04-30 22:03:36,661 INFO [train.py:904] (7/8) Epoch 19, batch 5700, loss[loss=0.2677, simple_loss=0.3316, pruned_loss=0.1019, over 11609.00 frames. ], tot_loss[loss=0.2145, simple_loss=0.2978, pruned_loss=0.06566, over 3048998.43 frames. ], batch size: 248, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:03:40,550 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3732, 4.6378, 4.4149, 4.4521, 4.1573, 4.0953, 4.1684, 4.6823], device='cuda:7'), covar=tensor([0.1201, 0.0891, 0.1091, 0.0911, 0.0869, 0.1597, 0.1112, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0634, 0.0782, 0.0643, 0.0581, 0.0490, 0.0497, 0.0646, 0.0601], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:04:55,857 INFO [train.py:904] (7/8) Epoch 19, batch 5750, loss[loss=0.1957, simple_loss=0.2872, pruned_loss=0.0521, over 16912.00 frames. ], tot_loss[loss=0.2164, simple_loss=0.2998, pruned_loss=0.06655, over 3047667.99 frames. ], batch size: 90, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:05:12,679 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.963e+02 2.939e+02 3.429e+02 4.149e+02 7.338e+02, threshold=6.857e+02, percent-clipped=0.0 2023-04-30 22:05:36,676 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-04-30 22:05:45,293 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188481.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:06:19,334 INFO [train.py:904] (7/8) Epoch 19, batch 5800, loss[loss=0.2064, simple_loss=0.2913, pruned_loss=0.06077, over 16927.00 frames. ], tot_loss[loss=0.2157, simple_loss=0.2998, pruned_loss=0.06575, over 3043766.84 frames. ], batch size: 116, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:07:03,973 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-04-30 22:07:24,120 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188542.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:07:38,959 INFO [train.py:904] (7/8) Epoch 19, batch 5850, loss[loss=0.2029, simple_loss=0.2896, pruned_loss=0.05813, over 16561.00 frames. ], tot_loss[loss=0.213, simple_loss=0.2975, pruned_loss=0.06427, over 3048772.20 frames. ], batch size: 62, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:07:57,314 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.681e+02 3.202e+02 3.753e+02 8.228e+02, threshold=6.404e+02, percent-clipped=1.0 2023-04-30 22:08:22,712 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5962, 1.7572, 2.1654, 2.5038, 2.4926, 2.8588, 1.8625, 2.8170], device='cuda:7'), covar=tensor([0.0197, 0.0481, 0.0320, 0.0297, 0.0302, 0.0180, 0.0526, 0.0143], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0188, 0.0176, 0.0178, 0.0189, 0.0148, 0.0191, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:08:59,608 INFO [train.py:904] (7/8) Epoch 19, batch 5900, loss[loss=0.2327, simple_loss=0.298, pruned_loss=0.08374, over 11266.00 frames. ], tot_loss[loss=0.213, simple_loss=0.297, pruned_loss=0.06447, over 3030102.81 frames. ], batch size: 246, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:10:01,935 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=188639.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:10:21,858 INFO [train.py:904] (7/8) Epoch 19, batch 5950, loss[loss=0.2725, simple_loss=0.3451, pruned_loss=0.09994, over 11497.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2977, pruned_loss=0.06332, over 3039781.90 frames. ], batch size: 247, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:10:40,535 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.478e+02 2.950e+02 3.832e+02 6.444e+02, threshold=5.900e+02, percent-clipped=1.0 2023-04-30 22:10:58,928 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3460, 3.2950, 3.2378, 3.4475, 3.4318, 3.2402, 3.4526, 3.4963], device='cuda:7'), covar=tensor([0.1349, 0.1130, 0.1512, 0.0828, 0.1020, 0.2580, 0.1202, 0.1115], device='cuda:7'), in_proj_covar=tensor([0.0608, 0.0746, 0.0877, 0.0768, 0.0573, 0.0602, 0.0616, 0.0719], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:10:59,020 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6423, 2.6010, 1.8331, 2.7292, 2.1271, 2.7577, 2.0346, 2.3196], device='cuda:7'), covar=tensor([0.0301, 0.0360, 0.1252, 0.0249, 0.0652, 0.0499, 0.1156, 0.0588], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0176, 0.0194, 0.0156, 0.0176, 0.0215, 0.0202, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:11:41,786 INFO [train.py:904] (7/8) Epoch 19, batch 6000, loss[loss=0.1771, simple_loss=0.2685, pruned_loss=0.04285, over 16701.00 frames. ], tot_loss[loss=0.2108, simple_loss=0.2964, pruned_loss=0.06261, over 3052593.51 frames. ], batch size: 89, lr: 3.56e-03, grad_scale: 8.0 2023-04-30 22:11:41,787 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 22:11:52,556 INFO [train.py:938] (7/8) Epoch 19, validation: loss=0.1523, simple_loss=0.2653, pruned_loss=0.01968, over 944034.00 frames. 2023-04-30 22:11:52,557 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 22:12:31,035 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-04-30 22:12:58,266 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8596, 3.2070, 3.1348, 2.1053, 2.9933, 3.1537, 3.0743, 1.8964], device='cuda:7'), covar=tensor([0.0557, 0.0060, 0.0068, 0.0425, 0.0110, 0.0115, 0.0095, 0.0455], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0079, 0.0080, 0.0133, 0.0095, 0.0106, 0.0092, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 22:13:00,918 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8681, 2.1033, 2.4189, 3.1101, 2.1871, 2.3130, 2.3110, 2.2152], device='cuda:7'), covar=tensor([0.1302, 0.3305, 0.2254, 0.0657, 0.3915, 0.2258, 0.2900, 0.3099], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0434, 0.0357, 0.0320, 0.0431, 0.0500, 0.0405, 0.0506], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:13:13,867 INFO [train.py:904] (7/8) Epoch 19, batch 6050, loss[loss=0.2109, simple_loss=0.2966, pruned_loss=0.06261, over 15474.00 frames. ], tot_loss[loss=0.2093, simple_loss=0.2947, pruned_loss=0.06198, over 3055207.57 frames. ], batch size: 190, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:13:33,108 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.719e+02 3.422e+02 3.983e+02 7.831e+02, threshold=6.844e+02, percent-clipped=4.0 2023-04-30 22:13:52,533 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-04-30 22:14:06,347 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-04-30 22:14:32,951 INFO [train.py:904] (7/8) Epoch 19, batch 6100, loss[loss=0.1878, simple_loss=0.2777, pruned_loss=0.04899, over 17262.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.294, pruned_loss=0.06113, over 3072738.96 frames. ], batch size: 52, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:15:33,788 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=188837.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:15:57,523 INFO [train.py:904] (7/8) Epoch 19, batch 6150, loss[loss=0.2231, simple_loss=0.2907, pruned_loss=0.07781, over 11669.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2914, pruned_loss=0.05978, over 3091971.71 frames. ], batch size: 246, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:16:16,885 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.888e+02 3.402e+02 4.182e+02 1.052e+03, threshold=6.804e+02, percent-clipped=3.0 2023-04-30 22:17:13,508 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2023-04-30 22:17:16,341 INFO [train.py:904] (7/8) Epoch 19, batch 6200, loss[loss=0.2208, simple_loss=0.2877, pruned_loss=0.07695, over 11515.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2902, pruned_loss=0.06, over 3068045.61 frames. ], batch size: 248, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:17:25,532 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-04-30 22:17:58,813 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188929.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:18:01,919 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=188931.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:18:04,172 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-04-30 22:18:15,036 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=188939.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:18:34,122 INFO [train.py:904] (7/8) Epoch 19, batch 6250, loss[loss=0.2079, simple_loss=0.296, pruned_loss=0.05991, over 15456.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2898, pruned_loss=0.05941, over 3084775.94 frames. ], batch size: 192, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:18:52,890 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 2.827e+02 3.410e+02 4.287e+02 1.283e+03, threshold=6.821e+02, percent-clipped=4.0 2023-04-30 22:19:28,072 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=188987.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:19:32,865 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188990.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:19:35,995 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=188992.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:19:38,535 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4691, 4.5313, 4.8990, 4.8682, 4.8742, 4.5597, 4.5306, 4.3934], device='cuda:7'), covar=tensor([0.0345, 0.0607, 0.0394, 0.0400, 0.0480, 0.0438, 0.0975, 0.0548], device='cuda:7'), in_proj_covar=tensor([0.0395, 0.0434, 0.0422, 0.0396, 0.0469, 0.0443, 0.0537, 0.0355], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 22:19:41,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0695, 5.5235, 5.7156, 5.4716, 5.5242, 6.0511, 5.5698, 5.3437], device='cuda:7'), covar=tensor([0.0935, 0.1798, 0.2263, 0.1783, 0.2210, 0.0880, 0.1467, 0.2160], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0575, 0.0634, 0.0484, 0.0639, 0.0668, 0.0498, 0.0644], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 22:19:51,028 INFO [train.py:904] (7/8) Epoch 19, batch 6300, loss[loss=0.2189, simple_loss=0.2944, pruned_loss=0.07171, over 11831.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2892, pruned_loss=0.05823, over 3096767.90 frames. ], batch size: 247, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:20:35,802 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189030.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:21:09,036 INFO [train.py:904] (7/8) Epoch 19, batch 6350, loss[loss=0.1919, simple_loss=0.2839, pruned_loss=0.04988, over 16894.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2893, pruned_loss=0.05853, over 3111438.60 frames. ], batch size: 96, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:21:27,083 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.691e+02 3.246e+02 4.132e+02 1.028e+03, threshold=6.492e+02, percent-clipped=3.0 2023-04-30 22:21:28,318 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189064.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:21:51,931 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189079.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:22:09,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8859, 2.7943, 2.8260, 2.1588, 2.6932, 2.1853, 2.8120, 2.9854], device='cuda:7'), covar=tensor([0.0267, 0.0884, 0.0576, 0.1797, 0.0766, 0.1059, 0.0530, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0161, 0.0167, 0.0152, 0.0145, 0.0129, 0.0143, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:22:11,150 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189091.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:22:11,442 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2023-04-30 22:22:26,889 INFO [train.py:904] (7/8) Epoch 19, batch 6400, loss[loss=0.1996, simple_loss=0.2796, pruned_loss=0.05985, over 16865.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.289, pruned_loss=0.0593, over 3111734.01 frames. ], batch size: 116, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:22:37,513 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9358, 2.7785, 2.8806, 2.1284, 2.7166, 2.2200, 2.8447, 3.0159], device='cuda:7'), covar=tensor([0.0243, 0.0822, 0.0506, 0.1716, 0.0736, 0.0942, 0.0507, 0.0629], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0161, 0.0167, 0.0152, 0.0145, 0.0129, 0.0143, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:23:02,783 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189125.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:23:21,511 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189137.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:23:25,702 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189140.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:23:43,276 INFO [train.py:904] (7/8) Epoch 19, batch 6450, loss[loss=0.1967, simple_loss=0.2849, pruned_loss=0.05429, over 16214.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2893, pruned_loss=0.05892, over 3105993.98 frames. ], batch size: 165, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:23:49,693 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-04-30 22:24:01,003 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.834e+02 2.902e+02 3.401e+02 4.159e+02 9.749e+02, threshold=6.803e+02, percent-clipped=3.0 2023-04-30 22:24:34,447 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:25:00,515 INFO [train.py:904] (7/8) Epoch 19, batch 6500, loss[loss=0.2, simple_loss=0.2845, pruned_loss=0.05779, over 16878.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2879, pruned_loss=0.05869, over 3096515.07 frames. ], batch size: 42, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:25:26,174 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-04-30 22:26:14,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1617, 4.9778, 5.1813, 5.3761, 5.5674, 4.9870, 5.5572, 5.5437], device='cuda:7'), covar=tensor([0.1947, 0.1319, 0.1606, 0.0622, 0.0639, 0.0825, 0.0661, 0.0640], device='cuda:7'), in_proj_covar=tensor([0.0610, 0.0748, 0.0878, 0.0766, 0.0574, 0.0602, 0.0617, 0.0719], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:26:14,697 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189248.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:26:19,681 INFO [train.py:904] (7/8) Epoch 19, batch 6550, loss[loss=0.2253, simple_loss=0.3166, pruned_loss=0.06701, over 15317.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2912, pruned_loss=0.06009, over 3090260.60 frames. ], batch size: 190, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:26:37,019 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.957e+02 2.752e+02 3.137e+02 4.130e+02 8.401e+02, threshold=6.273e+02, percent-clipped=2.0 2023-04-30 22:27:09,609 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189285.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:27:12,562 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189287.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:27:22,182 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-04-30 22:27:34,961 INFO [train.py:904] (7/8) Epoch 19, batch 6600, loss[loss=0.2033, simple_loss=0.2939, pruned_loss=0.05633, over 16465.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2935, pruned_loss=0.06037, over 3114145.69 frames. ], batch size: 75, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:27:45,671 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189309.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:28:44,339 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189347.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:28:50,951 INFO [train.py:904] (7/8) Epoch 19, batch 6650, loss[loss=0.2792, simple_loss=0.3387, pruned_loss=0.1099, over 11297.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2944, pruned_loss=0.06149, over 3108329.49 frames. ], batch size: 248, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:29:08,221 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.870e+02 3.524e+02 4.277e+02 9.308e+02, threshold=7.047e+02, percent-clipped=5.0 2023-04-30 22:29:19,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8984, 4.7217, 4.8323, 5.1315, 5.2726, 4.7715, 5.3806, 5.2859], device='cuda:7'), covar=tensor([0.1900, 0.1493, 0.1919, 0.0806, 0.0817, 0.0847, 0.0652, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0606, 0.0745, 0.0874, 0.0761, 0.0573, 0.0599, 0.0615, 0.0716], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:29:42,499 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189386.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:30:06,168 INFO [train.py:904] (7/8) Epoch 19, batch 6700, loss[loss=0.2055, simple_loss=0.2828, pruned_loss=0.06412, over 16798.00 frames. ], tot_loss[loss=0.2085, simple_loss=0.2931, pruned_loss=0.06194, over 3090799.91 frames. ], batch size: 39, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:30:13,002 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0544, 3.0617, 1.7854, 3.2413, 2.2579, 3.3086, 2.0186, 2.4958], device='cuda:7'), covar=tensor([0.0272, 0.0353, 0.1647, 0.0206, 0.0825, 0.0523, 0.1524, 0.0738], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0176, 0.0194, 0.0156, 0.0175, 0.0214, 0.0201, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:30:15,444 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189408.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:30:33,850 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189420.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:30:55,783 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-04-30 22:30:56,686 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189435.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:31:02,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2394, 3.1019, 3.3970, 1.6126, 3.4463, 3.5904, 2.8448, 2.6341], device='cuda:7'), covar=tensor([0.0859, 0.0268, 0.0219, 0.1363, 0.0112, 0.0187, 0.0410, 0.0530], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0138, 0.0078, 0.0122, 0.0126, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:31:21,925 INFO [train.py:904] (7/8) Epoch 19, batch 6750, loss[loss=0.2511, simple_loss=0.3156, pruned_loss=0.09331, over 11968.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2922, pruned_loss=0.06221, over 3088964.44 frames. ], batch size: 248, lr: 3.55e-03, grad_scale: 4.0 2023-04-30 22:31:42,236 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.911e+02 3.292e+02 3.890e+02 7.012e+02, threshold=6.585e+02, percent-clipped=0.0 2023-04-30 22:32:38,397 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189501.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:32:39,014 INFO [train.py:904] (7/8) Epoch 19, batch 6800, loss[loss=0.201, simple_loss=0.2898, pruned_loss=0.05612, over 16688.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2922, pruned_loss=0.06194, over 3089262.72 frames. ], batch size: 89, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:32:56,872 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-04-30 22:33:00,734 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.8561, 6.2074, 5.8751, 6.0652, 5.5602, 5.4702, 5.7018, 6.3057], device='cuda:7'), covar=tensor([0.1200, 0.0794, 0.1032, 0.0749, 0.0804, 0.0618, 0.1065, 0.0761], device='cuda:7'), in_proj_covar=tensor([0.0641, 0.0783, 0.0647, 0.0585, 0.0491, 0.0503, 0.0652, 0.0604], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:33:58,211 INFO [train.py:904] (7/8) Epoch 19, batch 6850, loss[loss=0.2444, simple_loss=0.3101, pruned_loss=0.08933, over 11632.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2938, pruned_loss=0.06259, over 3077262.26 frames. ], batch size: 247, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:34:13,642 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189562.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:34:17,029 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.732e+02 3.346e+02 4.467e+02 6.913e+02, threshold=6.693e+02, percent-clipped=4.0 2023-04-30 22:34:47,786 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189585.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:34:50,376 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189587.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:35:13,001 INFO [train.py:904] (7/8) Epoch 19, batch 6900, loss[loss=0.2036, simple_loss=0.2934, pruned_loss=0.05692, over 16463.00 frames. ], tot_loss[loss=0.2095, simple_loss=0.2958, pruned_loss=0.06163, over 3091981.49 frames. ], batch size: 146, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:35:17,331 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189604.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:35:25,685 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9803, 2.0876, 2.1956, 3.4147, 2.0437, 2.3387, 2.2044, 2.1928], device='cuda:7'), covar=tensor([0.1344, 0.3312, 0.2657, 0.0630, 0.4077, 0.2460, 0.3273, 0.3091], device='cuda:7'), in_proj_covar=tensor([0.0390, 0.0430, 0.0355, 0.0317, 0.0429, 0.0496, 0.0403, 0.0503], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:36:00,958 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189633.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:01,303 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7691, 2.7393, 2.6913, 4.6652, 2.6151, 2.9971, 2.7593, 2.7951], device='cuda:7'), covar=tensor([0.1027, 0.3001, 0.2407, 0.0376, 0.3520, 0.2072, 0.2824, 0.3061], device='cuda:7'), in_proj_covar=tensor([0.0390, 0.0430, 0.0355, 0.0317, 0.0429, 0.0495, 0.0403, 0.0503], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:36:04,228 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189635.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:12,818 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189640.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:36:27,621 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-04-30 22:36:29,242 INFO [train.py:904] (7/8) Epoch 19, batch 6950, loss[loss=0.2742, simple_loss=0.3359, pruned_loss=0.1063, over 11342.00 frames. ], tot_loss[loss=0.2123, simple_loss=0.2978, pruned_loss=0.06338, over 3084569.83 frames. ], batch size: 248, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:36:48,882 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.017e+02 3.156e+02 3.874e+02 4.949e+02 9.528e+02, threshold=7.748e+02, percent-clipped=6.0 2023-04-30 22:36:53,143 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-04-30 22:37:21,295 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189686.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:37:43,813 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189701.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:37:44,548 INFO [train.py:904] (7/8) Epoch 19, batch 7000, loss[loss=0.2304, simple_loss=0.2994, pruned_loss=0.08066, over 11327.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2973, pruned_loss=0.06223, over 3100307.07 frames. ], batch size: 246, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:37:47,786 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189703.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:37:49,929 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189704.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:13,454 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189720.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:25,574 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=189728.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:34,815 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189734.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:38:35,025 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0109, 2.3647, 2.3077, 2.8850, 2.0205, 3.1845, 1.8070, 2.7172], device='cuda:7'), covar=tensor([0.1177, 0.0637, 0.1086, 0.0191, 0.0143, 0.0360, 0.1449, 0.0721], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0173, 0.0195, 0.0185, 0.0207, 0.0215, 0.0198, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:38:36,125 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189735.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:39:01,647 INFO [train.py:904] (7/8) Epoch 19, batch 7050, loss[loss=0.2192, simple_loss=0.3028, pruned_loss=0.06777, over 15348.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2988, pruned_loss=0.06277, over 3079471.18 frames. ], batch size: 191, lr: 3.55e-03, grad_scale: 8.0 2023-04-30 22:39:22,081 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.118e+02 2.966e+02 3.403e+02 4.255e+02 9.294e+02, threshold=6.806e+02, percent-clipped=3.0 2023-04-30 22:39:23,231 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189765.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:39:27,448 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189768.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:39:50,864 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189783.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:40:01,258 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=189789.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:40:19,988 INFO [train.py:904] (7/8) Epoch 19, batch 7100, loss[loss=0.2444, simple_loss=0.3125, pruned_loss=0.0882, over 11450.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2972, pruned_loss=0.06269, over 3068120.64 frames. ], batch size: 250, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:40:33,222 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8020, 5.2759, 5.4274, 5.2064, 5.2944, 5.8223, 5.2735, 5.0906], device='cuda:7'), covar=tensor([0.1034, 0.1777, 0.1936, 0.1721, 0.2111, 0.0830, 0.1528, 0.2142], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0573, 0.0632, 0.0479, 0.0637, 0.0662, 0.0495, 0.0643], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 22:41:26,582 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3218, 1.5920, 2.0532, 2.1687, 2.2925, 2.5251, 1.6440, 2.3964], device='cuda:7'), covar=tensor([0.0218, 0.0512, 0.0278, 0.0321, 0.0287, 0.0187, 0.0513, 0.0140], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0188, 0.0174, 0.0177, 0.0189, 0.0148, 0.0190, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:41:38,324 INFO [train.py:904] (7/8) Epoch 19, batch 7150, loss[loss=0.2573, simple_loss=0.3184, pruned_loss=0.09805, over 11510.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2949, pruned_loss=0.06236, over 3072598.25 frames. ], batch size: 248, lr: 3.54e-03, grad_scale: 4.0 2023-04-30 22:41:45,580 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189857.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:41:49,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3873, 4.4720, 4.7794, 4.7152, 4.7633, 4.4271, 4.4526, 4.3467], device='cuda:7'), covar=tensor([0.0333, 0.0626, 0.0352, 0.0436, 0.0439, 0.0451, 0.0819, 0.0520], device='cuda:7'), in_proj_covar=tensor([0.0392, 0.0430, 0.0417, 0.0393, 0.0465, 0.0441, 0.0531, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 22:41:51,213 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-04-30 22:41:58,952 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.902e+02 2.755e+02 3.301e+02 3.879e+02 9.632e+02, threshold=6.601e+02, percent-clipped=3.0 2023-04-30 22:42:51,652 INFO [train.py:904] (7/8) Epoch 19, batch 7200, loss[loss=0.1772, simple_loss=0.2734, pruned_loss=0.04048, over 16719.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2922, pruned_loss=0.06048, over 3066072.52 frames. ], batch size: 134, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:42:55,730 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=189904.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:44:12,498 INFO [train.py:904] (7/8) Epoch 19, batch 7250, loss[loss=0.1974, simple_loss=0.2716, pruned_loss=0.06159, over 16251.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2899, pruned_loss=0.05927, over 3071173.27 frames. ], batch size: 165, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:44:12,830 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=189952.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:44:34,222 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.477e+02 2.873e+02 3.623e+02 8.553e+02, threshold=5.746e+02, percent-clipped=4.0 2023-04-30 22:45:19,999 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=189996.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:45:29,669 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3341, 3.3036, 3.8558, 1.8664, 4.0318, 3.9925, 2.8483, 2.8346], device='cuda:7'), covar=tensor([0.0961, 0.0305, 0.0208, 0.1287, 0.0073, 0.0156, 0.0483, 0.0535], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0106, 0.0096, 0.0138, 0.0078, 0.0122, 0.0126, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:45:32,090 INFO [train.py:904] (7/8) Epoch 19, batch 7300, loss[loss=0.2121, simple_loss=0.2954, pruned_loss=0.06437, over 16057.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.289, pruned_loss=0.05891, over 3076929.61 frames. ], batch size: 165, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:45:33,942 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190003.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:45:36,700 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-04-30 22:46:48,534 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190051.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:46:49,318 INFO [train.py:904] (7/8) Epoch 19, batch 7350, loss[loss=0.1909, simple_loss=0.2853, pruned_loss=0.04825, over 16776.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2899, pruned_loss=0.05974, over 3061729.03 frames. ], batch size: 83, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:47:01,739 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190060.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:47:10,691 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.253e+02 3.016e+02 3.386e+02 4.070e+02 1.285e+03, threshold=6.773e+02, percent-clipped=7.0 2023-04-30 22:47:41,037 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190084.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:48:08,860 INFO [train.py:904] (7/8) Epoch 19, batch 7400, loss[loss=0.2167, simple_loss=0.2986, pruned_loss=0.06745, over 16420.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2901, pruned_loss=0.05946, over 3078260.43 frames. ], batch size: 68, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:49:09,868 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190139.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 22:49:30,150 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190151.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:49:30,840 INFO [train.py:904] (7/8) Epoch 19, batch 7450, loss[loss=0.2012, simple_loss=0.2882, pruned_loss=0.05714, over 16775.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2918, pruned_loss=0.06107, over 3071701.53 frames. ], batch size: 83, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:49:41,014 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190157.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:49:55,758 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.975e+02 2.938e+02 3.493e+02 4.428e+02 1.014e+03, threshold=6.986e+02, percent-clipped=6.0 2023-04-30 22:50:51,802 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190200.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 22:50:54,219 INFO [train.py:904] (7/8) Epoch 19, batch 7500, loss[loss=0.1811, simple_loss=0.2776, pruned_loss=0.04232, over 16850.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.291, pruned_loss=0.05969, over 3066446.12 frames. ], batch size: 102, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:50:59,305 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190205.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:51:10,850 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190212.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:51:23,470 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-04-30 22:52:13,614 INFO [train.py:904] (7/8) Epoch 19, batch 7550, loss[loss=0.1843, simple_loss=0.2745, pruned_loss=0.04706, over 16913.00 frames. ], tot_loss[loss=0.206, simple_loss=0.291, pruned_loss=0.06047, over 3058868.90 frames. ], batch size: 96, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:52:34,942 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.643e+02 3.256e+02 4.054e+02 9.532e+02, threshold=6.511e+02, percent-clipped=2.0 2023-04-30 22:52:56,291 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1038, 5.6789, 5.8958, 5.5461, 5.6761, 6.2201, 5.6504, 5.4381], device='cuda:7'), covar=tensor([0.0889, 0.1812, 0.2010, 0.1836, 0.2258, 0.0846, 0.1676, 0.2428], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0574, 0.0631, 0.0479, 0.0638, 0.0661, 0.0495, 0.0643], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 22:53:22,566 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190296.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:53:31,291 INFO [train.py:904] (7/8) Epoch 19, batch 7600, loss[loss=0.2246, simple_loss=0.2994, pruned_loss=0.07489, over 15309.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2892, pruned_loss=0.05986, over 3070686.98 frames. ], batch size: 190, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:53:53,513 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190316.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:54:26,338 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4955, 3.5097, 3.4455, 2.6988, 3.3877, 2.0721, 3.0889, 2.6934], device='cuda:7'), covar=tensor([0.0156, 0.0121, 0.0171, 0.0218, 0.0104, 0.2223, 0.0132, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0143, 0.0189, 0.0172, 0.0164, 0.0199, 0.0178, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 22:54:37,074 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190344.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:54:45,292 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6518, 2.6246, 2.4217, 3.8136, 2.8855, 3.9455, 1.5086, 2.8960], device='cuda:7'), covar=tensor([0.1353, 0.0753, 0.1231, 0.0195, 0.0267, 0.0393, 0.1645, 0.0795], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0173, 0.0194, 0.0183, 0.0206, 0.0214, 0.0198, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-04-30 22:54:48,953 INFO [train.py:904] (7/8) Epoch 19, batch 7650, loss[loss=0.2105, simple_loss=0.2921, pruned_loss=0.0644, over 16620.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2899, pruned_loss=0.06021, over 3086783.04 frames. ], batch size: 134, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:55:02,255 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190360.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:55:10,295 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.023e+02 2.803e+02 3.556e+02 4.194e+02 6.713e+02, threshold=7.113e+02, percent-clipped=1.0 2023-04-30 22:55:27,606 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190377.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:55:38,700 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190384.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:56:02,488 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190399.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:56:06,248 INFO [train.py:904] (7/8) Epoch 19, batch 7700, loss[loss=0.206, simple_loss=0.2931, pruned_loss=0.05946, over 16642.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2903, pruned_loss=0.06125, over 3068097.88 frames. ], batch size: 62, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:56:15,130 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190408.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:56:50,414 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190432.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:57:21,413 INFO [train.py:904] (7/8) Epoch 19, batch 7750, loss[loss=0.1935, simple_loss=0.2827, pruned_loss=0.05218, over 16569.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2908, pruned_loss=0.0611, over 3088756.66 frames. ], batch size: 68, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:57:35,228 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190460.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:57:43,812 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.766e+02 2.725e+02 3.462e+02 4.075e+02 8.569e+02, threshold=6.924e+02, percent-clipped=2.0 2023-04-30 22:58:30,734 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190495.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 22:58:39,276 INFO [train.py:904] (7/8) Epoch 19, batch 7800, loss[loss=0.2706, simple_loss=0.3285, pruned_loss=0.1064, over 11656.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2921, pruned_loss=0.06191, over 3100111.53 frames. ], batch size: 248, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 22:58:47,784 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190507.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 22:59:57,144 INFO [train.py:904] (7/8) Epoch 19, batch 7850, loss[loss=0.2529, simple_loss=0.3147, pruned_loss=0.09553, over 11467.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2929, pruned_loss=0.06165, over 3090777.66 frames. ], batch size: 248, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:00:17,837 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.838e+02 3.506e+02 4.307e+02 8.316e+02, threshold=7.012e+02, percent-clipped=1.0 2023-04-30 23:00:22,529 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5492, 4.6234, 4.9712, 4.9228, 4.9395, 4.6317, 4.6470, 4.4986], device='cuda:7'), covar=tensor([0.0330, 0.0566, 0.0343, 0.0426, 0.0487, 0.0360, 0.0926, 0.0464], device='cuda:7'), in_proj_covar=tensor([0.0389, 0.0427, 0.0415, 0.0390, 0.0462, 0.0436, 0.0528, 0.0350], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-04-30 23:00:56,914 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190591.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 23:01:12,380 INFO [train.py:904] (7/8) Epoch 19, batch 7900, loss[loss=0.1892, simple_loss=0.2875, pruned_loss=0.04548, over 16807.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2919, pruned_loss=0.0609, over 3091393.56 frames. ], batch size: 83, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:01:23,372 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190609.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:02:32,483 INFO [train.py:904] (7/8) Epoch 19, batch 7950, loss[loss=0.2053, simple_loss=0.2929, pruned_loss=0.05881, over 16445.00 frames. ], tot_loss[loss=0.2075, simple_loss=0.2923, pruned_loss=0.06129, over 3098228.76 frames. ], batch size: 146, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:02:33,031 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190652.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 23:02:54,095 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.695e+02 3.262e+02 3.966e+02 7.542e+02, threshold=6.523e+02, percent-clipped=2.0 2023-04-30 23:03:01,394 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190670.0, num_to_drop=1, layers_to_drop={3} 2023-04-30 23:03:04,354 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190672.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:03:08,684 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=190675.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:03:49,834 INFO [train.py:904] (7/8) Epoch 19, batch 8000, loss[loss=0.2008, simple_loss=0.2818, pruned_loss=0.05992, over 17050.00 frames. ], tot_loss[loss=0.208, simple_loss=0.2927, pruned_loss=0.0616, over 3101558.20 frames. ], batch size: 53, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:04:44,444 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=190736.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:05:07,652 INFO [train.py:904] (7/8) Epoch 19, batch 8050, loss[loss=0.1912, simple_loss=0.281, pruned_loss=0.05071, over 16601.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2925, pruned_loss=0.06108, over 3089582.73 frames. ], batch size: 57, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:05:12,152 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190755.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:05:28,129 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.906e+02 2.693e+02 3.232e+02 4.056e+02 6.686e+02, threshold=6.463e+02, percent-clipped=1.0 2023-04-30 23:05:34,710 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-30 23:05:39,037 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5803, 4.5667, 4.4351, 3.6842, 4.4929, 1.5482, 4.2691, 4.1380], device='cuda:7'), covar=tensor([0.0091, 0.0079, 0.0187, 0.0364, 0.0100, 0.2938, 0.0133, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0143, 0.0190, 0.0172, 0.0165, 0.0199, 0.0178, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:06:12,513 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190795.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:06:22,314 INFO [train.py:904] (7/8) Epoch 19, batch 8100, loss[loss=0.2027, simple_loss=0.2855, pruned_loss=0.05994, over 15284.00 frames. ], tot_loss[loss=0.2073, simple_loss=0.2924, pruned_loss=0.06113, over 3089056.57 frames. ], batch size: 190, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:06:31,253 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190807.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:07:01,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2162, 3.6460, 3.6537, 2.3813, 3.3550, 3.6672, 3.3913, 2.0221], device='cuda:7'), covar=tensor([0.0542, 0.0055, 0.0055, 0.0408, 0.0103, 0.0120, 0.0097, 0.0465], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0079, 0.0080, 0.0133, 0.0094, 0.0107, 0.0091, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 23:07:25,606 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190843.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:07:39,359 INFO [train.py:904] (7/8) Epoch 19, batch 8150, loss[loss=0.1666, simple_loss=0.2639, pruned_loss=0.03467, over 16863.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2896, pruned_loss=0.05978, over 3109997.15 frames. ], batch size: 96, lr: 3.54e-03, grad_scale: 8.0 2023-04-30 23:07:45,309 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=190855.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:07:50,346 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-04-30 23:08:01,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.839e+02 2.760e+02 3.332e+02 4.142e+02 6.995e+02, threshold=6.664e+02, percent-clipped=2.0 2023-04-30 23:08:57,387 INFO [train.py:904] (7/8) Epoch 19, batch 8200, loss[loss=0.1935, simple_loss=0.2828, pruned_loss=0.05204, over 15320.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2867, pruned_loss=0.05867, over 3114421.32 frames. ], batch size: 191, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:10:12,364 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190947.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 23:10:19,467 INFO [train.py:904] (7/8) Epoch 19, batch 8250, loss[loss=0.1868, simple_loss=0.2726, pruned_loss=0.05049, over 16877.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.286, pruned_loss=0.05638, over 3115316.82 frames. ], batch size: 109, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:10:40,448 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=190965.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 23:10:41,155 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.606e+02 3.069e+02 3.589e+02 7.796e+02, threshold=6.137e+02, percent-clipped=2.0 2023-04-30 23:10:51,891 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=190972.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:11:02,507 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2356, 5.5697, 5.3344, 5.3833, 5.0191, 4.9945, 4.9803, 5.6497], device='cuda:7'), covar=tensor([0.1137, 0.0809, 0.0946, 0.0767, 0.0910, 0.0787, 0.1170, 0.0833], device='cuda:7'), in_proj_covar=tensor([0.0635, 0.0769, 0.0639, 0.0577, 0.0484, 0.0497, 0.0646, 0.0595], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:11:39,010 INFO [train.py:904] (7/8) Epoch 19, batch 8300, loss[loss=0.1892, simple_loss=0.2838, pruned_loss=0.04729, over 15308.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2837, pruned_loss=0.05387, over 3088289.08 frames. ], batch size: 191, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:12:10,264 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191020.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:12:27,706 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191031.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:13:00,824 INFO [train.py:904] (7/8) Epoch 19, batch 8350, loss[loss=0.1863, simple_loss=0.2851, pruned_loss=0.04378, over 16406.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2838, pruned_loss=0.05238, over 3074998.48 frames. ], batch size: 146, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:13:08,027 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191055.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:13:24,346 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.343e+02 2.792e+02 3.340e+02 8.341e+02, threshold=5.585e+02, percent-clipped=1.0 2023-04-30 23:13:28,644 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191068.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:13:30,113 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-04-30 23:14:22,694 INFO [train.py:904] (7/8) Epoch 19, batch 8400, loss[loss=0.1894, simple_loss=0.2797, pruned_loss=0.04957, over 16706.00 frames. ], tot_loss[loss=0.1916, simple_loss=0.2814, pruned_loss=0.05086, over 3038655.79 frames. ], batch size: 124, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:14:24,519 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191103.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:15:06,993 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191129.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:15:11,116 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191132.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:15:42,696 INFO [train.py:904] (7/8) Epoch 19, batch 8450, loss[loss=0.1565, simple_loss=0.2542, pruned_loss=0.0294, over 16446.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2789, pruned_loss=0.04891, over 3045623.68 frames. ], batch size: 68, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:16:06,337 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.247e+02 2.599e+02 3.228e+02 6.628e+02, threshold=5.197e+02, percent-clipped=1.0 2023-04-30 23:16:35,134 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0126, 2.0415, 2.1850, 3.5261, 2.0386, 2.3117, 2.2290, 2.1891], device='cuda:7'), covar=tensor([0.1244, 0.4098, 0.3019, 0.0591, 0.4725, 0.2817, 0.3656, 0.3794], device='cuda:7'), in_proj_covar=tensor([0.0382, 0.0424, 0.0349, 0.0312, 0.0421, 0.0486, 0.0395, 0.0494], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:16:36,957 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191185.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:16:50,129 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191193.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:17:04,717 INFO [train.py:904] (7/8) Epoch 19, batch 8500, loss[loss=0.1658, simple_loss=0.26, pruned_loss=0.03581, over 16589.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2749, pruned_loss=0.04642, over 3044374.12 frames. ], batch size: 68, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:18:22,796 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191246.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:18:25,058 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191247.0, num_to_drop=1, layers_to_drop={0} 2023-04-30 23:18:32,475 INFO [train.py:904] (7/8) Epoch 19, batch 8550, loss[loss=0.1843, simple_loss=0.2814, pruned_loss=0.0436, over 16852.00 frames. ], tot_loss[loss=0.1822, simple_loss=0.2732, pruned_loss=0.04563, over 3038862.18 frames. ], batch size: 116, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:18:46,144 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5218, 2.9141, 3.2933, 1.9432, 2.8084, 2.0644, 3.0243, 3.1008], device='cuda:7'), covar=tensor([0.0311, 0.0887, 0.0529, 0.2097, 0.0848, 0.1063, 0.0758, 0.0950], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0156, 0.0162, 0.0147, 0.0141, 0.0125, 0.0140, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 23:18:55,801 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4893, 3.4903, 3.4927, 2.6523, 3.4002, 2.0292, 3.1578, 2.7957], device='cuda:7'), covar=tensor([0.0131, 0.0123, 0.0172, 0.0177, 0.0111, 0.2337, 0.0130, 0.0243], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0141, 0.0185, 0.0167, 0.0162, 0.0196, 0.0174, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:18:57,580 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191265.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:19:00,620 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.304e+02 2.724e+02 3.153e+02 7.683e+02, threshold=5.448e+02, percent-clipped=1.0 2023-04-30 23:20:00,167 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191295.0, num_to_drop=1, layers_to_drop={1} 2023-04-30 23:20:13,465 INFO [train.py:904] (7/8) Epoch 19, batch 8600, loss[loss=0.1761, simple_loss=0.2757, pruned_loss=0.03829, over 16106.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2737, pruned_loss=0.04488, over 3030308.82 frames. ], batch size: 165, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:20:37,851 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191313.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:21:12,196 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191331.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:21:52,944 INFO [train.py:904] (7/8) Epoch 19, batch 8650, loss[loss=0.1969, simple_loss=0.293, pruned_loss=0.05037, over 11990.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2718, pruned_loss=0.04375, over 3007039.80 frames. ], batch size: 246, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:22:25,748 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=191365.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:22:29,950 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.144e+02 2.571e+02 3.162e+02 5.065e+02, threshold=5.143e+02, percent-clipped=0.0 2023-04-30 23:22:55,734 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191379.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:23:04,874 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7929, 1.3620, 1.6982, 1.7360, 1.8428, 1.8829, 1.6729, 1.8291], device='cuda:7'), covar=tensor([0.0269, 0.0423, 0.0221, 0.0286, 0.0291, 0.0196, 0.0393, 0.0132], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0186, 0.0171, 0.0175, 0.0187, 0.0145, 0.0188, 0.0139], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:23:39,029 INFO [train.py:904] (7/8) Epoch 19, batch 8700, loss[loss=0.1725, simple_loss=0.2671, pruned_loss=0.03894, over 15297.00 frames. ], tot_loss[loss=0.1771, simple_loss=0.2695, pruned_loss=0.04237, over 3016400.29 frames. ], batch size: 191, lr: 3.53e-03, grad_scale: 2.0 2023-04-30 23:24:20,727 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191424.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:24:23,702 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=191426.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:25:14,459 INFO [train.py:904] (7/8) Epoch 19, batch 8750, loss[loss=0.1577, simple_loss=0.2469, pruned_loss=0.03422, over 12293.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2689, pruned_loss=0.0416, over 3030562.79 frames. ], batch size: 248, lr: 3.53e-03, grad_scale: 2.0 2023-04-30 23:25:47,498 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-04-30 23:25:58,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.109e+02 2.725e+02 3.287e+02 5.129e+02, threshold=5.449e+02, percent-clipped=0.0 2023-04-30 23:26:40,164 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191488.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:26:53,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6740, 3.7053, 2.2986, 4.2882, 2.7729, 4.2262, 2.3956, 3.0144], device='cuda:7'), covar=tensor([0.0245, 0.0293, 0.1500, 0.0168, 0.0865, 0.0415, 0.1446, 0.0698], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0168, 0.0188, 0.0149, 0.0169, 0.0206, 0.0195, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-04-30 23:27:07,640 INFO [train.py:904] (7/8) Epoch 19, batch 8800, loss[loss=0.1704, simple_loss=0.2641, pruned_loss=0.03829, over 15289.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2668, pruned_loss=0.04026, over 3030671.43 frames. ], batch size: 191, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:28:29,947 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191541.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:28:51,738 INFO [train.py:904] (7/8) Epoch 19, batch 8850, loss[loss=0.1643, simple_loss=0.2692, pruned_loss=0.02971, over 15369.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2684, pruned_loss=0.03968, over 3016281.95 frames. ], batch size: 191, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:29:28,630 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.105e+02 2.539e+02 3.345e+02 5.950e+02, threshold=5.078e+02, percent-clipped=2.0 2023-04-30 23:30:39,783 INFO [train.py:904] (7/8) Epoch 19, batch 8900, loss[loss=0.1572, simple_loss=0.2594, pruned_loss=0.02745, over 17124.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2691, pruned_loss=0.03906, over 3031305.21 frames. ], batch size: 47, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:31:24,535 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-04-30 23:32:45,282 INFO [train.py:904] (7/8) Epoch 19, batch 8950, loss[loss=0.1764, simple_loss=0.2727, pruned_loss=0.04006, over 12172.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2688, pruned_loss=0.03908, over 3040898.91 frames. ], batch size: 248, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:33:21,164 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.162e+02 2.456e+02 2.877e+02 5.499e+02, threshold=4.911e+02, percent-clipped=2.0 2023-04-30 23:34:34,403 INFO [train.py:904] (7/8) Epoch 19, batch 9000, loss[loss=0.1778, simple_loss=0.2598, pruned_loss=0.04786, over 11908.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2662, pruned_loss=0.03842, over 3041720.02 frames. ], batch size: 248, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:34:34,404 INFO [train.py:929] (7/8) Computing validation loss 2023-04-30 23:34:44,208 INFO [train.py:938] (7/8) Epoch 19, validation: loss=0.1462, simple_loss=0.2506, pruned_loss=0.02087, over 944034.00 frames. 2023-04-30 23:34:44,209 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-04-30 23:35:13,770 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-04-30 23:35:26,244 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=191721.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:35:31,655 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191724.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:36:27,798 INFO [train.py:904] (7/8) Epoch 19, batch 9050, loss[loss=0.1969, simple_loss=0.2845, pruned_loss=0.05462, over 12862.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2672, pruned_loss=0.03891, over 3067948.33 frames. ], batch size: 248, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:37:04,258 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.085e+02 2.454e+02 3.075e+02 7.905e+02, threshold=4.907e+02, percent-clipped=4.0 2023-04-30 23:37:09,419 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191772.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:37:39,898 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191788.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:38:04,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7310, 4.5776, 4.7710, 4.9268, 5.0846, 4.5524, 5.0982, 5.0871], device='cuda:7'), covar=tensor([0.1746, 0.1176, 0.1512, 0.0673, 0.0498, 0.0835, 0.0442, 0.0639], device='cuda:7'), in_proj_covar=tensor([0.0584, 0.0724, 0.0845, 0.0742, 0.0555, 0.0581, 0.0596, 0.0696], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:38:10,491 INFO [train.py:904] (7/8) Epoch 19, batch 9100, loss[loss=0.1756, simple_loss=0.2681, pruned_loss=0.04151, over 16623.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2668, pruned_loss=0.03921, over 3061631.91 frames. ], batch size: 62, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:38:18,918 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-04-30 23:38:32,057 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5273, 1.7377, 2.1086, 2.4749, 2.4339, 2.7538, 1.8688, 2.7353], device='cuda:7'), covar=tensor([0.0198, 0.0504, 0.0333, 0.0284, 0.0330, 0.0181, 0.0486, 0.0139], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0184, 0.0170, 0.0172, 0.0185, 0.0143, 0.0187, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:39:32,447 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191836.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:39:41,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2228, 1.5769, 1.9374, 2.1547, 2.2701, 2.4475, 1.6301, 2.3893], device='cuda:7'), covar=tensor([0.0270, 0.0576, 0.0342, 0.0334, 0.0368, 0.0229, 0.0576, 0.0150], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0184, 0.0169, 0.0172, 0.0184, 0.0142, 0.0187, 0.0137], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:39:43,939 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=191841.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:40:08,188 INFO [train.py:904] (7/8) Epoch 19, batch 9150, loss[loss=0.1559, simple_loss=0.2481, pruned_loss=0.03188, over 12032.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2667, pruned_loss=0.03914, over 3037105.79 frames. ], batch size: 247, lr: 3.53e-03, grad_scale: 4.0 2023-04-30 23:40:46,345 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.243e+02 2.682e+02 3.704e+02 5.734e+02, threshold=5.364e+02, percent-clipped=7.0 2023-04-30 23:41:30,831 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=191889.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:41:52,738 INFO [train.py:904] (7/8) Epoch 19, batch 9200, loss[loss=0.1592, simple_loss=0.2573, pruned_loss=0.03052, over 15385.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2623, pruned_loss=0.03794, over 3061976.94 frames. ], batch size: 190, lr: 3.53e-03, grad_scale: 8.0 2023-04-30 23:42:35,558 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1586, 3.3559, 3.3915, 2.3121, 3.0806, 3.4192, 3.2818, 1.9379], device='cuda:7'), covar=tensor([0.0469, 0.0050, 0.0049, 0.0372, 0.0105, 0.0077, 0.0076, 0.0481], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0077, 0.0078, 0.0131, 0.0092, 0.0103, 0.0089, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 23:43:29,335 INFO [train.py:904] (7/8) Epoch 19, batch 9250, loss[loss=0.1693, simple_loss=0.2652, pruned_loss=0.0367, over 16464.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2625, pruned_loss=0.03815, over 3055753.51 frames. ], batch size: 147, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:43:44,749 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5988, 3.9371, 2.8961, 2.2293, 2.4038, 2.4283, 4.1428, 3.3062], device='cuda:7'), covar=tensor([0.3049, 0.0530, 0.1799, 0.2948, 0.2988, 0.2104, 0.0413, 0.1310], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0258, 0.0294, 0.0299, 0.0283, 0.0246, 0.0281, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-04-30 23:44:05,911 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.629e+02 2.331e+02 2.686e+02 3.317e+02 7.389e+02, threshold=5.371e+02, percent-clipped=4.0 2023-04-30 23:45:23,377 INFO [train.py:904] (7/8) Epoch 19, batch 9300, loss[loss=0.1565, simple_loss=0.2454, pruned_loss=0.03384, over 16619.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2612, pruned_loss=0.03757, over 3071613.65 frames. ], batch size: 62, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:45:25,353 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9410, 2.6742, 2.9484, 2.0766, 2.7653, 2.1503, 2.7236, 2.8684], device='cuda:7'), covar=tensor([0.0285, 0.0989, 0.0460, 0.1795, 0.0721, 0.0895, 0.0590, 0.0924], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0152, 0.0160, 0.0146, 0.0140, 0.0124, 0.0138, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 23:45:46,861 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0974, 2.6225, 2.6178, 1.9807, 2.7828, 2.8885, 2.4935, 2.5455], device='cuda:7'), covar=tensor([0.0614, 0.0223, 0.0229, 0.0953, 0.0095, 0.0201, 0.0406, 0.0385], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0102, 0.0091, 0.0133, 0.0074, 0.0115, 0.0121, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-04-30 23:46:09,769 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192021.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:47:09,787 INFO [train.py:904] (7/8) Epoch 19, batch 9350, loss[loss=0.1636, simple_loss=0.2555, pruned_loss=0.03587, over 16818.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2608, pruned_loss=0.03748, over 3059736.38 frames. ], batch size: 83, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:47:46,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192069.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:47:47,132 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 1.981e+02 2.500e+02 3.039e+02 5.486e+02, threshold=4.999e+02, percent-clipped=1.0 2023-04-30 23:47:52,139 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6046, 2.0351, 1.7660, 1.8445, 2.3517, 2.0141, 1.9896, 2.4748], device='cuda:7'), covar=tensor([0.0148, 0.0413, 0.0521, 0.0502, 0.0267, 0.0405, 0.0154, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0189, 0.0223, 0.0216, 0.0217, 0.0225, 0.0222, 0.0222, 0.0215], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:48:37,320 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9898, 2.8469, 2.6879, 2.0804, 2.5714, 2.8369, 2.7161, 1.8923], device='cuda:7'), covar=tensor([0.0398, 0.0069, 0.0067, 0.0347, 0.0143, 0.0095, 0.0103, 0.0455], device='cuda:7'), in_proj_covar=tensor([0.0131, 0.0076, 0.0077, 0.0130, 0.0092, 0.0102, 0.0088, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-04-30 23:48:41,502 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7308, 3.5448, 3.5825, 3.8814, 3.9570, 3.6144, 3.9166, 3.9763], device='cuda:7'), covar=tensor([0.1749, 0.1577, 0.2218, 0.1093, 0.0982, 0.2583, 0.1186, 0.1279], device='cuda:7'), in_proj_covar=tensor([0.0584, 0.0724, 0.0844, 0.0743, 0.0556, 0.0582, 0.0597, 0.0697], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:48:49,313 INFO [train.py:904] (7/8) Epoch 19, batch 9400, loss[loss=0.1693, simple_loss=0.2701, pruned_loss=0.03427, over 16189.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2607, pruned_loss=0.03715, over 3046399.60 frames. ], batch size: 165, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:48:50,418 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192102.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:50:29,966 INFO [train.py:904] (7/8) Epoch 19, batch 9450, loss[loss=0.1452, simple_loss=0.2437, pruned_loss=0.02333, over 16841.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2622, pruned_loss=0.03705, over 3059029.76 frames. ], batch size: 102, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:50:35,967 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192155.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:50:51,124 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192163.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:51:05,241 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.168e+02 2.516e+02 3.148e+02 5.649e+02, threshold=5.031e+02, percent-clipped=1.0 2023-04-30 23:51:54,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1137, 5.1337, 4.9698, 4.5242, 4.6258, 5.0667, 4.9329, 4.6986], device='cuda:7'), covar=tensor([0.0575, 0.0423, 0.0284, 0.0297, 0.0948, 0.0460, 0.0313, 0.0621], device='cuda:7'), in_proj_covar=tensor([0.0267, 0.0382, 0.0311, 0.0303, 0.0319, 0.0355, 0.0216, 0.0371], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-04-30 23:52:10,479 INFO [train.py:904] (7/8) Epoch 19, batch 9500, loss[loss=0.1571, simple_loss=0.259, pruned_loss=0.02762, over 15428.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2617, pruned_loss=0.03683, over 3057845.04 frames. ], batch size: 191, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:52:21,865 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5576, 3.5236, 3.5375, 2.9220, 3.4770, 1.9592, 3.2348, 2.9327], device='cuda:7'), covar=tensor([0.0126, 0.0112, 0.0163, 0.0206, 0.0105, 0.2314, 0.0122, 0.0227], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0139, 0.0181, 0.0163, 0.0159, 0.0194, 0.0171, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:52:40,460 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192216.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:52:45,063 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192219.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:52:50,017 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192221.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:52:57,814 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5016, 3.7011, 4.0707, 2.1657, 3.4169, 2.7038, 3.8873, 3.7686], device='cuda:7'), covar=tensor([0.0230, 0.0763, 0.0424, 0.1957, 0.0658, 0.0826, 0.0532, 0.1039], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0151, 0.0160, 0.0146, 0.0139, 0.0124, 0.0138, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 23:53:11,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3795, 3.4140, 2.6772, 2.0932, 2.1985, 2.3364, 3.4586, 3.0896], device='cuda:7'), covar=tensor([0.2912, 0.0612, 0.1806, 0.2815, 0.2667, 0.2043, 0.0466, 0.1187], device='cuda:7'), in_proj_covar=tensor([0.0313, 0.0254, 0.0289, 0.0293, 0.0277, 0.0242, 0.0276, 0.0314], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:53:55,148 INFO [train.py:904] (7/8) Epoch 19, batch 9550, loss[loss=0.1751, simple_loss=0.2648, pruned_loss=0.04272, over 12773.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2617, pruned_loss=0.03705, over 3073833.82 frames. ], batch size: 247, lr: 3.52e-03, grad_scale: 4.0 2023-04-30 23:54:34,515 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.006e+02 2.351e+02 2.778e+02 5.702e+02, threshold=4.701e+02, percent-clipped=1.0 2023-04-30 23:54:55,700 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192280.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:55:01,216 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192282.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:55:03,378 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9274, 2.6922, 2.9193, 2.0212, 2.7052, 2.1310, 2.6378, 2.8408], device='cuda:7'), covar=tensor([0.0287, 0.0837, 0.0478, 0.1828, 0.0727, 0.0906, 0.0614, 0.0914], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0151, 0.0159, 0.0146, 0.0138, 0.0123, 0.0138, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-04-30 23:55:09,673 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4184, 3.3891, 2.6829, 2.0554, 2.1478, 2.2311, 3.4634, 3.0379], device='cuda:7'), covar=tensor([0.2843, 0.0674, 0.1743, 0.2911, 0.2723, 0.2234, 0.0413, 0.1326], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0254, 0.0288, 0.0293, 0.0276, 0.0242, 0.0276, 0.0315], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:55:38,448 INFO [train.py:904] (7/8) Epoch 19, batch 9600, loss[loss=0.1688, simple_loss=0.2677, pruned_loss=0.03493, over 16593.00 frames. ], tot_loss[loss=0.169, simple_loss=0.263, pruned_loss=0.0375, over 3065475.42 frames. ], batch size: 68, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:57:27,648 INFO [train.py:904] (7/8) Epoch 19, batch 9650, loss[loss=0.1585, simple_loss=0.2593, pruned_loss=0.02887, over 16905.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2644, pruned_loss=0.03751, over 3072449.35 frames. ], batch size: 102, lr: 3.52e-03, grad_scale: 8.0 2023-04-30 23:58:09,767 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.165e+02 2.448e+02 2.941e+02 5.888e+02, threshold=4.896e+02, percent-clipped=1.0 2023-04-30 23:58:54,588 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192392.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:59:09,415 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7215, 4.7276, 4.5136, 3.9588, 4.5887, 1.6980, 4.3683, 4.3682], device='cuda:7'), covar=tensor([0.0101, 0.0093, 0.0190, 0.0288, 0.0125, 0.2668, 0.0129, 0.0209], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0139, 0.0181, 0.0162, 0.0159, 0.0194, 0.0171, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-04-30 23:59:09,444 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192399.0, num_to_drop=0, layers_to_drop=set() 2023-04-30 23:59:15,199 INFO [train.py:904] (7/8) Epoch 19, batch 9700, loss[loss=0.1649, simple_loss=0.2634, pruned_loss=0.03316, over 16739.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2633, pruned_loss=0.03711, over 3072979.87 frames. ], batch size: 83, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:00:19,303 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192433.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:00:29,853 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 00:00:57,139 INFO [train.py:904] (7/8) Epoch 19, batch 9750, loss[loss=0.1346, simple_loss=0.2243, pruned_loss=0.02247, over 17124.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2619, pruned_loss=0.03725, over 3050266.99 frames. ], batch size: 47, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:00:59,789 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192453.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:01:09,810 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192458.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:01:14,190 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192460.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:01:31,658 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.106e+02 2.517e+02 3.092e+02 5.794e+02, threshold=5.033e+02, percent-clipped=3.0 2023-05-01 00:02:23,668 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192494.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:02:37,068 INFO [train.py:904] (7/8) Epoch 19, batch 9800, loss[loss=0.155, simple_loss=0.262, pruned_loss=0.024, over 17149.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2626, pruned_loss=0.0367, over 3077525.74 frames. ], batch size: 48, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:02:55,416 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192511.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:04:23,525 INFO [train.py:904] (7/8) Epoch 19, batch 9850, loss[loss=0.1551, simple_loss=0.2455, pruned_loss=0.03239, over 12217.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2636, pruned_loss=0.03644, over 3073612.50 frames. ], batch size: 248, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:05:00,569 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 1.961e+02 2.506e+02 2.957e+02 4.630e+02, threshold=5.011e+02, percent-clipped=0.0 2023-05-01 00:05:11,855 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192575.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:05:15,474 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192577.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:05:48,852 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9480, 1.9177, 2.1056, 3.4224, 1.9014, 2.1160, 2.0898, 1.9637], device='cuda:7'), covar=tensor([0.1556, 0.4859, 0.3238, 0.0704, 0.5772, 0.3418, 0.4215, 0.4757], device='cuda:7'), in_proj_covar=tensor([0.0382, 0.0425, 0.0352, 0.0311, 0.0424, 0.0486, 0.0396, 0.0493], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:06:14,640 INFO [train.py:904] (7/8) Epoch 19, batch 9900, loss[loss=0.1657, simple_loss=0.2674, pruned_loss=0.03202, over 16907.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2636, pruned_loss=0.03625, over 3061472.64 frames. ], batch size: 96, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:06:47,318 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:08:13,292 INFO [train.py:904] (7/8) Epoch 19, batch 9950, loss[loss=0.1631, simple_loss=0.2622, pruned_loss=0.03197, over 16935.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.266, pruned_loss=0.0367, over 3078462.30 frames. ], batch size: 116, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:08:18,756 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4075, 4.5067, 4.6439, 4.4919, 4.5593, 5.0502, 4.6010, 4.3362], device='cuda:7'), covar=tensor([0.1371, 0.1917, 0.2183, 0.1972, 0.2533, 0.1026, 0.1501, 0.2141], device='cuda:7'), in_proj_covar=tensor([0.0374, 0.0541, 0.0599, 0.0450, 0.0601, 0.0629, 0.0471, 0.0601], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 00:08:26,872 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0823, 3.9891, 4.1567, 4.2658, 4.3857, 3.9818, 4.3816, 4.4018], device='cuda:7'), covar=tensor([0.1658, 0.1043, 0.1250, 0.0662, 0.0543, 0.1316, 0.0600, 0.0690], device='cuda:7'), in_proj_covar=tensor([0.0578, 0.0713, 0.0833, 0.0735, 0.0549, 0.0572, 0.0592, 0.0686], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:08:54,789 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.121e+02 2.565e+02 3.086e+02 6.326e+02, threshold=5.129e+02, percent-clipped=1.0 2023-05-01 00:09:04,495 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192673.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:09:15,092 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:09:53,626 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-01 00:10:14,425 INFO [train.py:904] (7/8) Epoch 19, batch 10000, loss[loss=0.1639, simple_loss=0.2526, pruned_loss=0.03757, over 13249.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2648, pruned_loss=0.03664, over 3087053.82 frames. ], batch size: 248, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:11:18,867 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-05-01 00:11:23,380 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192734.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:11:25,777 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192735.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:11:50,857 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192748.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:11:56,969 INFO [train.py:904] (7/8) Epoch 19, batch 10050, loss[loss=0.1809, simple_loss=0.2657, pruned_loss=0.04803, over 11955.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2654, pruned_loss=0.0369, over 3087956.69 frames. ], batch size: 250, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:12:03,071 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192755.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:12:10,593 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192758.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:12:32,891 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.015e+02 2.519e+02 2.909e+02 8.183e+02, threshold=5.037e+02, percent-clipped=1.0 2023-05-01 00:13:08,118 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192789.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:13:21,302 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192796.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:13:28,491 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9749, 5.2755, 5.0343, 5.0327, 4.7659, 4.8103, 4.5832, 5.3489], device='cuda:7'), covar=tensor([0.1173, 0.0883, 0.1015, 0.0792, 0.0780, 0.0806, 0.1198, 0.0867], device='cuda:7'), in_proj_covar=tensor([0.0613, 0.0754, 0.0612, 0.0560, 0.0473, 0.0482, 0.0628, 0.0578], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:13:30,585 INFO [train.py:904] (7/8) Epoch 19, batch 10100, loss[loss=0.1641, simple_loss=0.2546, pruned_loss=0.03681, over 16213.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2651, pruned_loss=0.03692, over 3076913.29 frames. ], batch size: 165, lr: 3.52e-03, grad_scale: 8.0 2023-05-01 00:13:36,106 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4082, 3.3113, 3.4598, 1.8844, 3.6361, 3.6936, 2.9141, 2.9162], device='cuda:7'), covar=tensor([0.0718, 0.0206, 0.0178, 0.1126, 0.0065, 0.0143, 0.0415, 0.0393], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0100, 0.0088, 0.0131, 0.0073, 0.0113, 0.0119, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 00:13:39,005 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192806.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:13:48,409 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192811.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:14:28,838 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=192831.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:15:13,702 INFO [train.py:904] (7/8) Epoch 20, batch 0, loss[loss=0.2347, simple_loss=0.2989, pruned_loss=0.08523, over 16880.00 frames. ], tot_loss[loss=0.2347, simple_loss=0.2989, pruned_loss=0.08523, over 16880.00 frames. ], batch size: 116, lr: 3.43e-03, grad_scale: 8.0 2023-05-01 00:15:13,702 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 00:15:21,164 INFO [train.py:938] (7/8) Epoch 20, validation: loss=0.146, simple_loss=0.2496, pruned_loss=0.02121, over 944034.00 frames. 2023-05-01 00:15:21,165 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 00:15:32,534 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192859.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:15:36,728 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0280, 3.9005, 4.4146, 2.1830, 4.5636, 4.6255, 3.1534, 3.5467], device='cuda:7'), covar=tensor([0.0684, 0.0235, 0.0169, 0.1088, 0.0056, 0.0110, 0.0443, 0.0365], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0101, 0.0089, 0.0133, 0.0074, 0.0114, 0.0120, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 00:15:49,200 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.412e+02 2.946e+02 3.638e+02 7.163e+02, threshold=5.893e+02, percent-clipped=6.0 2023-05-01 00:15:53,692 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192875.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:15:56,057 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=192877.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:16:17,611 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=192892.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 00:16:31,057 INFO [train.py:904] (7/8) Epoch 20, batch 50, loss[loss=0.1512, simple_loss=0.2455, pruned_loss=0.02848, over 17210.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2712, pruned_loss=0.04979, over 757177.84 frames. ], batch size: 44, lr: 3.43e-03, grad_scale: 2.0 2023-05-01 00:17:00,246 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192923.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:17:03,081 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=192925.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:17:21,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4869, 3.4469, 3.4744, 2.7918, 3.3734, 2.0912, 3.1677, 2.8118], device='cuda:7'), covar=tensor([0.0134, 0.0109, 0.0175, 0.0186, 0.0100, 0.2220, 0.0118, 0.0247], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0142, 0.0184, 0.0163, 0.0161, 0.0198, 0.0174, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:17:38,948 INFO [train.py:904] (7/8) Epoch 20, batch 100, loss[loss=0.1505, simple_loss=0.2316, pruned_loss=0.03469, over 15819.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2679, pruned_loss=0.04836, over 1322597.68 frames. ], batch size: 35, lr: 3.43e-03, grad_scale: 2.0 2023-05-01 00:17:48,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1338, 5.4885, 5.2192, 5.2278, 4.9417, 4.9477, 4.9570, 5.6083], device='cuda:7'), covar=tensor([0.1280, 0.0991, 0.1221, 0.0951, 0.0932, 0.0879, 0.1145, 0.1018], device='cuda:7'), in_proj_covar=tensor([0.0627, 0.0774, 0.0627, 0.0573, 0.0484, 0.0494, 0.0645, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:18:07,335 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.214e+02 2.624e+02 3.260e+02 6.598e+02, threshold=5.249e+02, percent-clipped=3.0 2023-05-01 00:18:07,676 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=192972.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:18:34,918 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3245, 5.2303, 5.1521, 4.5787, 4.7862, 5.1741, 5.1473, 4.7783], device='cuda:7'), covar=tensor([0.0569, 0.0414, 0.0294, 0.0379, 0.1066, 0.0428, 0.0283, 0.0731], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0387, 0.0316, 0.0306, 0.0324, 0.0359, 0.0218, 0.0377], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:18:37,556 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 00:18:41,897 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8178, 3.8981, 2.3546, 4.5952, 3.0210, 4.5023, 2.4925, 3.2856], device='cuda:7'), covar=tensor([0.0361, 0.0507, 0.1769, 0.0249, 0.0923, 0.0562, 0.1624, 0.0732], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0171, 0.0189, 0.0152, 0.0172, 0.0207, 0.0198, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:18:48,440 INFO [train.py:904] (7/8) Epoch 20, batch 150, loss[loss=0.1519, simple_loss=0.2422, pruned_loss=0.03079, over 16863.00 frames. ], tot_loss[loss=0.1805, simple_loss=0.2658, pruned_loss=0.04757, over 1773518.41 frames. ], batch size: 42, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:19:01,827 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-01 00:19:27,972 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193029.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:19:44,023 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193041.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:19:52,856 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193048.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:19:58,182 INFO [train.py:904] (7/8) Epoch 20, batch 200, loss[loss=0.1611, simple_loss=0.25, pruned_loss=0.03612, over 15960.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.265, pruned_loss=0.04699, over 2121391.28 frames. ], batch size: 35, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:20:03,508 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193055.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:20:05,229 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 00:20:27,495 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 2.323e+02 2.688e+02 3.539e+02 1.444e+03, threshold=5.377e+02, percent-clipped=5.0 2023-05-01 00:20:50,744 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193089.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:20:53,795 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193091.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:00,958 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193096.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:08,819 INFO [train.py:904] (7/8) Epoch 20, batch 250, loss[loss=0.1626, simple_loss=0.2568, pruned_loss=0.03419, over 15965.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2639, pruned_loss=0.04717, over 2383494.40 frames. ], batch size: 35, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:21:09,321 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193102.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:10,866 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193103.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:21:58,202 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193137.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:22:17,124 INFO [train.py:904] (7/8) Epoch 20, batch 300, loss[loss=0.16, simple_loss=0.2426, pruned_loss=0.03869, over 16544.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2608, pruned_loss=0.04572, over 2583130.17 frames. ], batch size: 68, lr: 3.42e-03, grad_scale: 1.0 2023-05-01 00:22:46,397 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.161e+02 2.496e+02 2.884e+02 5.011e+02, threshold=4.993e+02, percent-clipped=0.0 2023-05-01 00:23:05,578 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193187.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:23:28,170 INFO [train.py:904] (7/8) Epoch 20, batch 350, loss[loss=0.1428, simple_loss=0.2317, pruned_loss=0.02694, over 16779.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2584, pruned_loss=0.04428, over 2758628.95 frames. ], batch size: 39, lr: 3.42e-03, grad_scale: 1.0 2023-05-01 00:23:39,643 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:24:37,908 INFO [train.py:904] (7/8) Epoch 20, batch 400, loss[loss=0.1599, simple_loss=0.2444, pruned_loss=0.0377, over 15833.00 frames. ], tot_loss[loss=0.172, simple_loss=0.257, pruned_loss=0.04354, over 2874305.56 frames. ], batch size: 35, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:25:00,552 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9768, 2.9728, 2.7724, 4.6150, 3.7922, 4.3217, 1.8342, 3.1672], device='cuda:7'), covar=tensor([0.1277, 0.0704, 0.1093, 0.0219, 0.0214, 0.0411, 0.1510, 0.0770], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0171, 0.0192, 0.0181, 0.0199, 0.0212, 0.0197, 0.0189], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:25:05,945 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193272.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:25:06,043 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193272.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:25:06,672 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.085e+02 2.558e+02 2.961e+02 5.266e+02, threshold=5.116e+02, percent-clipped=1.0 2023-05-01 00:25:20,678 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6628, 2.2903, 2.2304, 4.5513, 2.3321, 2.7469, 2.4074, 2.4595], device='cuda:7'), covar=tensor([0.1098, 0.3786, 0.3207, 0.0433, 0.4138, 0.2610, 0.3495, 0.3702], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0436, 0.0362, 0.0321, 0.0432, 0.0500, 0.0406, 0.0509], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:25:23,470 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193284.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:25:26,151 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 00:25:46,662 INFO [train.py:904] (7/8) Epoch 20, batch 450, loss[loss=0.1694, simple_loss=0.2679, pruned_loss=0.03547, over 17095.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.256, pruned_loss=0.04329, over 2974681.66 frames. ], batch size: 47, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:26:01,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2825, 4.4415, 4.5322, 3.4646, 3.7548, 4.4384, 3.9978, 2.7972], device='cuda:7'), covar=tensor([0.0378, 0.0063, 0.0036, 0.0280, 0.0131, 0.0091, 0.0080, 0.0410], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0081, 0.0080, 0.0134, 0.0095, 0.0106, 0.0092, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 00:26:14,330 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193320.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:26:26,087 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193329.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:26:48,829 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193345.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:26:57,295 INFO [train.py:904] (7/8) Epoch 20, batch 500, loss[loss=0.1565, simple_loss=0.2405, pruned_loss=0.03621, over 16868.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2548, pruned_loss=0.04268, over 3044120.69 frames. ], batch size: 96, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:27:26,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.231e+02 2.613e+02 3.252e+02 5.197e+02, threshold=5.226e+02, percent-clipped=2.0 2023-05-01 00:27:31,787 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193377.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:27:51,582 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193391.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:27:59,468 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193397.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:28:07,094 INFO [train.py:904] (7/8) Epoch 20, batch 550, loss[loss=0.1633, simple_loss=0.26, pruned_loss=0.03332, over 17264.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2536, pruned_loss=0.04241, over 3105382.13 frames. ], batch size: 52, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:28:32,375 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193421.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:28:45,606 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-01 00:28:46,891 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3453, 2.4714, 2.9461, 3.1932, 3.0853, 3.7154, 2.8747, 3.5997], device='cuda:7'), covar=tensor([0.0209, 0.0389, 0.0281, 0.0253, 0.0269, 0.0159, 0.0360, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0189, 0.0174, 0.0177, 0.0190, 0.0147, 0.0191, 0.0141], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:28:56,784 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193439.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:28:58,196 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0430, 4.7388, 5.0546, 5.2589, 5.4918, 4.7781, 5.4877, 5.4485], device='cuda:7'), covar=tensor([0.1967, 0.1422, 0.1793, 0.0819, 0.0517, 0.1003, 0.0457, 0.0639], device='cuda:7'), in_proj_covar=tensor([0.0627, 0.0778, 0.0909, 0.0795, 0.0594, 0.0621, 0.0644, 0.0743], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:29:14,821 INFO [train.py:904] (7/8) Epoch 20, batch 600, loss[loss=0.1445, simple_loss=0.2355, pruned_loss=0.02674, over 17195.00 frames. ], tot_loss[loss=0.168, simple_loss=0.253, pruned_loss=0.04147, over 3163318.43 frames. ], batch size: 46, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:29:18,249 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193454.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:29:38,399 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-01 00:29:43,218 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.108e+02 2.456e+02 2.826e+02 6.166e+02, threshold=4.913e+02, percent-clipped=1.0 2023-05-01 00:29:55,236 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193482.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:30:03,392 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193487.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:30:21,028 INFO [train.py:904] (7/8) Epoch 20, batch 650, loss[loss=0.1676, simple_loss=0.247, pruned_loss=0.04403, over 12413.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2516, pruned_loss=0.04122, over 3192316.36 frames. ], batch size: 246, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:30:39,969 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193515.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 00:31:07,550 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193535.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:31:11,812 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4506, 3.4913, 4.0316, 1.9937, 3.2872, 2.5009, 3.9116, 3.6514], device='cuda:7'), covar=tensor([0.0262, 0.0978, 0.0464, 0.2118, 0.0768, 0.0946, 0.0581, 0.1160], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0158, 0.0165, 0.0151, 0.0144, 0.0127, 0.0142, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:31:29,638 INFO [train.py:904] (7/8) Epoch 20, batch 700, loss[loss=0.1569, simple_loss=0.2342, pruned_loss=0.03978, over 15875.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2509, pruned_loss=0.04078, over 3225336.79 frames. ], batch size: 35, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:31:48,975 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193567.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:31:57,463 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.109e+02 2.476e+02 3.086e+02 5.855e+02, threshold=4.951e+02, percent-clipped=1.0 2023-05-01 00:32:35,686 INFO [train.py:904] (7/8) Epoch 20, batch 750, loss[loss=0.1408, simple_loss=0.233, pruned_loss=0.02435, over 17251.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2515, pruned_loss=0.04081, over 3255379.64 frames. ], batch size: 45, lr: 3.42e-03, grad_scale: 2.0 2023-05-01 00:32:54,658 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:33:04,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2195, 3.3850, 3.6133, 2.1156, 3.0146, 2.5474, 3.6713, 3.6236], device='cuda:7'), covar=tensor([0.0271, 0.0946, 0.0624, 0.2042, 0.0917, 0.0946, 0.0603, 0.1170], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0158, 0.0165, 0.0151, 0.0144, 0.0127, 0.0143, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:33:28,508 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:33:42,968 INFO [train.py:904] (7/8) Epoch 20, batch 800, loss[loss=0.1587, simple_loss=0.2398, pruned_loss=0.03878, over 16818.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2515, pruned_loss=0.04055, over 3277184.33 frames. ], batch size: 102, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:33:53,258 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-01 00:34:05,419 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8004, 4.2482, 4.2936, 2.9881, 3.5611, 4.2057, 3.8564, 2.3513], device='cuda:7'), covar=tensor([0.0535, 0.0076, 0.0050, 0.0409, 0.0155, 0.0117, 0.0101, 0.0535], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0082, 0.0081, 0.0135, 0.0096, 0.0107, 0.0093, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 00:34:10,663 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.187e+02 2.610e+02 3.095e+02 1.121e+03, threshold=5.220e+02, percent-clipped=2.0 2023-05-01 00:34:16,588 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193676.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:34:45,116 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193697.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:34:52,182 INFO [train.py:904] (7/8) Epoch 20, batch 850, loss[loss=0.1446, simple_loss=0.2287, pruned_loss=0.03022, over 16987.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2516, pruned_loss=0.04053, over 3291745.47 frames. ], batch size: 41, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:34:57,954 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8545, 4.3166, 3.2084, 2.3343, 2.8073, 2.6368, 4.7432, 3.7539], device='cuda:7'), covar=tensor([0.2759, 0.0610, 0.1641, 0.2882, 0.2782, 0.1999, 0.0315, 0.1285], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0264, 0.0299, 0.0302, 0.0289, 0.0251, 0.0286, 0.0329], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 00:35:02,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9187, 4.4112, 3.2308, 2.4266, 2.8711, 2.6393, 4.7902, 3.7845], device='cuda:7'), covar=tensor([0.2726, 0.0618, 0.1688, 0.2772, 0.2771, 0.2018, 0.0340, 0.1312], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0264, 0.0298, 0.0302, 0.0289, 0.0251, 0.0286, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 00:35:51,453 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193745.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:35:59,341 INFO [train.py:904] (7/8) Epoch 20, batch 900, loss[loss=0.1538, simple_loss=0.2311, pruned_loss=0.03826, over 16519.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2507, pruned_loss=0.03976, over 3299482.11 frames. ], batch size: 75, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:36:28,230 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.015e+02 2.385e+02 2.719e+02 5.424e+02, threshold=4.769e+02, percent-clipped=3.0 2023-05-01 00:36:33,752 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193777.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:36:43,948 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6189, 4.4478, 4.6627, 4.8525, 4.9458, 4.5067, 4.9579, 4.9442], device='cuda:7'), covar=tensor([0.1929, 0.1627, 0.1975, 0.1023, 0.0840, 0.1159, 0.1178, 0.1234], device='cuda:7'), in_proj_covar=tensor([0.0634, 0.0788, 0.0920, 0.0803, 0.0600, 0.0626, 0.0650, 0.0751], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:37:09,306 INFO [train.py:904] (7/8) Epoch 20, batch 950, loss[loss=0.1515, simple_loss=0.2325, pruned_loss=0.03524, over 12174.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2505, pruned_loss=0.04062, over 3289196.52 frames. ], batch size: 246, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:37:20,552 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193810.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:37:21,732 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4396, 4.3521, 4.3051, 3.9873, 4.0742, 4.3617, 4.1384, 4.1165], device='cuda:7'), covar=tensor([0.0673, 0.0777, 0.0337, 0.0334, 0.0840, 0.0559, 0.0632, 0.0685], device='cuda:7'), in_proj_covar=tensor([0.0294, 0.0420, 0.0340, 0.0334, 0.0351, 0.0391, 0.0236, 0.0408], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:38:03,677 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.29 vs. limit=5.0 2023-05-01 00:38:09,572 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2462, 5.2119, 5.7109, 5.6969, 5.7127, 5.3677, 5.2962, 5.0885], device='cuda:7'), covar=tensor([0.0332, 0.0608, 0.0350, 0.0407, 0.0486, 0.0385, 0.0938, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0439, 0.0425, 0.0398, 0.0475, 0.0447, 0.0539, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 00:38:17,887 INFO [train.py:904] (7/8) Epoch 20, batch 1000, loss[loss=0.1544, simple_loss=0.2291, pruned_loss=0.03983, over 16244.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2495, pruned_loss=0.04036, over 3295943.70 frames. ], batch size: 165, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:38:39,300 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193867.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:38:45,944 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.639e+02 2.059e+02 2.451e+02 3.046e+02 5.383e+02, threshold=4.901e+02, percent-clipped=2.0 2023-05-01 00:39:00,672 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=193884.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:39:24,781 INFO [train.py:904] (7/8) Epoch 20, batch 1050, loss[loss=0.1616, simple_loss=0.2442, pruned_loss=0.03952, over 16566.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2489, pruned_loss=0.04031, over 3305894.78 frames. ], batch size: 75, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:39:43,721 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193915.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:40:18,192 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=193940.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:40:25,216 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=193945.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:40:35,458 INFO [train.py:904] (7/8) Epoch 20, batch 1100, loss[loss=0.1475, simple_loss=0.2324, pruned_loss=0.03124, over 17023.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2488, pruned_loss=0.03935, over 3316845.19 frames. ], batch size: 41, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:41:01,752 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=193971.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:41:04,221 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.207e+02 2.619e+02 3.357e+02 2.000e+03, threshold=5.237e+02, percent-clipped=3.0 2023-05-01 00:41:25,202 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=193988.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:41:46,950 INFO [train.py:904] (7/8) Epoch 20, batch 1150, loss[loss=0.1774, simple_loss=0.2501, pruned_loss=0.05235, over 16873.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2486, pruned_loss=0.03921, over 3322542.87 frames. ], batch size: 116, lr: 3.42e-03, grad_scale: 4.0 2023-05-01 00:42:23,256 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 00:42:56,055 INFO [train.py:904] (7/8) Epoch 20, batch 1200, loss[loss=0.1648, simple_loss=0.2451, pruned_loss=0.04222, over 16798.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2478, pruned_loss=0.0392, over 3316104.21 frames. ], batch size: 102, lr: 3.42e-03, grad_scale: 8.0 2023-05-01 00:43:25,045 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194072.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:43:25,796 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.319e+02 2.146e+02 2.588e+02 3.131e+02 5.620e+02, threshold=5.177e+02, percent-clipped=2.0 2023-05-01 00:43:33,176 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194077.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 00:43:46,153 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 00:44:06,935 INFO [train.py:904] (7/8) Epoch 20, batch 1250, loss[loss=0.166, simple_loss=0.2477, pruned_loss=0.0422, over 16575.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2483, pruned_loss=0.03964, over 3317797.25 frames. ], batch size: 146, lr: 3.42e-03, grad_scale: 8.0 2023-05-01 00:44:17,634 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194110.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:44:27,010 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0349, 4.0016, 3.9908, 3.3594, 3.9713, 1.8333, 3.7799, 3.5107], device='cuda:7'), covar=tensor([0.0133, 0.0124, 0.0169, 0.0262, 0.0096, 0.2539, 0.0132, 0.0236], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0151, 0.0196, 0.0174, 0.0172, 0.0207, 0.0185, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:44:38,203 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194125.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:44:49,867 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194133.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:45:02,350 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 00:45:15,925 INFO [train.py:904] (7/8) Epoch 20, batch 1300, loss[loss=0.159, simple_loss=0.2452, pruned_loss=0.03642, over 17245.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2487, pruned_loss=0.03978, over 3320804.51 frames. ], batch size: 44, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 00:45:26,576 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194158.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:45:46,649 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.297e+02 2.207e+02 2.541e+02 3.000e+02 4.873e+02, threshold=5.083e+02, percent-clipped=0.0 2023-05-01 00:46:27,325 INFO [train.py:904] (7/8) Epoch 20, batch 1350, loss[loss=0.1442, simple_loss=0.2344, pruned_loss=0.02702, over 17199.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2482, pruned_loss=0.03961, over 3319201.71 frames. ], batch size: 44, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:47:21,268 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194240.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:47:26,337 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7443, 2.6421, 2.5169, 4.0767, 3.3918, 4.0750, 1.5294, 2.8926], device='cuda:7'), covar=tensor([0.1394, 0.0688, 0.1144, 0.0194, 0.0152, 0.0337, 0.1535, 0.0855], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0173, 0.0192, 0.0184, 0.0201, 0.0214, 0.0198, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:47:31,557 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0937, 3.1702, 3.2608, 2.1754, 2.8861, 2.2751, 3.6621, 3.5198], device='cuda:7'), covar=tensor([0.0235, 0.0913, 0.0672, 0.1781, 0.0829, 0.1055, 0.0462, 0.0888], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0161, 0.0167, 0.0153, 0.0145, 0.0129, 0.0145, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:47:36,942 INFO [train.py:904] (7/8) Epoch 20, batch 1400, loss[loss=0.1818, simple_loss=0.2583, pruned_loss=0.05261, over 16741.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.249, pruned_loss=0.04011, over 3321889.15 frames. ], batch size: 124, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:48:03,422 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194271.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:48:06,434 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.050e+02 2.370e+02 3.098e+02 5.438e+02, threshold=4.739e+02, percent-clipped=2.0 2023-05-01 00:48:44,308 INFO [train.py:904] (7/8) Epoch 20, batch 1450, loss[loss=0.166, simple_loss=0.2535, pruned_loss=0.03928, over 17004.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2488, pruned_loss=0.03985, over 3327791.04 frames. ], batch size: 53, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:49:06,939 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194318.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:49:08,665 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194319.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:49:27,697 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9823, 5.0575, 5.4841, 5.4680, 5.4598, 5.1379, 5.0624, 4.8913], device='cuda:7'), covar=tensor([0.0348, 0.0621, 0.0434, 0.0470, 0.0495, 0.0412, 0.0920, 0.0456], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0444, 0.0429, 0.0402, 0.0479, 0.0452, 0.0545, 0.0364], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 00:49:47,642 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2023-05-01 00:49:54,002 INFO [train.py:904] (7/8) Epoch 20, batch 1500, loss[loss=0.1741, simple_loss=0.2522, pruned_loss=0.04803, over 16371.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2484, pruned_loss=0.03976, over 3328188.45 frames. ], batch size: 165, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:49:55,473 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194353.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:50:07,978 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194362.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:50:24,393 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.182e+02 2.566e+02 3.379e+02 8.536e+02, threshold=5.133e+02, percent-clipped=4.0 2023-05-01 00:50:31,247 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194379.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:50:41,768 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-01 00:51:01,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 00:51:03,657 INFO [train.py:904] (7/8) Epoch 20, batch 1550, loss[loss=0.1723, simple_loss=0.2796, pruned_loss=0.03246, over 17253.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2502, pruned_loss=0.04133, over 3325262.30 frames. ], batch size: 52, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:51:20,962 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:51:34,175 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194423.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 00:51:40,653 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194428.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:52:13,194 INFO [train.py:904] (7/8) Epoch 20, batch 1600, loss[loss=0.1966, simple_loss=0.2879, pruned_loss=0.05268, over 16645.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2517, pruned_loss=0.04156, over 3327358.19 frames. ], batch size: 57, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 00:52:43,866 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.299e+02 2.983e+02 3.595e+02 1.383e+03, threshold=5.966e+02, percent-clipped=6.0 2023-05-01 00:53:22,709 INFO [train.py:904] (7/8) Epoch 20, batch 1650, loss[loss=0.1617, simple_loss=0.2504, pruned_loss=0.03647, over 16846.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2526, pruned_loss=0.04199, over 3323567.35 frames. ], batch size: 42, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 00:53:47,449 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6653, 2.6963, 2.5101, 4.1289, 3.3389, 4.0608, 1.5821, 2.7209], device='cuda:7'), covar=tensor([0.1616, 0.0740, 0.1236, 0.0223, 0.0206, 0.0406, 0.1790, 0.0993], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0173, 0.0191, 0.0184, 0.0201, 0.0213, 0.0197, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 00:53:58,353 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 00:54:16,618 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194540.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:54:18,800 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5495, 5.9281, 5.6737, 5.7864, 5.3800, 5.2949, 5.3190, 6.0790], device='cuda:7'), covar=tensor([0.1325, 0.1052, 0.1073, 0.0844, 0.0922, 0.0754, 0.1202, 0.0983], device='cuda:7'), in_proj_covar=tensor([0.0670, 0.0826, 0.0674, 0.0615, 0.0517, 0.0523, 0.0691, 0.0632], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:54:33,475 INFO [train.py:904] (7/8) Epoch 20, batch 1700, loss[loss=0.1775, simple_loss=0.26, pruned_loss=0.04752, over 16758.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2546, pruned_loss=0.04258, over 3330149.93 frames. ], batch size: 89, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:55:05,703 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.259e+02 2.640e+02 3.354e+02 1.280e+03, threshold=5.281e+02, percent-clipped=2.0 2023-05-01 00:55:15,538 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.16 vs. limit=5.0 2023-05-01 00:55:25,665 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194588.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:55:28,423 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 00:55:43,642 INFO [train.py:904] (7/8) Epoch 20, batch 1750, loss[loss=0.1735, simple_loss=0.2678, pruned_loss=0.03964, over 16688.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2556, pruned_loss=0.04231, over 3338083.72 frames. ], batch size: 57, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:55:47,485 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2408, 5.2163, 4.9524, 4.4735, 5.1072, 1.6196, 4.8212, 4.8874], device='cuda:7'), covar=tensor([0.0090, 0.0082, 0.0232, 0.0423, 0.0118, 0.3204, 0.0150, 0.0222], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0152, 0.0197, 0.0177, 0.0174, 0.0208, 0.0187, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:55:56,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5178, 2.3156, 2.2147, 4.3309, 2.2005, 2.7319, 2.3713, 2.4894], device='cuda:7'), covar=tensor([0.1133, 0.3619, 0.3148, 0.0483, 0.4319, 0.2530, 0.3529, 0.3527], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0439, 0.0364, 0.0326, 0.0434, 0.0505, 0.0409, 0.0515], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 00:55:56,610 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 00:56:06,026 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8947, 4.4298, 3.1955, 2.2700, 2.7169, 2.6060, 4.8003, 3.6548], device='cuda:7'), covar=tensor([0.2883, 0.0584, 0.1743, 0.2926, 0.2868, 0.2129, 0.0373, 0.1443], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0266, 0.0300, 0.0304, 0.0291, 0.0252, 0.0287, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 00:56:27,143 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194634.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:56:48,923 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 00:56:51,755 INFO [train.py:904] (7/8) Epoch 20, batch 1800, loss[loss=0.1539, simple_loss=0.2445, pruned_loss=0.03166, over 17108.00 frames. ], tot_loss[loss=0.17, simple_loss=0.256, pruned_loss=0.04203, over 3341322.64 frames. ], batch size: 49, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:57:14,407 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 00:57:22,533 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194674.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:57:23,427 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.398e+02 2.901e+02 3.350e+02 9.637e+02, threshold=5.801e+02, percent-clipped=10.0 2023-05-01 00:57:35,443 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-05-01 00:57:50,488 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194695.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:57:58,022 INFO [train.py:904] (7/8) Epoch 20, batch 1850, loss[loss=0.1485, simple_loss=0.2415, pruned_loss=0.0278, over 17213.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2565, pruned_loss=0.04167, over 3338582.02 frames. ], batch size: 44, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:58:08,682 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194709.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:58:19,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 00:58:21,448 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194718.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 00:58:34,683 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194728.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 00:59:06,827 INFO [train.py:904] (7/8) Epoch 20, batch 1900, loss[loss=0.1743, simple_loss=0.2561, pruned_loss=0.04628, over 16159.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2552, pruned_loss=0.0408, over 3343506.60 frames. ], batch size: 164, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 00:59:24,219 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 00:59:33,825 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 00:59:33,919 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 00:59:38,384 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.030e+02 2.433e+02 3.010e+02 6.888e+02, threshold=4.866e+02, percent-clipped=1.0 2023-05-01 00:59:41,140 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=194776.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:00:16,333 INFO [train.py:904] (7/8) Epoch 20, batch 1950, loss[loss=0.181, simple_loss=0.2685, pruned_loss=0.04674, over 15650.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2551, pruned_loss=0.04097, over 3339277.79 frames. ], batch size: 191, lr: 3.41e-03, grad_scale: 4.0 2023-05-01 01:00:17,289 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-05-01 01:01:17,114 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=194847.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:01:23,590 INFO [train.py:904] (7/8) Epoch 20, batch 2000, loss[loss=0.1944, simple_loss=0.2652, pruned_loss=0.06179, over 16871.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.256, pruned_loss=0.04124, over 3321811.36 frames. ], batch size: 96, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:01:54,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.120e+02 2.565e+02 3.052e+02 4.628e+02, threshold=5.130e+02, percent-clipped=0.0 2023-05-01 01:02:14,998 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2882, 2.3013, 2.3678, 4.1171, 2.2807, 2.7042, 2.3079, 2.4904], device='cuda:7'), covar=tensor([0.1401, 0.3642, 0.2751, 0.0553, 0.3883, 0.2382, 0.3762, 0.2976], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0439, 0.0363, 0.0325, 0.0433, 0.0505, 0.0408, 0.0514], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:02:32,415 INFO [train.py:904] (7/8) Epoch 20, batch 2050, loss[loss=0.1888, simple_loss=0.2736, pruned_loss=0.05201, over 16771.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2564, pruned_loss=0.04194, over 3321270.34 frames. ], batch size: 57, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:02:41,407 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=194908.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:03:15,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9708, 4.3006, 4.3795, 3.5184, 3.6321, 4.3503, 3.8714, 2.6990], device='cuda:7'), covar=tensor([0.0404, 0.0062, 0.0039, 0.0251, 0.0125, 0.0083, 0.0087, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0082, 0.0081, 0.0134, 0.0096, 0.0108, 0.0093, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 01:03:41,680 INFO [train.py:904] (7/8) Epoch 20, batch 2100, loss[loss=0.1469, simple_loss=0.2379, pruned_loss=0.02794, over 17231.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2577, pruned_loss=0.04276, over 3321071.50 frames. ], batch size: 45, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:04:12,827 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=194974.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:04:13,685 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.067e+02 2.444e+02 3.012e+02 5.275e+02, threshold=4.887e+02, percent-clipped=1.0 2023-05-01 01:04:34,697 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=194990.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:04:50,822 INFO [train.py:904] (7/8) Epoch 20, batch 2150, loss[loss=0.2133, simple_loss=0.2876, pruned_loss=0.06954, over 11766.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2591, pruned_loss=0.04374, over 3315122.36 frames. ], batch size: 247, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:05:01,485 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195009.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:05:02,895 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.06 vs. limit=5.0 2023-05-01 01:05:04,328 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195011.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:05:13,342 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195018.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:05:19,274 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195022.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:05:43,323 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 01:05:50,360 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195044.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:06:01,916 INFO [train.py:904] (7/8) Epoch 20, batch 2200, loss[loss=0.1936, simple_loss=0.2707, pruned_loss=0.05826, over 16638.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.26, pruned_loss=0.04425, over 3311726.60 frames. ], batch size: 89, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:06:09,211 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195057.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:06:09,324 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9201, 5.0273, 5.4300, 5.4015, 5.4034, 5.0708, 5.0236, 4.7861], device='cuda:7'), covar=tensor([0.0350, 0.0551, 0.0373, 0.0451, 0.0491, 0.0374, 0.0933, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0410, 0.0452, 0.0435, 0.0410, 0.0485, 0.0462, 0.0551, 0.0370], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 01:06:21,179 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195066.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:06:28,940 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195071.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:06:30,134 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195072.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:06:33,766 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.892e+02 2.410e+02 2.761e+02 3.300e+02 5.006e+02, threshold=5.521e+02, percent-clipped=1.0 2023-05-01 01:07:10,872 INFO [train.py:904] (7/8) Epoch 20, batch 2250, loss[loss=0.192, simple_loss=0.2666, pruned_loss=0.05865, over 16724.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2597, pruned_loss=0.04387, over 3311965.83 frames. ], batch size: 134, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:07:15,528 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195105.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:07:54,749 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195132.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:08:21,156 INFO [train.py:904] (7/8) Epoch 20, batch 2300, loss[loss=0.1634, simple_loss=0.2504, pruned_loss=0.03819, over 15861.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2594, pruned_loss=0.04346, over 3320806.20 frames. ], batch size: 35, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:08:51,392 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.280e+02 2.663e+02 3.175e+02 5.300e+02, threshold=5.327e+02, percent-clipped=0.0 2023-05-01 01:09:29,600 INFO [train.py:904] (7/8) Epoch 20, batch 2350, loss[loss=0.1958, simple_loss=0.2852, pruned_loss=0.05324, over 16528.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2602, pruned_loss=0.0438, over 3324746.82 frames. ], batch size: 68, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:09:31,706 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:09:33,433 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 01:09:43,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0450, 4.4576, 4.5045, 3.4382, 3.7831, 4.4651, 3.9819, 2.6269], device='cuda:7'), covar=tensor([0.0413, 0.0062, 0.0039, 0.0285, 0.0132, 0.0092, 0.0079, 0.0433], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0082, 0.0081, 0.0132, 0.0095, 0.0107, 0.0093, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 01:09:57,443 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-01 01:10:36,705 INFO [train.py:904] (7/8) Epoch 20, batch 2400, loss[loss=0.1763, simple_loss=0.2526, pruned_loss=0.04999, over 16880.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2601, pruned_loss=0.04333, over 3322334.36 frames. ], batch size: 109, lr: 3.41e-03, grad_scale: 8.0 2023-05-01 01:11:07,016 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.253e+02 2.639e+02 3.077e+02 5.540e+02, threshold=5.278e+02, percent-clipped=1.0 2023-05-01 01:11:29,095 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195290.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:11:45,821 INFO [train.py:904] (7/8) Epoch 20, batch 2450, loss[loss=0.1878, simple_loss=0.278, pruned_loss=0.04882, over 17101.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.261, pruned_loss=0.0432, over 3335063.70 frames. ], batch size: 55, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:12:35,686 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195338.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:12:54,085 INFO [train.py:904] (7/8) Epoch 20, batch 2500, loss[loss=0.178, simple_loss=0.257, pruned_loss=0.04956, over 16786.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2605, pruned_loss=0.0432, over 3334892.40 frames. ], batch size: 83, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:12:55,492 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-01 01:13:07,759 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-01 01:13:16,677 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195367.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:13:26,928 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.185e+02 2.562e+02 3.023e+02 6.708e+02, threshold=5.124e+02, percent-clipped=4.0 2023-05-01 01:13:51,617 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=195392.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:14:01,563 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195400.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 01:14:04,083 INFO [train.py:904] (7/8) Epoch 20, batch 2550, loss[loss=0.1697, simple_loss=0.2481, pruned_loss=0.04568, over 16851.00 frames. ], tot_loss[loss=0.1749, simple_loss=0.2619, pruned_loss=0.04392, over 3328133.23 frames. ], batch size: 90, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:14:38,142 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195427.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:15:11,783 INFO [train.py:904] (7/8) Epoch 20, batch 2600, loss[loss=0.189, simple_loss=0.2632, pruned_loss=0.05742, over 16870.00 frames. ], tot_loss[loss=0.175, simple_loss=0.2625, pruned_loss=0.04379, over 3317127.61 frames. ], batch size: 116, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:15:13,409 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=195453.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:15:40,264 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3427, 4.6360, 4.4800, 4.4934, 4.2286, 4.1785, 4.2395, 4.6905], device='cuda:7'), covar=tensor([0.1289, 0.0999, 0.1088, 0.0864, 0.0822, 0.1446, 0.1068, 0.0930], device='cuda:7'), in_proj_covar=tensor([0.0661, 0.0817, 0.0665, 0.0610, 0.0511, 0.0515, 0.0681, 0.0624], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:15:42,990 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.224e+02 2.625e+02 3.381e+02 7.141e+02, threshold=5.251e+02, percent-clipped=4.0 2023-05-01 01:16:01,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4373, 4.3599, 4.3418, 4.0743, 4.1458, 4.4288, 4.1453, 4.1773], device='cuda:7'), covar=tensor([0.0689, 0.0745, 0.0281, 0.0273, 0.0700, 0.0490, 0.0642, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0432, 0.0349, 0.0344, 0.0361, 0.0400, 0.0241, 0.0417], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:16:20,747 INFO [train.py:904] (7/8) Epoch 20, batch 2650, loss[loss=0.1794, simple_loss=0.272, pruned_loss=0.04338, over 17012.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2614, pruned_loss=0.04269, over 3326135.55 frames. ], batch size: 55, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:16:22,195 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195503.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:16:24,584 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9812, 1.9828, 2.5681, 2.9032, 2.7759, 3.5316, 2.4746, 3.3858], device='cuda:7'), covar=tensor([0.0243, 0.0517, 0.0330, 0.0317, 0.0325, 0.0159, 0.0420, 0.0186], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0194, 0.0179, 0.0183, 0.0195, 0.0153, 0.0196, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:16:48,638 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4330, 4.4092, 4.8204, 4.7792, 4.8143, 4.4946, 4.4963, 4.3652], device='cuda:7'), covar=tensor([0.0354, 0.0668, 0.0386, 0.0432, 0.0528, 0.0429, 0.0888, 0.0574], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0448, 0.0432, 0.0406, 0.0480, 0.0457, 0.0548, 0.0367], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 01:17:03,631 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 01:17:21,417 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-05-01 01:17:28,766 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:17:29,630 INFO [train.py:904] (7/8) Epoch 20, batch 2700, loss[loss=0.1573, simple_loss=0.2501, pruned_loss=0.0322, over 16827.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2603, pruned_loss=0.04208, over 3321933.90 frames. ], batch size: 39, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:18:00,664 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.029e+02 2.571e+02 3.039e+02 8.808e+02, threshold=5.142e+02, percent-clipped=4.0 2023-05-01 01:18:31,061 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 01:18:39,439 INFO [train.py:904] (7/8) Epoch 20, batch 2750, loss[loss=0.1604, simple_loss=0.2589, pruned_loss=0.03101, over 17128.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2604, pruned_loss=0.04209, over 3321853.24 frames. ], batch size: 49, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:19:40,445 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4781, 5.8715, 5.5994, 5.6808, 5.2850, 5.3432, 5.2792, 6.0119], device='cuda:7'), covar=tensor([0.1362, 0.1024, 0.1088, 0.0906, 0.0876, 0.0678, 0.1203, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0667, 0.0822, 0.0670, 0.0615, 0.0515, 0.0519, 0.0686, 0.0628], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:19:47,490 INFO [train.py:904] (7/8) Epoch 20, batch 2800, loss[loss=0.1717, simple_loss=0.2529, pruned_loss=0.04522, over 16877.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.261, pruned_loss=0.04238, over 3321773.67 frames. ], batch size: 96, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:20:07,201 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195667.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:20:18,621 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 2.150e+02 2.557e+02 3.074e+02 6.713e+02, threshold=5.114e+02, percent-clipped=2.0 2023-05-01 01:20:40,213 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6780, 3.8528, 2.5340, 4.3923, 2.9184, 4.3423, 2.6250, 3.1440], device='cuda:7'), covar=tensor([0.0324, 0.0370, 0.1471, 0.0339, 0.0815, 0.0495, 0.1345, 0.0723], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0179, 0.0195, 0.0165, 0.0178, 0.0219, 0.0203, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:20:52,435 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195700.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 01:20:54,306 INFO [train.py:904] (7/8) Epoch 20, batch 2850, loss[loss=0.1621, simple_loss=0.26, pruned_loss=0.03209, over 17054.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2612, pruned_loss=0.04253, over 3306250.44 frames. ], batch size: 50, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:21:13,265 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195715.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:21:28,566 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-05-01 01:21:29,683 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=195727.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:21:57,089 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195748.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 01:21:57,098 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=195748.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:22:02,487 INFO [train.py:904] (7/8) Epoch 20, batch 2900, loss[loss=0.1992, simple_loss=0.2881, pruned_loss=0.05515, over 17127.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2606, pruned_loss=0.04301, over 3305903.00 frames. ], batch size: 47, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:22:33,093 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.294e+02 2.743e+02 3.539e+02 8.268e+02, threshold=5.486e+02, percent-clipped=5.0 2023-05-01 01:22:34,110 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=195775.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:22:44,781 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7300, 4.1389, 4.2269, 2.9119, 3.5868, 4.2534, 3.8675, 2.4805], device='cuda:7'), covar=tensor([0.0523, 0.0080, 0.0049, 0.0377, 0.0146, 0.0099, 0.0093, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0083, 0.0082, 0.0134, 0.0097, 0.0109, 0.0094, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 01:23:10,994 INFO [train.py:904] (7/8) Epoch 20, batch 2950, loss[loss=0.1977, simple_loss=0.272, pruned_loss=0.06175, over 16676.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2599, pruned_loss=0.04381, over 3308177.90 frames. ], batch size: 134, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:24:18,946 INFO [train.py:904] (7/8) Epoch 20, batch 3000, loss[loss=0.1938, simple_loss=0.2642, pruned_loss=0.06168, over 16864.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2598, pruned_loss=0.04385, over 3308094.25 frames. ], batch size: 116, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:24:18,946 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 01:24:25,474 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8622, 2.6813, 2.6388, 4.0697, 2.9091, 4.0145, 1.5823, 3.0502], device='cuda:7'), covar=tensor([0.1275, 0.0722, 0.1069, 0.0197, 0.0101, 0.0354, 0.1558, 0.0707], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0173, 0.0192, 0.0187, 0.0204, 0.0214, 0.0197, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:24:27,131 INFO [train.py:938] (7/8) Epoch 20, validation: loss=0.1354, simple_loss=0.2409, pruned_loss=0.01492, over 944034.00 frames. 2023-05-01 01:24:27,132 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 01:24:58,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.268e+02 2.640e+02 3.074e+02 4.713e+02, threshold=5.280e+02, percent-clipped=0.0 2023-05-01 01:25:38,249 INFO [train.py:904] (7/8) Epoch 20, batch 3050, loss[loss=0.1515, simple_loss=0.2471, pruned_loss=0.02795, over 17240.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2587, pruned_loss=0.04367, over 3313105.67 frames. ], batch size: 45, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:26:46,792 INFO [train.py:904] (7/8) Epoch 20, batch 3100, loss[loss=0.1676, simple_loss=0.2448, pruned_loss=0.04516, over 15405.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2586, pruned_loss=0.04427, over 3302580.07 frames. ], batch size: 190, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:26:48,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1560, 3.2183, 3.3578, 2.2457, 3.1265, 3.4548, 3.1221, 1.8856], device='cuda:7'), covar=tensor([0.0499, 0.0107, 0.0064, 0.0397, 0.0118, 0.0093, 0.0108, 0.0486], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0083, 0.0083, 0.0135, 0.0097, 0.0109, 0.0094, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 01:27:16,967 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.170e+02 2.517e+02 3.010e+02 4.589e+02, threshold=5.034e+02, percent-clipped=0.0 2023-05-01 01:27:48,232 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-05-01 01:27:55,521 INFO [train.py:904] (7/8) Epoch 20, batch 3150, loss[loss=0.1658, simple_loss=0.2659, pruned_loss=0.03289, over 16680.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2581, pruned_loss=0.04369, over 3305083.15 frames. ], batch size: 57, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:27:56,003 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1615, 4.0234, 4.4300, 2.0934, 4.6982, 4.7165, 3.2748, 3.5579], device='cuda:7'), covar=tensor([0.0710, 0.0241, 0.0214, 0.1224, 0.0064, 0.0145, 0.0457, 0.0435], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0109, 0.0098, 0.0140, 0.0080, 0.0125, 0.0128, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:28:30,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9466, 1.9468, 2.6081, 2.8301, 2.7334, 3.3868, 2.3379, 3.3995], device='cuda:7'), covar=tensor([0.0252, 0.0529, 0.0315, 0.0360, 0.0331, 0.0205, 0.0449, 0.0181], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0195, 0.0179, 0.0183, 0.0196, 0.0154, 0.0195, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:28:57,786 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=196048.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:29:03,225 INFO [train.py:904] (7/8) Epoch 20, batch 3200, loss[loss=0.1812, simple_loss=0.2758, pruned_loss=0.04325, over 17013.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2573, pruned_loss=0.04293, over 3303338.97 frames. ], batch size: 53, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:29:35,521 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.205e+02 2.543e+02 3.042e+02 4.560e+02, threshold=5.087e+02, percent-clipped=0.0 2023-05-01 01:30:04,491 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=196096.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:30:11,579 INFO [train.py:904] (7/8) Epoch 20, batch 3250, loss[loss=0.2028, simple_loss=0.2935, pruned_loss=0.056, over 17053.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2576, pruned_loss=0.04331, over 3310901.72 frames. ], batch size: 50, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:30:29,459 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3234, 5.2597, 5.0596, 4.5668, 5.1276, 2.2245, 4.8847, 5.0665], device='cuda:7'), covar=tensor([0.0124, 0.0103, 0.0239, 0.0458, 0.0134, 0.2507, 0.0163, 0.0212], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0154, 0.0200, 0.0180, 0.0177, 0.0209, 0.0190, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:31:19,983 INFO [train.py:904] (7/8) Epoch 20, batch 3300, loss[loss=0.1787, simple_loss=0.258, pruned_loss=0.04969, over 16744.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2586, pruned_loss=0.0434, over 3312487.08 frames. ], batch size: 83, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:31:52,358 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.123e+02 2.550e+02 3.106e+02 4.546e+02, threshold=5.100e+02, percent-clipped=0.0 2023-05-01 01:32:27,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7755, 4.2758, 3.0455, 2.2586, 2.7739, 2.6066, 4.6072, 3.6394], device='cuda:7'), covar=tensor([0.2878, 0.0601, 0.1712, 0.2779, 0.2521, 0.1938, 0.0362, 0.1234], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0270, 0.0303, 0.0307, 0.0297, 0.0255, 0.0292, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 01:32:28,250 INFO [train.py:904] (7/8) Epoch 20, batch 3350, loss[loss=0.1692, simple_loss=0.2642, pruned_loss=0.03713, over 17063.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2589, pruned_loss=0.04317, over 3318518.14 frames. ], batch size: 49, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:33:22,570 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8909, 1.9537, 2.5185, 2.8391, 2.7702, 3.1728, 2.1662, 3.2083], device='cuda:7'), covar=tensor([0.0218, 0.0499, 0.0334, 0.0282, 0.0310, 0.0208, 0.0501, 0.0162], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0194, 0.0178, 0.0183, 0.0196, 0.0154, 0.0195, 0.0149], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:33:35,777 INFO [train.py:904] (7/8) Epoch 20, batch 3400, loss[loss=0.169, simple_loss=0.2621, pruned_loss=0.038, over 17094.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.259, pruned_loss=0.04306, over 3324247.00 frames. ], batch size: 55, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:34:06,840 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 2.175e+02 2.675e+02 3.072e+02 5.322e+02, threshold=5.351e+02, percent-clipped=1.0 2023-05-01 01:34:44,341 INFO [train.py:904] (7/8) Epoch 20, batch 3450, loss[loss=0.1596, simple_loss=0.2403, pruned_loss=0.03939, over 16847.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2581, pruned_loss=0.04296, over 3319228.05 frames. ], batch size: 102, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:35:10,335 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196321.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:35:50,473 INFO [train.py:904] (7/8) Epoch 20, batch 3500, loss[loss=0.1814, simple_loss=0.2542, pruned_loss=0.05425, over 16887.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2563, pruned_loss=0.04227, over 3324782.50 frames. ], batch size: 109, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:35:55,616 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6355, 3.7799, 4.2545, 2.2212, 3.3909, 2.5399, 4.1527, 3.9705], device='cuda:7'), covar=tensor([0.0249, 0.0866, 0.0406, 0.2074, 0.0721, 0.0983, 0.0509, 0.1040], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0164, 0.0166, 0.0151, 0.0144, 0.0129, 0.0144, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:36:02,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2139, 4.0352, 4.2514, 4.3703, 4.4843, 4.0472, 4.2517, 4.4578], device='cuda:7'), covar=tensor([0.1497, 0.1156, 0.1213, 0.0710, 0.0582, 0.1287, 0.2079, 0.0762], device='cuda:7'), in_proj_covar=tensor([0.0665, 0.0823, 0.0962, 0.0844, 0.0628, 0.0657, 0.0673, 0.0781], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:36:04,140 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3779, 4.1433, 4.4498, 2.5178, 4.8004, 4.8157, 3.5052, 3.7595], device='cuda:7'), covar=tensor([0.0590, 0.0218, 0.0238, 0.0970, 0.0068, 0.0148, 0.0387, 0.0366], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0108, 0.0098, 0.0139, 0.0079, 0.0125, 0.0127, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:36:04,159 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196360.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:36:23,665 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.052e+02 2.465e+02 2.837e+02 5.038e+02, threshold=4.930e+02, percent-clipped=0.0 2023-05-01 01:36:31,287 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 01:36:35,851 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196382.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:37:01,682 INFO [train.py:904] (7/8) Epoch 20, batch 3550, loss[loss=0.1591, simple_loss=0.2505, pruned_loss=0.03381, over 16681.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2551, pruned_loss=0.0421, over 3309516.76 frames. ], batch size: 37, lr: 3.40e-03, grad_scale: 8.0 2023-05-01 01:37:09,754 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4193, 3.8695, 4.1398, 2.3463, 3.3103, 2.7987, 3.8182, 4.0138], device='cuda:7'), covar=tensor([0.0384, 0.0914, 0.0445, 0.1897, 0.0814, 0.0876, 0.0836, 0.1000], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0164, 0.0166, 0.0151, 0.0143, 0.0129, 0.0144, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:37:28,367 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196421.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:37:45,338 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7083, 1.8357, 2.3405, 2.5367, 2.6579, 2.7127, 1.6885, 2.7968], device='cuda:7'), covar=tensor([0.0156, 0.0478, 0.0297, 0.0286, 0.0245, 0.0232, 0.0609, 0.0155], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0193, 0.0178, 0.0182, 0.0196, 0.0154, 0.0194, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:38:10,222 INFO [train.py:904] (7/8) Epoch 20, batch 3600, loss[loss=0.1653, simple_loss=0.2512, pruned_loss=0.03968, over 15398.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2536, pruned_loss=0.04168, over 3323919.37 frames. ], batch size: 190, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 01:38:41,984 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.224e+02 2.585e+02 3.008e+02 4.978e+02, threshold=5.169e+02, percent-clipped=1.0 2023-05-01 01:39:20,691 INFO [train.py:904] (7/8) Epoch 20, batch 3650, loss[loss=0.1623, simple_loss=0.2578, pruned_loss=0.03338, over 17049.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2527, pruned_loss=0.04193, over 3312467.83 frames. ], batch size: 55, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 01:40:32,653 INFO [train.py:904] (7/8) Epoch 20, batch 3700, loss[loss=0.1662, simple_loss=0.2451, pruned_loss=0.0437, over 16793.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.252, pruned_loss=0.04341, over 3281494.02 frames. ], batch size: 124, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:40:33,918 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1135, 3.2417, 3.2945, 2.2278, 2.8819, 2.3865, 3.6859, 3.6147], device='cuda:7'), covar=tensor([0.0250, 0.0863, 0.0655, 0.1787, 0.0827, 0.0986, 0.0441, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0164, 0.0165, 0.0151, 0.0143, 0.0128, 0.0144, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:41:07,143 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.139e+02 2.467e+02 2.956e+02 4.680e+02, threshold=4.935e+02, percent-clipped=0.0 2023-05-01 01:41:11,783 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196578.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:41:47,079 INFO [train.py:904] (7/8) Epoch 20, batch 3750, loss[loss=0.1698, simple_loss=0.2398, pruned_loss=0.04988, over 16909.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2531, pruned_loss=0.04504, over 3264872.13 frames. ], batch size: 109, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:41:47,645 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4888, 3.7138, 3.8607, 2.7717, 3.5124, 3.9488, 3.6619, 2.2102], device='cuda:7'), covar=tensor([0.0472, 0.0110, 0.0048, 0.0330, 0.0094, 0.0082, 0.0084, 0.0444], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0082, 0.0081, 0.0133, 0.0096, 0.0107, 0.0093, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 01:42:39,579 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:42:58,325 INFO [train.py:904] (7/8) Epoch 20, batch 3800, loss[loss=0.185, simple_loss=0.259, pruned_loss=0.05554, over 16238.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2542, pruned_loss=0.04616, over 3268100.52 frames. ], batch size: 165, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:43:31,148 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.679e+02 2.212e+02 2.504e+02 2.913e+02 5.553e+02, threshold=5.009e+02, percent-clipped=1.0 2023-05-01 01:43:33,477 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6067, 4.4802, 4.6400, 4.8095, 4.9386, 4.4916, 4.7661, 4.9242], device='cuda:7'), covar=tensor([0.1661, 0.1188, 0.1419, 0.0737, 0.0612, 0.0962, 0.1517, 0.0734], device='cuda:7'), in_proj_covar=tensor([0.0654, 0.0813, 0.0947, 0.0829, 0.0621, 0.0646, 0.0666, 0.0772], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:43:35,323 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=196677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:44:10,776 INFO [train.py:904] (7/8) Epoch 20, batch 3850, loss[loss=0.1768, simple_loss=0.2451, pruned_loss=0.05425, over 16809.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2543, pruned_loss=0.0468, over 3277983.23 frames. ], batch size: 116, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:44:32,375 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=196716.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:44:42,552 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5266, 3.5198, 3.7677, 2.0497, 3.1447, 2.5937, 3.9623, 3.9561], device='cuda:7'), covar=tensor([0.0200, 0.0775, 0.0559, 0.2027, 0.0803, 0.0849, 0.0441, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0163, 0.0165, 0.0150, 0.0143, 0.0128, 0.0143, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:45:24,176 INFO [train.py:904] (7/8) Epoch 20, batch 3900, loss[loss=0.1799, simple_loss=0.2567, pruned_loss=0.05159, over 16471.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.254, pruned_loss=0.04725, over 3275753.47 frames. ], batch size: 146, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:45:57,455 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 2.027e+02 2.552e+02 2.919e+02 4.484e+02, threshold=5.103e+02, percent-clipped=0.0 2023-05-01 01:46:04,866 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196779.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:46:36,466 INFO [train.py:904] (7/8) Epoch 20, batch 3950, loss[loss=0.1755, simple_loss=0.2612, pruned_loss=0.04493, over 17055.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2539, pruned_loss=0.0477, over 3277959.15 frames. ], batch size: 55, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:46:55,807 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9421, 5.2763, 5.0693, 5.0152, 4.7905, 4.6890, 4.6988, 5.3845], device='cuda:7'), covar=tensor([0.1246, 0.0844, 0.0937, 0.0874, 0.0804, 0.1039, 0.1187, 0.0807], device='cuda:7'), in_proj_covar=tensor([0.0674, 0.0828, 0.0678, 0.0620, 0.0522, 0.0526, 0.0696, 0.0639], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:47:32,136 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196840.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:47:39,191 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7489, 3.7263, 1.9783, 4.0515, 2.8797, 4.0349, 2.1318, 2.9198], device='cuda:7'), covar=tensor([0.0231, 0.0351, 0.1834, 0.0238, 0.0693, 0.0474, 0.1764, 0.0722], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0178, 0.0195, 0.0164, 0.0177, 0.0219, 0.0203, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:47:48,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4432, 2.7993, 3.0236, 1.9543, 2.6989, 2.0993, 3.0714, 3.1055], device='cuda:7'), covar=tensor([0.0262, 0.0869, 0.0621, 0.1947, 0.0836, 0.1022, 0.0574, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0164, 0.0165, 0.0151, 0.0143, 0.0128, 0.0144, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 01:47:48,969 INFO [train.py:904] (7/8) Epoch 20, batch 4000, loss[loss=0.171, simple_loss=0.2595, pruned_loss=0.04122, over 15492.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2537, pruned_loss=0.04794, over 3272549.55 frames. ], batch size: 190, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:47:52,392 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196854.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:48:21,948 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.289e+02 2.579e+02 3.040e+02 5.791e+02, threshold=5.158e+02, percent-clipped=1.0 2023-05-01 01:48:32,440 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=196881.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:49:01,039 INFO [train.py:904] (7/8) Epoch 20, batch 4050, loss[loss=0.1968, simple_loss=0.2834, pruned_loss=0.05512, over 16480.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2545, pruned_loss=0.04758, over 3265576.92 frames. ], batch size: 75, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:49:19,926 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196915.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:49:47,745 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=196934.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:49:53,173 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-05-01 01:49:57,986 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=196942.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:50:07,133 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-01 01:50:12,378 INFO [train.py:904] (7/8) Epoch 20, batch 4100, loss[loss=0.2072, simple_loss=0.2953, pruned_loss=0.05952, over 15148.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2563, pruned_loss=0.04719, over 3262467.24 frames. ], batch size: 190, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:50:34,522 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9822, 5.3576, 5.5862, 5.2888, 5.4053, 5.9636, 5.4733, 5.1790], device='cuda:7'), covar=tensor([0.0933, 0.1639, 0.1784, 0.1818, 0.2179, 0.0791, 0.1376, 0.2210], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0600, 0.0658, 0.0502, 0.0664, 0.0693, 0.0513, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 01:50:41,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6185, 5.6318, 5.3737, 4.7816, 5.5966, 2.3428, 5.3163, 5.1296], device='cuda:7'), covar=tensor([0.0045, 0.0036, 0.0138, 0.0308, 0.0049, 0.2419, 0.0078, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0151, 0.0197, 0.0178, 0.0174, 0.0206, 0.0188, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:50:44,594 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 1.746e+02 2.102e+02 2.519e+02 4.672e+02, threshold=4.204e+02, percent-clipped=0.0 2023-05-01 01:50:48,529 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=196977.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:51:13,791 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6094, 4.6824, 4.8636, 4.6558, 4.6932, 5.2684, 4.7602, 4.4795], device='cuda:7'), covar=tensor([0.1252, 0.1869, 0.1954, 0.1961, 0.2603, 0.0953, 0.1502, 0.2341], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0601, 0.0658, 0.0503, 0.0664, 0.0694, 0.0513, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 01:51:23,809 INFO [train.py:904] (7/8) Epoch 20, batch 4150, loss[loss=0.2672, simple_loss=0.3252, pruned_loss=0.1046, over 11504.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2629, pruned_loss=0.04898, over 3249426.79 frames. ], batch size: 250, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:51:45,687 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197016.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:52:00,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197025.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:52:40,727 INFO [train.py:904] (7/8) Epoch 20, batch 4200, loss[loss=0.1891, simple_loss=0.279, pruned_loss=0.04962, over 16452.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2697, pruned_loss=0.05038, over 3222139.67 frames. ], batch size: 68, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:52:58,851 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197064.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:53:12,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.329e+02 2.798e+02 3.249e+02 9.837e+02, threshold=5.595e+02, percent-clipped=2.0 2023-05-01 01:53:51,305 INFO [train.py:904] (7/8) Epoch 20, batch 4250, loss[loss=0.202, simple_loss=0.289, pruned_loss=0.05748, over 16578.00 frames. ], tot_loss[loss=0.1869, simple_loss=0.2733, pruned_loss=0.0502, over 3220532.92 frames. ], batch size: 57, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:54:40,511 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197135.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:54:56,514 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7736, 1.3228, 1.6589, 1.6732, 1.8064, 1.9233, 1.6054, 1.7692], device='cuda:7'), covar=tensor([0.0237, 0.0400, 0.0237, 0.0292, 0.0262, 0.0186, 0.0422, 0.0142], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0193, 0.0178, 0.0182, 0.0195, 0.0152, 0.0194, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:55:04,303 INFO [train.py:904] (7/8) Epoch 20, batch 4300, loss[loss=0.1951, simple_loss=0.2856, pruned_loss=0.0523, over 16856.00 frames. ], tot_loss[loss=0.1865, simple_loss=0.2745, pruned_loss=0.04924, over 3211866.47 frames. ], batch size: 116, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:55:05,874 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 01:55:32,471 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3826, 5.3662, 5.0569, 4.4772, 5.3146, 1.7564, 5.0024, 4.9067], device='cuda:7'), covar=tensor([0.0061, 0.0056, 0.0175, 0.0313, 0.0063, 0.2856, 0.0092, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0151, 0.0196, 0.0177, 0.0173, 0.0206, 0.0187, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 01:55:37,987 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.176e+02 2.515e+02 3.039e+02 5.119e+02, threshold=5.030e+02, percent-clipped=0.0 2023-05-01 01:56:18,919 INFO [train.py:904] (7/8) Epoch 20, batch 4350, loss[loss=0.1915, simple_loss=0.2801, pruned_loss=0.05144, over 16728.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2782, pruned_loss=0.05073, over 3200846.70 frames. ], batch size: 89, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:56:29,943 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197210.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:57:05,390 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197234.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:57:09,110 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197237.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:57:31,883 INFO [train.py:904] (7/8) Epoch 20, batch 4400, loss[loss=0.1879, simple_loss=0.2742, pruned_loss=0.05075, over 16829.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.28, pruned_loss=0.05189, over 3194194.02 frames. ], batch size: 39, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:58:04,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.492e+02 2.097e+02 2.441e+02 2.762e+02 5.334e+02, threshold=4.883e+02, percent-clipped=1.0 2023-05-01 01:58:14,501 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197282.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 01:58:27,332 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-01 01:58:42,815 INFO [train.py:904] (7/8) Epoch 20, batch 4450, loss[loss=0.231, simple_loss=0.3206, pruned_loss=0.07073, over 16709.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2832, pruned_loss=0.05302, over 3202927.98 frames. ], batch size: 83, lr: 3.39e-03, grad_scale: 16.0 2023-05-01 01:59:14,457 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2023-05-01 01:59:55,991 INFO [train.py:904] (7/8) Epoch 20, batch 4500, loss[loss=0.2203, simple_loss=0.2887, pruned_loss=0.07597, over 11800.00 frames. ], tot_loss[loss=0.1959, simple_loss=0.2837, pruned_loss=0.05403, over 3207453.58 frames. ], batch size: 247, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:00:25,104 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0453, 5.0158, 4.7965, 4.1150, 4.9442, 1.8440, 4.6475, 4.3457], device='cuda:7'), covar=tensor([0.0050, 0.0045, 0.0125, 0.0288, 0.0052, 0.2832, 0.0082, 0.0215], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0151, 0.0196, 0.0177, 0.0173, 0.0205, 0.0187, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:00:30,612 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 1.873e+02 2.163e+02 2.505e+02 3.448e+02, threshold=4.325e+02, percent-clipped=0.0 2023-05-01 02:01:08,126 INFO [train.py:904] (7/8) Epoch 20, batch 4550, loss[loss=0.2272, simple_loss=0.3022, pruned_loss=0.07609, over 17038.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2847, pruned_loss=0.05508, over 3204826.48 frames. ], batch size: 53, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:01:27,179 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197415.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:01:42,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0290, 4.9008, 5.1065, 5.2310, 5.4311, 4.8152, 5.4280, 5.4384], device='cuda:7'), covar=tensor([0.1800, 0.1125, 0.1437, 0.0646, 0.0440, 0.0784, 0.0473, 0.0487], device='cuda:7'), in_proj_covar=tensor([0.0629, 0.0779, 0.0907, 0.0793, 0.0594, 0.0622, 0.0637, 0.0736], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:01:56,135 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197435.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:02:18,311 INFO [train.py:904] (7/8) Epoch 20, batch 4600, loss[loss=0.1943, simple_loss=0.2969, pruned_loss=0.04583, over 16904.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2853, pruned_loss=0.05485, over 3205313.55 frames. ], batch size: 96, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:02:43,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3860, 5.6890, 5.4851, 5.5285, 5.2132, 4.9037, 5.1570, 5.8265], device='cuda:7'), covar=tensor([0.1178, 0.0802, 0.0966, 0.0734, 0.0776, 0.0801, 0.1038, 0.0884], device='cuda:7'), in_proj_covar=tensor([0.0655, 0.0800, 0.0659, 0.0600, 0.0506, 0.0513, 0.0674, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:02:52,208 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 1.798e+02 2.178e+02 2.547e+02 5.479e+02, threshold=4.357e+02, percent-clipped=1.0 2023-05-01 02:02:52,834 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197476.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 02:03:03,001 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197483.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:03:30,133 INFO [train.py:904] (7/8) Epoch 20, batch 4650, loss[loss=0.1979, simple_loss=0.2802, pruned_loss=0.05782, over 16988.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2843, pruned_loss=0.05505, over 3205455.05 frames. ], batch size: 109, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:03:41,350 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197510.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:03:49,495 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197516.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:07,778 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0357, 4.1328, 3.8889, 3.5713, 3.6127, 4.0532, 3.6601, 3.7717], device='cuda:7'), covar=tensor([0.0514, 0.0413, 0.0249, 0.0261, 0.0635, 0.0339, 0.1059, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0411, 0.0333, 0.0330, 0.0346, 0.0379, 0.0231, 0.0398], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:04:21,486 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=197537.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:32,506 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197545.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:04:42,155 INFO [train.py:904] (7/8) Epoch 20, batch 4700, loss[loss=0.1878, simple_loss=0.2721, pruned_loss=0.05173, over 16480.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2817, pruned_loss=0.05411, over 3218254.70 frames. ], batch size: 68, lr: 3.39e-03, grad_scale: 8.0 2023-05-01 02:04:52,430 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197558.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:05:18,237 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.893e+02 2.227e+02 2.649e+02 4.562e+02, threshold=4.454e+02, percent-clipped=1.0 2023-05-01 02:05:19,892 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197577.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 02:05:25,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6086, 2.5150, 1.8673, 2.7138, 2.0834, 2.7402, 2.1491, 2.3911], device='cuda:7'), covar=tensor([0.0293, 0.0357, 0.1270, 0.0207, 0.0595, 0.0387, 0.1154, 0.0576], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0175, 0.0192, 0.0158, 0.0173, 0.0213, 0.0199, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 02:05:31,367 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=197585.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:05:55,398 INFO [train.py:904] (7/8) Epoch 20, batch 4750, loss[loss=0.1496, simple_loss=0.2396, pruned_loss=0.02981, over 16866.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2772, pruned_loss=0.05165, over 3220859.63 frames. ], batch size: 96, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:06:01,916 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197606.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:06:53,744 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.80 vs. limit=5.0 2023-05-01 02:07:05,644 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 02:07:08,962 INFO [train.py:904] (7/8) Epoch 20, batch 4800, loss[loss=0.1766, simple_loss=0.2695, pruned_loss=0.04184, over 15266.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2738, pruned_loss=0.04974, over 3222150.01 frames. ], batch size: 190, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:07:12,707 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1139, 2.1955, 2.2793, 3.8502, 2.1398, 2.5165, 2.3000, 2.3803], device='cuda:7'), covar=tensor([0.1418, 0.3614, 0.2911, 0.0534, 0.3959, 0.2575, 0.3611, 0.3127], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0440, 0.0361, 0.0323, 0.0432, 0.0507, 0.0409, 0.0516], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:07:45,096 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.191e+02 1.779e+02 2.011e+02 2.336e+02 4.336e+02, threshold=4.022e+02, percent-clipped=0.0 2023-05-01 02:08:24,128 INFO [train.py:904] (7/8) Epoch 20, batch 4850, loss[loss=0.1937, simple_loss=0.2854, pruned_loss=0.05098, over 15465.00 frames. ], tot_loss[loss=0.1862, simple_loss=0.2745, pruned_loss=0.04898, over 3207456.78 frames. ], batch size: 190, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:09:16,251 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 02:09:32,205 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 02:09:40,090 INFO [train.py:904] (7/8) Epoch 20, batch 4900, loss[loss=0.1618, simple_loss=0.2575, pruned_loss=0.03305, over 16492.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2736, pruned_loss=0.04765, over 3200655.97 frames. ], batch size: 62, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:10:08,663 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197771.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:10:15,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 1.929e+02 2.192e+02 2.700e+02 5.551e+02, threshold=4.384e+02, percent-clipped=1.0 2023-05-01 02:10:52,827 INFO [train.py:904] (7/8) Epoch 20, batch 4950, loss[loss=0.1707, simple_loss=0.2726, pruned_loss=0.0344, over 16756.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2735, pruned_loss=0.0469, over 3204337.37 frames. ], batch size: 89, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:12:04,542 INFO [train.py:904] (7/8) Epoch 20, batch 5000, loss[loss=0.1915, simple_loss=0.2814, pruned_loss=0.05074, over 16518.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.2748, pruned_loss=0.04707, over 3205090.07 frames. ], batch size: 75, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:12:33,418 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197872.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 02:12:38,859 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.376e+02 1.953e+02 2.351e+02 2.789e+02 5.042e+02, threshold=4.702e+02, percent-clipped=2.0 2023-05-01 02:13:01,331 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=197892.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:13:13,692 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=197901.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:13:14,486 INFO [train.py:904] (7/8) Epoch 20, batch 5050, loss[loss=0.1796, simple_loss=0.2647, pruned_loss=0.04726, over 16451.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2758, pruned_loss=0.04712, over 3215073.20 frames. ], batch size: 68, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:13:44,340 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1865, 5.1614, 5.0248, 4.6404, 4.6273, 5.0904, 5.0930, 4.7282], device='cuda:7'), covar=tensor([0.0602, 0.0400, 0.0301, 0.0284, 0.1127, 0.0451, 0.0240, 0.0622], device='cuda:7'), in_proj_covar=tensor([0.0288, 0.0415, 0.0336, 0.0331, 0.0347, 0.0385, 0.0231, 0.0401], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:14:25,050 INFO [train.py:904] (7/8) Epoch 20, batch 5100, loss[loss=0.1846, simple_loss=0.2732, pruned_loss=0.04799, over 16745.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2744, pruned_loss=0.04649, over 3197998.97 frames. ], batch size: 134, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:14:27,903 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=197953.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:15:00,611 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 1.957e+02 2.274e+02 2.606e+02 4.454e+02, threshold=4.549e+02, percent-clipped=0.0 2023-05-01 02:15:06,392 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6281, 4.8493, 4.9602, 4.7419, 4.8338, 5.3548, 4.7880, 4.5479], device='cuda:7'), covar=tensor([0.1162, 0.1911, 0.1614, 0.1886, 0.2287, 0.0843, 0.1394, 0.2232], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0570, 0.0623, 0.0474, 0.0633, 0.0662, 0.0490, 0.0638], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:15:41,439 INFO [train.py:904] (7/8) Epoch 20, batch 5150, loss[loss=0.185, simple_loss=0.2894, pruned_loss=0.04033, over 16892.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.274, pruned_loss=0.04592, over 3196985.39 frames. ], batch size: 96, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:16:52,362 INFO [train.py:904] (7/8) Epoch 20, batch 5200, loss[loss=0.159, simple_loss=0.2502, pruned_loss=0.03389, over 16765.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2722, pruned_loss=0.04493, over 3209353.94 frames. ], batch size: 83, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:17:20,265 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198071.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 02:17:27,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 1.904e+02 2.236e+02 2.771e+02 4.329e+02, threshold=4.472e+02, percent-clipped=0.0 2023-05-01 02:17:35,560 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198081.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:18:03,844 INFO [train.py:904] (7/8) Epoch 20, batch 5250, loss[loss=0.1585, simple_loss=0.2508, pruned_loss=0.03315, over 16706.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2687, pruned_loss=0.04404, over 3215083.86 frames. ], batch size: 83, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:18:27,797 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8724, 4.9710, 5.2935, 5.2769, 5.2748, 4.9692, 4.8715, 4.7159], device='cuda:7'), covar=tensor([0.0292, 0.0457, 0.0317, 0.0307, 0.0448, 0.0296, 0.0867, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0390, 0.0431, 0.0417, 0.0389, 0.0462, 0.0441, 0.0528, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 02:18:28,865 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198119.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 02:19:02,325 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198142.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:19:15,065 INFO [train.py:904] (7/8) Epoch 20, batch 5300, loss[loss=0.1546, simple_loss=0.238, pruned_loss=0.03557, over 16813.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2658, pruned_loss=0.04324, over 3214162.35 frames. ], batch size: 89, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:19:44,955 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198172.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:19:45,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0492, 2.1439, 2.1992, 3.6174, 2.0831, 2.5032, 2.3071, 2.3489], device='cuda:7'), covar=tensor([0.1356, 0.3432, 0.2839, 0.0566, 0.4044, 0.2419, 0.3309, 0.3049], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0438, 0.0360, 0.0322, 0.0430, 0.0505, 0.0408, 0.0512], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:19:49,805 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 1.825e+02 2.153e+02 2.534e+02 4.591e+02, threshold=4.307e+02, percent-clipped=1.0 2023-05-01 02:20:27,538 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198201.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:20:28,372 INFO [train.py:904] (7/8) Epoch 20, batch 5350, loss[loss=0.1766, simple_loss=0.2708, pruned_loss=0.04119, over 16800.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2648, pruned_loss=0.04281, over 3213361.79 frames. ], batch size: 83, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:20:54,795 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198220.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:21:35,356 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198248.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:21:37,094 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198249.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:21:40,992 INFO [train.py:904] (7/8) Epoch 20, batch 5400, loss[loss=0.2034, simple_loss=0.3007, pruned_loss=0.05304, over 15384.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2677, pruned_loss=0.04398, over 3194762.52 frames. ], batch size: 190, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:22:08,818 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-05-01 02:22:15,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 1.945e+02 2.236e+02 2.644e+02 4.931e+02, threshold=4.472e+02, percent-clipped=2.0 2023-05-01 02:22:57,922 INFO [train.py:904] (7/8) Epoch 20, batch 5450, loss[loss=0.1817, simple_loss=0.2756, pruned_loss=0.04392, over 17123.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2708, pruned_loss=0.04532, over 3182411.94 frames. ], batch size: 49, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:23:20,269 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 02:23:27,637 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9824, 2.1282, 2.2787, 3.5056, 2.0718, 2.4497, 2.2825, 2.2977], device='cuda:7'), covar=tensor([0.1336, 0.3271, 0.2685, 0.0557, 0.4056, 0.2301, 0.3222, 0.3186], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0437, 0.0360, 0.0322, 0.0430, 0.0505, 0.0408, 0.0513], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:24:14,769 INFO [train.py:904] (7/8) Epoch 20, batch 5500, loss[loss=0.2023, simple_loss=0.2963, pruned_loss=0.05413, over 16527.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2785, pruned_loss=0.05031, over 3128225.69 frames. ], batch size: 68, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:24:51,699 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.873e+02 3.524e+02 4.449e+02 7.452e+02, threshold=7.049e+02, percent-clipped=24.0 2023-05-01 02:25:00,790 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 02:25:34,177 INFO [train.py:904] (7/8) Epoch 20, batch 5550, loss[loss=0.2029, simple_loss=0.2931, pruned_loss=0.05631, over 15415.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2853, pruned_loss=0.05549, over 3111843.30 frames. ], batch size: 190, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:26:30,789 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198437.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:26:53,780 INFO [train.py:904] (7/8) Epoch 20, batch 5600, loss[loss=0.2202, simple_loss=0.303, pruned_loss=0.06872, over 16483.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2907, pruned_loss=0.06001, over 3077695.54 frames. ], batch size: 146, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:27:17,653 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3441, 3.2513, 3.6558, 1.7855, 3.8200, 3.8698, 2.9591, 2.8062], device='cuda:7'), covar=tensor([0.0874, 0.0289, 0.0200, 0.1248, 0.0069, 0.0165, 0.0424, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0097, 0.0139, 0.0080, 0.0124, 0.0128, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 02:27:32,096 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2023-05-01 02:27:34,784 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.278e+02 3.283e+02 3.655e+02 4.380e+02 7.107e+02, threshold=7.309e+02, percent-clipped=1.0 2023-05-01 02:28:14,287 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198499.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:28:18,615 INFO [train.py:904] (7/8) Epoch 20, batch 5650, loss[loss=0.2184, simple_loss=0.3129, pruned_loss=0.06197, over 16729.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.2946, pruned_loss=0.06361, over 3049398.53 frames. ], batch size: 76, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:28:31,498 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-01 02:28:50,063 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 02:29:32,479 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198548.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:29:36,992 INFO [train.py:904] (7/8) Epoch 20, batch 5700, loss[loss=0.209, simple_loss=0.304, pruned_loss=0.05699, over 16891.00 frames. ], tot_loss[loss=0.2149, simple_loss=0.2973, pruned_loss=0.06626, over 3024422.31 frames. ], batch size: 109, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:29:51,097 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198560.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:30:14,615 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.943e+02 3.352e+02 4.045e+02 5.075e+02 1.137e+03, threshold=8.090e+02, percent-clipped=5.0 2023-05-01 02:30:47,281 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198596.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:30:55,564 INFO [train.py:904] (7/8) Epoch 20, batch 5750, loss[loss=0.2264, simple_loss=0.2949, pruned_loss=0.07896, over 10945.00 frames. ], tot_loss[loss=0.2175, simple_loss=0.2999, pruned_loss=0.06759, over 3004203.55 frames. ], batch size: 247, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:31:17,316 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:31:20,737 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7547, 3.8397, 3.9166, 3.7211, 3.8548, 4.2502, 3.9022, 3.6465], device='cuda:7'), covar=tensor([0.2248, 0.2064, 0.2184, 0.2318, 0.2411, 0.1623, 0.1696, 0.2535], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0576, 0.0630, 0.0478, 0.0638, 0.0667, 0.0495, 0.0647], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:32:16,957 INFO [train.py:904] (7/8) Epoch 20, batch 5800, loss[loss=0.1987, simple_loss=0.2923, pruned_loss=0.0525, over 16843.00 frames. ], tot_loss[loss=0.2152, simple_loss=0.2989, pruned_loss=0.0657, over 3014186.71 frames. ], batch size: 116, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:32:39,106 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8531, 5.1624, 5.3046, 5.1161, 5.1535, 5.6838, 5.1669, 4.9601], device='cuda:7'), covar=tensor([0.1016, 0.1758, 0.2103, 0.1683, 0.2222, 0.0889, 0.1523, 0.2210], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0580, 0.0634, 0.0480, 0.0640, 0.0669, 0.0497, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:32:53,797 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.730e+02 3.384e+02 4.113e+02 5.841e+02, threshold=6.768e+02, percent-clipped=0.0 2023-05-01 02:32:54,380 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198676.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:32:54,520 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5452, 2.4580, 2.3303, 4.2552, 2.2772, 2.8393, 2.5002, 2.6794], device='cuda:7'), covar=tensor([0.1134, 0.3145, 0.2882, 0.0445, 0.3996, 0.2278, 0.3127, 0.2936], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0436, 0.0359, 0.0321, 0.0431, 0.0503, 0.0406, 0.0511], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:33:34,964 INFO [train.py:904] (7/8) Epoch 20, batch 5850, loss[loss=0.2019, simple_loss=0.2916, pruned_loss=0.0561, over 16295.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2967, pruned_loss=0.06382, over 3033031.04 frames. ], batch size: 165, lr: 3.38e-03, grad_scale: 8.0 2023-05-01 02:34:30,898 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=198737.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:34:32,590 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 02:34:53,519 INFO [train.py:904] (7/8) Epoch 20, batch 5900, loss[loss=0.2102, simple_loss=0.2886, pruned_loss=0.06585, over 15423.00 frames. ], tot_loss[loss=0.212, simple_loss=0.2965, pruned_loss=0.06378, over 3033297.90 frames. ], batch size: 190, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:35:34,252 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.837e+02 3.258e+02 4.009e+02 8.301e+02, threshold=6.515e+02, percent-clipped=1.0 2023-05-01 02:35:48,535 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=198785.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:36:11,769 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6938, 6.0436, 5.7578, 5.8094, 5.3963, 5.3195, 5.3029, 6.1389], device='cuda:7'), covar=tensor([0.1195, 0.0807, 0.0996, 0.0820, 0.0835, 0.0707, 0.1345, 0.0886], device='cuda:7'), in_proj_covar=tensor([0.0642, 0.0783, 0.0647, 0.0589, 0.0494, 0.0504, 0.0658, 0.0607], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:36:14,578 INFO [train.py:904] (7/8) Epoch 20, batch 5950, loss[loss=0.202, simple_loss=0.2931, pruned_loss=0.05548, over 16457.00 frames. ], tot_loss[loss=0.2109, simple_loss=0.297, pruned_loss=0.06243, over 3042987.42 frames. ], batch size: 68, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:36:19,104 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6768, 4.6727, 5.0502, 5.0126, 5.0398, 4.7471, 4.7029, 4.5314], device='cuda:7'), covar=tensor([0.0372, 0.0703, 0.0479, 0.0476, 0.0466, 0.0535, 0.1026, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0442, 0.0426, 0.0397, 0.0475, 0.0450, 0.0539, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 02:37:01,675 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2023-05-01 02:37:24,599 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6076, 2.4807, 2.3258, 3.5297, 2.4434, 3.7695, 1.4482, 2.7365], device='cuda:7'), covar=tensor([0.1366, 0.0795, 0.1263, 0.0182, 0.0192, 0.0365, 0.1731, 0.0817], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0173, 0.0193, 0.0185, 0.0205, 0.0212, 0.0199, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 02:37:31,024 INFO [train.py:904] (7/8) Epoch 20, batch 6000, loss[loss=0.1921, simple_loss=0.2814, pruned_loss=0.05138, over 16241.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2959, pruned_loss=0.06173, over 3056734.42 frames. ], batch size: 165, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:37:31,024 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 02:37:41,834 INFO [train.py:938] (7/8) Epoch 20, validation: loss=0.1516, simple_loss=0.2644, pruned_loss=0.01942, over 944034.00 frames. 2023-05-01 02:37:41,835 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 02:37:42,445 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198852.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:37:46,613 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198855.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:37:46,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6645, 1.7586, 1.6364, 1.5055, 1.9401, 1.6144, 1.5558, 1.8941], device='cuda:7'), covar=tensor([0.0192, 0.0297, 0.0406, 0.0347, 0.0220, 0.0267, 0.0189, 0.0222], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0227, 0.0220, 0.0220, 0.0230, 0.0227, 0.0229, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:38:08,118 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5662, 4.6844, 5.0224, 4.9445, 5.0052, 4.7280, 4.5597, 4.4892], device='cuda:7'), covar=tensor([0.0481, 0.0624, 0.0435, 0.0600, 0.0613, 0.0520, 0.1278, 0.0525], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0443, 0.0427, 0.0398, 0.0475, 0.0451, 0.0540, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 02:38:17,164 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 2.764e+02 3.257e+02 3.996e+02 6.001e+02, threshold=6.515e+02, percent-clipped=0.0 2023-05-01 02:38:28,096 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198883.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:38:56,346 INFO [train.py:904] (7/8) Epoch 20, batch 6050, loss[loss=0.1816, simple_loss=0.2884, pruned_loss=0.0374, over 16682.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2948, pruned_loss=0.06137, over 3068692.00 frames. ], batch size: 76, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:39:14,820 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198913.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:39:36,246 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=198927.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:39:57,606 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3465, 3.2842, 3.8278, 1.8279, 3.9932, 4.0474, 3.0016, 2.8327], device='cuda:7'), covar=tensor([0.0876, 0.0249, 0.0162, 0.1250, 0.0064, 0.0133, 0.0401, 0.0512], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0109, 0.0098, 0.0139, 0.0080, 0.0124, 0.0129, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 02:40:05,114 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198944.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:40:17,605 INFO [train.py:904] (7/8) Epoch 20, batch 6100, loss[loss=0.1867, simple_loss=0.2738, pruned_loss=0.0498, over 16927.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2935, pruned_loss=0.05992, over 3085380.71 frames. ], batch size: 109, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:40:46,262 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=198971.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:40:53,273 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.636e+02 3.021e+02 3.731e+02 7.049e+02, threshold=6.042e+02, percent-clipped=1.0 2023-05-01 02:41:01,053 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 02:41:10,901 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-05-01 02:41:12,482 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=198988.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:41:26,002 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7614, 3.0394, 3.1687, 1.9684, 2.7757, 2.1246, 3.3040, 3.3475], device='cuda:7'), covar=tensor([0.0301, 0.0854, 0.0661, 0.2097, 0.0902, 0.1058, 0.0648, 0.1027], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0162, 0.0165, 0.0151, 0.0143, 0.0128, 0.0144, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 02:41:32,342 INFO [train.py:904] (7/8) Epoch 20, batch 6150, loss[loss=0.1966, simple_loss=0.2814, pruned_loss=0.05587, over 16391.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2918, pruned_loss=0.05965, over 3075284.78 frames. ], batch size: 146, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:41:39,279 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1698, 4.3092, 4.0774, 3.8487, 3.6530, 4.2445, 3.9960, 3.8460], device='cuda:7'), covar=tensor([0.0765, 0.0619, 0.0399, 0.0378, 0.1066, 0.0567, 0.0751, 0.0724], device='cuda:7'), in_proj_covar=tensor([0.0284, 0.0410, 0.0332, 0.0325, 0.0342, 0.0379, 0.0228, 0.0396], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:41:47,564 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4212, 5.7168, 5.4519, 5.4726, 5.1105, 5.0143, 5.1170, 5.7957], device='cuda:7'), covar=tensor([0.1206, 0.0770, 0.1016, 0.0816, 0.0812, 0.0728, 0.1243, 0.0787], device='cuda:7'), in_proj_covar=tensor([0.0640, 0.0780, 0.0645, 0.0588, 0.0493, 0.0504, 0.0657, 0.0605], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:42:49,591 INFO [train.py:904] (7/8) Epoch 20, batch 6200, loss[loss=0.1817, simple_loss=0.2724, pruned_loss=0.04551, over 16613.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2898, pruned_loss=0.0593, over 3086094.90 frames. ], batch size: 57, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:43:19,309 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9762, 3.5159, 3.4856, 2.2730, 3.2839, 3.5395, 3.2798, 1.9198], device='cuda:7'), covar=tensor([0.0565, 0.0051, 0.0057, 0.0400, 0.0102, 0.0105, 0.0097, 0.0484], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0081, 0.0080, 0.0132, 0.0095, 0.0107, 0.0092, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:43:19,586 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 02:43:22,750 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3092, 4.3849, 4.7275, 4.6842, 4.6803, 4.3770, 4.3755, 4.2956], device='cuda:7'), covar=tensor([0.0355, 0.0609, 0.0345, 0.0386, 0.0493, 0.0444, 0.0984, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0442, 0.0426, 0.0398, 0.0476, 0.0451, 0.0541, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 02:43:28,025 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.934e+02 3.552e+02 4.446e+02 1.093e+03, threshold=7.104e+02, percent-clipped=2.0 2023-05-01 02:44:06,526 INFO [train.py:904] (7/8) Epoch 20, batch 6250, loss[loss=0.1678, simple_loss=0.2616, pruned_loss=0.03704, over 17016.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2894, pruned_loss=0.05857, over 3102191.11 frames. ], batch size: 50, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:44:56,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7414, 2.7485, 2.8774, 2.1083, 2.6692, 2.1345, 2.7855, 2.8935], device='cuda:7'), covar=tensor([0.0244, 0.0678, 0.0467, 0.1745, 0.0743, 0.0867, 0.0487, 0.0723], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0161, 0.0165, 0.0150, 0.0143, 0.0128, 0.0143, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 02:45:17,404 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 02:45:21,384 INFO [train.py:904] (7/8) Epoch 20, batch 6300, loss[loss=0.2003, simple_loss=0.2838, pruned_loss=0.05838, over 16675.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2886, pruned_loss=0.05799, over 3102309.18 frames. ], batch size: 134, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:45:26,468 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199155.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:45:31,656 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-05-01 02:45:39,631 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8266, 1.3830, 1.7061, 1.7108, 1.8374, 1.9211, 1.6388, 1.8219], device='cuda:7'), covar=tensor([0.0236, 0.0387, 0.0196, 0.0282, 0.0261, 0.0182, 0.0386, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0191, 0.0176, 0.0180, 0.0192, 0.0150, 0.0193, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:45:43,216 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6285, 4.8354, 4.9640, 4.7649, 4.8270, 5.3483, 4.8199, 4.6131], device='cuda:7'), covar=tensor([0.1173, 0.1671, 0.2199, 0.2072, 0.2380, 0.0952, 0.1569, 0.2376], device='cuda:7'), in_proj_covar=tensor([0.0403, 0.0584, 0.0642, 0.0485, 0.0644, 0.0673, 0.0501, 0.0651], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:45:59,387 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199176.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:46:00,085 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.632e+02 3.175e+02 3.943e+02 7.347e+02, threshold=6.350e+02, percent-clipped=2.0 2023-05-01 02:46:16,349 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5826, 4.7064, 4.9797, 4.9222, 5.0052, 4.7099, 4.6173, 4.5177], device='cuda:7'), covar=tensor([0.0476, 0.0695, 0.0481, 0.0580, 0.0598, 0.0543, 0.1184, 0.0558], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0439, 0.0423, 0.0396, 0.0473, 0.0448, 0.0537, 0.0360], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 02:46:38,975 INFO [train.py:904] (7/8) Epoch 20, batch 6350, loss[loss=0.2491, simple_loss=0.3217, pruned_loss=0.0882, over 15207.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2892, pruned_loss=0.05876, over 3103797.02 frames. ], batch size: 190, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 02:46:40,592 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:46:48,083 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199208.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:47:30,936 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199237.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:47:34,244 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199239.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:47:48,857 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-05-01 02:47:52,861 INFO [train.py:904] (7/8) Epoch 20, batch 6400, loss[loss=0.2659, simple_loss=0.33, pruned_loss=0.1009, over 10923.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2901, pruned_loss=0.06011, over 3101678.43 frames. ], batch size: 247, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:48:21,604 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199271.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:48:29,166 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.234e+02 2.937e+02 3.403e+02 4.348e+02 9.192e+02, threshold=6.807e+02, percent-clipped=3.0 2023-05-01 02:48:39,485 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199283.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:49:01,840 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 02:49:07,110 INFO [train.py:904] (7/8) Epoch 20, batch 6450, loss[loss=0.1991, simple_loss=0.2798, pruned_loss=0.05918, over 11621.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2905, pruned_loss=0.05966, over 3084840.58 frames. ], batch size: 248, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:49:12,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9888, 5.4158, 5.6207, 5.2751, 5.3821, 5.9557, 5.4149, 5.1866], device='cuda:7'), covar=tensor([0.0968, 0.1547, 0.1827, 0.1884, 0.2322, 0.0970, 0.1581, 0.2088], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0576, 0.0633, 0.0478, 0.0633, 0.0665, 0.0494, 0.0642], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:49:33,335 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199319.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:50:24,682 INFO [train.py:904] (7/8) Epoch 20, batch 6500, loss[loss=0.1716, simple_loss=0.2547, pruned_loss=0.04423, over 17208.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2882, pruned_loss=0.05887, over 3104061.27 frames. ], batch size: 44, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:50:33,585 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 02:51:02,008 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.751e+02 2.867e+02 3.442e+02 4.204e+02 8.359e+02, threshold=6.884e+02, percent-clipped=2.0 2023-05-01 02:51:39,152 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199400.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:51:41,759 INFO [train.py:904] (7/8) Epoch 20, batch 6550, loss[loss=0.1856, simple_loss=0.2938, pruned_loss=0.03873, over 16451.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2905, pruned_loss=0.05923, over 3105086.36 frames. ], batch size: 68, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:51:51,179 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-01 02:52:56,173 INFO [train.py:904] (7/8) Epoch 20, batch 6600, loss[loss=0.1965, simple_loss=0.2882, pruned_loss=0.05238, over 17007.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2919, pruned_loss=0.05893, over 3111628.10 frames. ], batch size: 55, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:53:05,805 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199458.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:53:10,073 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199461.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 02:53:33,380 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.722e+02 3.309e+02 3.917e+02 9.551e+02, threshold=6.618e+02, percent-clipped=2.0 2023-05-01 02:53:59,432 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199494.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:54:11,576 INFO [train.py:904] (7/8) Epoch 20, batch 6650, loss[loss=0.1988, simple_loss=0.2819, pruned_loss=0.05779, over 16504.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2922, pruned_loss=0.05949, over 3106319.36 frames. ], batch size: 68, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:54:20,160 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199508.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:54:37,237 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4614, 4.5522, 4.3742, 4.0975, 4.0434, 4.4768, 4.2529, 4.1673], device='cuda:7'), covar=tensor([0.0737, 0.0494, 0.0338, 0.0334, 0.0990, 0.0500, 0.0602, 0.0688], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0408, 0.0328, 0.0323, 0.0337, 0.0377, 0.0228, 0.0392], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:54:37,290 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199519.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 02:54:44,832 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8668, 1.3419, 1.7313, 1.6663, 1.7958, 1.9518, 1.5886, 1.7920], device='cuda:7'), covar=tensor([0.0239, 0.0425, 0.0218, 0.0311, 0.0263, 0.0169, 0.0435, 0.0130], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0192, 0.0176, 0.0181, 0.0192, 0.0150, 0.0193, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:54:56,500 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199532.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:55:00,541 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199535.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:55:07,043 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199539.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:55:25,190 INFO [train.py:904] (7/8) Epoch 20, batch 6700, loss[loss=0.2226, simple_loss=0.3095, pruned_loss=0.06785, over 16916.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2917, pruned_loss=0.06024, over 3101189.98 frames. ], batch size: 109, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:55:31,350 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199555.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 02:55:32,691 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199556.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:55:54,930 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4855, 2.4501, 2.4527, 4.3223, 2.3287, 2.9349, 2.4777, 2.6377], device='cuda:7'), covar=tensor([0.1202, 0.3251, 0.2756, 0.0480, 0.3897, 0.2177, 0.3344, 0.3069], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0437, 0.0359, 0.0321, 0.0431, 0.0505, 0.0407, 0.0513], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:56:02,764 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.649e+02 2.838e+02 3.650e+02 4.570e+02 8.139e+02, threshold=7.299e+02, percent-clipped=7.0 2023-05-01 02:56:13,374 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199583.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:56:18,218 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199587.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:56:31,533 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199596.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 02:56:32,878 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2411, 4.3048, 4.6426, 4.5896, 4.6172, 4.3329, 4.3056, 4.2345], device='cuda:7'), covar=tensor([0.0349, 0.0545, 0.0347, 0.0432, 0.0466, 0.0403, 0.0902, 0.0476], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0440, 0.0425, 0.0398, 0.0475, 0.0450, 0.0541, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 02:56:38,614 INFO [train.py:904] (7/8) Epoch 20, batch 6750, loss[loss=0.1838, simple_loss=0.269, pruned_loss=0.04929, over 16907.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2907, pruned_loss=0.05996, over 3111423.33 frames. ], batch size: 116, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:57:06,854 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7807, 4.7526, 4.5479, 3.0575, 3.9560, 4.5685, 3.9470, 2.9414], device='cuda:7'), covar=tensor([0.0502, 0.0027, 0.0035, 0.0344, 0.0100, 0.0108, 0.0089, 0.0342], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0081, 0.0080, 0.0133, 0.0096, 0.0107, 0.0092, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 02:57:20,747 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199631.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 02:57:53,092 INFO [train.py:904] (7/8) Epoch 20, batch 6800, loss[loss=0.2116, simple_loss=0.3028, pruned_loss=0.06019, over 16773.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2903, pruned_loss=0.05949, over 3115287.03 frames. ], batch size: 83, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 02:58:02,338 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4715, 1.6892, 2.1559, 2.4212, 2.4827, 2.7370, 1.8660, 2.6290], device='cuda:7'), covar=tensor([0.0216, 0.0502, 0.0309, 0.0321, 0.0323, 0.0186, 0.0519, 0.0164], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0192, 0.0176, 0.0180, 0.0193, 0.0150, 0.0193, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 02:58:29,264 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.711e+02 3.238e+02 3.835e+02 6.988e+02, threshold=6.476e+02, percent-clipped=0.0 2023-05-01 02:59:05,436 INFO [train.py:904] (7/8) Epoch 20, batch 6850, loss[loss=0.193, simple_loss=0.288, pruned_loss=0.04899, over 15307.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2909, pruned_loss=0.05963, over 3121161.37 frames. ], batch size: 190, lr: 3.37e-03, grad_scale: 8.0 2023-05-01 03:00:20,627 INFO [train.py:904] (7/8) Epoch 20, batch 6900, loss[loss=0.2556, simple_loss=0.3173, pruned_loss=0.09691, over 11568.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2933, pruned_loss=0.05952, over 3107956.88 frames. ], batch size: 248, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:00:26,880 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199756.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 03:00:59,801 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 2.558e+02 3.121e+02 3.929e+02 6.424e+02, threshold=6.242e+02, percent-clipped=0.0 2023-05-01 03:01:36,156 INFO [train.py:904] (7/8) Epoch 20, batch 6950, loss[loss=0.2194, simple_loss=0.3065, pruned_loss=0.06609, over 16686.00 frames. ], tot_loss[loss=0.2089, simple_loss=0.2951, pruned_loss=0.0613, over 3089949.81 frames. ], batch size: 124, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:01:55,061 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199814.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:02:22,400 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=199832.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:02:47,653 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199850.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 03:02:49,070 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7220, 4.1337, 3.1154, 2.3018, 2.7443, 2.6097, 4.3839, 3.5712], device='cuda:7'), covar=tensor([0.2904, 0.0588, 0.1688, 0.2948, 0.2592, 0.1890, 0.0513, 0.1310], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0267, 0.0301, 0.0307, 0.0294, 0.0253, 0.0292, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 03:02:49,638 INFO [train.py:904] (7/8) Epoch 20, batch 7000, loss[loss=0.1876, simple_loss=0.2869, pruned_loss=0.04417, over 17039.00 frames. ], tot_loss[loss=0.2086, simple_loss=0.2954, pruned_loss=0.06088, over 3090489.91 frames. ], batch size: 55, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:02:54,388 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-01 03:03:29,996 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.723e+02 2.829e+02 3.376e+02 4.341e+02 9.860e+02, threshold=6.751e+02, percent-clipped=7.0 2023-05-01 03:03:33,472 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=199880.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:03:49,935 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=199891.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:04:07,215 INFO [train.py:904] (7/8) Epoch 20, batch 7050, loss[loss=0.2604, simple_loss=0.3209, pruned_loss=0.1, over 11390.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2951, pruned_loss=0.06014, over 3102840.42 frames. ], batch size: 248, lr: 3.37e-03, grad_scale: 4.0 2023-05-01 03:04:18,524 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=199909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:05:25,773 INFO [train.py:904] (7/8) Epoch 20, batch 7100, loss[loss=0.2321, simple_loss=0.2947, pruned_loss=0.08471, over 11266.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2934, pruned_loss=0.0596, over 3101453.48 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:05:54,788 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=199970.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:06:05,961 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.782e+02 3.561e+02 4.163e+02 7.041e+02, threshold=7.121e+02, percent-clipped=1.0 2023-05-01 03:06:45,339 INFO [train.py:904] (7/8) Epoch 20, batch 7150, loss[loss=0.1958, simple_loss=0.2834, pruned_loss=0.05413, over 16678.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2918, pruned_loss=0.05951, over 3097475.80 frames. ], batch size: 134, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:07:08,776 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 03:07:59,662 INFO [train.py:904] (7/8) Epoch 20, batch 7200, loss[loss=0.17, simple_loss=0.2679, pruned_loss=0.03603, over 16732.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2898, pruned_loss=0.05829, over 3084266.98 frames. ], batch size: 83, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:08:06,696 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200056.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:08:40,492 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.812e+02 2.598e+02 3.072e+02 3.513e+02 6.350e+02, threshold=6.144e+02, percent-clipped=0.0 2023-05-01 03:09:16,826 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9806, 2.1271, 2.2898, 3.4272, 2.0848, 2.4049, 2.2508, 2.2626], device='cuda:7'), covar=tensor([0.1371, 0.3410, 0.2804, 0.0678, 0.4203, 0.2459, 0.3473, 0.3476], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0439, 0.0360, 0.0321, 0.0432, 0.0505, 0.0408, 0.0514], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:09:19,974 INFO [train.py:904] (7/8) Epoch 20, batch 7250, loss[loss=0.1752, simple_loss=0.2595, pruned_loss=0.04542, over 17266.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2883, pruned_loss=0.05782, over 3079936.20 frames. ], batch size: 52, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:09:23,388 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200104.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:09:37,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200114.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:10:32,785 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200150.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 03:10:35,114 INFO [train.py:904] (7/8) Epoch 20, batch 7300, loss[loss=0.2206, simple_loss=0.3036, pruned_loss=0.06882, over 15327.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2884, pruned_loss=0.05831, over 3073567.17 frames. ], batch size: 190, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:10:50,981 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200162.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:11:15,306 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=200177.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:11:15,884 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.904e+02 3.413e+02 4.614e+02 7.836e+02, threshold=6.825e+02, percent-clipped=9.0 2023-05-01 03:11:32,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4192, 3.5708, 3.7529, 2.2756, 3.1745, 2.5290, 3.7463, 3.8650], device='cuda:7'), covar=tensor([0.0255, 0.0890, 0.0549, 0.2029, 0.0809, 0.0953, 0.0696, 0.1014], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0163, 0.0167, 0.0152, 0.0144, 0.0130, 0.0144, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:11:37,010 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200191.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:11:47,791 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200198.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:11:52,594 INFO [train.py:904] (7/8) Epoch 20, batch 7350, loss[loss=0.1953, simple_loss=0.2819, pruned_loss=0.05433, over 16650.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2893, pruned_loss=0.05894, over 3064962.20 frames. ], batch size: 62, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:11:59,869 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8008, 2.6556, 1.8649, 2.2303, 3.0063, 2.5353, 3.2735, 3.2247], device='cuda:7'), covar=tensor([0.0108, 0.0519, 0.0804, 0.0630, 0.0331, 0.0588, 0.0333, 0.0295], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0228, 0.0220, 0.0220, 0.0229, 0.0228, 0.0227, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:12:51,296 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=200238.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:12:52,225 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200239.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:13:12,420 INFO [train.py:904] (7/8) Epoch 20, batch 7400, loss[loss=0.2636, simple_loss=0.3276, pruned_loss=0.0998, over 11387.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.291, pruned_loss=0.0599, over 3065324.32 frames. ], batch size: 247, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:13:23,785 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1257, 3.6688, 3.6529, 2.3017, 3.4434, 3.6743, 3.3884, 2.0271], device='cuda:7'), covar=tensor([0.0560, 0.0054, 0.0056, 0.0425, 0.0088, 0.0120, 0.0096, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0080, 0.0080, 0.0132, 0.0095, 0.0107, 0.0092, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 03:13:33,075 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=200265.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:13:52,749 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.893e+02 3.356e+02 4.091e+02 7.315e+02, threshold=6.713e+02, percent-clipped=2.0 2023-05-01 03:13:55,489 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=200279.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:14:32,188 INFO [train.py:904] (7/8) Epoch 20, batch 7450, loss[loss=0.188, simple_loss=0.2862, pruned_loss=0.04491, over 16831.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2926, pruned_loss=0.06082, over 3066212.38 frames. ], batch size: 83, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:15:13,574 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3222, 1.6151, 2.1071, 2.2946, 2.4035, 2.5961, 1.8424, 2.4969], device='cuda:7'), covar=tensor([0.0226, 0.0507, 0.0277, 0.0282, 0.0297, 0.0194, 0.0465, 0.0127], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0188, 0.0172, 0.0177, 0.0189, 0.0148, 0.0189, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:15:36,286 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=200340.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:15:54,175 INFO [train.py:904] (7/8) Epoch 20, batch 7500, loss[loss=0.2156, simple_loss=0.2906, pruned_loss=0.07028, over 11355.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2928, pruned_loss=0.06043, over 3040720.49 frames. ], batch size: 247, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:16:34,101 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 2.766e+02 3.444e+02 4.330e+02 6.961e+02, threshold=6.888e+02, percent-clipped=1.0 2023-05-01 03:16:37,898 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7631, 1.3660, 1.7143, 1.6073, 1.7634, 1.8658, 1.5960, 1.7258], device='cuda:7'), covar=tensor([0.0249, 0.0355, 0.0195, 0.0286, 0.0253, 0.0182, 0.0395, 0.0124], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0187, 0.0171, 0.0177, 0.0189, 0.0147, 0.0189, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:17:11,467 INFO [train.py:904] (7/8) Epoch 20, batch 7550, loss[loss=0.2166, simple_loss=0.3002, pruned_loss=0.0665, over 16718.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2911, pruned_loss=0.06037, over 3041739.29 frames. ], batch size: 124, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:17:58,557 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2249, 4.2981, 4.6106, 4.5708, 4.5893, 4.3046, 4.3087, 4.1916], device='cuda:7'), covar=tensor([0.0358, 0.0558, 0.0399, 0.0430, 0.0504, 0.0433, 0.0945, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0438, 0.0425, 0.0396, 0.0473, 0.0447, 0.0539, 0.0360], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 03:18:26,304 INFO [train.py:904] (7/8) Epoch 20, batch 7600, loss[loss=0.2064, simple_loss=0.2927, pruned_loss=0.06008, over 16509.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2911, pruned_loss=0.06144, over 3016881.38 frames. ], batch size: 75, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:19:06,495 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.777e+02 3.348e+02 4.097e+02 9.491e+02, threshold=6.697e+02, percent-clipped=3.0 2023-05-01 03:19:44,062 INFO [train.py:904] (7/8) Epoch 20, batch 7650, loss[loss=0.2159, simple_loss=0.3021, pruned_loss=0.06486, over 15265.00 frames. ], tot_loss[loss=0.2087, simple_loss=0.2922, pruned_loss=0.06257, over 3009604.35 frames. ], batch size: 191, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:20:33,352 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=200533.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:20:58,555 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8092, 2.1189, 2.3760, 3.1377, 2.2013, 2.3107, 2.3271, 2.2321], device='cuda:7'), covar=tensor([0.1315, 0.3165, 0.2467, 0.0686, 0.3998, 0.2413, 0.2950, 0.3337], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0438, 0.0359, 0.0321, 0.0431, 0.0505, 0.0408, 0.0513], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:21:02,178 INFO [train.py:904] (7/8) Epoch 20, batch 7700, loss[loss=0.2108, simple_loss=0.2926, pruned_loss=0.06444, over 16261.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2908, pruned_loss=0.06182, over 3048935.38 frames. ], batch size: 165, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:21:22,887 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200565.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:21:44,265 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 2.918e+02 3.681e+02 4.507e+02 8.362e+02, threshold=7.363e+02, percent-clipped=4.0 2023-05-01 03:22:21,790 INFO [train.py:904] (7/8) Epoch 20, batch 7750, loss[loss=0.2275, simple_loss=0.3013, pruned_loss=0.07683, over 11310.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2904, pruned_loss=0.06089, over 3062746.41 frames. ], batch size: 246, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:22:29,603 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-05-01 03:22:38,650 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200613.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:23:08,902 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2023-05-01 03:23:11,988 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=200635.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:23:26,647 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 03:23:28,164 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=200645.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:23:36,965 INFO [train.py:904] (7/8) Epoch 20, batch 7800, loss[loss=0.2059, simple_loss=0.293, pruned_loss=0.05944, over 16655.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2907, pruned_loss=0.06093, over 3083276.86 frames. ], batch size: 134, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:24:18,683 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.831e+02 3.422e+02 4.035e+02 8.624e+02, threshold=6.845e+02, percent-clipped=1.0 2023-05-01 03:24:19,855 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3742, 4.6581, 4.4698, 4.5051, 4.1973, 4.1323, 4.1243, 4.7239], device='cuda:7'), covar=tensor([0.1232, 0.0955, 0.1021, 0.0908, 0.0877, 0.1575, 0.1285, 0.0923], device='cuda:7'), in_proj_covar=tensor([0.0647, 0.0795, 0.0656, 0.0598, 0.0500, 0.0513, 0.0664, 0.0614], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:24:53,417 INFO [train.py:904] (7/8) Epoch 20, batch 7850, loss[loss=0.1923, simple_loss=0.2846, pruned_loss=0.04996, over 16716.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2908, pruned_loss=0.05995, over 3095876.41 frames. ], batch size: 134, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:25:00,398 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=200706.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:26:08,727 INFO [train.py:904] (7/8) Epoch 20, batch 7900, loss[loss=0.2234, simple_loss=0.3144, pruned_loss=0.06623, over 16297.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2906, pruned_loss=0.06023, over 3068556.32 frames. ], batch size: 165, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:26:49,224 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.098e+02 2.835e+02 3.346e+02 4.054e+02 6.354e+02, threshold=6.693e+02, percent-clipped=0.0 2023-05-01 03:26:52,640 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9152, 4.1860, 4.0069, 4.0574, 3.7519, 3.8056, 3.8482, 4.1834], device='cuda:7'), covar=tensor([0.1179, 0.0911, 0.1050, 0.0832, 0.0800, 0.1653, 0.1002, 0.1018], device='cuda:7'), in_proj_covar=tensor([0.0648, 0.0793, 0.0655, 0.0597, 0.0500, 0.0514, 0.0664, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:26:59,908 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5575, 2.5139, 2.2432, 3.5135, 2.2299, 3.6854, 1.5061, 2.7290], device='cuda:7'), covar=tensor([0.1506, 0.0817, 0.1308, 0.0214, 0.0177, 0.0424, 0.1814, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0186, 0.0208, 0.0214, 0.0200, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:27:27,132 INFO [train.py:904] (7/8) Epoch 20, batch 7950, loss[loss=0.2072, simple_loss=0.2965, pruned_loss=0.05896, over 16611.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2917, pruned_loss=0.06099, over 3070179.78 frames. ], batch size: 62, lr: 3.36e-03, grad_scale: 4.0 2023-05-01 03:28:16,250 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200833.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 03:28:44,617 INFO [train.py:904] (7/8) Epoch 20, batch 8000, loss[loss=0.2174, simple_loss=0.3008, pruned_loss=0.06698, over 16777.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2919, pruned_loss=0.0613, over 3074816.35 frames. ], batch size: 124, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:29:24,820 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.028e+02 3.046e+02 3.376e+02 4.028e+02 7.694e+02, threshold=6.753e+02, percent-clipped=2.0 2023-05-01 03:29:28,362 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200881.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 03:29:59,865 INFO [train.py:904] (7/8) Epoch 20, batch 8050, loss[loss=0.222, simple_loss=0.3119, pruned_loss=0.06604, over 16173.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2921, pruned_loss=0.0612, over 3064180.87 frames. ], batch size: 165, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:30:22,050 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5375, 3.5993, 2.0761, 4.1378, 2.7457, 4.0753, 2.3468, 2.8621], device='cuda:7'), covar=tensor([0.0296, 0.0411, 0.1739, 0.0206, 0.0814, 0.0536, 0.1466, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0173, 0.0192, 0.0155, 0.0174, 0.0213, 0.0199, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:30:36,836 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 03:30:50,128 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=200935.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:30:55,320 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 03:31:15,315 INFO [train.py:904] (7/8) Epoch 20, batch 8100, loss[loss=0.2208, simple_loss=0.3013, pruned_loss=0.07019, over 11464.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.292, pruned_loss=0.06072, over 3068912.36 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:31:57,146 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 2.634e+02 3.129e+02 3.819e+02 6.896e+02, threshold=6.259e+02, percent-clipped=1.0 2023-05-01 03:32:04,950 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=200983.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:32:31,958 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201001.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:32:32,048 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8862, 2.2745, 2.5127, 1.8091, 2.5707, 2.8032, 2.4263, 2.2349], device='cuda:7'), covar=tensor([0.0930, 0.0238, 0.0215, 0.1156, 0.0102, 0.0257, 0.0461, 0.0594], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0097, 0.0138, 0.0078, 0.0122, 0.0126, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:32:33,319 INFO [train.py:904] (7/8) Epoch 20, batch 8150, loss[loss=0.2033, simple_loss=0.2807, pruned_loss=0.06296, over 11885.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2897, pruned_loss=0.05993, over 3071781.31 frames. ], batch size: 248, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:32:55,782 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5102, 3.4975, 3.4877, 2.6881, 3.3685, 2.0711, 3.1496, 2.7983], device='cuda:7'), covar=tensor([0.0132, 0.0109, 0.0166, 0.0210, 0.0087, 0.2184, 0.0130, 0.0210], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0147, 0.0189, 0.0173, 0.0167, 0.0201, 0.0180, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:33:24,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7991, 5.1002, 4.8846, 4.8646, 4.6178, 4.5627, 4.5285, 5.1815], device='cuda:7'), covar=tensor([0.1248, 0.0828, 0.0995, 0.0923, 0.0861, 0.1163, 0.1270, 0.0907], device='cuda:7'), in_proj_covar=tensor([0.0649, 0.0791, 0.0655, 0.0598, 0.0500, 0.0514, 0.0664, 0.0613], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:33:52,119 INFO [train.py:904] (7/8) Epoch 20, batch 8200, loss[loss=0.1911, simple_loss=0.2824, pruned_loss=0.04987, over 16336.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2871, pruned_loss=0.05899, over 3077687.11 frames. ], batch size: 165, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:34:24,400 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201071.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:34:36,692 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.058e+02 2.732e+02 3.390e+02 4.593e+02 1.143e+03, threshold=6.781e+02, percent-clipped=6.0 2023-05-01 03:35:08,307 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4919, 3.3396, 2.7719, 2.0631, 2.1176, 2.2736, 3.4073, 2.9542], device='cuda:7'), covar=tensor([0.2999, 0.0629, 0.1729, 0.3284, 0.2995, 0.2273, 0.0436, 0.1509], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0269, 0.0304, 0.0310, 0.0297, 0.0256, 0.0293, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 03:35:15,219 INFO [train.py:904] (7/8) Epoch 20, batch 8250, loss[loss=0.1784, simple_loss=0.2735, pruned_loss=0.0417, over 16670.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2859, pruned_loss=0.05617, over 3061055.23 frames. ], batch size: 76, lr: 3.36e-03, grad_scale: 8.0 2023-05-01 03:36:06,010 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201132.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:36:11,445 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6314, 2.5705, 2.6469, 3.7066, 2.4389, 3.7861, 1.3607, 3.0165], device='cuda:7'), covar=tensor([0.1424, 0.0724, 0.0988, 0.0180, 0.0109, 0.0328, 0.1816, 0.0658], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0172, 0.0193, 0.0184, 0.0206, 0.0211, 0.0199, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:36:38,975 INFO [train.py:904] (7/8) Epoch 20, batch 8300, loss[loss=0.1646, simple_loss=0.2645, pruned_loss=0.03241, over 16416.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2835, pruned_loss=0.0539, over 3041881.36 frames. ], batch size: 68, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:37:09,101 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7275, 4.8996, 5.1025, 4.8577, 4.9145, 5.4902, 4.9868, 4.6842], device='cuda:7'), covar=tensor([0.1050, 0.1997, 0.2102, 0.1823, 0.2634, 0.0894, 0.1612, 0.2532], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0574, 0.0632, 0.0475, 0.0632, 0.0664, 0.0496, 0.0643], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 03:37:22,454 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.376e+02 2.857e+02 3.605e+02 7.537e+02, threshold=5.714e+02, percent-clipped=2.0 2023-05-01 03:38:00,768 INFO [train.py:904] (7/8) Epoch 20, batch 8350, loss[loss=0.186, simple_loss=0.2795, pruned_loss=0.04627, over 16406.00 frames. ], tot_loss[loss=0.1928, simple_loss=0.2823, pruned_loss=0.05171, over 3050391.99 frames. ], batch size: 68, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:39:10,179 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0301, 2.0944, 2.1684, 3.5441, 2.0851, 2.3824, 2.2202, 2.2000], device='cuda:7'), covar=tensor([0.1258, 0.3743, 0.3157, 0.0612, 0.4565, 0.2686, 0.3729, 0.3712], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0433, 0.0356, 0.0316, 0.0425, 0.0496, 0.0402, 0.0506], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:39:23,238 INFO [train.py:904] (7/8) Epoch 20, batch 8400, loss[loss=0.1646, simple_loss=0.2507, pruned_loss=0.03924, over 12523.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2795, pruned_loss=0.04964, over 3052764.20 frames. ], batch size: 246, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:39:26,529 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201253.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:40:00,788 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201274.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:40:08,502 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.430e+02 2.319e+02 2.651e+02 3.336e+02 5.393e+02, threshold=5.301e+02, percent-clipped=0.0 2023-05-01 03:40:44,373 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201301.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:40:45,037 INFO [train.py:904] (7/8) Epoch 20, batch 8450, loss[loss=0.1938, simple_loss=0.2765, pruned_loss=0.05557, over 12126.00 frames. ], tot_loss[loss=0.1872, simple_loss=0.2779, pruned_loss=0.04827, over 3049002.31 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:41:05,849 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201314.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:41:39,755 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201335.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:42:02,486 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201349.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:42:07,350 INFO [train.py:904] (7/8) Epoch 20, batch 8500, loss[loss=0.1834, simple_loss=0.2716, pruned_loss=0.04765, over 16703.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.274, pruned_loss=0.04581, over 3067633.03 frames. ], batch size: 76, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:42:50,128 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.179e+02 2.765e+02 3.393e+02 6.239e+02, threshold=5.531e+02, percent-clipped=4.0 2023-05-01 03:43:31,066 INFO [train.py:904] (7/8) Epoch 20, batch 8550, loss[loss=0.1794, simple_loss=0.2702, pruned_loss=0.04425, over 16475.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2721, pruned_loss=0.04455, over 3072076.84 frames. ], batch size: 68, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:44:19,135 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201427.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:44:41,299 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5961, 2.5754, 2.3087, 2.4665, 2.9953, 2.7065, 3.1089, 3.2308], device='cuda:7'), covar=tensor([0.0118, 0.0465, 0.0509, 0.0409, 0.0264, 0.0390, 0.0233, 0.0233], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0223, 0.0215, 0.0215, 0.0224, 0.0224, 0.0222, 0.0218], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:45:09,663 INFO [train.py:904] (7/8) Epoch 20, batch 8600, loss[loss=0.1713, simple_loss=0.2671, pruned_loss=0.03772, over 16866.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2724, pruned_loss=0.04387, over 3061419.95 frames. ], batch size: 90, lr: 3.35e-03, grad_scale: 8.0 2023-05-01 03:45:18,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7520, 3.7422, 2.3224, 4.2673, 2.8321, 4.1937, 2.4589, 3.0751], device='cuda:7'), covar=tensor([0.0230, 0.0328, 0.1440, 0.0179, 0.0759, 0.0406, 0.1411, 0.0671], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0168, 0.0186, 0.0150, 0.0169, 0.0206, 0.0194, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-01 03:45:27,584 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.45 vs. limit=5.0 2023-05-01 03:46:02,942 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.180e+02 2.549e+02 3.038e+02 6.489e+02, threshold=5.098e+02, percent-clipped=1.0 2023-05-01 03:46:48,565 INFO [train.py:904] (7/8) Epoch 20, batch 8650, loss[loss=0.1654, simple_loss=0.2567, pruned_loss=0.03704, over 12027.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2701, pruned_loss=0.04237, over 3045442.06 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:47:40,631 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-01 03:48:36,198 INFO [train.py:904] (7/8) Epoch 20, batch 8700, loss[loss=0.1834, simple_loss=0.2666, pruned_loss=0.05009, over 12578.00 frames. ], tot_loss[loss=0.1748, simple_loss=0.2676, pruned_loss=0.04105, over 3042625.69 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:49:28,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.047e+02 2.487e+02 3.171e+02 4.995e+02, threshold=4.975e+02, percent-clipped=0.0 2023-05-01 03:50:13,610 INFO [train.py:904] (7/8) Epoch 20, batch 8750, loss[loss=0.1937, simple_loss=0.2889, pruned_loss=0.04922, over 16239.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2675, pruned_loss=0.04043, over 3041492.01 frames. ], batch size: 165, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:50:31,553 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201609.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:50:53,765 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5647, 2.9813, 3.2511, 1.8655, 2.8264, 2.1608, 3.1452, 3.2020], device='cuda:7'), covar=tensor([0.0265, 0.0855, 0.0531, 0.2113, 0.0812, 0.0971, 0.0709, 0.0946], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0155, 0.0160, 0.0148, 0.0140, 0.0125, 0.0140, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:51:22,020 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=201630.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:51:51,033 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201644.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:52:07,688 INFO [train.py:904] (7/8) Epoch 20, batch 8800, loss[loss=0.1735, simple_loss=0.2613, pruned_loss=0.04285, over 15481.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2666, pruned_loss=0.03972, over 3042852.03 frames. ], batch size: 193, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:52:13,916 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6080, 3.6474, 3.4436, 3.1889, 3.2884, 3.5564, 3.3748, 3.4206], device='cuda:7'), covar=tensor([0.0537, 0.0613, 0.0262, 0.0277, 0.0551, 0.0438, 0.1351, 0.0483], device='cuda:7'), in_proj_covar=tensor([0.0276, 0.0395, 0.0318, 0.0315, 0.0328, 0.0364, 0.0220, 0.0380], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:52:18,438 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0422, 4.0249, 3.9623, 3.3100, 3.9680, 1.6902, 3.7784, 3.6214], device='cuda:7'), covar=tensor([0.0115, 0.0099, 0.0171, 0.0263, 0.0104, 0.2661, 0.0132, 0.0234], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0146, 0.0187, 0.0170, 0.0165, 0.0200, 0.0178, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:52:25,376 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201661.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:52:40,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9522, 2.1213, 2.3742, 3.2125, 2.2096, 2.3004, 2.3103, 2.1811], device='cuda:7'), covar=tensor([0.1315, 0.3789, 0.2627, 0.0688, 0.4224, 0.2812, 0.3500, 0.3794], device='cuda:7'), in_proj_covar=tensor([0.0386, 0.0432, 0.0355, 0.0314, 0.0425, 0.0495, 0.0401, 0.0504], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:53:06,177 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 03:53:06,518 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.243e+02 2.623e+02 3.212e+02 5.763e+02, threshold=5.246e+02, percent-clipped=4.0 2023-05-01 03:53:51,637 INFO [train.py:904] (7/8) Epoch 20, batch 8850, loss[loss=0.1676, simple_loss=0.2708, pruned_loss=0.03221, over 16457.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2691, pruned_loss=0.03897, over 3044842.73 frames. ], batch size: 147, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 03:53:56,477 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9799, 1.7820, 1.6834, 1.4564, 1.9469, 1.5354, 1.5568, 1.8780], device='cuda:7'), covar=tensor([0.0156, 0.0286, 0.0415, 0.0366, 0.0244, 0.0314, 0.0150, 0.0223], device='cuda:7'), in_proj_covar=tensor([0.0190, 0.0222, 0.0214, 0.0215, 0.0223, 0.0222, 0.0219, 0.0217], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 03:53:57,931 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201705.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:00,624 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201706.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:34,722 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201722.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:45,986 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201727.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:54:46,278 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 03:55:16,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0026, 2.6438, 2.9433, 2.0090, 2.7010, 2.0880, 2.7737, 2.8647], device='cuda:7'), covar=tensor([0.0292, 0.0961, 0.0495, 0.1974, 0.0765, 0.0981, 0.0681, 0.0973], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0156, 0.0161, 0.0148, 0.0140, 0.0126, 0.0140, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 03:55:38,109 INFO [train.py:904] (7/8) Epoch 20, batch 8900, loss[loss=0.1891, simple_loss=0.2827, pruned_loss=0.04776, over 15311.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2689, pruned_loss=0.03857, over 3026405.38 frames. ], batch size: 190, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 03:56:08,843 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=201767.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:56:27,848 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201775.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 03:56:46,609 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.216e+02 2.705e+02 3.264e+02 5.822e+02, threshold=5.410e+02, percent-clipped=1.0 2023-05-01 03:57:41,712 INFO [train.py:904] (7/8) Epoch 20, batch 8950, loss[loss=0.1766, simple_loss=0.2637, pruned_loss=0.04477, over 12597.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2685, pruned_loss=0.03883, over 3038947.32 frames. ], batch size: 246, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 03:59:30,705 INFO [train.py:904] (7/8) Epoch 20, batch 9000, loss[loss=0.1509, simple_loss=0.2427, pruned_loss=0.02955, over 16817.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2651, pruned_loss=0.03743, over 3051336.10 frames. ], batch size: 83, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 03:59:30,705 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 03:59:41,117 INFO [train.py:938] (7/8) Epoch 20, validation: loss=0.1464, simple_loss=0.2502, pruned_loss=0.02125, over 944034.00 frames. 2023-05-01 03:59:41,117 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 04:00:29,099 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-01 04:00:41,797 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.083e+02 2.552e+02 3.280e+02 1.556e+03, threshold=5.104e+02, percent-clipped=3.0 2023-05-01 04:01:24,591 INFO [train.py:904] (7/8) Epoch 20, batch 9050, loss[loss=0.1545, simple_loss=0.2473, pruned_loss=0.03083, over 17057.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2658, pruned_loss=0.03775, over 3045918.67 frames. ], batch size: 53, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 04:01:40,837 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:02:21,299 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=201930.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:03:12,554 INFO [train.py:904] (7/8) Epoch 20, batch 9100, loss[loss=0.1664, simple_loss=0.2553, pruned_loss=0.03877, over 12707.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2656, pruned_loss=0.03816, over 3072562.48 frames. ], batch size: 248, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 04:03:22,667 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201957.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:03:22,697 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8061, 3.8094, 3.9940, 3.7275, 3.8949, 4.3101, 3.9623, 3.6512], device='cuda:7'), covar=tensor([0.2000, 0.2620, 0.2263, 0.2807, 0.2806, 0.1604, 0.1714, 0.2806], device='cuda:7'), in_proj_covar=tensor([0.0382, 0.0557, 0.0612, 0.0460, 0.0609, 0.0640, 0.0479, 0.0619], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:03:32,174 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5757, 3.5466, 3.5301, 2.8276, 3.4663, 1.9663, 3.2554, 2.9060], device='cuda:7'), covar=tensor([0.0131, 0.0109, 0.0165, 0.0208, 0.0113, 0.2426, 0.0118, 0.0234], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0145, 0.0185, 0.0167, 0.0164, 0.0199, 0.0176, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:04:12,540 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 04:04:14,332 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=201978.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:04:23,058 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.025e+02 2.515e+02 2.947e+02 5.747e+02, threshold=5.029e+02, percent-clipped=1.0 2023-05-01 04:04:43,621 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=201990.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:05:11,358 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202000.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:05:14,327 INFO [train.py:904] (7/8) Epoch 20, batch 9150, loss[loss=0.1673, simple_loss=0.2566, pruned_loss=0.03902, over 16936.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2656, pruned_loss=0.03796, over 3051022.54 frames. ], batch size: 109, lr: 3.35e-03, grad_scale: 2.0 2023-05-01 04:05:48,718 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202017.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:06:44,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3753, 3.5160, 3.6865, 3.6681, 3.6842, 3.5186, 3.5413, 3.5712], device='cuda:7'), covar=tensor([0.0359, 0.0788, 0.0543, 0.0467, 0.0464, 0.0587, 0.0734, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0426, 0.0416, 0.0386, 0.0460, 0.0435, 0.0519, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 04:06:45,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6179, 3.5975, 3.5773, 2.9100, 3.4841, 1.9987, 3.2386, 2.9175], device='cuda:7'), covar=tensor([0.0129, 0.0125, 0.0183, 0.0208, 0.0109, 0.2375, 0.0129, 0.0230], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0144, 0.0184, 0.0166, 0.0164, 0.0198, 0.0175, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:06:49,009 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 04:06:59,308 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=202051.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:06:59,893 INFO [train.py:904] (7/8) Epoch 20, batch 9200, loss[loss=0.1453, simple_loss=0.23, pruned_loss=0.03028, over 12082.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.261, pruned_loss=0.03689, over 3034397.56 frames. ], batch size: 246, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:07:20,710 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202062.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:07:55,481 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.302e+02 2.670e+02 3.574e+02 1.158e+03, threshold=5.341e+02, percent-clipped=6.0 2023-05-01 04:08:33,860 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.53 vs. limit=2.0 2023-05-01 04:08:37,609 INFO [train.py:904] (7/8) Epoch 20, batch 9250, loss[loss=0.1809, simple_loss=0.2826, pruned_loss=0.03966, over 16224.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2611, pruned_loss=0.03709, over 3053251.81 frames. ], batch size: 165, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:10:29,973 INFO [train.py:904] (7/8) Epoch 20, batch 9300, loss[loss=0.1507, simple_loss=0.2457, pruned_loss=0.0278, over 16594.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2598, pruned_loss=0.0368, over 3039984.87 frames. ], batch size: 62, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:11:37,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 1.963e+02 2.278e+02 2.874e+02 5.860e+02, threshold=4.556e+02, percent-clipped=2.0 2023-05-01 04:11:45,573 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-05-01 04:12:16,150 INFO [train.py:904] (7/8) Epoch 20, batch 9350, loss[loss=0.1866, simple_loss=0.2868, pruned_loss=0.04322, over 16902.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2598, pruned_loss=0.03676, over 3035129.92 frames. ], batch size: 116, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:13:59,191 INFO [train.py:904] (7/8) Epoch 20, batch 9400, loss[loss=0.1653, simple_loss=0.2709, pruned_loss=0.02992, over 16908.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2596, pruned_loss=0.03649, over 3031590.04 frames. ], batch size: 96, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:14:20,306 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-01 04:15:00,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.064e+02 2.476e+02 2.960e+02 4.580e+02, threshold=4.951e+02, percent-clipped=1.0 2023-05-01 04:15:39,409 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202300.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:15:41,544 INFO [train.py:904] (7/8) Epoch 20, batch 9450, loss[loss=0.1578, simple_loss=0.262, pruned_loss=0.02683, over 17056.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2611, pruned_loss=0.03661, over 3019206.00 frames. ], batch size: 50, lr: 3.35e-03, grad_scale: 4.0 2023-05-01 04:16:13,667 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202317.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:16:16,337 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2645, 1.6089, 1.8505, 2.2048, 2.2450, 2.4205, 1.7539, 2.4107], device='cuda:7'), covar=tensor([0.0245, 0.0502, 0.0357, 0.0316, 0.0339, 0.0214, 0.0507, 0.0143], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0185, 0.0171, 0.0175, 0.0188, 0.0144, 0.0188, 0.0140], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:17:14,538 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=202346.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:18,058 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202348.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:25,758 INFO [train.py:904] (7/8) Epoch 20, batch 9500, loss[loss=0.1586, simple_loss=0.2575, pruned_loss=0.02986, over 16702.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2604, pruned_loss=0.03608, over 3055591.81 frames. ], batch size: 83, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:17:49,822 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202362.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:17:55,125 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202365.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:18:26,629 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.130e+02 2.550e+02 3.364e+02 1.443e+03, threshold=5.099e+02, percent-clipped=7.0 2023-05-01 04:18:56,041 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 04:19:13,227 INFO [train.py:904] (7/8) Epoch 20, batch 9550, loss[loss=0.1846, simple_loss=0.2852, pruned_loss=0.04201, over 16892.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2606, pruned_loss=0.03647, over 3058989.69 frames. ], batch size: 116, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:19:32,679 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202410.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:19:58,681 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8189, 3.8053, 3.9635, 3.7465, 3.8869, 4.3263, 3.9876, 3.6681], device='cuda:7'), covar=tensor([0.2105, 0.2377, 0.2189, 0.2426, 0.2784, 0.1392, 0.1495, 0.2439], device='cuda:7'), in_proj_covar=tensor([0.0377, 0.0552, 0.0608, 0.0456, 0.0607, 0.0637, 0.0478, 0.0612], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:20:55,246 INFO [train.py:904] (7/8) Epoch 20, batch 9600, loss[loss=0.1795, simple_loss=0.2784, pruned_loss=0.04035, over 16417.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2619, pruned_loss=0.0372, over 3050388.41 frames. ], batch size: 146, lr: 3.34e-03, grad_scale: 8.0 2023-05-01 04:21:53,898 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.243e+02 2.599e+02 3.368e+02 6.147e+02, threshold=5.198e+02, percent-clipped=3.0 2023-05-01 04:22:27,974 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 04:22:29,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7714, 4.7935, 4.6490, 4.3351, 4.3424, 4.7455, 4.5671, 4.4810], device='cuda:7'), covar=tensor([0.0627, 0.0742, 0.0320, 0.0296, 0.0835, 0.0541, 0.0404, 0.0678], device='cuda:7'), in_proj_covar=tensor([0.0272, 0.0388, 0.0315, 0.0310, 0.0320, 0.0358, 0.0217, 0.0373], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 04:22:41,349 INFO [train.py:904] (7/8) Epoch 20, batch 9650, loss[loss=0.1689, simple_loss=0.2585, pruned_loss=0.03963, over 12278.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2641, pruned_loss=0.03741, over 3057072.43 frames. ], batch size: 250, lr: 3.34e-03, grad_scale: 8.0 2023-05-01 04:24:31,601 INFO [train.py:904] (7/8) Epoch 20, batch 9700, loss[loss=0.157, simple_loss=0.256, pruned_loss=0.02901, over 16749.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.263, pruned_loss=0.03696, over 3074583.31 frames. ], batch size: 83, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:25:37,515 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.240e+02 2.594e+02 3.217e+02 6.091e+02, threshold=5.188e+02, percent-clipped=1.0 2023-05-01 04:26:15,974 INFO [train.py:904] (7/8) Epoch 20, batch 9750, loss[loss=0.1543, simple_loss=0.2447, pruned_loss=0.03191, over 16584.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2627, pruned_loss=0.03763, over 3071100.95 frames. ], batch size: 57, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:27:12,795 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7566, 3.6118, 3.7151, 3.9170, 3.9906, 3.6252, 3.9875, 4.0107], device='cuda:7'), covar=tensor([0.1684, 0.1360, 0.1682, 0.0904, 0.0831, 0.1955, 0.0848, 0.0983], device='cuda:7'), in_proj_covar=tensor([0.0590, 0.0725, 0.0845, 0.0745, 0.0563, 0.0582, 0.0599, 0.0694], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:27:23,758 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6001, 4.6858, 4.5030, 4.1240, 4.2573, 4.6041, 4.3916, 4.2964], device='cuda:7'), covar=tensor([0.0606, 0.0411, 0.0293, 0.0306, 0.0769, 0.0356, 0.0433, 0.0630], device='cuda:7'), in_proj_covar=tensor([0.0272, 0.0387, 0.0314, 0.0310, 0.0319, 0.0357, 0.0216, 0.0373], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 04:27:47,611 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=202646.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:27:57,089 INFO [train.py:904] (7/8) Epoch 20, batch 9800, loss[loss=0.1752, simple_loss=0.2783, pruned_loss=0.03605, over 16437.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2631, pruned_loss=0.03651, over 3097122.96 frames. ], batch size: 146, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:28:32,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1845, 2.1203, 2.2015, 3.7364, 2.0796, 2.4538, 2.2588, 2.2974], device='cuda:7'), covar=tensor([0.1305, 0.3767, 0.3079, 0.0563, 0.4322, 0.2624, 0.3778, 0.3527], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0431, 0.0357, 0.0315, 0.0425, 0.0493, 0.0402, 0.0502], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:28:57,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.120e+02 2.499e+02 2.833e+02 7.051e+02, threshold=4.998e+02, percent-clipped=2.0 2023-05-01 04:29:03,868 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0040, 4.2622, 4.0930, 4.1416, 3.8141, 3.8543, 3.8211, 4.2626], device='cuda:7'), covar=tensor([0.1096, 0.0865, 0.0906, 0.0736, 0.0787, 0.1554, 0.0979, 0.0949], device='cuda:7'), in_proj_covar=tensor([0.0623, 0.0759, 0.0622, 0.0570, 0.0485, 0.0493, 0.0638, 0.0592], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:29:03,956 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9316, 2.7634, 2.6620, 1.9328, 2.5647, 2.8436, 2.6431, 1.9308], device='cuda:7'), covar=tensor([0.0444, 0.0067, 0.0067, 0.0361, 0.0128, 0.0091, 0.0093, 0.0426], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0078, 0.0077, 0.0129, 0.0094, 0.0104, 0.0088, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:29:26,657 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=202694.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:29:42,368 INFO [train.py:904] (7/8) Epoch 20, batch 9850, loss[loss=0.1686, simple_loss=0.2646, pruned_loss=0.03631, over 15334.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2642, pruned_loss=0.03636, over 3089793.95 frames. ], batch size: 190, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:30:27,852 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5789, 3.6168, 2.7431, 2.1753, 2.2466, 2.3984, 3.7804, 3.1337], device='cuda:7'), covar=tensor([0.3073, 0.0646, 0.1918, 0.3355, 0.3004, 0.2118, 0.0417, 0.1355], device='cuda:7'), in_proj_covar=tensor([0.0316, 0.0260, 0.0295, 0.0299, 0.0281, 0.0249, 0.0283, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:30:32,482 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8884, 2.6462, 2.8651, 1.9750, 2.6240, 2.1001, 2.6917, 2.8418], device='cuda:7'), covar=tensor([0.0257, 0.0899, 0.0491, 0.1845, 0.0807, 0.0938, 0.0626, 0.0715], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0153, 0.0160, 0.0147, 0.0138, 0.0124, 0.0139, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 04:30:56,073 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 04:31:20,924 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9305, 4.9650, 4.8170, 4.3767, 4.5270, 4.8762, 4.7883, 4.5792], device='cuda:7'), covar=tensor([0.0549, 0.0342, 0.0279, 0.0301, 0.0759, 0.0366, 0.0276, 0.0573], device='cuda:7'), in_proj_covar=tensor([0.0271, 0.0385, 0.0313, 0.0308, 0.0318, 0.0355, 0.0215, 0.0372], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 04:31:34,898 INFO [train.py:904] (7/8) Epoch 20, batch 9900, loss[loss=0.1699, simple_loss=0.2746, pruned_loss=0.03256, over 16834.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2647, pruned_loss=0.03656, over 3075926.83 frames. ], batch size: 76, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:32:16,588 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3938, 3.5653, 2.7166, 2.0703, 2.2565, 2.3262, 3.7772, 3.1192], device='cuda:7'), covar=tensor([0.3024, 0.0565, 0.1685, 0.2763, 0.2795, 0.2089, 0.0358, 0.1299], device='cuda:7'), in_proj_covar=tensor([0.0315, 0.0259, 0.0294, 0.0298, 0.0280, 0.0248, 0.0282, 0.0321], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:32:49,189 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.190e+02 2.483e+02 3.109e+02 6.841e+02, threshold=4.966e+02, percent-clipped=3.0 2023-05-01 04:33:30,689 INFO [train.py:904] (7/8) Epoch 20, batch 9950, loss[loss=0.173, simple_loss=0.2632, pruned_loss=0.04142, over 12517.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2669, pruned_loss=0.037, over 3078264.09 frames. ], batch size: 250, lr: 3.34e-03, grad_scale: 2.0 2023-05-01 04:34:42,278 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=202830.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 04:35:32,310 INFO [train.py:904] (7/8) Epoch 20, batch 10000, loss[loss=0.1753, simple_loss=0.2764, pruned_loss=0.03714, over 16175.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2657, pruned_loss=0.03679, over 3090288.21 frames. ], batch size: 165, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:36:35,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3803, 5.7055, 5.4873, 5.5295, 5.1596, 5.1988, 5.0323, 5.7925], device='cuda:7'), covar=tensor([0.1087, 0.0824, 0.0846, 0.0739, 0.0765, 0.0655, 0.1018, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0620, 0.0758, 0.0619, 0.0570, 0.0484, 0.0493, 0.0636, 0.0590], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:36:36,183 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 2.233e+02 2.660e+02 3.205e+02 6.054e+02, threshold=5.319e+02, percent-clipped=4.0 2023-05-01 04:36:52,312 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=202891.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 04:37:11,287 INFO [train.py:904] (7/8) Epoch 20, batch 10050, loss[loss=0.1651, simple_loss=0.2617, pruned_loss=0.03426, over 16759.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2656, pruned_loss=0.03701, over 3076681.28 frames. ], batch size: 83, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:38:41,715 INFO [train.py:904] (7/8) Epoch 20, batch 10100, loss[loss=0.1701, simple_loss=0.2598, pruned_loss=0.04019, over 16929.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2656, pruned_loss=0.03697, over 3090426.49 frames. ], batch size: 109, lr: 3.34e-03, grad_scale: 4.0 2023-05-01 04:39:11,217 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 04:39:40,384 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.271e+02 2.682e+02 3.301e+02 6.522e+02, threshold=5.364e+02, percent-clipped=2.0 2023-05-01 04:39:56,178 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.25 vs. limit=5.0 2023-05-01 04:40:23,120 INFO [train.py:904] (7/8) Epoch 21, batch 0, loss[loss=0.228, simple_loss=0.2937, pruned_loss=0.08115, over 16716.00 frames. ], tot_loss[loss=0.228, simple_loss=0.2937, pruned_loss=0.08115, over 16716.00 frames. ], batch size: 134, lr: 3.26e-03, grad_scale: 8.0 2023-05-01 04:40:23,120 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 04:40:30,888 INFO [train.py:938] (7/8) Epoch 21, validation: loss=0.1451, simple_loss=0.2491, pruned_loss=0.02058, over 944034.00 frames. 2023-05-01 04:40:30,889 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 04:41:07,271 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1123, 4.8553, 5.1284, 5.2864, 5.4646, 4.7957, 5.4610, 5.4135], device='cuda:7'), covar=tensor([0.1871, 0.1232, 0.1648, 0.0706, 0.0599, 0.0784, 0.0489, 0.0695], device='cuda:7'), in_proj_covar=tensor([0.0594, 0.0726, 0.0847, 0.0750, 0.0565, 0.0582, 0.0603, 0.0699], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:41:19,636 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203038.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:41:36,914 INFO [train.py:904] (7/8) Epoch 21, batch 50, loss[loss=0.1892, simple_loss=0.2673, pruned_loss=0.0556, over 16728.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.272, pruned_loss=0.05171, over 753417.78 frames. ], batch size: 134, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:42:25,275 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203085.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:42:26,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.424e+02 2.967e+02 3.842e+02 7.470e+02, threshold=5.934e+02, percent-clipped=2.0 2023-05-01 04:42:28,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6304, 4.6406, 4.5841, 4.1635, 4.5768, 1.8927, 4.3960, 4.3013], device='cuda:7'), covar=tensor([0.0110, 0.0105, 0.0175, 0.0280, 0.0117, 0.2577, 0.0140, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0147, 0.0186, 0.0167, 0.0166, 0.0201, 0.0177, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:42:43,762 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203099.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:42:48,531 INFO [train.py:904] (7/8) Epoch 21, batch 100, loss[loss=0.2353, simple_loss=0.3173, pruned_loss=0.07665, over 12277.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2688, pruned_loss=0.04945, over 1316726.11 frames. ], batch size: 246, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:43:11,171 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 04:43:50,321 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203146.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:43:57,358 INFO [train.py:904] (7/8) Epoch 21, batch 150, loss[loss=0.194, simple_loss=0.2852, pruned_loss=0.05143, over 16484.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2649, pruned_loss=0.04672, over 1750169.34 frames. ], batch size: 75, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:44:31,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 04:44:39,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1375, 4.8159, 5.1608, 5.3092, 5.5411, 4.8520, 5.4646, 5.4850], device='cuda:7'), covar=tensor([0.1954, 0.1351, 0.1826, 0.0886, 0.0564, 0.0891, 0.0599, 0.0684], device='cuda:7'), in_proj_covar=tensor([0.0606, 0.0742, 0.0865, 0.0766, 0.0575, 0.0594, 0.0615, 0.0712], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:44:44,686 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.249e+02 2.595e+02 3.084e+02 6.012e+02, threshold=5.190e+02, percent-clipped=1.0 2023-05-01 04:44:45,063 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203186.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 04:45:04,864 INFO [train.py:904] (7/8) Epoch 21, batch 200, loss[loss=0.1887, simple_loss=0.2794, pruned_loss=0.04903, over 16442.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.265, pruned_loss=0.04624, over 2094263.02 frames. ], batch size: 68, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:45:05,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5430, 2.3581, 2.3956, 4.4194, 2.2716, 2.7532, 2.4081, 2.5317], device='cuda:7'), covar=tensor([0.1142, 0.3641, 0.2889, 0.0452, 0.4097, 0.2483, 0.3625, 0.3482], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0435, 0.0360, 0.0318, 0.0428, 0.0496, 0.0405, 0.0507], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:46:15,297 INFO [train.py:904] (7/8) Epoch 21, batch 250, loss[loss=0.1883, simple_loss=0.2675, pruned_loss=0.05452, over 16862.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2617, pruned_loss=0.04497, over 2373699.68 frames. ], batch size: 96, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:46:37,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9455, 4.1084, 2.6712, 4.6012, 3.2568, 4.5397, 2.6886, 3.3605], device='cuda:7'), covar=tensor([0.0289, 0.0342, 0.1497, 0.0311, 0.0770, 0.0512, 0.1394, 0.0670], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0174, 0.0193, 0.0157, 0.0175, 0.0210, 0.0203, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 04:46:44,173 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0176, 5.1168, 5.5502, 5.5144, 5.5426, 5.2016, 5.0651, 4.8745], device='cuda:7'), covar=tensor([0.0367, 0.0566, 0.0374, 0.0407, 0.0487, 0.0377, 0.1073, 0.0502], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0429, 0.0416, 0.0388, 0.0460, 0.0438, 0.0521, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 04:46:47,750 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3400, 2.5018, 2.8371, 3.1967, 2.9938, 3.6934, 2.7599, 3.7104], device='cuda:7'), covar=tensor([0.0201, 0.0397, 0.0290, 0.0267, 0.0304, 0.0153, 0.0388, 0.0134], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0187, 0.0174, 0.0178, 0.0190, 0.0147, 0.0191, 0.0142], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 04:47:00,834 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.163e+02 2.580e+02 3.064e+02 7.723e+02, threshold=5.160e+02, percent-clipped=4.0 2023-05-01 04:47:23,332 INFO [train.py:904] (7/8) Epoch 21, batch 300, loss[loss=0.1499, simple_loss=0.2293, pruned_loss=0.03521, over 15782.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2593, pruned_loss=0.04409, over 2578115.61 frames. ], batch size: 35, lr: 3.26e-03, grad_scale: 2.0 2023-05-01 04:47:45,466 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.68 vs. limit=5.0 2023-05-01 04:47:51,473 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 04:48:10,418 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 04:48:32,605 INFO [train.py:904] (7/8) Epoch 21, batch 350, loss[loss=0.1924, simple_loss=0.2901, pruned_loss=0.0474, over 16751.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2576, pruned_loss=0.04322, over 2737965.19 frames. ], batch size: 62, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:48:55,259 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9775, 4.5645, 3.2643, 2.4361, 2.7637, 2.5870, 4.8627, 3.7193], device='cuda:7'), covar=tensor([0.2794, 0.0510, 0.1678, 0.2899, 0.2958, 0.2171, 0.0305, 0.1390], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0266, 0.0301, 0.0306, 0.0288, 0.0254, 0.0290, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:49:17,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.171e+02 2.523e+02 3.052e+02 8.873e+02, threshold=5.047e+02, percent-clipped=2.0 2023-05-01 04:49:24,591 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.85 vs. limit=5.0 2023-05-01 04:49:30,109 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203394.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:49:41,966 INFO [train.py:904] (7/8) Epoch 21, batch 400, loss[loss=0.1923, simple_loss=0.2668, pruned_loss=0.05889, over 16683.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2563, pruned_loss=0.04295, over 2873459.52 frames. ], batch size: 134, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 04:50:37,070 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203441.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:50:48,104 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1093, 5.6247, 5.7530, 5.4646, 5.5701, 6.1635, 5.6554, 5.3812], device='cuda:7'), covar=tensor([0.0887, 0.1978, 0.2672, 0.2273, 0.2657, 0.0970, 0.1580, 0.2240], device='cuda:7'), in_proj_covar=tensor([0.0395, 0.0578, 0.0636, 0.0477, 0.0637, 0.0669, 0.0500, 0.0639], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:50:50,735 INFO [train.py:904] (7/8) Epoch 21, batch 450, loss[loss=0.1589, simple_loss=0.2417, pruned_loss=0.03809, over 16726.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2549, pruned_loss=0.04239, over 2972144.58 frames. ], batch size: 89, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 04:50:55,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6528, 3.8616, 2.9293, 2.1851, 2.4724, 2.3791, 3.9403, 3.2878], device='cuda:7'), covar=tensor([0.2783, 0.0595, 0.1683, 0.3091, 0.2730, 0.2138, 0.0531, 0.1541], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0267, 0.0303, 0.0308, 0.0291, 0.0256, 0.0291, 0.0333], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:51:10,992 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 04:51:11,839 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203467.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:51:37,698 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.094e+02 2.458e+02 3.011e+02 6.519e+02, threshold=4.916e+02, percent-clipped=3.0 2023-05-01 04:51:38,092 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=203486.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 04:51:59,116 INFO [train.py:904] (7/8) Epoch 21, batch 500, loss[loss=0.1761, simple_loss=0.2694, pruned_loss=0.0414, over 16695.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2537, pruned_loss=0.04161, over 3040713.16 frames. ], batch size: 62, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:52:36,272 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203528.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:52:43,133 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=203534.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 04:53:08,045 INFO [train.py:904] (7/8) Epoch 21, batch 550, loss[loss=0.1758, simple_loss=0.2566, pruned_loss=0.0475, over 16775.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2531, pruned_loss=0.04134, over 3107024.03 frames. ], batch size: 83, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:53:09,687 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8581, 4.2921, 4.5223, 2.5411, 4.7283, 4.8570, 3.4892, 3.5823], device='cuda:7'), covar=tensor([0.0947, 0.0148, 0.0193, 0.1083, 0.0075, 0.0116, 0.0371, 0.0455], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0108, 0.0095, 0.0139, 0.0078, 0.0122, 0.0127, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 04:53:52,878 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 04:53:56,916 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.204e+02 2.652e+02 3.296e+02 1.402e+03, threshold=5.303e+02, percent-clipped=5.0 2023-05-01 04:54:17,589 INFO [train.py:904] (7/8) Epoch 21, batch 600, loss[loss=0.1894, simple_loss=0.2608, pruned_loss=0.05903, over 16876.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2525, pruned_loss=0.04154, over 3145959.78 frames. ], batch size: 116, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:55:27,204 INFO [train.py:904] (7/8) Epoch 21, batch 650, loss[loss=0.165, simple_loss=0.2452, pruned_loss=0.04234, over 15568.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2512, pruned_loss=0.04123, over 3187226.82 frames. ], batch size: 191, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:55:50,963 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 04:56:01,827 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0409, 5.5248, 5.6560, 5.3850, 5.4278, 6.0430, 5.5765, 5.3028], device='cuda:7'), covar=tensor([0.1016, 0.1891, 0.2183, 0.2347, 0.2942, 0.0999, 0.1432, 0.2167], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0584, 0.0641, 0.0483, 0.0644, 0.0674, 0.0503, 0.0646], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 04:56:14,241 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.016e+02 2.395e+02 3.216e+02 7.414e+02, threshold=4.790e+02, percent-clipped=2.0 2023-05-01 04:56:22,668 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-05-01 04:56:24,918 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=203694.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:56:34,662 INFO [train.py:904] (7/8) Epoch 21, batch 700, loss[loss=0.1799, simple_loss=0.2562, pruned_loss=0.0518, over 16514.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2507, pruned_loss=0.04063, over 3219892.66 frames. ], batch size: 146, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:57:30,260 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=203741.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:57:31,299 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=203742.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:57:38,326 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203747.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:57:44,461 INFO [train.py:904] (7/8) Epoch 21, batch 750, loss[loss=0.1886, simple_loss=0.2659, pruned_loss=0.05564, over 16447.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.251, pruned_loss=0.04099, over 3245783.27 frames. ], batch size: 75, lr: 3.25e-03, grad_scale: 2.0 2023-05-01 04:58:03,823 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203766.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:58:33,074 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.172e+02 2.611e+02 3.083e+02 3.128e+03, threshold=5.221e+02, percent-clipped=7.0 2023-05-01 04:58:36,276 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=203789.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:58:54,761 INFO [train.py:904] (7/8) Epoch 21, batch 800, loss[loss=0.1796, simple_loss=0.2655, pruned_loss=0.04683, over 17055.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2506, pruned_loss=0.04072, over 3268988.87 frames. ], batch size: 55, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 04:59:03,718 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203808.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:59:24,822 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=203823.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:59:29,203 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2023-05-01 04:59:30,068 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203827.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 04:59:58,253 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203847.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:00:04,934 INFO [train.py:904] (7/8) Epoch 21, batch 850, loss[loss=0.1701, simple_loss=0.2506, pruned_loss=0.04478, over 16803.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2499, pruned_loss=0.0401, over 3286368.59 frames. ], batch size: 96, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:00:10,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0626, 5.5857, 5.6682, 5.3826, 5.4836, 6.0618, 5.5894, 5.2850], device='cuda:7'), covar=tensor([0.1037, 0.1937, 0.2297, 0.2106, 0.2589, 0.1094, 0.1480, 0.2250], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0589, 0.0649, 0.0487, 0.0650, 0.0680, 0.0509, 0.0653], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:00:29,520 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 05:00:53,314 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 1.964e+02 2.295e+02 2.674e+02 5.967e+02, threshold=4.591e+02, percent-clipped=3.0 2023-05-01 05:01:11,333 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=203900.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:01:12,971 INFO [train.py:904] (7/8) Epoch 21, batch 900, loss[loss=0.1704, simple_loss=0.248, pruned_loss=0.04638, over 16242.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2503, pruned_loss=0.03983, over 3299796.30 frames. ], batch size: 165, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:01:21,698 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203908.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:01:30,241 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-05-01 05:02:01,429 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1810, 2.0847, 1.7106, 1.8547, 2.3164, 2.0684, 2.1663, 2.4110], device='cuda:7'), covar=tensor([0.0254, 0.0407, 0.0540, 0.0445, 0.0240, 0.0371, 0.0216, 0.0286], device='cuda:7'), in_proj_covar=tensor([0.0206, 0.0238, 0.0227, 0.0227, 0.0237, 0.0235, 0.0238, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:02:07,258 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-05-01 05:02:21,446 INFO [train.py:904] (7/8) Epoch 21, batch 950, loss[loss=0.1682, simple_loss=0.2548, pruned_loss=0.04084, over 17261.00 frames. ], tot_loss[loss=0.165, simple_loss=0.25, pruned_loss=0.04001, over 3304502.95 frames. ], batch size: 52, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:02:34,003 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8547, 3.9272, 2.7330, 4.5962, 3.0994, 4.5581, 2.7205, 3.2549], device='cuda:7'), covar=tensor([0.0345, 0.0448, 0.1509, 0.0320, 0.0901, 0.0555, 0.1407, 0.0767], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0178, 0.0195, 0.0162, 0.0178, 0.0215, 0.0204, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:02:35,192 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=203961.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:02:40,864 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7550, 2.6006, 2.5894, 4.0486, 3.1548, 4.0202, 1.6164, 2.8846], device='cuda:7'), covar=tensor([0.1398, 0.0780, 0.1146, 0.0187, 0.0160, 0.0400, 0.1651, 0.0841], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0187, 0.0204, 0.0214, 0.0200, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:03:10,189 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.138e+02 2.475e+02 2.994e+02 7.321e+02, threshold=4.951e+02, percent-clipped=5.0 2023-05-01 05:03:34,207 INFO [train.py:904] (7/8) Epoch 21, batch 1000, loss[loss=0.1683, simple_loss=0.2548, pruned_loss=0.04088, over 16796.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2488, pruned_loss=0.04017, over 3303247.50 frames. ], batch size: 62, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:03:44,785 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8775, 2.0413, 2.3588, 2.6629, 2.7222, 2.6874, 2.0375, 2.9010], device='cuda:7'), covar=tensor([0.0185, 0.0449, 0.0348, 0.0256, 0.0304, 0.0300, 0.0525, 0.0187], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0192, 0.0178, 0.0182, 0.0194, 0.0152, 0.0195, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:04:10,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1817, 2.1254, 2.7475, 3.0316, 2.9021, 3.5851, 2.0909, 3.6350], device='cuda:7'), covar=tensor([0.0220, 0.0547, 0.0308, 0.0310, 0.0325, 0.0174, 0.0634, 0.0134], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0192, 0.0178, 0.0182, 0.0194, 0.0152, 0.0195, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:04:37,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1749, 5.0958, 4.9923, 4.5152, 4.6614, 5.0227, 4.9781, 4.6655], device='cuda:7'), covar=tensor([0.0583, 0.0536, 0.0323, 0.0347, 0.1012, 0.0482, 0.0369, 0.0840], device='cuda:7'), in_proj_covar=tensor([0.0293, 0.0419, 0.0339, 0.0335, 0.0348, 0.0390, 0.0233, 0.0408], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:04:43,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2997, 3.9957, 4.5067, 2.2913, 4.7319, 4.7998, 3.4190, 3.6894], device='cuda:7'), covar=tensor([0.0633, 0.0234, 0.0195, 0.1168, 0.0079, 0.0137, 0.0425, 0.0375], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0108, 0.0096, 0.0139, 0.0079, 0.0122, 0.0128, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:04:43,738 INFO [train.py:904] (7/8) Epoch 21, batch 1050, loss[loss=0.1596, simple_loss=0.2459, pruned_loss=0.03668, over 16838.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2484, pruned_loss=0.03984, over 3316839.56 frames. ], batch size: 42, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:05:09,509 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-05-01 05:05:10,425 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8042, 4.3371, 4.3660, 3.1128, 3.7032, 4.2962, 3.9418, 2.6418], device='cuda:7'), covar=tensor([0.0478, 0.0068, 0.0046, 0.0332, 0.0120, 0.0099, 0.0086, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0084, 0.0083, 0.0135, 0.0099, 0.0110, 0.0094, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 05:05:19,322 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=204077.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:05:34,282 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.155e+02 2.486e+02 2.944e+02 8.933e+02, threshold=4.972e+02, percent-clipped=2.0 2023-05-01 05:05:54,745 INFO [train.py:904] (7/8) Epoch 21, batch 1100, loss[loss=0.1761, simple_loss=0.2501, pruned_loss=0.05103, over 16902.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2472, pruned_loss=0.03957, over 3310739.85 frames. ], batch size: 116, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:05:56,315 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204103.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:06:01,809 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8915, 2.0006, 2.3673, 2.7017, 2.7107, 2.7774, 2.0508, 2.9796], device='cuda:7'), covar=tensor([0.0202, 0.0458, 0.0362, 0.0289, 0.0307, 0.0273, 0.0480, 0.0157], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0192, 0.0179, 0.0183, 0.0194, 0.0152, 0.0195, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:06:24,061 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204122.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:06:25,306 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204123.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:06:45,699 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=204138.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 05:07:05,266 INFO [train.py:904] (7/8) Epoch 21, batch 1150, loss[loss=0.1798, simple_loss=0.247, pruned_loss=0.05636, over 16875.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2469, pruned_loss=0.03912, over 3308107.77 frames. ], batch size: 116, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:07:21,037 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=204163.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:07:32,273 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204171.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:07:53,763 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.065e+02 2.500e+02 2.935e+02 5.015e+02, threshold=5.000e+02, percent-clipped=1.0 2023-05-01 05:08:07,196 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4206, 5.8473, 5.5892, 5.6515, 5.2921, 5.3571, 5.2693, 5.9486], device='cuda:7'), covar=tensor([0.1474, 0.1067, 0.1057, 0.0954, 0.0967, 0.0681, 0.1145, 0.0940], device='cuda:7'), in_proj_covar=tensor([0.0678, 0.0826, 0.0679, 0.0623, 0.0526, 0.0533, 0.0696, 0.0642], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:08:13,795 INFO [train.py:904] (7/8) Epoch 21, batch 1200, loss[loss=0.1521, simple_loss=0.2459, pruned_loss=0.02919, over 17177.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2468, pruned_loss=0.03889, over 3310988.78 frames. ], batch size: 46, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:08:15,246 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204203.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:08:15,368 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9408, 5.0822, 4.8398, 4.4708, 4.1199, 5.0788, 5.0343, 4.5661], device='cuda:7'), covar=tensor([0.0984, 0.0611, 0.0535, 0.0453, 0.1958, 0.0521, 0.0389, 0.0928], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0426, 0.0344, 0.0341, 0.0352, 0.0397, 0.0237, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:08:29,181 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2415, 5.8252, 5.9500, 5.6743, 5.7891, 6.2988, 5.8527, 5.5441], device='cuda:7'), covar=tensor([0.0883, 0.1915, 0.2146, 0.2049, 0.2537, 0.0890, 0.1506, 0.2221], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0594, 0.0654, 0.0491, 0.0656, 0.0683, 0.0515, 0.0659], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:08:44,973 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=204224.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:09:24,360 INFO [train.py:904] (7/8) Epoch 21, batch 1250, loss[loss=0.1483, simple_loss=0.2416, pruned_loss=0.02749, over 16682.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.248, pruned_loss=0.03987, over 3322162.85 frames. ], batch size: 57, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:09:30,691 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204256.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:10:12,299 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.450e+02 2.352e+02 2.804e+02 3.486e+02 9.364e+02, threshold=5.607e+02, percent-clipped=5.0 2023-05-01 05:10:33,933 INFO [train.py:904] (7/8) Epoch 21, batch 1300, loss[loss=0.189, simple_loss=0.2596, pruned_loss=0.05917, over 16413.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2478, pruned_loss=0.03988, over 3317409.10 frames. ], batch size: 146, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:11:42,900 INFO [train.py:904] (7/8) Epoch 21, batch 1350, loss[loss=0.1536, simple_loss=0.2555, pruned_loss=0.02581, over 17262.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2479, pruned_loss=0.03973, over 3309890.70 frames. ], batch size: 52, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:11:48,263 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7623, 4.8562, 5.1911, 5.2138, 5.2147, 4.8963, 4.8482, 4.6356], device='cuda:7'), covar=tensor([0.0375, 0.0594, 0.0497, 0.0400, 0.0495, 0.0417, 0.0836, 0.0505], device='cuda:7'), in_proj_covar=tensor([0.0410, 0.0454, 0.0439, 0.0409, 0.0486, 0.0462, 0.0549, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 05:12:31,607 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.188e+02 2.581e+02 2.996e+02 9.160e+02, threshold=5.162e+02, percent-clipped=1.0 2023-05-01 05:12:52,519 INFO [train.py:904] (7/8) Epoch 21, batch 1400, loss[loss=0.1662, simple_loss=0.2432, pruned_loss=0.04465, over 16630.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2478, pruned_loss=0.03933, over 3323475.29 frames. ], batch size: 76, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:12:54,620 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204403.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:13:20,798 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204422.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:13:35,322 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204433.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:14:00,513 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204451.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:14:01,408 INFO [train.py:904] (7/8) Epoch 21, batch 1450, loss[loss=0.1413, simple_loss=0.2282, pruned_loss=0.02721, over 16833.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2472, pruned_loss=0.03925, over 3329559.37 frames. ], batch size: 42, lr: 3.25e-03, grad_scale: 8.0 2023-05-01 05:14:26,870 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204470.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:14:50,913 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.087e+02 2.507e+02 3.018e+02 6.138e+02, threshold=5.015e+02, percent-clipped=2.0 2023-05-01 05:15:10,561 INFO [train.py:904] (7/8) Epoch 21, batch 1500, loss[loss=0.1454, simple_loss=0.2295, pruned_loss=0.03066, over 16991.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2471, pruned_loss=0.03903, over 3332567.40 frames. ], batch size: 41, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:15:12,770 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204503.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:15:18,822 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3810, 3.3627, 2.1196, 3.5491, 2.6330, 3.5360, 2.1562, 2.7349], device='cuda:7'), covar=tensor([0.0285, 0.0420, 0.1567, 0.0360, 0.0785, 0.0825, 0.1519, 0.0767], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0178, 0.0195, 0.0164, 0.0178, 0.0217, 0.0204, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:15:33,855 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=204519.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:15:37,221 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6686, 4.2304, 4.2399, 3.0225, 3.6114, 4.1685, 3.8183, 2.6128], device='cuda:7'), covar=tensor([0.0511, 0.0084, 0.0055, 0.0355, 0.0124, 0.0117, 0.0088, 0.0435], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0084, 0.0083, 0.0135, 0.0099, 0.0109, 0.0094, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 05:16:17,195 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:16:18,104 INFO [train.py:904] (7/8) Epoch 21, batch 1550, loss[loss=0.127, simple_loss=0.2209, pruned_loss=0.01658, over 17216.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2482, pruned_loss=0.03985, over 3333592.34 frames. ], batch size: 45, lr: 3.25e-03, grad_scale: 4.0 2023-05-01 05:16:23,717 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204556.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 05:17:07,002 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.391e+02 2.852e+02 3.354e+02 7.624e+02, threshold=5.704e+02, percent-clipped=5.0 2023-05-01 05:17:26,293 INFO [train.py:904] (7/8) Epoch 21, batch 1600, loss[loss=0.1661, simple_loss=0.2595, pruned_loss=0.0363, over 17122.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2503, pruned_loss=0.04056, over 3326995.22 frames. ], batch size: 47, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:17:29,638 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204604.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 05:18:13,250 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7863, 2.9628, 3.2487, 2.1160, 2.7726, 2.2119, 3.3316, 3.3498], device='cuda:7'), covar=tensor([0.0261, 0.0899, 0.0570, 0.1828, 0.0842, 0.1019, 0.0573, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0163, 0.0167, 0.0153, 0.0144, 0.0130, 0.0145, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:18:13,455 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 05:18:35,820 INFO [train.py:904] (7/8) Epoch 21, batch 1650, loss[loss=0.1921, simple_loss=0.2613, pruned_loss=0.0614, over 16895.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2518, pruned_loss=0.04094, over 3325150.74 frames. ], batch size: 116, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:19:25,765 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.742e+02 2.251e+02 2.625e+02 3.143e+02 5.667e+02, threshold=5.251e+02, percent-clipped=0.0 2023-05-01 05:19:27,442 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1497, 3.1949, 3.3427, 2.3593, 3.0975, 3.4522, 3.1834, 2.0277], device='cuda:7'), covar=tensor([0.0535, 0.0125, 0.0060, 0.0392, 0.0126, 0.0097, 0.0095, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0084, 0.0083, 0.0135, 0.0099, 0.0110, 0.0094, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 05:19:45,546 INFO [train.py:904] (7/8) Epoch 21, batch 1700, loss[loss=0.1884, simple_loss=0.2664, pruned_loss=0.05524, over 16806.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2533, pruned_loss=0.04116, over 3324809.71 frames. ], batch size: 96, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:20:00,655 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4255, 4.0798, 4.5764, 2.4967, 4.8966, 4.8789, 3.5717, 3.7715], device='cuda:7'), covar=tensor([0.0613, 0.0243, 0.0219, 0.1131, 0.0075, 0.0194, 0.0385, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0108, 0.0096, 0.0138, 0.0079, 0.0122, 0.0127, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 05:20:29,952 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204733.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:20:55,688 INFO [train.py:904] (7/8) Epoch 21, batch 1750, loss[loss=0.184, simple_loss=0.2635, pruned_loss=0.05221, over 16407.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2555, pruned_loss=0.04185, over 3319112.37 frames. ], batch size: 146, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:21:06,622 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 05:21:18,087 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3684, 2.3322, 2.3460, 4.1811, 2.2704, 2.7314, 2.3381, 2.4949], device='cuda:7'), covar=tensor([0.1357, 0.3697, 0.2930, 0.0556, 0.4018, 0.2536, 0.3869, 0.3273], device='cuda:7'), in_proj_covar=tensor([0.0403, 0.0448, 0.0369, 0.0331, 0.0436, 0.0515, 0.0418, 0.0525], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:21:37,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204781.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:21:46,297 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.363e+02 2.773e+02 3.345e+02 7.836e+02, threshold=5.546e+02, percent-clipped=6.0 2023-05-01 05:22:06,989 INFO [train.py:904] (7/8) Epoch 21, batch 1800, loss[loss=0.1684, simple_loss=0.2524, pruned_loss=0.04223, over 16561.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2559, pruned_loss=0.0415, over 3311268.86 frames. ], batch size: 75, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:22:07,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7959, 4.4164, 3.0687, 2.4046, 2.7733, 2.6343, 4.7338, 3.5697], device='cuda:7'), covar=tensor([0.2831, 0.0557, 0.1742, 0.2759, 0.2668, 0.1958, 0.0349, 0.1360], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0270, 0.0304, 0.0309, 0.0295, 0.0257, 0.0294, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:22:30,015 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=204819.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:22:30,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0258, 3.0233, 3.1306, 2.1397, 2.9548, 3.2495, 3.0041, 1.9592], device='cuda:7'), covar=tensor([0.0531, 0.0154, 0.0068, 0.0425, 0.0146, 0.0122, 0.0113, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0085, 0.0083, 0.0136, 0.0100, 0.0110, 0.0094, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0004], device='cuda:7') 2023-05-01 05:22:32,948 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=204821.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:22:39,242 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 05:23:15,229 INFO [train.py:904] (7/8) Epoch 21, batch 1850, loss[loss=0.1682, simple_loss=0.2593, pruned_loss=0.03853, over 16718.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2565, pruned_loss=0.04157, over 3308191.86 frames. ], batch size: 57, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:23:37,480 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=204867.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:23:51,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7906, 3.9910, 3.0262, 2.3383, 2.5347, 2.4023, 4.1285, 3.4091], device='cuda:7'), covar=tensor([0.2715, 0.0581, 0.1727, 0.3074, 0.2845, 0.2062, 0.0474, 0.1563], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0271, 0.0305, 0.0310, 0.0296, 0.0257, 0.0295, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:23:58,573 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=204882.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:24:06,222 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.082e+02 2.497e+02 3.026e+02 6.100e+02, threshold=4.995e+02, percent-clipped=2.0 2023-05-01 05:24:26,180 INFO [train.py:904] (7/8) Epoch 21, batch 1900, loss[loss=0.1421, simple_loss=0.2263, pruned_loss=0.02893, over 16792.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2552, pruned_loss=0.04059, over 3314474.93 frames. ], batch size: 39, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:25:10,808 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2568, 5.6026, 5.3495, 5.4558, 5.1131, 5.0969, 5.0290, 5.7122], device='cuda:7'), covar=tensor([0.1434, 0.1057, 0.1174, 0.0909, 0.0877, 0.0821, 0.1241, 0.1069], device='cuda:7'), in_proj_covar=tensor([0.0682, 0.0832, 0.0684, 0.0627, 0.0530, 0.0534, 0.0701, 0.0643], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:25:34,281 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6219, 2.4344, 2.4497, 4.5295, 2.3709, 2.8848, 2.4610, 2.5553], device='cuda:7'), covar=tensor([0.1186, 0.3618, 0.3027, 0.0443, 0.4064, 0.2441, 0.3548, 0.3852], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0450, 0.0370, 0.0332, 0.0437, 0.0516, 0.0419, 0.0527], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:25:35,967 INFO [train.py:904] (7/8) Epoch 21, batch 1950, loss[loss=0.179, simple_loss=0.2687, pruned_loss=0.04464, over 16545.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2549, pruned_loss=0.04018, over 3316623.57 frames. ], batch size: 68, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:26:05,658 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0121, 3.3153, 3.3138, 2.1311, 2.8810, 2.3691, 3.4279, 3.6085], device='cuda:7'), covar=tensor([0.0285, 0.0868, 0.0656, 0.1914, 0.0859, 0.0982, 0.0664, 0.0856], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0163, 0.0167, 0.0153, 0.0144, 0.0129, 0.0144, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:26:26,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.195e+02 2.614e+02 3.172e+02 4.787e+02, threshold=5.228e+02, percent-clipped=0.0 2023-05-01 05:26:44,686 INFO [train.py:904] (7/8) Epoch 21, batch 2000, loss[loss=0.1855, simple_loss=0.2605, pruned_loss=0.05527, over 16764.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.256, pruned_loss=0.0405, over 3315344.81 frames. ], batch size: 124, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:27:55,531 INFO [train.py:904] (7/8) Epoch 21, batch 2050, loss[loss=0.188, simple_loss=0.2697, pruned_loss=0.05315, over 16497.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2561, pruned_loss=0.04056, over 3311347.49 frames. ], batch size: 75, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:28:01,837 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2089, 3.2565, 3.2536, 5.2144, 4.4146, 4.6230, 2.0338, 3.5998], device='cuda:7'), covar=tensor([0.1204, 0.0681, 0.0965, 0.0172, 0.0252, 0.0362, 0.1473, 0.0678], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0190, 0.0205, 0.0215, 0.0201, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:28:02,214 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 05:28:44,314 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.164e+02 2.478e+02 3.098e+02 5.896e+02, threshold=4.956e+02, percent-clipped=1.0 2023-05-01 05:29:00,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0137, 5.0608, 5.4728, 5.4591, 5.4954, 5.1206, 5.0861, 4.8296], device='cuda:7'), covar=tensor([0.0352, 0.0494, 0.0404, 0.0457, 0.0446, 0.0383, 0.0923, 0.0478], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0456, 0.0444, 0.0414, 0.0490, 0.0466, 0.0557, 0.0373], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 05:29:04,148 INFO [train.py:904] (7/8) Epoch 21, batch 2100, loss[loss=0.1468, simple_loss=0.2382, pruned_loss=0.02769, over 17129.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.256, pruned_loss=0.0409, over 3320580.10 frames. ], batch size: 47, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:29:07,914 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 05:29:15,401 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205109.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:30:03,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8110, 3.9230, 2.6162, 4.5684, 3.0486, 4.4893, 2.7056, 3.2851], device='cuda:7'), covar=tensor([0.0347, 0.0398, 0.1515, 0.0302, 0.0874, 0.0563, 0.1426, 0.0724], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0196, 0.0166, 0.0179, 0.0220, 0.0205, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:30:06,312 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-01 05:30:14,932 INFO [train.py:904] (7/8) Epoch 21, batch 2150, loss[loss=0.1518, simple_loss=0.2466, pruned_loss=0.02849, over 17175.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2577, pruned_loss=0.0413, over 3327288.98 frames. ], batch size: 46, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:30:20,539 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-01 05:30:39,986 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205170.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:30:50,430 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=205177.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:31:04,864 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.289e+02 2.743e+02 3.268e+02 8.955e+02, threshold=5.486e+02, percent-clipped=2.0 2023-05-01 05:31:19,799 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205198.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:31:24,884 INFO [train.py:904] (7/8) Epoch 21, batch 2200, loss[loss=0.1744, simple_loss=0.2689, pruned_loss=0.03994, over 16695.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2585, pruned_loss=0.04212, over 3320243.01 frames. ], batch size: 57, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:31:36,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7190, 3.7850, 2.3488, 4.3881, 2.9455, 4.3567, 2.6748, 3.2380], device='cuda:7'), covar=tensor([0.0323, 0.0411, 0.1598, 0.0363, 0.0838, 0.0545, 0.1382, 0.0691], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0198, 0.0167, 0.0180, 0.0221, 0.0206, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:31:53,571 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-05-01 05:32:34,221 INFO [train.py:904] (7/8) Epoch 21, batch 2250, loss[loss=0.2035, simple_loss=0.2883, pruned_loss=0.05939, over 11692.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2589, pruned_loss=0.04247, over 3320420.06 frames. ], batch size: 246, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:32:45,109 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:33:23,478 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.631e+02 2.149e+02 2.507e+02 2.983e+02 5.754e+02, threshold=5.014e+02, percent-clipped=1.0 2023-05-01 05:33:33,426 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8315, 4.2481, 3.1509, 2.3809, 2.6825, 2.5679, 4.5620, 3.5921], device='cuda:7'), covar=tensor([0.2820, 0.0561, 0.1706, 0.2786, 0.2925, 0.2066, 0.0365, 0.1296], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0270, 0.0303, 0.0308, 0.0296, 0.0257, 0.0293, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:33:44,284 INFO [train.py:904] (7/8) Epoch 21, batch 2300, loss[loss=0.1585, simple_loss=0.254, pruned_loss=0.03152, over 17236.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2591, pruned_loss=0.04282, over 3317969.63 frames. ], batch size: 52, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:34:53,189 INFO [train.py:904] (7/8) Epoch 21, batch 2350, loss[loss=0.2079, simple_loss=0.2937, pruned_loss=0.06108, over 16127.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.26, pruned_loss=0.04348, over 3318948.58 frames. ], batch size: 164, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:35:42,803 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.163e+02 2.457e+02 2.980e+02 4.846e+02, threshold=4.914e+02, percent-clipped=0.0 2023-05-01 05:35:47,920 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 05:36:02,966 INFO [train.py:904] (7/8) Epoch 21, batch 2400, loss[loss=0.1768, simple_loss=0.2718, pruned_loss=0.04086, over 17137.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.2608, pruned_loss=0.04378, over 3316679.40 frames. ], batch size: 48, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:37:10,852 INFO [train.py:904] (7/8) Epoch 21, batch 2450, loss[loss=0.17, simple_loss=0.2618, pruned_loss=0.03911, over 17202.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2606, pruned_loss=0.04337, over 3320100.21 frames. ], batch size: 46, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:37:29,570 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=205465.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 05:37:33,978 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0465, 3.1673, 3.3627, 1.7378, 3.5048, 3.5804, 2.8377, 2.4603], device='cuda:7'), covar=tensor([0.1147, 0.0235, 0.0202, 0.1321, 0.0120, 0.0201, 0.0455, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0098, 0.0140, 0.0080, 0.0125, 0.0129, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:37:45,907 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=205477.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:38:00,844 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.346e+02 2.648e+02 3.284e+02 5.369e+02, threshold=5.296e+02, percent-clipped=1.0 2023-05-01 05:38:21,271 INFO [train.py:904] (7/8) Epoch 21, batch 2500, loss[loss=0.1734, simple_loss=0.2567, pruned_loss=0.0451, over 16555.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2604, pruned_loss=0.04321, over 3325711.85 frames. ], batch size: 75, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:38:24,346 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 05:38:34,974 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5328, 3.5814, 2.1663, 3.8217, 2.7842, 3.7885, 2.3242, 2.8593], device='cuda:7'), covar=tensor([0.0293, 0.0397, 0.1534, 0.0321, 0.0771, 0.0708, 0.1336, 0.0679], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0180, 0.0197, 0.0166, 0.0179, 0.0220, 0.0205, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:38:35,270 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.65 vs. limit=5.0 2023-05-01 05:38:53,422 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=205525.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:39:30,351 INFO [train.py:904] (7/8) Epoch 21, batch 2550, loss[loss=0.1793, simple_loss=0.2629, pruned_loss=0.04787, over 15744.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2602, pruned_loss=0.04294, over 3333423.24 frames. ], batch size: 191, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:39:33,601 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=205554.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:39:41,907 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-05-01 05:40:08,418 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5003, 4.8009, 4.5879, 4.6359, 4.3561, 4.2977, 4.3181, 4.8809], device='cuda:7'), covar=tensor([0.1293, 0.0925, 0.1031, 0.0830, 0.0860, 0.1509, 0.1133, 0.0861], device='cuda:7'), in_proj_covar=tensor([0.0676, 0.0832, 0.0682, 0.0627, 0.0528, 0.0532, 0.0697, 0.0640], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:40:19,336 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.328e+02 2.679e+02 3.297e+02 1.189e+03, threshold=5.358e+02, percent-clipped=3.0 2023-05-01 05:40:38,680 INFO [train.py:904] (7/8) Epoch 21, batch 2600, loss[loss=0.1892, simple_loss=0.2633, pruned_loss=0.05755, over 16698.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2602, pruned_loss=0.04239, over 3332420.45 frames. ], batch size: 134, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:41:18,062 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-05-01 05:41:49,815 INFO [train.py:904] (7/8) Epoch 21, batch 2650, loss[loss=0.1856, simple_loss=0.2851, pruned_loss=0.0431, over 16904.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2603, pruned_loss=0.04195, over 3334474.50 frames. ], batch size: 96, lr: 3.24e-03, grad_scale: 4.0 2023-05-01 05:41:54,587 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205655.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:42:40,191 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.048e+02 2.349e+02 2.910e+02 5.870e+02, threshold=4.699e+02, percent-clipped=1.0 2023-05-01 05:43:00,024 INFO [train.py:904] (7/8) Epoch 21, batch 2700, loss[loss=0.1483, simple_loss=0.2495, pruned_loss=0.02352, over 17128.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.26, pruned_loss=0.04134, over 3342294.47 frames. ], batch size: 46, lr: 3.24e-03, grad_scale: 4.0 2023-05-01 05:43:18,678 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205716.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:43:49,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2052, 4.0689, 4.2758, 4.3900, 4.5008, 4.0729, 4.2725, 4.4936], device='cuda:7'), covar=tensor([0.1632, 0.1202, 0.1315, 0.0725, 0.0643, 0.1341, 0.2710, 0.0754], device='cuda:7'), in_proj_covar=tensor([0.0662, 0.0815, 0.0954, 0.0836, 0.0621, 0.0654, 0.0672, 0.0778], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:44:00,191 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205744.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:44:09,878 INFO [train.py:904] (7/8) Epoch 21, batch 2750, loss[loss=0.1709, simple_loss=0.2773, pruned_loss=0.03224, over 17073.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2604, pruned_loss=0.04134, over 3334769.50 frames. ], batch size: 50, lr: 3.24e-03, grad_scale: 4.0 2023-05-01 05:44:10,228 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9324, 5.2232, 5.5221, 5.5036, 5.5758, 5.2122, 4.8870, 4.8831], device='cuda:7'), covar=tensor([0.0609, 0.0673, 0.0539, 0.0648, 0.0683, 0.0588, 0.1505, 0.0623], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0457, 0.0444, 0.0414, 0.0491, 0.0466, 0.0556, 0.0373], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 05:44:13,656 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0421, 4.5821, 4.4930, 3.2832, 3.7764, 4.4700, 3.9524, 2.5359], device='cuda:7'), covar=tensor([0.0464, 0.0052, 0.0045, 0.0337, 0.0121, 0.0090, 0.0086, 0.0467], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0084, 0.0083, 0.0134, 0.0099, 0.0110, 0.0095, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 05:44:29,188 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=205765.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:45:02,029 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.167e+02 2.431e+02 2.778e+02 3.742e+02, threshold=4.863e+02, percent-clipped=0.0 2023-05-01 05:45:19,232 INFO [train.py:904] (7/8) Epoch 21, batch 2800, loss[loss=0.1869, simple_loss=0.2911, pruned_loss=0.04138, over 16738.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.26, pruned_loss=0.04108, over 3334167.47 frames. ], batch size: 57, lr: 3.24e-03, grad_scale: 8.0 2023-05-01 05:45:24,492 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205805.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:45:36,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=205813.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:45:56,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7232, 3.0066, 3.0389, 4.8861, 4.1060, 4.3984, 1.4405, 3.5470], device='cuda:7'), covar=tensor([0.1367, 0.0709, 0.1021, 0.0204, 0.0217, 0.0347, 0.1667, 0.0639], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0175, 0.0195, 0.0192, 0.0208, 0.0217, 0.0203, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:46:29,085 INFO [train.py:904] (7/8) Epoch 21, batch 2850, loss[loss=0.1663, simple_loss=0.2481, pruned_loss=0.04225, over 16367.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2601, pruned_loss=0.04144, over 3332008.84 frames. ], batch size: 146, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:46:31,700 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=205854.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:46:32,181 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-01 05:46:43,345 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 05:46:57,586 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8497, 4.9397, 5.3424, 5.3223, 5.3458, 5.0261, 4.9576, 4.7725], device='cuda:7'), covar=tensor([0.0360, 0.0578, 0.0393, 0.0412, 0.0457, 0.0364, 0.0896, 0.0470], device='cuda:7'), in_proj_covar=tensor([0.0419, 0.0463, 0.0450, 0.0419, 0.0497, 0.0471, 0.0563, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 05:47:09,561 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2759, 3.5229, 3.8993, 2.2757, 3.0397, 2.4516, 3.7616, 3.6746], device='cuda:7'), covar=tensor([0.0300, 0.0931, 0.0497, 0.1907, 0.0882, 0.0989, 0.0593, 0.1095], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0164, 0.0168, 0.0153, 0.0145, 0.0130, 0.0144, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:47:17,821 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 2.169e+02 2.532e+02 3.008e+02 5.585e+02, threshold=5.064e+02, percent-clipped=2.0 2023-05-01 05:47:19,402 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=205890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:47:29,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8685, 2.8369, 2.6441, 4.2541, 3.5265, 4.1315, 1.6623, 2.9346], device='cuda:7'), covar=tensor([0.1335, 0.0629, 0.1051, 0.0163, 0.0126, 0.0350, 0.1507, 0.0808], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0175, 0.0195, 0.0192, 0.0208, 0.0217, 0.0202, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 05:47:36,339 INFO [train.py:904] (7/8) Epoch 21, batch 2900, loss[loss=0.1543, simple_loss=0.2378, pruned_loss=0.03538, over 16913.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2575, pruned_loss=0.041, over 3334136.14 frames. ], batch size: 96, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:47:37,316 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=205902.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:48:44,810 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=205951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:48:45,585 INFO [train.py:904] (7/8) Epoch 21, batch 2950, loss[loss=0.1418, simple_loss=0.2288, pruned_loss=0.02745, over 16753.00 frames. ], tot_loss[loss=0.17, simple_loss=0.257, pruned_loss=0.0415, over 3335058.24 frames. ], batch size: 39, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:49:13,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3956, 5.3828, 5.1976, 4.6599, 5.1804, 2.1189, 4.9356, 5.1798], device='cuda:7'), covar=tensor([0.0079, 0.0067, 0.0187, 0.0355, 0.0103, 0.2465, 0.0127, 0.0166], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0157, 0.0199, 0.0180, 0.0178, 0.0210, 0.0190, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:49:36,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6342, 2.6935, 2.2601, 2.3868, 2.9716, 2.6475, 3.2427, 3.1591], device='cuda:7'), covar=tensor([0.0153, 0.0411, 0.0516, 0.0474, 0.0278, 0.0398, 0.0228, 0.0275], device='cuda:7'), in_proj_covar=tensor([0.0214, 0.0241, 0.0230, 0.0231, 0.0241, 0.0241, 0.0244, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:49:36,952 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.251e+02 2.659e+02 3.342e+02 9.466e+02, threshold=5.318e+02, percent-clipped=4.0 2023-05-01 05:49:58,057 INFO [train.py:904] (7/8) Epoch 21, batch 3000, loss[loss=0.1518, simple_loss=0.2359, pruned_loss=0.03382, over 16871.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.258, pruned_loss=0.0427, over 3333997.36 frames. ], batch size: 42, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:49:58,058 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 05:50:06,476 INFO [train.py:938] (7/8) Epoch 21, validation: loss=0.1356, simple_loss=0.2408, pruned_loss=0.01521, over 944034.00 frames. 2023-05-01 05:50:06,477 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 05:50:18,462 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206011.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:51:14,523 INFO [train.py:904] (7/8) Epoch 21, batch 3050, loss[loss=0.191, simple_loss=0.2836, pruned_loss=0.04918, over 16640.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2576, pruned_loss=0.04261, over 3323895.52 frames. ], batch size: 57, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:51:21,086 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=206056.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:51:34,095 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 05:51:40,139 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8822, 4.2946, 4.2738, 3.0865, 3.6234, 4.2269, 3.8482, 2.2932], device='cuda:7'), covar=tensor([0.0453, 0.0066, 0.0056, 0.0344, 0.0125, 0.0111, 0.0086, 0.0491], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0084, 0.0083, 0.0134, 0.0098, 0.0109, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:51:51,314 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4219, 5.8264, 5.5675, 5.6693, 5.2378, 5.3098, 5.1937, 5.9775], device='cuda:7'), covar=tensor([0.1431, 0.1033, 0.1097, 0.0842, 0.0911, 0.0677, 0.1223, 0.1005], device='cuda:7'), in_proj_covar=tensor([0.0682, 0.0841, 0.0693, 0.0631, 0.0535, 0.0538, 0.0704, 0.0650], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:52:05,517 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.000e+02 2.421e+02 3.043e+02 5.222e+02, threshold=4.843e+02, percent-clipped=0.0 2023-05-01 05:52:22,008 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206100.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:52:24,073 INFO [train.py:904] (7/8) Epoch 21, batch 3100, loss[loss=0.168, simple_loss=0.2444, pruned_loss=0.0458, over 16738.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2573, pruned_loss=0.04247, over 3326835.90 frames. ], batch size: 134, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:52:43,770 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=206117.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:53:24,843 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 05:53:30,991 INFO [train.py:904] (7/8) Epoch 21, batch 3150, loss[loss=0.1578, simple_loss=0.2605, pruned_loss=0.02756, over 17041.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2563, pruned_loss=0.04187, over 3333164.43 frames. ], batch size: 50, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:53:31,484 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4675, 5.4405, 5.3235, 4.8721, 4.9671, 5.3943, 5.3531, 4.9933], device='cuda:7'), covar=tensor([0.0640, 0.0482, 0.0286, 0.0312, 0.1033, 0.0442, 0.0251, 0.0775], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0441, 0.0358, 0.0353, 0.0367, 0.0410, 0.0246, 0.0431], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 05:53:31,825 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 05:53:42,659 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 05:54:22,975 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.110e+02 2.360e+02 2.759e+02 5.694e+02, threshold=4.721e+02, percent-clipped=1.0 2023-05-01 05:54:35,022 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=206197.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:54:41,550 INFO [train.py:904] (7/8) Epoch 21, batch 3200, loss[loss=0.1726, simple_loss=0.2659, pruned_loss=0.03964, over 16636.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2554, pruned_loss=0.04126, over 3329620.60 frames. ], batch size: 62, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:55:01,234 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 05:55:12,653 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9828, 2.2043, 2.5720, 3.0480, 2.8001, 3.5048, 2.3779, 3.4495], device='cuda:7'), covar=tensor([0.0256, 0.0526, 0.0359, 0.0290, 0.0370, 0.0171, 0.0493, 0.0161], device='cuda:7'), in_proj_covar=tensor([0.0189, 0.0195, 0.0180, 0.0186, 0.0197, 0.0155, 0.0197, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 05:55:27,122 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 05:55:33,624 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 05:55:43,203 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206246.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:55:51,283 INFO [train.py:904] (7/8) Epoch 21, batch 3250, loss[loss=0.1534, simple_loss=0.2503, pruned_loss=0.02821, over 17223.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.256, pruned_loss=0.04166, over 3330033.66 frames. ], batch size: 45, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:55:59,473 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=206258.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:55:59,824 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.05 vs. limit=5.0 2023-05-01 05:56:42,634 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.332e+02 2.138e+02 2.458e+02 2.946e+02 5.794e+02, threshold=4.917e+02, percent-clipped=2.0 2023-05-01 05:57:00,582 INFO [train.py:904] (7/8) Epoch 21, batch 3300, loss[loss=0.1844, simple_loss=0.2773, pruned_loss=0.04574, over 17031.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.258, pruned_loss=0.04224, over 3324238.84 frames. ], batch size: 55, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:57:13,838 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206311.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:58:09,887 INFO [train.py:904] (7/8) Epoch 21, batch 3350, loss[loss=0.2, simple_loss=0.278, pruned_loss=0.06105, over 16724.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2572, pruned_loss=0.04126, over 3330025.11 frames. ], batch size: 134, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:58:20,196 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206359.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:59:00,091 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.208e+02 2.654e+02 3.136e+02 4.728e+02, threshold=5.309e+02, percent-clipped=0.0 2023-05-01 05:59:16,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206400.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:59:18,376 INFO [train.py:904] (7/8) Epoch 21, batch 3400, loss[loss=0.175, simple_loss=0.2664, pruned_loss=0.04181, over 16547.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2579, pruned_loss=0.04191, over 3311017.61 frames. ], batch size: 62, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 05:59:31,259 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0450, 5.1039, 5.5473, 5.5258, 5.5528, 5.1925, 5.1236, 4.9599], device='cuda:7'), covar=tensor([0.0379, 0.0595, 0.0396, 0.0456, 0.0551, 0.0380, 0.1030, 0.0457], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0464, 0.0452, 0.0419, 0.0497, 0.0474, 0.0568, 0.0379], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 05:59:32,345 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206412.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 05:59:37,081 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 06:00:07,497 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9213, 4.3128, 4.3698, 3.2848, 3.6171, 4.2713, 3.7391, 2.5744], device='cuda:7'), covar=tensor([0.0440, 0.0070, 0.0048, 0.0301, 0.0133, 0.0098, 0.0099, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0084, 0.0083, 0.0134, 0.0098, 0.0109, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:00:11,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5257, 3.5006, 3.4907, 2.7293, 3.2411, 2.0720, 3.0676, 2.6741], device='cuda:7'), covar=tensor([0.0142, 0.0118, 0.0183, 0.0223, 0.0105, 0.2400, 0.0135, 0.0236], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0158, 0.0200, 0.0181, 0.0179, 0.0210, 0.0190, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:00:21,657 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206448.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:00:26,898 INFO [train.py:904] (7/8) Epoch 21, batch 3450, loss[loss=0.1759, simple_loss=0.2753, pruned_loss=0.03818, over 17030.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2562, pruned_loss=0.04129, over 3309477.82 frames. ], batch size: 50, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:01:17,176 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.080e+02 2.437e+02 2.944e+02 7.361e+02, threshold=4.875e+02, percent-clipped=3.0 2023-05-01 06:01:36,858 INFO [train.py:904] (7/8) Epoch 21, batch 3500, loss[loss=0.1949, simple_loss=0.2839, pruned_loss=0.05297, over 16471.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2555, pruned_loss=0.04077, over 3320005.09 frames. ], batch size: 75, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:02:12,285 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 06:02:37,901 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:02:44,994 INFO [train.py:904] (7/8) Epoch 21, batch 3550, loss[loss=0.1602, simple_loss=0.2481, pruned_loss=0.03613, over 17228.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2544, pruned_loss=0.0403, over 3320284.48 frames. ], batch size: 45, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:02:47,153 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=206553.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:03:37,042 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.127e+02 2.401e+02 2.794e+02 4.680e+02, threshold=4.802e+02, percent-clipped=0.0 2023-05-01 06:03:44,321 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:03:55,940 INFO [train.py:904] (7/8) Epoch 21, batch 3600, loss[loss=0.1453, simple_loss=0.2288, pruned_loss=0.03087, over 16469.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.253, pruned_loss=0.04032, over 3302890.49 frames. ], batch size: 68, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:04:14,499 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0767, 3.0134, 2.6464, 4.4569, 3.7031, 4.2509, 1.7069, 3.1578], device='cuda:7'), covar=tensor([0.1208, 0.0655, 0.1116, 0.0194, 0.0171, 0.0375, 0.1533, 0.0757], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0175, 0.0194, 0.0192, 0.0207, 0.0217, 0.0201, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 06:04:54,301 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5681, 2.3253, 1.8352, 2.0946, 2.6325, 2.4446, 2.6199, 2.7811], device='cuda:7'), covar=tensor([0.0223, 0.0415, 0.0571, 0.0484, 0.0274, 0.0334, 0.0220, 0.0281], device='cuda:7'), in_proj_covar=tensor([0.0215, 0.0240, 0.0229, 0.0229, 0.0241, 0.0240, 0.0244, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:05:06,977 INFO [train.py:904] (7/8) Epoch 21, batch 3650, loss[loss=0.1794, simple_loss=0.2476, pruned_loss=0.05563, over 16878.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2527, pruned_loss=0.04112, over 3293062.81 frames. ], batch size: 109, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:05:59,788 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.224e+02 2.566e+02 3.274e+02 6.534e+02, threshold=5.132e+02, percent-clipped=5.0 2023-05-01 06:06:18,969 INFO [train.py:904] (7/8) Epoch 21, batch 3700, loss[loss=0.1611, simple_loss=0.2405, pruned_loss=0.04088, over 16824.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2515, pruned_loss=0.04243, over 3272326.17 frames. ], batch size: 102, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:06:34,752 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1454, 5.4939, 5.2192, 5.2842, 5.0073, 4.9302, 4.9787, 5.5749], device='cuda:7'), covar=tensor([0.1214, 0.0798, 0.1084, 0.0789, 0.0768, 0.0991, 0.1205, 0.0844], device='cuda:7'), in_proj_covar=tensor([0.0679, 0.0834, 0.0688, 0.0630, 0.0531, 0.0535, 0.0699, 0.0648], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:06:34,800 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206712.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:07:32,564 INFO [train.py:904] (7/8) Epoch 21, batch 3750, loss[loss=0.1723, simple_loss=0.2606, pruned_loss=0.04203, over 16448.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2522, pruned_loss=0.0436, over 3279660.38 frames. ], batch size: 68, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:07:45,864 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206760.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:08:26,979 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.267e+02 2.726e+02 3.357e+02 9.074e+02, threshold=5.452e+02, percent-clipped=3.0 2023-05-01 06:08:45,466 INFO [train.py:904] (7/8) Epoch 21, batch 3800, loss[loss=0.1837, simple_loss=0.2519, pruned_loss=0.05777, over 16752.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2526, pruned_loss=0.04481, over 3285803.66 frames. ], batch size: 83, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:08:48,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5991, 3.8819, 4.0315, 2.9418, 3.7609, 4.1385, 3.7177, 2.5657], device='cuda:7'), covar=tensor([0.0522, 0.0297, 0.0059, 0.0356, 0.0105, 0.0094, 0.0099, 0.0442], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0084, 0.0082, 0.0133, 0.0097, 0.0109, 0.0094, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:09:18,993 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 06:09:32,819 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=206834.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:09:58,314 INFO [train.py:904] (7/8) Epoch 21, batch 3850, loss[loss=0.1976, simple_loss=0.2791, pruned_loss=0.05808, over 12403.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2537, pruned_loss=0.04576, over 3273418.10 frames. ], batch size: 246, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:10:00,742 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=206853.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:10:52,983 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.673e+02 2.267e+02 2.812e+02 3.321e+02 1.032e+03, threshold=5.623e+02, percent-clipped=3.0 2023-05-01 06:11:00,976 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=206895.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 06:11:09,944 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=206901.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:11:10,723 INFO [train.py:904] (7/8) Epoch 21, batch 3900, loss[loss=0.1626, simple_loss=0.2385, pruned_loss=0.04332, over 15677.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2529, pruned_loss=0.04616, over 3275352.60 frames. ], batch size: 191, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:12:24,777 INFO [train.py:904] (7/8) Epoch 21, batch 3950, loss[loss=0.1512, simple_loss=0.2283, pruned_loss=0.03706, over 16853.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2528, pruned_loss=0.04691, over 3261832.77 frames. ], batch size: 96, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:12:25,376 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7965, 3.8723, 2.9140, 2.3548, 2.5029, 2.4385, 3.9293, 3.4198], device='cuda:7'), covar=tensor([0.2532, 0.0569, 0.1711, 0.3021, 0.2749, 0.2061, 0.0492, 0.1364], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0273, 0.0307, 0.0313, 0.0301, 0.0261, 0.0297, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:12:30,378 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5867, 4.6938, 4.8365, 4.6949, 4.7318, 5.2663, 4.8104, 4.5018], device='cuda:7'), covar=tensor([0.1660, 0.2074, 0.2207, 0.2090, 0.2632, 0.1042, 0.1625, 0.2599], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0605, 0.0662, 0.0500, 0.0665, 0.0692, 0.0519, 0.0665], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:12:35,868 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6606, 3.7054, 2.3755, 4.0612, 2.9193, 4.0713, 2.4215, 2.9755], device='cuda:7'), covar=tensor([0.0254, 0.0368, 0.1381, 0.0239, 0.0744, 0.0551, 0.1324, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0177, 0.0193, 0.0165, 0.0178, 0.0217, 0.0200, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 06:13:16,344 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.243e+02 2.750e+02 3.370e+02 5.125e+02, threshold=5.500e+02, percent-clipped=0.0 2023-05-01 06:13:35,027 INFO [train.py:904] (7/8) Epoch 21, batch 4000, loss[loss=0.1669, simple_loss=0.2409, pruned_loss=0.04648, over 16758.00 frames. ], tot_loss[loss=0.174, simple_loss=0.253, pruned_loss=0.04752, over 3269070.73 frames. ], batch size: 124, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:13:53,454 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7149, 3.8088, 2.8521, 2.2874, 2.4176, 2.3932, 3.8508, 3.3179], device='cuda:7'), covar=tensor([0.2503, 0.0525, 0.1655, 0.2833, 0.2730, 0.2045, 0.0526, 0.1308], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0273, 0.0307, 0.0313, 0.0301, 0.0261, 0.0297, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:14:07,930 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2554, 2.2863, 2.3590, 3.9931, 2.1807, 2.6452, 2.3702, 2.4376], device='cuda:7'), covar=tensor([0.1313, 0.3525, 0.2771, 0.0513, 0.3899, 0.2377, 0.3389, 0.3166], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0453, 0.0371, 0.0334, 0.0439, 0.0522, 0.0421, 0.0530], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:14:45,538 INFO [train.py:904] (7/8) Epoch 21, batch 4050, loss[loss=0.1646, simple_loss=0.2489, pruned_loss=0.04015, over 16700.00 frames. ], tot_loss[loss=0.1732, simple_loss=0.2535, pruned_loss=0.04642, over 3285918.77 frames. ], batch size: 76, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:15:37,540 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 1.811e+02 2.022e+02 2.362e+02 4.277e+02, threshold=4.045e+02, percent-clipped=0.0 2023-05-01 06:15:56,001 INFO [train.py:904] (7/8) Epoch 21, batch 4100, loss[loss=0.1804, simple_loss=0.2635, pruned_loss=0.04865, over 16723.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2545, pruned_loss=0.04553, over 3275622.59 frames. ], batch size: 76, lr: 3.23e-03, grad_scale: 8.0 2023-05-01 06:16:27,966 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0478, 4.2159, 4.4346, 4.4197, 4.4133, 4.1711, 4.1647, 4.0934], device='cuda:7'), covar=tensor([0.0326, 0.0419, 0.0343, 0.0339, 0.0491, 0.0367, 0.0782, 0.0473], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0460, 0.0445, 0.0413, 0.0491, 0.0467, 0.0556, 0.0375], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 06:17:10,034 INFO [train.py:904] (7/8) Epoch 21, batch 4150, loss[loss=0.1892, simple_loss=0.2872, pruned_loss=0.04561, over 16720.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.262, pruned_loss=0.04819, over 3247146.77 frames. ], batch size: 89, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:17:38,263 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0415, 4.0035, 3.8663, 2.8340, 3.9145, 1.6473, 3.6703, 3.2719], device='cuda:7'), covar=tensor([0.0132, 0.0119, 0.0219, 0.0428, 0.0106, 0.3496, 0.0147, 0.0399], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0158, 0.0200, 0.0182, 0.0180, 0.0211, 0.0191, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:17:46,416 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=207176.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:18:04,517 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.144e+02 2.646e+02 3.287e+02 6.191e+02, threshold=5.292e+02, percent-clipped=8.0 2023-05-01 06:18:07,780 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=207190.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 06:18:24,677 INFO [train.py:904] (7/8) Epoch 21, batch 4200, loss[loss=0.1991, simple_loss=0.2944, pruned_loss=0.05191, over 16476.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2692, pruned_loss=0.05035, over 3190606.98 frames. ], batch size: 75, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:19:18,470 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=207237.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:19:26,242 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1240, 4.2661, 4.5023, 4.4851, 4.5142, 4.2510, 4.1722, 4.1954], device='cuda:7'), covar=tensor([0.0354, 0.0544, 0.0496, 0.0457, 0.0450, 0.0393, 0.1193, 0.0525], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0458, 0.0443, 0.0412, 0.0489, 0.0466, 0.0554, 0.0373], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 06:19:40,059 INFO [train.py:904] (7/8) Epoch 21, batch 4250, loss[loss=0.1665, simple_loss=0.2591, pruned_loss=0.03693, over 17094.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2725, pruned_loss=0.0502, over 3171845.98 frames. ], batch size: 47, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:19:42,592 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 06:19:50,505 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3126, 2.3558, 2.3204, 3.9655, 2.1916, 2.5808, 2.4404, 2.4749], device='cuda:7'), covar=tensor([0.1295, 0.3649, 0.2749, 0.0555, 0.3997, 0.2512, 0.3381, 0.3356], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0450, 0.0368, 0.0331, 0.0435, 0.0518, 0.0417, 0.0525], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:20:10,911 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 06:20:35,823 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.204e+02 2.470e+02 3.047e+02 4.735e+02, threshold=4.941e+02, percent-clipped=0.0 2023-05-01 06:20:55,799 INFO [train.py:904] (7/8) Epoch 21, batch 4300, loss[loss=0.1889, simple_loss=0.2815, pruned_loss=0.04813, over 16965.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2737, pruned_loss=0.04947, over 3161641.70 frames. ], batch size: 116, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:22:07,521 INFO [train.py:904] (7/8) Epoch 21, batch 4350, loss[loss=0.2055, simple_loss=0.2891, pruned_loss=0.06094, over 12190.00 frames. ], tot_loss[loss=0.1887, simple_loss=0.277, pruned_loss=0.05023, over 3158213.31 frames. ], batch size: 246, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:22:57,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6184, 3.7404, 2.3710, 4.4055, 2.9128, 4.3219, 2.4817, 3.0279], device='cuda:7'), covar=tensor([0.0294, 0.0361, 0.1617, 0.0122, 0.0801, 0.0443, 0.1426, 0.0754], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0177, 0.0193, 0.0163, 0.0177, 0.0216, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 06:23:02,765 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.219e+02 2.598e+02 3.036e+02 7.307e+02, threshold=5.195e+02, percent-clipped=1.0 2023-05-01 06:23:22,054 INFO [train.py:904] (7/8) Epoch 21, batch 4400, loss[loss=0.2044, simple_loss=0.2942, pruned_loss=0.05733, over 16323.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.279, pruned_loss=0.05125, over 3156725.78 frames. ], batch size: 35, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:24:37,207 INFO [train.py:904] (7/8) Epoch 21, batch 4450, loss[loss=0.2061, simple_loss=0.3032, pruned_loss=0.05454, over 16765.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2819, pruned_loss=0.052, over 3165045.13 frames. ], batch size: 83, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:24:58,684 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 06:25:31,517 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.026e+02 2.353e+02 2.743e+02 3.990e+02, threshold=4.706e+02, percent-clipped=0.0 2023-05-01 06:25:33,786 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=207490.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 06:25:50,502 INFO [train.py:904] (7/8) Epoch 21, batch 4500, loss[loss=0.2131, simple_loss=0.2937, pruned_loss=0.06628, over 16941.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2829, pruned_loss=0.05294, over 3185540.97 frames. ], batch size: 109, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:26:35,660 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=207532.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:26:43,762 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=207538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:26:59,938 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7512, 2.8492, 2.5857, 4.4584, 3.3336, 3.9140, 1.8530, 2.8909], device='cuda:7'), covar=tensor([0.1425, 0.0814, 0.1233, 0.0169, 0.0292, 0.0444, 0.1663, 0.0884], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0175, 0.0193, 0.0190, 0.0207, 0.0215, 0.0200, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 06:27:03,821 INFO [train.py:904] (7/8) Epoch 21, batch 4550, loss[loss=0.2017, simple_loss=0.2928, pruned_loss=0.05529, over 16788.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2839, pruned_loss=0.05403, over 3199562.38 frames. ], batch size: 89, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:27:57,549 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 1.874e+02 2.187e+02 2.648e+02 6.639e+02, threshold=4.374e+02, percent-clipped=1.0 2023-05-01 06:28:11,151 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 06:28:16,272 INFO [train.py:904] (7/8) Epoch 21, batch 4600, loss[loss=0.1824, simple_loss=0.2735, pruned_loss=0.0457, over 17237.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2852, pruned_loss=0.05442, over 3212280.80 frames. ], batch size: 52, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:29:28,376 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1170, 5.6688, 5.9171, 5.5466, 5.6256, 6.1907, 5.7245, 5.5708], device='cuda:7'), covar=tensor([0.0857, 0.1612, 0.1834, 0.1778, 0.2172, 0.0756, 0.1161, 0.1916], device='cuda:7'), in_proj_covar=tensor([0.0403, 0.0584, 0.0637, 0.0487, 0.0647, 0.0672, 0.0498, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:29:29,140 INFO [train.py:904] (7/8) Epoch 21, batch 4650, loss[loss=0.1896, simple_loss=0.2776, pruned_loss=0.05081, over 17158.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.284, pruned_loss=0.05427, over 3233857.78 frames. ], batch size: 46, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:29:53,356 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7862, 4.7608, 4.4961, 3.7472, 4.6532, 1.6445, 4.4065, 3.9744], device='cuda:7'), covar=tensor([0.0050, 0.0042, 0.0146, 0.0216, 0.0052, 0.2987, 0.0080, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0154, 0.0196, 0.0177, 0.0175, 0.0207, 0.0186, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:30:23,483 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 1.925e+02 2.221e+02 2.572e+02 4.933e+02, threshold=4.441e+02, percent-clipped=1.0 2023-05-01 06:30:42,424 INFO [train.py:904] (7/8) Epoch 21, batch 4700, loss[loss=0.1962, simple_loss=0.282, pruned_loss=0.05518, over 16702.00 frames. ], tot_loss[loss=0.1943, simple_loss=0.2818, pruned_loss=0.05341, over 3233739.60 frames. ], batch size: 124, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:30:59,613 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8611, 4.0293, 3.0116, 2.4972, 2.8948, 2.5164, 4.6698, 3.5804], device='cuda:7'), covar=tensor([0.2690, 0.0690, 0.1791, 0.2492, 0.2582, 0.1999, 0.0399, 0.1201], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0269, 0.0303, 0.0310, 0.0297, 0.0257, 0.0295, 0.0337], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:31:12,060 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9938, 5.0530, 4.8975, 4.5390, 4.5152, 4.9625, 4.7960, 4.6289], device='cuda:7'), covar=tensor([0.0630, 0.0513, 0.0286, 0.0261, 0.1042, 0.0569, 0.0318, 0.0614], device='cuda:7'), in_proj_covar=tensor([0.0291, 0.0417, 0.0338, 0.0333, 0.0348, 0.0388, 0.0231, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:31:56,746 INFO [train.py:904] (7/8) Epoch 21, batch 4750, loss[loss=0.1704, simple_loss=0.2585, pruned_loss=0.04114, over 16738.00 frames. ], tot_loss[loss=0.1896, simple_loss=0.2773, pruned_loss=0.05098, over 3222511.75 frames. ], batch size: 124, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:32:36,037 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1935, 4.3594, 4.1085, 3.8761, 3.6681, 4.2542, 3.9855, 3.9023], device='cuda:7'), covar=tensor([0.0735, 0.0637, 0.0395, 0.0346, 0.1108, 0.0585, 0.0794, 0.0712], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0418, 0.0339, 0.0333, 0.0348, 0.0388, 0.0231, 0.0407], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:32:50,130 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 1.797e+02 2.146e+02 2.444e+02 4.649e+02, threshold=4.292e+02, percent-clipped=1.0 2023-05-01 06:33:10,710 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.73 vs. limit=5.0 2023-05-01 06:33:11,168 INFO [train.py:904] (7/8) Epoch 21, batch 4800, loss[loss=0.1747, simple_loss=0.2754, pruned_loss=0.03704, over 16698.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2736, pruned_loss=0.04867, over 3225655.01 frames. ], batch size: 124, lr: 3.22e-03, grad_scale: 16.0 2023-05-01 06:33:50,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2441, 2.2314, 2.4512, 3.9924, 2.1848, 2.5558, 2.3272, 2.4113], device='cuda:7'), covar=tensor([0.1423, 0.3724, 0.2673, 0.0531, 0.4114, 0.2579, 0.3610, 0.3265], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0446, 0.0365, 0.0328, 0.0435, 0.0515, 0.0415, 0.0521], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:33:55,577 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=207832.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:34:24,568 INFO [train.py:904] (7/8) Epoch 21, batch 4850, loss[loss=0.2119, simple_loss=0.2931, pruned_loss=0.06534, over 12403.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2742, pruned_loss=0.04831, over 3196926.12 frames. ], batch size: 246, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:35:08,243 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=207880.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:35:12,912 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-05-01 06:35:22,069 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 1.907e+02 2.222e+02 2.565e+02 4.310e+02, threshold=4.444e+02, percent-clipped=1.0 2023-05-01 06:35:27,213 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4138, 3.3656, 3.4510, 3.5159, 3.5617, 3.2996, 3.5397, 3.6340], device='cuda:7'), covar=tensor([0.1207, 0.0893, 0.1039, 0.0617, 0.0569, 0.2203, 0.0994, 0.0692], device='cuda:7'), in_proj_covar=tensor([0.0618, 0.0764, 0.0894, 0.0783, 0.0585, 0.0614, 0.0632, 0.0733], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:35:40,339 INFO [train.py:904] (7/8) Epoch 21, batch 4900, loss[loss=0.1865, simple_loss=0.2811, pruned_loss=0.04594, over 16225.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.273, pruned_loss=0.04718, over 3190681.30 frames. ], batch size: 165, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:35:49,358 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8296, 3.8864, 4.1390, 4.1131, 4.1029, 3.8911, 3.8935, 3.8912], device='cuda:7'), covar=tensor([0.0313, 0.0579, 0.0349, 0.0361, 0.0446, 0.0386, 0.0818, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0446, 0.0432, 0.0401, 0.0476, 0.0454, 0.0542, 0.0364], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 06:36:52,781 INFO [train.py:904] (7/8) Epoch 21, batch 4950, loss[loss=0.2062, simple_loss=0.2925, pruned_loss=0.05997, over 12237.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2723, pruned_loss=0.0462, over 3199605.16 frames. ], batch size: 246, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:36:56,319 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3446, 3.4752, 3.6209, 3.5940, 3.5999, 3.4269, 3.4505, 3.4874], device='cuda:7'), covar=tensor([0.0388, 0.0570, 0.0421, 0.0411, 0.0474, 0.0489, 0.0793, 0.0520], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0445, 0.0432, 0.0400, 0.0475, 0.0452, 0.0539, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 06:37:47,806 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 1.964e+02 2.322e+02 2.960e+02 4.980e+02, threshold=4.644e+02, percent-clipped=1.0 2023-05-01 06:38:08,414 INFO [train.py:904] (7/8) Epoch 21, batch 5000, loss[loss=0.1773, simple_loss=0.2715, pruned_loss=0.04157, over 17219.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.274, pruned_loss=0.04672, over 3193445.11 frames. ], batch size: 44, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:38:53,167 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=208032.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:39:21,549 INFO [train.py:904] (7/8) Epoch 21, batch 5050, loss[loss=0.1697, simple_loss=0.2669, pruned_loss=0.03624, over 16904.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2747, pruned_loss=0.04649, over 3216241.15 frames. ], batch size: 96, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:39:21,989 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9389, 3.3987, 3.3879, 2.1033, 3.0835, 3.3784, 3.1190, 1.8528], device='cuda:7'), covar=tensor([0.0584, 0.0065, 0.0055, 0.0437, 0.0114, 0.0111, 0.0120, 0.0505], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0082, 0.0082, 0.0133, 0.0097, 0.0108, 0.0093, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:39:36,435 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3611, 4.4885, 4.2653, 4.0006, 3.9299, 4.3458, 4.1373, 4.1082], device='cuda:7'), covar=tensor([0.0639, 0.0462, 0.0318, 0.0269, 0.0940, 0.0529, 0.0537, 0.0584], device='cuda:7'), in_proj_covar=tensor([0.0288, 0.0416, 0.0338, 0.0331, 0.0346, 0.0388, 0.0230, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:40:06,774 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1552, 4.2931, 4.0720, 3.7992, 3.7533, 4.1557, 3.8745, 3.9286], device='cuda:7'), covar=tensor([0.0660, 0.0434, 0.0321, 0.0277, 0.0907, 0.0496, 0.0771, 0.0580], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0414, 0.0336, 0.0330, 0.0345, 0.0386, 0.0230, 0.0403], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:40:18,556 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 1.998e+02 2.352e+02 2.773e+02 5.877e+02, threshold=4.704e+02, percent-clipped=4.0 2023-05-01 06:40:23,294 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=208093.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 06:40:35,287 INFO [train.py:904] (7/8) Epoch 21, batch 5100, loss[loss=0.1965, simple_loss=0.2792, pruned_loss=0.05688, over 11774.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2731, pruned_loss=0.04575, over 3221050.70 frames. ], batch size: 247, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:41:48,637 INFO [train.py:904] (7/8) Epoch 21, batch 5150, loss[loss=0.1636, simple_loss=0.2552, pruned_loss=0.03601, over 17267.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.273, pruned_loss=0.04506, over 3229595.87 frames. ], batch size: 52, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:42:00,384 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5560, 3.5119, 2.6953, 2.1633, 2.3706, 2.4107, 3.6482, 3.2313], device='cuda:7'), covar=tensor([0.2710, 0.0610, 0.1782, 0.2756, 0.2572, 0.1935, 0.0513, 0.1129], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0270, 0.0305, 0.0311, 0.0297, 0.0257, 0.0297, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:42:35,458 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 06:42:43,673 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 1.920e+02 2.264e+02 2.595e+02 3.716e+02, threshold=4.527e+02, percent-clipped=0.0 2023-05-01 06:43:01,063 INFO [train.py:904] (7/8) Epoch 21, batch 5200, loss[loss=0.183, simple_loss=0.2708, pruned_loss=0.04764, over 16519.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2714, pruned_loss=0.04462, over 3221920.16 frames. ], batch size: 75, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:44:11,877 INFO [train.py:904] (7/8) Epoch 21, batch 5250, loss[loss=0.1671, simple_loss=0.2487, pruned_loss=0.04279, over 16373.00 frames. ], tot_loss[loss=0.179, simple_loss=0.269, pruned_loss=0.04447, over 3209950.36 frames. ], batch size: 35, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:44:22,164 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 06:44:32,525 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=208265.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:45:07,261 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 1.864e+02 2.120e+02 2.516e+02 5.355e+02, threshold=4.239e+02, percent-clipped=2.0 2023-05-01 06:45:25,342 INFO [train.py:904] (7/8) Epoch 21, batch 5300, loss[loss=0.1548, simple_loss=0.2416, pruned_loss=0.03401, over 16450.00 frames. ], tot_loss[loss=0.1761, simple_loss=0.2657, pruned_loss=0.04324, over 3207808.18 frames. ], batch size: 146, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:45:32,280 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-01 06:46:00,467 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=208326.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:46:37,971 INFO [train.py:904] (7/8) Epoch 21, batch 5350, loss[loss=0.1642, simple_loss=0.2581, pruned_loss=0.03517, over 16758.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2636, pruned_loss=0.04195, over 3221381.58 frames. ], batch size: 39, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:47:33,149 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=208388.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 06:47:34,969 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.963e+02 2.291e+02 2.599e+02 4.161e+02, threshold=4.581e+02, percent-clipped=0.0 2023-05-01 06:47:53,659 INFO [train.py:904] (7/8) Epoch 21, batch 5400, loss[loss=0.1702, simple_loss=0.2656, pruned_loss=0.03743, over 17037.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.266, pruned_loss=0.04259, over 3209984.23 frames. ], batch size: 50, lr: 3.22e-03, grad_scale: 8.0 2023-05-01 06:48:12,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5255, 4.4913, 4.4078, 3.1026, 4.4553, 1.4709, 4.0999, 4.0716], device='cuda:7'), covar=tensor([0.0172, 0.0144, 0.0247, 0.0809, 0.0164, 0.3861, 0.0236, 0.0401], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0151, 0.0193, 0.0175, 0.0172, 0.0203, 0.0183, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:49:10,989 INFO [train.py:904] (7/8) Epoch 21, batch 5450, loss[loss=0.2076, simple_loss=0.2981, pruned_loss=0.05856, over 16355.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2689, pruned_loss=0.04414, over 3203316.06 frames. ], batch size: 146, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:49:48,388 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 06:50:09,144 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.337e+02 2.736e+02 3.594e+02 9.529e+02, threshold=5.472e+02, percent-clipped=14.0 2023-05-01 06:50:28,303 INFO [train.py:904] (7/8) Epoch 21, batch 5500, loss[loss=0.2158, simple_loss=0.3041, pruned_loss=0.06378, over 16725.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2758, pruned_loss=0.04844, over 3168275.50 frames. ], batch size: 134, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:51:24,866 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 06:51:47,653 INFO [train.py:904] (7/8) Epoch 21, batch 5550, loss[loss=0.213, simple_loss=0.298, pruned_loss=0.06397, over 16397.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2838, pruned_loss=0.05407, over 3148329.05 frames. ], batch size: 146, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:52:17,891 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 06:52:27,813 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=208576.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:52:43,688 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 06:52:49,286 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 3.115e+02 3.921e+02 4.748e+02 1.231e+03, threshold=7.841e+02, percent-clipped=12.0 2023-05-01 06:52:50,111 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.81 vs. limit=5.0 2023-05-01 06:53:07,718 INFO [train.py:904] (7/8) Epoch 21, batch 5600, loss[loss=0.2121, simple_loss=0.2964, pruned_loss=0.06395, over 16427.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2881, pruned_loss=0.05714, over 3133805.98 frames. ], batch size: 146, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:53:39,439 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=208621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:54:07,063 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=208637.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:54:29,867 INFO [train.py:904] (7/8) Epoch 21, batch 5650, loss[loss=0.2204, simple_loss=0.3007, pruned_loss=0.07005, over 15250.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.293, pruned_loss=0.06145, over 3084782.42 frames. ], batch size: 190, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:54:47,213 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7642, 3.1974, 3.2258, 2.0213, 2.7255, 2.1924, 3.2235, 3.4491], device='cuda:7'), covar=tensor([0.0327, 0.0801, 0.0597, 0.2042, 0.0982, 0.1004, 0.0753, 0.0957], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0163, 0.0167, 0.0152, 0.0145, 0.0130, 0.0143, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 06:54:52,088 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2032, 3.1728, 3.5172, 1.7047, 3.6647, 3.6965, 2.8637, 2.6349], device='cuda:7'), covar=tensor([0.0906, 0.0279, 0.0207, 0.1268, 0.0078, 0.0203, 0.0459, 0.0545], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0108, 0.0096, 0.0138, 0.0080, 0.0124, 0.0128, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 06:54:58,003 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 06:55:17,112 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.83 vs. limit=5.0 2023-05-01 06:55:28,208 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=208688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:55:30,665 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.148e+02 3.160e+02 3.845e+02 4.644e+02 8.553e+02, threshold=7.690e+02, percent-clipped=3.0 2023-05-01 06:55:37,746 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2525, 3.4400, 3.5711, 3.5473, 3.5701, 3.3758, 3.4266, 3.4579], device='cuda:7'), covar=tensor([0.0439, 0.0678, 0.0487, 0.0464, 0.0520, 0.0596, 0.0824, 0.0578], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0444, 0.0430, 0.0400, 0.0474, 0.0453, 0.0540, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 06:55:50,863 INFO [train.py:904] (7/8) Epoch 21, batch 5700, loss[loss=0.2242, simple_loss=0.3093, pruned_loss=0.06957, over 16379.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2949, pruned_loss=0.06356, over 3070285.43 frames. ], batch size: 146, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:56:07,249 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8996, 4.0213, 4.3002, 4.2635, 4.2759, 3.9931, 4.0086, 3.9961], device='cuda:7'), covar=tensor([0.0380, 0.0646, 0.0397, 0.0424, 0.0505, 0.0489, 0.1009, 0.0589], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0445, 0.0431, 0.0400, 0.0475, 0.0454, 0.0541, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 06:56:45,736 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=208736.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 06:57:10,974 INFO [train.py:904] (7/8) Epoch 21, batch 5750, loss[loss=0.2493, simple_loss=0.3168, pruned_loss=0.09085, over 11369.00 frames. ], tot_loss[loss=0.2141, simple_loss=0.2979, pruned_loss=0.06513, over 3059135.10 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:58:13,921 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.275e+02 2.983e+02 3.733e+02 4.569e+02 7.128e+02, threshold=7.466e+02, percent-clipped=0.0 2023-05-01 06:58:33,855 INFO [train.py:904] (7/8) Epoch 21, batch 5800, loss[loss=0.1727, simple_loss=0.2659, pruned_loss=0.03968, over 16732.00 frames. ], tot_loss[loss=0.2113, simple_loss=0.2967, pruned_loss=0.06295, over 3082757.60 frames. ], batch size: 89, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 06:58:36,253 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 06:58:47,176 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6908, 4.7436, 4.5828, 4.2638, 4.2449, 4.6637, 4.4567, 4.3806], device='cuda:7'), covar=tensor([0.0686, 0.0704, 0.0319, 0.0332, 0.1015, 0.0571, 0.0505, 0.0646], device='cuda:7'), in_proj_covar=tensor([0.0288, 0.0418, 0.0337, 0.0333, 0.0346, 0.0388, 0.0232, 0.0405], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 06:59:43,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3702, 3.2769, 2.6351, 2.1156, 2.2709, 2.2818, 3.3576, 3.0798], device='cuda:7'), covar=tensor([0.3023, 0.0674, 0.1806, 0.2794, 0.2461, 0.2127, 0.0527, 0.1281], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0267, 0.0301, 0.0307, 0.0293, 0.0254, 0.0293, 0.0332], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 06:59:52,594 INFO [train.py:904] (7/8) Epoch 21, batch 5850, loss[loss=0.1966, simple_loss=0.287, pruned_loss=0.05309, over 16317.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2943, pruned_loss=0.06094, over 3097604.71 frames. ], batch size: 165, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:00:33,281 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 07:00:53,529 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.655e+02 2.740e+02 3.068e+02 3.692e+02 6.527e+02, threshold=6.136e+02, percent-clipped=0.0 2023-05-01 07:01:01,950 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3953, 5.4167, 5.1181, 4.5093, 5.3081, 1.7335, 5.0202, 4.8424], device='cuda:7'), covar=tensor([0.0075, 0.0066, 0.0167, 0.0360, 0.0072, 0.2884, 0.0143, 0.0225], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0150, 0.0192, 0.0173, 0.0169, 0.0200, 0.0181, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:01:12,718 INFO [train.py:904] (7/8) Epoch 21, batch 5900, loss[loss=0.24, simple_loss=0.3083, pruned_loss=0.08589, over 11636.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2941, pruned_loss=0.06065, over 3105617.09 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:01:21,999 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-01 07:01:48,547 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=208921.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:01:54,124 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9179, 5.2437, 4.9721, 5.0218, 4.7487, 4.7007, 4.6040, 5.3492], device='cuda:7'), covar=tensor([0.1248, 0.0854, 0.0984, 0.0920, 0.0828, 0.0986, 0.1291, 0.0830], device='cuda:7'), in_proj_covar=tensor([0.0650, 0.0790, 0.0661, 0.0600, 0.0502, 0.0510, 0.0664, 0.0616], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:02:00,570 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7199, 3.8034, 3.9097, 3.7515, 3.8442, 4.2291, 3.8863, 3.6414], device='cuda:7'), covar=tensor([0.2377, 0.2254, 0.2458, 0.2506, 0.2792, 0.1890, 0.1665, 0.2580], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0586, 0.0642, 0.0490, 0.0647, 0.0674, 0.0505, 0.0653], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 07:02:05,310 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=208932.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:02:21,236 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0226, 3.2129, 3.3400, 2.1759, 3.1286, 3.3781, 3.1534, 1.8828], device='cuda:7'), covar=tensor([0.0547, 0.0101, 0.0071, 0.0419, 0.0111, 0.0124, 0.0103, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0083, 0.0083, 0.0133, 0.0097, 0.0108, 0.0093, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 07:02:36,231 INFO [train.py:904] (7/8) Epoch 21, batch 5950, loss[loss=0.1886, simple_loss=0.2835, pruned_loss=0.04689, over 16155.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2945, pruned_loss=0.05975, over 3093229.88 frames. ], batch size: 35, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:02:39,123 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0730, 5.6548, 5.8782, 5.5668, 5.6684, 6.2009, 5.6934, 5.4743], device='cuda:7'), covar=tensor([0.0985, 0.2005, 0.2324, 0.2016, 0.2381, 0.0871, 0.1553, 0.2291], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0586, 0.0641, 0.0490, 0.0647, 0.0674, 0.0505, 0.0653], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 07:03:02,963 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=208969.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:03:36,546 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.651e+02 3.328e+02 3.727e+02 9.744e+02, threshold=6.656e+02, percent-clipped=4.0 2023-05-01 07:03:56,988 INFO [train.py:904] (7/8) Epoch 21, batch 6000, loss[loss=0.2351, simple_loss=0.3076, pruned_loss=0.08131, over 11623.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2935, pruned_loss=0.05892, over 3112035.66 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:03:56,988 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 07:04:08,272 INFO [train.py:938] (7/8) Epoch 21, validation: loss=0.1512, simple_loss=0.2639, pruned_loss=0.01924, over 944034.00 frames. 2023-05-01 07:04:08,273 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 07:04:10,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5477, 3.5353, 3.4908, 2.7841, 3.3995, 2.0630, 3.1761, 2.8183], device='cuda:7'), covar=tensor([0.0159, 0.0127, 0.0196, 0.0228, 0.0104, 0.2303, 0.0151, 0.0272], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0150, 0.0192, 0.0173, 0.0170, 0.0201, 0.0181, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:04:12,433 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 07:05:25,689 INFO [train.py:904] (7/8) Epoch 21, batch 6050, loss[loss=0.2199, simple_loss=0.2924, pruned_loss=0.07374, over 11621.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2927, pruned_loss=0.0591, over 3098750.70 frames. ], batch size: 248, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:06:26,994 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.899e+02 2.646e+02 3.170e+02 4.163e+02 9.783e+02, threshold=6.340e+02, percent-clipped=1.0 2023-05-01 07:06:45,928 INFO [train.py:904] (7/8) Epoch 21, batch 6100, loss[loss=0.2339, simple_loss=0.3063, pruned_loss=0.08077, over 11832.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2928, pruned_loss=0.05826, over 3119230.43 frames. ], batch size: 246, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:06:47,196 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5147, 4.3779, 4.5381, 4.7005, 4.8497, 4.3846, 4.8462, 4.8668], device='cuda:7'), covar=tensor([0.1876, 0.1201, 0.1563, 0.0726, 0.0603, 0.0985, 0.0619, 0.0647], device='cuda:7'), in_proj_covar=tensor([0.0621, 0.0765, 0.0894, 0.0780, 0.0587, 0.0617, 0.0634, 0.0735], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:07:07,200 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-05-01 07:08:04,726 INFO [train.py:904] (7/8) Epoch 21, batch 6150, loss[loss=0.1777, simple_loss=0.2662, pruned_loss=0.04457, over 16173.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.291, pruned_loss=0.05786, over 3101640.03 frames. ], batch size: 35, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:08:45,845 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8702, 3.1930, 3.1801, 2.0095, 2.9745, 3.1931, 2.9783, 1.9595], device='cuda:7'), covar=tensor([0.0574, 0.0064, 0.0075, 0.0451, 0.0112, 0.0130, 0.0111, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0082, 0.0082, 0.0133, 0.0097, 0.0108, 0.0093, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 07:08:58,929 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2023-05-01 07:09:04,300 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.750e+02 2.685e+02 3.365e+02 4.144e+02 7.779e+02, threshold=6.731e+02, percent-clipped=2.0 2023-05-01 07:09:23,303 INFO [train.py:904] (7/8) Epoch 21, batch 6200, loss[loss=0.2063, simple_loss=0.2773, pruned_loss=0.06766, over 11639.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.289, pruned_loss=0.05774, over 3090562.74 frames. ], batch size: 247, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:09:23,830 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209202.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:10:07,072 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3050, 4.3741, 4.1895, 3.9700, 3.9378, 4.2952, 4.0664, 4.0236], device='cuda:7'), covar=tensor([0.0629, 0.0619, 0.0304, 0.0297, 0.0759, 0.0514, 0.0641, 0.0639], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0415, 0.0335, 0.0331, 0.0343, 0.0385, 0.0230, 0.0402], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:10:10,203 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209232.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:10:28,269 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-05-01 07:10:35,188 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0461, 2.0891, 2.2055, 3.5578, 2.0406, 2.4297, 2.2317, 2.2604], device='cuda:7'), covar=tensor([0.1469, 0.3657, 0.3038, 0.0624, 0.4351, 0.2617, 0.3724, 0.3339], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0444, 0.0364, 0.0325, 0.0434, 0.0511, 0.0414, 0.0519], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:10:39,918 INFO [train.py:904] (7/8) Epoch 21, batch 6250, loss[loss=0.1767, simple_loss=0.2767, pruned_loss=0.03833, over 16525.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2882, pruned_loss=0.05693, over 3120057.42 frames. ], batch size: 75, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:10:46,191 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209255.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:10:57,648 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209263.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:11:22,160 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=209280.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:11:32,778 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-01 07:11:35,897 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.681e+02 3.167e+02 3.896e+02 8.805e+02, threshold=6.333e+02, percent-clipped=2.0 2023-05-01 07:11:37,216 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0175, 4.8297, 5.0272, 5.1968, 5.3818, 4.7368, 5.3998, 5.3816], device='cuda:7'), covar=tensor([0.1905, 0.1197, 0.1698, 0.0765, 0.0616, 0.0913, 0.0652, 0.0538], device='cuda:7'), in_proj_covar=tensor([0.0622, 0.0766, 0.0894, 0.0780, 0.0585, 0.0616, 0.0635, 0.0736], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:11:50,422 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209298.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:11:54,999 INFO [train.py:904] (7/8) Epoch 21, batch 6300, loss[loss=0.2152, simple_loss=0.2861, pruned_loss=0.07219, over 11282.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2878, pruned_loss=0.05622, over 3127101.08 frames. ], batch size: 247, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:12:17,981 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209316.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:13:13,585 INFO [train.py:904] (7/8) Epoch 21, batch 6350, loss[loss=0.1972, simple_loss=0.2887, pruned_loss=0.05286, over 16551.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.289, pruned_loss=0.05771, over 3112134.71 frames. ], batch size: 68, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:13:25,012 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209359.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:14:13,989 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 3.041e+02 3.551e+02 4.346e+02 9.300e+02, threshold=7.102e+02, percent-clipped=4.0 2023-05-01 07:14:31,829 INFO [train.py:904] (7/8) Epoch 21, batch 6400, loss[loss=0.1829, simple_loss=0.2749, pruned_loss=0.0454, over 16912.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2891, pruned_loss=0.05869, over 3114488.42 frames. ], batch size: 96, lr: 3.21e-03, grad_scale: 8.0 2023-05-01 07:15:48,019 INFO [train.py:904] (7/8) Epoch 21, batch 6450, loss[loss=0.2089, simple_loss=0.2974, pruned_loss=0.06017, over 16465.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2895, pruned_loss=0.05856, over 3112947.62 frames. ], batch size: 146, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:16:10,783 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209466.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:16:52,761 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 2.825e+02 3.422e+02 4.079e+02 7.372e+02, threshold=6.844e+02, percent-clipped=3.0 2023-05-01 07:17:08,467 INFO [train.py:904] (7/8) Epoch 21, batch 6500, loss[loss=0.1851, simple_loss=0.2815, pruned_loss=0.04439, over 16689.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2881, pruned_loss=0.05829, over 3087853.93 frames. ], batch size: 89, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:17:21,415 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209510.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:17:48,033 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209527.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:18:20,951 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8505, 3.8584, 2.5199, 4.4654, 2.9871, 4.3447, 2.3661, 3.0160], device='cuda:7'), covar=tensor([0.0244, 0.0360, 0.1506, 0.0211, 0.0725, 0.0644, 0.1710, 0.0819], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0175, 0.0193, 0.0160, 0.0174, 0.0214, 0.0199, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:18:28,677 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9415, 4.9472, 4.8374, 4.0574, 4.8236, 1.9085, 4.5231, 4.5036], device='cuda:7'), covar=tensor([0.0099, 0.0090, 0.0185, 0.0384, 0.0091, 0.2858, 0.0139, 0.0208], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0152, 0.0195, 0.0175, 0.0172, 0.0204, 0.0184, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:18:29,361 INFO [train.py:904] (7/8) Epoch 21, batch 6550, loss[loss=0.2094, simple_loss=0.3193, pruned_loss=0.0498, over 16421.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.291, pruned_loss=0.05859, over 3107912.61 frames. ], batch size: 75, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:18:40,052 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209558.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:19:01,146 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209571.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:19:04,341 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.22 vs. limit=5.0 2023-05-01 07:19:33,743 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.646e+02 3.318e+02 3.875e+02 1.016e+03, threshold=6.636e+02, percent-clipped=2.0 2023-05-01 07:19:49,530 INFO [train.py:904] (7/8) Epoch 21, batch 6600, loss[loss=0.2125, simple_loss=0.3045, pruned_loss=0.06027, over 16852.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.293, pruned_loss=0.05937, over 3088656.62 frames. ], batch size: 42, lr: 3.21e-03, grad_scale: 4.0 2023-05-01 07:19:55,984 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209606.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:20:03,748 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209611.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:20:06,093 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-01 07:20:33,049 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 07:20:46,751 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6489, 4.6398, 4.4972, 3.7905, 4.5764, 1.6773, 4.3412, 4.2295], device='cuda:7'), covar=tensor([0.0102, 0.0081, 0.0178, 0.0385, 0.0092, 0.2893, 0.0131, 0.0257], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0152, 0.0196, 0.0176, 0.0173, 0.0205, 0.0185, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:20:59,573 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3195, 4.3044, 4.2214, 3.4876, 4.2830, 1.7543, 4.0387, 3.8928], device='cuda:7'), covar=tensor([0.0121, 0.0102, 0.0183, 0.0356, 0.0093, 0.2909, 0.0147, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0152, 0.0196, 0.0176, 0.0173, 0.0205, 0.0185, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:21:08,006 INFO [train.py:904] (7/8) Epoch 21, batch 6650, loss[loss=0.1928, simple_loss=0.2767, pruned_loss=0.05449, over 16859.00 frames. ], tot_loss[loss=0.207, simple_loss=0.2932, pruned_loss=0.06039, over 3077304.45 frames. ], batch size: 116, lr: 3.21e-03, grad_scale: 2.0 2023-05-01 07:21:12,310 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209654.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:21:33,103 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209667.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:22:05,637 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=209688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:22:11,939 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.980e+02 3.003e+02 3.834e+02 5.149e+02 1.142e+03, threshold=7.669e+02, percent-clipped=10.0 2023-05-01 07:22:25,649 INFO [train.py:904] (7/8) Epoch 21, batch 6700, loss[loss=0.1874, simple_loss=0.274, pruned_loss=0.05041, over 15381.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2917, pruned_loss=0.06005, over 3100197.07 frames. ], batch size: 190, lr: 3.21e-03, grad_scale: 2.0 2023-05-01 07:23:11,479 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8737, 3.1712, 3.3128, 1.9334, 2.8655, 2.0603, 3.3480, 3.4202], device='cuda:7'), covar=tensor([0.0230, 0.0788, 0.0580, 0.2072, 0.0834, 0.1048, 0.0608, 0.0906], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0163, 0.0167, 0.0152, 0.0144, 0.0129, 0.0143, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:23:40,266 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=209749.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:23:44,129 INFO [train.py:904] (7/8) Epoch 21, batch 6750, loss[loss=0.1969, simple_loss=0.2735, pruned_loss=0.0601, over 17126.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2919, pruned_loss=0.06093, over 3080869.23 frames. ], batch size: 48, lr: 3.20e-03, grad_scale: 2.0 2023-05-01 07:24:47,249 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.958e+02 2.606e+02 3.292e+02 3.909e+02 7.681e+02, threshold=6.584e+02, percent-clipped=1.0 2023-05-01 07:25:01,552 INFO [train.py:904] (7/8) Epoch 21, batch 6800, loss[loss=0.1974, simple_loss=0.289, pruned_loss=0.05288, over 16266.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2906, pruned_loss=0.06017, over 3100060.77 frames. ], batch size: 165, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:25:33,969 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209822.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:26:20,416 INFO [train.py:904] (7/8) Epoch 21, batch 6850, loss[loss=0.1991, simple_loss=0.3103, pruned_loss=0.04394, over 16729.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2923, pruned_loss=0.0601, over 3116090.40 frames. ], batch size: 89, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:26:29,687 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209858.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:26:37,384 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9805, 3.8698, 4.4140, 2.1111, 4.6433, 4.6522, 3.6044, 3.2333], device='cuda:7'), covar=tensor([0.0701, 0.0226, 0.0135, 0.1137, 0.0046, 0.0110, 0.0250, 0.0485], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0105, 0.0095, 0.0136, 0.0078, 0.0122, 0.0126, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 07:26:42,204 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209866.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:27:21,614 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.648e+02 3.249e+02 3.781e+02 8.526e+02, threshold=6.499e+02, percent-clipped=1.0 2023-05-01 07:27:29,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5360, 4.6199, 4.4014, 4.1110, 4.0911, 4.5129, 4.3089, 4.2275], device='cuda:7'), covar=tensor([0.0785, 0.0972, 0.0378, 0.0387, 0.1001, 0.0832, 0.0747, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0410, 0.0331, 0.0327, 0.0341, 0.0381, 0.0228, 0.0399], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:27:34,659 INFO [train.py:904] (7/8) Epoch 21, batch 6900, loss[loss=0.2481, simple_loss=0.3202, pruned_loss=0.08799, over 11862.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2938, pruned_loss=0.05882, over 3140176.19 frames. ], batch size: 246, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:27:41,307 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=209906.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:27:49,080 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209911.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:28:51,094 INFO [train.py:904] (7/8) Epoch 21, batch 6950, loss[loss=0.2237, simple_loss=0.3047, pruned_loss=0.07132, over 16699.00 frames. ], tot_loss[loss=0.2077, simple_loss=0.2947, pruned_loss=0.06031, over 3127916.84 frames. ], batch size: 124, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:28:54,543 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=209954.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:29:02,108 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=209959.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:29:06,371 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=209962.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:29:51,933 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.035e+02 3.251e+02 3.766e+02 4.564e+02 9.580e+02, threshold=7.532e+02, percent-clipped=9.0 2023-05-01 07:30:09,079 INFO [train.py:904] (7/8) Epoch 21, batch 7000, loss[loss=0.2174, simple_loss=0.2954, pruned_loss=0.0697, over 11810.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2954, pruned_loss=0.06062, over 3103393.05 frames. ], batch size: 249, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:30:09,362 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210002.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:31:11,197 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=210044.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:31:11,416 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5208, 3.0370, 3.1598, 1.8850, 2.6790, 1.8808, 3.2568, 3.2259], device='cuda:7'), covar=tensor([0.0294, 0.0810, 0.0615, 0.2174, 0.0932, 0.1140, 0.0645, 0.0935], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0162, 0.0166, 0.0151, 0.0143, 0.0128, 0.0142, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:31:22,819 INFO [train.py:904] (7/8) Epoch 21, batch 7050, loss[loss=0.2065, simple_loss=0.2924, pruned_loss=0.06031, over 16642.00 frames. ], tot_loss[loss=0.2082, simple_loss=0.296, pruned_loss=0.06024, over 3113624.32 frames. ], batch size: 57, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:31:28,119 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8750, 2.2101, 2.3774, 1.8507, 2.5809, 2.7268, 2.3584, 2.2037], device='cuda:7'), covar=tensor([0.0934, 0.0264, 0.0231, 0.1045, 0.0123, 0.0243, 0.0474, 0.0560], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0096, 0.0136, 0.0078, 0.0122, 0.0127, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 07:31:30,646 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210057.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:32:24,163 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.868e+02 2.885e+02 3.523e+02 4.096e+02 8.338e+02, threshold=7.047e+02, percent-clipped=1.0 2023-05-01 07:32:29,111 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7551, 2.4080, 1.9991, 1.9692, 2.6901, 2.3126, 2.5834, 2.8227], device='cuda:7'), covar=tensor([0.0203, 0.0389, 0.0504, 0.0512, 0.0240, 0.0365, 0.0233, 0.0262], device='cuda:7'), in_proj_covar=tensor([0.0202, 0.0230, 0.0222, 0.0223, 0.0231, 0.0229, 0.0230, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:32:37,568 INFO [train.py:904] (7/8) Epoch 21, batch 7100, loss[loss=0.2106, simple_loss=0.2908, pruned_loss=0.06513, over 16910.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2948, pruned_loss=0.06072, over 3097407.90 frames. ], batch size: 109, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:33:03,347 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210118.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:33:10,048 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210122.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:33:55,429 INFO [train.py:904] (7/8) Epoch 21, batch 7150, loss[loss=0.186, simple_loss=0.2763, pruned_loss=0.0478, over 16909.00 frames. ], tot_loss[loss=0.2074, simple_loss=0.2929, pruned_loss=0.06096, over 3069566.53 frames. ], batch size: 96, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:34:16,404 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210166.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:34:21,341 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210170.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:34:33,629 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1794, 2.9862, 3.2685, 1.7169, 3.4823, 3.5144, 2.7341, 2.6532], device='cuda:7'), covar=tensor([0.0860, 0.0299, 0.0214, 0.1268, 0.0073, 0.0168, 0.0462, 0.0481], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0106, 0.0096, 0.0136, 0.0078, 0.0122, 0.0126, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 07:34:50,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9728, 2.2894, 2.3154, 2.7101, 1.9631, 3.1874, 1.7759, 2.7197], device='cuda:7'), covar=tensor([0.1142, 0.0676, 0.1147, 0.0187, 0.0111, 0.0358, 0.1520, 0.0736], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0174, 0.0196, 0.0190, 0.0207, 0.0215, 0.0202, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:34:53,946 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.076e+02 2.584e+02 3.186e+02 4.207e+02 8.537e+02, threshold=6.371e+02, percent-clipped=2.0 2023-05-01 07:35:07,924 INFO [train.py:904] (7/8) Epoch 21, batch 7200, loss[loss=0.1859, simple_loss=0.2726, pruned_loss=0.0496, over 15146.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2905, pruned_loss=0.05911, over 3063832.93 frames. ], batch size: 190, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:35:27,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210214.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:36:28,296 INFO [train.py:904] (7/8) Epoch 21, batch 7250, loss[loss=0.1731, simple_loss=0.2546, pruned_loss=0.04575, over 16612.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2878, pruned_loss=0.05773, over 3057154.73 frames. ], batch size: 62, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:36:43,839 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210262.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:37:29,738 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 07:37:31,239 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2012, 3.6710, 3.3722, 1.8893, 2.7536, 1.9582, 3.5284, 3.8660], device='cuda:7'), covar=tensor([0.0250, 0.0739, 0.0792, 0.2665, 0.1238, 0.1435, 0.0709, 0.0931], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0164, 0.0168, 0.0153, 0.0145, 0.0129, 0.0144, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:37:31,436 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.99 vs. limit=5.0 2023-05-01 07:37:31,773 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 2.676e+02 3.072e+02 3.848e+02 7.717e+02, threshold=6.145e+02, percent-clipped=2.0 2023-05-01 07:37:44,664 INFO [train.py:904] (7/8) Epoch 21, batch 7300, loss[loss=0.183, simple_loss=0.2712, pruned_loss=0.04741, over 16629.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2876, pruned_loss=0.05798, over 3043600.01 frames. ], batch size: 57, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:37:47,425 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-05-01 07:37:58,861 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210310.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:38:48,819 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210344.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:38:59,616 INFO [train.py:904] (7/8) Epoch 21, batch 7350, loss[loss=0.2033, simple_loss=0.293, pruned_loss=0.05678, over 16333.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2886, pruned_loss=0.05855, over 3030998.69 frames. ], batch size: 165, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:40:00,159 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210392.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:40:02,084 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.648e+02 3.204e+02 3.841e+02 6.410e+02, threshold=6.409e+02, percent-clipped=2.0 2023-05-01 07:40:14,865 INFO [train.py:904] (7/8) Epoch 21, batch 7400, loss[loss=0.2633, simple_loss=0.3257, pruned_loss=0.1004, over 11556.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2893, pruned_loss=0.05895, over 3030880.74 frames. ], batch size: 248, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:40:32,840 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=210413.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:41:20,258 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5988, 2.5376, 2.5579, 4.5050, 2.4694, 2.9958, 2.5951, 2.7352], device='cuda:7'), covar=tensor([0.1205, 0.3209, 0.2610, 0.0413, 0.3703, 0.2278, 0.3179, 0.2873], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0445, 0.0362, 0.0324, 0.0434, 0.0511, 0.0414, 0.0517], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:41:32,233 INFO [train.py:904] (7/8) Epoch 21, batch 7450, loss[loss=0.1931, simple_loss=0.2885, pruned_loss=0.04886, over 16446.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2899, pruned_loss=0.05904, over 3060377.47 frames. ], batch size: 68, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:41:42,858 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1756, 3.7099, 3.8542, 2.4062, 3.5182, 3.8700, 3.4660, 1.9823], device='cuda:7'), covar=tensor([0.0572, 0.0068, 0.0052, 0.0455, 0.0099, 0.0106, 0.0092, 0.0508], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0082, 0.0083, 0.0134, 0.0096, 0.0108, 0.0093, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 07:41:44,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9983, 5.0104, 4.8684, 4.5298, 4.5480, 4.9484, 4.8729, 4.6088], device='cuda:7'), covar=tensor([0.0690, 0.0808, 0.0311, 0.0311, 0.0990, 0.0623, 0.0310, 0.0700], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0408, 0.0327, 0.0323, 0.0338, 0.0378, 0.0227, 0.0394], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:42:10,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1268, 5.7025, 5.9399, 5.5397, 5.7041, 6.2424, 5.6991, 5.4415], device='cuda:7'), covar=tensor([0.0929, 0.1992, 0.2477, 0.2080, 0.2503, 0.1079, 0.1729, 0.2519], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0589, 0.0648, 0.0491, 0.0648, 0.0681, 0.0507, 0.0661], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 07:42:42,642 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.331e+02 3.072e+02 3.553e+02 4.443e+02 7.195e+02, threshold=7.106e+02, percent-clipped=1.0 2023-05-01 07:42:53,258 INFO [train.py:904] (7/8) Epoch 21, batch 7500, loss[loss=0.1736, simple_loss=0.2567, pruned_loss=0.04526, over 16522.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.29, pruned_loss=0.05874, over 3048808.04 frames. ], batch size: 68, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:43:35,982 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9214, 4.9007, 5.2873, 5.2568, 5.3019, 4.9360, 4.9204, 4.7115], device='cuda:7'), covar=tensor([0.0312, 0.0496, 0.0337, 0.0385, 0.0504, 0.0360, 0.1028, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0440, 0.0429, 0.0400, 0.0475, 0.0451, 0.0537, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 07:44:09,158 INFO [train.py:904] (7/8) Epoch 21, batch 7550, loss[loss=0.2093, simple_loss=0.2918, pruned_loss=0.06345, over 16882.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2897, pruned_loss=0.05949, over 3023454.53 frames. ], batch size: 109, lr: 3.20e-03, grad_scale: 4.0 2023-05-01 07:44:09,622 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4945, 4.5593, 4.3733, 4.0767, 4.0988, 4.4799, 4.2074, 4.1782], device='cuda:7'), covar=tensor([0.0581, 0.0438, 0.0282, 0.0281, 0.0808, 0.0421, 0.0611, 0.0597], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0405, 0.0326, 0.0321, 0.0336, 0.0375, 0.0226, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:45:11,351 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.786e+02 2.775e+02 3.392e+02 4.108e+02 6.854e+02, threshold=6.785e+02, percent-clipped=0.0 2023-05-01 07:45:23,238 INFO [train.py:904] (7/8) Epoch 21, batch 7600, loss[loss=0.2154, simple_loss=0.3018, pruned_loss=0.06445, over 16269.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2894, pruned_loss=0.05974, over 3045823.72 frames. ], batch size: 165, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:46:08,236 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210632.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:46:37,240 INFO [train.py:904] (7/8) Epoch 21, batch 7650, loss[loss=0.2387, simple_loss=0.3193, pruned_loss=0.079, over 15372.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2901, pruned_loss=0.06005, over 3058409.93 frames. ], batch size: 190, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:46:59,515 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210666.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:47:25,974 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210682.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:47:42,146 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210693.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:47:42,760 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 3.052e+02 3.513e+02 4.198e+02 7.732e+02, threshold=7.025e+02, percent-clipped=2.0 2023-05-01 07:47:55,353 INFO [train.py:904] (7/8) Epoch 21, batch 7700, loss[loss=0.1949, simple_loss=0.2838, pruned_loss=0.05299, over 15500.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2901, pruned_loss=0.0605, over 3058350.08 frames. ], batch size: 191, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:48:12,316 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=210713.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:48:17,654 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 07:48:34,129 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210727.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:48:59,646 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210743.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:49:12,537 INFO [train.py:904] (7/8) Epoch 21, batch 7750, loss[loss=0.2278, simple_loss=0.2989, pruned_loss=0.07838, over 11310.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2905, pruned_loss=0.06036, over 3066731.74 frames. ], batch size: 250, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:49:27,500 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=210761.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:49:31,264 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5655, 4.6320, 4.4441, 4.1723, 4.1534, 4.5543, 4.2384, 4.2517], device='cuda:7'), covar=tensor([0.0589, 0.0528, 0.0280, 0.0274, 0.0842, 0.0501, 0.0565, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0279, 0.0405, 0.0326, 0.0320, 0.0335, 0.0375, 0.0225, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:49:33,040 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210764.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:50:18,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8389, 2.8210, 2.5890, 4.8201, 3.6273, 4.2532, 1.8851, 3.1049], device='cuda:7'), covar=tensor([0.1321, 0.0778, 0.1245, 0.0188, 0.0315, 0.0390, 0.1518, 0.0817], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0175, 0.0196, 0.0190, 0.0207, 0.0216, 0.0202, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:50:18,622 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 2.795e+02 3.327e+02 4.093e+02 7.547e+02, threshold=6.654e+02, percent-clipped=2.0 2023-05-01 07:50:31,011 INFO [train.py:904] (7/8) Epoch 21, batch 7800, loss[loss=0.1754, simple_loss=0.2727, pruned_loss=0.03902, over 16725.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2908, pruned_loss=0.06052, over 3083056.38 frames. ], batch size: 83, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:50:31,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4409, 3.2646, 3.5769, 1.8148, 3.7601, 3.8071, 2.9326, 2.8067], device='cuda:7'), covar=tensor([0.0809, 0.0275, 0.0194, 0.1252, 0.0076, 0.0180, 0.0432, 0.0460], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0096, 0.0137, 0.0079, 0.0123, 0.0127, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 07:51:07,596 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=210825.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:51:48,919 INFO [train.py:904] (7/8) Epoch 21, batch 7850, loss[loss=0.1804, simple_loss=0.2808, pruned_loss=0.04001, over 16791.00 frames. ], tot_loss[loss=0.206, simple_loss=0.2913, pruned_loss=0.06038, over 3063246.17 frames. ], batch size: 102, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:52:31,801 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3631, 2.9206, 3.0655, 1.9774, 2.7981, 2.1164, 3.0268, 3.1800], device='cuda:7'), covar=tensor([0.0285, 0.0820, 0.0628, 0.2050, 0.0826, 0.1056, 0.0646, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0163, 0.0166, 0.0151, 0.0144, 0.0128, 0.0143, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:52:54,101 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.040e+02 2.722e+02 3.190e+02 3.938e+02 6.813e+02, threshold=6.379e+02, percent-clipped=1.0 2023-05-01 07:53:05,686 INFO [train.py:904] (7/8) Epoch 21, batch 7900, loss[loss=0.192, simple_loss=0.2824, pruned_loss=0.05083, over 16480.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2902, pruned_loss=0.05966, over 3065656.37 frames. ], batch size: 68, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:54:24,392 INFO [train.py:904] (7/8) Epoch 21, batch 7950, loss[loss=0.2732, simple_loss=0.3276, pruned_loss=0.1094, over 11270.00 frames. ], tot_loss[loss=0.2067, simple_loss=0.2917, pruned_loss=0.06087, over 3062615.03 frames. ], batch size: 246, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:55:14,462 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=210984.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:55:18,443 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 07:55:20,869 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=210988.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:55:25,354 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0384, 5.0567, 4.8185, 3.5054, 4.9315, 1.5967, 4.5076, 4.5274], device='cuda:7'), covar=tensor([0.0209, 0.0171, 0.0293, 0.0832, 0.0160, 0.3650, 0.0241, 0.0396], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0151, 0.0195, 0.0174, 0.0172, 0.0204, 0.0182, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:55:29,115 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.961e+02 3.312e+02 4.005e+02 9.237e+02, threshold=6.624e+02, percent-clipped=4.0 2023-05-01 07:55:41,333 INFO [train.py:904] (7/8) Epoch 21, batch 8000, loss[loss=0.2332, simple_loss=0.3159, pruned_loss=0.07527, over 15255.00 frames. ], tot_loss[loss=0.2081, simple_loss=0.2926, pruned_loss=0.06178, over 3048674.96 frames. ], batch size: 190, lr: 3.20e-03, grad_scale: 8.0 2023-05-01 07:56:12,620 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211022.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:56:31,482 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8736, 2.7157, 2.8347, 2.1075, 2.7375, 2.1623, 2.6548, 2.8999], device='cuda:7'), covar=tensor([0.0271, 0.0768, 0.0455, 0.1775, 0.0719, 0.0866, 0.0530, 0.0683], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0162, 0.0166, 0.0151, 0.0143, 0.0128, 0.0142, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:56:35,933 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211038.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:56:45,520 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=211045.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:56:56,541 INFO [train.py:904] (7/8) Epoch 21, batch 8050, loss[loss=0.2164, simple_loss=0.2974, pruned_loss=0.06772, over 16848.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2922, pruned_loss=0.06105, over 3055270.31 frames. ], batch size: 116, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 07:57:36,696 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6069, 1.5635, 2.1723, 2.5113, 2.5316, 2.8995, 1.7723, 2.8331], device='cuda:7'), covar=tensor([0.0201, 0.0650, 0.0314, 0.0326, 0.0310, 0.0179, 0.0657, 0.0155], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0191, 0.0176, 0.0182, 0.0194, 0.0151, 0.0194, 0.0148], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 07:57:59,280 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.685e+02 2.757e+02 3.257e+02 3.943e+02 6.625e+02, threshold=6.515e+02, percent-clipped=2.0 2023-05-01 07:58:10,498 INFO [train.py:904] (7/8) Epoch 21, batch 8100, loss[loss=0.1889, simple_loss=0.2781, pruned_loss=0.04988, over 16611.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2912, pruned_loss=0.05972, over 3080446.61 frames. ], batch size: 62, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 07:58:25,916 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2183, 3.4296, 3.5854, 1.9894, 3.1372, 2.3581, 3.5392, 3.6294], device='cuda:7'), covar=tensor([0.0227, 0.0776, 0.0568, 0.2116, 0.0800, 0.0967, 0.0599, 0.0890], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0163, 0.0166, 0.0152, 0.0144, 0.0129, 0.0142, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 07:58:38,287 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211120.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 07:59:01,936 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.64 vs. limit=5.0 2023-05-01 07:59:21,508 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-05-01 07:59:22,872 INFO [train.py:904] (7/8) Epoch 21, batch 8150, loss[loss=0.1878, simple_loss=0.265, pruned_loss=0.05533, over 17036.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2889, pruned_loss=0.05897, over 3078967.53 frames. ], batch size: 55, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:00:09,891 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 08:00:27,471 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.753e+02 3.326e+02 4.060e+02 8.278e+02, threshold=6.652e+02, percent-clipped=2.0 2023-05-01 08:00:40,754 INFO [train.py:904] (7/8) Epoch 21, batch 8200, loss[loss=0.2061, simple_loss=0.2912, pruned_loss=0.06052, over 15442.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2865, pruned_loss=0.0582, over 3110050.77 frames. ], batch size: 190, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:01:59,684 INFO [train.py:904] (7/8) Epoch 21, batch 8250, loss[loss=0.1787, simple_loss=0.2643, pruned_loss=0.04654, over 12085.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2852, pruned_loss=0.0557, over 3066516.82 frames. ], batch size: 248, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:02:56,881 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211288.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:03:06,693 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.346e+02 2.848e+02 3.670e+02 8.296e+02, threshold=5.695e+02, percent-clipped=3.0 2023-05-01 08:03:18,688 INFO [train.py:904] (7/8) Epoch 21, batch 8300, loss[loss=0.1659, simple_loss=0.269, pruned_loss=0.03137, over 16254.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.283, pruned_loss=0.053, over 3075860.42 frames. ], batch size: 165, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:03:51,867 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211322.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:12,347 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211336.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:16,089 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211338.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:18,492 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:04:36,937 INFO [train.py:904] (7/8) Epoch 21, batch 8350, loss[loss=0.2066, simple_loss=0.2986, pruned_loss=0.05729, over 16181.00 frames. ], tot_loss[loss=0.1924, simple_loss=0.2823, pruned_loss=0.05125, over 3079300.48 frames. ], batch size: 165, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:05:05,458 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211370.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:05:30,361 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211386.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:05:30,535 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=211386.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:05:43,491 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.192e+02 2.619e+02 3.085e+02 8.186e+02, threshold=5.238e+02, percent-clipped=3.0 2023-05-01 08:05:46,429 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-05-01 08:05:55,513 INFO [train.py:904] (7/8) Epoch 21, batch 8400, loss[loss=0.167, simple_loss=0.2519, pruned_loss=0.04108, over 12201.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2797, pruned_loss=0.04925, over 3081469.48 frames. ], batch size: 248, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:06:10,946 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=211412.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:06:22,575 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211420.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:07:04,431 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=211447.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:07:10,969 INFO [train.py:904] (7/8) Epoch 21, batch 8450, loss[loss=0.1788, simple_loss=0.261, pruned_loss=0.04829, over 12455.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2776, pruned_loss=0.04747, over 3073245.59 frames. ], batch size: 247, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:07:36,210 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211468.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:07:41,996 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5079, 3.0808, 2.7498, 2.3130, 2.2215, 2.3124, 3.0707, 2.9187], device='cuda:7'), covar=tensor([0.2592, 0.0757, 0.1529, 0.2753, 0.2909, 0.2303, 0.0496, 0.1363], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0265, 0.0300, 0.0307, 0.0292, 0.0254, 0.0290, 0.0329], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 08:07:45,014 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=211473.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:08:18,594 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.143e+02 2.501e+02 3.093e+02 5.428e+02, threshold=5.001e+02, percent-clipped=1.0 2023-05-01 08:08:30,002 INFO [train.py:904] (7/8) Epoch 21, batch 8500, loss[loss=0.1718, simple_loss=0.2505, pruned_loss=0.04658, over 11991.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.274, pruned_loss=0.04555, over 3052693.67 frames. ], batch size: 248, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:09:00,063 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6402, 3.7059, 2.3712, 4.1193, 2.8930, 4.1247, 2.5018, 3.1433], device='cuda:7'), covar=tensor([0.0254, 0.0316, 0.1429, 0.0210, 0.0704, 0.0421, 0.1390, 0.0602], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0170, 0.0189, 0.0155, 0.0172, 0.0210, 0.0197, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 08:09:54,391 INFO [train.py:904] (7/8) Epoch 21, batch 8550, loss[loss=0.1607, simple_loss=0.251, pruned_loss=0.0352, over 12123.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2717, pruned_loss=0.04431, over 3035263.21 frames. ], batch size: 248, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:11:18,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.272e+02 2.648e+02 3.114e+02 4.500e+02, threshold=5.297e+02, percent-clipped=0.0 2023-05-01 08:11:32,300 INFO [train.py:904] (7/8) Epoch 21, batch 8600, loss[loss=0.1709, simple_loss=0.2677, pruned_loss=0.03703, over 16598.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2722, pruned_loss=0.04341, over 3041501.97 frames. ], batch size: 62, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:11:35,722 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7846, 2.6433, 2.2969, 3.8774, 2.3449, 3.8975, 1.5396, 2.8797], device='cuda:7'), covar=tensor([0.1337, 0.0753, 0.1292, 0.0181, 0.0096, 0.0313, 0.1660, 0.0731], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0172, 0.0193, 0.0186, 0.0203, 0.0212, 0.0199, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 08:12:48,844 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=211640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:12:57,042 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3498, 2.9461, 2.6442, 2.2070, 2.1126, 2.2646, 2.8772, 2.7290], device='cuda:7'), covar=tensor([0.2785, 0.0686, 0.1735, 0.2973, 0.2907, 0.2232, 0.0481, 0.1550], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0264, 0.0301, 0.0307, 0.0291, 0.0254, 0.0290, 0.0329], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 08:13:09,691 INFO [train.py:904] (7/8) Epoch 21, batch 8650, loss[loss=0.1625, simple_loss=0.2626, pruned_loss=0.03117, over 16464.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2704, pruned_loss=0.04176, over 3052476.74 frames. ], batch size: 68, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:14:01,648 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7616, 2.0286, 2.2430, 2.7952, 2.6832, 3.0516, 2.0401, 2.9798], device='cuda:7'), covar=tensor([0.0191, 0.0550, 0.0355, 0.0286, 0.0324, 0.0176, 0.0542, 0.0186], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0189, 0.0173, 0.0178, 0.0191, 0.0148, 0.0191, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:14:30,914 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=211688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:14:43,486 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.069e+02 2.535e+02 3.155e+02 6.979e+02, threshold=5.069e+02, percent-clipped=2.0 2023-05-01 08:14:57,335 INFO [train.py:904] (7/8) Epoch 21, batch 8700, loss[loss=0.1741, simple_loss=0.2706, pruned_loss=0.03878, over 15249.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2678, pruned_loss=0.04062, over 3047148.94 frames. ], batch size: 191, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:16:10,367 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211742.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:16:31,419 INFO [train.py:904] (7/8) Epoch 21, batch 8750, loss[loss=0.1902, simple_loss=0.2858, pruned_loss=0.04737, over 15383.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2679, pruned_loss=0.03997, over 3059907.31 frames. ], batch size: 190, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:16:56,173 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9381, 3.1945, 2.7023, 5.0709, 3.8438, 4.4419, 1.7197, 3.2140], device='cuda:7'), covar=tensor([0.1215, 0.0661, 0.1104, 0.0179, 0.0165, 0.0311, 0.1474, 0.0643], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0170, 0.0191, 0.0184, 0.0201, 0.0210, 0.0198, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 08:17:13,766 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=211768.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:18:09,824 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.163e+02 2.564e+02 3.220e+02 5.959e+02, threshold=5.128e+02, percent-clipped=1.0 2023-05-01 08:18:23,762 INFO [train.py:904] (7/8) Epoch 21, batch 8800, loss[loss=0.1746, simple_loss=0.258, pruned_loss=0.04556, over 12096.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2668, pruned_loss=0.039, over 3080496.25 frames. ], batch size: 247, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:18:51,203 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6741, 1.9520, 2.1810, 2.6652, 2.5338, 2.9515, 1.9899, 2.9654], device='cuda:7'), covar=tensor([0.0226, 0.0512, 0.0396, 0.0296, 0.0339, 0.0213, 0.0506, 0.0155], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0189, 0.0173, 0.0178, 0.0192, 0.0148, 0.0190, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:19:02,545 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 08:20:07,852 INFO [train.py:904] (7/8) Epoch 21, batch 8850, loss[loss=0.1865, simple_loss=0.2861, pruned_loss=0.0435, over 16914.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2695, pruned_loss=0.03834, over 3087411.05 frames. ], batch size: 116, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:20:17,788 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4114, 3.4667, 3.6553, 3.6428, 3.6718, 3.5061, 3.5193, 3.5168], device='cuda:7'), covar=tensor([0.0376, 0.0656, 0.0536, 0.0496, 0.0438, 0.0511, 0.0713, 0.0491], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0436, 0.0423, 0.0394, 0.0469, 0.0442, 0.0524, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 08:20:20,013 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5968, 4.5926, 4.3943, 3.3083, 4.5080, 1.3740, 4.1254, 4.1266], device='cuda:7'), covar=tensor([0.0132, 0.0101, 0.0225, 0.0646, 0.0139, 0.3760, 0.0193, 0.0391], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0147, 0.0188, 0.0168, 0.0167, 0.0199, 0.0177, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:20:39,433 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4093, 2.4604, 2.1242, 2.2673, 2.7883, 2.5108, 2.8503, 3.0397], device='cuda:7'), covar=tensor([0.0153, 0.0481, 0.0562, 0.0486, 0.0350, 0.0431, 0.0295, 0.0295], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0225, 0.0218, 0.0218, 0.0226, 0.0224, 0.0223, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:20:41,986 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.32 vs. limit=5.0 2023-05-01 08:21:18,610 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 08:21:26,011 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8845, 2.7490, 2.7447, 2.0162, 2.5498, 2.8192, 2.7092, 1.9367], device='cuda:7'), covar=tensor([0.0458, 0.0084, 0.0077, 0.0375, 0.0153, 0.0095, 0.0104, 0.0452], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0079, 0.0080, 0.0131, 0.0095, 0.0104, 0.0091, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 08:21:27,496 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5928, 3.9224, 2.8658, 2.2037, 2.3663, 2.3812, 4.1363, 3.2852], device='cuda:7'), covar=tensor([0.3060, 0.0595, 0.1939, 0.2962, 0.2934, 0.2180, 0.0436, 0.1316], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0262, 0.0299, 0.0304, 0.0287, 0.0252, 0.0287, 0.0325], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 08:21:38,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 2.129e+02 2.519e+02 2.912e+02 9.426e+02, threshold=5.038e+02, percent-clipped=1.0 2023-05-01 08:21:54,333 INFO [train.py:904] (7/8) Epoch 21, batch 8900, loss[loss=0.183, simple_loss=0.2778, pruned_loss=0.04409, over 16904.00 frames. ], tot_loss[loss=0.172, simple_loss=0.269, pruned_loss=0.03751, over 3079601.84 frames. ], batch size: 116, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:23:57,823 INFO [train.py:904] (7/8) Epoch 21, batch 8950, loss[loss=0.1562, simple_loss=0.2487, pruned_loss=0.03188, over 15375.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2687, pruned_loss=0.03811, over 3062461.37 frames. ], batch size: 192, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:25:29,297 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.092e+02 2.341e+02 2.927e+02 4.703e+02, threshold=4.682e+02, percent-clipped=0.0 2023-05-01 08:25:46,911 INFO [train.py:904] (7/8) Epoch 21, batch 9000, loss[loss=0.1463, simple_loss=0.2415, pruned_loss=0.02558, over 16740.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2644, pruned_loss=0.03665, over 3060793.02 frames. ], batch size: 83, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:25:46,912 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 08:25:56,445 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9504, 2.3259, 2.0337, 2.1077, 2.6193, 2.3221, 2.5001, 2.7641], device='cuda:7'), covar=tensor([0.0169, 0.0419, 0.0521, 0.0451, 0.0283, 0.0372, 0.0195, 0.0261], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0225, 0.0217, 0.0217, 0.0225, 0.0224, 0.0223, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:25:57,432 INFO [train.py:938] (7/8) Epoch 21, validation: loss=0.1457, simple_loss=0.2498, pruned_loss=0.02077, over 944034.00 frames. 2023-05-01 08:25:57,432 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 08:27:20,397 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212042.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:27:39,214 INFO [train.py:904] (7/8) Epoch 21, batch 9050, loss[loss=0.1789, simple_loss=0.2668, pruned_loss=0.04548, over 16540.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2658, pruned_loss=0.03757, over 3051303.22 frames. ], batch size: 148, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:28:15,049 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212068.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:28:58,994 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212090.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:29:10,428 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.192e+02 2.458e+02 2.901e+02 5.041e+02, threshold=4.916e+02, percent-clipped=1.0 2023-05-01 08:29:13,823 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1629, 4.0550, 4.2250, 4.3668, 4.4695, 4.0435, 4.4601, 4.5052], device='cuda:7'), covar=tensor([0.1692, 0.1127, 0.1408, 0.0689, 0.0545, 0.1254, 0.0629, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0594, 0.0736, 0.0851, 0.0749, 0.0564, 0.0594, 0.0609, 0.0709], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:29:18,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1013, 2.1495, 2.1540, 3.6288, 2.0659, 2.4174, 2.2790, 2.2709], device='cuda:7'), covar=tensor([0.1245, 0.3644, 0.3119, 0.0571, 0.4476, 0.2613, 0.3670, 0.3455], device='cuda:7'), in_proj_covar=tensor([0.0388, 0.0435, 0.0359, 0.0316, 0.0427, 0.0498, 0.0407, 0.0507], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:29:26,249 INFO [train.py:904] (7/8) Epoch 21, batch 9100, loss[loss=0.1556, simple_loss=0.2525, pruned_loss=0.0293, over 17175.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2653, pruned_loss=0.03783, over 3067137.01 frames. ], batch size: 46, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:29:27,957 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-05-01 08:29:30,990 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212104.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 08:29:53,537 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212116.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:30:23,665 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6929, 4.6614, 4.4544, 3.9008, 4.6030, 1.7235, 4.3265, 4.2463], device='cuda:7'), covar=tensor([0.0104, 0.0127, 0.0203, 0.0334, 0.0110, 0.2785, 0.0145, 0.0252], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0147, 0.0189, 0.0168, 0.0167, 0.0200, 0.0177, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:30:52,437 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4363, 1.7691, 2.0790, 2.4823, 2.4566, 2.8067, 2.0079, 2.6602], device='cuda:7'), covar=tensor([0.0278, 0.0549, 0.0389, 0.0331, 0.0376, 0.0192, 0.0497, 0.0176], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0187, 0.0171, 0.0176, 0.0190, 0.0146, 0.0189, 0.0143], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:31:22,748 INFO [train.py:904] (7/8) Epoch 21, batch 9150, loss[loss=0.1654, simple_loss=0.2616, pruned_loss=0.03457, over 16246.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2655, pruned_loss=0.03759, over 3061564.50 frames. ], batch size: 166, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:31:52,687 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212165.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 08:32:16,134 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7612, 4.9926, 5.1808, 4.8977, 5.0134, 5.5155, 5.0297, 4.7369], device='cuda:7'), covar=tensor([0.0954, 0.1783, 0.1844, 0.1893, 0.2147, 0.0786, 0.1504, 0.2323], device='cuda:7'), in_proj_covar=tensor([0.0385, 0.0565, 0.0622, 0.0471, 0.0622, 0.0654, 0.0493, 0.0632], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 08:32:57,280 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.079e+02 2.542e+02 3.144e+02 5.396e+02, threshold=5.084e+02, percent-clipped=1.0 2023-05-01 08:33:09,952 INFO [train.py:904] (7/8) Epoch 21, batch 9200, loss[loss=0.1691, simple_loss=0.256, pruned_loss=0.04113, over 17069.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2609, pruned_loss=0.03639, over 3078842.58 frames. ], batch size: 53, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:33:28,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0158, 1.8378, 1.6531, 1.4933, 1.9849, 1.6547, 1.5671, 1.9250], device='cuda:7'), covar=tensor([0.0217, 0.0305, 0.0428, 0.0391, 0.0248, 0.0305, 0.0157, 0.0226], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0224, 0.0217, 0.0217, 0.0224, 0.0224, 0.0222, 0.0217], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:34:48,828 INFO [train.py:904] (7/8) Epoch 21, batch 9250, loss[loss=0.171, simple_loss=0.2697, pruned_loss=0.03618, over 16780.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2607, pruned_loss=0.0366, over 3053890.05 frames. ], batch size: 124, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:34:51,566 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212253.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:36:25,639 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 2.162e+02 2.598e+02 3.159e+02 6.088e+02, threshold=5.195e+02, percent-clipped=3.0 2023-05-01 08:36:39,750 INFO [train.py:904] (7/8) Epoch 21, batch 9300, loss[loss=0.1495, simple_loss=0.2446, pruned_loss=0.02722, over 16707.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2593, pruned_loss=0.03607, over 3044377.78 frames. ], batch size: 134, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:37:07,701 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212314.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 08:38:22,842 INFO [train.py:904] (7/8) Epoch 21, batch 9350, loss[loss=0.1857, simple_loss=0.2734, pruned_loss=0.04903, over 12257.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2595, pruned_loss=0.03631, over 3040577.62 frames. ], batch size: 246, lr: 3.19e-03, grad_scale: 8.0 2023-05-01 08:38:28,776 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 08:39:47,853 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.212e+02 2.571e+02 3.078e+02 5.703e+02, threshold=5.142e+02, percent-clipped=1.0 2023-05-01 08:40:00,112 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2725, 2.9888, 3.1350, 1.6939, 3.2782, 3.3770, 2.7603, 2.6658], device='cuda:7'), covar=tensor([0.0726, 0.0254, 0.0202, 0.1198, 0.0102, 0.0168, 0.0464, 0.0421], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0102, 0.0091, 0.0133, 0.0075, 0.0117, 0.0123, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 08:40:03,005 INFO [train.py:904] (7/8) Epoch 21, batch 9400, loss[loss=0.1482, simple_loss=0.2363, pruned_loss=0.03005, over 12444.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2589, pruned_loss=0.03597, over 3031415.35 frames. ], batch size: 248, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:41:18,514 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212439.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:41:42,918 INFO [train.py:904] (7/8) Epoch 21, batch 9450, loss[loss=0.1641, simple_loss=0.2621, pruned_loss=0.03308, over 15508.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2608, pruned_loss=0.0361, over 3030753.36 frames. ], batch size: 191, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:41:49,219 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1421, 2.5289, 2.6740, 1.8987, 2.7911, 2.8724, 2.5158, 2.4988], device='cuda:7'), covar=tensor([0.0672, 0.0235, 0.0222, 0.1026, 0.0115, 0.0250, 0.0435, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0103, 0.0091, 0.0133, 0.0075, 0.0117, 0.0123, 0.0123], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 08:41:58,699 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=212460.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 08:43:11,246 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 1.974e+02 2.463e+02 3.006e+02 5.675e+02, threshold=4.927e+02, percent-clipped=1.0 2023-05-01 08:43:21,842 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212500.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:43:25,068 INFO [train.py:904] (7/8) Epoch 21, batch 9500, loss[loss=0.168, simple_loss=0.2577, pruned_loss=0.03911, over 16982.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2596, pruned_loss=0.03541, over 3049424.26 frames. ], batch size: 109, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:45:05,780 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6289, 3.6986, 3.4821, 3.1877, 3.2933, 3.6089, 3.3809, 3.4511], device='cuda:7'), covar=tensor([0.0512, 0.0517, 0.0302, 0.0262, 0.0466, 0.0463, 0.1280, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0273, 0.0395, 0.0321, 0.0316, 0.0327, 0.0367, 0.0222, 0.0382], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 08:45:12,080 INFO [train.py:904] (7/8) Epoch 21, batch 9550, loss[loss=0.1781, simple_loss=0.2841, pruned_loss=0.03606, over 16127.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2601, pruned_loss=0.03584, over 3050493.98 frames. ], batch size: 165, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:45:43,546 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2249, 2.1590, 2.1748, 3.9299, 2.1974, 2.4625, 2.2880, 2.3041], device='cuda:7'), covar=tensor([0.1300, 0.4031, 0.3240, 0.0526, 0.4464, 0.2879, 0.3864, 0.3947], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0435, 0.0359, 0.0316, 0.0428, 0.0497, 0.0407, 0.0507], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 08:46:39,907 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.109e+02 2.597e+02 3.170e+02 5.717e+02, threshold=5.194e+02, percent-clipped=4.0 2023-05-01 08:46:50,972 INFO [train.py:904] (7/8) Epoch 21, batch 9600, loss[loss=0.1782, simple_loss=0.2758, pruned_loss=0.04031, over 16709.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2613, pruned_loss=0.03636, over 3050802.25 frames. ], batch size: 83, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:47:06,249 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=212609.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 08:48:04,642 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-01 08:48:38,613 INFO [train.py:904] (7/8) Epoch 21, batch 9650, loss[loss=0.1749, simple_loss=0.2678, pruned_loss=0.04102, over 16870.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.263, pruned_loss=0.03688, over 3056850.15 frames. ], batch size: 116, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:49:53,876 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-01 08:50:11,919 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.547e+02 2.239e+02 2.609e+02 3.201e+02 8.250e+02, threshold=5.217e+02, percent-clipped=3.0 2023-05-01 08:50:21,203 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212698.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:50:28,039 INFO [train.py:904] (7/8) Epoch 21, batch 9700, loss[loss=0.1703, simple_loss=0.2654, pruned_loss=0.03764, over 16156.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2625, pruned_loss=0.03682, over 3053329.13 frames. ], batch size: 165, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:50:46,091 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212711.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:52:10,277 INFO [train.py:904] (7/8) Epoch 21, batch 9750, loss[loss=0.1582, simple_loss=0.2586, pruned_loss=0.02888, over 15331.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2617, pruned_loss=0.03676, over 3068045.65 frames. ], batch size: 191, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:52:25,427 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212759.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:52:27,274 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212760.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 08:52:49,209 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212772.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:52:53,749 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=212774.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:52:59,372 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 08:53:36,937 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.010e+02 2.340e+02 2.880e+02 4.703e+02, threshold=4.680e+02, percent-clipped=0.0 2023-05-01 08:53:37,562 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=212795.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:53:48,984 INFO [train.py:904] (7/8) Epoch 21, batch 9800, loss[loss=0.1723, simple_loss=0.2699, pruned_loss=0.03735, over 17181.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2625, pruned_loss=0.03622, over 3086031.70 frames. ], batch size: 46, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:54:01,428 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212808.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 08:54:51,643 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=212835.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:55:31,550 INFO [train.py:904] (7/8) Epoch 21, batch 9850, loss[loss=0.1817, simple_loss=0.2741, pruned_loss=0.04461, over 16615.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2637, pruned_loss=0.03616, over 3084634.69 frames. ], batch size: 134, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:57:07,914 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.176e+02 2.812e+02 3.349e+02 5.684e+02, threshold=5.624e+02, percent-clipped=7.0 2023-05-01 08:57:21,379 INFO [train.py:904] (7/8) Epoch 21, batch 9900, loss[loss=0.1524, simple_loss=0.2435, pruned_loss=0.0307, over 12259.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2641, pruned_loss=0.03624, over 3083994.95 frames. ], batch size: 247, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:57:38,633 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=212909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 08:59:17,581 INFO [train.py:904] (7/8) Epoch 21, batch 9950, loss[loss=0.1873, simple_loss=0.2906, pruned_loss=0.04202, over 15430.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2664, pruned_loss=0.0364, over 3089857.44 frames. ], batch size: 191, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 08:59:29,578 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=212957.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:00:04,105 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 09:00:53,115 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6154, 2.0878, 1.7096, 1.8364, 2.3924, 2.0853, 2.0293, 2.4867], device='cuda:7'), covar=tensor([0.0164, 0.0411, 0.0573, 0.0513, 0.0295, 0.0394, 0.0230, 0.0272], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0226, 0.0219, 0.0219, 0.0226, 0.0225, 0.0222, 0.0219], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:00:58,653 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 2.203e+02 2.698e+02 3.794e+02 1.048e+03, threshold=5.396e+02, percent-clipped=2.0 2023-05-01 09:01:17,082 INFO [train.py:904] (7/8) Epoch 21, batch 10000, loss[loss=0.146, simple_loss=0.2426, pruned_loss=0.02475, over 17226.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2643, pruned_loss=0.0357, over 3116033.98 frames. ], batch size: 45, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:01:33,546 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 09:02:00,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2536, 3.5443, 3.6636, 1.8118, 3.8327, 4.0534, 3.0742, 2.9356], device='cuda:7'), covar=tensor([0.1045, 0.0165, 0.0160, 0.1257, 0.0072, 0.0130, 0.0354, 0.0520], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0102, 0.0089, 0.0132, 0.0075, 0.0116, 0.0122, 0.0122], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 09:02:58,016 INFO [train.py:904] (7/8) Epoch 21, batch 10050, loss[loss=0.159, simple_loss=0.259, pruned_loss=0.02953, over 16654.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2641, pruned_loss=0.03567, over 3113943.63 frames. ], batch size: 134, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:03:03,008 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213054.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:03:28,753 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213067.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:04:21,049 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.063e+02 2.495e+02 3.076e+02 5.792e+02, threshold=4.990e+02, percent-clipped=2.0 2023-05-01 09:04:21,624 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213095.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:04:32,168 INFO [train.py:904] (7/8) Epoch 21, batch 10100, loss[loss=0.1707, simple_loss=0.2524, pruned_loss=0.04454, over 12881.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2645, pruned_loss=0.03602, over 3104858.92 frames. ], batch size: 248, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:05:27,346 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213130.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:05:42,672 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213143.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:05:51,231 INFO [train.py:904] (7/8) Epoch 21, batch 10150, loss[loss=0.1528, simple_loss=0.2393, pruned_loss=0.03312, over 12519.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2623, pruned_loss=0.0358, over 3062111.04 frames. ], batch size: 248, lr: 3.18e-03, grad_scale: 8.0 2023-05-01 09:06:16,187 INFO [train.py:904] (7/8) Epoch 22, batch 0, loss[loss=0.1828, simple_loss=0.2733, pruned_loss=0.0462, over 16871.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2733, pruned_loss=0.0462, over 16871.00 frames. ], batch size: 42, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:06:16,187 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 09:06:23,637 INFO [train.py:938] (7/8) Epoch 22, validation: loss=0.1457, simple_loss=0.2489, pruned_loss=0.0212, over 944034.00 frames. 2023-05-01 09:06:23,637 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 09:07:26,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 2.320e+02 2.870e+02 3.515e+02 6.456e+02, threshold=5.740e+02, percent-clipped=5.0 2023-05-01 09:07:34,022 INFO [train.py:904] (7/8) Epoch 22, batch 50, loss[loss=0.2018, simple_loss=0.2749, pruned_loss=0.06437, over 16768.00 frames. ], tot_loss[loss=0.1831, simple_loss=0.2688, pruned_loss=0.04875, over 750361.51 frames. ], batch size: 83, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:08:41,862 INFO [train.py:904] (7/8) Epoch 22, batch 100, loss[loss=0.1944, simple_loss=0.2646, pruned_loss=0.06206, over 16852.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2658, pruned_loss=0.04639, over 1324266.72 frames. ], batch size: 96, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:09:44,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4213, 2.3147, 2.2129, 4.2280, 2.3149, 2.7421, 2.3807, 2.4954], device='cuda:7'), covar=tensor([0.1210, 0.3616, 0.3005, 0.0493, 0.4009, 0.2574, 0.3401, 0.3364], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0438, 0.0361, 0.0320, 0.0431, 0.0499, 0.0409, 0.0511], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:09:44,707 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.194e+02 2.643e+02 3.119e+02 6.629e+02, threshold=5.286e+02, percent-clipped=1.0 2023-05-01 09:09:51,962 INFO [train.py:904] (7/8) Epoch 22, batch 150, loss[loss=0.1821, simple_loss=0.2639, pruned_loss=0.05012, over 16862.00 frames. ], tot_loss[loss=0.1779, simple_loss=0.2644, pruned_loss=0.04573, over 1768772.07 frames. ], batch size: 116, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:11:00,689 INFO [train.py:904] (7/8) Epoch 22, batch 200, loss[loss=0.1867, simple_loss=0.2691, pruned_loss=0.05218, over 16898.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2648, pruned_loss=0.04651, over 2113975.23 frames. ], batch size: 90, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:11:02,275 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213354.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:11:19,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8939, 4.4612, 4.3446, 3.1097, 3.8966, 4.4198, 4.1028, 2.3399], device='cuda:7'), covar=tensor([0.0599, 0.0122, 0.0082, 0.0468, 0.0155, 0.0134, 0.0103, 0.0685], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0082, 0.0082, 0.0133, 0.0097, 0.0106, 0.0093, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:11:20,338 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213367.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:11:33,943 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8703, 4.6542, 4.7960, 5.1214, 5.2886, 4.7087, 5.3466, 5.2901], device='cuda:7'), covar=tensor([0.2158, 0.1565, 0.2450, 0.1040, 0.0917, 0.1043, 0.0742, 0.0922], device='cuda:7'), in_proj_covar=tensor([0.0611, 0.0752, 0.0874, 0.0764, 0.0580, 0.0606, 0.0630, 0.0727], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:11:55,610 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3362, 3.4199, 3.8560, 2.1280, 2.9978, 2.3694, 3.7586, 3.5421], device='cuda:7'), covar=tensor([0.0296, 0.1095, 0.0581, 0.2211, 0.0963, 0.1136, 0.0748, 0.1361], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0158, 0.0163, 0.0150, 0.0142, 0.0128, 0.0140, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 09:12:00,920 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.291e+02 2.770e+02 3.341e+02 7.349e+02, threshold=5.539e+02, percent-clipped=1.0 2023-05-01 09:12:07,194 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213402.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:12:08,118 INFO [train.py:904] (7/8) Epoch 22, batch 250, loss[loss=0.1754, simple_loss=0.2519, pruned_loss=0.04948, over 16908.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2625, pruned_loss=0.0456, over 2385893.45 frames. ], batch size: 116, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:12:25,587 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213415.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:12:46,330 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=213430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:12:51,627 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7749, 3.9034, 2.5913, 4.4722, 3.1080, 4.4263, 2.6587, 3.1586], device='cuda:7'), covar=tensor([0.0307, 0.0406, 0.1531, 0.0394, 0.0783, 0.0534, 0.1513, 0.0789], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0175, 0.0194, 0.0159, 0.0177, 0.0214, 0.0202, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 09:12:53,083 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2023-05-01 09:13:17,500 INFO [train.py:904] (7/8) Epoch 22, batch 300, loss[loss=0.1407, simple_loss=0.2367, pruned_loss=0.02232, over 17134.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2604, pruned_loss=0.04425, over 2601896.29 frames. ], batch size: 47, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:13:52,066 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=213478.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:13:55,804 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8281, 1.9097, 2.3064, 2.6749, 2.6782, 2.7488, 1.9243, 2.9200], device='cuda:7'), covar=tensor([0.0209, 0.0504, 0.0385, 0.0292, 0.0350, 0.0283, 0.0566, 0.0184], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0190, 0.0176, 0.0180, 0.0194, 0.0150, 0.0193, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:14:19,529 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.218e+02 2.593e+02 2.976e+02 4.668e+02, threshold=5.187e+02, percent-clipped=0.0 2023-05-01 09:14:25,248 INFO [train.py:904] (7/8) Epoch 22, batch 350, loss[loss=0.153, simple_loss=0.2466, pruned_loss=0.02974, over 17118.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2582, pruned_loss=0.04357, over 2765558.23 frames. ], batch size: 48, lr: 3.10e-03, grad_scale: 1.0 2023-05-01 09:15:34,185 INFO [train.py:904] (7/8) Epoch 22, batch 400, loss[loss=0.151, simple_loss=0.2393, pruned_loss=0.03134, over 16860.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2572, pruned_loss=0.04345, over 2879195.68 frames. ], batch size: 42, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:15:52,087 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=213566.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:16:19,686 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6130, 4.8006, 4.9422, 4.7750, 4.8237, 5.4104, 4.8781, 4.5544], device='cuda:7'), covar=tensor([0.1676, 0.2080, 0.2503, 0.2241, 0.2707, 0.1091, 0.1984, 0.2791], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0590, 0.0653, 0.0492, 0.0651, 0.0687, 0.0514, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:16:31,609 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 09:16:36,629 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 2.265e+02 2.651e+02 3.318e+02 1.993e+03, threshold=5.302e+02, percent-clipped=3.0 2023-05-01 09:16:43,120 INFO [train.py:904] (7/8) Epoch 22, batch 450, loss[loss=0.1503, simple_loss=0.2448, pruned_loss=0.02797, over 17124.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2555, pruned_loss=0.04253, over 2979548.36 frames. ], batch size: 49, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:17:01,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4252, 4.2821, 4.4906, 4.6502, 4.7564, 4.2938, 4.5946, 4.7494], device='cuda:7'), covar=tensor([0.1823, 0.1425, 0.1444, 0.0767, 0.0586, 0.1200, 0.2747, 0.0964], device='cuda:7'), in_proj_covar=tensor([0.0623, 0.0767, 0.0893, 0.0780, 0.0590, 0.0621, 0.0643, 0.0741], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:17:05,350 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-01 09:17:16,902 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=213627.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:17:53,052 INFO [train.py:904] (7/8) Epoch 22, batch 500, loss[loss=0.1681, simple_loss=0.2558, pruned_loss=0.04016, over 12202.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2537, pruned_loss=0.04157, over 3052425.02 frames. ], batch size: 247, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:17:59,986 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 09:18:24,179 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8148, 2.7041, 2.6148, 1.9121, 2.5785, 2.6951, 2.6170, 1.8692], device='cuda:7'), covar=tensor([0.0514, 0.0108, 0.0100, 0.0431, 0.0159, 0.0152, 0.0134, 0.0486], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0083, 0.0083, 0.0135, 0.0098, 0.0107, 0.0094, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:18:54,888 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.187e+02 2.448e+02 2.877e+02 6.308e+02, threshold=4.896e+02, percent-clipped=1.0 2023-05-01 09:19:01,577 INFO [train.py:904] (7/8) Epoch 22, batch 550, loss[loss=0.1729, simple_loss=0.2519, pruned_loss=0.04697, over 16823.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2528, pruned_loss=0.04091, over 3110524.09 frames. ], batch size: 90, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:19:19,186 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 09:19:38,502 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1200, 5.6477, 5.7895, 5.5667, 5.6156, 6.1937, 5.5442, 5.2993], device='cuda:7'), covar=tensor([0.0905, 0.1914, 0.2725, 0.1783, 0.2620, 0.0958, 0.1751, 0.2140], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0594, 0.0658, 0.0495, 0.0656, 0.0690, 0.0517, 0.0659], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:20:10,216 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 09:20:10,676 INFO [train.py:904] (7/8) Epoch 22, batch 600, loss[loss=0.1495, simple_loss=0.2336, pruned_loss=0.03276, over 16790.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2531, pruned_loss=0.04141, over 3155715.67 frames. ], batch size: 39, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:21:10,509 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8761, 5.3025, 5.4408, 5.1896, 5.2937, 5.8844, 5.3236, 5.0795], device='cuda:7'), covar=tensor([0.1306, 0.2077, 0.2658, 0.2210, 0.2911, 0.1117, 0.1742, 0.2260], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0596, 0.0659, 0.0496, 0.0657, 0.0692, 0.0517, 0.0660], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:21:13,519 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.232e+02 2.565e+02 3.000e+02 5.415e+02, threshold=5.129e+02, percent-clipped=2.0 2023-05-01 09:21:21,592 INFO [train.py:904] (7/8) Epoch 22, batch 650, loss[loss=0.1585, simple_loss=0.2471, pruned_loss=0.03492, over 17234.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2514, pruned_loss=0.04095, over 3197959.76 frames. ], batch size: 45, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:22:30,265 INFO [train.py:904] (7/8) Epoch 22, batch 700, loss[loss=0.1584, simple_loss=0.2506, pruned_loss=0.03305, over 17121.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2507, pruned_loss=0.0412, over 3217535.31 frames. ], batch size: 48, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:23:35,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.030e+02 2.447e+02 3.057e+02 6.587e+02, threshold=4.894e+02, percent-clipped=3.0 2023-05-01 09:23:37,038 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4609, 5.8995, 5.3584, 5.7624, 5.2928, 5.1801, 5.4400, 5.9809], device='cuda:7'), covar=tensor([0.2434, 0.1403, 0.2784, 0.1484, 0.1750, 0.1490, 0.2316, 0.1995], device='cuda:7'), in_proj_covar=tensor([0.0677, 0.0825, 0.0680, 0.0628, 0.0527, 0.0534, 0.0692, 0.0647], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:23:41,791 INFO [train.py:904] (7/8) Epoch 22, batch 750, loss[loss=0.1751, simple_loss=0.2709, pruned_loss=0.0396, over 16664.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.251, pruned_loss=0.04097, over 3243006.96 frames. ], batch size: 62, lr: 3.10e-03, grad_scale: 2.0 2023-05-01 09:24:08,350 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=213922.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:24:32,792 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4961, 4.4506, 4.4376, 3.4459, 4.4451, 1.6004, 4.1064, 3.9871], device='cuda:7'), covar=tensor([0.0243, 0.0188, 0.0265, 0.0672, 0.0185, 0.3753, 0.0258, 0.0370], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0154, 0.0196, 0.0173, 0.0174, 0.0207, 0.0184, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:24:53,418 INFO [train.py:904] (7/8) Epoch 22, batch 800, loss[loss=0.168, simple_loss=0.2477, pruned_loss=0.04416, over 16831.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2501, pruned_loss=0.04035, over 3251725.06 frames. ], batch size: 83, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:25:56,896 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.094e+02 2.441e+02 2.760e+02 5.264e+02, threshold=4.882e+02, percent-clipped=1.0 2023-05-01 09:26:06,628 INFO [train.py:904] (7/8) Epoch 22, batch 850, loss[loss=0.1703, simple_loss=0.2494, pruned_loss=0.0456, over 16704.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2498, pruned_loss=0.04026, over 3271168.18 frames. ], batch size: 124, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:26:29,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6610, 3.8125, 2.5062, 4.2937, 2.9987, 4.2719, 2.6289, 3.1711], device='cuda:7'), covar=tensor([0.0336, 0.0410, 0.1606, 0.0325, 0.0828, 0.0590, 0.1443, 0.0743], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0178, 0.0197, 0.0164, 0.0179, 0.0219, 0.0204, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 09:27:17,248 INFO [train.py:904] (7/8) Epoch 22, batch 900, loss[loss=0.1949, simple_loss=0.2645, pruned_loss=0.0627, over 15705.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2497, pruned_loss=0.03983, over 3277011.85 frames. ], batch size: 191, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:27:55,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-01 09:28:19,801 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 1.995e+02 2.382e+02 2.694e+02 6.707e+02, threshold=4.763e+02, percent-clipped=1.0 2023-05-01 09:28:27,533 INFO [train.py:904] (7/8) Epoch 22, batch 950, loss[loss=0.1571, simple_loss=0.2503, pruned_loss=0.03193, over 17039.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2497, pruned_loss=0.03972, over 3290082.84 frames. ], batch size: 50, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:29:03,372 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=214129.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:29:35,597 INFO [train.py:904] (7/8) Epoch 22, batch 1000, loss[loss=0.1496, simple_loss=0.2469, pruned_loss=0.02612, over 17128.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2481, pruned_loss=0.03894, over 3296638.41 frames. ], batch size: 49, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:30:29,137 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=214190.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:30:39,711 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.141e+02 2.430e+02 2.890e+02 5.407e+02, threshold=4.860e+02, percent-clipped=2.0 2023-05-01 09:30:46,656 INFO [train.py:904] (7/8) Epoch 22, batch 1050, loss[loss=0.1844, simple_loss=0.2553, pruned_loss=0.05678, over 16821.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2482, pruned_loss=0.0389, over 3309249.23 frames. ], batch size: 116, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:31:08,609 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=214218.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:31:12,711 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0482, 4.4501, 4.4790, 3.1945, 3.7049, 4.4187, 3.9520, 2.7996], device='cuda:7'), covar=tensor([0.0449, 0.0062, 0.0046, 0.0367, 0.0148, 0.0083, 0.0093, 0.0418], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0083, 0.0084, 0.0134, 0.0098, 0.0108, 0.0094, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:31:13,735 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=214222.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:31:50,041 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2580, 5.2543, 5.1422, 4.6209, 4.7868, 5.1312, 5.1059, 4.7214], device='cuda:7'), covar=tensor([0.0639, 0.0521, 0.0309, 0.0349, 0.1134, 0.0513, 0.0344, 0.0867], device='cuda:7'), in_proj_covar=tensor([0.0297, 0.0432, 0.0347, 0.0345, 0.0356, 0.0402, 0.0238, 0.0417], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:31:56,105 INFO [train.py:904] (7/8) Epoch 22, batch 1100, loss[loss=0.1529, simple_loss=0.2298, pruned_loss=0.03804, over 16309.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2476, pruned_loss=0.03885, over 3315056.70 frames. ], batch size: 165, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:32:04,012 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5084, 2.3947, 2.5008, 4.4356, 2.4393, 2.8117, 2.5289, 2.5688], device='cuda:7'), covar=tensor([0.1271, 0.3735, 0.2977, 0.0461, 0.4065, 0.2611, 0.3414, 0.3805], device='cuda:7'), in_proj_covar=tensor([0.0403, 0.0449, 0.0370, 0.0330, 0.0439, 0.0514, 0.0420, 0.0527], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:32:19,740 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=214270.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:32:31,802 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=214279.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:32:57,758 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.112e+02 2.466e+02 2.921e+02 8.908e+02, threshold=4.932e+02, percent-clipped=7.0 2023-05-01 09:33:03,842 INFO [train.py:904] (7/8) Epoch 22, batch 1150, loss[loss=0.1616, simple_loss=0.2417, pruned_loss=0.04072, over 16840.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2469, pruned_loss=0.03826, over 3312546.46 frames. ], batch size: 39, lr: 3.10e-03, grad_scale: 4.0 2023-05-01 09:34:15,652 INFO [train.py:904] (7/8) Epoch 22, batch 1200, loss[loss=0.1756, simple_loss=0.2659, pruned_loss=0.04264, over 17049.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2464, pruned_loss=0.03782, over 3316013.22 frames. ], batch size: 53, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:35:18,113 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.085e+02 2.414e+02 2.922e+02 4.636e+02, threshold=4.828e+02, percent-clipped=0.0 2023-05-01 09:35:25,091 INFO [train.py:904] (7/8) Epoch 22, batch 1250, loss[loss=0.1773, simple_loss=0.2475, pruned_loss=0.05355, over 16512.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2469, pruned_loss=0.03827, over 3317458.07 frames. ], batch size: 75, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:36:35,056 INFO [train.py:904] (7/8) Epoch 22, batch 1300, loss[loss=0.1541, simple_loss=0.2367, pruned_loss=0.03573, over 16760.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2472, pruned_loss=0.03851, over 3325820.78 frames. ], batch size: 102, lr: 3.10e-03, grad_scale: 8.0 2023-05-01 09:37:18,499 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=214485.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:37:34,615 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.092e+02 2.483e+02 2.893e+02 4.900e+02, threshold=4.965e+02, percent-clipped=1.0 2023-05-01 09:37:42,346 INFO [train.py:904] (7/8) Epoch 22, batch 1350, loss[loss=0.1632, simple_loss=0.2443, pruned_loss=0.04102, over 16730.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2483, pruned_loss=0.03887, over 3332315.13 frames. ], batch size: 124, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:37:55,088 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6533, 3.6634, 2.7863, 2.2202, 2.3740, 2.3587, 3.7460, 3.1715], device='cuda:7'), covar=tensor([0.2711, 0.0594, 0.1794, 0.3135, 0.2941, 0.2312, 0.0526, 0.1598], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0271, 0.0307, 0.0313, 0.0297, 0.0261, 0.0295, 0.0338], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:38:48,217 INFO [train.py:904] (7/8) Epoch 22, batch 1400, loss[loss=0.1595, simple_loss=0.2361, pruned_loss=0.0415, over 16891.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2479, pruned_loss=0.03869, over 3335557.65 frames. ], batch size: 116, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:39:17,613 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=214574.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:39:50,087 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.086e+02 2.334e+02 2.820e+02 4.887e+02, threshold=4.669e+02, percent-clipped=0.0 2023-05-01 09:39:57,066 INFO [train.py:904] (7/8) Epoch 22, batch 1450, loss[loss=0.2039, simple_loss=0.2721, pruned_loss=0.06787, over 16527.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2481, pruned_loss=0.03887, over 3330895.38 frames. ], batch size: 75, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:41:07,199 INFO [train.py:904] (7/8) Epoch 22, batch 1500, loss[loss=0.1768, simple_loss=0.2574, pruned_loss=0.0481, over 16840.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2481, pruned_loss=0.03873, over 3326622.31 frames. ], batch size: 96, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:42:07,566 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-05-01 09:42:10,193 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.149e+02 2.472e+02 3.063e+02 5.864e+02, threshold=4.945e+02, percent-clipped=3.0 2023-05-01 09:42:16,358 INFO [train.py:904] (7/8) Epoch 22, batch 1550, loss[loss=0.1581, simple_loss=0.2531, pruned_loss=0.03154, over 17203.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2491, pruned_loss=0.03973, over 3333710.84 frames. ], batch size: 45, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:43:04,992 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7444, 2.6875, 2.5073, 2.6474, 3.0579, 2.7757, 3.3329, 3.2504], device='cuda:7'), covar=tensor([0.0140, 0.0407, 0.0449, 0.0414, 0.0250, 0.0389, 0.0242, 0.0258], device='cuda:7'), in_proj_covar=tensor([0.0216, 0.0240, 0.0230, 0.0230, 0.0240, 0.0239, 0.0242, 0.0236], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:43:27,998 INFO [train.py:904] (7/8) Epoch 22, batch 1600, loss[loss=0.1801, simple_loss=0.2707, pruned_loss=0.04473, over 17015.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2508, pruned_loss=0.0403, over 3330049.92 frames. ], batch size: 55, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:44:12,470 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=214785.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:44:32,065 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.192e+02 2.616e+02 3.002e+02 5.148e+02, threshold=5.232e+02, percent-clipped=2.0 2023-05-01 09:44:32,928 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-01 09:44:37,875 INFO [train.py:904] (7/8) Epoch 22, batch 1650, loss[loss=0.1481, simple_loss=0.2287, pruned_loss=0.03375, over 16754.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2523, pruned_loss=0.04066, over 3327783.03 frames. ], batch size: 83, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:44:46,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8862, 4.6970, 4.9290, 5.1074, 5.3304, 4.7067, 5.3120, 5.2973], device='cuda:7'), covar=tensor([0.2014, 0.1477, 0.1957, 0.0896, 0.0621, 0.0974, 0.0553, 0.0660], device='cuda:7'), in_proj_covar=tensor([0.0651, 0.0800, 0.0935, 0.0814, 0.0614, 0.0642, 0.0665, 0.0774], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:45:20,796 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=214833.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:45:47,652 INFO [train.py:904] (7/8) Epoch 22, batch 1700, loss[loss=0.2032, simple_loss=0.2796, pruned_loss=0.06335, over 16720.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2536, pruned_loss=0.04132, over 3322146.79 frames. ], batch size: 134, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:46:07,525 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1980, 4.0420, 4.2251, 4.3604, 4.4856, 4.0262, 4.3016, 4.4754], device='cuda:7'), covar=tensor([0.1484, 0.1218, 0.1273, 0.0698, 0.0543, 0.1342, 0.2060, 0.0750], device='cuda:7'), in_proj_covar=tensor([0.0652, 0.0802, 0.0937, 0.0815, 0.0614, 0.0644, 0.0666, 0.0775], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:46:18,582 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=214874.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:46:23,890 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0934, 2.2350, 2.6900, 2.9829, 2.9232, 3.5420, 2.5260, 3.5375], device='cuda:7'), covar=tensor([0.0256, 0.0525, 0.0349, 0.0362, 0.0333, 0.0196, 0.0454, 0.0181], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0195, 0.0181, 0.0185, 0.0199, 0.0155, 0.0197, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:46:45,063 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-05-01 09:46:53,105 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.472e+02 2.965e+02 3.692e+02 1.718e+03, threshold=5.930e+02, percent-clipped=4.0 2023-05-01 09:46:58,560 INFO [train.py:904] (7/8) Epoch 22, batch 1750, loss[loss=0.1651, simple_loss=0.2608, pruned_loss=0.0347, over 17074.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2535, pruned_loss=0.0407, over 3329750.59 frames. ], batch size: 55, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:47:25,707 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=214922.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:48:06,751 INFO [train.py:904] (7/8) Epoch 22, batch 1800, loss[loss=0.1944, simple_loss=0.2773, pruned_loss=0.05579, over 12093.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2538, pruned_loss=0.0404, over 3330734.90 frames. ], batch size: 247, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:48:38,746 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2023-05-01 09:48:45,990 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8315, 5.1864, 4.9620, 4.9678, 4.7164, 4.7034, 4.6504, 5.3088], device='cuda:7'), covar=tensor([0.1306, 0.0899, 0.0993, 0.0866, 0.0844, 0.1063, 0.1206, 0.0980], device='cuda:7'), in_proj_covar=tensor([0.0684, 0.0839, 0.0692, 0.0638, 0.0534, 0.0540, 0.0704, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:49:04,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9471, 4.0202, 4.2927, 4.2888, 4.3115, 4.0153, 4.0732, 4.0149], device='cuda:7'), covar=tensor([0.0385, 0.0604, 0.0433, 0.0402, 0.0515, 0.0477, 0.0762, 0.0555], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0459, 0.0447, 0.0415, 0.0493, 0.0470, 0.0552, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 09:49:13,319 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.061e+02 2.528e+02 2.863e+02 4.992e+02, threshold=5.055e+02, percent-clipped=0.0 2023-05-01 09:49:18,206 INFO [train.py:904] (7/8) Epoch 22, batch 1850, loss[loss=0.1616, simple_loss=0.2444, pruned_loss=0.03942, over 16833.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2549, pruned_loss=0.04106, over 3326084.14 frames. ], batch size: 102, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:49:46,632 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5599, 4.9494, 4.4354, 4.7621, 4.5164, 4.4930, 4.5306, 5.0494], device='cuda:7'), covar=tensor([0.2491, 0.1720, 0.2701, 0.1770, 0.1838, 0.1898, 0.2533, 0.2033], device='cuda:7'), in_proj_covar=tensor([0.0684, 0.0840, 0.0693, 0.0638, 0.0534, 0.0540, 0.0705, 0.0658], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 09:50:27,235 INFO [train.py:904] (7/8) Epoch 22, batch 1900, loss[loss=0.1591, simple_loss=0.2556, pruned_loss=0.03132, over 17125.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2542, pruned_loss=0.04051, over 3321883.01 frames. ], batch size: 48, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:50:39,552 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 09:51:31,794 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.086e+02 2.418e+02 2.993e+02 8.338e+02, threshold=4.836e+02, percent-clipped=2.0 2023-05-01 09:51:36,135 INFO [train.py:904] (7/8) Epoch 22, batch 1950, loss[loss=0.153, simple_loss=0.2481, pruned_loss=0.02896, over 17125.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2542, pruned_loss=0.04012, over 3321298.66 frames. ], batch size: 47, lr: 3.09e-03, grad_scale: 4.0 2023-05-01 09:51:41,242 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3163, 3.0319, 3.3427, 1.8028, 3.4410, 3.5080, 2.7949, 2.5319], device='cuda:7'), covar=tensor([0.0848, 0.0275, 0.0207, 0.1197, 0.0119, 0.0214, 0.0435, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0099, 0.0140, 0.0081, 0.0127, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 09:52:22,899 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.60 vs. limit=5.0 2023-05-01 09:52:40,919 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8615, 5.1610, 5.2993, 5.0927, 5.0946, 5.7105, 5.2015, 4.9171], device='cuda:7'), covar=tensor([0.1317, 0.1938, 0.2333, 0.1866, 0.2653, 0.1022, 0.1575, 0.2322], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0614, 0.0673, 0.0509, 0.0677, 0.0707, 0.0531, 0.0679], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:52:44,790 INFO [train.py:904] (7/8) Epoch 22, batch 2000, loss[loss=0.1678, simple_loss=0.2534, pruned_loss=0.04114, over 16537.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2547, pruned_loss=0.04049, over 3325691.70 frames. ], batch size: 68, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:53:48,738 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.173e+02 2.667e+02 3.241e+02 6.066e+02, threshold=5.334e+02, percent-clipped=3.0 2023-05-01 09:53:53,932 INFO [train.py:904] (7/8) Epoch 22, batch 2050, loss[loss=0.1728, simple_loss=0.2591, pruned_loss=0.0432, over 16399.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2547, pruned_loss=0.04093, over 3320639.52 frames. ], batch size: 68, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:54:31,429 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215230.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 09:54:34,054 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 09:55:01,198 INFO [train.py:904] (7/8) Epoch 22, batch 2100, loss[loss=0.1505, simple_loss=0.2506, pruned_loss=0.02516, over 17066.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2556, pruned_loss=0.04156, over 3308231.16 frames. ], batch size: 50, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:55:37,030 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0519, 3.1768, 3.3871, 2.1464, 2.9489, 2.3469, 3.5485, 3.5023], device='cuda:7'), covar=tensor([0.0265, 0.0888, 0.0640, 0.1962, 0.0828, 0.1028, 0.0533, 0.0954], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0164, 0.0167, 0.0155, 0.0145, 0.0131, 0.0144, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 09:55:54,444 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215291.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 09:56:04,633 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.033e+02 2.346e+02 2.893e+02 7.245e+02, threshold=4.692e+02, percent-clipped=2.0 2023-05-01 09:56:09,033 INFO [train.py:904] (7/8) Epoch 22, batch 2150, loss[loss=0.1512, simple_loss=0.2438, pruned_loss=0.02925, over 17168.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2565, pruned_loss=0.04148, over 3317410.64 frames. ], batch size: 46, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:56:16,256 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2392, 5.2353, 5.1201, 4.6196, 4.7542, 5.1984, 5.1123, 4.7563], device='cuda:7'), covar=tensor([0.0620, 0.0607, 0.0332, 0.0404, 0.1154, 0.0518, 0.0372, 0.0896], device='cuda:7'), in_proj_covar=tensor([0.0307, 0.0444, 0.0357, 0.0356, 0.0366, 0.0411, 0.0245, 0.0428], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 09:56:40,338 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215325.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 09:56:48,990 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 09:56:51,024 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215333.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 09:57:18,612 INFO [train.py:904] (7/8) Epoch 22, batch 2200, loss[loss=0.1449, simple_loss=0.2292, pruned_loss=0.0303, over 16996.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.257, pruned_loss=0.04159, over 3322070.98 frames. ], batch size: 41, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:58:02,279 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215386.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 09:58:14,384 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215394.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 09:58:20,774 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.171e+02 2.606e+02 3.111e+02 9.974e+02, threshold=5.211e+02, percent-clipped=4.0 2023-05-01 09:58:24,664 INFO [train.py:904] (7/8) Epoch 22, batch 2250, loss[loss=0.1536, simple_loss=0.2396, pruned_loss=0.0338, over 15836.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2577, pruned_loss=0.0422, over 3307914.65 frames. ], batch size: 35, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:58:34,406 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 09:59:36,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7798, 2.9244, 3.2367, 2.0161, 2.7298, 2.1917, 3.3368, 3.2515], device='cuda:7'), covar=tensor([0.0263, 0.0958, 0.0578, 0.1984, 0.0905, 0.1011, 0.0561, 0.0907], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0164, 0.0167, 0.0154, 0.0145, 0.0130, 0.0144, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 09:59:36,622 INFO [train.py:904] (7/8) Epoch 22, batch 2300, loss[loss=0.1745, simple_loss=0.256, pruned_loss=0.04645, over 16438.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.257, pruned_loss=0.04178, over 3309446.75 frames. ], batch size: 146, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 09:59:43,420 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9935, 4.7969, 5.0179, 5.2355, 5.4651, 4.8306, 5.4326, 5.4470], device='cuda:7'), covar=tensor([0.2009, 0.1404, 0.1982, 0.0798, 0.0583, 0.0858, 0.0530, 0.0635], device='cuda:7'), in_proj_covar=tensor([0.0659, 0.0817, 0.0954, 0.0825, 0.0623, 0.0656, 0.0677, 0.0784], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:00:27,248 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1910, 2.2339, 2.2611, 3.8406, 2.2217, 2.5520, 2.2708, 2.3741], device='cuda:7'), covar=tensor([0.1506, 0.3454, 0.3012, 0.0661, 0.3912, 0.2642, 0.3870, 0.3142], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0452, 0.0372, 0.0331, 0.0439, 0.0519, 0.0423, 0.0530], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:00:42,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.161e+02 2.569e+02 3.134e+02 6.590e+02, threshold=5.137e+02, percent-clipped=2.0 2023-05-01 10:00:46,498 INFO [train.py:904] (7/8) Epoch 22, batch 2350, loss[loss=0.1517, simple_loss=0.2399, pruned_loss=0.03175, over 15983.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2576, pruned_loss=0.04218, over 3303861.33 frames. ], batch size: 35, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:00:46,953 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215503.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:01:48,961 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215548.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:01:55,009 INFO [train.py:904] (7/8) Epoch 22, batch 2400, loss[loss=0.1426, simple_loss=0.2379, pruned_loss=0.0236, over 16837.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2576, pruned_loss=0.0422, over 3310113.16 frames. ], batch size: 42, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:02:11,590 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215564.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:02:12,739 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215565.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:02:29,264 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 10:02:41,712 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215586.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:02:59,710 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.111e+02 2.344e+02 3.035e+02 7.498e+02, threshold=4.688e+02, percent-clipped=4.0 2023-05-01 10:03:04,617 INFO [train.py:904] (7/8) Epoch 22, batch 2450, loss[loss=0.1545, simple_loss=0.2393, pruned_loss=0.03482, over 16993.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.258, pruned_loss=0.04182, over 3317867.04 frames. ], batch size: 41, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:03:08,505 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7283, 2.6331, 2.3215, 2.3997, 2.9936, 2.7360, 3.3602, 3.2193], device='cuda:7'), covar=tensor([0.0170, 0.0448, 0.0532, 0.0499, 0.0302, 0.0407, 0.0244, 0.0277], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0241, 0.0231, 0.0232, 0.0242, 0.0240, 0.0243, 0.0237], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:03:12,000 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215609.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:03:15,501 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-05-01 10:03:35,638 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215626.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:04:11,703 INFO [train.py:904] (7/8) Epoch 22, batch 2500, loss[loss=0.1494, simple_loss=0.2475, pruned_loss=0.02561, over 17111.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2581, pruned_loss=0.04148, over 3318493.06 frames. ], batch size: 48, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:04:34,285 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 10:04:36,349 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.63 vs. limit=5.0 2023-05-01 10:04:38,387 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8882, 2.0052, 2.3558, 2.7031, 2.7348, 2.7416, 1.9972, 3.0285], device='cuda:7'), covar=tensor([0.0186, 0.0516, 0.0387, 0.0316, 0.0348, 0.0335, 0.0547, 0.0164], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0196, 0.0182, 0.0187, 0.0200, 0.0157, 0.0198, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:04:51,433 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215681.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:05:02,673 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215689.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 10:05:15,513 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.601e+02 2.254e+02 2.666e+02 2.993e+02 6.188e+02, threshold=5.333e+02, percent-clipped=1.0 2023-05-01 10:05:20,567 INFO [train.py:904] (7/8) Epoch 22, batch 2550, loss[loss=0.1827, simple_loss=0.2862, pruned_loss=0.03965, over 17247.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2588, pruned_loss=0.04188, over 3324610.38 frames. ], batch size: 52, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:05:25,720 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9066, 5.1680, 5.3829, 5.1370, 5.1371, 5.7654, 5.2094, 4.8642], device='cuda:7'), covar=tensor([0.1219, 0.2036, 0.2381, 0.1926, 0.2741, 0.1041, 0.1558, 0.2436], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0618, 0.0679, 0.0514, 0.0684, 0.0714, 0.0535, 0.0684], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:06:09,954 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6798, 6.0544, 5.7913, 5.8808, 5.3732, 5.4880, 5.4965, 6.2051], device='cuda:7'), covar=tensor([0.1416, 0.0914, 0.1132, 0.0933, 0.1019, 0.0683, 0.1228, 0.0953], device='cuda:7'), in_proj_covar=tensor([0.0688, 0.0844, 0.0696, 0.0643, 0.0535, 0.0541, 0.0706, 0.0660], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:06:30,382 INFO [train.py:904] (7/8) Epoch 22, batch 2600, loss[loss=0.1726, simple_loss=0.2712, pruned_loss=0.03701, over 16719.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2583, pruned_loss=0.04167, over 3318519.42 frames. ], batch size: 57, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:07:34,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3693, 2.5457, 2.1777, 2.3924, 2.8959, 2.6187, 3.0217, 3.0553], device='cuda:7'), covar=tensor([0.0229, 0.0440, 0.0538, 0.0460, 0.0283, 0.0373, 0.0275, 0.0277], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0241, 0.0231, 0.0232, 0.0242, 0.0241, 0.0244, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:07:36,117 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.223e+02 2.543e+02 2.941e+02 8.287e+02, threshold=5.085e+02, percent-clipped=5.0 2023-05-01 10:07:40,011 INFO [train.py:904] (7/8) Epoch 22, batch 2650, loss[loss=0.1756, simple_loss=0.2814, pruned_loss=0.03492, over 17269.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2591, pruned_loss=0.04146, over 3320922.38 frames. ], batch size: 52, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:08:05,905 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3521, 4.2624, 4.2435, 3.9911, 4.0219, 4.3552, 4.0418, 4.1091], device='cuda:7'), covar=tensor([0.0749, 0.0928, 0.0385, 0.0322, 0.0824, 0.0517, 0.0795, 0.0662], device='cuda:7'), in_proj_covar=tensor([0.0307, 0.0446, 0.0357, 0.0356, 0.0368, 0.0412, 0.0245, 0.0430], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:08:16,692 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=215829.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:08:24,019 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1777, 5.1346, 4.9191, 4.4325, 5.0323, 2.0759, 4.7691, 4.8412], device='cuda:7'), covar=tensor([0.0088, 0.0090, 0.0212, 0.0389, 0.0105, 0.2609, 0.0140, 0.0196], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0158, 0.0201, 0.0179, 0.0179, 0.0210, 0.0190, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:08:48,050 INFO [train.py:904] (7/8) Epoch 22, batch 2700, loss[loss=0.1805, simple_loss=0.2619, pruned_loss=0.04952, over 15600.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2593, pruned_loss=0.041, over 3328439.88 frames. ], batch size: 191, lr: 3.09e-03, grad_scale: 8.0 2023-05-01 10:08:57,519 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215859.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:09:34,749 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=215886.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:09:39,332 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=215890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:09:53,923 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.142e+02 2.465e+02 2.972e+02 5.160e+02, threshold=4.929e+02, percent-clipped=1.0 2023-05-01 10:09:57,375 INFO [train.py:904] (7/8) Epoch 22, batch 2750, loss[loss=0.1725, simple_loss=0.2649, pruned_loss=0.04004, over 16094.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2595, pruned_loss=0.04087, over 3323845.42 frames. ], batch size: 35, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:09:59,491 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215904.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 10:10:21,992 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=215921.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:10:38,797 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=215934.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:10:51,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1233, 5.0939, 4.8393, 4.3480, 4.9589, 1.8450, 4.6442, 4.6319], device='cuda:7'), covar=tensor([0.0086, 0.0078, 0.0216, 0.0390, 0.0103, 0.3015, 0.0148, 0.0249], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0157, 0.0200, 0.0178, 0.0178, 0.0209, 0.0189, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:11:05,013 INFO [train.py:904] (7/8) Epoch 22, batch 2800, loss[loss=0.1597, simple_loss=0.2435, pruned_loss=0.03799, over 16792.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2597, pruned_loss=0.04115, over 3321427.12 frames. ], batch size: 39, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:11:22,068 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5813, 4.9276, 4.7480, 4.7176, 4.4874, 4.4115, 4.4197, 5.0268], device='cuda:7'), covar=tensor([0.1235, 0.0898, 0.0967, 0.0859, 0.0819, 0.1336, 0.1268, 0.0878], device='cuda:7'), in_proj_covar=tensor([0.0699, 0.0859, 0.0709, 0.0653, 0.0544, 0.0551, 0.0719, 0.0671], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:11:23,355 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3158, 3.3262, 2.1256, 3.5816, 2.5811, 3.5371, 2.1569, 2.7378], device='cuda:7'), covar=tensor([0.0317, 0.0425, 0.1532, 0.0317, 0.0859, 0.0873, 0.1476, 0.0725], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0197, 0.0168, 0.0180, 0.0223, 0.0205, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:11:42,413 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=215981.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 10:11:53,908 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=215989.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 10:12:07,422 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.162e+02 2.449e+02 3.092e+02 5.987e+02, threshold=4.899e+02, percent-clipped=3.0 2023-05-01 10:12:14,956 INFO [train.py:904] (7/8) Epoch 22, batch 2850, loss[loss=0.1554, simple_loss=0.2514, pruned_loss=0.02968, over 17217.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2595, pruned_loss=0.04104, over 3324682.80 frames. ], batch size: 52, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:12:51,518 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216028.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:12:52,546 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216029.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:13:04,675 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:13:26,223 INFO [train.py:904] (7/8) Epoch 22, batch 2900, loss[loss=0.1559, simple_loss=0.2369, pruned_loss=0.03752, over 16441.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2592, pruned_loss=0.0422, over 3304516.40 frames. ], batch size: 75, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:14:03,353 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5572, 3.4614, 2.7681, 2.1981, 2.3101, 2.3279, 3.5886, 3.1000], device='cuda:7'), covar=tensor([0.2727, 0.0675, 0.1692, 0.2934, 0.2809, 0.2104, 0.0556, 0.1499], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0272, 0.0307, 0.0314, 0.0298, 0.0262, 0.0296, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:14:17,836 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216089.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:14:31,569 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.264e+02 2.773e+02 3.437e+02 6.159e+02, threshold=5.547e+02, percent-clipped=1.0 2023-05-01 10:14:35,745 INFO [train.py:904] (7/8) Epoch 22, batch 2950, loss[loss=0.1906, simple_loss=0.2637, pruned_loss=0.05872, over 16953.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2591, pruned_loss=0.04323, over 3290221.39 frames. ], batch size: 116, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:15:00,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0526, 5.1761, 4.9517, 4.6289, 4.1787, 5.1755, 5.1817, 4.6624], device='cuda:7'), covar=tensor([0.0892, 0.0623, 0.0504, 0.0432, 0.2073, 0.0528, 0.0302, 0.0885], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0448, 0.0358, 0.0359, 0.0370, 0.0415, 0.0247, 0.0433], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:15:45,385 INFO [train.py:904] (7/8) Epoch 22, batch 3000, loss[loss=0.168, simple_loss=0.2458, pruned_loss=0.04515, over 16746.00 frames. ], tot_loss[loss=0.1724, simple_loss=0.2585, pruned_loss=0.04321, over 3295678.18 frames. ], batch size: 124, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:15:45,386 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 10:15:54,109 INFO [train.py:938] (7/8) Epoch 22, validation: loss=0.1347, simple_loss=0.2399, pruned_loss=0.0148, over 944034.00 frames. 2023-05-01 10:15:54,110 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 10:16:02,695 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216159.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:16:39,538 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=216185.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:16:59,839 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.146e+02 2.550e+02 3.086e+02 1.158e+03, threshold=5.100e+02, percent-clipped=1.0 2023-05-01 10:17:03,731 INFO [train.py:904] (7/8) Epoch 22, batch 3050, loss[loss=0.1814, simple_loss=0.273, pruned_loss=0.04491, over 17271.00 frames. ], tot_loss[loss=0.172, simple_loss=0.258, pruned_loss=0.04303, over 3293738.05 frames. ], batch size: 52, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:17:05,350 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216204.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:17:09,670 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216207.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:17:22,496 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.57 vs. limit=2.0 2023-05-01 10:17:29,414 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216221.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:17:32,037 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 10:17:32,962 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8941, 2.6854, 2.7039, 2.0132, 2.6206, 2.8092, 2.6075, 1.9693], device='cuda:7'), covar=tensor([0.0433, 0.0116, 0.0089, 0.0361, 0.0138, 0.0124, 0.0132, 0.0373], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0083, 0.0084, 0.0132, 0.0099, 0.0109, 0.0094, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:17:36,596 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6434, 2.4447, 2.3857, 4.5239, 2.4269, 2.8099, 2.4876, 2.6471], device='cuda:7'), covar=tensor([0.1208, 0.3691, 0.3112, 0.0469, 0.4022, 0.2623, 0.3445, 0.3544], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0452, 0.0371, 0.0332, 0.0439, 0.0520, 0.0423, 0.0530], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:18:11,437 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216252.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:18:12,222 INFO [train.py:904] (7/8) Epoch 22, batch 3100, loss[loss=0.1864, simple_loss=0.2578, pruned_loss=0.05747, over 16857.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2577, pruned_loss=0.04334, over 3293122.68 frames. ], batch size: 116, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:18:33,938 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216269.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:19:16,929 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.131e+02 2.523e+02 3.440e+02 6.792e+02, threshold=5.047e+02, percent-clipped=4.0 2023-05-01 10:19:21,083 INFO [train.py:904] (7/8) Epoch 22, batch 3150, loss[loss=0.2072, simple_loss=0.2831, pruned_loss=0.06565, over 12287.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2566, pruned_loss=0.04286, over 3293814.76 frames. ], batch size: 247, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:20:29,467 INFO [train.py:904] (7/8) Epoch 22, batch 3200, loss[loss=0.1781, simple_loss=0.2503, pruned_loss=0.05291, over 16728.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2549, pruned_loss=0.04188, over 3303878.61 frames. ], batch size: 124, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:20:57,910 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.04 vs. limit=5.0 2023-05-01 10:21:13,955 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=216384.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:21:36,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.182e+02 2.419e+02 2.918e+02 5.281e+02, threshold=4.837e+02, percent-clipped=1.0 2023-05-01 10:21:40,334 INFO [train.py:904] (7/8) Epoch 22, batch 3250, loss[loss=0.213, simple_loss=0.2885, pruned_loss=0.06872, over 16651.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2556, pruned_loss=0.04197, over 3300963.55 frames. ], batch size: 134, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:22:30,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1540, 3.3570, 3.5698, 2.4048, 3.2415, 3.6527, 3.3133, 1.9797], device='cuda:7'), covar=tensor([0.0536, 0.0115, 0.0061, 0.0399, 0.0121, 0.0090, 0.0104, 0.0481], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0083, 0.0084, 0.0132, 0.0099, 0.0109, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:22:52,223 INFO [train.py:904] (7/8) Epoch 22, batch 3300, loss[loss=0.1806, simple_loss=0.279, pruned_loss=0.04107, over 16681.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2568, pruned_loss=0.04231, over 3303490.24 frames. ], batch size: 57, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:23:36,593 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216485.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:23:48,351 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2096, 5.2298, 5.6447, 5.6339, 5.6737, 5.3031, 5.2438, 5.0345], device='cuda:7'), covar=tensor([0.0362, 0.0556, 0.0433, 0.0474, 0.0532, 0.0384, 0.0963, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0468, 0.0454, 0.0422, 0.0502, 0.0479, 0.0563, 0.0382], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 10:23:50,949 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 10:23:56,631 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 2.110e+02 2.633e+02 3.149e+02 8.538e+02, threshold=5.267e+02, percent-clipped=2.0 2023-05-01 10:24:00,076 INFO [train.py:904] (7/8) Epoch 22, batch 3350, loss[loss=0.1665, simple_loss=0.2655, pruned_loss=0.03377, over 16601.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2574, pruned_loss=0.04238, over 3303133.50 frames. ], batch size: 57, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:24:42,591 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216533.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:25:10,645 INFO [train.py:904] (7/8) Epoch 22, batch 3400, loss[loss=0.1645, simple_loss=0.2533, pruned_loss=0.03788, over 16547.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2569, pruned_loss=0.04186, over 3310111.38 frames. ], batch size: 75, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:25:47,308 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5321, 2.3562, 2.3242, 4.4009, 2.2379, 2.7857, 2.4094, 2.5244], device='cuda:7'), covar=tensor([0.1320, 0.3735, 0.3288, 0.0478, 0.4518, 0.2710, 0.3831, 0.3924], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0453, 0.0374, 0.0334, 0.0442, 0.0523, 0.0426, 0.0533], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:26:02,913 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216591.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:26:15,742 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.115e+02 2.508e+02 2.941e+02 4.033e+02, threshold=5.017e+02, percent-clipped=0.0 2023-05-01 10:26:19,666 INFO [train.py:904] (7/8) Epoch 22, batch 3450, loss[loss=0.1678, simple_loss=0.2527, pruned_loss=0.04145, over 16441.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2561, pruned_loss=0.04148, over 3315693.76 frames. ], batch size: 68, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:26:34,105 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3770, 4.1322, 4.4256, 2.3421, 4.6398, 4.8098, 3.6200, 3.8675], device='cuda:7'), covar=tensor([0.0677, 0.0237, 0.0274, 0.1247, 0.0104, 0.0164, 0.0398, 0.0376], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0110, 0.0100, 0.0140, 0.0082, 0.0130, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:26:51,252 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4487, 4.4056, 4.8183, 4.7973, 4.8347, 4.5190, 4.5197, 4.3786], device='cuda:7'), covar=tensor([0.0405, 0.0784, 0.0424, 0.0478, 0.0454, 0.0487, 0.0821, 0.0690], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0471, 0.0456, 0.0423, 0.0505, 0.0482, 0.0566, 0.0385], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 10:27:29,268 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216652.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:27:29,990 INFO [train.py:904] (7/8) Epoch 22, batch 3500, loss[loss=0.1454, simple_loss=0.2321, pruned_loss=0.02942, over 16832.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2547, pruned_loss=0.04129, over 3313649.28 frames. ], batch size: 42, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:27:32,743 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216655.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:28:12,667 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=216684.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:28:21,608 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 10:28:35,768 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.548e+02 1.992e+02 2.337e+02 2.852e+02 6.760e+02, threshold=4.673e+02, percent-clipped=2.0 2023-05-01 10:28:39,278 INFO [train.py:904] (7/8) Epoch 22, batch 3550, loss[loss=0.1412, simple_loss=0.2307, pruned_loss=0.02589, over 16831.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2542, pruned_loss=0.04089, over 3300155.64 frames. ], batch size: 42, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:28:42,529 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216705.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:28:57,296 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216716.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:29:20,539 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=216732.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:29:42,250 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-05-01 10:29:49,496 INFO [train.py:904] (7/8) Epoch 22, batch 3600, loss[loss=0.1354, simple_loss=0.2292, pruned_loss=0.02079, over 15762.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.253, pruned_loss=0.04019, over 3305881.33 frames. ], batch size: 35, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:29:56,444 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 10:30:02,264 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3613, 4.1262, 4.1033, 4.5047, 4.6159, 4.2471, 4.4686, 4.5741], device='cuda:7'), covar=tensor([0.1651, 0.1521, 0.2136, 0.1033, 0.0933, 0.1586, 0.2690, 0.1336], device='cuda:7'), in_proj_covar=tensor([0.0669, 0.0831, 0.0968, 0.0844, 0.0636, 0.0667, 0.0688, 0.0798], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:30:08,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=216766.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:30:17,174 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 10:30:17,990 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1766, 3.8039, 4.3534, 2.3017, 4.5320, 4.6382, 3.3618, 3.6555], device='cuda:7'), covar=tensor([0.0646, 0.0269, 0.0218, 0.1142, 0.0075, 0.0176, 0.0436, 0.0364], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0109, 0.0100, 0.0140, 0.0082, 0.0129, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:31:00,615 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.119e+02 2.478e+02 2.922e+02 5.214e+02, threshold=4.955e+02, percent-clipped=3.0 2023-05-01 10:31:03,556 INFO [train.py:904] (7/8) Epoch 22, batch 3650, loss[loss=0.1599, simple_loss=0.2324, pruned_loss=0.04367, over 16391.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.252, pruned_loss=0.04073, over 3307030.53 frames. ], batch size: 68, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:31:31,350 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4725, 5.8828, 5.6338, 5.6694, 5.2800, 5.2343, 5.2525, 6.0148], device='cuda:7'), covar=tensor([0.1338, 0.0972, 0.1040, 0.0920, 0.0881, 0.0735, 0.1184, 0.0878], device='cuda:7'), in_proj_covar=tensor([0.0699, 0.0856, 0.0706, 0.0650, 0.0543, 0.0549, 0.0717, 0.0670], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:32:00,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6537, 3.6890, 2.2696, 3.9640, 2.8644, 3.9065, 2.4750, 2.9895], device='cuda:7'), covar=tensor([0.0241, 0.0396, 0.1491, 0.0295, 0.0787, 0.0682, 0.1312, 0.0664], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0196, 0.0169, 0.0179, 0.0223, 0.0204, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:32:18,436 INFO [train.py:904] (7/8) Epoch 22, batch 3700, loss[loss=0.1595, simple_loss=0.2351, pruned_loss=0.04194, over 16615.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2512, pruned_loss=0.04237, over 3286928.74 frames. ], batch size: 89, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:33:31,446 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.238e+02 2.459e+02 3.071e+02 7.166e+02, threshold=4.918e+02, percent-clipped=1.0 2023-05-01 10:33:32,648 INFO [train.py:904] (7/8) Epoch 22, batch 3750, loss[loss=0.1941, simple_loss=0.2786, pruned_loss=0.05484, over 16403.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2522, pruned_loss=0.04363, over 3272786.08 frames. ], batch size: 146, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:34:36,641 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=216947.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:34:44,491 INFO [train.py:904] (7/8) Epoch 22, batch 3800, loss[loss=0.1963, simple_loss=0.2649, pruned_loss=0.0639, over 16873.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2525, pruned_loss=0.04488, over 3277541.73 frames. ], batch size: 109, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:35:03,048 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=216965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:35:55,033 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.325e+02 2.685e+02 3.161e+02 5.857e+02, threshold=5.370e+02, percent-clipped=3.0 2023-05-01 10:35:56,830 INFO [train.py:904] (7/8) Epoch 22, batch 3850, loss[loss=0.1752, simple_loss=0.246, pruned_loss=0.05216, over 16767.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2534, pruned_loss=0.0459, over 3278186.10 frames. ], batch size: 83, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:36:08,465 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217011.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 10:36:29,907 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217026.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:36:34,162 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2023-05-01 10:37:04,869 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-05-01 10:37:09,497 INFO [train.py:904] (7/8) Epoch 22, batch 3900, loss[loss=0.1727, simple_loss=0.2507, pruned_loss=0.04735, over 16732.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2528, pruned_loss=0.04621, over 3282911.56 frames. ], batch size: 134, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:37:22,103 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217061.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:38:21,561 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.294e+02 2.626e+02 3.091e+02 8.707e+02, threshold=5.252e+02, percent-clipped=1.0 2023-05-01 10:38:22,865 INFO [train.py:904] (7/8) Epoch 22, batch 3950, loss[loss=0.1777, simple_loss=0.2574, pruned_loss=0.04902, over 17119.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2523, pruned_loss=0.04658, over 3276689.15 frames. ], batch size: 48, lr: 3.08e-03, grad_scale: 4.0 2023-05-01 10:38:49,012 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6573, 4.6125, 4.5893, 4.0031, 4.5987, 1.8722, 4.3739, 4.2114], device='cuda:7'), covar=tensor([0.0159, 0.0147, 0.0201, 0.0401, 0.0114, 0.2804, 0.0180, 0.0245], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0160, 0.0204, 0.0183, 0.0182, 0.0212, 0.0194, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:39:35,768 INFO [train.py:904] (7/8) Epoch 22, batch 4000, loss[loss=0.1967, simple_loss=0.2729, pruned_loss=0.06028, over 16511.00 frames. ], tot_loss[loss=0.1723, simple_loss=0.2516, pruned_loss=0.04646, over 3288465.37 frames. ], batch size: 146, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:40:10,366 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4453, 2.3019, 1.9130, 2.0369, 2.6258, 2.4307, 2.5272, 2.7417], device='cuda:7'), covar=tensor([0.0302, 0.0436, 0.0569, 0.0507, 0.0248, 0.0365, 0.0214, 0.0293], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0239, 0.0230, 0.0230, 0.0240, 0.0239, 0.0244, 0.0237], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:40:48,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.995e+02 2.386e+02 2.991e+02 7.237e+02, threshold=4.771e+02, percent-clipped=2.0 2023-05-01 10:40:49,972 INFO [train.py:904] (7/8) Epoch 22, batch 4050, loss[loss=0.1702, simple_loss=0.2526, pruned_loss=0.04388, over 17230.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2525, pruned_loss=0.04588, over 3292865.81 frames. ], batch size: 45, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:41:55,324 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217247.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:42:02,488 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2023-05-01 10:42:04,923 INFO [train.py:904] (7/8) Epoch 22, batch 4100, loss[loss=0.1884, simple_loss=0.2685, pruned_loss=0.05416, over 16793.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2541, pruned_loss=0.04543, over 3295779.31 frames. ], batch size: 39, lr: 3.08e-03, grad_scale: 8.0 2023-05-01 10:43:03,750 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7412, 3.1729, 3.3758, 2.0041, 2.8804, 2.1743, 3.2754, 3.3400], device='cuda:7'), covar=tensor([0.0278, 0.0835, 0.0538, 0.2057, 0.0887, 0.1035, 0.0697, 0.1015], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0165, 0.0167, 0.0154, 0.0146, 0.0130, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:43:11,028 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217295.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:43:21,263 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.082e+02 2.531e+02 2.908e+02 7.878e+02, threshold=5.062e+02, percent-clipped=6.0 2023-05-01 10:43:23,211 INFO [train.py:904] (7/8) Epoch 22, batch 4150, loss[loss=0.1969, simple_loss=0.2833, pruned_loss=0.05522, over 17060.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2613, pruned_loss=0.04798, over 3252142.45 frames. ], batch size: 53, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:43:36,161 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217311.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:43:52,139 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217321.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:44:13,408 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5369, 3.6017, 3.3369, 2.9602, 3.1948, 3.5186, 3.2894, 3.3291], device='cuda:7'), covar=tensor([0.0561, 0.0544, 0.0271, 0.0248, 0.0514, 0.0453, 0.1546, 0.0471], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0443, 0.0355, 0.0355, 0.0364, 0.0411, 0.0243, 0.0426], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:44:39,666 INFO [train.py:904] (7/8) Epoch 22, batch 4200, loss[loss=0.1898, simple_loss=0.2864, pruned_loss=0.04659, over 16877.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2686, pruned_loss=0.04948, over 3241638.73 frames. ], batch size: 109, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:44:42,496 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.91 vs. limit=5.0 2023-05-01 10:44:50,073 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217359.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 10:44:52,715 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217361.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:45:27,759 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2359, 3.2713, 1.9967, 3.5308, 2.4746, 3.6235, 2.2278, 2.7243], device='cuda:7'), covar=tensor([0.0332, 0.0364, 0.1710, 0.0338, 0.0844, 0.0577, 0.1458, 0.0764], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0177, 0.0194, 0.0165, 0.0177, 0.0220, 0.0201, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:45:29,556 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2760, 3.5190, 3.2547, 3.0391, 2.8854, 3.4288, 3.1793, 3.1926], device='cuda:7'), covar=tensor([0.1060, 0.0916, 0.0545, 0.0456, 0.1230, 0.0659, 0.2522, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0441, 0.0353, 0.0353, 0.0362, 0.0409, 0.0242, 0.0422], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:45:32,810 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217388.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:45:53,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.187e+02 2.459e+02 3.018e+02 5.838e+02, threshold=4.919e+02, percent-clipped=1.0 2023-05-01 10:45:55,233 INFO [train.py:904] (7/8) Epoch 22, batch 4250, loss[loss=0.1694, simple_loss=0.2642, pruned_loss=0.0373, over 17102.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2717, pruned_loss=0.04886, over 3238693.38 frames. ], batch size: 47, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:46:04,664 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217409.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:47:04,214 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217449.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:47:09,154 INFO [train.py:904] (7/8) Epoch 22, batch 4300, loss[loss=0.1918, simple_loss=0.2764, pruned_loss=0.05362, over 11522.00 frames. ], tot_loss[loss=0.1845, simple_loss=0.273, pruned_loss=0.04794, over 3230968.95 frames. ], batch size: 248, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:47:36,581 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3353, 3.4458, 2.0011, 4.0092, 2.5769, 3.9662, 2.1149, 2.6710], device='cuda:7'), covar=tensor([0.0360, 0.0402, 0.1835, 0.0158, 0.0938, 0.0582, 0.1688, 0.0906], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0177, 0.0194, 0.0165, 0.0177, 0.0220, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:47:49,160 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7414, 3.7346, 3.9286, 3.6563, 3.8472, 4.2494, 3.8960, 3.5340], device='cuda:7'), covar=tensor([0.2194, 0.2210, 0.2118, 0.2319, 0.2484, 0.1789, 0.1520, 0.2828], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0597, 0.0653, 0.0494, 0.0660, 0.0690, 0.0515, 0.0664], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:47:51,975 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 10:48:07,903 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217492.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:48:23,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.190e+02 2.513e+02 2.861e+02 5.953e+02, threshold=5.026e+02, percent-clipped=1.0 2023-05-01 10:48:24,302 INFO [train.py:904] (7/8) Epoch 22, batch 4350, loss[loss=0.2056, simple_loss=0.2866, pruned_loss=0.06233, over 16459.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2761, pruned_loss=0.04879, over 3229694.34 frames. ], batch size: 146, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:49:39,993 INFO [train.py:904] (7/8) Epoch 22, batch 4400, loss[loss=0.2022, simple_loss=0.2904, pruned_loss=0.05702, over 16910.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2784, pruned_loss=0.04977, over 3232809.76 frames. ], batch size: 109, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:49:41,105 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217553.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:50:09,331 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8001, 3.0866, 3.2783, 1.9142, 2.7640, 2.0731, 3.3083, 3.2946], device='cuda:7'), covar=tensor([0.0208, 0.0759, 0.0564, 0.2119, 0.0888, 0.1024, 0.0578, 0.0910], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0164, 0.0166, 0.0153, 0.0145, 0.0130, 0.0143, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:50:52,394 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 2.074e+02 2.354e+02 2.786e+02 4.691e+02, threshold=4.709e+02, percent-clipped=0.0 2023-05-01 10:50:53,489 INFO [train.py:904] (7/8) Epoch 22, batch 4450, loss[loss=0.2027, simple_loss=0.2928, pruned_loss=0.05629, over 16505.00 frames. ], tot_loss[loss=0.1918, simple_loss=0.2815, pruned_loss=0.05107, over 3222362.74 frames. ], batch size: 68, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:51:09,767 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3413, 4.2201, 4.3821, 4.5120, 4.6337, 4.2256, 4.5970, 4.6509], device='cuda:7'), covar=tensor([0.1536, 0.1075, 0.1225, 0.0594, 0.0442, 0.1069, 0.0651, 0.0526], device='cuda:7'), in_proj_covar=tensor([0.0641, 0.0794, 0.0921, 0.0805, 0.0610, 0.0637, 0.0660, 0.0763], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:51:20,416 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=217621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:51:39,635 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1114, 2.4635, 2.6354, 1.9556, 2.7376, 2.8062, 2.4537, 2.3613], device='cuda:7'), covar=tensor([0.0724, 0.0277, 0.0225, 0.0924, 0.0117, 0.0231, 0.0474, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0109, 0.0098, 0.0139, 0.0081, 0.0127, 0.0130, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 10:51:48,881 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0949, 2.2484, 2.2441, 3.8066, 2.1501, 2.5558, 2.3394, 2.3615], device='cuda:7'), covar=tensor([0.1318, 0.3159, 0.2774, 0.0547, 0.4046, 0.2298, 0.2976, 0.3388], device='cuda:7'), in_proj_covar=tensor([0.0404, 0.0451, 0.0368, 0.0330, 0.0438, 0.0520, 0.0422, 0.0528], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:52:08,356 INFO [train.py:904] (7/8) Epoch 22, batch 4500, loss[loss=0.1889, simple_loss=0.2751, pruned_loss=0.05131, over 16394.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.2826, pruned_loss=0.05225, over 3222377.67 frames. ], batch size: 146, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:52:24,076 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217664.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:52:31,379 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0612, 2.2133, 2.1761, 3.6070, 2.1053, 2.5369, 2.3081, 2.3473], device='cuda:7'), covar=tensor([0.1394, 0.3221, 0.2895, 0.0618, 0.4082, 0.2331, 0.3046, 0.3366], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0452, 0.0369, 0.0330, 0.0439, 0.0521, 0.0423, 0.0529], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:52:32,287 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=217669.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:52:34,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7101, 4.5111, 4.3556, 2.8872, 3.8550, 4.4670, 3.8431, 2.6690], device='cuda:7'), covar=tensor([0.0505, 0.0026, 0.0041, 0.0395, 0.0092, 0.0075, 0.0093, 0.0382], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0083, 0.0085, 0.0134, 0.0099, 0.0110, 0.0096, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 10:53:11,560 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-01 10:53:18,214 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 1.861e+02 2.083e+02 2.379e+02 4.476e+02, threshold=4.167e+02, percent-clipped=0.0 2023-05-01 10:53:19,296 INFO [train.py:904] (7/8) Epoch 22, batch 4550, loss[loss=0.218, simple_loss=0.3008, pruned_loss=0.06764, over 17274.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2837, pruned_loss=0.05315, over 3238666.27 frames. ], batch size: 52, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:53:53,286 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217725.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:54:19,814 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217744.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:54:31,798 INFO [train.py:904] (7/8) Epoch 22, batch 4600, loss[loss=0.1981, simple_loss=0.2872, pruned_loss=0.05447, over 15404.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2845, pruned_loss=0.05331, over 3234701.73 frames. ], batch size: 191, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:54:59,671 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.34 vs. limit=5.0 2023-05-01 10:55:41,528 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 1.780e+02 2.045e+02 2.386e+02 4.197e+02, threshold=4.091e+02, percent-clipped=1.0 2023-05-01 10:55:42,809 INFO [train.py:904] (7/8) Epoch 22, batch 4650, loss[loss=0.2155, simple_loss=0.2941, pruned_loss=0.06848, over 11687.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.2839, pruned_loss=0.05364, over 3198659.37 frames. ], batch size: 248, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:55:49,327 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8904, 2.6789, 2.1276, 2.4342, 3.1998, 2.7448, 3.4613, 3.4524], device='cuda:7'), covar=tensor([0.0080, 0.0495, 0.0678, 0.0498, 0.0245, 0.0407, 0.0200, 0.0219], device='cuda:7'), in_proj_covar=tensor([0.0213, 0.0236, 0.0227, 0.0228, 0.0237, 0.0237, 0.0240, 0.0234], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:56:19,125 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8911, 4.1653, 4.0064, 4.0652, 3.7099, 3.7808, 3.8003, 4.1546], device='cuda:7'), covar=tensor([0.1136, 0.0841, 0.0962, 0.0747, 0.0841, 0.1673, 0.0904, 0.1007], device='cuda:7'), in_proj_covar=tensor([0.0673, 0.0822, 0.0680, 0.0627, 0.0523, 0.0531, 0.0689, 0.0647], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:56:26,525 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8969, 3.9768, 3.0980, 2.4538, 2.7686, 2.4780, 4.6227, 3.5272], device='cuda:7'), covar=tensor([0.2792, 0.0673, 0.1762, 0.2507, 0.2568, 0.2244, 0.0420, 0.1259], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0270, 0.0306, 0.0316, 0.0299, 0.0261, 0.0296, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 10:56:46,916 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=217848.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 10:56:52,997 INFO [train.py:904] (7/8) Epoch 22, batch 4700, loss[loss=0.1982, simple_loss=0.275, pruned_loss=0.06067, over 11657.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2807, pruned_loss=0.05233, over 3198126.72 frames. ], batch size: 247, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:57:45,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4626, 5.4702, 5.3623, 4.6239, 5.3964, 2.1247, 5.1481, 5.2274], device='cuda:7'), covar=tensor([0.0086, 0.0075, 0.0145, 0.0478, 0.0086, 0.2564, 0.0108, 0.0172], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0159, 0.0203, 0.0182, 0.0180, 0.0211, 0.0192, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 10:57:59,234 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=217898.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 10:58:04,667 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 1.895e+02 2.114e+02 2.418e+02 3.922e+02, threshold=4.227e+02, percent-clipped=0.0 2023-05-01 10:58:05,931 INFO [train.py:904] (7/8) Epoch 22, batch 4750, loss[loss=0.1906, simple_loss=0.2795, pruned_loss=0.05085, over 17015.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.277, pruned_loss=0.05044, over 3199531.53 frames. ], batch size: 53, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:58:54,159 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 10:59:17,341 INFO [train.py:904] (7/8) Epoch 22, batch 4800, loss[loss=0.171, simple_loss=0.2658, pruned_loss=0.03814, over 16858.00 frames. ], tot_loss[loss=0.1847, simple_loss=0.2731, pruned_loss=0.04817, over 3216026.86 frames. ], batch size: 116, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 10:59:27,659 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=217959.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 10:59:55,152 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1428, 2.2398, 2.2886, 3.9622, 2.1896, 2.5774, 2.3257, 2.4441], device='cuda:7'), covar=tensor([0.1415, 0.3649, 0.2864, 0.0527, 0.3897, 0.2517, 0.3555, 0.3107], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0448, 0.0367, 0.0327, 0.0435, 0.0516, 0.0419, 0.0524], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:00:36,339 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.164e+02 1.772e+02 2.165e+02 2.536e+02 4.360e+02, threshold=4.330e+02, percent-clipped=2.0 2023-05-01 11:00:36,355 INFO [train.py:904] (7/8) Epoch 22, batch 4850, loss[loss=0.1979, simple_loss=0.2829, pruned_loss=0.05643, over 12221.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2735, pruned_loss=0.04706, over 3219390.46 frames. ], batch size: 248, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:00:39,266 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6617, 3.4973, 3.9640, 1.9113, 4.0948, 4.1585, 3.1024, 3.0534], device='cuda:7'), covar=tensor([0.0823, 0.0274, 0.0173, 0.1301, 0.0072, 0.0131, 0.0421, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0109, 0.0098, 0.0139, 0.0081, 0.0126, 0.0129, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:01:01,666 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218020.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:01:29,350 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3458, 4.1559, 4.1684, 2.8949, 3.6710, 4.2333, 3.6000, 2.3910], device='cuda:7'), covar=tensor([0.0568, 0.0049, 0.0042, 0.0360, 0.0102, 0.0094, 0.0115, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0081, 0.0083, 0.0131, 0.0097, 0.0108, 0.0093, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:01:38,617 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218044.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:01:51,569 INFO [train.py:904] (7/8) Epoch 22, batch 4900, loss[loss=0.172, simple_loss=0.2708, pruned_loss=0.03658, over 16364.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2725, pruned_loss=0.04565, over 3224878.59 frames. ], batch size: 146, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:02:11,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6720, 2.2990, 2.1734, 3.1105, 1.6784, 3.4573, 1.5224, 2.7115], device='cuda:7'), covar=tensor([0.1483, 0.0954, 0.1387, 0.0164, 0.0129, 0.0388, 0.1835, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0173, 0.0193, 0.0190, 0.0203, 0.0214, 0.0200, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:02:31,139 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4010, 3.2026, 2.6194, 2.1322, 2.1666, 2.2469, 3.3534, 2.9311], device='cuda:7'), covar=tensor([0.2953, 0.0809, 0.1849, 0.2919, 0.2461, 0.2082, 0.0574, 0.1273], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0269, 0.0305, 0.0315, 0.0299, 0.0260, 0.0295, 0.0339], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:02:34,962 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 11:02:35,733 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2594, 2.9971, 3.3444, 1.6816, 3.4370, 3.4774, 2.8058, 2.5290], device='cuda:7'), covar=tensor([0.0869, 0.0294, 0.0180, 0.1280, 0.0086, 0.0158, 0.0434, 0.0581], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0108, 0.0098, 0.0138, 0.0081, 0.0125, 0.0129, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:02:49,305 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218092.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:03:01,688 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218100.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:03:05,450 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.042e+02 2.302e+02 2.921e+02 6.655e+02, threshold=4.605e+02, percent-clipped=4.0 2023-05-01 11:03:05,471 INFO [train.py:904] (7/8) Epoch 22, batch 4950, loss[loss=0.1714, simple_loss=0.2702, pruned_loss=0.03629, over 16871.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2725, pruned_loss=0.04576, over 3205533.42 frames. ], batch size: 96, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:04:11,532 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:04:18,898 INFO [train.py:904] (7/8) Epoch 22, batch 5000, loss[loss=0.1638, simple_loss=0.2695, pruned_loss=0.02908, over 16810.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2739, pruned_loss=0.04593, over 3211872.04 frames. ], batch size: 102, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:04:30,061 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218161.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:05:01,513 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218182.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:05:20,601 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.47 vs. limit=5.0 2023-05-01 11:05:22,063 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218196.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:05:32,360 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.007e+02 2.287e+02 2.605e+02 5.161e+02, threshold=4.575e+02, percent-clipped=1.0 2023-05-01 11:05:32,376 INFO [train.py:904] (7/8) Epoch 22, batch 5050, loss[loss=0.204, simple_loss=0.2879, pruned_loss=0.06006, over 11977.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2746, pruned_loss=0.04599, over 3212681.30 frames. ], batch size: 247, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:05:42,379 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218210.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:06:30,533 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218243.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:06:44,152 INFO [train.py:904] (7/8) Epoch 22, batch 5100, loss[loss=0.1497, simple_loss=0.2439, pruned_loss=0.02771, over 16740.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2728, pruned_loss=0.04544, over 3210971.69 frames. ], batch size: 83, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:06:45,824 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218254.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 11:06:48,274 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7328, 3.9422, 4.0145, 2.5263, 3.3347, 2.8636, 3.9762, 4.0645], device='cuda:7'), covar=tensor([0.0226, 0.0693, 0.0569, 0.1808, 0.0813, 0.0805, 0.0561, 0.0869], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0165, 0.0168, 0.0153, 0.0146, 0.0130, 0.0144, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:07:10,549 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218271.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:07:57,332 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 1.925e+02 2.150e+02 2.531e+02 3.604e+02, threshold=4.300e+02, percent-clipped=0.0 2023-05-01 11:07:57,353 INFO [train.py:904] (7/8) Epoch 22, batch 5150, loss[loss=0.1839, simple_loss=0.2837, pruned_loss=0.04198, over 16889.00 frames. ], tot_loss[loss=0.1814, simple_loss=0.273, pruned_loss=0.04492, over 3198884.81 frames. ], batch size: 116, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:08:23,166 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218320.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:09:11,242 INFO [train.py:904] (7/8) Epoch 22, batch 5200, loss[loss=0.1814, simple_loss=0.2678, pruned_loss=0.0475, over 16417.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2725, pruned_loss=0.04475, over 3191594.95 frames. ], batch size: 146, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:09:33,136 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218368.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:10:23,970 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 1.955e+02 2.305e+02 2.711e+02 5.108e+02, threshold=4.610e+02, percent-clipped=2.0 2023-05-01 11:10:23,986 INFO [train.py:904] (7/8) Epoch 22, batch 5250, loss[loss=0.17, simple_loss=0.2585, pruned_loss=0.04076, over 16688.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2694, pruned_loss=0.04393, over 3216889.04 frames. ], batch size: 76, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:11:15,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0758, 3.9660, 4.1214, 4.3039, 4.4248, 4.0552, 4.3362, 4.4611], device='cuda:7'), covar=tensor([0.1725, 0.1259, 0.1650, 0.0809, 0.0625, 0.1566, 0.0941, 0.0695], device='cuda:7'), in_proj_covar=tensor([0.0630, 0.0782, 0.0909, 0.0793, 0.0597, 0.0628, 0.0649, 0.0749], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:11:37,157 INFO [train.py:904] (7/8) Epoch 22, batch 5300, loss[loss=0.166, simple_loss=0.2527, pruned_loss=0.03965, over 16747.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2659, pruned_loss=0.04303, over 3217482.30 frames. ], batch size: 134, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:11:39,617 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.56 vs. limit=2.0 2023-05-01 11:11:41,616 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218456.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:12:47,930 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0206, 1.9734, 2.6062, 2.9538, 2.8685, 3.4790, 2.2504, 3.3820], device='cuda:7'), covar=tensor([0.0244, 0.0570, 0.0364, 0.0324, 0.0310, 0.0167, 0.0565, 0.0147], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0196, 0.0182, 0.0186, 0.0200, 0.0155, 0.0199, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:12:51,259 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 1.955e+02 2.262e+02 2.726e+02 6.747e+02, threshold=4.525e+02, percent-clipped=3.0 2023-05-01 11:12:51,280 INFO [train.py:904] (7/8) Epoch 22, batch 5350, loss[loss=0.2054, simple_loss=0.2969, pruned_loss=0.05693, over 15287.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.264, pruned_loss=0.04223, over 3223250.65 frames. ], batch size: 190, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:13:01,240 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218509.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:13:43,240 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:14:03,271 INFO [train.py:904] (7/8) Epoch 22, batch 5400, loss[loss=0.1753, simple_loss=0.2747, pruned_loss=0.03794, over 16527.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2659, pruned_loss=0.04262, over 3206069.20 frames. ], batch size: 68, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:14:06,198 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218554.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 11:14:23,163 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218566.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 11:14:29,588 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218570.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:14:35,234 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218574.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:15:02,085 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2062, 4.3451, 4.4684, 4.2152, 4.3344, 4.8384, 4.3693, 4.0599], device='cuda:7'), covar=tensor([0.1813, 0.1870, 0.2238, 0.2063, 0.2501, 0.1076, 0.1578, 0.2562], device='cuda:7'), in_proj_covar=tensor([0.0409, 0.0580, 0.0637, 0.0486, 0.0648, 0.0675, 0.0508, 0.0654], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:15:06,903 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6163, 1.8699, 2.3120, 2.6000, 2.6210, 3.1097, 2.0602, 3.0041], device='cuda:7'), covar=tensor([0.0244, 0.0515, 0.0347, 0.0352, 0.0314, 0.0179, 0.0511, 0.0154], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0195, 0.0181, 0.0186, 0.0199, 0.0154, 0.0198, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:15:18,767 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218602.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 11:15:19,500 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.152e+02 2.360e+02 2.669e+02 5.554e+02, threshold=4.720e+02, percent-clipped=2.0 2023-05-01 11:15:19,521 INFO [train.py:904] (7/8) Epoch 22, batch 5450, loss[loss=0.2035, simple_loss=0.291, pruned_loss=0.05799, over 16909.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2691, pruned_loss=0.04445, over 3213562.42 frames. ], batch size: 96, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:15:36,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3554, 4.2122, 4.3882, 4.5437, 4.7263, 4.2614, 4.6560, 4.7278], device='cuda:7'), covar=tensor([0.1670, 0.1300, 0.1456, 0.0751, 0.0543, 0.1071, 0.0724, 0.0615], device='cuda:7'), in_proj_covar=tensor([0.0632, 0.0786, 0.0912, 0.0796, 0.0598, 0.0631, 0.0651, 0.0752], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:16:10,998 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=218635.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:16:38,008 INFO [train.py:904] (7/8) Epoch 22, batch 5500, loss[loss=0.2251, simple_loss=0.3104, pruned_loss=0.06986, over 16548.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2757, pruned_loss=0.04846, over 3177880.99 frames. ], batch size: 68, lr: 3.07e-03, grad_scale: 8.0 2023-05-01 11:16:48,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4012, 4.4309, 4.2707, 3.9876, 3.9709, 4.3695, 4.0907, 4.1229], device='cuda:7'), covar=tensor([0.0561, 0.0618, 0.0297, 0.0317, 0.0869, 0.0546, 0.0634, 0.0576], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0431, 0.0344, 0.0343, 0.0353, 0.0398, 0.0236, 0.0411], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:17:20,024 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0317, 3.0614, 1.8707, 3.2553, 2.3667, 3.3569, 2.0779, 2.5581], device='cuda:7'), covar=tensor([0.0311, 0.0409, 0.1603, 0.0213, 0.0807, 0.0479, 0.1469, 0.0700], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0176, 0.0194, 0.0162, 0.0177, 0.0216, 0.0201, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:17:57,931 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.055e+02 2.950e+02 3.517e+02 4.298e+02 7.348e+02, threshold=7.033e+02, percent-clipped=15.0 2023-05-01 11:17:57,953 INFO [train.py:904] (7/8) Epoch 22, batch 5550, loss[loss=0.2259, simple_loss=0.3025, pruned_loss=0.07461, over 16313.00 frames. ], tot_loss[loss=0.1946, simple_loss=0.2829, pruned_loss=0.05314, over 3148338.79 frames. ], batch size: 146, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:19:20,451 INFO [train.py:904] (7/8) Epoch 22, batch 5600, loss[loss=0.2097, simple_loss=0.296, pruned_loss=0.06172, over 16261.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2874, pruned_loss=0.05677, over 3131718.66 frames. ], batch size: 165, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:19:26,482 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218756.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:20:41,878 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.037e+02 3.163e+02 3.914e+02 4.827e+02 8.104e+02, threshold=7.827e+02, percent-clipped=4.0 2023-05-01 11:20:41,902 INFO [train.py:904] (7/8) Epoch 22, batch 5650, loss[loss=0.2739, simple_loss=0.3438, pruned_loss=0.102, over 15280.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2933, pruned_loss=0.06171, over 3088001.74 frames. ], batch size: 190, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:20:44,450 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218804.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:21:35,495 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218838.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:21:56,896 INFO [train.py:904] (7/8) Epoch 22, batch 5700, loss[loss=0.219, simple_loss=0.3076, pruned_loss=0.06523, over 15364.00 frames. ], tot_loss[loss=0.2098, simple_loss=0.2946, pruned_loss=0.06254, over 3093016.98 frames. ], batch size: 190, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:22:16,244 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218865.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:22:18,254 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=218866.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:22:48,657 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218886.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:23:13,877 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 2.926e+02 3.484e+02 4.272e+02 7.645e+02, threshold=6.969e+02, percent-clipped=0.0 2023-05-01 11:23:13,893 INFO [train.py:904] (7/8) Epoch 22, batch 5750, loss[loss=0.2147, simple_loss=0.2824, pruned_loss=0.07353, over 11130.00 frames. ], tot_loss[loss=0.2131, simple_loss=0.2972, pruned_loss=0.06451, over 3052209.61 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:23:32,711 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=218914.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:23:55,976 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2407, 3.7345, 3.6791, 2.2116, 3.1602, 2.5431, 3.7385, 3.7896], device='cuda:7'), covar=tensor([0.0311, 0.0687, 0.0657, 0.2003, 0.0823, 0.1038, 0.0576, 0.0887], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0163, 0.0166, 0.0153, 0.0145, 0.0130, 0.0143, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:23:57,794 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=218930.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:24:09,050 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5516, 3.4570, 3.4936, 2.7563, 3.3457, 2.0569, 3.1794, 2.7637], device='cuda:7'), covar=tensor([0.0152, 0.0137, 0.0188, 0.0215, 0.0105, 0.2396, 0.0142, 0.0248], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0156, 0.0199, 0.0178, 0.0176, 0.0208, 0.0187, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:24:34,384 INFO [train.py:904] (7/8) Epoch 22, batch 5800, loss[loss=0.1975, simple_loss=0.2865, pruned_loss=0.05426, over 16905.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2967, pruned_loss=0.06309, over 3057434.42 frames. ], batch size: 116, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:24:38,748 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7637, 6.1109, 5.8136, 5.8864, 5.5567, 5.4492, 5.4825, 6.2070], device='cuda:7'), covar=tensor([0.1287, 0.0794, 0.1011, 0.0913, 0.0803, 0.0638, 0.1203, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0663, 0.0810, 0.0672, 0.0616, 0.0512, 0.0521, 0.0675, 0.0633], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:24:54,940 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:25:24,837 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=218984.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:25:53,555 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 2.954e+02 3.399e+02 3.875e+02 5.875e+02, threshold=6.799e+02, percent-clipped=0.0 2023-05-01 11:25:53,591 INFO [train.py:904] (7/8) Epoch 22, batch 5850, loss[loss=0.2289, simple_loss=0.2977, pruned_loss=0.08004, over 11226.00 frames. ], tot_loss[loss=0.2088, simple_loss=0.2945, pruned_loss=0.06155, over 3064309.36 frames. ], batch size: 246, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:26:21,592 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219021.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:26:30,050 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219026.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:27:01,371 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 11:27:02,417 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219045.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:27:14,929 INFO [train.py:904] (7/8) Epoch 22, batch 5900, loss[loss=0.2182, simple_loss=0.2987, pruned_loss=0.0689, over 16376.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2939, pruned_loss=0.06129, over 3076737.46 frames. ], batch size: 146, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:27:36,239 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-05-01 11:27:49,534 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219072.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:28:04,756 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219082.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:28:35,887 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.790e+02 2.529e+02 2.968e+02 3.874e+02 6.567e+02, threshold=5.937e+02, percent-clipped=0.0 2023-05-01 11:28:35,908 INFO [train.py:904] (7/8) Epoch 22, batch 5950, loss[loss=0.1906, simple_loss=0.2824, pruned_loss=0.04945, over 16782.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2945, pruned_loss=0.05984, over 3082351.39 frames. ], batch size: 83, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:29:24,565 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219133.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:29:51,910 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3670, 2.9469, 2.6489, 2.3143, 2.2477, 2.2679, 2.9286, 2.8259], device='cuda:7'), covar=tensor([0.2491, 0.0644, 0.1581, 0.2386, 0.2275, 0.2096, 0.0491, 0.1194], device='cuda:7'), in_proj_covar=tensor([0.0325, 0.0268, 0.0304, 0.0313, 0.0296, 0.0258, 0.0295, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:29:57,674 INFO [train.py:904] (7/8) Epoch 22, batch 6000, loss[loss=0.1861, simple_loss=0.2762, pruned_loss=0.04802, over 16275.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2938, pruned_loss=0.05957, over 3065686.83 frames. ], batch size: 165, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:29:57,674 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 11:30:07,623 INFO [train.py:938] (7/8) Epoch 22, validation: loss=0.1507, simple_loss=0.2632, pruned_loss=0.01907, over 944034.00 frames. 2023-05-01 11:30:07,624 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 11:30:27,726 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219165.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:31:28,925 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.979e+02 2.852e+02 3.411e+02 4.220e+02 7.320e+02, threshold=6.821e+02, percent-clipped=6.0 2023-05-01 11:31:28,945 INFO [train.py:904] (7/8) Epoch 22, batch 6050, loss[loss=0.1918, simple_loss=0.2873, pruned_loss=0.04812, over 16646.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2916, pruned_loss=0.05855, over 3076161.26 frames. ], batch size: 62, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:31:29,697 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:31:42,182 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2976, 3.7495, 3.6456, 2.1394, 3.1765, 2.5857, 3.6410, 3.8840], device='cuda:7'), covar=tensor([0.0276, 0.0694, 0.0647, 0.2119, 0.0833, 0.1000, 0.0631, 0.0879], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0164, 0.0168, 0.0154, 0.0146, 0.0130, 0.0144, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:31:45,631 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219213.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:32:11,192 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219230.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:32:39,118 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5141, 2.9858, 3.0876, 1.9644, 2.7776, 2.1049, 3.1824, 3.1928], device='cuda:7'), covar=tensor([0.0275, 0.0788, 0.0647, 0.2171, 0.0901, 0.1060, 0.0620, 0.0961], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0168, 0.0154, 0.0146, 0.0131, 0.0144, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:32:46,221 INFO [train.py:904] (7/8) Epoch 22, batch 6100, loss[loss=0.179, simple_loss=0.2734, pruned_loss=0.04228, over 16881.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2908, pruned_loss=0.05745, over 3107459.90 frames. ], batch size: 96, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:33:05,416 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219264.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:33:26,410 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219278.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:34:04,119 INFO [train.py:904] (7/8) Epoch 22, batch 6150, loss[loss=0.2179, simple_loss=0.2954, pruned_loss=0.0702, over 11702.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2894, pruned_loss=0.05727, over 3102634.77 frames. ], batch size: 248, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:34:05,867 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.700e+02 2.531e+02 2.938e+02 3.735e+02 5.885e+02, threshold=5.877e+02, percent-clipped=0.0 2023-05-01 11:34:33,726 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219321.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:35:03,745 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:35:06,982 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219342.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:35:22,814 INFO [train.py:904] (7/8) Epoch 22, batch 6200, loss[loss=0.2085, simple_loss=0.2931, pruned_loss=0.06195, over 16483.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2871, pruned_loss=0.05637, over 3128290.86 frames. ], batch size: 146, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:36:02,570 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219377.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:36:10,569 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 11:36:19,110 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 11:36:42,480 INFO [train.py:904] (7/8) Epoch 22, batch 6250, loss[loss=0.1779, simple_loss=0.2685, pruned_loss=0.04365, over 17037.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2867, pruned_loss=0.05617, over 3130301.00 frames. ], batch size: 50, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:36:43,086 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219403.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:36:43,735 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.827e+02 2.722e+02 3.215e+02 3.900e+02 7.497e+02, threshold=6.430e+02, percent-clipped=6.0 2023-05-01 11:36:47,417 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3520, 3.4552, 3.6118, 3.5922, 3.6083, 3.4248, 3.4502, 3.4898], device='cuda:7'), covar=tensor([0.0444, 0.0720, 0.0503, 0.0485, 0.0555, 0.0579, 0.0838, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0451, 0.0437, 0.0404, 0.0484, 0.0458, 0.0543, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 11:37:20,981 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219428.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:37:57,570 INFO [train.py:904] (7/8) Epoch 22, batch 6300, loss[loss=0.1728, simple_loss=0.27, pruned_loss=0.03779, over 16800.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2864, pruned_loss=0.05533, over 3129627.71 frames. ], batch size: 102, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:37:58,316 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8126, 2.9540, 2.6893, 4.8657, 3.5977, 4.2044, 1.6652, 3.0449], device='cuda:7'), covar=tensor([0.1365, 0.0744, 0.1231, 0.0150, 0.0428, 0.0454, 0.1672, 0.0885], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0174, 0.0195, 0.0190, 0.0207, 0.0215, 0.0202, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:38:47,325 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 11:38:56,008 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219490.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:39:15,309 INFO [train.py:904] (7/8) Epoch 22, batch 6350, loss[loss=0.1876, simple_loss=0.272, pruned_loss=0.05167, over 17018.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2874, pruned_loss=0.05669, over 3111157.36 frames. ], batch size: 50, lr: 3.06e-03, grad_scale: 4.0 2023-05-01 11:39:16,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.781e+02 2.840e+02 3.669e+02 4.695e+02 9.321e+02, threshold=7.339e+02, percent-clipped=9.0 2023-05-01 11:39:27,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0760, 2.1355, 2.2407, 3.5864, 2.1469, 2.4956, 2.2524, 2.2915], device='cuda:7'), covar=tensor([0.1285, 0.3356, 0.2780, 0.0577, 0.3866, 0.2376, 0.3417, 0.3138], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0447, 0.0366, 0.0325, 0.0435, 0.0514, 0.0419, 0.0521], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:39:41,696 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4540, 4.4100, 4.3005, 3.4709, 4.3765, 1.6857, 4.1520, 3.9121], device='cuda:7'), covar=tensor([0.0103, 0.0097, 0.0185, 0.0330, 0.0093, 0.2770, 0.0127, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0155, 0.0198, 0.0178, 0.0174, 0.0206, 0.0186, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:40:18,923 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-01 11:40:28,706 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:40:31,084 INFO [train.py:904] (7/8) Epoch 22, batch 6400, loss[loss=0.2342, simple_loss=0.3054, pruned_loss=0.08144, over 11580.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2875, pruned_loss=0.05739, over 3118388.05 frames. ], batch size: 247, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:40:40,471 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219559.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:41:13,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7089, 4.8993, 5.0769, 4.8184, 4.8929, 5.4659, 4.9454, 4.6772], device='cuda:7'), covar=tensor([0.1146, 0.1942, 0.2735, 0.2033, 0.2491, 0.0916, 0.1494, 0.2438], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0591, 0.0654, 0.0495, 0.0658, 0.0684, 0.0514, 0.0663], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:41:45,423 INFO [train.py:904] (7/8) Epoch 22, batch 6450, loss[loss=0.1873, simple_loss=0.2959, pruned_loss=0.03933, over 16851.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2877, pruned_loss=0.05664, over 3111787.57 frames. ], batch size: 102, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:41:47,190 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.783e+02 2.795e+02 3.528e+02 4.334e+02 7.123e+02, threshold=7.056e+02, percent-clipped=0.0 2023-05-01 11:42:13,368 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:42:45,358 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:43:04,573 INFO [train.py:904] (7/8) Epoch 22, batch 6500, loss[loss=0.2043, simple_loss=0.2905, pruned_loss=0.05903, over 16761.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2859, pruned_loss=0.05584, over 3116311.48 frames. ], batch size: 124, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:43:30,547 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219669.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:43:43,509 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:44:01,335 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:44:19,772 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219698.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 11:44:23,496 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-05-01 11:44:26,643 INFO [train.py:904] (7/8) Epoch 22, batch 6550, loss[loss=0.1947, simple_loss=0.3044, pruned_loss=0.04248, over 16239.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2888, pruned_loss=0.05656, over 3118451.78 frames. ], batch size: 165, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:44:28,428 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.583e+02 3.105e+02 3.738e+02 8.268e+02, threshold=6.210e+02, percent-clipped=2.0 2023-05-01 11:45:03,980 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219725.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:45:08,809 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219728.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:45:46,190 INFO [train.py:904] (7/8) Epoch 22, batch 6600, loss[loss=0.2114, simple_loss=0.2933, pruned_loss=0.06472, over 16233.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.291, pruned_loss=0.05728, over 3125657.68 frames. ], batch size: 165, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:46:02,228 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219763.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:46:23,294 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219776.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:47:06,520 INFO [train.py:904] (7/8) Epoch 22, batch 6650, loss[loss=0.1988, simple_loss=0.2894, pruned_loss=0.05414, over 16424.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2916, pruned_loss=0.05859, over 3108162.54 frames. ], batch size: 146, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:47:07,643 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.837e+02 2.943e+02 3.397e+02 4.500e+02 8.184e+02, threshold=6.793e+02, percent-clipped=7.0 2023-05-01 11:47:39,782 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219824.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:10,757 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=219846.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:21,250 INFO [train.py:904] (7/8) Epoch 22, batch 6700, loss[loss=0.2127, simple_loss=0.3, pruned_loss=0.06271, over 15427.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.29, pruned_loss=0.05863, over 3097829.58 frames. ], batch size: 191, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:48:27,690 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=219857.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:30,152 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219859.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:48:53,557 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1992, 4.2642, 4.0839, 3.8271, 3.8386, 4.2087, 3.8582, 3.9421], device='cuda:7'), covar=tensor([0.0594, 0.0585, 0.0305, 0.0299, 0.0748, 0.0503, 0.0860, 0.0655], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0427, 0.0340, 0.0337, 0.0346, 0.0392, 0.0234, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 11:49:26,601 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9037, 3.9775, 2.8914, 2.4674, 2.7378, 2.5977, 4.3360, 3.5850], device='cuda:7'), covar=tensor([0.2559, 0.0654, 0.1875, 0.2515, 0.2488, 0.1929, 0.0384, 0.1150], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0269, 0.0305, 0.0315, 0.0298, 0.0260, 0.0296, 0.0337], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:49:36,043 INFO [train.py:904] (7/8) Epoch 22, batch 6750, loss[loss=0.1886, simple_loss=0.2683, pruned_loss=0.05444, over 16637.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2889, pruned_loss=0.05864, over 3106903.74 frames. ], batch size: 57, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:49:37,878 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 2.924e+02 3.493e+02 4.502e+02 8.889e+02, threshold=6.985e+02, percent-clipped=2.0 2023-05-01 11:49:43,915 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=219907.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:50:00,728 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=219918.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:50:50,226 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2825, 4.3415, 4.6366, 4.5946, 4.6181, 4.3303, 4.3267, 4.2142], device='cuda:7'), covar=tensor([0.0358, 0.0604, 0.0388, 0.0428, 0.0525, 0.0425, 0.0988, 0.0591], device='cuda:7'), in_proj_covar=tensor([0.0409, 0.0454, 0.0439, 0.0408, 0.0489, 0.0462, 0.0547, 0.0369], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 11:50:52,007 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0474, 5.4881, 5.6628, 5.3757, 5.4681, 5.9791, 5.4546, 5.2522], device='cuda:7'), covar=tensor([0.0902, 0.1592, 0.2032, 0.1733, 0.2108, 0.0911, 0.1551, 0.2387], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0585, 0.0647, 0.0487, 0.0650, 0.0676, 0.0511, 0.0656], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 11:50:54,132 INFO [train.py:904] (7/8) Epoch 22, batch 6800, loss[loss=0.2005, simple_loss=0.2864, pruned_loss=0.05734, over 16733.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2888, pruned_loss=0.05895, over 3093584.13 frames. ], batch size: 124, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:52:03,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4394, 3.5689, 3.7779, 2.0808, 3.1243, 2.5851, 3.7791, 3.8408], device='cuda:7'), covar=tensor([0.0298, 0.0879, 0.0607, 0.2174, 0.0859, 0.0954, 0.0709, 0.1018], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0169, 0.0155, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 11:52:05,039 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=219998.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:52:15,016 INFO [train.py:904] (7/8) Epoch 22, batch 6850, loss[loss=0.1996, simple_loss=0.2937, pruned_loss=0.05276, over 17090.00 frames. ], tot_loss[loss=0.2052, simple_loss=0.2908, pruned_loss=0.05977, over 3092176.16 frames. ], batch size: 49, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:52:16,800 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.941e+02 2.883e+02 3.357e+02 3.930e+02 6.765e+02, threshold=6.713e+02, percent-clipped=0.0 2023-05-01 11:52:41,012 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 11:53:21,303 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220046.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 11:53:31,120 INFO [train.py:904] (7/8) Epoch 22, batch 6900, loss[loss=0.2952, simple_loss=0.3467, pruned_loss=0.1218, over 11556.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2927, pruned_loss=0.05898, over 3099485.69 frames. ], batch size: 246, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:54:24,707 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220087.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:54:41,936 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220098.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:54:50,004 INFO [train.py:904] (7/8) Epoch 22, batch 6950, loss[loss=0.1954, simple_loss=0.2904, pruned_loss=0.05015, over 16683.00 frames. ], tot_loss[loss=0.2072, simple_loss=0.2941, pruned_loss=0.06009, over 3099063.96 frames. ], batch size: 89, lr: 3.06e-03, grad_scale: 8.0 2023-05-01 11:54:51,083 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.857e+02 2.861e+02 3.409e+02 4.251e+02 1.328e+03, threshold=6.818e+02, percent-clipped=1.0 2023-05-01 11:55:14,115 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220119.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:55:54,663 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220146.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:55:57,605 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:56:03,780 INFO [train.py:904] (7/8) Epoch 22, batch 7000, loss[loss=0.2199, simple_loss=0.3118, pruned_loss=0.06397, over 16864.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2943, pruned_loss=0.05969, over 3094345.39 frames. ], batch size: 116, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 11:56:14,388 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220159.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:56:42,733 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-01 11:57:05,684 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220194.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:57:18,850 INFO [train.py:904] (7/8) Epoch 22, batch 7050, loss[loss=0.1944, simple_loss=0.2892, pruned_loss=0.04982, over 16842.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2945, pruned_loss=0.05909, over 3103034.00 frames. ], batch size: 102, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 11:57:21,955 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 2.577e+02 3.204e+02 3.781e+02 6.030e+02, threshold=6.408e+02, percent-clipped=0.0 2023-05-01 11:57:35,250 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220213.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:57:59,358 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220229.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:58:36,513 INFO [train.py:904] (7/8) Epoch 22, batch 7100, loss[loss=0.2228, simple_loss=0.2853, pruned_loss=0.08018, over 11694.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2923, pruned_loss=0.05859, over 3101833.32 frames. ], batch size: 248, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 11:58:45,457 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220258.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:59:35,835 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220290.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 11:59:47,014 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 11:59:54,486 INFO [train.py:904] (7/8) Epoch 22, batch 7150, loss[loss=0.202, simple_loss=0.2904, pruned_loss=0.0568, over 16706.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2903, pruned_loss=0.05826, over 3121015.48 frames. ], batch size: 62, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 11:59:58,136 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.701e+02 3.416e+02 4.420e+02 1.024e+03, threshold=6.833e+02, percent-clipped=4.0 2023-05-01 12:00:09,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3456, 5.6319, 5.3907, 5.4129, 5.0651, 4.9887, 5.0727, 5.7364], device='cuda:7'), covar=tensor([0.1254, 0.0836, 0.1126, 0.0834, 0.0870, 0.0812, 0.1177, 0.0926], device='cuda:7'), in_proj_covar=tensor([0.0673, 0.0819, 0.0679, 0.0625, 0.0518, 0.0530, 0.0686, 0.0642], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:00:20,953 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220319.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:00:55,330 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 12:01:07,951 INFO [train.py:904] (7/8) Epoch 22, batch 7200, loss[loss=0.169, simple_loss=0.263, pruned_loss=0.03751, over 17217.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2883, pruned_loss=0.05674, over 3115355.17 frames. ], batch size: 45, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:01:40,680 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-05-01 12:02:28,583 INFO [train.py:904] (7/8) Epoch 22, batch 7250, loss[loss=0.1709, simple_loss=0.2616, pruned_loss=0.04011, over 16867.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2866, pruned_loss=0.0562, over 3095854.66 frames. ], batch size: 102, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:02:30,896 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.548e+02 3.054e+02 3.716e+02 9.083e+02, threshold=6.108e+02, percent-clipped=2.0 2023-05-01 12:02:53,111 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220419.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:03:30,329 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220443.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:03:45,057 INFO [train.py:904] (7/8) Epoch 22, batch 7300, loss[loss=0.2133, simple_loss=0.3078, pruned_loss=0.0594, over 16860.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2871, pruned_loss=0.05681, over 3084169.45 frames. ], batch size: 109, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:03:47,416 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220454.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:04:07,596 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220467.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:04:24,498 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-01 12:05:02,412 INFO [train.py:904] (7/8) Epoch 22, batch 7350, loss[loss=0.2066, simple_loss=0.2912, pruned_loss=0.061, over 15350.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2887, pruned_loss=0.05819, over 3055067.82 frames. ], batch size: 190, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:05:05,572 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.713e+02 3.260e+02 3.989e+02 1.125e+03, threshold=6.520e+02, percent-clipped=5.0 2023-05-01 12:05:18,398 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220513.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:06:20,498 INFO [train.py:904] (7/8) Epoch 22, batch 7400, loss[loss=0.2114, simple_loss=0.295, pruned_loss=0.0639, over 15212.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.289, pruned_loss=0.05832, over 3064995.93 frames. ], batch size: 190, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:06:25,423 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220555.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:06:35,403 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220561.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:07:13,092 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220585.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:07:40,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0094, 3.4318, 3.5173, 2.2744, 3.2494, 3.5304, 3.2830, 2.0142], device='cuda:7'), covar=tensor([0.0588, 0.0077, 0.0063, 0.0434, 0.0108, 0.0112, 0.0100, 0.0499], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0083, 0.0084, 0.0133, 0.0098, 0.0109, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 12:07:41,490 INFO [train.py:904] (7/8) Epoch 22, batch 7450, loss[loss=0.2071, simple_loss=0.3056, pruned_loss=0.05431, over 16412.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2902, pruned_loss=0.05892, over 3078387.30 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:07:43,948 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.745e+02 3.256e+02 3.859e+02 7.938e+02, threshold=6.512e+02, percent-clipped=1.0 2023-05-01 12:07:46,447 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1477, 5.1711, 5.5873, 5.5470, 5.5891, 5.2060, 5.1371, 4.8319], device='cuda:7'), covar=tensor([0.0311, 0.0467, 0.0319, 0.0392, 0.0419, 0.0351, 0.0955, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0451, 0.0436, 0.0405, 0.0484, 0.0461, 0.0546, 0.0369], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 12:08:01,238 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220614.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:08:04,664 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220616.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 12:08:47,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1073, 5.3751, 5.1152, 5.1453, 4.8903, 4.7633, 4.7682, 5.4567], device='cuda:7'), covar=tensor([0.1167, 0.0868, 0.1064, 0.0964, 0.0845, 0.1054, 0.1232, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0668, 0.0814, 0.0675, 0.0620, 0.0515, 0.0528, 0.0682, 0.0638], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:09:01,337 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220651.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:09:03,157 INFO [train.py:904] (7/8) Epoch 22, batch 7500, loss[loss=0.1854, simple_loss=0.2613, pruned_loss=0.0547, over 17115.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2905, pruned_loss=0.05837, over 3089010.37 frames. ], batch size: 47, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:10:21,156 INFO [train.py:904] (7/8) Epoch 22, batch 7550, loss[loss=0.2371, simple_loss=0.2979, pruned_loss=0.0881, over 11261.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2901, pruned_loss=0.0591, over 3066746.71 frames. ], batch size: 248, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:10:24,493 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.979e+02 2.918e+02 3.506e+02 4.599e+02 8.060e+02, threshold=7.013e+02, percent-clipped=6.0 2023-05-01 12:10:25,036 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8414, 5.1043, 4.9005, 4.8788, 4.6193, 4.5654, 4.5446, 5.2037], device='cuda:7'), covar=tensor([0.1180, 0.0838, 0.1014, 0.0908, 0.0830, 0.0995, 0.1186, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0670, 0.0818, 0.0676, 0.0622, 0.0516, 0.0530, 0.0684, 0.0640], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:10:36,076 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220712.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:11:00,950 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.52 vs. limit=2.0 2023-05-01 12:11:23,709 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220743.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:11:38,535 INFO [train.py:904] (7/8) Epoch 22, batch 7600, loss[loss=0.203, simple_loss=0.2929, pruned_loss=0.05653, over 16745.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2891, pruned_loss=0.05911, over 3069754.90 frames. ], batch size: 124, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:11:40,749 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220754.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:12:37,571 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220791.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:12:55,481 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220802.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:12:56,375 INFO [train.py:904] (7/8) Epoch 22, batch 7650, loss[loss=0.1965, simple_loss=0.2864, pruned_loss=0.05326, over 16488.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2889, pruned_loss=0.05896, over 3090633.77 frames. ], batch size: 75, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:12:59,173 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.956e+02 2.818e+02 3.454e+02 4.152e+02 9.180e+02, threshold=6.907e+02, percent-clipped=1.0 2023-05-01 12:13:27,123 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220822.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:13:54,462 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 12:14:11,869 INFO [train.py:904] (7/8) Epoch 22, batch 7700, loss[loss=0.1834, simple_loss=0.2736, pruned_loss=0.04664, over 16392.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2886, pruned_loss=0.05939, over 3086645.24 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:14:20,471 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5722, 4.4785, 4.6290, 4.7868, 4.9785, 4.4627, 4.9597, 4.9748], device='cuda:7'), covar=tensor([0.1957, 0.1140, 0.1564, 0.0803, 0.0665, 0.1130, 0.0692, 0.0671], device='cuda:7'), in_proj_covar=tensor([0.0626, 0.0773, 0.0895, 0.0782, 0.0596, 0.0621, 0.0649, 0.0746], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:14:56,186 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 12:14:58,294 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220883.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:15:01,883 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220885.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:15:29,149 INFO [train.py:904] (7/8) Epoch 22, batch 7750, loss[loss=0.202, simple_loss=0.2953, pruned_loss=0.05438, over 16443.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.289, pruned_loss=0.05965, over 3070945.06 frames. ], batch size: 68, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:15:30,888 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=220904.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:15:32,047 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.706e+02 3.407e+02 4.136e+02 9.876e+02, threshold=6.815e+02, percent-clipped=5.0 2023-05-01 12:15:40,154 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=220911.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 12:15:45,823 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=220914.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:16:15,123 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220933.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:16:42,231 INFO [train.py:904] (7/8) Epoch 22, batch 7800, loss[loss=0.2369, simple_loss=0.3184, pruned_loss=0.07766, over 15400.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2896, pruned_loss=0.05991, over 3075311.86 frames. ], batch size: 190, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:16:56,581 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=220962.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:17:00,352 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=220965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:17:13,362 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4811, 3.4412, 2.6782, 2.1347, 2.2161, 2.2290, 3.5642, 3.1226], device='cuda:7'), covar=tensor([0.2998, 0.0653, 0.1943, 0.2917, 0.2679, 0.2326, 0.0517, 0.1375], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0267, 0.0304, 0.0313, 0.0297, 0.0259, 0.0295, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 12:17:55,731 INFO [train.py:904] (7/8) Epoch 22, batch 7850, loss[loss=0.1808, simple_loss=0.2775, pruned_loss=0.04211, over 16487.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.29, pruned_loss=0.05966, over 3054080.17 frames. ], batch size: 68, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:17:58,015 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.766e+02 3.554e+02 4.297e+02 6.446e+02, threshold=7.108e+02, percent-clipped=0.0 2023-05-01 12:18:02,919 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221007.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:18:49,454 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8415, 2.7141, 2.8275, 2.1094, 2.6660, 2.1560, 2.7173, 2.9394], device='cuda:7'), covar=tensor([0.0282, 0.0823, 0.0503, 0.1801, 0.0818, 0.0941, 0.0574, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0164, 0.0167, 0.0154, 0.0145, 0.0130, 0.0143, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:19:08,777 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4070, 2.1028, 1.7539, 1.8528, 2.4283, 2.0987, 2.1725, 2.5483], device='cuda:7'), covar=tensor([0.0218, 0.0410, 0.0538, 0.0484, 0.0278, 0.0382, 0.0234, 0.0276], device='cuda:7'), in_proj_covar=tensor([0.0205, 0.0231, 0.0223, 0.0222, 0.0233, 0.0230, 0.0231, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:19:09,448 INFO [train.py:904] (7/8) Epoch 22, batch 7900, loss[loss=0.2124, simple_loss=0.3007, pruned_loss=0.06208, over 16225.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2895, pruned_loss=0.05886, over 3068100.83 frames. ], batch size: 165, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:19:10,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6179, 2.5985, 1.9556, 2.6996, 2.1843, 2.7894, 2.1067, 2.3607], device='cuda:7'), covar=tensor([0.0307, 0.0382, 0.1243, 0.0254, 0.0666, 0.0574, 0.1196, 0.0587], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0175, 0.0193, 0.0160, 0.0175, 0.0215, 0.0201, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:19:14,081 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0855, 2.4231, 2.6364, 1.9383, 2.7097, 2.8186, 2.4109, 2.3780], device='cuda:7'), covar=tensor([0.0702, 0.0254, 0.0226, 0.0945, 0.0123, 0.0284, 0.0496, 0.0450], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0107, 0.0097, 0.0138, 0.0080, 0.0124, 0.0128, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:19:38,593 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2246, 4.2075, 4.0987, 3.3005, 4.1466, 1.6208, 3.9485, 3.7717], device='cuda:7'), covar=tensor([0.0127, 0.0113, 0.0195, 0.0374, 0.0109, 0.2914, 0.0138, 0.0266], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0154, 0.0198, 0.0178, 0.0173, 0.0207, 0.0186, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:19:54,321 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4847, 5.4446, 5.1458, 4.4710, 5.2808, 2.0606, 5.0635, 5.0744], device='cuda:7'), covar=tensor([0.0083, 0.0086, 0.0228, 0.0469, 0.0106, 0.2626, 0.0130, 0.0198], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0154, 0.0198, 0.0178, 0.0174, 0.0207, 0.0186, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:20:27,110 INFO [train.py:904] (7/8) Epoch 22, batch 7950, loss[loss=0.1848, simple_loss=0.2672, pruned_loss=0.05121, over 17111.00 frames. ], tot_loss[loss=0.2042, simple_loss=0.2896, pruned_loss=0.05941, over 3051972.64 frames. ], batch size: 47, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:20:32,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 2.674e+02 3.039e+02 3.526e+02 5.687e+02, threshold=6.078e+02, percent-clipped=0.0 2023-05-01 12:21:30,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7406, 2.5279, 2.3058, 3.5816, 2.4017, 3.7922, 1.4463, 2.7071], device='cuda:7'), covar=tensor([0.1363, 0.0824, 0.1295, 0.0226, 0.0196, 0.0398, 0.1803, 0.0845], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0190, 0.0207, 0.0214, 0.0202, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:21:41,835 INFO [train.py:904] (7/8) Epoch 22, batch 8000, loss[loss=0.1907, simple_loss=0.2789, pruned_loss=0.05126, over 16896.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2905, pruned_loss=0.06063, over 3040005.35 frames. ], batch size: 96, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:22:16,131 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2451, 4.8608, 4.7221, 3.4468, 4.1587, 4.7704, 3.9884, 2.9670], device='cuda:7'), covar=tensor([0.0415, 0.0042, 0.0040, 0.0311, 0.0098, 0.0080, 0.0094, 0.0371], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0084, 0.0084, 0.0134, 0.0098, 0.0111, 0.0095, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 12:22:19,288 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221178.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:22:41,485 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6571, 2.3107, 2.2681, 3.4002, 2.2115, 3.7585, 1.5098, 2.6630], device='cuda:7'), covar=tensor([0.1419, 0.0956, 0.1347, 0.0237, 0.0198, 0.0405, 0.1779, 0.0876], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0174, 0.0194, 0.0189, 0.0207, 0.0214, 0.0202, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:22:42,582 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5938, 4.6507, 4.4634, 4.1502, 4.1369, 4.5698, 4.2916, 4.2544], device='cuda:7'), covar=tensor([0.0694, 0.0717, 0.0342, 0.0355, 0.1005, 0.0585, 0.0484, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0287, 0.0423, 0.0337, 0.0333, 0.0342, 0.0388, 0.0233, 0.0403], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:22:55,175 INFO [train.py:904] (7/8) Epoch 22, batch 8050, loss[loss=0.2133, simple_loss=0.3045, pruned_loss=0.06103, over 16391.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2907, pruned_loss=0.06049, over 3030379.73 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:23:01,254 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.760e+02 2.816e+02 3.381e+02 4.131e+02 1.220e+03, threshold=6.763e+02, percent-clipped=7.0 2023-05-01 12:23:07,943 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:23:11,881 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 12:23:23,694 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-05-01 12:24:08,792 INFO [train.py:904] (7/8) Epoch 22, batch 8100, loss[loss=0.206, simple_loss=0.291, pruned_loss=0.06052, over 16914.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2905, pruned_loss=0.05983, over 3056180.05 frames. ], batch size: 109, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:24:18,137 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:24:20,008 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221260.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:25:11,710 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 12:25:16,914 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5514, 3.5641, 2.8034, 2.2596, 2.3807, 2.3690, 3.8365, 3.2337], device='cuda:7'), covar=tensor([0.3117, 0.0654, 0.1833, 0.2646, 0.2681, 0.2189, 0.0485, 0.1411], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0267, 0.0304, 0.0313, 0.0297, 0.0260, 0.0295, 0.0336], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 12:25:24,348 INFO [train.py:904] (7/8) Epoch 22, batch 8150, loss[loss=0.1931, simple_loss=0.2822, pruned_loss=0.05205, over 15475.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2873, pruned_loss=0.05818, over 3079080.71 frames. ], batch size: 190, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:25:31,012 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.655e+02 3.146e+02 3.866e+02 6.459e+02, threshold=6.292e+02, percent-clipped=0.0 2023-05-01 12:25:32,109 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221307.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:25:34,349 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 12:26:41,094 INFO [train.py:904] (7/8) Epoch 22, batch 8200, loss[loss=0.199, simple_loss=0.2924, pruned_loss=0.05277, over 16430.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2846, pruned_loss=0.05704, over 3109271.39 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:26:44,117 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221355.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:27:58,681 INFO [train.py:904] (7/8) Epoch 22, batch 8250, loss[loss=0.1939, simple_loss=0.2876, pruned_loss=0.05006, over 16401.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2845, pruned_loss=0.0548, over 3098126.93 frames. ], batch size: 146, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:28:05,606 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.638e+02 3.060e+02 3.657e+02 7.056e+02, threshold=6.120e+02, percent-clipped=1.0 2023-05-01 12:29:17,817 INFO [train.py:904] (7/8) Epoch 22, batch 8300, loss[loss=0.1993, simple_loss=0.2931, pruned_loss=0.05278, over 16891.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2822, pruned_loss=0.05213, over 3083239.39 frames. ], batch size: 116, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:29:57,616 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221478.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:30:38,305 INFO [train.py:904] (7/8) Epoch 22, batch 8350, loss[loss=0.2207, simple_loss=0.2913, pruned_loss=0.07504, over 11925.00 frames. ], tot_loss[loss=0.1911, simple_loss=0.2817, pruned_loss=0.05027, over 3075840.16 frames. ], batch size: 248, lr: 3.05e-03, grad_scale: 4.0 2023-05-01 12:30:43,700 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.123e+02 2.436e+02 3.235e+02 5.612e+02, threshold=4.873e+02, percent-clipped=0.0 2023-05-01 12:30:44,337 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=221507.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:30:54,329 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 12:31:11,610 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 12:31:15,670 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221526.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:31:56,143 INFO [train.py:904] (7/8) Epoch 22, batch 8400, loss[loss=0.1666, simple_loss=0.2603, pruned_loss=0.03644, over 16545.00 frames. ], tot_loss[loss=0.1882, simple_loss=0.2794, pruned_loss=0.0485, over 3088469.65 frames. ], batch size: 62, lr: 3.05e-03, grad_scale: 8.0 2023-05-01 12:32:09,042 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=221560.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:32:21,854 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=221568.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:33:17,384 INFO [train.py:904] (7/8) Epoch 22, batch 8450, loss[loss=0.155, simple_loss=0.255, pruned_loss=0.02756, over 16799.00 frames. ], tot_loss[loss=0.1859, simple_loss=0.2778, pruned_loss=0.04697, over 3073247.78 frames. ], batch size: 83, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:33:24,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.610e+02 2.296e+02 2.720e+02 3.439e+02 5.542e+02, threshold=5.440e+02, percent-clipped=2.0 2023-05-01 12:33:26,674 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=221608.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:34:02,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7746, 4.0848, 3.1438, 2.2959, 2.4245, 2.5710, 4.3320, 3.4361], device='cuda:7'), covar=tensor([0.2849, 0.0448, 0.1638, 0.3004, 0.3207, 0.2112, 0.0314, 0.1305], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0263, 0.0300, 0.0309, 0.0292, 0.0256, 0.0291, 0.0330], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 12:34:38,844 INFO [train.py:904] (7/8) Epoch 22, batch 8500, loss[loss=0.174, simple_loss=0.2651, pruned_loss=0.04146, over 16847.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2741, pruned_loss=0.04487, over 3071030.82 frames. ], batch size: 116, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:34:45,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8651, 3.7564, 3.8551, 4.0991, 4.1801, 3.7677, 4.1974, 4.1893], device='cuda:7'), covar=tensor([0.2032, 0.1416, 0.1876, 0.0880, 0.0834, 0.1828, 0.0830, 0.1030], device='cuda:7'), in_proj_covar=tensor([0.0622, 0.0768, 0.0891, 0.0775, 0.0592, 0.0620, 0.0643, 0.0748], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:34:55,033 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-05-01 12:36:02,541 INFO [train.py:904] (7/8) Epoch 22, batch 8550, loss[loss=0.1787, simple_loss=0.2775, pruned_loss=0.03992, over 16783.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2719, pruned_loss=0.04421, over 3033474.04 frames. ], batch size: 124, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:36:07,341 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.87 vs. limit=5.0 2023-05-01 12:36:10,061 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.073e+02 2.527e+02 2.950e+02 5.682e+02, threshold=5.053e+02, percent-clipped=1.0 2023-05-01 12:36:16,359 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 12:37:41,424 INFO [train.py:904] (7/8) Epoch 22, batch 8600, loss[loss=0.1871, simple_loss=0.2888, pruned_loss=0.04267, over 15219.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2718, pruned_loss=0.04298, over 3038257.38 frames. ], batch size: 190, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:37:53,303 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=221758.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:38:11,436 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-01 12:38:33,400 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7836, 4.5948, 4.8720, 5.0070, 5.1925, 4.6089, 5.1801, 5.1711], device='cuda:7'), covar=tensor([0.1899, 0.1256, 0.1519, 0.0726, 0.0488, 0.0944, 0.0463, 0.0605], device='cuda:7'), in_proj_covar=tensor([0.0618, 0.0763, 0.0884, 0.0771, 0.0589, 0.0615, 0.0640, 0.0742], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:39:07,370 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4029, 5.7508, 5.5145, 5.5825, 5.2206, 5.2061, 5.1251, 5.8813], device='cuda:7'), covar=tensor([0.1203, 0.0836, 0.0998, 0.0843, 0.0806, 0.0694, 0.1228, 0.0827], device='cuda:7'), in_proj_covar=tensor([0.0657, 0.0800, 0.0662, 0.0607, 0.0503, 0.0518, 0.0668, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:39:21,622 INFO [train.py:904] (7/8) Epoch 22, batch 8650, loss[loss=0.1613, simple_loss=0.2603, pruned_loss=0.03114, over 15408.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2695, pruned_loss=0.0415, over 3039504.74 frames. ], batch size: 190, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:39:30,561 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.228e+02 2.506e+02 3.024e+02 6.242e+02, threshold=5.012e+02, percent-clipped=2.0 2023-05-01 12:39:58,994 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=221819.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:41:04,026 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9946, 4.2665, 4.1283, 4.1499, 3.8153, 3.8738, 3.9043, 4.2750], device='cuda:7'), covar=tensor([0.1159, 0.1008, 0.0957, 0.0821, 0.0835, 0.1777, 0.0958, 0.1015], device='cuda:7'), in_proj_covar=tensor([0.0657, 0.0799, 0.0661, 0.0607, 0.0503, 0.0517, 0.0668, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:41:06,656 INFO [train.py:904] (7/8) Epoch 22, batch 8700, loss[loss=0.1675, simple_loss=0.2592, pruned_loss=0.03789, over 16728.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2669, pruned_loss=0.04059, over 3037665.91 frames. ], batch size: 134, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:41:24,450 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 12:41:27,842 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=221863.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:42:16,873 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=221890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:42:42,867 INFO [train.py:904] (7/8) Epoch 22, batch 8750, loss[loss=0.1799, simple_loss=0.282, pruned_loss=0.03885, over 16909.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2667, pruned_loss=0.0401, over 3049588.65 frames. ], batch size: 116, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:42:53,176 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.058e+02 2.561e+02 3.126e+02 5.404e+02, threshold=5.122e+02, percent-clipped=2.0 2023-05-01 12:42:54,686 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0952, 4.0681, 3.9407, 3.2440, 4.0083, 1.7703, 3.7984, 3.5903], device='cuda:7'), covar=tensor([0.0090, 0.0088, 0.0169, 0.0247, 0.0093, 0.2751, 0.0118, 0.0230], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0151, 0.0193, 0.0172, 0.0170, 0.0203, 0.0182, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:44:31,277 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=221951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:44:34,621 INFO [train.py:904] (7/8) Epoch 22, batch 8800, loss[loss=0.1711, simple_loss=0.2701, pruned_loss=0.03601, over 16753.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2647, pruned_loss=0.03853, over 3061037.60 frames. ], batch size: 124, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:44:46,849 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8497, 3.0742, 2.7709, 4.9388, 3.5978, 4.4359, 1.6617, 3.1681], device='cuda:7'), covar=tensor([0.1427, 0.0748, 0.1115, 0.0142, 0.0197, 0.0366, 0.1749, 0.0719], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0172, 0.0192, 0.0185, 0.0202, 0.0211, 0.0200, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:46:22,311 INFO [train.py:904] (7/8) Epoch 22, batch 8850, loss[loss=0.1541, simple_loss=0.2489, pruned_loss=0.0297, over 12668.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2669, pruned_loss=0.03822, over 3044807.93 frames. ], batch size: 247, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:46:28,904 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.177e+02 2.540e+02 3.054e+02 9.241e+02, threshold=5.079e+02, percent-clipped=4.0 2023-05-01 12:46:35,814 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222009.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:46:56,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9334, 2.1803, 2.3226, 3.0202, 1.7731, 3.2732, 1.6817, 2.7015], device='cuda:7'), covar=tensor([0.1217, 0.0752, 0.1088, 0.0166, 0.0081, 0.0346, 0.1569, 0.0710], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0172, 0.0192, 0.0184, 0.0202, 0.0211, 0.0200, 0.0190], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:46:59,554 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4597, 4.4203, 4.7831, 4.7650, 4.7893, 4.4968, 4.4396, 4.3703], device='cuda:7'), covar=tensor([0.0298, 0.0521, 0.0458, 0.0424, 0.0477, 0.0363, 0.0967, 0.0452], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0443, 0.0430, 0.0397, 0.0475, 0.0450, 0.0531, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 12:48:07,873 INFO [train.py:904] (7/8) Epoch 22, batch 8900, loss[loss=0.1561, simple_loss=0.2476, pruned_loss=0.03232, over 12982.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2681, pruned_loss=0.03777, over 3069317.30 frames. ], batch size: 248, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:48:35,346 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2023-05-01 12:48:43,918 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222070.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:48:45,845 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4762, 2.3845, 2.8548, 3.3786, 3.0027, 3.8485, 2.6734, 3.7779], device='cuda:7'), covar=tensor([0.0147, 0.0475, 0.0313, 0.0239, 0.0306, 0.0143, 0.0436, 0.0149], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0188, 0.0174, 0.0178, 0.0191, 0.0148, 0.0191, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 12:49:24,699 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 12:50:04,406 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1497, 3.3412, 3.7009, 2.2575, 3.0884, 2.4298, 3.6519, 3.4693], device='cuda:7'), covar=tensor([0.0238, 0.0819, 0.0504, 0.1910, 0.0743, 0.0884, 0.0599, 0.1023], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0158, 0.0162, 0.0150, 0.0141, 0.0126, 0.0139, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:50:11,691 INFO [train.py:904] (7/8) Epoch 22, batch 8950, loss[loss=0.1515, simple_loss=0.2501, pruned_loss=0.02646, over 16749.00 frames. ], tot_loss[loss=0.1721, simple_loss=0.2678, pruned_loss=0.03822, over 3063498.92 frames. ], batch size: 83, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:50:23,616 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.055e+02 2.537e+02 3.242e+02 5.018e+02, threshold=5.074e+02, percent-clipped=0.0 2023-05-01 12:50:37,677 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222114.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:52:03,810 INFO [train.py:904] (7/8) Epoch 22, batch 9000, loss[loss=0.1688, simple_loss=0.2528, pruned_loss=0.04242, over 11976.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2648, pruned_loss=0.03724, over 3043554.94 frames. ], batch size: 246, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:52:03,811 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 12:52:14,710 INFO [train.py:938] (7/8) Epoch 22, validation: loss=0.1453, simple_loss=0.2492, pruned_loss=0.0207, over 944034.00 frames. 2023-05-01 12:52:14,710 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 12:52:36,562 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222163.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:53:22,605 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222185.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:53:58,641 INFO [train.py:904] (7/8) Epoch 22, batch 9050, loss[loss=0.1678, simple_loss=0.2563, pruned_loss=0.03968, over 12662.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2661, pruned_loss=0.03785, over 3059341.03 frames. ], batch size: 246, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:54:04,293 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222205.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:54:09,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.330e+02 2.762e+02 3.325e+02 5.542e+02, threshold=5.524e+02, percent-clipped=4.0 2023-05-01 12:54:13,531 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 12:54:17,178 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:55:31,427 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222246.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:55:31,627 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222246.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 12:55:44,669 INFO [train.py:904] (7/8) Epoch 22, batch 9100, loss[loss=0.1701, simple_loss=0.2716, pruned_loss=0.03431, over 16185.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2658, pruned_loss=0.03842, over 3054096.85 frames. ], batch size: 165, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:56:11,381 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222266.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 12:56:48,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4205, 3.0286, 2.7191, 2.2436, 2.1567, 2.2905, 3.0109, 2.8070], device='cuda:7'), covar=tensor([0.2664, 0.0671, 0.1573, 0.2572, 0.2634, 0.2152, 0.0485, 0.1408], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0260, 0.0298, 0.0306, 0.0286, 0.0254, 0.0288, 0.0327], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 12:57:42,776 INFO [train.py:904] (7/8) Epoch 22, batch 9150, loss[loss=0.1576, simple_loss=0.2498, pruned_loss=0.03267, over 16990.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.266, pruned_loss=0.03784, over 3057779.23 frames. ], batch size: 55, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 12:57:53,767 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.620e+02 2.177e+02 2.613e+02 3.280e+02 4.905e+02, threshold=5.227e+02, percent-clipped=0.0 2023-05-01 12:58:12,974 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 12:59:04,642 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 12:59:27,010 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9360, 2.6350, 2.8991, 2.1684, 2.6429, 2.1430, 2.7074, 2.8432], device='cuda:7'), covar=tensor([0.0272, 0.1039, 0.0521, 0.1828, 0.0827, 0.0977, 0.0639, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0157, 0.0161, 0.0149, 0.0140, 0.0125, 0.0138, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 12:59:28,951 INFO [train.py:904] (7/8) Epoch 22, batch 9200, loss[loss=0.1786, simple_loss=0.2707, pruned_loss=0.04327, over 16675.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2621, pruned_loss=0.03707, over 3057099.59 frames. ], batch size: 134, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 12:59:52,296 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222365.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:01:05,896 INFO [train.py:904] (7/8) Epoch 22, batch 9250, loss[loss=0.151, simple_loss=0.2339, pruned_loss=0.03407, over 12285.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2615, pruned_loss=0.03674, over 3064981.13 frames. ], batch size: 247, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:01:07,658 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.11 vs. limit=5.0 2023-05-01 13:01:16,247 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.130e+02 2.525e+02 3.050e+02 5.911e+02, threshold=5.049e+02, percent-clipped=2.0 2023-05-01 13:01:28,887 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:02:56,979 INFO [train.py:904] (7/8) Epoch 22, batch 9300, loss[loss=0.1593, simple_loss=0.2511, pruned_loss=0.03378, over 16689.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2594, pruned_loss=0.03601, over 3073499.74 frames. ], batch size: 134, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:03:10,207 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 13:03:17,565 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222462.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:03:23,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2773, 2.2031, 2.1653, 3.9596, 2.1755, 2.5413, 2.3226, 2.3450], device='cuda:7'), covar=tensor([0.1210, 0.3691, 0.3254, 0.0480, 0.4230, 0.2704, 0.3699, 0.3571], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0438, 0.0361, 0.0318, 0.0428, 0.0504, 0.0411, 0.0511], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:03:36,686 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 13:03:43,393 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2869, 2.2246, 2.2788, 4.0275, 2.1574, 2.5410, 2.3318, 2.3774], device='cuda:7'), covar=tensor([0.1255, 0.3767, 0.3227, 0.0510, 0.4467, 0.2797, 0.3771, 0.3667], device='cuda:7'), in_proj_covar=tensor([0.0391, 0.0439, 0.0361, 0.0318, 0.0429, 0.0504, 0.0411, 0.0512], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:04:40,779 INFO [train.py:904] (7/8) Epoch 22, batch 9350, loss[loss=0.1817, simple_loss=0.2731, pruned_loss=0.04513, over 16175.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2592, pruned_loss=0.03595, over 3075937.80 frames. ], batch size: 165, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:04:43,298 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3196, 4.3588, 4.6778, 4.6745, 4.6690, 4.4576, 4.4083, 4.3547], device='cuda:7'), covar=tensor([0.0471, 0.1346, 0.0619, 0.0626, 0.0562, 0.0673, 0.0996, 0.0617], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0441, 0.0428, 0.0397, 0.0473, 0.0450, 0.0530, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:04:49,916 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 1.994e+02 2.436e+02 3.069e+02 5.740e+02, threshold=4.871e+02, percent-clipped=2.0 2023-05-01 13:05:23,317 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222523.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:05:58,164 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222541.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 13:06:07,173 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9270, 2.2231, 2.3080, 2.9497, 1.8079, 3.2650, 1.7306, 2.7257], device='cuda:7'), covar=tensor([0.1261, 0.0750, 0.1087, 0.0139, 0.0071, 0.0321, 0.1562, 0.0729], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0172, 0.0193, 0.0183, 0.0200, 0.0212, 0.0201, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 13:06:08,609 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:06:20,191 INFO [train.py:904] (7/8) Epoch 22, batch 9400, loss[loss=0.1558, simple_loss=0.2645, pruned_loss=0.02352, over 16851.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2587, pruned_loss=0.03558, over 3063219.72 frames. ], batch size: 96, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:06:38,190 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222561.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:06:55,504 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222570.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:07:24,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3735, 3.4790, 3.6730, 3.6481, 3.6608, 3.4887, 3.5314, 3.5479], device='cuda:7'), covar=tensor([0.0468, 0.0945, 0.0732, 0.0847, 0.0838, 0.1000, 0.0908, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0437, 0.0424, 0.0394, 0.0470, 0.0446, 0.0526, 0.0359], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:07:24,493 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222584.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 13:07:43,456 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:08:00,588 INFO [train.py:904] (7/8) Epoch 22, batch 9450, loss[loss=0.1559, simple_loss=0.2514, pruned_loss=0.03019, over 16412.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.26, pruned_loss=0.03581, over 3042286.78 frames. ], batch size: 68, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:08:08,417 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.238e+02 2.730e+02 3.189e+02 9.782e+02, threshold=5.460e+02, percent-clipped=5.0 2023-05-01 13:08:26,480 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5303, 3.5954, 2.1688, 4.1035, 2.7193, 3.9871, 2.4068, 2.9414], device='cuda:7'), covar=tensor([0.0332, 0.0439, 0.1754, 0.0261, 0.0834, 0.0657, 0.1554, 0.0833], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0171, 0.0188, 0.0156, 0.0172, 0.0208, 0.0198, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 13:08:56,855 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=222631.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:09:11,554 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1967, 5.2606, 5.6614, 5.6177, 5.6376, 5.3527, 5.1958, 5.0838], device='cuda:7'), covar=tensor([0.0355, 0.0719, 0.0543, 0.0677, 0.0654, 0.0609, 0.0981, 0.0479], device='cuda:7'), in_proj_covar=tensor([0.0395, 0.0438, 0.0426, 0.0395, 0.0471, 0.0447, 0.0527, 0.0359], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:09:40,759 INFO [train.py:904] (7/8) Epoch 22, batch 9500, loss[loss=0.1604, simple_loss=0.253, pruned_loss=0.03391, over 16688.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.259, pruned_loss=0.03522, over 3049441.64 frames. ], batch size: 134, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:10:06,952 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222665.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:11:22,397 INFO [train.py:904] (7/8) Epoch 22, batch 9550, loss[loss=0.1856, simple_loss=0.2813, pruned_loss=0.04494, over 16783.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2589, pruned_loss=0.03559, over 3037000.30 frames. ], batch size: 124, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:11:34,508 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.117e+02 2.369e+02 3.001e+02 4.954e+02, threshold=4.737e+02, percent-clipped=0.0 2023-05-01 13:11:44,580 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222713.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:12:03,471 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6263, 4.8045, 4.9225, 4.7418, 4.8070, 5.3050, 4.7996, 4.4888], device='cuda:7'), covar=tensor([0.1062, 0.1875, 0.2203, 0.1950, 0.2279, 0.0898, 0.1542, 0.2393], device='cuda:7'), in_proj_covar=tensor([0.0387, 0.0563, 0.0625, 0.0465, 0.0620, 0.0651, 0.0493, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 13:12:46,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1529, 3.4534, 3.5884, 3.5627, 3.5806, 3.4530, 3.2326, 3.4998], device='cuda:7'), covar=tensor([0.0669, 0.0949, 0.0717, 0.0808, 0.0855, 0.0819, 0.1389, 0.0699], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0438, 0.0426, 0.0394, 0.0471, 0.0447, 0.0528, 0.0359], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:13:00,669 INFO [train.py:904] (7/8) Epoch 22, batch 9600, loss[loss=0.1722, simple_loss=0.2775, pruned_loss=0.03346, over 16882.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2608, pruned_loss=0.03633, over 3038569.67 frames. ], batch size: 102, lr: 3.04e-03, grad_scale: 8.0 2023-05-01 13:13:04,215 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0531, 3.4330, 3.6426, 2.1255, 2.9722, 2.2864, 3.4963, 3.5145], device='cuda:7'), covar=tensor([0.0255, 0.0779, 0.0554, 0.2045, 0.0868, 0.1040, 0.0605, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0157, 0.0162, 0.0150, 0.0141, 0.0126, 0.0139, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 13:13:17,812 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 13:13:47,336 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5000, 4.0856, 4.1214, 2.6946, 3.6224, 4.1197, 3.8129, 2.5571], device='cuda:7'), covar=tensor([0.0541, 0.0046, 0.0046, 0.0423, 0.0102, 0.0086, 0.0070, 0.0444], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0081, 0.0082, 0.0131, 0.0095, 0.0106, 0.0092, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 13:14:44,335 INFO [train.py:904] (7/8) Epoch 22, batch 9650, loss[loss=0.1653, simple_loss=0.2549, pruned_loss=0.03786, over 12071.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2625, pruned_loss=0.03636, over 3053850.71 frames. ], batch size: 248, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:14:58,749 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.257e+02 2.597e+02 3.466e+02 1.012e+03, threshold=5.195e+02, percent-clipped=6.0 2023-05-01 13:15:21,353 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9161, 1.2930, 1.6913, 1.7226, 1.9036, 1.9780, 1.5018, 1.8792], device='cuda:7'), covar=tensor([0.0267, 0.0492, 0.0261, 0.0308, 0.0307, 0.0217, 0.0604, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0186, 0.0174, 0.0177, 0.0190, 0.0147, 0.0190, 0.0145], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:15:52,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1409, 3.2145, 1.9423, 3.4865, 2.3312, 3.4517, 2.0634, 2.6717], device='cuda:7'), covar=tensor([0.0335, 0.0406, 0.1713, 0.0264, 0.0885, 0.0606, 0.1730, 0.0771], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0171, 0.0187, 0.0155, 0.0172, 0.0207, 0.0198, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:16:03,821 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222841.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 13:16:27,398 INFO [train.py:904] (7/8) Epoch 22, batch 9700, loss[loss=0.1589, simple_loss=0.2573, pruned_loss=0.03029, over 16742.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2622, pruned_loss=0.03645, over 3076518.15 frames. ], batch size: 83, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:16:43,332 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=222861.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:17:16,362 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1253, 1.5862, 1.9284, 2.0910, 2.2396, 2.3096, 1.7940, 2.1707], device='cuda:7'), covar=tensor([0.0253, 0.0488, 0.0326, 0.0351, 0.0354, 0.0227, 0.0557, 0.0158], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0187, 0.0174, 0.0177, 0.0191, 0.0147, 0.0191, 0.0146], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:17:23,134 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222879.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 13:17:42,992 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222889.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:18:08,542 INFO [train.py:904] (7/8) Epoch 22, batch 9750, loss[loss=0.1215, simple_loss=0.2124, pruned_loss=0.01531, over 17133.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2609, pruned_loss=0.03626, over 3079166.86 frames. ], batch size: 47, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:18:18,160 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.037e+02 2.441e+02 3.104e+02 5.362e+02, threshold=4.883e+02, percent-clipped=3.0 2023-05-01 13:18:19,215 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=222909.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:18:52,783 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=222926.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:19:04,357 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-05-01 13:19:24,010 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8225, 3.6775, 3.8875, 3.9913, 4.0964, 3.6956, 4.0636, 4.1152], device='cuda:7'), covar=tensor([0.1574, 0.1225, 0.1254, 0.0647, 0.0532, 0.1829, 0.0677, 0.0682], device='cuda:7'), in_proj_covar=tensor([0.0608, 0.0753, 0.0872, 0.0760, 0.0578, 0.0605, 0.0632, 0.0730], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:19:24,050 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222940.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:19:45,661 INFO [train.py:904] (7/8) Epoch 22, batch 9800, loss[loss=0.1625, simple_loss=0.2478, pruned_loss=0.03856, over 12322.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2617, pruned_loss=0.03574, over 3087328.90 frames. ], batch size: 247, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:20:15,167 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222969.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:20:40,851 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=222984.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:21:23,940 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223001.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:21:26,487 INFO [train.py:904] (7/8) Epoch 22, batch 9850, loss[loss=0.1548, simple_loss=0.254, pruned_loss=0.02779, over 16859.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2623, pruned_loss=0.03529, over 3074968.21 frames. ], batch size: 102, lr: 3.04e-03, grad_scale: 4.0 2023-05-01 13:21:37,445 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.181e+02 2.565e+02 3.075e+02 7.070e+02, threshold=5.130e+02, percent-clipped=2.0 2023-05-01 13:21:50,299 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1209, 3.9943, 4.2173, 4.3432, 4.4838, 4.0299, 4.4693, 4.4967], device='cuda:7'), covar=tensor([0.1784, 0.1225, 0.1450, 0.0713, 0.0525, 0.1461, 0.0583, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0609, 0.0750, 0.0869, 0.0758, 0.0577, 0.0603, 0.0630, 0.0729], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:22:19,907 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223030.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:23:01,416 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223045.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:23:15,943 INFO [train.py:904] (7/8) Epoch 22, batch 9900, loss[loss=0.17, simple_loss=0.2702, pruned_loss=0.03489, over 16515.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2626, pruned_loss=0.03513, over 3071466.68 frames. ], batch size: 62, lr: 3.03e-03, grad_scale: 4.0 2023-05-01 13:23:27,678 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 13:23:44,385 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=223064.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:25:13,522 INFO [train.py:904] (7/8) Epoch 22, batch 9950, loss[loss=0.1639, simple_loss=0.268, pruned_loss=0.02991, over 16193.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2647, pruned_loss=0.0357, over 3064958.29 frames. ], batch size: 165, lr: 3.03e-03, grad_scale: 4.0 2023-05-01 13:25:27,902 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.209e+02 2.588e+02 3.009e+02 4.332e+02, threshold=5.177e+02, percent-clipped=0.0 2023-05-01 13:25:33,368 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-01 13:26:07,266 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223125.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:26:16,796 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 13:27:13,205 INFO [train.py:904] (7/8) Epoch 22, batch 10000, loss[loss=0.1656, simple_loss=0.2674, pruned_loss=0.03191, over 16846.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2631, pruned_loss=0.03519, over 3068988.79 frames. ], batch size: 90, lr: 3.03e-03, grad_scale: 8.0 2023-05-01 13:27:52,326 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=223173.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:28:05,113 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223179.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:28:54,099 INFO [train.py:904] (7/8) Epoch 22, batch 10050, loss[loss=0.1618, simple_loss=0.2579, pruned_loss=0.0329, over 12329.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2632, pruned_loss=0.03539, over 3048533.08 frames. ], batch size: 248, lr: 3.03e-03, grad_scale: 8.0 2023-05-01 13:29:04,331 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.078e+02 2.413e+02 2.769e+02 4.519e+02, threshold=4.827e+02, percent-clipped=0.0 2023-05-01 13:29:40,815 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223226.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:29:42,095 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223227.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:29:54,687 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223234.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:30:27,331 INFO [train.py:904] (7/8) Epoch 22, batch 10100, loss[loss=0.154, simple_loss=0.2433, pruned_loss=0.03232, over 15252.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2639, pruned_loss=0.03607, over 3032345.61 frames. ], batch size: 191, lr: 3.03e-03, grad_scale: 8.0 2023-05-01 13:31:10,810 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223274.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:31:26,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9253, 3.7874, 4.0468, 4.1434, 4.2475, 3.8330, 4.2480, 4.2714], device='cuda:7'), covar=tensor([0.1907, 0.1296, 0.1486, 0.0725, 0.0608, 0.1442, 0.0608, 0.0815], device='cuda:7'), in_proj_covar=tensor([0.0606, 0.0746, 0.0864, 0.0756, 0.0576, 0.0600, 0.0627, 0.0724], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:31:30,204 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-05-01 13:31:39,361 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223296.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:32:13,437 INFO [train.py:904] (7/8) Epoch 23, batch 0, loss[loss=0.1717, simple_loss=0.2625, pruned_loss=0.04047, over 17108.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2625, pruned_loss=0.04047, over 17108.00 frames. ], batch size: 47, lr: 2.97e-03, grad_scale: 8.0 2023-05-01 13:32:13,437 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 13:32:20,849 INFO [train.py:938] (7/8) Epoch 23, validation: loss=0.1454, simple_loss=0.2487, pruned_loss=0.02111, over 944034.00 frames. 2023-05-01 13:32:20,850 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 13:32:28,408 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.627e+02 3.041e+02 3.739e+02 7.225e+02, threshold=6.083e+02, percent-clipped=7.0 2023-05-01 13:32:40,846 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0082, 5.1469, 5.5113, 5.4584, 5.5558, 5.2512, 5.0301, 5.0166], device='cuda:7'), covar=tensor([0.0527, 0.0721, 0.0469, 0.0605, 0.0629, 0.0482, 0.1277, 0.0457], device='cuda:7'), in_proj_covar=tensor([0.0392, 0.0436, 0.0426, 0.0393, 0.0467, 0.0445, 0.0524, 0.0357], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:32:49,534 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223325.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:33:09,294 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:33:26,110 INFO [train.py:904] (7/8) Epoch 23, batch 50, loss[loss=0.1666, simple_loss=0.2574, pruned_loss=0.03791, over 15939.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2691, pruned_loss=0.04844, over 756921.49 frames. ], batch size: 35, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:34:32,077 INFO [train.py:904] (7/8) Epoch 23, batch 100, loss[loss=0.1441, simple_loss=0.2358, pruned_loss=0.02617, over 16999.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.263, pruned_loss=0.04534, over 1337420.33 frames. ], batch size: 41, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:34:35,425 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9132, 4.8426, 4.6188, 3.6257, 4.6529, 1.7437, 4.2770, 4.4617], device='cuda:7'), covar=tensor([0.0191, 0.0173, 0.0327, 0.0788, 0.0209, 0.3540, 0.0310, 0.0464], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0152, 0.0192, 0.0168, 0.0170, 0.0204, 0.0182, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:34:42,063 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.242e+02 2.764e+02 3.330e+02 6.883e+02, threshold=5.527e+02, percent-clipped=1.0 2023-05-01 13:34:55,446 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223420.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:35:38,701 INFO [train.py:904] (7/8) Epoch 23, batch 150, loss[loss=0.152, simple_loss=0.2378, pruned_loss=0.03314, over 16801.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2603, pruned_loss=0.04339, over 1784188.93 frames. ], batch size: 42, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:36:11,230 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 13:36:47,617 INFO [train.py:904] (7/8) Epoch 23, batch 200, loss[loss=0.2118, simple_loss=0.2962, pruned_loss=0.06374, over 16345.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2627, pruned_loss=0.04419, over 2127654.81 frames. ], batch size: 145, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:36:51,624 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5903, 2.3914, 2.4044, 4.4433, 2.4116, 2.7444, 2.4630, 2.5516], device='cuda:7'), covar=tensor([0.1173, 0.3735, 0.3015, 0.0454, 0.4080, 0.2592, 0.3399, 0.3662], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0446, 0.0368, 0.0323, 0.0435, 0.0511, 0.0418, 0.0520], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:36:57,905 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 2.279e+02 2.532e+02 3.033e+02 5.752e+02, threshold=5.065e+02, percent-clipped=1.0 2023-05-01 13:37:21,193 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223529.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 13:37:52,723 INFO [train.py:904] (7/8) Epoch 23, batch 250, loss[loss=0.1663, simple_loss=0.2527, pruned_loss=0.04, over 17163.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2618, pruned_loss=0.04433, over 2401596.48 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:37:54,527 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-01 13:38:29,070 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=223578.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:38:32,467 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8804, 2.8078, 2.7875, 4.7604, 3.8404, 4.2438, 1.8081, 3.1945], device='cuda:7'), covar=tensor([0.1345, 0.0811, 0.1123, 0.0220, 0.0221, 0.0414, 0.1582, 0.0768], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0174, 0.0194, 0.0186, 0.0200, 0.0214, 0.0203, 0.0192], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 13:38:54,798 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223596.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:39:04,058 INFO [train.py:904] (7/8) Epoch 23, batch 300, loss[loss=0.1564, simple_loss=0.249, pruned_loss=0.0319, over 17111.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2595, pruned_loss=0.04324, over 2610517.14 frames. ], batch size: 47, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:39:14,755 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.219e+02 2.672e+02 3.069e+02 6.824e+02, threshold=5.344e+02, percent-clipped=3.0 2023-05-01 13:39:35,518 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223625.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:39:51,604 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 13:39:54,077 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=223639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:39:55,298 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223640.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:40:01,752 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223644.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:40:13,662 INFO [train.py:904] (7/8) Epoch 23, batch 350, loss[loss=0.2054, simple_loss=0.277, pruned_loss=0.06686, over 12442.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2566, pruned_loss=0.04195, over 2764403.74 frames. ], batch size: 247, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:40:42,277 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223673.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:41:03,082 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223688.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:41:22,673 INFO [train.py:904] (7/8) Epoch 23, batch 400, loss[loss=0.1538, simple_loss=0.2571, pruned_loss=0.02526, over 17136.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2546, pruned_loss=0.04132, over 2891327.44 frames. ], batch size: 49, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:41:27,433 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3643, 5.6862, 5.4089, 5.4800, 5.0833, 5.1198, 5.0704, 5.7962], device='cuda:7'), covar=tensor([0.1427, 0.1023, 0.1190, 0.0893, 0.0936, 0.0792, 0.1340, 0.1045], device='cuda:7'), in_proj_covar=tensor([0.0671, 0.0815, 0.0671, 0.0618, 0.0514, 0.0526, 0.0685, 0.0636], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:41:34,908 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.389e+02 2.103e+02 2.400e+02 2.968e+02 1.328e+03, threshold=4.800e+02, percent-clipped=3.0 2023-05-01 13:41:47,195 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223720.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:41:52,327 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2023-05-01 13:41:54,625 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-05-01 13:42:32,063 INFO [train.py:904] (7/8) Epoch 23, batch 450, loss[loss=0.1482, simple_loss=0.2275, pruned_loss=0.03448, over 16750.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2528, pruned_loss=0.0401, over 2992354.74 frames. ], batch size: 83, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:42:52,162 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223768.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:43:03,143 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5300, 4.4674, 4.4147, 4.0896, 4.1370, 4.5113, 4.1996, 4.1928], device='cuda:7'), covar=tensor([0.0726, 0.0942, 0.0401, 0.0370, 0.0937, 0.0511, 0.0580, 0.0827], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0427, 0.0341, 0.0336, 0.0347, 0.0394, 0.0232, 0.0407], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:43:40,975 INFO [train.py:904] (7/8) Epoch 23, batch 500, loss[loss=0.1599, simple_loss=0.2482, pruned_loss=0.0358, over 16521.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2515, pruned_loss=0.03924, over 3068314.77 frames. ], batch size: 68, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:43:52,983 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.054e+02 2.358e+02 2.805e+02 6.007e+02, threshold=4.715e+02, percent-clipped=4.0 2023-05-01 13:43:56,023 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1413, 4.2219, 4.5782, 4.5385, 4.6122, 4.3043, 4.2029, 4.2451], device='cuda:7'), covar=tensor([0.0636, 0.0983, 0.0628, 0.0679, 0.0736, 0.0754, 0.1349, 0.0699], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0459, 0.0446, 0.0413, 0.0489, 0.0468, 0.0550, 0.0374], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 13:44:16,911 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=223829.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 13:44:25,970 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4108, 2.2022, 2.3148, 4.2150, 2.3307, 2.5783, 2.3227, 2.4303], device='cuda:7'), covar=tensor([0.1285, 0.3866, 0.3236, 0.0531, 0.4205, 0.2824, 0.3667, 0.3757], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0451, 0.0370, 0.0328, 0.0439, 0.0517, 0.0422, 0.0526], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:44:49,728 INFO [train.py:904] (7/8) Epoch 23, batch 550, loss[loss=0.1494, simple_loss=0.2373, pruned_loss=0.03076, over 17292.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2505, pruned_loss=0.0386, over 3130384.60 frames. ], batch size: 43, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:45:13,151 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.00 vs. limit=5.0 2023-05-01 13:45:17,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1070, 5.0809, 4.9699, 4.5142, 4.6351, 5.0382, 4.8854, 4.6802], device='cuda:7'), covar=tensor([0.0570, 0.0539, 0.0313, 0.0312, 0.0985, 0.0450, 0.0363, 0.0731], device='cuda:7'), in_proj_covar=tensor([0.0294, 0.0431, 0.0344, 0.0340, 0.0350, 0.0397, 0.0235, 0.0410], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:45:24,265 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=223877.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 13:45:50,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7553, 2.5520, 2.6590, 4.9585, 3.9394, 4.3585, 1.7141, 3.2398], device='cuda:7'), covar=tensor([0.1450, 0.0993, 0.1277, 0.0203, 0.0190, 0.0403, 0.1741, 0.0752], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0174, 0.0193, 0.0188, 0.0201, 0.0215, 0.0203, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 13:45:59,119 INFO [train.py:904] (7/8) Epoch 23, batch 600, loss[loss=0.1514, simple_loss=0.2453, pruned_loss=0.02872, over 17125.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2504, pruned_loss=0.03922, over 3164589.07 frames. ], batch size: 49, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:46:10,993 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.619e+02 2.055e+02 2.460e+02 3.168e+02 5.428e+02, threshold=4.920e+02, percent-clipped=4.0 2023-05-01 13:46:20,890 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7497, 4.0356, 2.7959, 2.3586, 2.5407, 2.4485, 4.3084, 3.2955], device='cuda:7'), covar=tensor([0.3190, 0.0899, 0.2089, 0.2897, 0.3059, 0.2302, 0.0530, 0.1655], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0267, 0.0304, 0.0313, 0.0294, 0.0260, 0.0295, 0.0337], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 13:46:42,829 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=223934.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:47:08,255 INFO [train.py:904] (7/8) Epoch 23, batch 650, loss[loss=0.1479, simple_loss=0.2462, pruned_loss=0.02483, over 17130.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2491, pruned_loss=0.03891, over 3199952.13 frames. ], batch size: 48, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:48:22,016 INFO [train.py:904] (7/8) Epoch 23, batch 700, loss[loss=0.1875, simple_loss=0.2799, pruned_loss=0.04758, over 17099.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2491, pruned_loss=0.03925, over 3231605.69 frames. ], batch size: 55, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:48:35,478 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.336e+02 2.741e+02 3.391e+02 8.377e+02, threshold=5.482e+02, percent-clipped=4.0 2023-05-01 13:48:38,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 13:48:41,653 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2023-05-01 13:49:24,348 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5497, 5.9869, 5.7524, 5.7379, 5.3034, 5.3730, 5.3884, 6.0764], device='cuda:7'), covar=tensor([0.1469, 0.0913, 0.1079, 0.0818, 0.1027, 0.0679, 0.1296, 0.1034], device='cuda:7'), in_proj_covar=tensor([0.0683, 0.0830, 0.0683, 0.0630, 0.0523, 0.0535, 0.0699, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:49:33,346 INFO [train.py:904] (7/8) Epoch 23, batch 750, loss[loss=0.16, simple_loss=0.2552, pruned_loss=0.03235, over 16718.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2506, pruned_loss=0.03972, over 3243517.32 frames. ], batch size: 62, lr: 2.96e-03, grad_scale: 2.0 2023-05-01 13:50:41,982 INFO [train.py:904] (7/8) Epoch 23, batch 800, loss[loss=0.16, simple_loss=0.2575, pruned_loss=0.0313, over 16631.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2507, pruned_loss=0.04006, over 3253318.68 frames. ], batch size: 62, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:50:54,820 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.213e+02 2.619e+02 3.114e+02 4.585e+02, threshold=5.238e+02, percent-clipped=0.0 2023-05-01 13:51:51,765 INFO [train.py:904] (7/8) Epoch 23, batch 850, loss[loss=0.1412, simple_loss=0.2273, pruned_loss=0.02758, over 16294.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2507, pruned_loss=0.03965, over 3269852.09 frames. ], batch size: 36, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:52:36,173 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6204, 2.4357, 1.9639, 2.2113, 2.7543, 2.5082, 2.7060, 2.8936], device='cuda:7'), covar=tensor([0.0231, 0.0415, 0.0582, 0.0456, 0.0242, 0.0361, 0.0222, 0.0269], device='cuda:7'), in_proj_covar=tensor([0.0218, 0.0241, 0.0232, 0.0232, 0.0242, 0.0241, 0.0241, 0.0235], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 13:53:00,765 INFO [train.py:904] (7/8) Epoch 23, batch 900, loss[loss=0.1564, simple_loss=0.2373, pruned_loss=0.03774, over 12443.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.25, pruned_loss=0.03891, over 3281258.26 frames. ], batch size: 247, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:53:14,896 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.103e+02 2.438e+02 2.948e+02 5.610e+02, threshold=4.876e+02, percent-clipped=1.0 2023-05-01 13:53:15,372 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224212.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:53:26,047 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 13:53:46,041 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=224234.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:54:10,208 INFO [train.py:904] (7/8) Epoch 23, batch 950, loss[loss=0.1612, simple_loss=0.2554, pruned_loss=0.0335, over 17151.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.25, pruned_loss=0.03866, over 3295188.43 frames. ], batch size: 47, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:54:38,591 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224273.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:54:51,189 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=224282.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 13:55:20,388 INFO [train.py:904] (7/8) Epoch 23, batch 1000, loss[loss=0.173, simple_loss=0.2497, pruned_loss=0.04815, over 16906.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2486, pruned_loss=0.03912, over 3301563.76 frames. ], batch size: 109, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:55:33,107 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-05-01 13:55:33,535 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.242e+02 2.591e+02 2.992e+02 5.709e+02, threshold=5.183e+02, percent-clipped=2.0 2023-05-01 13:56:17,390 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 13:56:31,388 INFO [train.py:904] (7/8) Epoch 23, batch 1050, loss[loss=0.159, simple_loss=0.2562, pruned_loss=0.03084, over 17058.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2484, pruned_loss=0.03925, over 3304678.68 frames. ], batch size: 53, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:57:42,166 INFO [train.py:904] (7/8) Epoch 23, batch 1100, loss[loss=0.1852, simple_loss=0.2789, pruned_loss=0.04571, over 16753.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2482, pruned_loss=0.03918, over 3314183.36 frames. ], batch size: 57, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:57:54,069 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.008e+02 2.334e+02 2.683e+02 4.688e+02, threshold=4.667e+02, percent-clipped=0.0 2023-05-01 13:58:51,588 INFO [train.py:904] (7/8) Epoch 23, batch 1150, loss[loss=0.1657, simple_loss=0.2448, pruned_loss=0.04327, over 16773.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2476, pruned_loss=0.03882, over 3306414.92 frames. ], batch size: 102, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 13:59:33,616 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.05 vs. limit=5.0 2023-05-01 14:00:00,972 INFO [train.py:904] (7/8) Epoch 23, batch 1200, loss[loss=0.1473, simple_loss=0.2318, pruned_loss=0.03138, over 16836.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2469, pruned_loss=0.03821, over 3310846.33 frames. ], batch size: 42, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:00:14,528 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.279e+02 2.643e+02 3.361e+02 1.197e+03, threshold=5.285e+02, percent-clipped=8.0 2023-05-01 14:00:53,773 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7887, 3.7065, 3.7764, 3.6681, 3.7774, 4.2258, 3.8618, 3.5536], device='cuda:7'), covar=tensor([0.2051, 0.2708, 0.2779, 0.2493, 0.3099, 0.2288, 0.1740, 0.2706], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0603, 0.0670, 0.0496, 0.0669, 0.0694, 0.0523, 0.0664], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:01:10,510 INFO [train.py:904] (7/8) Epoch 23, batch 1250, loss[loss=0.1659, simple_loss=0.2496, pruned_loss=0.04107, over 17220.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2471, pruned_loss=0.03843, over 3316931.16 frames. ], batch size: 44, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:01:14,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1950, 5.9124, 6.0253, 5.7218, 5.9123, 6.3602, 5.8655, 5.6067], device='cuda:7'), covar=tensor([0.0864, 0.1766, 0.2246, 0.1948, 0.2430, 0.0899, 0.1390, 0.2138], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0601, 0.0667, 0.0494, 0.0666, 0.0692, 0.0521, 0.0661], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:01:22,903 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224562.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:01:31,872 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=224568.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:01:52,180 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 14:02:07,622 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 14:02:20,415 INFO [train.py:904] (7/8) Epoch 23, batch 1300, loss[loss=0.122, simple_loss=0.2079, pruned_loss=0.018, over 16847.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2466, pruned_loss=0.03821, over 3318158.84 frames. ], batch size: 39, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:02:33,673 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.195e+02 2.523e+02 3.004e+02 8.579e+02, threshold=5.045e+02, percent-clipped=3.0 2023-05-01 14:02:49,073 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224623.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:02:53,156 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7038, 3.7060, 2.0236, 3.9324, 2.7986, 3.9006, 2.0224, 2.8749], device='cuda:7'), covar=tensor([0.0242, 0.0340, 0.1752, 0.0311, 0.0773, 0.0494, 0.1848, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0197, 0.0169, 0.0180, 0.0221, 0.0207, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 14:03:29,367 INFO [train.py:904] (7/8) Epoch 23, batch 1350, loss[loss=0.1456, simple_loss=0.2336, pruned_loss=0.02882, over 17019.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2464, pruned_loss=0.03749, over 3328662.02 frames. ], batch size: 41, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:03:49,047 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 14:04:40,094 INFO [train.py:904] (7/8) Epoch 23, batch 1400, loss[loss=0.1738, simple_loss=0.257, pruned_loss=0.04531, over 17171.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2467, pruned_loss=0.0377, over 3319959.67 frames. ], batch size: 46, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:04:52,700 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.431e+02 2.135e+02 2.533e+02 2.912e+02 4.513e+02, threshold=5.066e+02, percent-clipped=0.0 2023-05-01 14:05:14,017 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224726.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:05:33,797 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=224740.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:05:50,559 INFO [train.py:904] (7/8) Epoch 23, batch 1450, loss[loss=0.1545, simple_loss=0.2375, pruned_loss=0.03575, over 16778.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2456, pruned_loss=0.03816, over 3324278.41 frames. ], batch size: 39, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:06:01,581 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0134, 4.9507, 4.7713, 4.2356, 4.8904, 1.9266, 4.6304, 4.5509], device='cuda:7'), covar=tensor([0.0134, 0.0113, 0.0226, 0.0373, 0.0111, 0.2799, 0.0160, 0.0241], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0162, 0.0204, 0.0180, 0.0182, 0.0214, 0.0194, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:06:32,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6713, 4.5874, 4.5600, 3.5276, 4.5932, 1.7209, 4.1843, 4.1492], device='cuda:7'), covar=tensor([0.0232, 0.0174, 0.0264, 0.0704, 0.0193, 0.3534, 0.0267, 0.0377], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0162, 0.0205, 0.0181, 0.0182, 0.0215, 0.0194, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:06:39,759 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224787.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:06:39,823 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0627, 4.6353, 3.2261, 2.4358, 2.8832, 2.6639, 4.8980, 3.6693], device='cuda:7'), covar=tensor([0.2592, 0.0463, 0.1847, 0.2850, 0.2930, 0.2104, 0.0299, 0.1336], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0270, 0.0309, 0.0317, 0.0299, 0.0264, 0.0299, 0.0343], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:06:59,987 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=224801.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:07:01,752 INFO [train.py:904] (7/8) Epoch 23, batch 1500, loss[loss=0.1476, simple_loss=0.2289, pruned_loss=0.03321, over 12006.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2449, pruned_loss=0.03802, over 3313726.04 frames. ], batch size: 246, lr: 2.96e-03, grad_scale: 8.0 2023-05-01 14:07:03,425 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1760, 5.2356, 5.5912, 5.5816, 5.6321, 5.2813, 5.2158, 5.0427], device='cuda:7'), covar=tensor([0.0302, 0.0502, 0.0412, 0.0426, 0.0396, 0.0355, 0.0864, 0.0400], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0470, 0.0458, 0.0423, 0.0502, 0.0481, 0.0563, 0.0383], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 14:07:16,528 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.087e+02 2.500e+02 3.086e+02 8.479e+02, threshold=5.000e+02, percent-clipped=3.0 2023-05-01 14:07:44,784 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8646, 2.9324, 2.6531, 5.0273, 3.9313, 4.3755, 1.6724, 3.1430], device='cuda:7'), covar=tensor([0.1473, 0.0841, 0.1313, 0.0220, 0.0302, 0.0447, 0.1807, 0.0850], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0176, 0.0196, 0.0192, 0.0205, 0.0217, 0.0205, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 14:08:14,260 INFO [train.py:904] (7/8) Epoch 23, batch 1550, loss[loss=0.1819, simple_loss=0.2516, pruned_loss=0.05609, over 16882.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2466, pruned_loss=0.0391, over 3317869.04 frames. ], batch size: 116, lr: 2.96e-03, grad_scale: 4.0 2023-05-01 14:08:35,278 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=224868.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:09:23,925 INFO [train.py:904] (7/8) Epoch 23, batch 1600, loss[loss=0.1341, simple_loss=0.2184, pruned_loss=0.02495, over 16786.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2486, pruned_loss=0.03957, over 3323179.80 frames. ], batch size: 39, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:09:36,821 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.147e+02 2.697e+02 3.461e+02 9.919e+02, threshold=5.394e+02, percent-clipped=4.0 2023-05-01 14:09:41,303 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=224916.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:09:43,710 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=224918.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:09:54,894 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 14:10:32,923 INFO [train.py:904] (7/8) Epoch 23, batch 1650, loss[loss=0.1685, simple_loss=0.2644, pruned_loss=0.03632, over 17020.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2508, pruned_loss=0.04062, over 3316145.92 frames. ], batch size: 50, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:11:41,682 INFO [train.py:904] (7/8) Epoch 23, batch 1700, loss[loss=0.1499, simple_loss=0.2379, pruned_loss=0.031, over 17197.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2527, pruned_loss=0.04111, over 3302084.24 frames. ], batch size: 44, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:11:56,161 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.319e+02 2.651e+02 3.431e+02 9.066e+02, threshold=5.301e+02, percent-clipped=1.0 2023-05-01 14:12:13,288 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5237, 2.6067, 2.2805, 2.4641, 2.9116, 2.6671, 3.0964, 3.0734], device='cuda:7'), covar=tensor([0.0184, 0.0429, 0.0530, 0.0456, 0.0319, 0.0390, 0.0332, 0.0293], device='cuda:7'), in_proj_covar=tensor([0.0220, 0.0242, 0.0232, 0.0232, 0.0243, 0.0242, 0.0244, 0.0237], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:12:52,536 INFO [train.py:904] (7/8) Epoch 23, batch 1750, loss[loss=0.1687, simple_loss=0.2488, pruned_loss=0.04424, over 16754.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2544, pruned_loss=0.04168, over 3292914.70 frames. ], batch size: 83, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:13:33,051 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225082.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:13:34,955 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3932, 5.3277, 5.2417, 4.6732, 4.8379, 5.2622, 5.2265, 4.8502], device='cuda:7'), covar=tensor([0.0607, 0.0582, 0.0340, 0.0381, 0.1199, 0.0553, 0.0269, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0448, 0.0357, 0.0354, 0.0362, 0.0414, 0.0242, 0.0426], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:13:52,629 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225096.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:14:01,725 INFO [train.py:904] (7/8) Epoch 23, batch 1800, loss[loss=0.1823, simple_loss=0.2837, pruned_loss=0.04047, over 16694.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2555, pruned_loss=0.0415, over 3286697.72 frames. ], batch size: 57, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:14:02,789 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225103.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:14:15,809 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.184e+02 2.489e+02 3.045e+02 6.303e+02, threshold=4.978e+02, percent-clipped=2.0 2023-05-01 14:14:35,786 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0078, 4.7732, 5.0720, 5.2462, 5.4766, 4.7780, 5.4194, 5.4515], device='cuda:7'), covar=tensor([0.2002, 0.1528, 0.2009, 0.0906, 0.0579, 0.0938, 0.0567, 0.0642], device='cuda:7'), in_proj_covar=tensor([0.0674, 0.0832, 0.0964, 0.0842, 0.0636, 0.0668, 0.0692, 0.0803], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:14:57,330 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225142.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:15:12,135 INFO [train.py:904] (7/8) Epoch 23, batch 1850, loss[loss=0.1556, simple_loss=0.2425, pruned_loss=0.0344, over 16779.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2568, pruned_loss=0.04184, over 3281310.82 frames. ], batch size: 83, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:15:27,798 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225164.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:16:22,256 INFO [train.py:904] (7/8) Epoch 23, batch 1900, loss[loss=0.1676, simple_loss=0.2614, pruned_loss=0.03688, over 17037.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2551, pruned_loss=0.04085, over 3292877.96 frames. ], batch size: 50, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:16:22,751 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225203.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:16:36,681 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 2.136e+02 2.472e+02 2.864e+02 6.129e+02, threshold=4.945e+02, percent-clipped=2.0 2023-05-01 14:16:44,002 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225218.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:17:11,927 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 14:17:31,829 INFO [train.py:904] (7/8) Epoch 23, batch 1950, loss[loss=0.1756, simple_loss=0.2668, pruned_loss=0.04226, over 17016.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2553, pruned_loss=0.04074, over 3295314.33 frames. ], batch size: 55, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:17:49,241 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3576, 2.2174, 2.2727, 4.0810, 2.2189, 2.5266, 2.3182, 2.3599], device='cuda:7'), covar=tensor([0.1369, 0.4061, 0.3214, 0.0574, 0.4438, 0.2974, 0.3785, 0.3995], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0456, 0.0374, 0.0332, 0.0441, 0.0526, 0.0428, 0.0532], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:17:50,100 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225266.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:18:24,611 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2240, 4.0713, 4.2896, 4.4146, 4.5381, 4.1499, 4.3295, 4.5287], device='cuda:7'), covar=tensor([0.1822, 0.1319, 0.1549, 0.0813, 0.0633, 0.1283, 0.2350, 0.0758], device='cuda:7'), in_proj_covar=tensor([0.0677, 0.0838, 0.0971, 0.0844, 0.0639, 0.0672, 0.0694, 0.0805], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:18:37,348 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7817, 2.6473, 2.3621, 2.5736, 2.9958, 2.7565, 3.2848, 3.1980], device='cuda:7'), covar=tensor([0.0146, 0.0442, 0.0549, 0.0464, 0.0315, 0.0437, 0.0277, 0.0290], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0242, 0.0232, 0.0232, 0.0243, 0.0242, 0.0243, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:18:42,395 INFO [train.py:904] (7/8) Epoch 23, batch 2000, loss[loss=0.1695, simple_loss=0.2588, pruned_loss=0.04007, over 17191.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2545, pruned_loss=0.04055, over 3302302.03 frames. ], batch size: 44, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:18:56,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.089e+02 2.408e+02 2.851e+02 8.100e+02, threshold=4.816e+02, percent-clipped=1.0 2023-05-01 14:19:52,859 INFO [train.py:904] (7/8) Epoch 23, batch 2050, loss[loss=0.1888, simple_loss=0.2671, pruned_loss=0.05521, over 16490.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2552, pruned_loss=0.04096, over 3304479.21 frames. ], batch size: 146, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:20:02,854 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-01 14:20:35,695 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225382.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:20:54,954 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225396.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:21:04,273 INFO [train.py:904] (7/8) Epoch 23, batch 2100, loss[loss=0.2027, simple_loss=0.289, pruned_loss=0.0582, over 16533.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2566, pruned_loss=0.042, over 3306460.96 frames. ], batch size: 68, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:21:09,881 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1468, 2.1805, 2.6865, 3.1409, 3.1288, 3.5968, 2.0588, 3.6572], device='cuda:7'), covar=tensor([0.0231, 0.0607, 0.0383, 0.0312, 0.0301, 0.0206, 0.0747, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0195, 0.0183, 0.0188, 0.0200, 0.0158, 0.0199, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:21:18,936 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.181e+02 2.818e+02 3.430e+02 7.396e+02, threshold=5.635e+02, percent-clipped=7.0 2023-05-01 14:21:29,385 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 14:21:44,081 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:21:57,158 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225439.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:03,552 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225444.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:15,767 INFO [train.py:904] (7/8) Epoch 23, batch 2150, loss[loss=0.179, simple_loss=0.2548, pruned_loss=0.05162, over 16508.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.257, pruned_loss=0.04182, over 3320485.00 frames. ], batch size: 146, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:22:25,333 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225459.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:25,395 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4372, 4.3833, 4.3733, 4.0772, 4.1400, 4.3907, 4.1293, 4.1852], device='cuda:7'), covar=tensor([0.0647, 0.0806, 0.0310, 0.0338, 0.0746, 0.0560, 0.0661, 0.0618], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0456, 0.0363, 0.0361, 0.0367, 0.0421, 0.0246, 0.0433], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:22:29,371 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225462.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:22:57,601 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5089, 3.7230, 4.1784, 2.3000, 3.2993, 2.7641, 4.0241, 3.8629], device='cuda:7'), covar=tensor([0.0274, 0.0894, 0.0467, 0.2060, 0.0822, 0.0919, 0.0580, 0.1118], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0168, 0.0154, 0.0146, 0.0130, 0.0144, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 14:23:19,534 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225498.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:22,098 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225500.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:26,046 INFO [train.py:904] (7/8) Epoch 23, batch 2200, loss[loss=0.1639, simple_loss=0.2596, pruned_loss=0.03414, over 17061.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2574, pruned_loss=0.0421, over 3320365.05 frames. ], batch size: 55, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:23:29,410 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225505.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:40,047 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.156e+02 2.513e+02 2.761e+02 5.563e+02, threshold=5.025e+02, percent-clipped=0.0 2023-05-01 14:23:53,565 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225522.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:23:55,311 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225523.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:24:36,374 INFO [train.py:904] (7/8) Epoch 23, batch 2250, loss[loss=0.1567, simple_loss=0.2562, pruned_loss=0.02854, over 17050.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2589, pruned_loss=0.04257, over 3326124.37 frames. ], batch size: 50, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:24:40,095 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5637, 1.8188, 2.2103, 2.3731, 2.5790, 2.5660, 1.9150, 2.6739], device='cuda:7'), covar=tensor([0.0216, 0.0524, 0.0346, 0.0347, 0.0312, 0.0325, 0.0541, 0.0197], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0188, 0.0200, 0.0157, 0.0199, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:24:45,169 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225559.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 14:24:54,978 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225566.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:25:18,800 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225583.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:25:46,980 INFO [train.py:904] (7/8) Epoch 23, batch 2300, loss[loss=0.18, simple_loss=0.2573, pruned_loss=0.05132, over 15555.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2586, pruned_loss=0.04224, over 3320530.46 frames. ], batch size: 191, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:25:56,972 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 14:26:01,556 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.669e+02 2.061e+02 2.466e+02 3.026e+02 4.196e+02, threshold=4.932e+02, percent-clipped=0.0 2023-05-01 14:26:11,652 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225620.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 14:26:58,889 INFO [train.py:904] (7/8) Epoch 23, batch 2350, loss[loss=0.1893, simple_loss=0.267, pruned_loss=0.05577, over 16667.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.258, pruned_loss=0.04215, over 3323670.74 frames. ], batch size: 89, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:27:34,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3313, 2.1607, 2.3623, 4.0638, 2.1083, 2.3759, 2.3031, 2.3190], device='cuda:7'), covar=tensor([0.1609, 0.4483, 0.3332, 0.0727, 0.5259, 0.3431, 0.3962, 0.4572], device='cuda:7'), in_proj_covar=tensor([0.0410, 0.0459, 0.0376, 0.0335, 0.0443, 0.0530, 0.0431, 0.0537], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:28:10,324 INFO [train.py:904] (7/8) Epoch 23, batch 2400, loss[loss=0.1408, simple_loss=0.2289, pruned_loss=0.02631, over 17218.00 frames. ], tot_loss[loss=0.171, simple_loss=0.258, pruned_loss=0.04199, over 3329850.25 frames. ], batch size: 44, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:28:23,216 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.701e+02 2.227e+02 2.621e+02 3.138e+02 6.803e+02, threshold=5.242e+02, percent-clipped=3.0 2023-05-01 14:29:17,854 INFO [train.py:904] (7/8) Epoch 23, batch 2450, loss[loss=0.1842, simple_loss=0.2762, pruned_loss=0.04613, over 17141.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2587, pruned_loss=0.04223, over 3329023.90 frames. ], batch size: 48, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:29:26,331 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225759.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:10,521 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.21 vs. limit=5.0 2023-05-01 14:30:18,105 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225795.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:22,204 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=225798.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:29,836 INFO [train.py:904] (7/8) Epoch 23, batch 2500, loss[loss=0.1858, simple_loss=0.2818, pruned_loss=0.04494, over 17063.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2583, pruned_loss=0.04196, over 3335384.54 frames. ], batch size: 55, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:30:36,121 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225807.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:30:43,416 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.105e+02 2.544e+02 3.075e+02 7.140e+02, threshold=5.088e+02, percent-clipped=2.0 2023-05-01 14:30:51,426 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225818.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:31:18,765 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6360, 3.8081, 2.4640, 4.3767, 2.9491, 4.3506, 2.4430, 3.1035], device='cuda:7'), covar=tensor([0.0345, 0.0435, 0.1596, 0.0395, 0.0874, 0.0512, 0.1579, 0.0782], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0181, 0.0197, 0.0171, 0.0180, 0.0223, 0.0207, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 14:31:29,062 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=225846.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:31:38,345 INFO [train.py:904] (7/8) Epoch 23, batch 2550, loss[loss=0.1853, simple_loss=0.272, pruned_loss=0.04931, over 16426.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2591, pruned_loss=0.04235, over 3328964.58 frames. ], batch size: 68, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:31:49,570 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225861.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:31:50,904 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225862.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:32:04,200 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2655, 4.5704, 4.5303, 3.3751, 3.7418, 4.4888, 3.9828, 2.8840], device='cuda:7'), covar=tensor([0.0404, 0.0068, 0.0043, 0.0319, 0.0135, 0.0091, 0.0098, 0.0424], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0085, 0.0086, 0.0135, 0.0099, 0.0111, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 14:32:14,115 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225878.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:32:15,370 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2967, 3.3038, 3.4023, 2.3676, 3.1849, 3.5396, 3.1773, 1.7931], device='cuda:7'), covar=tensor([0.0521, 0.0148, 0.0087, 0.0430, 0.0136, 0.0112, 0.0140, 0.0634], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0085, 0.0086, 0.0135, 0.0099, 0.0111, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 14:32:17,444 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9936, 4.5135, 3.3265, 2.5393, 2.8329, 2.7599, 4.8553, 3.6835], device='cuda:7'), covar=tensor([0.2811, 0.0537, 0.1689, 0.2909, 0.3122, 0.1973, 0.0338, 0.1451], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0272, 0.0309, 0.0318, 0.0301, 0.0265, 0.0301, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:32:38,077 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=225895.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:32:49,288 INFO [train.py:904] (7/8) Epoch 23, batch 2600, loss[loss=0.1541, simple_loss=0.2412, pruned_loss=0.03349, over 16879.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2588, pruned_loss=0.04215, over 3325246.94 frames. ], batch size: 96, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:33:03,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.265e+02 2.588e+02 2.921e+02 5.309e+02, threshold=5.176e+02, percent-clipped=1.0 2023-05-01 14:33:06,168 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=225915.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 14:33:17,370 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225923.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:33:58,804 INFO [train.py:904] (7/8) Epoch 23, batch 2650, loss[loss=0.1941, simple_loss=0.2853, pruned_loss=0.05142, over 16552.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2588, pruned_loss=0.04144, over 3330971.22 frames. ], batch size: 75, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:34:04,295 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=225956.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:34:31,970 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8997, 1.9853, 2.3688, 2.7013, 2.7582, 2.7167, 2.0505, 2.9913], device='cuda:7'), covar=tensor([0.0209, 0.0516, 0.0380, 0.0328, 0.0328, 0.0338, 0.0560, 0.0167], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0196, 0.0183, 0.0188, 0.0201, 0.0159, 0.0199, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:35:12,278 INFO [train.py:904] (7/8) Epoch 23, batch 2700, loss[loss=0.1803, simple_loss=0.2622, pruned_loss=0.04919, over 16271.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2591, pruned_loss=0.041, over 3336150.56 frames. ], batch size: 165, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:35:25,734 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.491e+02 2.127e+02 2.428e+02 2.889e+02 7.591e+02, threshold=4.856e+02, percent-clipped=2.0 2023-05-01 14:36:06,284 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=226041.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:36:23,072 INFO [train.py:904] (7/8) Epoch 23, batch 2750, loss[loss=0.1764, simple_loss=0.2703, pruned_loss=0.04124, over 16056.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2597, pruned_loss=0.04084, over 3331374.69 frames. ], batch size: 35, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:36:38,078 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6383, 3.6583, 2.3376, 3.9373, 2.9171, 3.9301, 2.4606, 2.9732], device='cuda:7'), covar=tensor([0.0268, 0.0406, 0.1507, 0.0319, 0.0700, 0.0628, 0.1392, 0.0693], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0196, 0.0171, 0.0179, 0.0222, 0.0206, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 14:37:21,976 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226095.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:37:31,960 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=226102.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:37:32,648 INFO [train.py:904] (7/8) Epoch 23, batch 2800, loss[loss=0.1783, simple_loss=0.2557, pruned_loss=0.05039, over 16390.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2587, pruned_loss=0.04051, over 3323295.71 frames. ], batch size: 146, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:37:40,572 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 14:37:47,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.139e+02 2.068e+02 2.362e+02 2.950e+02 8.233e+02, threshold=4.725e+02, percent-clipped=4.0 2023-05-01 14:37:54,866 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226118.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:38:29,333 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226143.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:38:42,393 INFO [train.py:904] (7/8) Epoch 23, batch 2850, loss[loss=0.1636, simple_loss=0.2444, pruned_loss=0.04138, over 16552.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2582, pruned_loss=0.04055, over 3320205.34 frames. ], batch size: 146, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:38:54,339 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226161.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:39:00,769 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226166.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:39:18,096 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226178.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:39:51,840 INFO [train.py:904] (7/8) Epoch 23, batch 2900, loss[loss=0.1765, simple_loss=0.2824, pruned_loss=0.03524, over 17148.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2576, pruned_loss=0.04051, over 3325062.70 frames. ], batch size: 49, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:40:00,213 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226209.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:00,478 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3103, 3.3929, 3.7538, 2.2539, 3.1720, 2.5316, 3.8108, 3.7669], device='cuda:7'), covar=tensor([0.0231, 0.0895, 0.0524, 0.1952, 0.0838, 0.0952, 0.0466, 0.0816], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0167, 0.0168, 0.0154, 0.0146, 0.0131, 0.0144, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 14:40:05,816 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.036e+02 2.506e+02 2.935e+02 4.421e+02, threshold=5.013e+02, percent-clipped=0.0 2023-05-01 14:40:09,254 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226215.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 14:40:13,517 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226218.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:24,172 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226226.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:34,166 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=226233.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:40:58,442 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226251.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:41:00,394 INFO [train.py:904] (7/8) Epoch 23, batch 2950, loss[loss=0.1898, simple_loss=0.2678, pruned_loss=0.05588, over 16771.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.257, pruned_loss=0.04108, over 3323053.97 frames. ], batch size: 124, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:41:14,674 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226263.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 14:41:36,942 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1540, 3.9028, 3.9490, 4.3055, 4.4245, 4.0798, 4.2156, 4.3869], device='cuda:7'), covar=tensor([0.1858, 0.1520, 0.2202, 0.1080, 0.0945, 0.1825, 0.3081, 0.1217], device='cuda:7'), in_proj_covar=tensor([0.0675, 0.0839, 0.0970, 0.0847, 0.0643, 0.0673, 0.0697, 0.0808], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:41:56,424 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=226294.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:42:08,780 INFO [train.py:904] (7/8) Epoch 23, batch 3000, loss[loss=0.1506, simple_loss=0.2358, pruned_loss=0.03275, over 16881.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2569, pruned_loss=0.04212, over 3316226.24 frames. ], batch size: 96, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:42:08,780 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 14:42:17,865 INFO [train.py:938] (7/8) Epoch 23, validation: loss=0.1344, simple_loss=0.2397, pruned_loss=0.01456, over 944034.00 frames. 2023-05-01 14:42:17,866 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 14:42:30,994 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.698e+02 2.313e+02 2.743e+02 3.282e+02 6.136e+02, threshold=5.486e+02, percent-clipped=4.0 2023-05-01 14:43:26,884 INFO [train.py:904] (7/8) Epoch 23, batch 3050, loss[loss=0.1628, simple_loss=0.2469, pruned_loss=0.03931, over 16263.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2568, pruned_loss=0.04199, over 3315524.67 frames. ], batch size: 165, lr: 2.95e-03, grad_scale: 8.0 2023-05-01 14:43:32,254 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7637, 4.2366, 4.2941, 2.9462, 3.5369, 4.2736, 3.8154, 2.5191], device='cuda:7'), covar=tensor([0.0490, 0.0076, 0.0052, 0.0394, 0.0152, 0.0094, 0.0097, 0.0455], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 14:43:55,990 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9743, 4.6381, 4.6309, 5.1025, 5.3386, 4.7904, 5.2631, 5.2813], device='cuda:7'), covar=tensor([0.1783, 0.1359, 0.2591, 0.1023, 0.0791, 0.1220, 0.0862, 0.0961], device='cuda:7'), in_proj_covar=tensor([0.0677, 0.0842, 0.0974, 0.0849, 0.0645, 0.0675, 0.0698, 0.0810], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:43:58,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0394, 4.4909, 4.4273, 3.1148, 3.7503, 4.4444, 3.9803, 2.6327], device='cuda:7'), covar=tensor([0.0495, 0.0061, 0.0056, 0.0407, 0.0147, 0.0108, 0.0107, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0097, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 14:44:29,381 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226397.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:44:37,979 INFO [train.py:904] (7/8) Epoch 23, batch 3100, loss[loss=0.1612, simple_loss=0.2396, pruned_loss=0.04134, over 16691.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2564, pruned_loss=0.04176, over 3311472.99 frames. ], batch size: 89, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:44:51,601 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.140e+02 2.727e+02 3.303e+02 6.006e+02, threshold=5.454e+02, percent-clipped=2.0 2023-05-01 14:45:47,156 INFO [train.py:904] (7/8) Epoch 23, batch 3150, loss[loss=0.1712, simple_loss=0.2661, pruned_loss=0.03816, over 17128.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2562, pruned_loss=0.04154, over 3310488.66 frames. ], batch size: 49, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:46:54,863 INFO [train.py:904] (7/8) Epoch 23, batch 3200, loss[loss=0.1811, simple_loss=0.2623, pruned_loss=0.04994, over 15508.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.255, pruned_loss=0.04089, over 3314901.53 frames. ], batch size: 190, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:47:09,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.052e+02 2.422e+02 2.820e+02 4.460e+02, threshold=4.844e+02, percent-clipped=0.0 2023-05-01 14:47:15,896 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226518.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:47:45,932 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8736, 4.3560, 3.1002, 2.3249, 2.6971, 2.6052, 4.7319, 3.5773], device='cuda:7'), covar=tensor([0.2788, 0.0533, 0.1727, 0.2820, 0.2788, 0.2072, 0.0309, 0.1345], device='cuda:7'), in_proj_covar=tensor([0.0330, 0.0272, 0.0310, 0.0319, 0.0303, 0.0266, 0.0302, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:48:02,145 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226551.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:48:04,035 INFO [train.py:904] (7/8) Epoch 23, batch 3250, loss[loss=0.128, simple_loss=0.2134, pruned_loss=0.02132, over 16975.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.255, pruned_loss=0.04092, over 3320424.47 frames. ], batch size: 41, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:48:22,344 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226566.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:48:54,467 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=226589.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:48:58,898 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3534, 5.2558, 5.0757, 4.4731, 5.1471, 2.0248, 4.9561, 5.1002], device='cuda:7'), covar=tensor([0.0084, 0.0090, 0.0202, 0.0424, 0.0116, 0.2686, 0.0145, 0.0191], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0166, 0.0207, 0.0184, 0.0186, 0.0215, 0.0197, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:49:09,557 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226599.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:49:14,399 INFO [train.py:904] (7/8) Epoch 23, batch 3300, loss[loss=0.2016, simple_loss=0.2823, pruned_loss=0.06052, over 16392.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.256, pruned_loss=0.04113, over 3310955.56 frames. ], batch size: 146, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:49:24,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3818, 4.3156, 4.2743, 3.7034, 4.3092, 1.7525, 4.1021, 3.8449], device='cuda:7'), covar=tensor([0.0126, 0.0125, 0.0187, 0.0301, 0.0108, 0.2921, 0.0127, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0166, 0.0207, 0.0184, 0.0185, 0.0214, 0.0197, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:49:27,931 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.105e+02 2.619e+02 3.003e+02 4.974e+02, threshold=5.238e+02, percent-clipped=1.0 2023-05-01 14:50:23,123 INFO [train.py:904] (7/8) Epoch 23, batch 3350, loss[loss=0.1769, simple_loss=0.2787, pruned_loss=0.03759, over 17247.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.256, pruned_loss=0.04118, over 3314692.30 frames. ], batch size: 52, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:51:25,664 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226697.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:51:34,604 INFO [train.py:904] (7/8) Epoch 23, batch 3400, loss[loss=0.1659, simple_loss=0.2423, pruned_loss=0.04472, over 16908.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2558, pruned_loss=0.04101, over 3308577.21 frames. ], batch size: 109, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:51:36,984 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9247, 4.8456, 4.7763, 4.1516, 4.8041, 1.9413, 4.5839, 4.5453], device='cuda:7'), covar=tensor([0.0122, 0.0104, 0.0189, 0.0390, 0.0104, 0.2663, 0.0152, 0.0223], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0166, 0.0207, 0.0184, 0.0185, 0.0215, 0.0197, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:51:47,767 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.087e+02 2.446e+02 3.167e+02 4.781e+02, threshold=4.893e+02, percent-clipped=0.0 2023-05-01 14:52:33,286 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226745.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:52:38,611 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6646, 2.4980, 2.4705, 4.4999, 2.4873, 2.9008, 2.5799, 2.6354], device='cuda:7'), covar=tensor([0.1249, 0.3645, 0.3059, 0.0525, 0.4099, 0.2632, 0.3387, 0.3493], device='cuda:7'), in_proj_covar=tensor([0.0409, 0.0456, 0.0375, 0.0335, 0.0441, 0.0527, 0.0428, 0.0535], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:52:43,441 INFO [train.py:904] (7/8) Epoch 23, batch 3450, loss[loss=0.1599, simple_loss=0.2401, pruned_loss=0.03988, over 16842.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.255, pruned_loss=0.04071, over 3303190.54 frames. ], batch size: 96, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:53:52,948 INFO [train.py:904] (7/8) Epoch 23, batch 3500, loss[loss=0.1713, simple_loss=0.2778, pruned_loss=0.03236, over 17123.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2538, pruned_loss=0.03992, over 3313691.71 frames. ], batch size: 48, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:54:07,206 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.101e+02 2.445e+02 3.055e+02 4.723e+02, threshold=4.890e+02, percent-clipped=0.0 2023-05-01 14:55:03,891 INFO [train.py:904] (7/8) Epoch 23, batch 3550, loss[loss=0.1513, simple_loss=0.2371, pruned_loss=0.03274, over 16973.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2522, pruned_loss=0.03911, over 3316975.59 frames. ], batch size: 41, lr: 2.94e-03, grad_scale: 16.0 2023-05-01 14:55:10,135 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-01 14:55:45,825 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 14:55:48,372 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9988, 2.1101, 2.5182, 2.8737, 2.8970, 2.9712, 2.1303, 3.1158], device='cuda:7'), covar=tensor([0.0175, 0.0455, 0.0328, 0.0276, 0.0308, 0.0309, 0.0556, 0.0197], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0198, 0.0186, 0.0191, 0.0204, 0.0161, 0.0202, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 14:55:52,880 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=226889.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:56:03,020 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4095, 5.3600, 5.2886, 4.7226, 4.8822, 5.3281, 5.2068, 4.9098], device='cuda:7'), covar=tensor([0.0571, 0.0573, 0.0291, 0.0350, 0.1106, 0.0469, 0.0298, 0.0868], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0465, 0.0369, 0.0368, 0.0375, 0.0428, 0.0251, 0.0442], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:56:12,772 INFO [train.py:904] (7/8) Epoch 23, batch 3600, loss[loss=0.1524, simple_loss=0.2344, pruned_loss=0.03523, over 16482.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2508, pruned_loss=0.03864, over 3312349.29 frames. ], batch size: 68, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:56:28,219 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.070e+02 2.545e+02 2.955e+02 4.911e+02, threshold=5.089e+02, percent-clipped=1.0 2023-05-01 14:57:03,053 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=226937.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:57:24,675 INFO [train.py:904] (7/8) Epoch 23, batch 3650, loss[loss=0.1588, simple_loss=0.2335, pruned_loss=0.04202, over 16697.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.25, pruned_loss=0.03949, over 3298004.76 frames. ], batch size: 89, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:58:20,361 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=226989.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:58:27,843 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9091, 5.0229, 5.2174, 5.0108, 5.0535, 5.6295, 5.1225, 4.8633], device='cuda:7'), covar=tensor([0.1278, 0.2026, 0.2011, 0.2114, 0.2505, 0.1021, 0.1673, 0.2388], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0617, 0.0681, 0.0508, 0.0680, 0.0703, 0.0534, 0.0681], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 14:58:40,559 INFO [train.py:904] (7/8) Epoch 23, batch 3700, loss[loss=0.1631, simple_loss=0.2392, pruned_loss=0.0435, over 16778.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2493, pruned_loss=0.0408, over 3293225.35 frames. ], batch size: 102, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 14:58:56,578 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.184e+02 2.522e+02 2.938e+02 6.329e+02, threshold=5.043e+02, percent-clipped=1.0 2023-05-01 14:59:35,570 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4022, 3.4328, 3.5830, 2.5815, 3.2703, 3.6964, 3.3924, 2.2789], device='cuda:7'), covar=tensor([0.0477, 0.0105, 0.0062, 0.0360, 0.0118, 0.0097, 0.0097, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 14:59:50,715 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227050.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 14:59:53,703 INFO [train.py:904] (7/8) Epoch 23, batch 3750, loss[loss=0.1648, simple_loss=0.2373, pruned_loss=0.04614, over 16718.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2501, pruned_loss=0.04198, over 3274564.25 frames. ], batch size: 124, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:00:30,451 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 15:00:42,451 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8542, 4.9710, 5.1659, 4.9255, 5.0088, 5.5763, 5.0452, 4.7742], device='cuda:7'), covar=tensor([0.1272, 0.1897, 0.2132, 0.2262, 0.2585, 0.0980, 0.1679, 0.2689], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0611, 0.0674, 0.0505, 0.0674, 0.0698, 0.0529, 0.0676], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 15:01:07,862 INFO [train.py:904] (7/8) Epoch 23, batch 3800, loss[loss=0.1687, simple_loss=0.252, pruned_loss=0.04272, over 16372.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2517, pruned_loss=0.04341, over 3263178.08 frames. ], batch size: 146, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:01:25,258 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.697e+02 2.146e+02 2.630e+02 3.388e+02 6.503e+02, threshold=5.260e+02, percent-clipped=3.0 2023-05-01 15:02:21,622 INFO [train.py:904] (7/8) Epoch 23, batch 3850, loss[loss=0.1757, simple_loss=0.2598, pruned_loss=0.04581, over 16815.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2514, pruned_loss=0.04387, over 3270572.17 frames. ], batch size: 96, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:03:34,977 INFO [train.py:904] (7/8) Epoch 23, batch 3900, loss[loss=0.1727, simple_loss=0.2498, pruned_loss=0.04784, over 16334.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.251, pruned_loss=0.04441, over 3277179.35 frames. ], batch size: 35, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:03:51,365 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 15:03:51,651 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.638e+02 2.230e+02 2.530e+02 3.049e+02 6.179e+02, threshold=5.060e+02, percent-clipped=1.0 2023-05-01 15:04:47,733 INFO [train.py:904] (7/8) Epoch 23, batch 3950, loss[loss=0.1647, simple_loss=0.2355, pruned_loss=0.04699, over 16920.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2509, pruned_loss=0.04526, over 3275011.79 frames. ], batch size: 90, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:06:00,643 INFO [train.py:904] (7/8) Epoch 23, batch 4000, loss[loss=0.1899, simple_loss=0.2728, pruned_loss=0.05346, over 12388.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2509, pruned_loss=0.04562, over 3272304.50 frames. ], batch size: 246, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:06:07,473 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9852, 5.2523, 5.4440, 5.2075, 5.2445, 5.8172, 5.2632, 4.9773], device='cuda:7'), covar=tensor([0.1029, 0.1772, 0.1868, 0.1935, 0.2469, 0.0907, 0.1516, 0.2306], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0614, 0.0678, 0.0508, 0.0676, 0.0701, 0.0531, 0.0678], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 15:06:17,148 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.172e+02 2.478e+02 2.977e+02 5.073e+02, threshold=4.957e+02, percent-clipped=1.0 2023-05-01 15:07:01,815 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=227345.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:07:13,398 INFO [train.py:904] (7/8) Epoch 23, batch 4050, loss[loss=0.1681, simple_loss=0.2575, pruned_loss=0.03934, over 16746.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2523, pruned_loss=0.04521, over 3276267.96 frames. ], batch size: 83, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:07:48,481 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1692, 5.1620, 4.9164, 4.1547, 5.0981, 1.7148, 4.8351, 4.4895], device='cuda:7'), covar=tensor([0.0071, 0.0060, 0.0194, 0.0407, 0.0079, 0.3290, 0.0119, 0.0316], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0166, 0.0208, 0.0184, 0.0185, 0.0214, 0.0197, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:08:27,046 INFO [train.py:904] (7/8) Epoch 23, batch 4100, loss[loss=0.183, simple_loss=0.2702, pruned_loss=0.04788, over 16696.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2538, pruned_loss=0.04456, over 3269227.82 frames. ], batch size: 124, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:08:39,151 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=227411.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:08:42,772 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.752e+02 1.921e+02 2.258e+02 5.560e+02, threshold=3.841e+02, percent-clipped=1.0 2023-05-01 15:09:45,396 INFO [train.py:904] (7/8) Epoch 23, batch 4150, loss[loss=0.1863, simple_loss=0.2788, pruned_loss=0.04692, over 17158.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2609, pruned_loss=0.04708, over 3243454.06 frames. ], batch size: 46, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:10:16,201 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227472.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:10:22,913 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8708, 2.1628, 2.4404, 3.1243, 2.2074, 2.3513, 2.3245, 2.2882], device='cuda:7'), covar=tensor([0.1408, 0.3386, 0.2398, 0.0759, 0.3811, 0.2300, 0.3372, 0.3186], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0458, 0.0374, 0.0334, 0.0440, 0.0529, 0.0428, 0.0535], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:11:03,942 INFO [train.py:904] (7/8) Epoch 23, batch 4200, loss[loss=0.1794, simple_loss=0.2795, pruned_loss=0.03964, over 16667.00 frames. ], tot_loss[loss=0.1826, simple_loss=0.2679, pruned_loss=0.04862, over 3214905.90 frames. ], batch size: 89, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:11:20,464 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.273e+02 2.688e+02 3.076e+02 6.689e+02, threshold=5.376e+02, percent-clipped=9.0 2023-05-01 15:12:16,799 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-05-01 15:12:19,956 INFO [train.py:904] (7/8) Epoch 23, batch 4250, loss[loss=0.1924, simple_loss=0.2827, pruned_loss=0.05109, over 16572.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.272, pruned_loss=0.04912, over 3193732.77 frames. ], batch size: 57, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:13:10,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 15:13:36,404 INFO [train.py:904] (7/8) Epoch 23, batch 4300, loss[loss=0.207, simple_loss=0.2941, pruned_loss=0.05998, over 16721.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2736, pruned_loss=0.0486, over 3188178.18 frames. ], batch size: 134, lr: 2.94e-03, grad_scale: 4.0 2023-05-01 15:13:55,084 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.112e+02 2.458e+02 3.022e+02 6.068e+02, threshold=4.917e+02, percent-clipped=1.0 2023-05-01 15:14:42,488 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=227645.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:14:53,554 INFO [train.py:904] (7/8) Epoch 23, batch 4350, loss[loss=0.1715, simple_loss=0.2646, pruned_loss=0.03916, over 17232.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2762, pruned_loss=0.0495, over 3179471.33 frames. ], batch size: 44, lr: 2.94e-03, grad_scale: 4.0 2023-05-01 15:15:45,929 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 15:15:55,219 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=227693.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:15:55,385 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=227693.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:16:05,244 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-05-01 15:16:09,821 INFO [train.py:904] (7/8) Epoch 23, batch 4400, loss[loss=0.2015, simple_loss=0.285, pruned_loss=0.05902, over 16659.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2782, pruned_loss=0.05029, over 3185498.34 frames. ], batch size: 57, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:16:27,164 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.099e+02 2.601e+02 2.882e+02 6.298e+02, threshold=5.202e+02, percent-clipped=2.0 2023-05-01 15:17:22,049 INFO [train.py:904] (7/8) Epoch 23, batch 4450, loss[loss=0.1944, simple_loss=0.296, pruned_loss=0.04641, over 16897.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2818, pruned_loss=0.0518, over 3182461.21 frames. ], batch size: 96, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:17:23,638 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227754.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:17:42,536 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=227767.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:18:15,067 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8594, 3.9299, 4.1468, 4.1194, 4.1388, 3.9424, 3.9512, 3.9129], device='cuda:7'), covar=tensor([0.0318, 0.0472, 0.0359, 0.0400, 0.0415, 0.0389, 0.0750, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0459, 0.0443, 0.0413, 0.0491, 0.0465, 0.0549, 0.0372], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 15:18:35,089 INFO [train.py:904] (7/8) Epoch 23, batch 4500, loss[loss=0.1931, simple_loss=0.2823, pruned_loss=0.0519, over 16768.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2826, pruned_loss=0.05271, over 3192619.62 frames. ], batch size: 76, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:18:52,365 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 1.777e+02 2.085e+02 2.523e+02 4.139e+02, threshold=4.170e+02, percent-clipped=0.0 2023-05-01 15:19:11,722 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=227827.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:19:48,152 INFO [train.py:904] (7/8) Epoch 23, batch 4550, loss[loss=0.1936, simple_loss=0.2765, pruned_loss=0.05537, over 17093.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2841, pruned_loss=0.05397, over 3198146.23 frames. ], batch size: 49, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:20:39,347 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=227888.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:20:55,825 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3763, 1.8703, 2.8471, 3.3065, 3.0141, 3.6745, 1.9344, 3.7376], device='cuda:7'), covar=tensor([0.0163, 0.0625, 0.0314, 0.0254, 0.0282, 0.0143, 0.0737, 0.0102], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0182, 0.0187, 0.0201, 0.0158, 0.0198, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:21:00,788 INFO [train.py:904] (7/8) Epoch 23, batch 4600, loss[loss=0.1876, simple_loss=0.2748, pruned_loss=0.0502, over 15534.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2849, pruned_loss=0.0542, over 3203149.43 frames. ], batch size: 191, lr: 2.94e-03, grad_scale: 8.0 2023-05-01 15:21:18,035 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6784, 3.3891, 3.9378, 1.7703, 4.1399, 4.0971, 2.9676, 3.0277], device='cuda:7'), covar=tensor([0.0803, 0.0313, 0.0224, 0.1346, 0.0074, 0.0175, 0.0512, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0109, 0.0099, 0.0139, 0.0082, 0.0127, 0.0129, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:21:18,784 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.195e+02 1.806e+02 2.082e+02 2.426e+02 3.731e+02, threshold=4.163e+02, percent-clipped=0.0 2023-05-01 15:21:35,648 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3351, 5.3355, 5.1953, 4.8506, 4.8855, 5.2469, 5.0713, 4.9254], device='cuda:7'), covar=tensor([0.0473, 0.0283, 0.0221, 0.0251, 0.0802, 0.0265, 0.0261, 0.0531], device='cuda:7'), in_proj_covar=tensor([0.0299, 0.0439, 0.0350, 0.0349, 0.0355, 0.0404, 0.0238, 0.0417], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:21:53,240 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9403, 5.2819, 5.5013, 5.1545, 5.2839, 5.8587, 5.2934, 4.9730], device='cuda:7'), covar=tensor([0.1092, 0.1752, 0.1772, 0.1825, 0.2283, 0.0820, 0.1420, 0.2391], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0601, 0.0661, 0.0495, 0.0658, 0.0689, 0.0517, 0.0664], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 15:22:14,205 INFO [train.py:904] (7/8) Epoch 23, batch 4650, loss[loss=0.1868, simple_loss=0.267, pruned_loss=0.05326, over 16651.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2833, pruned_loss=0.05405, over 3221796.09 frames. ], batch size: 57, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:23:02,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2222, 4.3082, 4.1071, 3.8133, 3.8302, 4.2303, 3.8753, 3.9727], device='cuda:7'), covar=tensor([0.0546, 0.0497, 0.0264, 0.0291, 0.0728, 0.0406, 0.0902, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0438, 0.0349, 0.0349, 0.0353, 0.0404, 0.0238, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:23:28,870 INFO [train.py:904] (7/8) Epoch 23, batch 4700, loss[loss=0.1581, simple_loss=0.2573, pruned_loss=0.02942, over 16810.00 frames. ], tot_loss[loss=0.1927, simple_loss=0.2804, pruned_loss=0.05249, over 3225477.46 frames. ], batch size: 102, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:23:34,996 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228007.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:23:45,413 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 1.844e+02 2.189e+02 2.555e+02 4.222e+02, threshold=4.378e+02, percent-clipped=1.0 2023-05-01 15:24:35,573 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228049.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:24:41,827 INFO [train.py:904] (7/8) Epoch 23, batch 4750, loss[loss=0.1651, simple_loss=0.2561, pruned_loss=0.03708, over 16472.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2763, pruned_loss=0.05035, over 3221343.39 frames. ], batch size: 68, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:24:42,948 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9375, 1.3032, 1.7546, 1.7524, 1.8392, 1.9657, 1.5291, 1.9210], device='cuda:7'), covar=tensor([0.0243, 0.0493, 0.0246, 0.0285, 0.0294, 0.0208, 0.0602, 0.0171], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0182, 0.0188, 0.0201, 0.0158, 0.0199, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:25:02,217 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228067.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:25:03,371 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228068.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:25:19,541 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228080.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:25:54,062 INFO [train.py:904] (7/8) Epoch 23, batch 4800, loss[loss=0.1793, simple_loss=0.2764, pruned_loss=0.04104, over 15235.00 frames. ], tot_loss[loss=0.1849, simple_loss=0.2728, pruned_loss=0.04852, over 3203221.10 frames. ], batch size: 190, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:26:10,774 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.116e+02 1.852e+02 2.092e+02 2.435e+02 6.400e+02, threshold=4.184e+02, percent-clipped=1.0 2023-05-01 15:26:11,946 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228115.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:26:50,584 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228141.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:27:07,823 INFO [train.py:904] (7/8) Epoch 23, batch 4850, loss[loss=0.1713, simple_loss=0.2701, pruned_loss=0.03632, over 16666.00 frames. ], tot_loss[loss=0.1844, simple_loss=0.2735, pruned_loss=0.04764, over 3196216.00 frames. ], batch size: 83, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:27:42,714 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3785, 2.1845, 1.8109, 1.9158, 2.4872, 2.1586, 2.0610, 2.6358], device='cuda:7'), covar=tensor([0.0201, 0.0520, 0.0641, 0.0540, 0.0274, 0.0432, 0.0185, 0.0295], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0237, 0.0227, 0.0229, 0.0237, 0.0237, 0.0237, 0.0235], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:27:54,600 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228183.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:28:23,612 INFO [train.py:904] (7/8) Epoch 23, batch 4900, loss[loss=0.1785, simple_loss=0.2649, pruned_loss=0.04602, over 16668.00 frames. ], tot_loss[loss=0.1832, simple_loss=0.2728, pruned_loss=0.04683, over 3189594.64 frames. ], batch size: 68, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:28:42,125 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 1.946e+02 2.242e+02 2.632e+02 4.949e+02, threshold=4.484e+02, percent-clipped=2.0 2023-05-01 15:29:36,722 INFO [train.py:904] (7/8) Epoch 23, batch 4950, loss[loss=0.1631, simple_loss=0.2632, pruned_loss=0.0315, over 16821.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2714, pruned_loss=0.04535, over 3204613.34 frames. ], batch size: 96, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:30:51,758 INFO [train.py:904] (7/8) Epoch 23, batch 5000, loss[loss=0.1716, simple_loss=0.2682, pruned_loss=0.03743, over 16798.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2729, pruned_loss=0.0453, over 3202315.57 frames. ], batch size: 83, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:31:09,999 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 1.949e+02 2.305e+02 2.743e+02 3.734e+02, threshold=4.611e+02, percent-clipped=0.0 2023-05-01 15:31:18,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8036, 3.9266, 2.3869, 4.7759, 2.9949, 4.5758, 2.5654, 3.1414], device='cuda:7'), covar=tensor([0.0284, 0.0313, 0.1676, 0.0106, 0.0814, 0.0430, 0.1402, 0.0754], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0175, 0.0192, 0.0164, 0.0175, 0.0216, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:31:30,428 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228329.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:31:36,856 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228333.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:32:00,026 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228349.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:32:04,592 INFO [train.py:904] (7/8) Epoch 23, batch 5050, loss[loss=0.1833, simple_loss=0.2627, pruned_loss=0.05194, over 16677.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2736, pruned_loss=0.04527, over 3213932.02 frames. ], batch size: 57, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:32:19,779 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228363.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:32:39,618 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6653, 3.7522, 2.2871, 4.4198, 2.9040, 4.2868, 2.4200, 2.9827], device='cuda:7'), covar=tensor([0.0265, 0.0327, 0.1580, 0.0119, 0.0829, 0.0398, 0.1436, 0.0801], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0175, 0.0192, 0.0164, 0.0175, 0.0216, 0.0200, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:32:58,947 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228390.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:33:00,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0532, 2.1829, 2.2187, 3.6880, 2.0713, 2.5256, 2.2867, 2.3640], device='cuda:7'), covar=tensor([0.1444, 0.3635, 0.2997, 0.0575, 0.3981, 0.2624, 0.3592, 0.3191], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0455, 0.0371, 0.0330, 0.0436, 0.0523, 0.0424, 0.0530], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:33:04,500 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228394.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:33:09,601 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228397.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:33:17,247 INFO [train.py:904] (7/8) Epoch 23, batch 5100, loss[loss=0.1954, simple_loss=0.2842, pruned_loss=0.05331, over 16888.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2717, pruned_loss=0.04458, over 3219922.78 frames. ], batch size: 109, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:33:22,491 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8589, 4.7154, 4.9339, 5.1044, 5.3300, 4.7489, 5.3457, 5.3116], device='cuda:7'), covar=tensor([0.1947, 0.1344, 0.1634, 0.0675, 0.0455, 0.0747, 0.0459, 0.0561], device='cuda:7'), in_proj_covar=tensor([0.0642, 0.0796, 0.0918, 0.0802, 0.0611, 0.0637, 0.0658, 0.0766], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:33:34,756 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 1.954e+02 2.359e+02 2.853e+02 6.747e+02, threshold=4.718e+02, percent-clipped=3.0 2023-05-01 15:33:55,471 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228429.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:34:06,450 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228436.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:34:30,938 INFO [train.py:904] (7/8) Epoch 23, batch 5150, loss[loss=0.172, simple_loss=0.275, pruned_loss=0.03448, over 16232.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2717, pruned_loss=0.04439, over 3186574.17 frames. ], batch size: 165, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:35:13,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3936, 4.1265, 4.0720, 2.7186, 3.6468, 4.1299, 3.6383, 2.4180], device='cuda:7'), covar=tensor([0.0601, 0.0044, 0.0044, 0.0404, 0.0100, 0.0096, 0.0105, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0085, 0.0087, 0.0135, 0.0100, 0.0112, 0.0096, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 15:35:15,346 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228483.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:35:25,949 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228490.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:35:44,394 INFO [train.py:904] (7/8) Epoch 23, batch 5200, loss[loss=0.165, simple_loss=0.2534, pruned_loss=0.03829, over 16864.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2705, pruned_loss=0.04404, over 3186433.69 frames. ], batch size: 109, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:35:49,296 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6633, 2.6368, 2.4081, 4.1125, 2.8966, 3.8825, 1.5369, 2.8832], device='cuda:7'), covar=tensor([0.1418, 0.0803, 0.1303, 0.0133, 0.0256, 0.0417, 0.1740, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0176, 0.0196, 0.0192, 0.0205, 0.0215, 0.0204, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:36:01,316 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.022e+02 2.336e+02 2.717e+02 6.475e+02, threshold=4.673e+02, percent-clipped=1.0 2023-05-01 15:36:25,682 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228531.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:36:47,504 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8366, 1.4144, 1.6552, 1.7764, 1.9148, 1.9810, 1.5983, 1.8804], device='cuda:7'), covar=tensor([0.0249, 0.0442, 0.0250, 0.0334, 0.0301, 0.0242, 0.0470, 0.0158], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0189, 0.0202, 0.0158, 0.0199, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:36:57,638 INFO [train.py:904] (7/8) Epoch 23, batch 5250, loss[loss=0.1557, simple_loss=0.2502, pruned_loss=0.0306, over 16501.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.268, pruned_loss=0.04367, over 3182811.92 frames. ], batch size: 68, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:37:40,976 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5776, 5.5263, 5.4417, 5.0235, 5.0552, 5.4271, 5.3897, 5.1444], device='cuda:7'), covar=tensor([0.0594, 0.0452, 0.0270, 0.0270, 0.1059, 0.0478, 0.0224, 0.0604], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0439, 0.0348, 0.0345, 0.0352, 0.0405, 0.0238, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:38:10,579 INFO [train.py:904] (7/8) Epoch 23, batch 5300, loss[loss=0.1479, simple_loss=0.2324, pruned_loss=0.03165, over 17222.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2642, pruned_loss=0.04239, over 3197431.45 frames. ], batch size: 43, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:38:28,439 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 1.965e+02 2.226e+02 2.604e+02 5.338e+02, threshold=4.452e+02, percent-clipped=3.0 2023-05-01 15:38:36,093 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 15:39:02,146 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6533, 2.3920, 2.4121, 3.6246, 2.3804, 3.8067, 1.3987, 2.8001], device='cuda:7'), covar=tensor([0.1398, 0.0856, 0.1232, 0.0159, 0.0182, 0.0353, 0.1789, 0.0809], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0175, 0.0195, 0.0191, 0.0205, 0.0214, 0.0203, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:39:23,394 INFO [train.py:904] (7/8) Epoch 23, batch 5350, loss[loss=0.1743, simple_loss=0.2701, pruned_loss=0.03928, over 16397.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2627, pruned_loss=0.04179, over 3203322.19 frames. ], batch size: 146, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:39:38,585 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228663.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:10,728 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228685.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:16,858 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228689.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:36,799 INFO [train.py:904] (7/8) Epoch 23, batch 5400, loss[loss=0.1971, simple_loss=0.2824, pruned_loss=0.05589, over 16827.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2657, pruned_loss=0.04269, over 3214409.93 frames. ], batch size: 42, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:40:38,555 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 15:40:48,846 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228711.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:40:54,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 1.938e+02 2.190e+02 2.538e+02 4.586e+02, threshold=4.379e+02, percent-clipped=1.0 2023-05-01 15:40:55,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2375, 4.0167, 3.9860, 2.4604, 3.5293, 4.0110, 3.5940, 2.1291], device='cuda:7'), covar=tensor([0.0581, 0.0047, 0.0044, 0.0450, 0.0105, 0.0081, 0.0101, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0084, 0.0086, 0.0135, 0.0099, 0.0110, 0.0096, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 15:41:27,343 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228736.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:41:54,056 INFO [train.py:904] (7/8) Epoch 23, batch 5450, loss[loss=0.2131, simple_loss=0.2921, pruned_loss=0.06701, over 11881.00 frames. ], tot_loss[loss=0.1784, simple_loss=0.2686, pruned_loss=0.04412, over 3205458.16 frames. ], batch size: 246, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:42:43,125 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=228784.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:42:44,418 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=228785.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:43:12,594 INFO [train.py:904] (7/8) Epoch 23, batch 5500, loss[loss=0.2463, simple_loss=0.3222, pruned_loss=0.08517, over 15389.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2751, pruned_loss=0.04751, over 3190599.55 frames. ], batch size: 191, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:43:32,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.710e+02 2.602e+02 3.171e+02 3.709e+02 7.104e+02, threshold=6.343e+02, percent-clipped=12.0 2023-05-01 15:44:31,572 INFO [train.py:904] (7/8) Epoch 23, batch 5550, loss[loss=0.1779, simple_loss=0.2677, pruned_loss=0.04409, over 16986.00 frames. ], tot_loss[loss=0.1931, simple_loss=0.282, pruned_loss=0.0521, over 3177271.19 frames. ], batch size: 41, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:45:52,976 INFO [train.py:904] (7/8) Epoch 23, batch 5600, loss[loss=0.2003, simple_loss=0.2913, pruned_loss=0.05466, over 16347.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2866, pruned_loss=0.05609, over 3140277.90 frames. ], batch size: 146, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:46:10,854 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.072e+02 3.210e+02 3.734e+02 4.915e+02 1.270e+03, threshold=7.469e+02, percent-clipped=7.0 2023-05-01 15:46:32,422 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8338, 1.4009, 1.7306, 1.7589, 1.8618, 1.9655, 1.6579, 1.8647], device='cuda:7'), covar=tensor([0.0245, 0.0377, 0.0200, 0.0263, 0.0261, 0.0156, 0.0400, 0.0147], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0194, 0.0181, 0.0186, 0.0199, 0.0156, 0.0198, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:46:45,651 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=228935.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:47:13,523 INFO [train.py:904] (7/8) Epoch 23, batch 5650, loss[loss=0.2123, simple_loss=0.2916, pruned_loss=0.06652, over 15342.00 frames. ], tot_loss[loss=0.2054, simple_loss=0.2913, pruned_loss=0.05968, over 3129527.34 frames. ], batch size: 190, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:48:04,600 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228985.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:48:11,781 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=228989.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:48:22,401 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=228996.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:48:32,734 INFO [train.py:904] (7/8) Epoch 23, batch 5700, loss[loss=0.236, simple_loss=0.3274, pruned_loss=0.07229, over 16532.00 frames. ], tot_loss[loss=0.2079, simple_loss=0.2932, pruned_loss=0.06135, over 3100264.55 frames. ], batch size: 75, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:48:50,703 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.137e+02 3.035e+02 3.772e+02 4.443e+02 6.175e+02, threshold=7.544e+02, percent-clipped=0.0 2023-05-01 15:49:18,829 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229033.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:49:22,040 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0924, 2.4119, 2.3968, 2.6903, 2.0426, 3.1514, 1.8162, 2.7989], device='cuda:7'), covar=tensor([0.1095, 0.0561, 0.1020, 0.0186, 0.0126, 0.0365, 0.1459, 0.0648], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0175, 0.0196, 0.0192, 0.0206, 0.0215, 0.0204, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:49:25,417 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:49:30,371 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-01 15:49:31,567 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-01 15:49:50,906 INFO [train.py:904] (7/8) Epoch 23, batch 5750, loss[loss=0.2534, simple_loss=0.3083, pruned_loss=0.09923, over 11031.00 frames. ], tot_loss[loss=0.2114, simple_loss=0.2958, pruned_loss=0.06351, over 3052337.16 frames. ], batch size: 246, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:50:44,858 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=229085.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 15:51:13,985 INFO [train.py:904] (7/8) Epoch 23, batch 5800, loss[loss=0.1962, simple_loss=0.2782, pruned_loss=0.05711, over 11777.00 frames. ], tot_loss[loss=0.2094, simple_loss=0.295, pruned_loss=0.06195, over 3061732.75 frames. ], batch size: 246, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:51:32,217 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.995e+02 2.918e+02 3.323e+02 4.052e+02 8.500e+02, threshold=6.645e+02, percent-clipped=1.0 2023-05-01 15:52:01,503 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229133.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 15:52:31,392 INFO [train.py:904] (7/8) Epoch 23, batch 5850, loss[loss=0.202, simple_loss=0.2901, pruned_loss=0.05695, over 15324.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2933, pruned_loss=0.06025, over 3076157.26 frames. ], batch size: 190, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:53:22,309 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-01 15:53:53,147 INFO [train.py:904] (7/8) Epoch 23, batch 5900, loss[loss=0.1884, simple_loss=0.2847, pruned_loss=0.04605, over 16823.00 frames. ], tot_loss[loss=0.2058, simple_loss=0.2922, pruned_loss=0.05974, over 3076413.79 frames. ], batch size: 102, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:54:16,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.674e+02 2.520e+02 3.106e+02 3.673e+02 6.247e+02, threshold=6.213e+02, percent-clipped=0.0 2023-05-01 15:55:17,148 INFO [train.py:904] (7/8) Epoch 23, batch 5950, loss[loss=0.1802, simple_loss=0.2734, pruned_loss=0.04346, over 16642.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2932, pruned_loss=0.0587, over 3091340.89 frames. ], batch size: 62, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:56:09,070 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6502, 3.5199, 4.0203, 1.9840, 4.1415, 4.1815, 3.0836, 3.0215], device='cuda:7'), covar=tensor([0.0813, 0.0279, 0.0205, 0.1228, 0.0079, 0.0154, 0.0420, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0108, 0.0099, 0.0139, 0.0082, 0.0127, 0.0128, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:56:18,283 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229291.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:56:33,844 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229300.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:56:37,761 INFO [train.py:904] (7/8) Epoch 23, batch 6000, loss[loss=0.1924, simple_loss=0.2861, pruned_loss=0.04938, over 16276.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2925, pruned_loss=0.05842, over 3099659.18 frames. ], batch size: 165, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:56:37,762 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 15:56:49,496 INFO [train.py:938] (7/8) Epoch 23, validation: loss=0.1497, simple_loss=0.2623, pruned_loss=0.01859, over 944034.00 frames. 2023-05-01 15:56:49,497 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 15:57:07,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.923e+02 2.769e+02 3.211e+02 3.808e+02 9.716e+02, threshold=6.422e+02, percent-clipped=3.0 2023-05-01 15:57:08,367 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0904, 2.4811, 2.6108, 1.9207, 2.7533, 2.8211, 2.4540, 2.3167], device='cuda:7'), covar=tensor([0.0726, 0.0268, 0.0247, 0.1009, 0.0133, 0.0286, 0.0466, 0.0526], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0099, 0.0140, 0.0082, 0.0127, 0.0128, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 15:57:13,074 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9217, 4.9615, 4.8224, 4.4886, 4.4872, 4.8628, 4.6772, 4.5965], device='cuda:7'), covar=tensor([0.0711, 0.0623, 0.0288, 0.0337, 0.0945, 0.0519, 0.0442, 0.0682], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0439, 0.0346, 0.0343, 0.0351, 0.0402, 0.0236, 0.0412], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:57:27,293 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9458, 2.0115, 2.5879, 2.9738, 2.7529, 3.4295, 2.2217, 3.4231], device='cuda:7'), covar=tensor([0.0228, 0.0524, 0.0327, 0.0302, 0.0343, 0.0149, 0.0507, 0.0121], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0193, 0.0180, 0.0184, 0.0198, 0.0155, 0.0197, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:58:06,107 INFO [train.py:904] (7/8) Epoch 23, batch 6050, loss[loss=0.1813, simple_loss=0.2805, pruned_loss=0.04106, over 16691.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2916, pruned_loss=0.05797, over 3112600.38 frames. ], batch size: 134, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:58:21,090 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229361.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 15:58:54,733 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8871, 5.1829, 4.9423, 4.9797, 4.7292, 4.6718, 4.5620, 5.2829], device='cuda:7'), covar=tensor([0.1319, 0.0834, 0.0998, 0.0846, 0.0854, 0.0974, 0.1324, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0674, 0.0820, 0.0678, 0.0624, 0.0519, 0.0528, 0.0690, 0.0644], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 15:59:09,965 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 15:59:21,993 INFO [train.py:904] (7/8) Epoch 23, batch 6100, loss[loss=0.1765, simple_loss=0.2653, pruned_loss=0.04385, over 16601.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2908, pruned_loss=0.0573, over 3118512.82 frames. ], batch size: 57, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 15:59:40,513 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.924e+02 2.664e+02 3.486e+02 4.048e+02 8.551e+02, threshold=6.973e+02, percent-clipped=3.0 2023-05-01 16:00:22,355 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229443.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:00:36,342 INFO [train.py:904] (7/8) Epoch 23, batch 6150, loss[loss=0.2027, simple_loss=0.2904, pruned_loss=0.05751, over 16789.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2889, pruned_loss=0.05658, over 3126038.01 frames. ], batch size: 124, lr: 2.93e-03, grad_scale: 8.0 2023-05-01 16:01:15,805 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229477.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:01:53,549 INFO [train.py:904] (7/8) Epoch 23, batch 6200, loss[loss=0.1975, simple_loss=0.2832, pruned_loss=0.05587, over 15318.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2868, pruned_loss=0.05645, over 3117577.13 frames. ], batch size: 190, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:01:55,800 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229504.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:02:14,505 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.788e+02 3.260e+02 3.956e+02 6.763e+02, threshold=6.520e+02, percent-clipped=0.0 2023-05-01 16:02:38,231 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9710, 4.9362, 4.7677, 4.0303, 4.8592, 1.8570, 4.6140, 4.4515], device='cuda:7'), covar=tensor([0.0122, 0.0125, 0.0208, 0.0432, 0.0111, 0.2739, 0.0216, 0.0251], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0160, 0.0201, 0.0179, 0.0178, 0.0208, 0.0190, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:02:50,022 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:03:11,724 INFO [train.py:904] (7/8) Epoch 23, batch 6250, loss[loss=0.1744, simple_loss=0.2665, pruned_loss=0.04114, over 16546.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2865, pruned_loss=0.05664, over 3093438.41 frames. ], batch size: 68, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:03:49,507 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229577.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:04:09,781 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=229591.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:04:28,583 INFO [train.py:904] (7/8) Epoch 23, batch 6300, loss[loss=0.2002, simple_loss=0.2902, pruned_loss=0.05506, over 16796.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2862, pruned_loss=0.05592, over 3098883.71 frames. ], batch size: 83, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:04:50,640 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.619e+02 3.224e+02 4.167e+02 1.378e+03, threshold=6.449e+02, percent-clipped=8.0 2023-05-01 16:05:27,316 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229638.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:05:28,546 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=229639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:05:48,016 INFO [train.py:904] (7/8) Epoch 23, batch 6350, loss[loss=0.1952, simple_loss=0.2751, pruned_loss=0.05764, over 16653.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2875, pruned_loss=0.05752, over 3086329.25 frames. ], batch size: 134, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:05:50,507 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8313, 3.8537, 4.1331, 4.0969, 4.1087, 3.8943, 3.8922, 3.9231], device='cuda:7'), covar=tensor([0.0424, 0.0854, 0.0500, 0.0523, 0.0549, 0.0572, 0.0936, 0.0579], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0461, 0.0448, 0.0417, 0.0493, 0.0469, 0.0555, 0.0375], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 16:05:53,578 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229656.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:05:58,043 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229659.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:07:04,904 INFO [train.py:904] (7/8) Epoch 23, batch 6400, loss[loss=0.1945, simple_loss=0.287, pruned_loss=0.05096, over 17198.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2877, pruned_loss=0.05827, over 3088736.19 frames. ], batch size: 46, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:07:24,871 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.907e+02 2.905e+02 3.417e+02 4.252e+02 8.377e+02, threshold=6.835e+02, percent-clipped=2.0 2023-05-01 16:07:32,465 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229720.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 16:08:21,147 INFO [train.py:904] (7/8) Epoch 23, batch 6450, loss[loss=0.2056, simple_loss=0.2948, pruned_loss=0.05817, over 16856.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2875, pruned_loss=0.05734, over 3099528.73 frames. ], batch size: 116, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:08:58,156 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9823, 3.2290, 3.5122, 2.0533, 2.9454, 2.2164, 3.4911, 3.4602], device='cuda:7'), covar=tensor([0.0263, 0.0793, 0.0534, 0.2126, 0.0884, 0.1026, 0.0583, 0.0946], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0167, 0.0168, 0.0154, 0.0147, 0.0131, 0.0143, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:09:34,337 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229799.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:09:40,121 INFO [train.py:904] (7/8) Epoch 23, batch 6500, loss[loss=0.2191, simple_loss=0.2964, pruned_loss=0.07083, over 16924.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2861, pruned_loss=0.05705, over 3092330.48 frames. ], batch size: 109, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:09:59,298 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.807e+02 2.714e+02 3.280e+02 3.954e+02 7.541e+02, threshold=6.561e+02, percent-clipped=1.0 2023-05-01 16:10:25,487 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229833.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:10:57,914 INFO [train.py:904] (7/8) Epoch 23, batch 6550, loss[loss=0.1946, simple_loss=0.2945, pruned_loss=0.04733, over 16545.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2882, pruned_loss=0.05752, over 3096092.63 frames. ], batch size: 62, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:11:07,653 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6954, 4.8637, 5.0148, 4.8392, 4.9016, 5.4142, 4.9082, 4.6103], device='cuda:7'), covar=tensor([0.1085, 0.1754, 0.2178, 0.1904, 0.2124, 0.0907, 0.1497, 0.2368], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0598, 0.0658, 0.0492, 0.0655, 0.0688, 0.0515, 0.0663], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 16:12:02,442 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229894.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:12:15,590 INFO [train.py:904] (7/8) Epoch 23, batch 6600, loss[loss=0.216, simple_loss=0.3077, pruned_loss=0.0621, over 16894.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2897, pruned_loss=0.05764, over 3095585.71 frames. ], batch size: 96, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:12:35,467 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 2.706e+02 3.377e+02 4.261e+02 8.660e+02, threshold=6.754e+02, percent-clipped=4.0 2023-05-01 16:13:02,423 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=229933.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:13:21,451 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4989, 3.4578, 3.4591, 2.6756, 3.2875, 2.1119, 3.1213, 2.8401], device='cuda:7'), covar=tensor([0.0162, 0.0143, 0.0188, 0.0266, 0.0122, 0.2207, 0.0146, 0.0228], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0160, 0.0200, 0.0178, 0.0177, 0.0207, 0.0189, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:13:33,863 INFO [train.py:904] (7/8) Epoch 23, batch 6650, loss[loss=0.1801, simple_loss=0.2695, pruned_loss=0.04532, over 16710.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2898, pruned_loss=0.05856, over 3076349.72 frames. ], batch size: 68, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:13:37,607 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=229955.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:13:38,717 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=229956.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:14:26,157 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=229987.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:14:52,384 INFO [train.py:904] (7/8) Epoch 23, batch 6700, loss[loss=0.1879, simple_loss=0.2736, pruned_loss=0.05106, over 16286.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2884, pruned_loss=0.05848, over 3089543.72 frames. ], batch size: 165, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:14:55,508 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230004.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:15:04,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5440, 2.8826, 2.8951, 5.0374, 4.0945, 4.1641, 1.6088, 3.2607], device='cuda:7'), covar=tensor([0.1432, 0.0770, 0.1164, 0.0131, 0.0314, 0.0401, 0.1601, 0.0772], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0177, 0.0198, 0.0193, 0.0208, 0.0218, 0.0206, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:15:07,543 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2023-05-01 16:15:11,345 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230015.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:15:12,081 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.536e+02 3.063e+02 3.747e+02 8.042e+02, threshold=6.126e+02, percent-clipped=2.0 2023-05-01 16:15:33,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7107, 1.7710, 1.6164, 1.4966, 1.9298, 1.5860, 1.5978, 1.8749], device='cuda:7'), covar=tensor([0.0200, 0.0312, 0.0438, 0.0357, 0.0235, 0.0278, 0.0175, 0.0216], device='cuda:7'), in_proj_covar=tensor([0.0213, 0.0235, 0.0226, 0.0228, 0.0237, 0.0234, 0.0235, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:16:02,794 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230048.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:16:09,701 INFO [train.py:904] (7/8) Epoch 23, batch 6750, loss[loss=0.2244, simple_loss=0.2991, pruned_loss=0.07485, over 12001.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2883, pruned_loss=0.05897, over 3098488.07 frames. ], batch size: 248, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:17:19,760 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230099.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:17:24,164 INFO [train.py:904] (7/8) Epoch 23, batch 6800, loss[loss=0.2315, simple_loss=0.3134, pruned_loss=0.07476, over 16816.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2888, pruned_loss=0.05884, over 3114789.63 frames. ], batch size: 116, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:17:43,642 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 2.682e+02 3.194e+02 4.039e+02 7.212e+02, threshold=6.387e+02, percent-clipped=3.0 2023-05-01 16:18:11,097 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230133.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:18:31,928 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230147.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:18:36,460 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230150.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:18:40,520 INFO [train.py:904] (7/8) Epoch 23, batch 6850, loss[loss=0.2432, simple_loss=0.3173, pruned_loss=0.08457, over 11592.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2893, pruned_loss=0.05883, over 3109260.72 frames. ], batch size: 248, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:18:52,653 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.15 vs. limit=5.0 2023-05-01 16:19:23,365 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230181.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:19:56,860 INFO [train.py:904] (7/8) Epoch 23, batch 6900, loss[loss=0.1855, simple_loss=0.2807, pruned_loss=0.04518, over 16861.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2914, pruned_loss=0.05855, over 3118683.28 frames. ], batch size: 96, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:20:11,440 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:20:20,079 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.571e+02 3.126e+02 3.927e+02 7.395e+02, threshold=6.253e+02, percent-clipped=1.0 2023-05-01 16:20:36,659 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1333, 3.9776, 4.1949, 4.3272, 4.4589, 4.0294, 4.4101, 4.4819], device='cuda:7'), covar=tensor([0.1826, 0.1320, 0.1477, 0.0751, 0.0636, 0.1371, 0.0822, 0.0752], device='cuda:7'), in_proj_covar=tensor([0.0629, 0.0783, 0.0898, 0.0789, 0.0599, 0.0624, 0.0651, 0.0755], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:20:45,532 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230233.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:20:59,719 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230241.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:21:04,385 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5133, 3.7821, 3.5729, 1.8297, 2.9500, 2.1883, 3.7310, 3.9877], device='cuda:7'), covar=tensor([0.0241, 0.0739, 0.0692, 0.2550, 0.1053, 0.1180, 0.0626, 0.0953], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0167, 0.0169, 0.0155, 0.0147, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:21:13,682 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230250.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:21:18,000 INFO [train.py:904] (7/8) Epoch 23, batch 6950, loss[loss=0.2071, simple_loss=0.2903, pruned_loss=0.06194, over 16638.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2931, pruned_loss=0.06022, over 3100360.54 frames. ], batch size: 62, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:21:36,859 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3070, 1.6016, 2.0088, 2.2108, 2.3142, 2.5485, 1.7888, 2.4705], device='cuda:7'), covar=tensor([0.0250, 0.0565, 0.0318, 0.0385, 0.0346, 0.0230, 0.0560, 0.0171], device='cuda:7'), in_proj_covar=tensor([0.0190, 0.0193, 0.0179, 0.0185, 0.0199, 0.0156, 0.0198, 0.0154], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:22:01,759 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230281.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:22:34,168 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230302.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:22:34,887 INFO [train.py:904] (7/8) Epoch 23, batch 7000, loss[loss=0.1912, simple_loss=0.2868, pruned_loss=0.04783, over 16306.00 frames. ], tot_loss[loss=0.2063, simple_loss=0.2934, pruned_loss=0.05966, over 3078609.91 frames. ], batch size: 165, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:22:53,540 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230315.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:22:56,097 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.940e+02 2.703e+02 3.306e+02 4.193e+02 6.685e+02, threshold=6.612e+02, percent-clipped=2.0 2023-05-01 16:23:21,255 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0930, 5.1020, 4.9011, 4.2359, 5.0077, 1.8414, 4.7447, 4.5926], device='cuda:7'), covar=tensor([0.0094, 0.0097, 0.0213, 0.0386, 0.0100, 0.2795, 0.0144, 0.0229], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0159, 0.0200, 0.0177, 0.0176, 0.0206, 0.0188, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:23:35,810 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230343.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:23:36,071 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1399, 2.2195, 2.2521, 3.6823, 2.1289, 2.5854, 2.3045, 2.3349], device='cuda:7'), covar=tensor([0.1401, 0.3331, 0.2904, 0.0598, 0.4109, 0.2425, 0.3333, 0.3402], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0449, 0.0367, 0.0325, 0.0435, 0.0516, 0.0421, 0.0525], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:23:48,229 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4879, 3.4596, 3.4545, 2.7392, 3.2997, 2.1326, 3.1127, 2.8194], device='cuda:7'), covar=tensor([0.0176, 0.0148, 0.0186, 0.0234, 0.0113, 0.2337, 0.0146, 0.0260], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0159, 0.0200, 0.0177, 0.0176, 0.0206, 0.0188, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:23:50,832 INFO [train.py:904] (7/8) Epoch 23, batch 7050, loss[loss=0.196, simple_loss=0.2929, pruned_loss=0.04956, over 16482.00 frames. ], tot_loss[loss=0.2066, simple_loss=0.2942, pruned_loss=0.05948, over 3078206.66 frames. ], batch size: 75, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:24:05,850 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230363.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:25:04,988 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 16:25:07,485 INFO [train.py:904] (7/8) Epoch 23, batch 7100, loss[loss=0.1874, simple_loss=0.2835, pruned_loss=0.04561, over 16905.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2931, pruned_loss=0.05965, over 3057458.73 frames. ], batch size: 96, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:25:09,409 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2871, 3.4761, 3.5480, 2.1553, 3.0091, 2.3925, 3.7444, 3.7950], device='cuda:7'), covar=tensor([0.0268, 0.0849, 0.0619, 0.2129, 0.0913, 0.1003, 0.0579, 0.0932], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0167, 0.0169, 0.0155, 0.0147, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:25:30,798 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.925e+02 3.270e+02 3.973e+02 7.550e+02, threshold=6.541e+02, percent-clipped=2.0 2023-05-01 16:25:38,723 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4327, 3.5160, 2.1175, 3.9849, 2.7072, 3.9613, 2.2344, 2.7910], device='cuda:7'), covar=tensor([0.0284, 0.0360, 0.1632, 0.0195, 0.0795, 0.0563, 0.1610, 0.0802], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0174, 0.0192, 0.0162, 0.0174, 0.0215, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:26:22,863 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9298, 2.8325, 2.7915, 2.1167, 2.6824, 2.2226, 2.7507, 3.0159], device='cuda:7'), covar=tensor([0.0308, 0.0650, 0.0571, 0.1780, 0.0792, 0.0845, 0.0600, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0169, 0.0155, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:26:24,970 INFO [train.py:904] (7/8) Epoch 23, batch 7150, loss[loss=0.2714, simple_loss=0.3308, pruned_loss=0.106, over 11932.00 frames. ], tot_loss[loss=0.205, simple_loss=0.2912, pruned_loss=0.05937, over 3068855.06 frames. ], batch size: 247, lr: 2.92e-03, grad_scale: 4.0 2023-05-01 16:26:49,337 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230468.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:27:42,221 INFO [train.py:904] (7/8) Epoch 23, batch 7200, loss[loss=0.1954, simple_loss=0.2908, pruned_loss=0.05001, over 16283.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2885, pruned_loss=0.05714, over 3079595.61 frames. ], batch size: 146, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:27:47,862 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230506.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:27:49,204 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5210, 4.5361, 4.3742, 3.1461, 4.4561, 1.4956, 4.1252, 4.0554], device='cuda:7'), covar=tensor([0.0189, 0.0161, 0.0295, 0.0809, 0.0166, 0.3851, 0.0245, 0.0436], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0159, 0.0199, 0.0176, 0.0175, 0.0206, 0.0187, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:28:03,477 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.863e+02 2.554e+02 3.131e+02 3.804e+02 6.997e+02, threshold=6.261e+02, percent-clipped=1.0 2023-05-01 16:28:16,871 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230525.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:28:23,120 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230529.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 16:28:23,393 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 16:28:30,109 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230533.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:28:53,525 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9213, 2.7861, 2.6981, 1.9556, 2.5747, 2.7318, 2.6139, 1.9242], device='cuda:7'), covar=tensor([0.0480, 0.0091, 0.0090, 0.0403, 0.0144, 0.0142, 0.0138, 0.0417], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0086, 0.0087, 0.0136, 0.0100, 0.0112, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 16:28:59,055 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230550.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:29:00,337 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8256, 3.8432, 4.1207, 4.0897, 4.1313, 3.8974, 3.9063, 3.8967], device='cuda:7'), covar=tensor([0.0369, 0.0627, 0.0430, 0.0472, 0.0430, 0.0426, 0.0862, 0.0520], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0462, 0.0447, 0.0415, 0.0491, 0.0469, 0.0556, 0.0374], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 16:29:02,535 INFO [train.py:904] (7/8) Epoch 23, batch 7250, loss[loss=0.1628, simple_loss=0.2542, pruned_loss=0.0357, over 16775.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2861, pruned_loss=0.05591, over 3085921.85 frames. ], batch size: 83, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:29:25,640 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 16:29:34,262 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7905, 3.8583, 4.1176, 4.0859, 4.1257, 3.8880, 3.8883, 3.8912], device='cuda:7'), covar=tensor([0.0407, 0.0664, 0.0478, 0.0485, 0.0500, 0.0475, 0.0954, 0.0550], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0462, 0.0447, 0.0415, 0.0492, 0.0470, 0.0556, 0.0374], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 16:29:43,937 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4581, 3.5366, 2.0423, 4.0367, 2.6552, 3.9817, 2.2712, 2.7504], device='cuda:7'), covar=tensor([0.0332, 0.0406, 0.1854, 0.0215, 0.0907, 0.0682, 0.1627, 0.0941], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0176, 0.0194, 0.0163, 0.0175, 0.0216, 0.0202, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:29:51,471 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230586.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:03,628 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:05,311 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7842, 4.7906, 4.6081, 3.9170, 4.7176, 1.5815, 4.4840, 4.2589], device='cuda:7'), covar=tensor([0.0102, 0.0088, 0.0208, 0.0384, 0.0091, 0.3133, 0.0153, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0159, 0.0199, 0.0177, 0.0175, 0.0206, 0.0187, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:30:08,953 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230597.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:10,103 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230598.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:30:18,077 INFO [train.py:904] (7/8) Epoch 23, batch 7300, loss[loss=0.1844, simple_loss=0.2767, pruned_loss=0.04604, over 16742.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2863, pruned_loss=0.05591, over 3087914.48 frames. ], batch size: 89, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:30:39,633 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.739e+02 2.620e+02 3.228e+02 4.011e+02 8.387e+02, threshold=6.456e+02, percent-clipped=2.0 2023-05-01 16:31:19,692 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230643.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:31:33,677 INFO [train.py:904] (7/8) Epoch 23, batch 7350, loss[loss=0.1799, simple_loss=0.2715, pruned_loss=0.04416, over 16503.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2874, pruned_loss=0.05671, over 3072695.08 frames. ], batch size: 68, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:32:33,479 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230691.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:32:53,499 INFO [train.py:904] (7/8) Epoch 23, batch 7400, loss[loss=0.2147, simple_loss=0.2935, pruned_loss=0.06797, over 11270.00 frames. ], tot_loss[loss=0.2018, simple_loss=0.2888, pruned_loss=0.05736, over 3067764.76 frames. ], batch size: 248, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:33:16,064 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.016e+02 2.706e+02 3.245e+02 4.052e+02 5.999e+02, threshold=6.490e+02, percent-clipped=0.0 2023-05-01 16:34:13,433 INFO [train.py:904] (7/8) Epoch 23, batch 7450, loss[loss=0.197, simple_loss=0.2809, pruned_loss=0.05653, over 17062.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2899, pruned_loss=0.0587, over 3054157.17 frames. ], batch size: 50, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:35:05,683 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5543, 3.5978, 2.7542, 2.1541, 2.3918, 2.3903, 3.8094, 3.2163], device='cuda:7'), covar=tensor([0.3096, 0.0698, 0.1944, 0.2926, 0.2742, 0.2096, 0.0566, 0.1368], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0271, 0.0308, 0.0318, 0.0300, 0.0264, 0.0300, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 16:35:19,270 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230792.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:35:35,348 INFO [train.py:904] (7/8) Epoch 23, batch 7500, loss[loss=0.2129, simple_loss=0.2928, pruned_loss=0.06644, over 16608.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2905, pruned_loss=0.05868, over 3035360.95 frames. ], batch size: 134, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:35:40,979 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230806.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:35:58,127 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.010e+02 2.839e+02 3.579e+02 4.414e+02 1.002e+03, threshold=7.158e+02, percent-clipped=2.0 2023-05-01 16:36:09,726 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230824.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 16:36:52,039 INFO [train.py:904] (7/8) Epoch 23, batch 7550, loss[loss=0.1825, simple_loss=0.273, pruned_loss=0.04602, over 16531.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2886, pruned_loss=0.058, over 3054410.46 frames. ], batch size: 75, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:36:52,568 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230853.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:36:53,665 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230854.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:32,281 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3110, 3.9808, 4.0055, 2.5136, 3.5921, 4.0730, 3.5367, 2.2174], device='cuda:7'), covar=tensor([0.0581, 0.0060, 0.0055, 0.0459, 0.0120, 0.0121, 0.0121, 0.0473], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0085, 0.0086, 0.0135, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 16:37:33,362 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230881.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:38,972 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 16:37:42,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9377, 4.0284, 2.6602, 4.8584, 3.2330, 4.7669, 2.9155, 3.2627], device='cuda:7'), covar=tensor([0.0311, 0.0386, 0.1585, 0.0274, 0.0770, 0.0413, 0.1281, 0.0756], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0176, 0.0194, 0.0164, 0.0176, 0.0217, 0.0203, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:37:45,799 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=230889.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:47,873 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=230890.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:37:58,365 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=230897.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:38:06,492 INFO [train.py:904] (7/8) Epoch 23, batch 7600, loss[loss=0.1845, simple_loss=0.2722, pruned_loss=0.04841, over 16509.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2879, pruned_loss=0.05819, over 3071995.04 frames. ], batch size: 62, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:38:27,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 2.696e+02 3.168e+02 3.892e+02 6.127e+02, threshold=6.336e+02, percent-clipped=0.0 2023-05-01 16:38:43,030 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 16:39:10,346 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=230945.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:39:19,785 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5691, 4.7399, 4.8956, 4.6792, 4.7139, 5.2589, 4.7470, 4.4916], device='cuda:7'), covar=tensor([0.1312, 0.1935, 0.2453, 0.1949, 0.2511, 0.0970, 0.1703, 0.2538], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0602, 0.0664, 0.0495, 0.0661, 0.0690, 0.0518, 0.0663], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 16:39:19,886 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=230951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:39:21,942 INFO [train.py:904] (7/8) Epoch 23, batch 7650, loss[loss=0.1916, simple_loss=0.2782, pruned_loss=0.05247, over 16397.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2885, pruned_loss=0.05888, over 3072420.37 frames. ], batch size: 35, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:39:43,785 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 16:40:26,150 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9968, 2.8499, 2.7716, 2.0278, 2.6573, 2.1180, 2.7950, 3.0187], device='cuda:7'), covar=tensor([0.0263, 0.0836, 0.0571, 0.2079, 0.0935, 0.1043, 0.0646, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0167, 0.0169, 0.0155, 0.0147, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:40:35,949 INFO [train.py:904] (7/8) Epoch 23, batch 7700, loss[loss=0.2017, simple_loss=0.2857, pruned_loss=0.05886, over 15207.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2887, pruned_loss=0.05957, over 3061189.09 frames. ], batch size: 190, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:40:43,382 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4546, 2.6658, 2.6699, 4.4414, 2.5408, 2.9608, 2.6506, 2.8373], device='cuda:7'), covar=tensor([0.1327, 0.3163, 0.2561, 0.0462, 0.3667, 0.2208, 0.3171, 0.2933], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0449, 0.0367, 0.0324, 0.0433, 0.0516, 0.0420, 0.0524], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:40:57,684 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 2.849e+02 3.493e+02 4.384e+02 8.642e+02, threshold=6.986e+02, percent-clipped=7.0 2023-05-01 16:41:28,233 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5789, 2.6213, 2.7528, 4.4215, 3.4068, 4.0058, 1.5205, 3.2579], device='cuda:7'), covar=tensor([0.1558, 0.0836, 0.1176, 0.0181, 0.0352, 0.0434, 0.1816, 0.0760], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0177, 0.0198, 0.0194, 0.0209, 0.0217, 0.0206, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:41:31,537 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3575, 3.0495, 3.3902, 1.8517, 3.4854, 3.5894, 2.8402, 2.6584], device='cuda:7'), covar=tensor([0.0800, 0.0305, 0.0221, 0.1222, 0.0096, 0.0179, 0.0464, 0.0535], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0098, 0.0138, 0.0081, 0.0125, 0.0127, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 16:41:53,918 INFO [train.py:904] (7/8) Epoch 23, batch 7750, loss[loss=0.2561, simple_loss=0.312, pruned_loss=0.1001, over 11180.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2887, pruned_loss=0.05928, over 3073313.80 frames. ], batch size: 250, lr: 2.92e-03, grad_scale: 8.0 2023-05-01 16:42:37,557 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2023-05-01 16:43:09,657 INFO [train.py:904] (7/8) Epoch 23, batch 7800, loss[loss=0.2493, simple_loss=0.311, pruned_loss=0.09377, over 11747.00 frames. ], tot_loss[loss=0.2039, simple_loss=0.2892, pruned_loss=0.05925, over 3081585.58 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:43:30,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.091e+02 2.907e+02 3.422e+02 4.020e+02 9.124e+02, threshold=6.845e+02, percent-clipped=1.0 2023-05-01 16:43:41,784 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231124.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:44:17,272 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=231148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:44:24,903 INFO [train.py:904] (7/8) Epoch 23, batch 7850, loss[loss=0.1959, simple_loss=0.2853, pruned_loss=0.05327, over 16491.00 frames. ], tot_loss[loss=0.2038, simple_loss=0.2896, pruned_loss=0.05903, over 3076134.79 frames. ], batch size: 68, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:44:52,237 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231172.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:45:05,985 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231181.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:45:16,804 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231189.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:45:37,214 INFO [train.py:904] (7/8) Epoch 23, batch 7900, loss[loss=0.1834, simple_loss=0.2769, pruned_loss=0.04494, over 16761.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2889, pruned_loss=0.05841, over 3098172.54 frames. ], batch size: 76, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:45:57,448 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.744e+02 2.682e+02 3.251e+02 4.238e+02 6.670e+02, threshold=6.501e+02, percent-clipped=0.0 2023-05-01 16:46:16,886 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231229.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:31,397 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231237.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:44,794 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=231246.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:52,818 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=231251.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:46:54,853 INFO [train.py:904] (7/8) Epoch 23, batch 7950, loss[loss=0.2504, simple_loss=0.3097, pruned_loss=0.09558, over 11869.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2893, pruned_loss=0.05885, over 3101258.79 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:48:11,524 INFO [train.py:904] (7/8) Epoch 23, batch 8000, loss[loss=0.2075, simple_loss=0.295, pruned_loss=0.06001, over 16922.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2903, pruned_loss=0.05964, over 3106138.49 frames. ], batch size: 116, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:48:26,473 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=231312.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:48:33,452 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.670e+02 3.387e+02 4.030e+02 6.156e+02, threshold=6.774e+02, percent-clipped=0.0 2023-05-01 16:49:26,918 INFO [train.py:904] (7/8) Epoch 23, batch 8050, loss[loss=0.2199, simple_loss=0.2924, pruned_loss=0.07365, over 11777.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2901, pruned_loss=0.05933, over 3092610.12 frames. ], batch size: 247, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:50:04,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9996, 3.0529, 1.7218, 3.2215, 2.3209, 3.2418, 1.9459, 2.4692], device='cuda:7'), covar=tensor([0.0308, 0.0414, 0.1949, 0.0258, 0.0890, 0.0653, 0.1742, 0.0846], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0174, 0.0193, 0.0163, 0.0175, 0.0216, 0.0201, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:50:09,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3447, 3.4582, 3.6116, 1.9874, 3.2154, 2.5122, 3.7524, 3.8066], device='cuda:7'), covar=tensor([0.0234, 0.0853, 0.0596, 0.2275, 0.0837, 0.0937, 0.0583, 0.0950], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0165, 0.0168, 0.0154, 0.0146, 0.0130, 0.0143, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:50:11,514 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6745, 4.6570, 4.5170, 3.8232, 4.5899, 1.5789, 4.3785, 4.2322], device='cuda:7'), covar=tensor([0.0102, 0.0096, 0.0195, 0.0363, 0.0092, 0.2976, 0.0141, 0.0253], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0159, 0.0200, 0.0177, 0.0176, 0.0208, 0.0188, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:50:42,769 INFO [train.py:904] (7/8) Epoch 23, batch 8100, loss[loss=0.2111, simple_loss=0.2923, pruned_loss=0.06492, over 16564.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2895, pruned_loss=0.05819, over 3113712.19 frames. ], batch size: 62, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:51:04,337 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.823e+02 2.502e+02 3.046e+02 3.676e+02 7.002e+02, threshold=6.092e+02, percent-clipped=1.0 2023-05-01 16:51:16,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1945, 4.3377, 4.5061, 4.2391, 4.3803, 4.8253, 4.3543, 4.0879], device='cuda:7'), covar=tensor([0.1872, 0.1989, 0.2257, 0.2026, 0.2461, 0.1103, 0.1789, 0.2541], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0602, 0.0665, 0.0497, 0.0662, 0.0691, 0.0518, 0.0665], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 16:51:51,846 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231448.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:51:59,319 INFO [train.py:904] (7/8) Epoch 23, batch 8150, loss[loss=0.1878, simple_loss=0.2697, pruned_loss=0.05296, over 16739.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2865, pruned_loss=0.05728, over 3117923.88 frames. ], batch size: 124, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:53:05,170 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231496.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:53:14,930 INFO [train.py:904] (7/8) Epoch 23, batch 8200, loss[loss=0.2006, simple_loss=0.289, pruned_loss=0.05611, over 15275.00 frames. ], tot_loss[loss=0.1987, simple_loss=0.284, pruned_loss=0.05668, over 3116207.42 frames. ], batch size: 190, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:53:38,111 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.780e+02 3.369e+02 4.151e+02 6.479e+02, threshold=6.737e+02, percent-clipped=3.0 2023-05-01 16:54:18,813 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 16:54:26,264 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:54:37,512 INFO [train.py:904] (7/8) Epoch 23, batch 8250, loss[loss=0.188, simple_loss=0.2854, pruned_loss=0.04527, over 16452.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2833, pruned_loss=0.05414, over 3109774.00 frames. ], batch size: 146, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:55:43,801 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231594.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:55:57,553 INFO [train.py:904] (7/8) Epoch 23, batch 8300, loss[loss=0.1692, simple_loss=0.2531, pruned_loss=0.04268, over 12017.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2808, pruned_loss=0.05133, over 3092660.19 frames. ], batch size: 247, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 16:56:05,057 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=231607.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:56:22,159 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 2.236e+02 2.512e+02 2.973e+02 5.364e+02, threshold=5.024e+02, percent-clipped=0.0 2023-05-01 16:56:29,056 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0508, 4.0153, 3.9504, 3.1016, 3.9691, 1.7406, 3.7644, 3.5174], device='cuda:7'), covar=tensor([0.0120, 0.0118, 0.0183, 0.0263, 0.0102, 0.3145, 0.0147, 0.0306], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0159, 0.0200, 0.0177, 0.0176, 0.0208, 0.0188, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 16:56:42,881 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6865, 3.4272, 3.7671, 1.9710, 3.8683, 3.9558, 3.1449, 3.0538], device='cuda:7'), covar=tensor([0.0635, 0.0220, 0.0181, 0.1126, 0.0083, 0.0153, 0.0341, 0.0398], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0106, 0.0096, 0.0136, 0.0080, 0.0123, 0.0124, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 16:57:11,273 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=231647.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 16:57:14,767 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6731, 2.6410, 1.9570, 2.8157, 2.1740, 2.8247, 2.1703, 2.4595], device='cuda:7'), covar=tensor([0.0289, 0.0304, 0.1188, 0.0295, 0.0654, 0.0455, 0.1198, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0175, 0.0193, 0.0163, 0.0176, 0.0216, 0.0203, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:57:19,290 INFO [train.py:904] (7/8) Epoch 23, batch 8350, loss[loss=0.1778, simple_loss=0.2819, pruned_loss=0.03689, over 16707.00 frames. ], tot_loss[loss=0.1897, simple_loss=0.2804, pruned_loss=0.04951, over 3091432.70 frames. ], batch size: 89, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 16:58:23,399 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9648, 2.8008, 2.4375, 3.7241, 2.3439, 3.8584, 1.7356, 2.9027], device='cuda:7'), covar=tensor([0.1226, 0.0657, 0.1166, 0.0186, 0.0115, 0.0487, 0.1485, 0.0738], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0174, 0.0195, 0.0190, 0.0205, 0.0214, 0.0202, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 16:58:41,948 INFO [train.py:904] (7/8) Epoch 23, batch 8400, loss[loss=0.1773, simple_loss=0.2646, pruned_loss=0.04502, over 12125.00 frames. ], tot_loss[loss=0.1868, simple_loss=0.2782, pruned_loss=0.04773, over 3079398.76 frames. ], batch size: 246, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 16:58:50,848 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=231708.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 16:59:06,592 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.191e+02 2.741e+02 3.304e+02 6.516e+02, threshold=5.483e+02, percent-clipped=5.0 2023-05-01 17:00:04,866 INFO [train.py:904] (7/8) Epoch 23, batch 8450, loss[loss=0.1754, simple_loss=0.2696, pruned_loss=0.0406, over 16055.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2761, pruned_loss=0.0459, over 3087497.71 frames. ], batch size: 35, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:01:24,909 INFO [train.py:904] (7/8) Epoch 23, batch 8500, loss[loss=0.1575, simple_loss=0.2368, pruned_loss=0.03908, over 11592.00 frames. ], tot_loss[loss=0.1806, simple_loss=0.2731, pruned_loss=0.04406, over 3086582.30 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:01:48,518 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.174e+02 2.671e+02 3.276e+02 7.093e+02, threshold=5.341e+02, percent-clipped=2.0 2023-05-01 17:02:48,234 INFO [train.py:904] (7/8) Epoch 23, batch 8550, loss[loss=0.1973, simple_loss=0.2913, pruned_loss=0.05161, over 15289.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.271, pruned_loss=0.04316, over 3078166.94 frames. ], batch size: 191, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:03:04,772 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1226, 2.0473, 2.1076, 3.7060, 2.0685, 2.4010, 2.1837, 2.2006], device='cuda:7'), covar=tensor([0.1301, 0.4200, 0.3260, 0.0564, 0.4583, 0.2840, 0.4004, 0.3849], device='cuda:7'), in_proj_covar=tensor([0.0394, 0.0444, 0.0364, 0.0320, 0.0428, 0.0509, 0.0414, 0.0517], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:03:09,024 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5766, 3.6457, 3.3924, 3.0455, 3.2031, 3.5222, 3.3383, 3.3812], device='cuda:7'), covar=tensor([0.0589, 0.0707, 0.0308, 0.0302, 0.0510, 0.0528, 0.1282, 0.0546], device='cuda:7'), in_proj_covar=tensor([0.0286, 0.0428, 0.0332, 0.0333, 0.0338, 0.0390, 0.0229, 0.0401], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:04:28,278 INFO [train.py:904] (7/8) Epoch 23, batch 8600, loss[loss=0.1622, simple_loss=0.2531, pruned_loss=0.03567, over 12701.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2717, pruned_loss=0.04242, over 3073687.89 frames. ], batch size: 246, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:04:36,605 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=231907.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:05:00,383 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.309e+02 2.625e+02 3.302e+02 5.565e+02, threshold=5.250e+02, percent-clipped=1.0 2023-05-01 17:05:08,978 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4336, 4.5325, 4.3489, 3.9573, 4.0239, 4.4128, 4.1324, 4.1385], device='cuda:7'), covar=tensor([0.0543, 0.0472, 0.0280, 0.0314, 0.0742, 0.0467, 0.0659, 0.0680], device='cuda:7'), in_proj_covar=tensor([0.0284, 0.0426, 0.0331, 0.0331, 0.0336, 0.0387, 0.0228, 0.0399], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:06:03,608 INFO [train.py:904] (7/8) Epoch 23, batch 8650, loss[loss=0.1631, simple_loss=0.2665, pruned_loss=0.02983, over 16424.00 frames. ], tot_loss[loss=0.1757, simple_loss=0.2695, pruned_loss=0.04095, over 3070287.69 frames. ], batch size: 146, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:06:10,285 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=231955.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:07:29,600 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-01 17:07:37,576 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 17:07:53,209 INFO [train.py:904] (7/8) Epoch 23, batch 8700, loss[loss=0.1756, simple_loss=0.2576, pruned_loss=0.04685, over 12258.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.2668, pruned_loss=0.04014, over 3043860.94 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:07:54,844 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232003.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 17:08:13,671 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4684, 3.5106, 3.7225, 3.6979, 3.6962, 3.5407, 3.5684, 3.6147], device='cuda:7'), covar=tensor([0.0454, 0.1006, 0.0606, 0.0621, 0.0659, 0.0740, 0.0800, 0.0478], device='cuda:7'), in_proj_covar=tensor([0.0406, 0.0452, 0.0440, 0.0407, 0.0484, 0.0458, 0.0543, 0.0368], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 17:08:23,061 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.142e+02 2.520e+02 3.226e+02 6.058e+02, threshold=5.039e+02, percent-clipped=2.0 2023-05-01 17:09:27,797 INFO [train.py:904] (7/8) Epoch 23, batch 8750, loss[loss=0.1799, simple_loss=0.2791, pruned_loss=0.04036, over 16146.00 frames. ], tot_loss[loss=0.1734, simple_loss=0.2671, pruned_loss=0.03987, over 3055294.17 frames. ], batch size: 165, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:10:47,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5089, 4.3528, 4.5845, 4.6701, 4.8654, 4.3895, 4.8911, 4.8861], device='cuda:7'), covar=tensor([0.1832, 0.1305, 0.1479, 0.0795, 0.0524, 0.0963, 0.0616, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0618, 0.0765, 0.0880, 0.0775, 0.0589, 0.0617, 0.0643, 0.0750], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:10:53,024 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6976, 4.9874, 4.7891, 4.7718, 4.5328, 4.4805, 4.3841, 5.0619], device='cuda:7'), covar=tensor([0.1145, 0.0848, 0.0959, 0.0825, 0.0788, 0.1197, 0.1255, 0.0847], device='cuda:7'), in_proj_covar=tensor([0.0667, 0.0810, 0.0670, 0.0621, 0.0513, 0.0528, 0.0682, 0.0636], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:11:15,152 INFO [train.py:904] (7/8) Epoch 23, batch 8800, loss[loss=0.1585, simple_loss=0.2566, pruned_loss=0.03021, over 16806.00 frames. ], tot_loss[loss=0.1712, simple_loss=0.2653, pruned_loss=0.03854, over 3072085.20 frames. ], batch size: 90, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:11:46,654 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.145e+02 2.530e+02 3.020e+02 7.216e+02, threshold=5.060e+02, percent-clipped=1.0 2023-05-01 17:12:57,919 INFO [train.py:904] (7/8) Epoch 23, batch 8850, loss[loss=0.1865, simple_loss=0.2864, pruned_loss=0.04329, over 16788.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2676, pruned_loss=0.0381, over 3054484.64 frames. ], batch size: 124, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:13:30,972 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6839, 2.0630, 1.7756, 1.9042, 2.4108, 2.0466, 2.0549, 2.4771], device='cuda:7'), covar=tensor([0.0194, 0.0482, 0.0560, 0.0496, 0.0316, 0.0448, 0.0202, 0.0297], device='cuda:7'), in_proj_covar=tensor([0.0206, 0.0230, 0.0220, 0.0223, 0.0230, 0.0229, 0.0227, 0.0225], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:14:42,080 INFO [train.py:904] (7/8) Epoch 23, batch 8900, loss[loss=0.194, simple_loss=0.2908, pruned_loss=0.04861, over 15510.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2674, pruned_loss=0.03736, over 3038748.73 frames. ], batch size: 192, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:15:12,057 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.232e+02 2.655e+02 3.319e+02 5.405e+02, threshold=5.309e+02, percent-clipped=1.0 2023-05-01 17:16:33,805 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232247.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:16:45,334 INFO [train.py:904] (7/8) Epoch 23, batch 8950, loss[loss=0.1702, simple_loss=0.2641, pruned_loss=0.03814, over 16342.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2665, pruned_loss=0.03721, over 3057237.38 frames. ], batch size: 146, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:17:58,913 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 17:18:14,778 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 17:18:31,859 INFO [train.py:904] (7/8) Epoch 23, batch 9000, loss[loss=0.1514, simple_loss=0.246, pruned_loss=0.0284, over 12189.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2635, pruned_loss=0.03591, over 3045962.81 frames. ], batch size: 250, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:18:31,860 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 17:18:42,674 INFO [train.py:938] (7/8) Epoch 23, validation: loss=0.1452, simple_loss=0.249, pruned_loss=0.02066, over 944034.00 frames. 2023-05-01 17:18:42,675 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 17:18:43,887 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=232303.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 17:18:54,251 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232308.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:19:18,068 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 1.963e+02 2.332e+02 2.758e+02 5.487e+02, threshold=4.665e+02, percent-clipped=1.0 2023-05-01 17:19:40,986 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 17:20:24,509 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=232351.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:20:28,123 INFO [train.py:904] (7/8) Epoch 23, batch 9050, loss[loss=0.1714, simple_loss=0.2614, pruned_loss=0.04067, over 16662.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2645, pruned_loss=0.03661, over 3041537.47 frames. ], batch size: 62, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:20:50,509 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4226, 2.2978, 2.2026, 4.1551, 2.1831, 2.6788, 2.3556, 2.4291], device='cuda:7'), covar=tensor([0.1131, 0.3465, 0.3279, 0.0423, 0.4363, 0.2399, 0.3851, 0.3203], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0446, 0.0367, 0.0320, 0.0431, 0.0510, 0.0417, 0.0518], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:21:02,911 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8198, 2.7941, 2.6172, 1.9974, 2.5518, 2.8107, 2.6565, 1.7959], device='cuda:7'), covar=tensor([0.0528, 0.0087, 0.0081, 0.0404, 0.0136, 0.0117, 0.0106, 0.0537], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0083, 0.0084, 0.0132, 0.0097, 0.0108, 0.0093, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 17:21:23,586 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232381.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:21:45,195 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1613, 2.1557, 2.0857, 3.7040, 2.0644, 2.4940, 2.2772, 2.2592], device='cuda:7'), covar=tensor([0.1243, 0.3680, 0.3371, 0.0559, 0.4412, 0.2473, 0.3755, 0.3464], device='cuda:7'), in_proj_covar=tensor([0.0396, 0.0444, 0.0365, 0.0320, 0.0429, 0.0508, 0.0416, 0.0516], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:21:47,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0451, 4.1336, 3.9566, 3.6310, 3.6693, 4.0591, 3.7242, 3.8278], device='cuda:7'), covar=tensor([0.0603, 0.0660, 0.0380, 0.0344, 0.0689, 0.0546, 0.1053, 0.0687], device='cuda:7'), in_proj_covar=tensor([0.0282, 0.0419, 0.0329, 0.0328, 0.0332, 0.0383, 0.0225, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 17:22:12,834 INFO [train.py:904] (7/8) Epoch 23, batch 9100, loss[loss=0.1666, simple_loss=0.275, pruned_loss=0.02917, over 16913.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2643, pruned_loss=0.03715, over 3045106.82 frames. ], batch size: 90, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:22:15,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5745, 4.5502, 4.3899, 3.7758, 4.4859, 1.6338, 4.2523, 4.2046], device='cuda:7'), covar=tensor([0.0108, 0.0134, 0.0201, 0.0375, 0.0107, 0.2884, 0.0152, 0.0242], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0158, 0.0198, 0.0174, 0.0174, 0.0207, 0.0186, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:22:46,302 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.157e+02 2.616e+02 3.171e+02 8.876e+02, threshold=5.231e+02, percent-clipped=3.0 2023-05-01 17:23:41,051 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8401, 3.7066, 3.9089, 3.9902, 4.0839, 3.7261, 4.0411, 4.1276], device='cuda:7'), covar=tensor([0.1601, 0.1162, 0.1299, 0.0718, 0.0561, 0.1781, 0.0755, 0.0714], device='cuda:7'), in_proj_covar=tensor([0.0620, 0.0767, 0.0880, 0.0777, 0.0589, 0.0617, 0.0643, 0.0751], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:23:46,943 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232442.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 17:24:08,727 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8092, 2.7106, 2.6162, 1.9521, 2.4891, 2.7482, 2.5872, 1.9033], device='cuda:7'), covar=tensor([0.0465, 0.0085, 0.0086, 0.0384, 0.0160, 0.0116, 0.0116, 0.0479], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0084, 0.0084, 0.0133, 0.0098, 0.0109, 0.0094, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 17:24:09,375 INFO [train.py:904] (7/8) Epoch 23, batch 9150, loss[loss=0.1817, simple_loss=0.2614, pruned_loss=0.05103, over 11842.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2644, pruned_loss=0.03655, over 3053190.08 frames. ], batch size: 250, lr: 2.91e-03, grad_scale: 4.0 2023-05-01 17:25:00,477 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232476.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:25:07,463 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0654, 4.1440, 3.9589, 3.6794, 3.7193, 4.0633, 3.7530, 3.8716], device='cuda:7'), covar=tensor([0.0565, 0.0610, 0.0295, 0.0292, 0.0648, 0.0464, 0.0963, 0.0600], device='cuda:7'), in_proj_covar=tensor([0.0283, 0.0419, 0.0329, 0.0328, 0.0332, 0.0383, 0.0225, 0.0396], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 17:25:46,783 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2258, 3.3893, 3.6688, 2.2008, 3.0975, 2.4471, 3.6153, 3.6007], device='cuda:7'), covar=tensor([0.0244, 0.0843, 0.0495, 0.1975, 0.0812, 0.0966, 0.0626, 0.0933], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0159, 0.0163, 0.0150, 0.0142, 0.0127, 0.0140, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 17:25:48,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7085, 3.0134, 3.3506, 1.9574, 2.8785, 2.1693, 3.1994, 3.2158], device='cuda:7'), covar=tensor([0.0348, 0.0942, 0.0517, 0.2193, 0.0865, 0.1077, 0.0748, 0.1001], device='cuda:7'), in_proj_covar=tensor([0.0152, 0.0159, 0.0163, 0.0150, 0.0142, 0.0127, 0.0140, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 17:25:52,925 INFO [train.py:904] (7/8) Epoch 23, batch 9200, loss[loss=0.1729, simple_loss=0.2615, pruned_loss=0.04214, over 15321.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.261, pruned_loss=0.03617, over 3063072.15 frames. ], batch size: 192, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:26:21,485 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232518.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:26:24,114 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.032e+02 2.532e+02 2.962e+02 8.451e+02, threshold=5.064e+02, percent-clipped=2.0 2023-05-01 17:26:44,446 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 17:26:57,142 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232537.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:27:27,742 INFO [train.py:904] (7/8) Epoch 23, batch 9250, loss[loss=0.1634, simple_loss=0.247, pruned_loss=0.03988, over 12258.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2601, pruned_loss=0.03623, over 3036376.26 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:28:23,255 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=232579.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:29:17,468 INFO [train.py:904] (7/8) Epoch 23, batch 9300, loss[loss=0.169, simple_loss=0.2464, pruned_loss=0.04582, over 12342.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2581, pruned_loss=0.03529, over 3033308.75 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:29:18,427 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232603.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:29:58,639 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.317e+02 2.074e+02 2.482e+02 2.986e+02 5.209e+02, threshold=4.963e+02, percent-clipped=1.0 2023-05-01 17:30:49,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5784, 4.7292, 4.4742, 4.0797, 4.0217, 4.6203, 4.4091, 4.2172], device='cuda:7'), covar=tensor([0.0669, 0.0648, 0.0430, 0.0425, 0.1217, 0.0561, 0.0548, 0.0809], device='cuda:7'), in_proj_covar=tensor([0.0285, 0.0421, 0.0331, 0.0329, 0.0333, 0.0385, 0.0227, 0.0397], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:31:04,802 INFO [train.py:904] (7/8) Epoch 23, batch 9350, loss[loss=0.1596, simple_loss=0.248, pruned_loss=0.03558, over 12303.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.258, pruned_loss=0.03526, over 3032331.09 frames. ], batch size: 248, lr: 2.91e-03, grad_scale: 8.0 2023-05-01 17:31:50,483 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-01 17:32:47,940 INFO [train.py:904] (7/8) Epoch 23, batch 9400, loss[loss=0.1589, simple_loss=0.2467, pruned_loss=0.03561, over 12563.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2581, pruned_loss=0.03524, over 3017318.62 frames. ], batch size: 248, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:33:21,842 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.048e+02 2.308e+02 2.875e+02 5.038e+02, threshold=4.615e+02, percent-clipped=1.0 2023-05-01 17:33:57,949 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232737.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 17:34:29,663 INFO [train.py:904] (7/8) Epoch 23, batch 9450, loss[loss=0.1686, simple_loss=0.2591, pruned_loss=0.03905, over 16924.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2596, pruned_loss=0.03541, over 3020437.08 frames. ], batch size: 116, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:35:04,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 17:36:09,228 INFO [train.py:904] (7/8) Epoch 23, batch 9500, loss[loss=0.1725, simple_loss=0.2677, pruned_loss=0.03861, over 16237.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2592, pruned_loss=0.0353, over 3023642.67 frames. ], batch size: 165, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:36:18,550 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3681, 4.2041, 4.4315, 4.5393, 4.6854, 4.2445, 4.7028, 4.7167], device='cuda:7'), covar=tensor([0.1784, 0.1243, 0.1483, 0.0754, 0.0527, 0.1059, 0.0567, 0.0722], device='cuda:7'), in_proj_covar=tensor([0.0610, 0.0756, 0.0866, 0.0765, 0.0581, 0.0607, 0.0635, 0.0740], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:36:44,543 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.194e+02 2.525e+02 3.286e+02 6.197e+02, threshold=5.051e+02, percent-clipped=6.0 2023-05-01 17:37:08,361 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232832.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:37:54,049 INFO [train.py:904] (7/8) Epoch 23, batch 9550, loss[loss=0.1869, simple_loss=0.2899, pruned_loss=0.04196, over 15263.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2591, pruned_loss=0.03544, over 3024675.01 frames. ], batch size: 191, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:38:13,691 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1202, 3.3112, 3.3793, 1.6838, 3.4873, 3.7084, 3.0355, 2.6685], device='cuda:7'), covar=tensor([0.1076, 0.0196, 0.0199, 0.1404, 0.0076, 0.0144, 0.0382, 0.0596], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0103, 0.0093, 0.0134, 0.0078, 0.0119, 0.0122, 0.0124], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 17:38:38,052 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=232874.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:39:05,525 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3205, 4.3369, 4.7044, 4.7003, 4.7073, 4.4192, 4.4166, 4.3644], device='cuda:7'), covar=tensor([0.0484, 0.0938, 0.0679, 0.0723, 0.0747, 0.0676, 0.1020, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0442, 0.0434, 0.0397, 0.0474, 0.0450, 0.0530, 0.0362], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 17:39:09,282 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1260, 2.1440, 2.1756, 3.7123, 2.0916, 2.4933, 2.2710, 2.2762], device='cuda:7'), covar=tensor([0.1205, 0.3741, 0.3074, 0.0535, 0.4344, 0.2579, 0.3649, 0.3482], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0443, 0.0365, 0.0318, 0.0429, 0.0506, 0.0415, 0.0514], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:39:11,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3415, 4.3053, 4.1910, 3.5538, 4.2445, 1.6502, 4.0639, 3.9605], device='cuda:7'), covar=tensor([0.0120, 0.0123, 0.0212, 0.0303, 0.0125, 0.2920, 0.0152, 0.0276], device='cuda:7'), in_proj_covar=tensor([0.0165, 0.0157, 0.0196, 0.0172, 0.0173, 0.0206, 0.0185, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:39:33,912 INFO [train.py:904] (7/8) Epoch 23, batch 9600, loss[loss=0.1674, simple_loss=0.2555, pruned_loss=0.03966, over 12501.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2608, pruned_loss=0.03614, over 3031530.58 frames. ], batch size: 248, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:39:34,763 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=232903.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:40:05,775 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.131e+02 2.488e+02 2.932e+02 6.217e+02, threshold=4.975e+02, percent-clipped=3.0 2023-05-01 17:40:07,403 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7202, 3.6899, 3.9950, 1.7860, 4.1544, 4.2670, 3.2469, 3.1124], device='cuda:7'), covar=tensor([0.0798, 0.0253, 0.0218, 0.1495, 0.0083, 0.0139, 0.0379, 0.0501], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0103, 0.0093, 0.0134, 0.0078, 0.0120, 0.0122, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 17:40:12,527 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9857, 5.2871, 5.0890, 5.0622, 4.7960, 4.7448, 4.6713, 5.4069], device='cuda:7'), covar=tensor([0.1222, 0.0945, 0.1053, 0.0893, 0.0801, 0.1052, 0.1257, 0.0886], device='cuda:7'), in_proj_covar=tensor([0.0664, 0.0806, 0.0666, 0.0619, 0.0511, 0.0524, 0.0679, 0.0637], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:40:15,354 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-01 17:40:27,921 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5105, 2.0297, 1.7222, 1.7119, 2.2561, 1.9252, 1.9145, 2.3353], device='cuda:7'), covar=tensor([0.0190, 0.0435, 0.0553, 0.0519, 0.0289, 0.0417, 0.0196, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0227, 0.0219, 0.0220, 0.0227, 0.0228, 0.0222, 0.0221], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:41:09,513 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2582, 3.9911, 4.0448, 4.3976, 4.5656, 4.2060, 4.6063, 4.5736], device='cuda:7'), covar=tensor([0.1884, 0.1489, 0.2531, 0.1170, 0.0879, 0.1607, 0.0900, 0.1199], device='cuda:7'), in_proj_covar=tensor([0.0612, 0.0757, 0.0868, 0.0767, 0.0583, 0.0607, 0.0635, 0.0741], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:41:17,233 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=232951.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:41:20,706 INFO [train.py:904] (7/8) Epoch 23, batch 9650, loss[loss=0.161, simple_loss=0.2599, pruned_loss=0.03102, over 15210.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2621, pruned_loss=0.03613, over 3038242.40 frames. ], batch size: 190, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:41:48,128 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.16 vs. limit=2.0 2023-05-01 17:42:39,059 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0497, 5.3256, 5.1320, 5.1210, 4.8503, 4.7599, 4.6643, 5.4116], device='cuda:7'), covar=tensor([0.1211, 0.0842, 0.0997, 0.0834, 0.0692, 0.0958, 0.1196, 0.0893], device='cuda:7'), in_proj_covar=tensor([0.0663, 0.0804, 0.0663, 0.0616, 0.0510, 0.0521, 0.0677, 0.0635], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:42:42,000 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=232990.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:43:09,190 INFO [train.py:904] (7/8) Epoch 23, batch 9700, loss[loss=0.1745, simple_loss=0.2787, pruned_loss=0.03514, over 15266.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2614, pruned_loss=0.03622, over 3049940.49 frames. ], batch size: 190, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:43:40,110 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233018.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:43:43,788 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.165e+02 2.462e+02 3.116e+02 5.593e+02, threshold=4.924e+02, percent-clipped=3.0 2023-05-01 17:44:17,993 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0580, 1.8731, 1.6701, 1.4963, 2.0100, 1.6752, 1.6630, 1.9411], device='cuda:7'), covar=tensor([0.0163, 0.0327, 0.0437, 0.0393, 0.0238, 0.0276, 0.0153, 0.0217], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0227, 0.0219, 0.0220, 0.0227, 0.0227, 0.0221, 0.0221], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:44:22,606 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:44:50,834 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233051.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 17:44:53,545 INFO [train.py:904] (7/8) Epoch 23, batch 9750, loss[loss=0.1468, simple_loss=0.2356, pruned_loss=0.02899, over 12283.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2598, pruned_loss=0.03602, over 3066390.43 frames. ], batch size: 247, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:44:54,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4331, 3.5154, 2.7447, 2.1126, 2.2158, 2.3748, 3.7353, 3.0869], device='cuda:7'), covar=tensor([0.3329, 0.0660, 0.1861, 0.3048, 0.3018, 0.2272, 0.0411, 0.1478], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0261, 0.0299, 0.0308, 0.0286, 0.0256, 0.0289, 0.0329], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 17:45:14,265 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8076, 2.6445, 2.2597, 3.9055, 2.2803, 3.7347, 1.4875, 2.7008], device='cuda:7'), covar=tensor([0.1295, 0.0725, 0.1279, 0.0168, 0.0111, 0.0461, 0.1641, 0.0828], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0172, 0.0194, 0.0185, 0.0198, 0.0211, 0.0202, 0.0191], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 17:45:43,987 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233079.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:45:59,983 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233085.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:46:29,609 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2181, 4.3121, 4.1470, 3.8519, 3.8776, 4.2268, 3.9067, 3.9517], device='cuda:7'), covar=tensor([0.0587, 0.0654, 0.0314, 0.0286, 0.0779, 0.0494, 0.0800, 0.0592], device='cuda:7'), in_proj_covar=tensor([0.0281, 0.0415, 0.0328, 0.0326, 0.0329, 0.0380, 0.0224, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 17:46:31,587 INFO [train.py:904] (7/8) Epoch 23, batch 9800, loss[loss=0.1457, simple_loss=0.2368, pruned_loss=0.02729, over 12239.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2602, pruned_loss=0.03503, over 3086173.32 frames. ], batch size: 247, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:47:03,696 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.589e+02 2.119e+02 2.583e+02 3.356e+02 7.260e+02, threshold=5.167e+02, percent-clipped=1.0 2023-05-01 17:47:27,294 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233132.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:47:35,964 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9333, 4.9242, 5.3128, 5.2813, 5.3244, 5.0466, 4.9785, 4.8621], device='cuda:7'), covar=tensor([0.0354, 0.0651, 0.0421, 0.0403, 0.0476, 0.0397, 0.0895, 0.0401], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0443, 0.0434, 0.0397, 0.0473, 0.0450, 0.0529, 0.0363], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 17:47:48,030 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5056, 4.6077, 4.4046, 4.0832, 4.1658, 4.5150, 4.2839, 4.2153], device='cuda:7'), covar=tensor([0.0538, 0.0412, 0.0292, 0.0307, 0.0714, 0.0400, 0.0515, 0.0602], device='cuda:7'), in_proj_covar=tensor([0.0280, 0.0412, 0.0327, 0.0324, 0.0327, 0.0378, 0.0222, 0.0389], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 17:48:10,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9870, 2.7207, 2.8584, 2.0861, 2.7032, 2.1935, 2.6720, 2.9159], device='cuda:7'), covar=tensor([0.0371, 0.0888, 0.0556, 0.1886, 0.0874, 0.0933, 0.0686, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0157, 0.0163, 0.0150, 0.0142, 0.0126, 0.0139, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 17:48:16,812 INFO [train.py:904] (7/8) Epoch 23, batch 9850, loss[loss=0.1587, simple_loss=0.2604, pruned_loss=0.02849, over 16379.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.261, pruned_loss=0.03467, over 3083734.68 frames. ], batch size: 146, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:49:01,263 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233174.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:49:13,903 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233180.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:50:08,901 INFO [train.py:904] (7/8) Epoch 23, batch 9900, loss[loss=0.1686, simple_loss=0.2698, pruned_loss=0.03367, over 16606.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2616, pruned_loss=0.03454, over 3086588.73 frames. ], batch size: 62, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:50:21,530 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0397, 3.0102, 1.9484, 3.2974, 2.2659, 3.2928, 2.1735, 2.5979], device='cuda:7'), covar=tensor([0.0326, 0.0447, 0.1622, 0.0266, 0.0846, 0.0607, 0.1487, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0168, 0.0186, 0.0156, 0.0169, 0.0205, 0.0197, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-01 17:50:46,348 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.034e+02 2.487e+02 3.058e+02 5.815e+02, threshold=4.974e+02, percent-clipped=3.0 2023-05-01 17:50:53,601 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233222.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:51:09,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1730, 3.1707, 1.9579, 3.5107, 2.3915, 3.4921, 2.1750, 2.6365], device='cuda:7'), covar=tensor([0.0320, 0.0421, 0.1682, 0.0233, 0.0870, 0.0513, 0.1531, 0.0782], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0168, 0.0186, 0.0155, 0.0169, 0.0205, 0.0197, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-01 17:51:45,257 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4059, 1.6731, 2.0118, 2.3638, 2.4007, 2.6314, 1.7940, 2.6078], device='cuda:7'), covar=tensor([0.0217, 0.0540, 0.0370, 0.0318, 0.0336, 0.0191, 0.0567, 0.0155], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0187, 0.0173, 0.0177, 0.0192, 0.0149, 0.0192, 0.0147], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:52:06,040 INFO [train.py:904] (7/8) Epoch 23, batch 9950, loss[loss=0.1626, simple_loss=0.2565, pruned_loss=0.03431, over 16547.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2635, pruned_loss=0.03497, over 3085168.11 frames. ], batch size: 68, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:52:10,798 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9652, 2.6910, 2.8874, 1.9759, 2.7267, 2.1440, 2.7113, 2.8898], device='cuda:7'), covar=tensor([0.0291, 0.0908, 0.0550, 0.1901, 0.0773, 0.0950, 0.0622, 0.0800], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0156, 0.0162, 0.0149, 0.0141, 0.0125, 0.0139, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 17:52:18,777 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0239, 1.7658, 1.5910, 1.4263, 1.9782, 1.6356, 1.5861, 1.9446], device='cuda:7'), covar=tensor([0.0251, 0.0404, 0.0537, 0.0470, 0.0273, 0.0319, 0.0207, 0.0233], device='cuda:7'), in_proj_covar=tensor([0.0205, 0.0229, 0.0221, 0.0223, 0.0229, 0.0230, 0.0223, 0.0223], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 17:52:21,024 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:53:11,749 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-05-01 17:54:01,236 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 17:54:07,454 INFO [train.py:904] (7/8) Epoch 23, batch 10000, loss[loss=0.1633, simple_loss=0.2559, pruned_loss=0.03538, over 13006.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2626, pruned_loss=0.03495, over 3088643.97 frames. ], batch size: 248, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:54:40,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.128e+02 2.777e+02 3.344e+02 7.240e+02, threshold=5.555e+02, percent-clipped=2.0 2023-05-01 17:54:41,863 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233320.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:55:26,224 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2409, 4.2998, 4.1044, 3.7898, 3.8626, 4.2038, 3.8796, 3.9469], device='cuda:7'), covar=tensor([0.0536, 0.0638, 0.0313, 0.0306, 0.0677, 0.0521, 0.0771, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0278, 0.0411, 0.0325, 0.0323, 0.0326, 0.0377, 0.0222, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:7') 2023-05-01 17:55:36,810 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233346.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 17:55:47,980 INFO [train.py:904] (7/8) Epoch 23, batch 10050, loss[loss=0.1667, simple_loss=0.2642, pruned_loss=0.03457, over 15221.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2627, pruned_loss=0.03513, over 3075545.15 frames. ], batch size: 191, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:56:31,056 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233374.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 17:57:20,842 INFO [train.py:904] (7/8) Epoch 23, batch 10100, loss[loss=0.1553, simple_loss=0.2501, pruned_loss=0.03021, over 16721.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2629, pruned_loss=0.03532, over 3054252.36 frames. ], batch size: 76, lr: 2.90e-03, grad_scale: 8.0 2023-05-01 17:57:53,981 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.206e+02 2.577e+02 3.013e+02 4.747e+02, threshold=5.154e+02, percent-clipped=0.0 2023-05-01 17:59:06,335 INFO [train.py:904] (7/8) Epoch 24, batch 0, loss[loss=0.1617, simple_loss=0.2538, pruned_loss=0.03481, over 17278.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2538, pruned_loss=0.03481, over 17278.00 frames. ], batch size: 52, lr: 2.84e-03, grad_scale: 8.0 2023-05-01 17:59:06,335 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 17:59:14,245 INFO [train.py:938] (7/8) Epoch 24, validation: loss=0.145, simple_loss=0.2483, pruned_loss=0.02085, over 944034.00 frames. 2023-05-01 17:59:14,246 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 18:00:06,228 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7103, 2.6085, 2.3067, 2.4958, 2.9745, 2.6602, 3.2968, 3.1632], device='cuda:7'), covar=tensor([0.0170, 0.0455, 0.0560, 0.0505, 0.0346, 0.0482, 0.0254, 0.0330], device='cuda:7'), in_proj_covar=tensor([0.0207, 0.0231, 0.0222, 0.0224, 0.0230, 0.0231, 0.0225, 0.0224], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:00:23,698 INFO [train.py:904] (7/8) Epoch 24, batch 50, loss[loss=0.2083, simple_loss=0.2885, pruned_loss=0.0641, over 16706.00 frames. ], tot_loss[loss=0.1858, simple_loss=0.2733, pruned_loss=0.04913, over 742790.97 frames. ], batch size: 134, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:00:52,611 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.410e+02 2.907e+02 3.501e+02 5.437e+02, threshold=5.814e+02, percent-clipped=4.0 2023-05-01 18:01:32,990 INFO [train.py:904] (7/8) Epoch 24, batch 100, loss[loss=0.1821, simple_loss=0.2619, pruned_loss=0.05112, over 16756.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2649, pruned_loss=0.04565, over 1317021.83 frames. ], batch size: 83, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:01:53,778 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6801, 3.8157, 2.4470, 4.3544, 3.0434, 4.2357, 2.4912, 3.1503], device='cuda:7'), covar=tensor([0.0326, 0.0419, 0.1580, 0.0326, 0.0830, 0.0622, 0.1638, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0166, 0.0171, 0.0189, 0.0159, 0.0172, 0.0210, 0.0199, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:02:40,665 INFO [train.py:904] (7/8) Epoch 24, batch 150, loss[loss=0.1551, simple_loss=0.2377, pruned_loss=0.03626, over 17207.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2631, pruned_loss=0.04528, over 1753715.66 frames. ], batch size: 44, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:02:56,767 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:03:08,349 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.737e+02 2.160e+02 2.597e+02 3.268e+02 6.927e+02, threshold=5.194e+02, percent-clipped=3.0 2023-05-01 18:03:30,598 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233639.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:03:39,450 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233646.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:03:48,630 INFO [train.py:904] (7/8) Epoch 24, batch 200, loss[loss=0.1788, simple_loss=0.2575, pruned_loss=0.05009, over 16847.00 frames. ], tot_loss[loss=0.1763, simple_loss=0.2631, pruned_loss=0.04473, over 2106202.16 frames. ], batch size: 102, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:04:18,064 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233674.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:04:34,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0397, 2.2800, 2.5656, 3.0073, 2.8842, 3.4703, 2.3084, 3.4799], device='cuda:7'), covar=tensor([0.0288, 0.0532, 0.0371, 0.0358, 0.0357, 0.0211, 0.0538, 0.0179], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0192, 0.0178, 0.0181, 0.0197, 0.0153, 0.0196, 0.0151], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:04:45,850 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233694.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:04:54,361 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233700.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 18:04:58,754 INFO [train.py:904] (7/8) Epoch 24, batch 250, loss[loss=0.1585, simple_loss=0.2619, pruned_loss=0.02757, over 16739.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2601, pruned_loss=0.04448, over 2376146.01 frames. ], batch size: 57, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:05:07,854 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=233710.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:05:25,969 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233722.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:05:26,869 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.284e+02 2.752e+02 3.292e+02 5.130e+02, threshold=5.503e+02, percent-clipped=0.0 2023-05-01 18:05:48,954 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 18:06:08,093 INFO [train.py:904] (7/8) Epoch 24, batch 300, loss[loss=0.1532, simple_loss=0.2374, pruned_loss=0.03447, over 16694.00 frames. ], tot_loss[loss=0.1714, simple_loss=0.2572, pruned_loss=0.04276, over 2588675.92 frames. ], batch size: 76, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:06:33,486 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=233771.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:06:45,327 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6177, 2.5203, 2.6543, 4.6048, 2.4807, 2.8401, 2.6245, 2.6600], device='cuda:7'), covar=tensor([0.1258, 0.3781, 0.2988, 0.0440, 0.4104, 0.2677, 0.3595, 0.3568], device='cuda:7'), in_proj_covar=tensor([0.0400, 0.0450, 0.0372, 0.0324, 0.0435, 0.0513, 0.0422, 0.0524], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:06:57,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3693, 3.9160, 4.3888, 2.3292, 4.5599, 4.6490, 3.4740, 3.5576], device='cuda:7'), covar=tensor([0.0590, 0.0254, 0.0208, 0.1143, 0.0070, 0.0158, 0.0394, 0.0394], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0106, 0.0096, 0.0138, 0.0080, 0.0124, 0.0125, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 18:07:16,410 INFO [train.py:904] (7/8) Epoch 24, batch 350, loss[loss=0.1618, simple_loss=0.2632, pruned_loss=0.03019, over 17052.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2545, pruned_loss=0.04119, over 2744917.68 frames. ], batch size: 50, lr: 2.84e-03, grad_scale: 1.0 2023-05-01 18:07:43,749 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 2.126e+02 2.409e+02 2.975e+02 6.590e+02, threshold=4.818e+02, percent-clipped=2.0 2023-05-01 18:08:10,073 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 18:08:25,450 INFO [train.py:904] (7/8) Epoch 24, batch 400, loss[loss=0.182, simple_loss=0.2589, pruned_loss=0.05257, over 16427.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2532, pruned_loss=0.04105, over 2868531.98 frames. ], batch size: 146, lr: 2.84e-03, grad_scale: 2.0 2023-05-01 18:09:08,270 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6937, 6.0838, 5.8224, 5.8572, 5.4499, 5.5489, 5.4822, 6.2031], device='cuda:7'), covar=tensor([0.1417, 0.1024, 0.1063, 0.0817, 0.0955, 0.0647, 0.1296, 0.0906], device='cuda:7'), in_proj_covar=tensor([0.0681, 0.0825, 0.0680, 0.0633, 0.0524, 0.0532, 0.0698, 0.0650], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:09:32,985 INFO [train.py:904] (7/8) Epoch 24, batch 450, loss[loss=0.1546, simple_loss=0.2357, pruned_loss=0.03671, over 16499.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2523, pruned_loss=0.04062, over 2964938.40 frames. ], batch size: 75, lr: 2.84e-03, grad_scale: 2.0 2023-05-01 18:09:47,264 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1063, 3.7520, 4.2480, 2.2458, 4.3913, 4.4982, 3.2903, 3.4241], device='cuda:7'), covar=tensor([0.0707, 0.0279, 0.0241, 0.1201, 0.0091, 0.0201, 0.0461, 0.0430], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0107, 0.0097, 0.0139, 0.0081, 0.0125, 0.0126, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 18:09:50,207 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=233915.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:09:59,639 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.087e+02 2.389e+02 2.800e+02 4.652e+02, threshold=4.779e+02, percent-clipped=0.0 2023-05-01 18:10:39,005 INFO [train.py:904] (7/8) Epoch 24, batch 500, loss[loss=0.1523, simple_loss=0.2398, pruned_loss=0.03245, over 16350.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2513, pruned_loss=0.03999, over 3049741.37 frames. ], batch size: 36, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:10:48,838 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-01 18:10:54,543 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=233963.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:11:20,719 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-01 18:11:23,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7766, 4.9342, 5.0617, 4.9043, 4.9032, 5.5134, 5.0313, 4.7055], device='cuda:7'), covar=tensor([0.1410, 0.2040, 0.2663, 0.2195, 0.2656, 0.1060, 0.1882, 0.2819], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0599, 0.0662, 0.0496, 0.0662, 0.0691, 0.0515, 0.0660], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 18:11:37,683 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=233995.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:11:50,843 INFO [train.py:904] (7/8) Epoch 24, batch 550, loss[loss=0.1625, simple_loss=0.2491, pruned_loss=0.03798, over 16449.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2504, pruned_loss=0.03945, over 3106709.41 frames. ], batch size: 75, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:12:17,098 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.103e+02 2.436e+02 3.122e+02 5.008e+02, threshold=4.871e+02, percent-clipped=2.0 2023-05-01 18:12:20,376 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234025.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:12:36,738 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4860, 3.4408, 3.9963, 2.1298, 3.2999, 2.5612, 3.9057, 3.6733], device='cuda:7'), covar=tensor([0.0310, 0.1078, 0.0518, 0.2195, 0.0856, 0.1042, 0.0616, 0.1219], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0163, 0.0167, 0.0154, 0.0146, 0.0129, 0.0143, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:12:57,937 INFO [train.py:904] (7/8) Epoch 24, batch 600, loss[loss=0.1725, simple_loss=0.2546, pruned_loss=0.04518, over 16482.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2494, pruned_loss=0.03911, over 3165361.93 frames. ], batch size: 146, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:13:17,464 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234066.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:13:45,362 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234086.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:14:06,106 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 18:14:08,037 INFO [train.py:904] (7/8) Epoch 24, batch 650, loss[loss=0.1496, simple_loss=0.2312, pruned_loss=0.03402, over 15419.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2482, pruned_loss=0.03821, over 3196941.87 frames. ], batch size: 190, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:14:10,993 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-05-01 18:14:36,191 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 2.015e+02 2.288e+02 2.913e+02 5.815e+02, threshold=4.575e+02, percent-clipped=2.0 2023-05-01 18:15:16,431 INFO [train.py:904] (7/8) Epoch 24, batch 700, loss[loss=0.1506, simple_loss=0.2337, pruned_loss=0.0338, over 17040.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2479, pruned_loss=0.03778, over 3230647.35 frames. ], batch size: 41, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:15:24,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5492, 5.9845, 5.6966, 5.7662, 5.2774, 5.3496, 5.3617, 6.1179], device='cuda:7'), covar=tensor([0.1561, 0.0987, 0.1084, 0.0847, 0.1030, 0.0733, 0.1275, 0.0979], device='cuda:7'), in_proj_covar=tensor([0.0685, 0.0833, 0.0686, 0.0638, 0.0529, 0.0536, 0.0705, 0.0655], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:16:23,797 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234202.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:16:24,497 INFO [train.py:904] (7/8) Epoch 24, batch 750, loss[loss=0.1678, simple_loss=0.2656, pruned_loss=0.03503, over 17041.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2486, pruned_loss=0.03831, over 3248710.13 frames. ], batch size: 50, lr: 2.83e-03, grad_scale: 2.0 2023-05-01 18:16:44,357 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-01 18:16:45,126 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1726, 3.2390, 3.3483, 2.2458, 3.0877, 3.4061, 3.1330, 1.9839], device='cuda:7'), covar=tensor([0.0535, 0.0123, 0.0074, 0.0464, 0.0147, 0.0128, 0.0139, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0086, 0.0087, 0.0135, 0.0099, 0.0111, 0.0096, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 18:16:52,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.040e+02 2.434e+02 2.884e+02 6.768e+02, threshold=4.867e+02, percent-clipped=4.0 2023-05-01 18:17:04,924 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234231.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:17:33,980 INFO [train.py:904] (7/8) Epoch 24, batch 800, loss[loss=0.1377, simple_loss=0.2241, pruned_loss=0.0257, over 16811.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2489, pruned_loss=0.039, over 3257890.54 frames. ], batch size: 42, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:17:48,323 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234263.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:18:00,206 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-05-01 18:18:29,150 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234292.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:18:33,017 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234295.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:18:43,654 INFO [train.py:904] (7/8) Epoch 24, batch 850, loss[loss=0.1791, simple_loss=0.2498, pruned_loss=0.05422, over 16846.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2481, pruned_loss=0.03873, over 3275987.84 frames. ], batch size: 116, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:19:11,881 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.010e+02 2.479e+02 2.918e+02 4.237e+02, threshold=4.958e+02, percent-clipped=0.0 2023-05-01 18:19:40,598 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234343.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:19:49,570 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1587, 5.1400, 5.0091, 4.5227, 4.6451, 5.0410, 4.9978, 4.6891], device='cuda:7'), covar=tensor([0.0632, 0.0635, 0.0345, 0.0347, 0.1154, 0.0512, 0.0373, 0.0747], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0444, 0.0351, 0.0350, 0.0353, 0.0407, 0.0238, 0.0420], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:19:52,517 INFO [train.py:904] (7/8) Epoch 24, batch 900, loss[loss=0.1817, simple_loss=0.2524, pruned_loss=0.05546, over 16900.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2478, pruned_loss=0.03886, over 3279322.48 frames. ], batch size: 96, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:20:11,531 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234366.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:20:33,697 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234381.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:21:02,829 INFO [train.py:904] (7/8) Epoch 24, batch 950, loss[loss=0.1573, simple_loss=0.2377, pruned_loss=0.03842, over 15314.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2476, pruned_loss=0.03901, over 3284472.71 frames. ], batch size: 190, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:21:17,861 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:21:30,310 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.144e+02 2.493e+02 3.109e+02 1.071e+03, threshold=4.986e+02, percent-clipped=5.0 2023-05-01 18:21:37,444 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9573, 2.1183, 2.4993, 2.8329, 2.6889, 3.3586, 2.3442, 3.3598], device='cuda:7'), covar=tensor([0.0318, 0.0513, 0.0366, 0.0372, 0.0396, 0.0202, 0.0487, 0.0204], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0195, 0.0181, 0.0186, 0.0200, 0.0158, 0.0198, 0.0155], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:21:38,414 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5668, 4.5939, 4.9244, 4.9236, 4.9656, 4.6130, 4.6377, 4.4719], device='cuda:7'), covar=tensor([0.0461, 0.0908, 0.0497, 0.0512, 0.0585, 0.0607, 0.1045, 0.0737], device='cuda:7'), in_proj_covar=tensor([0.0427, 0.0473, 0.0462, 0.0424, 0.0503, 0.0483, 0.0565, 0.0386], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 18:22:10,617 INFO [train.py:904] (7/8) Epoch 24, batch 1000, loss[loss=0.1607, simple_loss=0.2534, pruned_loss=0.034, over 17042.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2469, pruned_loss=0.03911, over 3294872.34 frames. ], batch size: 55, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:22:50,683 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6572, 2.5960, 1.8690, 2.7298, 2.1704, 2.8000, 2.1553, 2.3477], device='cuda:7'), covar=tensor([0.0365, 0.0431, 0.1463, 0.0307, 0.0753, 0.0564, 0.1344, 0.0746], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0179, 0.0196, 0.0169, 0.0178, 0.0220, 0.0205, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:23:22,529 INFO [train.py:904] (7/8) Epoch 24, batch 1050, loss[loss=0.1462, simple_loss=0.2259, pruned_loss=0.03321, over 16190.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2468, pruned_loss=0.03893, over 3287507.59 frames. ], batch size: 165, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:23:28,504 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-01 18:23:50,908 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.145e+02 2.609e+02 3.101e+02 1.491e+03, threshold=5.219e+02, percent-clipped=5.0 2023-05-01 18:24:30,873 INFO [train.py:904] (7/8) Epoch 24, batch 1100, loss[loss=0.1531, simple_loss=0.2411, pruned_loss=0.0326, over 16450.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2462, pruned_loss=0.03832, over 3292189.60 frames. ], batch size: 75, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:24:37,571 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234558.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:25:18,101 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=234587.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:25:38,660 INFO [train.py:904] (7/8) Epoch 24, batch 1150, loss[loss=0.135, simple_loss=0.2229, pruned_loss=0.02356, over 16996.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2463, pruned_loss=0.03822, over 3304410.44 frames. ], batch size: 41, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:25:53,422 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3214, 3.6060, 4.0328, 2.2615, 3.3241, 2.3958, 3.7861, 3.7595], device='cuda:7'), covar=tensor([0.0289, 0.0879, 0.0477, 0.2148, 0.0787, 0.1110, 0.0588, 0.1085], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0168, 0.0155, 0.0146, 0.0130, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:26:06,115 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.089e+02 2.435e+02 2.920e+02 5.927e+02, threshold=4.869e+02, percent-clipped=1.0 2023-05-01 18:26:33,572 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 18:26:46,796 INFO [train.py:904] (7/8) Epoch 24, batch 1200, loss[loss=0.1817, simple_loss=0.2479, pruned_loss=0.05779, over 16901.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2458, pruned_loss=0.03788, over 3310929.69 frames. ], batch size: 109, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:27:09,117 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234669.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:27:25,016 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234681.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:27:53,735 INFO [train.py:904] (7/8) Epoch 24, batch 1250, loss[loss=0.1606, simple_loss=0.2556, pruned_loss=0.03284, over 16726.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2458, pruned_loss=0.0377, over 3312026.35 frames. ], batch size: 57, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:28:21,122 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.044e+02 2.334e+02 2.766e+02 4.867e+02, threshold=4.669e+02, percent-clipped=0.0 2023-05-01 18:28:25,788 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234726.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:28:30,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234729.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:28:31,750 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234730.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:28:35,935 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4647, 3.5378, 4.1225, 2.1363, 3.2984, 2.6010, 3.8570, 3.7480], device='cuda:7'), covar=tensor([0.0309, 0.1022, 0.0438, 0.2190, 0.0819, 0.0971, 0.0601, 0.1136], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0169, 0.0155, 0.0147, 0.0131, 0.0145, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:28:37,180 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-01 18:28:39,566 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.16 vs. limit=5.0 2023-05-01 18:29:02,400 INFO [train.py:904] (7/8) Epoch 24, batch 1300, loss[loss=0.1281, simple_loss=0.2121, pruned_loss=0.02202, over 16799.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.2455, pruned_loss=0.03811, over 3310007.17 frames. ], batch size: 39, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:29:27,396 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5041, 4.5944, 4.7002, 4.5366, 4.5854, 5.1429, 4.6492, 4.2621], device='cuda:7'), covar=tensor([0.1807, 0.2099, 0.2444, 0.2422, 0.2665, 0.1183, 0.1753, 0.2929], device='cuda:7'), in_proj_covar=tensor([0.0418, 0.0611, 0.0675, 0.0508, 0.0673, 0.0705, 0.0526, 0.0674], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 18:29:41,655 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234782.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:29:47,411 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234787.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:30:10,002 INFO [train.py:904] (7/8) Epoch 24, batch 1350, loss[loss=0.1937, simple_loss=0.2565, pruned_loss=0.06543, over 16758.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.246, pruned_loss=0.03844, over 3305625.97 frames. ], batch size: 124, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:30:38,936 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.157e+02 2.428e+02 2.986e+02 5.268e+02, threshold=4.855e+02, percent-clipped=4.0 2023-05-01 18:30:41,867 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1775, 3.9656, 4.3578, 2.4997, 4.5694, 4.6299, 3.4015, 3.5250], device='cuda:7'), covar=tensor([0.0653, 0.0262, 0.0223, 0.1106, 0.0080, 0.0166, 0.0418, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0100, 0.0141, 0.0083, 0.0129, 0.0129, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:30:45,922 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 18:31:07,929 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234843.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:31:19,644 INFO [train.py:904] (7/8) Epoch 24, batch 1400, loss[loss=0.1796, simple_loss=0.2534, pruned_loss=0.05288, over 16485.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.246, pruned_loss=0.03817, over 3317260.63 frames. ], batch size: 146, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:31:28,000 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234858.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:31:59,524 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.70 vs. limit=5.0 2023-05-01 18:32:00,170 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234882.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:32:07,412 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=234887.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:32:29,523 INFO [train.py:904] (7/8) Epoch 24, batch 1450, loss[loss=0.1614, simple_loss=0.2521, pruned_loss=0.03536, over 16721.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2448, pruned_loss=0.03811, over 3302003.05 frames. ], batch size: 62, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:32:34,694 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234906.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:32:42,064 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-01 18:32:51,103 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-05-01 18:32:58,967 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.989e+02 2.305e+02 2.603e+02 6.556e+02, threshold=4.610e+02, percent-clipped=2.0 2023-05-01 18:33:15,415 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=234935.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:33:26,791 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=234943.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:33:38,971 INFO [train.py:904] (7/8) Epoch 24, batch 1500, loss[loss=0.1542, simple_loss=0.2393, pruned_loss=0.03454, over 16837.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.245, pruned_loss=0.03836, over 3309218.84 frames. ], batch size: 42, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:33:44,917 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.95 vs. limit=5.0 2023-05-01 18:33:46,405 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234958.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:33:49,598 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=234960.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:34:06,150 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 18:34:13,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8748, 2.0080, 2.4186, 2.6558, 2.7427, 2.6889, 1.9916, 2.9385], device='cuda:7'), covar=tensor([0.0212, 0.0505, 0.0353, 0.0334, 0.0315, 0.0344, 0.0588, 0.0187], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0198, 0.0183, 0.0189, 0.0203, 0.0161, 0.0201, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:34:28,948 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3367, 2.3382, 2.3181, 4.1171, 2.2883, 2.6586, 2.3602, 2.4973], device='cuda:7'), covar=tensor([0.1371, 0.3699, 0.3293, 0.0577, 0.4201, 0.2681, 0.3700, 0.3686], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0460, 0.0378, 0.0333, 0.0443, 0.0526, 0.0431, 0.0538], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:34:49,724 INFO [train.py:904] (7/8) Epoch 24, batch 1550, loss[loss=0.1561, simple_loss=0.2502, pruned_loss=0.03101, over 17203.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.247, pruned_loss=0.03974, over 3309866.83 frames. ], batch size: 46, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:35:12,943 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235019.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:35:15,744 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235021.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:35:18,855 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.608e+02 2.203e+02 2.509e+02 3.045e+02 6.004e+02, threshold=5.019e+02, percent-clipped=5.0 2023-05-01 18:35:20,359 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235025.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 18:35:21,550 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9750, 5.3832, 5.5350, 5.2911, 5.3496, 5.9582, 5.4397, 5.1050], device='cuda:7'), covar=tensor([0.1169, 0.2090, 0.2465, 0.2064, 0.2906, 0.1023, 0.1596, 0.2558], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0616, 0.0679, 0.0510, 0.0679, 0.0710, 0.0529, 0.0682], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 18:35:58,106 INFO [train.py:904] (7/8) Epoch 24, batch 1600, loss[loss=0.1863, simple_loss=0.2711, pruned_loss=0.05071, over 16451.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2497, pruned_loss=0.0412, over 3310400.03 frames. ], batch size: 68, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:36:38,017 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235082.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:37:06,940 INFO [train.py:904] (7/8) Epoch 24, batch 1650, loss[loss=0.1439, simple_loss=0.2334, pruned_loss=0.02722, over 16849.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2514, pruned_loss=0.04144, over 3309950.99 frames. ], batch size: 42, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:37:35,283 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.234e+02 2.602e+02 3.440e+02 9.714e+02, threshold=5.204e+02, percent-clipped=4.0 2023-05-01 18:37:37,093 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-01 18:37:55,439 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235138.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:38:16,060 INFO [train.py:904] (7/8) Epoch 24, batch 1700, loss[loss=0.1851, simple_loss=0.2621, pruned_loss=0.05409, over 16787.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2519, pruned_loss=0.04099, over 3314997.46 frames. ], batch size: 83, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:39:24,682 INFO [train.py:904] (7/8) Epoch 24, batch 1750, loss[loss=0.1879, simple_loss=0.2815, pruned_loss=0.04711, over 16673.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2527, pruned_loss=0.04105, over 3317625.33 frames. ], batch size: 57, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:39:52,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.256e+02 2.911e+02 3.628e+02 6.601e+02, threshold=5.823e+02, percent-clipped=7.0 2023-05-01 18:40:12,950 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235238.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:40:19,703 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2023-05-01 18:40:33,181 INFO [train.py:904] (7/8) Epoch 24, batch 1800, loss[loss=0.1937, simple_loss=0.2707, pruned_loss=0.05836, over 16908.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2541, pruned_loss=0.04102, over 3315973.59 frames. ], batch size: 109, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:40:36,455 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235255.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:40:57,014 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 18:41:30,432 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-01 18:41:34,391 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4668, 2.3751, 2.3798, 4.3682, 2.3778, 2.7995, 2.4605, 2.6241], device='cuda:7'), covar=tensor([0.1309, 0.3852, 0.3117, 0.0497, 0.4171, 0.2628, 0.3605, 0.3728], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0462, 0.0379, 0.0334, 0.0444, 0.0528, 0.0433, 0.0539], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:41:40,807 INFO [train.py:904] (7/8) Epoch 24, batch 1850, loss[loss=0.1784, simple_loss=0.2633, pruned_loss=0.04681, over 16499.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2552, pruned_loss=0.04137, over 3318992.17 frames. ], batch size: 68, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:41:56,551 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235314.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:41:59,483 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235316.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:41:59,616 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235316.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:42:11,308 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.161e+02 2.587e+02 3.187e+02 4.962e+02, threshold=5.175e+02, percent-clipped=0.0 2023-05-01 18:42:12,322 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235325.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:42:49,556 INFO [train.py:904] (7/8) Epoch 24, batch 1900, loss[loss=0.1711, simple_loss=0.2449, pruned_loss=0.04865, over 16716.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2546, pruned_loss=0.04112, over 3323208.77 frames. ], batch size: 124, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:43:01,057 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235360.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 18:43:14,758 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3739, 4.2068, 4.4578, 4.5716, 4.6777, 4.2541, 4.5359, 4.6933], device='cuda:7'), covar=tensor([0.1625, 0.1257, 0.1243, 0.0707, 0.0567, 0.1157, 0.2299, 0.0685], device='cuda:7'), in_proj_covar=tensor([0.0677, 0.0840, 0.0963, 0.0849, 0.0644, 0.0667, 0.0700, 0.0814], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:43:17,656 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235373.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:43:31,588 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235382.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:44:00,957 INFO [train.py:904] (7/8) Epoch 24, batch 1950, loss[loss=0.1676, simple_loss=0.2575, pruned_loss=0.0389, over 16658.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2543, pruned_loss=0.04093, over 3327636.25 frames. ], batch size: 62, lr: 2.83e-03, grad_scale: 4.0 2023-05-01 18:44:26,877 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235421.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:44:31,115 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.100e+02 2.476e+02 2.993e+02 5.203e+02, threshold=4.952e+02, percent-clipped=1.0 2023-05-01 18:44:37,952 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:44:43,447 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-05-01 18:44:49,804 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235438.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:45:08,810 INFO [train.py:904] (7/8) Epoch 24, batch 2000, loss[loss=0.1709, simple_loss=0.266, pruned_loss=0.03794, over 12131.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2539, pruned_loss=0.04028, over 3317723.07 frames. ], batch size: 246, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:45:53,864 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235486.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:46:09,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1466, 4.0066, 4.2105, 4.3116, 4.3716, 3.9966, 4.1979, 4.3955], device='cuda:7'), covar=tensor([0.1477, 0.1122, 0.1185, 0.0667, 0.0613, 0.1408, 0.2066, 0.0665], device='cuda:7'), in_proj_covar=tensor([0.0682, 0.0845, 0.0971, 0.0853, 0.0648, 0.0670, 0.0704, 0.0818], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:46:12,014 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-01 18:46:16,402 INFO [train.py:904] (7/8) Epoch 24, batch 2050, loss[loss=0.1806, simple_loss=0.2694, pruned_loss=0.0459, over 17000.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2542, pruned_loss=0.04008, over 3320613.79 frames. ], batch size: 55, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:46:46,285 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.111e+02 2.349e+02 3.134e+02 8.501e+02, threshold=4.699e+02, percent-clipped=2.0 2023-05-01 18:47:03,591 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235538.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:47:24,556 INFO [train.py:904] (7/8) Epoch 24, batch 2100, loss[loss=0.1473, simple_loss=0.2385, pruned_loss=0.02802, over 16784.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2547, pruned_loss=0.04019, over 3321317.96 frames. ], batch size: 39, lr: 2.83e-03, grad_scale: 8.0 2023-05-01 18:48:10,093 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235586.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:48:31,076 INFO [train.py:904] (7/8) Epoch 24, batch 2150, loss[loss=0.1829, simple_loss=0.2631, pruned_loss=0.05134, over 16971.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2551, pruned_loss=0.04036, over 3328279.60 frames. ], batch size: 116, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:48:43,109 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235611.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:48:47,644 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235614.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:48:49,858 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:49:02,119 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.182e+02 2.677e+02 3.005e+02 4.991e+02, threshold=5.354e+02, percent-clipped=2.0 2023-05-01 18:49:25,101 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-05-01 18:49:41,319 INFO [train.py:904] (7/8) Epoch 24, batch 2200, loss[loss=0.1396, simple_loss=0.2293, pruned_loss=0.02497, over 17226.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2556, pruned_loss=0.04047, over 3323517.92 frames. ], batch size: 45, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:49:54,421 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235662.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:49:57,765 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235664.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:50:14,585 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3001, 5.2191, 5.1408, 4.6081, 4.7893, 5.1741, 5.0681, 4.7706], device='cuda:7'), covar=tensor([0.0563, 0.0561, 0.0297, 0.0394, 0.1074, 0.0526, 0.0366, 0.0784], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0461, 0.0363, 0.0361, 0.0365, 0.0420, 0.0246, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 18:50:18,255 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235679.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:50:50,809 INFO [train.py:904] (7/8) Epoch 24, batch 2250, loss[loss=0.1942, simple_loss=0.2702, pruned_loss=0.05906, over 16754.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2554, pruned_loss=0.04051, over 3325174.62 frames. ], batch size: 124, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:51:01,533 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-01 18:51:08,553 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=235716.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 18:51:20,162 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 2.086e+02 2.517e+02 3.047e+02 4.896e+02, threshold=5.035e+02, percent-clipped=0.0 2023-05-01 18:51:41,239 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235740.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:51:57,980 INFO [train.py:904] (7/8) Epoch 24, batch 2300, loss[loss=0.1623, simple_loss=0.2428, pruned_loss=0.0409, over 16287.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2559, pruned_loss=0.04073, over 3327384.53 frames. ], batch size: 165, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:52:11,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6463, 3.3470, 3.7445, 2.0104, 3.8180, 3.8210, 3.1148, 2.8168], device='cuda:7'), covar=tensor([0.0725, 0.0253, 0.0181, 0.1126, 0.0103, 0.0194, 0.0407, 0.0452], device='cuda:7'), in_proj_covar=tensor([0.0146, 0.0108, 0.0099, 0.0139, 0.0082, 0.0127, 0.0128, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 18:52:50,327 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=235791.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:53:08,148 INFO [train.py:904] (7/8) Epoch 24, batch 2350, loss[loss=0.1721, simple_loss=0.2676, pruned_loss=0.03834, over 16760.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2562, pruned_loss=0.04144, over 3322410.73 frames. ], batch size: 57, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:53:31,834 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7265, 2.5664, 2.5117, 3.8888, 3.1066, 3.9226, 1.5208, 2.9122], device='cuda:7'), covar=tensor([0.1453, 0.0716, 0.1196, 0.0221, 0.0178, 0.0401, 0.1694, 0.0839], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0177, 0.0196, 0.0195, 0.0205, 0.0217, 0.0205, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:53:37,794 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.284e+02 2.753e+02 3.332e+02 6.201e+02, threshold=5.507e+02, percent-clipped=2.0 2023-05-01 18:54:14,676 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=235852.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:54:15,319 INFO [train.py:904] (7/8) Epoch 24, batch 2400, loss[loss=0.2018, simple_loss=0.2762, pruned_loss=0.06368, over 16747.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2577, pruned_loss=0.0418, over 3317979.42 frames. ], batch size: 124, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:54:30,902 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9175, 3.0492, 3.2009, 2.1044, 2.7971, 2.2058, 3.4537, 3.4366], device='cuda:7'), covar=tensor([0.0255, 0.0930, 0.0655, 0.1928, 0.0926, 0.1058, 0.0538, 0.0824], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0167, 0.0168, 0.0155, 0.0147, 0.0131, 0.0145, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 18:55:22,152 INFO [train.py:904] (7/8) Epoch 24, batch 2450, loss[loss=0.1817, simple_loss=0.2638, pruned_loss=0.04978, over 16897.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2578, pruned_loss=0.04148, over 3319535.14 frames. ], batch size: 109, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:55:33,276 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=235911.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:55:51,880 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.168e+02 2.510e+02 2.858e+02 5.891e+02, threshold=5.019e+02, percent-clipped=1.0 2023-05-01 18:56:28,805 INFO [train.py:904] (7/8) Epoch 24, batch 2500, loss[loss=0.1501, simple_loss=0.2466, pruned_loss=0.02679, over 17114.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2575, pruned_loss=0.04076, over 3318422.04 frames. ], batch size: 47, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:56:29,748 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6154, 4.9445, 4.7551, 4.7535, 4.5057, 4.4274, 4.4002, 5.0463], device='cuda:7'), covar=tensor([0.1241, 0.0909, 0.1021, 0.0884, 0.0831, 0.1309, 0.1219, 0.0819], device='cuda:7'), in_proj_covar=tensor([0.0710, 0.0866, 0.0710, 0.0664, 0.0549, 0.0553, 0.0729, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 18:56:38,291 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=235959.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:57:41,348 INFO [train.py:904] (7/8) Epoch 24, batch 2550, loss[loss=0.1517, simple_loss=0.2349, pruned_loss=0.03425, over 16695.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2575, pruned_loss=0.04088, over 3321072.23 frames. ], batch size: 89, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:57:59,241 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236016.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:58:01,864 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=236018.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:58:10,536 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.492e+02 2.266e+02 2.526e+02 3.100e+02 4.908e+02, threshold=5.053e+02, percent-clipped=0.0 2023-05-01 18:58:24,325 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=236035.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:58:49,127 INFO [train.py:904] (7/8) Epoch 24, batch 2600, loss[loss=0.147, simple_loss=0.2402, pruned_loss=0.02685, over 16808.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2571, pruned_loss=0.04051, over 3326180.35 frames. ], batch size: 39, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 18:58:53,953 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-05-01 18:59:03,970 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236064.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 18:59:25,564 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=236079.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 18:59:58,194 INFO [train.py:904] (7/8) Epoch 24, batch 2650, loss[loss=0.1547, simple_loss=0.2444, pruned_loss=0.03248, over 17220.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2579, pruned_loss=0.04039, over 3318588.84 frames. ], batch size: 44, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:00:27,606 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.167e+02 2.518e+02 3.194e+02 5.192e+02, threshold=5.036e+02, percent-clipped=2.0 2023-05-01 19:00:55,076 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.41 vs. limit=5.0 2023-05-01 19:00:58,565 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=236147.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:01:05,917 INFO [train.py:904] (7/8) Epoch 24, batch 2700, loss[loss=0.1766, simple_loss=0.2767, pruned_loss=0.03821, over 16642.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2584, pruned_loss=0.03989, over 3323924.31 frames. ], batch size: 57, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:01:14,286 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7514, 2.7129, 2.3170, 2.6703, 3.0180, 2.8573, 3.3038, 3.2611], device='cuda:7'), covar=tensor([0.0173, 0.0467, 0.0549, 0.0454, 0.0305, 0.0402, 0.0290, 0.0295], device='cuda:7'), in_proj_covar=tensor([0.0229, 0.0246, 0.0235, 0.0237, 0.0247, 0.0246, 0.0247, 0.0244], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:02:15,471 INFO [train.py:904] (7/8) Epoch 24, batch 2750, loss[loss=0.162, simple_loss=0.2575, pruned_loss=0.03329, over 16026.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2578, pruned_loss=0.03912, over 3330356.19 frames. ], batch size: 35, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:02:16,869 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0431, 5.3388, 5.1251, 5.1010, 4.8372, 4.7912, 4.8109, 5.4519], device='cuda:7'), covar=tensor([0.1274, 0.0960, 0.1087, 0.0890, 0.0902, 0.1085, 0.1257, 0.0961], device='cuda:7'), in_proj_covar=tensor([0.0719, 0.0876, 0.0719, 0.0672, 0.0555, 0.0559, 0.0735, 0.0684], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:02:44,673 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.298e+02 1.972e+02 2.253e+02 2.676e+02 4.118e+02, threshold=4.507e+02, percent-clipped=0.0 2023-05-01 19:03:22,969 INFO [train.py:904] (7/8) Epoch 24, batch 2800, loss[loss=0.1762, simple_loss=0.2572, pruned_loss=0.04755, over 16460.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2579, pruned_loss=0.03933, over 3328899.94 frames. ], batch size: 146, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:04:32,923 INFO [train.py:904] (7/8) Epoch 24, batch 2850, loss[loss=0.2025, simple_loss=0.2855, pruned_loss=0.05974, over 15400.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2566, pruned_loss=0.03916, over 3327224.63 frames. ], batch size: 190, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:05:03,662 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-05-01 19:05:04,035 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.044e+02 2.504e+02 3.029e+02 5.589e+02, threshold=5.008e+02, percent-clipped=4.0 2023-05-01 19:05:17,518 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236335.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:05:43,134 INFO [train.py:904] (7/8) Epoch 24, batch 2900, loss[loss=0.1616, simple_loss=0.2339, pruned_loss=0.04462, over 16884.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2566, pruned_loss=0.0398, over 3325134.48 frames. ], batch size: 116, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:06:13,276 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=236374.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:06:25,556 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236383.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:06:45,446 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1613, 5.1393, 5.0453, 4.6216, 4.6897, 5.1106, 5.0311, 4.7338], device='cuda:7'), covar=tensor([0.0617, 0.0505, 0.0309, 0.0354, 0.1084, 0.0425, 0.0355, 0.0714], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0466, 0.0366, 0.0366, 0.0371, 0.0424, 0.0250, 0.0442], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:06:53,579 INFO [train.py:904] (7/8) Epoch 24, batch 2950, loss[loss=0.2045, simple_loss=0.2785, pruned_loss=0.06526, over 15603.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2571, pruned_loss=0.04079, over 3321647.15 frames. ], batch size: 191, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:07:24,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.404e+02 2.771e+02 3.339e+02 1.037e+03, threshold=5.543e+02, percent-clipped=6.0 2023-05-01 19:07:30,632 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9420, 2.6817, 2.5686, 1.9400, 2.5542, 2.7381, 2.5402, 1.9462], device='cuda:7'), covar=tensor([0.0421, 0.0119, 0.0103, 0.0385, 0.0171, 0.0130, 0.0148, 0.0403], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0088, 0.0136, 0.0101, 0.0113, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 19:07:54,771 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236447.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:08:02,714 INFO [train.py:904] (7/8) Epoch 24, batch 3000, loss[loss=0.1687, simple_loss=0.2481, pruned_loss=0.0447, over 15476.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2577, pruned_loss=0.04139, over 3321975.90 frames. ], batch size: 190, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:08:02,714 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 19:08:12,055 INFO [train.py:938] (7/8) Epoch 24, validation: loss=0.1342, simple_loss=0.2393, pruned_loss=0.0145, over 944034.00 frames. 2023-05-01 19:08:12,056 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 19:08:46,492 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1272, 5.2859, 5.0514, 4.6450, 4.2911, 5.2685, 5.2510, 4.7290], device='cuda:7'), covar=tensor([0.0907, 0.0598, 0.0453, 0.0457, 0.1966, 0.0469, 0.0288, 0.0818], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0466, 0.0367, 0.0366, 0.0371, 0.0425, 0.0250, 0.0443], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:09:12,148 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236495.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:09:22,992 INFO [train.py:904] (7/8) Epoch 24, batch 3050, loss[loss=0.1826, simple_loss=0.2603, pruned_loss=0.05248, over 16904.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2574, pruned_loss=0.04149, over 3322273.09 frames. ], batch size: 90, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:09:53,425 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.094e+02 2.423e+02 2.785e+02 4.383e+02, threshold=4.846e+02, percent-clipped=1.0 2023-05-01 19:10:02,745 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0192, 4.7723, 5.0580, 5.2313, 5.4459, 4.7804, 5.3841, 5.4375], device='cuda:7'), covar=tensor([0.1929, 0.1435, 0.1856, 0.0835, 0.0530, 0.1031, 0.0607, 0.0548], device='cuda:7'), in_proj_covar=tensor([0.0686, 0.0850, 0.0978, 0.0854, 0.0652, 0.0679, 0.0707, 0.0823], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:10:32,486 INFO [train.py:904] (7/8) Epoch 24, batch 3100, loss[loss=0.169, simple_loss=0.264, pruned_loss=0.037, over 17085.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2556, pruned_loss=0.04106, over 3326511.31 frames. ], batch size: 53, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:11:43,888 INFO [train.py:904] (7/8) Epoch 24, batch 3150, loss[loss=0.1618, simple_loss=0.2591, pruned_loss=0.03226, over 17282.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2548, pruned_loss=0.04054, over 3315451.17 frames. ], batch size: 52, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:12:13,747 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.103e+02 2.556e+02 2.857e+02 5.533e+02, threshold=5.112e+02, percent-clipped=2.0 2023-05-01 19:12:49,306 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2652, 5.7028, 5.8599, 5.6038, 5.6370, 6.2192, 5.7373, 5.4468], device='cuda:7'), covar=tensor([0.0979, 0.2152, 0.2688, 0.2167, 0.2773, 0.0999, 0.1519, 0.2547], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0624, 0.0687, 0.0517, 0.0689, 0.0718, 0.0538, 0.0685], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:12:52,331 INFO [train.py:904] (7/8) Epoch 24, batch 3200, loss[loss=0.1736, simple_loss=0.2549, pruned_loss=0.04615, over 16595.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2534, pruned_loss=0.04049, over 3314445.07 frames. ], batch size: 89, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:13:21,998 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=236674.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:13:34,890 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4885, 4.3171, 4.5175, 4.6682, 4.8054, 4.3364, 4.6448, 4.7900], device='cuda:7'), covar=tensor([0.1674, 0.1172, 0.1411, 0.0672, 0.0560, 0.1203, 0.1593, 0.0643], device='cuda:7'), in_proj_covar=tensor([0.0685, 0.0850, 0.0976, 0.0851, 0.0652, 0.0678, 0.0707, 0.0822], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:13:50,723 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 19:14:01,579 INFO [train.py:904] (7/8) Epoch 24, batch 3250, loss[loss=0.2045, simple_loss=0.2766, pruned_loss=0.06624, over 12251.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2537, pruned_loss=0.04066, over 3311756.17 frames. ], batch size: 246, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:14:27,660 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=236722.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:14:31,667 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 2.232e+02 2.615e+02 3.014e+02 5.902e+02, threshold=5.231e+02, percent-clipped=1.0 2023-05-01 19:15:11,546 INFO [train.py:904] (7/8) Epoch 24, batch 3300, loss[loss=0.1601, simple_loss=0.2479, pruned_loss=0.03618, over 16847.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2549, pruned_loss=0.04056, over 3316860.05 frames. ], batch size: 83, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:16:21,168 INFO [train.py:904] (7/8) Epoch 24, batch 3350, loss[loss=0.1424, simple_loss=0.2298, pruned_loss=0.02748, over 16430.00 frames. ], tot_loss[loss=0.168, simple_loss=0.255, pruned_loss=0.04056, over 3313662.63 frames. ], batch size: 68, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:16:51,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 2.070e+02 2.526e+02 2.944e+02 6.359e+02, threshold=5.052e+02, percent-clipped=3.0 2023-05-01 19:16:54,180 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-01 19:17:04,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1657, 5.1584, 4.9255, 4.3104, 5.0427, 1.8218, 4.7712, 4.7807], device='cuda:7'), covar=tensor([0.0111, 0.0086, 0.0236, 0.0494, 0.0118, 0.3161, 0.0182, 0.0260], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0168, 0.0210, 0.0186, 0.0186, 0.0216, 0.0199, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:17:15,799 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-01 19:17:33,212 INFO [train.py:904] (7/8) Epoch 24, batch 3400, loss[loss=0.1681, simple_loss=0.2646, pruned_loss=0.03582, over 17124.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2545, pruned_loss=0.04044, over 3314634.61 frames. ], batch size: 47, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:17:40,350 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4204, 4.4607, 4.8117, 4.7859, 4.8281, 4.5313, 4.5121, 4.4201], device='cuda:7'), covar=tensor([0.0377, 0.0639, 0.0393, 0.0450, 0.0484, 0.0431, 0.0859, 0.0623], device='cuda:7'), in_proj_covar=tensor([0.0429, 0.0478, 0.0465, 0.0428, 0.0510, 0.0485, 0.0570, 0.0390], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 19:17:44,511 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7733, 5.1076, 4.8701, 4.8945, 4.6519, 4.5913, 4.5757, 5.1899], device='cuda:7'), covar=tensor([0.1290, 0.0867, 0.1091, 0.0840, 0.0832, 0.1221, 0.1237, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0715, 0.0871, 0.0717, 0.0670, 0.0555, 0.0557, 0.0731, 0.0680], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:18:31,104 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=236892.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:18:45,607 INFO [train.py:904] (7/8) Epoch 24, batch 3450, loss[loss=0.1612, simple_loss=0.2425, pruned_loss=0.03995, over 16780.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2536, pruned_loss=0.04032, over 3307344.88 frames. ], batch size: 39, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:19:09,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0225, 2.8992, 2.7097, 4.4882, 3.7093, 4.2123, 1.7627, 3.1416], device='cuda:7'), covar=tensor([0.1264, 0.0731, 0.1124, 0.0234, 0.0211, 0.0409, 0.1587, 0.0770], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0177, 0.0197, 0.0197, 0.0207, 0.0219, 0.0205, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:19:15,850 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.094e+02 2.355e+02 2.776e+02 4.395e+02, threshold=4.710e+02, percent-clipped=0.0 2023-05-01 19:19:39,792 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7752, 3.8765, 2.9601, 2.2679, 2.4344, 2.4110, 4.0177, 3.2946], device='cuda:7'), covar=tensor([0.2823, 0.0546, 0.1682, 0.3019, 0.2893, 0.2221, 0.0468, 0.1654], device='cuda:7'), in_proj_covar=tensor([0.0330, 0.0272, 0.0309, 0.0320, 0.0302, 0.0267, 0.0298, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:19:55,278 INFO [train.py:904] (7/8) Epoch 24, batch 3500, loss[loss=0.2111, simple_loss=0.2879, pruned_loss=0.06716, over 15431.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2523, pruned_loss=0.03939, over 3320065.66 frames. ], batch size: 190, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:19:56,891 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=236953.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 19:20:06,990 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0780, 4.0616, 3.9561, 3.6862, 3.7553, 4.0682, 3.6736, 3.8878], device='cuda:7'), covar=tensor([0.0640, 0.0779, 0.0313, 0.0297, 0.0610, 0.0475, 0.1069, 0.0577], device='cuda:7'), in_proj_covar=tensor([0.0316, 0.0469, 0.0368, 0.0368, 0.0373, 0.0426, 0.0251, 0.0445], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:21:06,718 INFO [train.py:904] (7/8) Epoch 24, batch 3550, loss[loss=0.1795, simple_loss=0.2569, pruned_loss=0.05102, over 16251.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.252, pruned_loss=0.03932, over 3312465.96 frames. ], batch size: 165, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:21:19,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9075, 5.2490, 4.9792, 5.0105, 4.7483, 4.7103, 4.6341, 5.3166], device='cuda:7'), covar=tensor([0.1158, 0.0773, 0.1023, 0.0880, 0.0809, 0.1125, 0.1262, 0.0773], device='cuda:7'), in_proj_covar=tensor([0.0718, 0.0873, 0.0719, 0.0673, 0.0557, 0.0558, 0.0734, 0.0682], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:21:24,777 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7779, 5.1651, 5.5009, 5.4240, 5.4848, 5.1214, 4.7702, 4.8602], device='cuda:7'), covar=tensor([0.0673, 0.0756, 0.0546, 0.0693, 0.0769, 0.0622, 0.1530, 0.0625], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0484, 0.0470, 0.0432, 0.0515, 0.0491, 0.0577, 0.0394], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 19:21:35,733 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 1.933e+02 2.243e+02 2.593e+02 4.523e+02, threshold=4.485e+02, percent-clipped=0.0 2023-05-01 19:22:15,053 INFO [train.py:904] (7/8) Epoch 24, batch 3600, loss[loss=0.1801, simple_loss=0.2471, pruned_loss=0.05649, over 16743.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.252, pruned_loss=0.04008, over 3309731.82 frames. ], batch size: 124, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:23:26,132 INFO [train.py:904] (7/8) Epoch 24, batch 3650, loss[loss=0.1472, simple_loss=0.2299, pruned_loss=0.03225, over 16735.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2507, pruned_loss=0.04007, over 3297527.92 frames. ], batch size: 39, lr: 2.82e-03, grad_scale: 8.0 2023-05-01 19:23:58,728 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.153e+02 2.502e+02 3.114e+02 9.969e+02, threshold=5.004e+02, percent-clipped=5.0 2023-05-01 19:24:39,868 INFO [train.py:904] (7/8) Epoch 24, batch 3700, loss[loss=0.1806, simple_loss=0.2649, pruned_loss=0.0482, over 16655.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2494, pruned_loss=0.04121, over 3276123.75 frames. ], batch size: 57, lr: 2.82e-03, grad_scale: 4.0 2023-05-01 19:25:53,330 INFO [train.py:904] (7/8) Epoch 24, batch 3750, loss[loss=0.1719, simple_loss=0.2526, pruned_loss=0.04553, over 16869.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2504, pruned_loss=0.04264, over 3266175.50 frames. ], batch size: 109, lr: 2.82e-03, grad_scale: 4.0 2023-05-01 19:26:15,442 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1039, 3.9404, 4.0969, 4.3208, 4.3660, 4.0704, 4.2379, 4.4298], device='cuda:7'), covar=tensor([0.1775, 0.1449, 0.1753, 0.0899, 0.0864, 0.1437, 0.2290, 0.1004], device='cuda:7'), in_proj_covar=tensor([0.0682, 0.0845, 0.0974, 0.0849, 0.0650, 0.0673, 0.0704, 0.0820], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:26:25,690 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.223e+02 2.609e+02 3.048e+02 5.921e+02, threshold=5.219e+02, percent-clipped=3.0 2023-05-01 19:26:57,502 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=237248.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 19:26:59,379 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4887, 4.3866, 4.4038, 4.1455, 4.1949, 4.4641, 4.1524, 4.2489], device='cuda:7'), covar=tensor([0.0559, 0.0793, 0.0271, 0.0266, 0.0598, 0.0479, 0.0602, 0.0533], device='cuda:7'), in_proj_covar=tensor([0.0313, 0.0466, 0.0365, 0.0364, 0.0368, 0.0421, 0.0247, 0.0440], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:27:03,608 INFO [train.py:904] (7/8) Epoch 24, batch 3800, loss[loss=0.1711, simple_loss=0.2581, pruned_loss=0.04207, over 16471.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2515, pruned_loss=0.04393, over 3265789.51 frames. ], batch size: 75, lr: 2.82e-03, grad_scale: 4.0 2023-05-01 19:27:11,160 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3902, 4.4323, 4.7638, 4.7273, 4.7768, 4.4589, 4.5004, 4.2866], device='cuda:7'), covar=tensor([0.0349, 0.0601, 0.0363, 0.0379, 0.0497, 0.0474, 0.0791, 0.0673], device='cuda:7'), in_proj_covar=tensor([0.0432, 0.0483, 0.0467, 0.0431, 0.0512, 0.0487, 0.0573, 0.0392], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 19:27:25,245 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2706, 2.3911, 2.5288, 4.1823, 2.3283, 2.7584, 2.4323, 2.5531], device='cuda:7'), covar=tensor([0.1457, 0.3432, 0.2782, 0.0555, 0.3714, 0.2416, 0.3702, 0.2824], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0465, 0.0381, 0.0337, 0.0444, 0.0534, 0.0435, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:28:20,072 INFO [train.py:904] (7/8) Epoch 24, batch 3850, loss[loss=0.1944, simple_loss=0.261, pruned_loss=0.06394, over 16309.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2512, pruned_loss=0.04428, over 3274881.73 frames. ], batch size: 165, lr: 2.81e-03, grad_scale: 4.0 2023-05-01 19:28:52,887 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.317e+02 2.601e+02 3.052e+02 4.766e+02, threshold=5.202e+02, percent-clipped=0.0 2023-05-01 19:28:53,440 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1468, 2.4618, 2.7064, 1.8871, 2.8033, 2.7897, 2.4932, 2.3393], device='cuda:7'), covar=tensor([0.0701, 0.0307, 0.0195, 0.0992, 0.0126, 0.0251, 0.0432, 0.0448], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0110, 0.0100, 0.0140, 0.0083, 0.0129, 0.0129, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:29:00,975 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4143, 1.8479, 2.1407, 2.3415, 2.5086, 2.4467, 1.8908, 2.5679], device='cuda:7'), covar=tensor([0.0207, 0.0480, 0.0313, 0.0325, 0.0317, 0.0368, 0.0532, 0.0177], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0196, 0.0185, 0.0189, 0.0204, 0.0163, 0.0201, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:29:30,869 INFO [train.py:904] (7/8) Epoch 24, batch 3900, loss[loss=0.1841, simple_loss=0.2517, pruned_loss=0.05825, over 16740.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2515, pruned_loss=0.04519, over 3280041.83 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 4.0 2023-05-01 19:29:58,136 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7299, 4.6720, 4.6137, 3.4508, 4.6962, 1.7691, 4.3175, 4.1239], device='cuda:7'), covar=tensor([0.0211, 0.0170, 0.0276, 0.0800, 0.0173, 0.3620, 0.0250, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0168, 0.0210, 0.0186, 0.0186, 0.0215, 0.0198, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:30:43,391 INFO [train.py:904] (7/8) Epoch 24, batch 3950, loss[loss=0.1652, simple_loss=0.2399, pruned_loss=0.04523, over 16727.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.251, pruned_loss=0.04582, over 3286581.52 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 4.0 2023-05-01 19:31:17,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.223e+02 2.645e+02 3.283e+02 6.262e+02, threshold=5.289e+02, percent-clipped=1.0 2023-05-01 19:31:57,084 INFO [train.py:904] (7/8) Epoch 24, batch 4000, loss[loss=0.1778, simple_loss=0.2621, pruned_loss=0.04677, over 16547.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2514, pruned_loss=0.04614, over 3278698.11 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:32:54,243 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237491.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:33:06,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8447, 4.0498, 2.5539, 4.8158, 3.1603, 4.7921, 2.7873, 3.1857], device='cuda:7'), covar=tensor([0.0314, 0.0330, 0.1660, 0.0111, 0.0753, 0.0245, 0.1426, 0.0769], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0181, 0.0197, 0.0174, 0.0179, 0.0223, 0.0205, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:33:10,938 INFO [train.py:904] (7/8) Epoch 24, batch 4050, loss[loss=0.173, simple_loss=0.2557, pruned_loss=0.04513, over 16691.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.252, pruned_loss=0.04508, over 3276740.23 frames. ], batch size: 57, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:33:32,977 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2090, 2.1737, 1.8024, 1.9005, 2.4213, 2.1307, 2.0527, 2.5649], device='cuda:7'), covar=tensor([0.0185, 0.0466, 0.0558, 0.0509, 0.0283, 0.0379, 0.0196, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0228, 0.0244, 0.0234, 0.0235, 0.0245, 0.0245, 0.0246, 0.0243], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:33:43,735 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 1.828e+02 2.122e+02 2.508e+02 6.067e+02, threshold=4.243e+02, percent-clipped=1.0 2023-05-01 19:34:01,595 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-01 19:34:17,622 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=237548.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 19:34:23,851 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237552.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:34:24,476 INFO [train.py:904] (7/8) Epoch 24, batch 4100, loss[loss=0.1866, simple_loss=0.2688, pruned_loss=0.05223, over 12369.00 frames. ], tot_loss[loss=0.1717, simple_loss=0.2541, pruned_loss=0.04463, over 3273951.06 frames. ], batch size: 246, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:34:38,982 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237563.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:35:09,683 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8540, 3.9595, 4.1360, 4.0975, 4.1287, 3.9372, 3.9324, 3.9442], device='cuda:7'), covar=tensor([0.0340, 0.0489, 0.0382, 0.0418, 0.0486, 0.0389, 0.0813, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0426, 0.0476, 0.0460, 0.0423, 0.0506, 0.0481, 0.0565, 0.0386], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 19:35:26,324 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7744, 1.4305, 1.6999, 1.7509, 1.7775, 1.9489, 1.6368, 1.8141], device='cuda:7'), covar=tensor([0.0270, 0.0375, 0.0209, 0.0295, 0.0263, 0.0199, 0.0424, 0.0131], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0195, 0.0184, 0.0189, 0.0204, 0.0163, 0.0201, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:35:30,754 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=237596.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:35:37,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9484, 5.2052, 4.9944, 5.0566, 4.7522, 4.7027, 4.7245, 5.3490], device='cuda:7'), covar=tensor([0.1354, 0.0930, 0.1057, 0.0810, 0.0822, 0.0914, 0.1022, 0.0875], device='cuda:7'), in_proj_covar=tensor([0.0707, 0.0858, 0.0709, 0.0664, 0.0547, 0.0551, 0.0724, 0.0671], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:35:40,166 INFO [train.py:904] (7/8) Epoch 24, batch 4150, loss[loss=0.2, simple_loss=0.2969, pruned_loss=0.05154, over 16219.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2612, pruned_loss=0.04682, over 3249382.11 frames. ], batch size: 165, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:36:14,659 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237624.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:36:16,372 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.139e+02 2.639e+02 3.206e+02 5.465e+02, threshold=5.279e+02, percent-clipped=4.0 2023-05-01 19:36:22,848 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-01 19:36:56,395 INFO [train.py:904] (7/8) Epoch 24, batch 4200, loss[loss=0.1934, simple_loss=0.2924, pruned_loss=0.04721, over 16987.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.268, pruned_loss=0.0485, over 3225149.25 frames. ], batch size: 41, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:37:14,868 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 19:37:46,512 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3596, 3.2833, 3.2380, 3.4636, 3.4659, 3.2562, 3.4487, 3.5326], device='cuda:7'), covar=tensor([0.1298, 0.1299, 0.1643, 0.0888, 0.0962, 0.2893, 0.1409, 0.1180], device='cuda:7'), in_proj_covar=tensor([0.0670, 0.0829, 0.0953, 0.0834, 0.0637, 0.0661, 0.0692, 0.0805], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:38:10,766 INFO [train.py:904] (7/8) Epoch 24, batch 4250, loss[loss=0.1629, simple_loss=0.2568, pruned_loss=0.03452, over 17262.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2715, pruned_loss=0.04806, over 3217548.71 frames. ], batch size: 52, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:38:45,326 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.159e+02 2.541e+02 2.876e+02 4.427e+02, threshold=5.081e+02, percent-clipped=0.0 2023-05-01 19:39:26,168 INFO [train.py:904] (7/8) Epoch 24, batch 4300, loss[loss=0.1841, simple_loss=0.2804, pruned_loss=0.04393, over 16401.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2725, pruned_loss=0.04736, over 3222563.33 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:40:00,886 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6536, 2.5227, 2.3211, 3.8517, 3.0209, 3.8104, 1.4873, 2.7690], device='cuda:7'), covar=tensor([0.1355, 0.0829, 0.1340, 0.0185, 0.0301, 0.0400, 0.1721, 0.0869], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0177, 0.0195, 0.0194, 0.0205, 0.0215, 0.0203, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:40:23,440 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237790.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:40:41,742 INFO [train.py:904] (7/8) Epoch 24, batch 4350, loss[loss=0.1816, simple_loss=0.2793, pruned_loss=0.04191, over 16847.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2761, pruned_loss=0.04859, over 3216742.59 frames. ], batch size: 96, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:41:03,168 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3218, 3.0596, 3.4749, 1.6905, 3.5849, 3.6145, 2.7705, 2.6879], device='cuda:7'), covar=tensor([0.0836, 0.0327, 0.0226, 0.1249, 0.0092, 0.0149, 0.0517, 0.0488], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0111, 0.0101, 0.0141, 0.0083, 0.0129, 0.0130, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:41:14,404 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.621e+02 2.331e+02 2.533e+02 3.016e+02 1.010e+03, threshold=5.065e+02, percent-clipped=1.0 2023-05-01 19:41:45,928 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=237847.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:41:49,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9460, 4.1972, 4.0142, 4.0667, 3.7812, 3.8056, 3.8188, 4.2041], device='cuda:7'), covar=tensor([0.1029, 0.0846, 0.0983, 0.0752, 0.0695, 0.1769, 0.0899, 0.0921], device='cuda:7'), in_proj_covar=tensor([0.0696, 0.0842, 0.0697, 0.0654, 0.0537, 0.0542, 0.0711, 0.0661], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:41:52,167 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237851.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:41:54,774 INFO [train.py:904] (7/8) Epoch 24, batch 4400, loss[loss=0.2265, simple_loss=0.3154, pruned_loss=0.06876, over 16483.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2783, pruned_loss=0.05006, over 3208302.56 frames. ], batch size: 35, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:43:05,641 INFO [train.py:904] (7/8) Epoch 24, batch 4450, loss[loss=0.1869, simple_loss=0.275, pruned_loss=0.04935, over 12088.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2813, pruned_loss=0.05121, over 3220601.69 frames. ], batch size: 248, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:43:18,973 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=237912.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:43:28,636 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=237919.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 19:43:38,899 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 1.917e+02 2.137e+02 2.680e+02 5.470e+02, threshold=4.273e+02, percent-clipped=1.0 2023-05-01 19:44:16,783 INFO [train.py:904] (7/8) Epoch 24, batch 4500, loss[loss=0.2019, simple_loss=0.2925, pruned_loss=0.0557, over 16864.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2817, pruned_loss=0.05168, over 3221042.77 frames. ], batch size: 102, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:44:47,403 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=237973.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:45:00,017 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5299, 5.7836, 5.5250, 5.6172, 5.2782, 5.1634, 5.1845, 5.9497], device='cuda:7'), covar=tensor([0.1176, 0.0774, 0.1077, 0.0750, 0.0789, 0.0643, 0.1102, 0.0803], device='cuda:7'), in_proj_covar=tensor([0.0694, 0.0840, 0.0696, 0.0652, 0.0537, 0.0541, 0.0709, 0.0661], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:45:32,225 INFO [train.py:904] (7/8) Epoch 24, batch 4550, loss[loss=0.1998, simple_loss=0.2796, pruned_loss=0.06003, over 12496.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2826, pruned_loss=0.05273, over 3228721.91 frames. ], batch size: 248, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:45:49,394 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-05-01 19:46:04,574 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 1.751e+02 2.009e+02 2.366e+02 4.725e+02, threshold=4.018e+02, percent-clipped=1.0 2023-05-01 19:46:20,468 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-05-01 19:46:44,403 INFO [train.py:904] (7/8) Epoch 24, batch 4600, loss[loss=0.204, simple_loss=0.2938, pruned_loss=0.05715, over 16855.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.2833, pruned_loss=0.05301, over 3223019.50 frames. ], batch size: 116, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:46:49,540 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7189, 4.0267, 2.6278, 2.4442, 2.5602, 2.3069, 4.1939, 3.2270], device='cuda:7'), covar=tensor([0.3169, 0.0659, 0.2197, 0.2629, 0.2764, 0.2481, 0.0612, 0.1432], device='cuda:7'), in_proj_covar=tensor([0.0330, 0.0272, 0.0308, 0.0319, 0.0302, 0.0266, 0.0298, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 19:47:00,821 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238064.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:47:15,262 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238074.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:47:19,042 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 19:47:35,718 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4216, 4.6936, 4.5237, 4.5159, 4.2283, 4.1780, 4.2346, 4.7494], device='cuda:7'), covar=tensor([0.1172, 0.0817, 0.0957, 0.0777, 0.0802, 0.1500, 0.1030, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0691, 0.0837, 0.0694, 0.0650, 0.0535, 0.0539, 0.0707, 0.0659], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:47:43,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3758, 4.4663, 4.2563, 3.9397, 3.9982, 4.3682, 3.9526, 4.1140], device='cuda:7'), covar=tensor([0.0471, 0.0359, 0.0230, 0.0243, 0.0609, 0.0285, 0.0788, 0.0523], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0444, 0.0349, 0.0349, 0.0352, 0.0402, 0.0239, 0.0420], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:47:50,335 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9622, 3.9993, 2.3612, 4.9187, 3.2130, 4.7455, 2.6411, 3.1391], device='cuda:7'), covar=tensor([0.0274, 0.0352, 0.1833, 0.0139, 0.0784, 0.0349, 0.1534, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0178, 0.0194, 0.0169, 0.0177, 0.0220, 0.0203, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:47:53,682 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-01 19:47:56,663 INFO [train.py:904] (7/8) Epoch 24, batch 4650, loss[loss=0.185, simple_loss=0.2694, pruned_loss=0.05024, over 16562.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2822, pruned_loss=0.05297, over 3207138.26 frames. ], batch size: 62, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:48:29,693 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238125.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:48:30,378 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.898e+02 2.163e+02 2.416e+02 3.900e+02, threshold=4.325e+02, percent-clipped=0.0 2023-05-01 19:48:30,843 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238126.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:48:43,743 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238135.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:48:59,366 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238146.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:49:00,602 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238147.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:49:05,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1703, 2.2949, 2.3490, 3.8684, 2.2688, 2.6834, 2.4127, 2.4791], device='cuda:7'), covar=tensor([0.1446, 0.3642, 0.2932, 0.0588, 0.3963, 0.2545, 0.3587, 0.3293], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0462, 0.0377, 0.0333, 0.0443, 0.0531, 0.0432, 0.0541], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:49:09,310 INFO [train.py:904] (7/8) Epoch 24, batch 4700, loss[loss=0.1719, simple_loss=0.2585, pruned_loss=0.04268, over 16542.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2794, pruned_loss=0.05167, over 3201535.45 frames. ], batch size: 62, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:49:10,468 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0197, 2.1787, 2.2001, 3.6329, 2.1789, 2.5362, 2.3278, 2.3567], device='cuda:7'), covar=tensor([0.1551, 0.3479, 0.3095, 0.0639, 0.4072, 0.2500, 0.3550, 0.3443], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0462, 0.0377, 0.0333, 0.0443, 0.0531, 0.0432, 0.0541], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:49:10,906 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 19:49:38,891 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4020, 1.7439, 2.0866, 2.3952, 2.5109, 2.7198, 1.8483, 2.6447], device='cuda:7'), covar=tensor([0.0277, 0.0548, 0.0356, 0.0385, 0.0358, 0.0238, 0.0576, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0182, 0.0187, 0.0202, 0.0161, 0.0199, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:49:58,937 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238187.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:50:09,179 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238195.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:50:20,670 INFO [train.py:904] (7/8) Epoch 24, batch 4750, loss[loss=0.1891, simple_loss=0.2715, pruned_loss=0.05334, over 11887.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2749, pruned_loss=0.04915, over 3209793.48 frames. ], batch size: 247, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:50:43,598 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238219.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 19:50:53,745 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 1.823e+02 2.174e+02 2.542e+02 5.265e+02, threshold=4.348e+02, percent-clipped=2.0 2023-05-01 19:51:31,474 INFO [train.py:904] (7/8) Epoch 24, batch 4800, loss[loss=0.187, simple_loss=0.2817, pruned_loss=0.04617, over 15376.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2714, pruned_loss=0.04727, over 3210439.48 frames. ], batch size: 190, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:51:52,869 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238267.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:51:54,742 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238268.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:52:46,492 INFO [train.py:904] (7/8) Epoch 24, batch 4850, loss[loss=0.1863, simple_loss=0.2808, pruned_loss=0.04592, over 16705.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2728, pruned_loss=0.04728, over 3173166.03 frames. ], batch size: 134, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:53:22,734 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.326e+02 1.927e+02 2.251e+02 2.681e+02 4.052e+02, threshold=4.502e+02, percent-clipped=0.0 2023-05-01 19:53:37,169 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6611, 4.7119, 5.0180, 4.9747, 4.9742, 4.6925, 4.6592, 4.5526], device='cuda:7'), covar=tensor([0.0273, 0.0523, 0.0297, 0.0324, 0.0370, 0.0315, 0.0808, 0.0409], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0459, 0.0447, 0.0411, 0.0492, 0.0466, 0.0549, 0.0374], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 19:54:04,052 INFO [train.py:904] (7/8) Epoch 24, batch 4900, loss[loss=0.1692, simple_loss=0.2579, pruned_loss=0.0403, over 16719.00 frames. ], tot_loss[loss=0.1816, simple_loss=0.2716, pruned_loss=0.04575, over 3177159.51 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:54:43,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4337, 4.4079, 4.2914, 3.2693, 4.3872, 1.5336, 4.1214, 3.9119], device='cuda:7'), covar=tensor([0.0125, 0.0120, 0.0204, 0.0582, 0.0122, 0.3355, 0.0172, 0.0377], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0164, 0.0205, 0.0181, 0.0180, 0.0210, 0.0193, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:55:17,238 INFO [train.py:904] (7/8) Epoch 24, batch 4950, loss[loss=0.1993, simple_loss=0.2785, pruned_loss=0.06005, over 16968.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2708, pruned_loss=0.04481, over 3178622.77 frames. ], batch size: 41, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:55:28,423 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 19:55:41,598 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238420.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:55:41,744 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238420.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:55:49,725 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.002e+02 2.407e+02 2.827e+02 4.284e+02, threshold=4.815e+02, percent-clipped=0.0 2023-05-01 19:55:57,439 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238430.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:56:20,355 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238446.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:56:29,589 INFO [train.py:904] (7/8) Epoch 24, batch 5000, loss[loss=0.2023, simple_loss=0.2845, pruned_loss=0.06003, over 17049.00 frames. ], tot_loss[loss=0.1815, simple_loss=0.2726, pruned_loss=0.04521, over 3183061.24 frames. ], batch size: 55, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:57:11,155 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238481.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 19:57:12,065 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238482.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:57:31,130 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238494.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:57:42,737 INFO [train.py:904] (7/8) Epoch 24, batch 5050, loss[loss=0.1753, simple_loss=0.2673, pruned_loss=0.04164, over 16750.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2734, pruned_loss=0.04498, over 3199322.62 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:58:15,457 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.006e+02 2.316e+02 2.692e+02 7.292e+02, threshold=4.632e+02, percent-clipped=1.0 2023-05-01 19:58:29,407 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7253, 2.6219, 2.7150, 4.6048, 3.4153, 4.0411, 1.6581, 2.9389], device='cuda:7'), covar=tensor([0.1357, 0.0821, 0.1106, 0.0121, 0.0193, 0.0326, 0.1606, 0.0833], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0178, 0.0196, 0.0195, 0.0205, 0.0216, 0.0205, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 19:58:53,441 INFO [train.py:904] (7/8) Epoch 24, batch 5100, loss[loss=0.1635, simple_loss=0.2686, pruned_loss=0.02924, over 16888.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.271, pruned_loss=0.04385, over 3220323.07 frames. ], batch size: 90, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 19:59:16,761 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238568.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 19:59:50,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2700, 1.5997, 1.9556, 2.1823, 2.3239, 2.5054, 1.7913, 2.4292], device='cuda:7'), covar=tensor([0.0236, 0.0555, 0.0338, 0.0384, 0.0357, 0.0227, 0.0540, 0.0149], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0183, 0.0187, 0.0202, 0.0161, 0.0199, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 19:59:54,152 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6041, 1.7522, 2.1640, 2.5231, 2.5644, 2.9087, 1.8403, 2.8731], device='cuda:7'), covar=tensor([0.0234, 0.0581, 0.0383, 0.0368, 0.0368, 0.0213, 0.0619, 0.0128], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0183, 0.0186, 0.0202, 0.0161, 0.0199, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:00:07,249 INFO [train.py:904] (7/8) Epoch 24, batch 5150, loss[loss=0.1783, simple_loss=0.2769, pruned_loss=0.03986, over 16415.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2719, pruned_loss=0.04357, over 3220745.37 frames. ], batch size: 146, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:00:26,118 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:00:36,833 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4056, 3.3408, 3.4306, 3.5175, 3.5572, 3.2951, 3.5312, 3.6141], device='cuda:7'), covar=tensor([0.1216, 0.0947, 0.1066, 0.0660, 0.0605, 0.2649, 0.0962, 0.0726], device='cuda:7'), in_proj_covar=tensor([0.0642, 0.0793, 0.0913, 0.0803, 0.0613, 0.0633, 0.0662, 0.0773], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:00:38,785 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.286e+02 1.999e+02 2.272e+02 2.615e+02 4.924e+02, threshold=4.544e+02, percent-clipped=1.0 2023-05-01 20:00:39,396 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9703, 2.7586, 2.8709, 2.0977, 2.6707, 2.1802, 2.7615, 2.9578], device='cuda:7'), covar=tensor([0.0281, 0.0781, 0.0513, 0.1723, 0.0829, 0.0921, 0.0537, 0.0680], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0168, 0.0154, 0.0145, 0.0130, 0.0143, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:01:17,683 INFO [train.py:904] (7/8) Epoch 24, batch 5200, loss[loss=0.1784, simple_loss=0.2678, pruned_loss=0.04447, over 16250.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2708, pruned_loss=0.04315, over 3204636.36 frames. ], batch size: 165, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:01:28,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7338, 4.6711, 4.6130, 3.5582, 4.6260, 1.6238, 4.3476, 4.3118], device='cuda:7'), covar=tensor([0.0120, 0.0122, 0.0216, 0.0633, 0.0145, 0.3141, 0.0180, 0.0300], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0162, 0.0203, 0.0180, 0.0179, 0.0209, 0.0191, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:01:32,628 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 20:01:55,286 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8240, 2.0076, 2.4005, 2.7828, 2.8019, 3.2011, 2.0556, 3.1957], device='cuda:7'), covar=tensor([0.0226, 0.0557, 0.0393, 0.0347, 0.0344, 0.0206, 0.0601, 0.0141], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0195, 0.0183, 0.0187, 0.0203, 0.0161, 0.0199, 0.0158], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:01:57,540 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238681.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:02:28,279 INFO [train.py:904] (7/8) Epoch 24, batch 5250, loss[loss=0.1855, simple_loss=0.2757, pruned_loss=0.0476, over 12594.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2683, pruned_loss=0.04288, over 3203017.85 frames. ], batch size: 246, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:02:30,112 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3505, 3.4620, 2.0449, 3.7948, 2.5702, 3.7807, 2.2909, 2.7423], device='cuda:7'), covar=tensor([0.0292, 0.0339, 0.1662, 0.0177, 0.0876, 0.0493, 0.1543, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0177, 0.0192, 0.0165, 0.0176, 0.0216, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:02:52,105 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238720.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:03:01,286 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 1.983e+02 2.265e+02 2.655e+02 9.176e+02, threshold=4.530e+02, percent-clipped=2.0 2023-05-01 20:03:07,692 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238730.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:03:25,253 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238742.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:03:26,593 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5408, 3.5344, 3.9297, 2.1976, 3.3064, 2.6444, 3.9221, 3.8731], device='cuda:7'), covar=tensor([0.0193, 0.0838, 0.0502, 0.2048, 0.0797, 0.0893, 0.0515, 0.0960], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0168, 0.0154, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:03:39,628 INFO [train.py:904] (7/8) Epoch 24, batch 5300, loss[loss=0.1626, simple_loss=0.2532, pruned_loss=0.03597, over 16224.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2646, pruned_loss=0.04174, over 3209568.92 frames. ], batch size: 165, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:03:47,569 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3852, 3.5042, 3.6499, 3.6236, 3.6233, 3.4914, 3.4956, 3.5335], device='cuda:7'), covar=tensor([0.0384, 0.0705, 0.0449, 0.0422, 0.0545, 0.0457, 0.0736, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0463, 0.0451, 0.0412, 0.0497, 0.0469, 0.0552, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 20:04:01,108 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238768.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:04:12,308 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=238776.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 20:04:15,139 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238778.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:04:20,293 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=238782.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:04:50,573 INFO [train.py:904] (7/8) Epoch 24, batch 5350, loss[loss=0.1867, simple_loss=0.2839, pruned_loss=0.04477, over 16786.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2632, pruned_loss=0.04134, over 3212784.49 frames. ], batch size: 124, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:05:21,928 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 1.926e+02 2.251e+02 2.610e+02 1.024e+03, threshold=4.501e+02, percent-clipped=4.0 2023-05-01 20:05:27,688 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=238830.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:05:59,875 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 20:06:01,811 INFO [train.py:904] (7/8) Epoch 24, batch 5400, loss[loss=0.1846, simple_loss=0.2769, pruned_loss=0.04618, over 16678.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2656, pruned_loss=0.04186, over 3208062.94 frames. ], batch size: 134, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:07:04,627 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6376, 3.6943, 2.7520, 2.2809, 2.4410, 2.3994, 4.0071, 3.3323], device='cuda:7'), covar=tensor([0.2800, 0.0618, 0.1853, 0.2682, 0.2625, 0.2069, 0.0412, 0.1252], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0271, 0.0307, 0.0318, 0.0300, 0.0265, 0.0298, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 20:07:11,347 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0541, 3.9524, 4.0963, 4.2292, 4.3461, 3.9665, 4.2868, 4.3677], device='cuda:7'), covar=tensor([0.1652, 0.1184, 0.1523, 0.0766, 0.0629, 0.1493, 0.1015, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0649, 0.0799, 0.0921, 0.0810, 0.0617, 0.0639, 0.0667, 0.0779], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:07:18,022 INFO [train.py:904] (7/8) Epoch 24, batch 5450, loss[loss=0.1848, simple_loss=0.2812, pruned_loss=0.04424, over 16841.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2687, pruned_loss=0.04334, over 3213877.46 frames. ], batch size: 83, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:07:54,289 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.137e+02 2.656e+02 3.537e+02 7.468e+02, threshold=5.312e+02, percent-clipped=13.0 2023-05-01 20:08:11,062 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=238936.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:08:18,339 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5894, 4.6662, 4.4523, 4.1776, 4.1782, 4.5674, 4.3500, 4.2916], device='cuda:7'), covar=tensor([0.0614, 0.0533, 0.0298, 0.0308, 0.0842, 0.0501, 0.0498, 0.0571], device='cuda:7'), in_proj_covar=tensor([0.0298, 0.0448, 0.0351, 0.0351, 0.0354, 0.0408, 0.0238, 0.0423], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:08:28,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6411, 2.5563, 1.9182, 2.7002, 2.1248, 2.7869, 2.1828, 2.3932], device='cuda:7'), covar=tensor([0.0304, 0.0359, 0.1141, 0.0286, 0.0593, 0.0496, 0.1044, 0.0537], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0177, 0.0193, 0.0166, 0.0176, 0.0216, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:08:35,280 INFO [train.py:904] (7/8) Epoch 24, batch 5500, loss[loss=0.2002, simple_loss=0.2919, pruned_loss=0.05425, over 16875.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2755, pruned_loss=0.04744, over 3180628.54 frames. ], batch size: 116, lr: 2.81e-03, grad_scale: 8.0 2023-05-01 20:09:44,166 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=238997.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 20:09:53,117 INFO [train.py:904] (7/8) Epoch 24, batch 5550, loss[loss=0.2413, simple_loss=0.3172, pruned_loss=0.08276, over 11463.00 frames. ], tot_loss[loss=0.1935, simple_loss=0.2825, pruned_loss=0.05222, over 3156201.56 frames. ], batch size: 247, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:10:06,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8847, 2.1937, 2.4830, 3.1062, 2.2243, 2.3671, 2.3585, 2.3177], device='cuda:7'), covar=tensor([0.1362, 0.3121, 0.2177, 0.0722, 0.3719, 0.2135, 0.2821, 0.2936], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0456, 0.0373, 0.0329, 0.0436, 0.0523, 0.0427, 0.0532], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:10:27,985 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7857, 3.2443, 3.2644, 2.0500, 2.8647, 2.1841, 3.2984, 3.5256], device='cuda:7'), covar=tensor([0.0383, 0.0827, 0.0616, 0.2110, 0.0938, 0.1020, 0.0781, 0.0916], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0167, 0.0154, 0.0146, 0.0130, 0.0143, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:10:30,509 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.205e+02 3.032e+02 3.628e+02 4.510e+02 7.943e+02, threshold=7.255e+02, percent-clipped=11.0 2023-05-01 20:10:48,593 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=239037.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:11:12,512 INFO [train.py:904] (7/8) Epoch 24, batch 5600, loss[loss=0.1908, simple_loss=0.2723, pruned_loss=0.05464, over 16673.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.287, pruned_loss=0.05599, over 3131374.42 frames. ], batch size: 62, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:11:52,284 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=239076.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 20:12:26,622 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 20:12:36,920 INFO [train.py:904] (7/8) Epoch 24, batch 5650, loss[loss=0.2986, simple_loss=0.3531, pruned_loss=0.1221, over 11367.00 frames. ], tot_loss[loss=0.2075, simple_loss=0.2931, pruned_loss=0.06096, over 3082849.82 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:12:54,725 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-05-01 20:13:11,865 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=239124.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:13:14,770 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.394e+02 3.339e+02 4.323e+02 5.614e+02 1.229e+03, threshold=8.646e+02, percent-clipped=9.0 2023-05-01 20:13:47,386 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3418, 4.5419, 4.6818, 4.4642, 4.5249, 5.0363, 4.5622, 4.2898], device='cuda:7'), covar=tensor([0.1503, 0.1882, 0.2205, 0.1978, 0.2346, 0.1035, 0.1670, 0.2491], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0602, 0.0661, 0.0496, 0.0660, 0.0692, 0.0519, 0.0659], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 20:13:55,709 INFO [train.py:904] (7/8) Epoch 24, batch 5700, loss[loss=0.2124, simple_loss=0.2977, pruned_loss=0.06354, over 16926.00 frames. ], tot_loss[loss=0.211, simple_loss=0.2954, pruned_loss=0.06329, over 3058601.95 frames. ], batch size: 109, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:15:14,262 INFO [train.py:904] (7/8) Epoch 24, batch 5750, loss[loss=0.2059, simple_loss=0.2897, pruned_loss=0.06105, over 17012.00 frames. ], tot_loss[loss=0.2142, simple_loss=0.2983, pruned_loss=0.06506, over 3040191.43 frames. ], batch size: 55, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:15:53,492 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.857e+02 3.000e+02 3.580e+02 4.341e+02 9.766e+02, threshold=7.159e+02, percent-clipped=1.0 2023-05-01 20:16:37,343 INFO [train.py:904] (7/8) Epoch 24, batch 5800, loss[loss=0.2013, simple_loss=0.2862, pruned_loss=0.05818, over 16213.00 frames. ], tot_loss[loss=0.2122, simple_loss=0.2977, pruned_loss=0.06337, over 3058914.19 frames. ], batch size: 35, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:17:40,244 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=239292.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:17:56,794 INFO [train.py:904] (7/8) Epoch 24, batch 5850, loss[loss=0.1849, simple_loss=0.2771, pruned_loss=0.04632, over 16473.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2944, pruned_loss=0.06118, over 3068680.69 frames. ], batch size: 146, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:18:33,901 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.891e+02 3.438e+02 4.490e+02 7.448e+02, threshold=6.875e+02, percent-clipped=1.0 2023-05-01 20:18:53,095 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=239337.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:19:19,097 INFO [train.py:904] (7/8) Epoch 24, batch 5900, loss[loss=0.2425, simple_loss=0.3084, pruned_loss=0.08833, over 11397.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2943, pruned_loss=0.06115, over 3074166.19 frames. ], batch size: 248, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:19:20,055 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1745, 3.1910, 1.8602, 3.4310, 2.4110, 3.4852, 2.0906, 2.6046], device='cuda:7'), covar=tensor([0.0320, 0.0417, 0.1891, 0.0317, 0.0904, 0.0707, 0.1720, 0.0888], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0196, 0.0169, 0.0179, 0.0219, 0.0205, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:19:29,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6965, 4.3866, 4.3661, 2.9839, 3.9936, 4.3924, 3.9096, 2.3943], device='cuda:7'), covar=tensor([0.0538, 0.0049, 0.0051, 0.0386, 0.0090, 0.0119, 0.0091, 0.0488], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0086, 0.0134, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 20:19:33,218 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7002, 4.5553, 4.7122, 4.8940, 5.1018, 4.6130, 5.0857, 5.0822], device='cuda:7'), covar=tensor([0.2050, 0.1307, 0.1981, 0.0954, 0.0701, 0.1089, 0.0940, 0.0685], device='cuda:7'), in_proj_covar=tensor([0.0641, 0.0790, 0.0910, 0.0800, 0.0608, 0.0630, 0.0660, 0.0771], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:20:14,708 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=239385.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:20:42,782 INFO [train.py:904] (7/8) Epoch 24, batch 5950, loss[loss=0.2053, simple_loss=0.2876, pruned_loss=0.06155, over 11699.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2946, pruned_loss=0.0598, over 3069897.98 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:21:03,783 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3943, 3.4308, 2.0332, 3.8076, 2.6535, 3.8382, 2.2853, 2.7877], device='cuda:7'), covar=tensor([0.0319, 0.0433, 0.1814, 0.0255, 0.0853, 0.0585, 0.1542, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0179, 0.0195, 0.0168, 0.0178, 0.0218, 0.0204, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:21:12,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9816, 3.1233, 3.1974, 2.0584, 2.9556, 3.1855, 2.9749, 1.9618], device='cuda:7'), covar=tensor([0.0569, 0.0083, 0.0077, 0.0463, 0.0130, 0.0135, 0.0124, 0.0479], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0086, 0.0134, 0.0099, 0.0111, 0.0097, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 20:21:21,255 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.785e+02 2.477e+02 3.014e+02 3.724e+02 7.833e+02, threshold=6.029e+02, percent-clipped=2.0 2023-05-01 20:22:03,343 INFO [train.py:904] (7/8) Epoch 24, batch 6000, loss[loss=0.1928, simple_loss=0.2785, pruned_loss=0.05348, over 16292.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2926, pruned_loss=0.0586, over 3098950.75 frames. ], batch size: 165, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:22:03,343 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 20:22:14,266 INFO [train.py:938] (7/8) Epoch 24, validation: loss=0.1493, simple_loss=0.2618, pruned_loss=0.01837, over 944034.00 frames. 2023-05-01 20:22:14,267 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 20:23:32,270 INFO [train.py:904] (7/8) Epoch 24, batch 6050, loss[loss=0.1959, simple_loss=0.2907, pruned_loss=0.05057, over 16721.00 frames. ], tot_loss[loss=0.2032, simple_loss=0.2909, pruned_loss=0.05778, over 3102687.18 frames. ], batch size: 76, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:24:09,879 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.713e+02 3.363e+02 4.065e+02 7.660e+02, threshold=6.725e+02, percent-clipped=3.0 2023-05-01 20:24:51,439 INFO [train.py:904] (7/8) Epoch 24, batch 6100, loss[loss=0.1932, simple_loss=0.2814, pruned_loss=0.05256, over 16900.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.29, pruned_loss=0.05717, over 3082347.44 frames. ], batch size: 109, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:25:56,292 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=239592.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:26:14,100 INFO [train.py:904] (7/8) Epoch 24, batch 6150, loss[loss=0.2182, simple_loss=0.2928, pruned_loss=0.07179, over 11711.00 frames. ], tot_loss[loss=0.2004, simple_loss=0.2878, pruned_loss=0.05649, over 3077349.92 frames. ], batch size: 248, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:26:25,198 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3230, 2.3979, 2.8346, 3.2786, 3.1133, 3.8006, 2.3673, 3.7151], device='cuda:7'), covar=tensor([0.0199, 0.0466, 0.0324, 0.0289, 0.0283, 0.0131, 0.0505, 0.0128], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0194, 0.0182, 0.0185, 0.0201, 0.0160, 0.0199, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:26:41,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9248, 5.2498, 4.7888, 5.0399, 4.7494, 4.5825, 4.7499, 5.2874], device='cuda:7'), covar=tensor([0.2018, 0.1270, 0.2044, 0.1429, 0.1427, 0.1556, 0.2276, 0.1642], device='cuda:7'), in_proj_covar=tensor([0.0689, 0.0829, 0.0688, 0.0641, 0.0530, 0.0531, 0.0697, 0.0650], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:26:45,943 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=239622.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:26:53,239 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.890e+02 2.573e+02 3.097e+02 3.625e+02 7.083e+02, threshold=6.193e+02, percent-clipped=1.0 2023-05-01 20:27:14,544 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=239640.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 20:27:34,960 INFO [train.py:904] (7/8) Epoch 24, batch 6200, loss[loss=0.1982, simple_loss=0.2806, pruned_loss=0.05792, over 16847.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2855, pruned_loss=0.05566, over 3098306.38 frames. ], batch size: 109, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:28:13,165 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.11 vs. limit=2.0 2023-05-01 20:28:22,590 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=239683.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:28:52,029 INFO [train.py:904] (7/8) Epoch 24, batch 6250, loss[loss=0.2131, simple_loss=0.3026, pruned_loss=0.0618, over 16941.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2853, pruned_loss=0.05529, over 3112545.83 frames. ], batch size: 109, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:29:29,986 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 2.614e+02 3.057e+02 3.659e+02 7.500e+02, threshold=6.114e+02, percent-clipped=2.0 2023-05-01 20:29:44,029 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1072, 2.4654, 2.5613, 1.8984, 2.7097, 2.7775, 2.4382, 2.3676], device='cuda:7'), covar=tensor([0.0699, 0.0262, 0.0222, 0.0994, 0.0125, 0.0284, 0.0472, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0109, 0.0100, 0.0139, 0.0083, 0.0129, 0.0129, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:30:05,928 INFO [train.py:904] (7/8) Epoch 24, batch 6300, loss[loss=0.2264, simple_loss=0.2972, pruned_loss=0.07782, over 11907.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2855, pruned_loss=0.05518, over 3095107.24 frames. ], batch size: 249, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:30:20,218 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7559, 1.9771, 2.3819, 2.7419, 2.6616, 3.0885, 2.0813, 3.0481], device='cuda:7'), covar=tensor([0.0243, 0.0543, 0.0386, 0.0359, 0.0350, 0.0203, 0.0590, 0.0174], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0182, 0.0185, 0.0202, 0.0160, 0.0200, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:30:47,829 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7782, 3.2943, 3.3165, 2.0963, 2.9206, 2.3946, 3.3268, 3.5515], device='cuda:7'), covar=tensor([0.0305, 0.0756, 0.0663, 0.2104, 0.0945, 0.1001, 0.0676, 0.0849], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0169, 0.0155, 0.0147, 0.0132, 0.0144, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:31:13,276 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 20:31:24,190 INFO [train.py:904] (7/8) Epoch 24, batch 6350, loss[loss=0.2105, simple_loss=0.2939, pruned_loss=0.0635, over 16372.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2856, pruned_loss=0.05542, over 3110783.88 frames. ], batch size: 146, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:31:25,586 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-01 20:32:03,932 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.770e+02 2.790e+02 3.396e+02 4.270e+02 1.026e+03, threshold=6.791e+02, percent-clipped=1.0 2023-05-01 20:32:15,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5117, 3.4225, 3.4547, 2.6963, 3.3327, 2.1262, 3.1219, 2.6563], device='cuda:7'), covar=tensor([0.0173, 0.0153, 0.0194, 0.0226, 0.0109, 0.2282, 0.0160, 0.0239], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0165, 0.0206, 0.0182, 0.0181, 0.0212, 0.0193, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:32:17,787 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1934, 2.2394, 2.2479, 3.8736, 2.1844, 2.6321, 2.3114, 2.4155], device='cuda:7'), covar=tensor([0.1308, 0.3497, 0.2964, 0.0520, 0.4005, 0.2313, 0.3530, 0.3137], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0456, 0.0373, 0.0329, 0.0437, 0.0522, 0.0427, 0.0533], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:32:42,138 INFO [train.py:904] (7/8) Epoch 24, batch 6400, loss[loss=0.1971, simple_loss=0.293, pruned_loss=0.05057, over 16682.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2867, pruned_loss=0.05697, over 3091887.30 frames. ], batch size: 68, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:33:21,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4483, 5.3627, 5.2542, 3.9538, 5.2707, 1.6152, 4.9018, 4.7053], device='cuda:7'), covar=tensor([0.0167, 0.0166, 0.0224, 0.0746, 0.0168, 0.3419, 0.0287, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0165, 0.0206, 0.0182, 0.0181, 0.0212, 0.0193, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:33:58,190 INFO [train.py:904] (7/8) Epoch 24, batch 6450, loss[loss=0.1832, simple_loss=0.2691, pruned_loss=0.04864, over 15182.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2865, pruned_loss=0.05624, over 3092838.56 frames. ], batch size: 190, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:34:27,883 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 20:34:37,423 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.672e+02 3.087e+02 3.802e+02 7.341e+02, threshold=6.174e+02, percent-clipped=2.0 2023-05-01 20:34:38,265 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0235, 3.2484, 3.2102, 2.1137, 3.0377, 3.2317, 3.0121, 1.9550], device='cuda:7'), covar=tensor([0.0572, 0.0076, 0.0073, 0.0457, 0.0123, 0.0123, 0.0128, 0.0509], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0087, 0.0134, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 20:35:08,753 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-01 20:35:16,109 INFO [train.py:904] (7/8) Epoch 24, batch 6500, loss[loss=0.1875, simple_loss=0.2775, pruned_loss=0.04873, over 16702.00 frames. ], tot_loss[loss=0.1973, simple_loss=0.2842, pruned_loss=0.05522, over 3098051.52 frames. ], batch size: 134, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:35:37,018 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-01 20:35:55,185 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=239978.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:36:09,106 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7279, 3.0660, 3.2724, 1.9381, 2.8628, 2.1287, 3.3314, 3.3458], device='cuda:7'), covar=tensor([0.0285, 0.0852, 0.0608, 0.2203, 0.0850, 0.1080, 0.0621, 0.0865], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0166, 0.0169, 0.0155, 0.0147, 0.0132, 0.0144, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:36:39,181 INFO [train.py:904] (7/8) Epoch 24, batch 6550, loss[loss=0.2231, simple_loss=0.2999, pruned_loss=0.07316, over 11762.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2878, pruned_loss=0.05704, over 3080921.07 frames. ], batch size: 247, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:37:16,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.521e+02 3.197e+02 3.720e+02 9.360e+02, threshold=6.395e+02, percent-clipped=2.0 2023-05-01 20:37:54,314 INFO [train.py:904] (7/8) Epoch 24, batch 6600, loss[loss=0.2182, simple_loss=0.2887, pruned_loss=0.07388, over 11750.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2901, pruned_loss=0.05739, over 3076771.93 frames. ], batch size: 247, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:38:04,204 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.10 vs. limit=5.0 2023-05-01 20:39:11,707 INFO [train.py:904] (7/8) Epoch 24, batch 6650, loss[loss=0.2641, simple_loss=0.3204, pruned_loss=0.1039, over 11436.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2907, pruned_loss=0.05862, over 3080537.83 frames. ], batch size: 247, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:39:28,206 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3001, 5.3730, 5.7573, 5.7290, 5.7625, 5.4304, 5.3544, 5.0058], device='cuda:7'), covar=tensor([0.0313, 0.0405, 0.0300, 0.0291, 0.0454, 0.0313, 0.0917, 0.0488], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0466, 0.0454, 0.0415, 0.0499, 0.0472, 0.0556, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 20:39:50,369 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.847e+02 2.925e+02 3.532e+02 4.365e+02 9.170e+02, threshold=7.065e+02, percent-clipped=1.0 2023-05-01 20:40:28,853 INFO [train.py:904] (7/8) Epoch 24, batch 6700, loss[loss=0.2216, simple_loss=0.2967, pruned_loss=0.07331, over 11562.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2893, pruned_loss=0.05876, over 3063048.98 frames. ], batch size: 248, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:40:50,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1256, 2.0223, 1.7571, 1.7718, 2.2927, 1.9696, 1.9651, 2.3715], device='cuda:7'), covar=tensor([0.0212, 0.0441, 0.0571, 0.0502, 0.0261, 0.0373, 0.0206, 0.0266], device='cuda:7'), in_proj_covar=tensor([0.0215, 0.0235, 0.0226, 0.0227, 0.0236, 0.0234, 0.0234, 0.0232], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:40:58,206 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6446, 4.6158, 4.4738, 3.7161, 4.5770, 1.6002, 4.2772, 4.1129], device='cuda:7'), covar=tensor([0.0092, 0.0084, 0.0193, 0.0375, 0.0088, 0.2989, 0.0129, 0.0255], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0165, 0.0206, 0.0182, 0.0180, 0.0211, 0.0193, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:41:15,865 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0659, 4.0291, 3.9463, 3.1704, 4.0039, 1.7376, 3.7772, 3.4599], device='cuda:7'), covar=tensor([0.0137, 0.0114, 0.0209, 0.0301, 0.0111, 0.2849, 0.0145, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0165, 0.0206, 0.0182, 0.0181, 0.0211, 0.0193, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:41:45,714 INFO [train.py:904] (7/8) Epoch 24, batch 6750, loss[loss=0.1949, simple_loss=0.2803, pruned_loss=0.0548, over 15164.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2884, pruned_loss=0.0586, over 3078564.75 frames. ], batch size: 191, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:42:23,537 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.933e+02 2.874e+02 3.326e+02 4.186e+02 6.587e+02, threshold=6.652e+02, percent-clipped=0.0 2023-05-01 20:43:01,414 INFO [train.py:904] (7/8) Epoch 24, batch 6800, loss[loss=0.2378, simple_loss=0.3193, pruned_loss=0.07814, over 11765.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.289, pruned_loss=0.05837, over 3093852.80 frames. ], batch size: 248, lr: 2.80e-03, grad_scale: 8.0 2023-05-01 20:43:32,899 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4969, 3.5243, 2.7391, 2.2213, 2.2985, 2.3599, 3.6859, 3.1356], device='cuda:7'), covar=tensor([0.3015, 0.0623, 0.1749, 0.2630, 0.2719, 0.2182, 0.0435, 0.1366], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0273, 0.0308, 0.0321, 0.0302, 0.0266, 0.0300, 0.0343], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 20:43:42,195 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=240278.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:44:21,125 INFO [train.py:904] (7/8) Epoch 24, batch 6850, loss[loss=0.1964, simple_loss=0.2926, pruned_loss=0.05013, over 16350.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2896, pruned_loss=0.05848, over 3093978.35 frames. ], batch size: 165, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:44:56,068 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=240326.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:45:00,026 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.623e+02 3.086e+02 3.760e+02 7.465e+02, threshold=6.171e+02, percent-clipped=2.0 2023-05-01 20:45:17,596 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=240340.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 20:45:35,447 INFO [train.py:904] (7/8) Epoch 24, batch 6900, loss[loss=0.238, simple_loss=0.338, pruned_loss=0.06899, over 16389.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2923, pruned_loss=0.05819, over 3098408.07 frames. ], batch size: 146, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:45:44,233 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2932, 3.9701, 4.0167, 2.4820, 3.6573, 4.0375, 3.6011, 2.2413], device='cuda:7'), covar=tensor([0.0604, 0.0073, 0.0060, 0.0486, 0.0108, 0.0123, 0.0115, 0.0493], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0086, 0.0086, 0.0134, 0.0099, 0.0111, 0.0096, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 20:46:05,022 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.64 vs. limit=2.0 2023-05-01 20:46:50,840 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=240401.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 20:46:52,768 INFO [train.py:904] (7/8) Epoch 24, batch 6950, loss[loss=0.1932, simple_loss=0.2863, pruned_loss=0.05004, over 16751.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2938, pruned_loss=0.05945, over 3094568.12 frames. ], batch size: 89, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:47:33,343 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.139e+02 2.974e+02 3.564e+02 4.384e+02 8.202e+02, threshold=7.128e+02, percent-clipped=2.0 2023-05-01 20:47:43,387 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1083, 4.1618, 4.4543, 4.4211, 4.4308, 4.1857, 4.1783, 4.1106], device='cuda:7'), covar=tensor([0.0371, 0.0592, 0.0404, 0.0391, 0.0477, 0.0417, 0.0954, 0.0532], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0463, 0.0450, 0.0414, 0.0496, 0.0470, 0.0553, 0.0375], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 20:48:07,679 INFO [train.py:904] (7/8) Epoch 24, batch 7000, loss[loss=0.1822, simple_loss=0.2804, pruned_loss=0.042, over 17041.00 frames. ], tot_loss[loss=0.2064, simple_loss=0.2944, pruned_loss=0.05922, over 3092035.16 frames. ], batch size: 50, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:48:10,126 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9879, 4.8492, 4.7715, 3.2622, 4.3611, 4.8576, 4.2650, 2.7417], device='cuda:7'), covar=tensor([0.0502, 0.0061, 0.0048, 0.0366, 0.0076, 0.0100, 0.0074, 0.0456], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0086, 0.0086, 0.0134, 0.0099, 0.0111, 0.0096, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 20:49:23,910 INFO [train.py:904] (7/8) Epoch 24, batch 7050, loss[loss=0.1952, simple_loss=0.2785, pruned_loss=0.05594, over 16688.00 frames. ], tot_loss[loss=0.2076, simple_loss=0.2955, pruned_loss=0.05992, over 3077253.40 frames. ], batch size: 57, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:49:39,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1729, 4.2494, 4.5434, 4.5042, 4.5038, 4.2359, 4.2317, 4.1649], device='cuda:7'), covar=tensor([0.0347, 0.0592, 0.0353, 0.0390, 0.0519, 0.0414, 0.0923, 0.0529], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0465, 0.0451, 0.0416, 0.0499, 0.0473, 0.0555, 0.0378], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 20:50:06,657 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.536e+02 3.115e+02 4.008e+02 6.614e+02, threshold=6.229e+02, percent-clipped=0.0 2023-05-01 20:50:21,015 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5269, 3.0399, 3.2157, 2.0428, 2.8020, 2.1075, 3.1839, 3.2614], device='cuda:7'), covar=tensor([0.0296, 0.0822, 0.0554, 0.2055, 0.0888, 0.1036, 0.0653, 0.0919], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0169, 0.0155, 0.0147, 0.0132, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 20:50:42,275 INFO [train.py:904] (7/8) Epoch 24, batch 7100, loss[loss=0.2322, simple_loss=0.2958, pruned_loss=0.08432, over 10933.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2938, pruned_loss=0.0599, over 3057652.11 frames. ], batch size: 246, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:51:22,044 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-01 20:51:59,042 INFO [train.py:904] (7/8) Epoch 24, batch 7150, loss[loss=0.2109, simple_loss=0.2881, pruned_loss=0.0669, over 16223.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2919, pruned_loss=0.05974, over 3050241.25 frames. ], batch size: 165, lr: 2.80e-03, grad_scale: 2.0 2023-05-01 20:52:39,166 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.962e+02 2.941e+02 3.518e+02 4.081e+02 6.999e+02, threshold=7.036e+02, percent-clipped=1.0 2023-05-01 20:52:54,827 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-01 20:53:12,644 INFO [train.py:904] (7/8) Epoch 24, batch 7200, loss[loss=0.181, simple_loss=0.2824, pruned_loss=0.03987, over 15458.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2892, pruned_loss=0.05759, over 3058032.78 frames. ], batch size: 191, lr: 2.80e-03, grad_scale: 4.0 2023-05-01 20:53:23,828 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-05-01 20:53:45,085 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-01 20:53:49,934 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5392, 4.8258, 4.4724, 4.6828, 4.3942, 4.3167, 4.4010, 4.9197], device='cuda:7'), covar=tensor([0.2252, 0.1485, 0.2257, 0.1458, 0.1562, 0.1893, 0.2175, 0.1728], device='cuda:7'), in_proj_covar=tensor([0.0688, 0.0826, 0.0685, 0.0640, 0.0528, 0.0532, 0.0696, 0.0650], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:54:10,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8938, 2.7662, 2.6437, 1.8896, 2.6087, 2.7250, 2.5938, 1.9215], device='cuda:7'), covar=tensor([0.0453, 0.0093, 0.0090, 0.0393, 0.0142, 0.0129, 0.0136, 0.0405], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0085, 0.0086, 0.0133, 0.0098, 0.0110, 0.0095, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 20:54:21,211 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=240696.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 20:54:32,010 INFO [train.py:904] (7/8) Epoch 24, batch 7250, loss[loss=0.2139, simple_loss=0.2978, pruned_loss=0.06499, over 11623.00 frames. ], tot_loss[loss=0.2, simple_loss=0.287, pruned_loss=0.05651, over 3051666.24 frames. ], batch size: 247, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:55:12,204 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.642e+02 3.264e+02 4.366e+02 7.559e+02, threshold=6.528e+02, percent-clipped=1.0 2023-05-01 20:55:45,073 INFO [train.py:904] (7/8) Epoch 24, batch 7300, loss[loss=0.2228, simple_loss=0.309, pruned_loss=0.06832, over 15351.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2867, pruned_loss=0.05645, over 3046082.88 frames. ], batch size: 190, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:56:10,751 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-01 20:56:42,359 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8551, 1.3688, 1.7763, 1.6889, 1.7935, 1.9745, 1.6338, 1.8045], device='cuda:7'), covar=tensor([0.0256, 0.0402, 0.0234, 0.0311, 0.0277, 0.0185, 0.0442, 0.0134], device='cuda:7'), in_proj_covar=tensor([0.0190, 0.0193, 0.0180, 0.0183, 0.0199, 0.0158, 0.0197, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 20:57:02,843 INFO [train.py:904] (7/8) Epoch 24, batch 7350, loss[loss=0.1947, simple_loss=0.2862, pruned_loss=0.05162, over 16672.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2872, pruned_loss=0.05743, over 3031865.29 frames. ], batch size: 62, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:57:44,660 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.735e+02 3.201e+02 4.098e+02 9.519e+02, threshold=6.402e+02, percent-clipped=6.0 2023-05-01 20:58:18,679 INFO [train.py:904] (7/8) Epoch 24, batch 7400, loss[loss=0.2381, simple_loss=0.3038, pruned_loss=0.08622, over 11351.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2885, pruned_loss=0.05789, over 3055009.87 frames. ], batch size: 246, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 20:59:34,864 INFO [train.py:904] (7/8) Epoch 24, batch 7450, loss[loss=0.2258, simple_loss=0.2999, pruned_loss=0.07587, over 11436.00 frames. ], tot_loss[loss=0.2045, simple_loss=0.2904, pruned_loss=0.05929, over 3066537.11 frames. ], batch size: 248, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:00:19,564 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.977e+02 2.843e+02 3.471e+02 4.367e+02 9.441e+02, threshold=6.943e+02, percent-clipped=5.0 2023-05-01 21:00:22,621 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=240932.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:00:43,038 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 21:00:55,597 INFO [train.py:904] (7/8) Epoch 24, batch 7500, loss[loss=0.2025, simple_loss=0.2834, pruned_loss=0.06075, over 17061.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2901, pruned_loss=0.05839, over 3058915.80 frames. ], batch size: 55, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:01:19,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8237, 3.8749, 3.9539, 3.7616, 3.8650, 4.2392, 3.9196, 3.6119], device='cuda:7'), covar=tensor([0.2032, 0.2101, 0.2601, 0.2389, 0.2570, 0.1823, 0.1651, 0.2493], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0609, 0.0673, 0.0500, 0.0663, 0.0699, 0.0523, 0.0667], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 21:01:46,117 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=240986.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:01:57,386 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=240993.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 21:02:00,998 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=240996.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:02:11,657 INFO [train.py:904] (7/8) Epoch 24, batch 7550, loss[loss=0.1898, simple_loss=0.2788, pruned_loss=0.05046, over 16669.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2894, pruned_loss=0.05878, over 3040103.81 frames. ], batch size: 134, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:02:53,685 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.697e+02 3.200e+02 3.848e+02 7.597e+02, threshold=6.400e+02, percent-clipped=1.0 2023-05-01 21:03:15,916 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=241044.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:03:20,302 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241047.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:03:29,258 INFO [train.py:904] (7/8) Epoch 24, batch 7600, loss[loss=0.1976, simple_loss=0.2822, pruned_loss=0.05651, over 16956.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2886, pruned_loss=0.05931, over 3026541.00 frames. ], batch size: 109, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:04:14,926 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9404, 3.8184, 3.9530, 4.1202, 4.2018, 3.8587, 4.1320, 4.2167], device='cuda:7'), covar=tensor([0.1757, 0.1327, 0.1640, 0.0844, 0.0766, 0.1874, 0.1110, 0.0863], device='cuda:7'), in_proj_covar=tensor([0.0633, 0.0783, 0.0899, 0.0788, 0.0607, 0.0627, 0.0657, 0.0762], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:04:47,289 INFO [train.py:904] (7/8) Epoch 24, batch 7650, loss[loss=0.2121, simple_loss=0.2987, pruned_loss=0.06276, over 16391.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2882, pruned_loss=0.05884, over 3041581.18 frames. ], batch size: 146, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:05:30,620 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.802e+02 2.751e+02 3.488e+02 4.821e+02 1.001e+03, threshold=6.976e+02, percent-clipped=5.0 2023-05-01 21:06:04,726 INFO [train.py:904] (7/8) Epoch 24, batch 7700, loss[loss=0.215, simple_loss=0.2906, pruned_loss=0.06971, over 16537.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2877, pruned_loss=0.05899, over 3054723.35 frames. ], batch size: 62, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:06:58,096 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8553, 2.7152, 2.6376, 1.9134, 2.5963, 2.6986, 2.6024, 1.9216], device='cuda:7'), covar=tensor([0.0533, 0.0110, 0.0106, 0.0437, 0.0149, 0.0144, 0.0132, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0086, 0.0087, 0.0135, 0.0099, 0.0110, 0.0095, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 21:07:09,422 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3666, 2.9506, 2.6955, 2.2698, 2.2731, 2.3053, 2.9774, 2.8452], device='cuda:7'), covar=tensor([0.2494, 0.0704, 0.1563, 0.2406, 0.2155, 0.2102, 0.0526, 0.1227], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0269, 0.0306, 0.0317, 0.0299, 0.0265, 0.0298, 0.0339], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 21:07:23,963 INFO [train.py:904] (7/8) Epoch 24, batch 7750, loss[loss=0.1893, simple_loss=0.2868, pruned_loss=0.04593, over 16924.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2884, pruned_loss=0.05924, over 3049493.91 frames. ], batch size: 90, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:08:06,389 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.032e+02 2.844e+02 3.408e+02 3.883e+02 6.664e+02, threshold=6.815e+02, percent-clipped=0.0 2023-05-01 21:08:38,523 INFO [train.py:904] (7/8) Epoch 24, batch 7800, loss[loss=0.1919, simple_loss=0.2831, pruned_loss=0.05036, over 16459.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2893, pruned_loss=0.05934, over 3059871.24 frames. ], batch size: 75, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:08:52,961 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2782, 5.6405, 5.3235, 5.3848, 5.0789, 5.0242, 4.9511, 5.7157], device='cuda:7'), covar=tensor([0.1384, 0.0896, 0.1097, 0.0938, 0.0913, 0.0777, 0.1265, 0.0922], device='cuda:7'), in_proj_covar=tensor([0.0689, 0.0825, 0.0687, 0.0642, 0.0527, 0.0534, 0.0696, 0.0648], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:09:34,097 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241288.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:09:56,321 INFO [train.py:904] (7/8) Epoch 24, batch 7850, loss[loss=0.214, simple_loss=0.3002, pruned_loss=0.06385, over 16288.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2904, pruned_loss=0.05944, over 3056522.84 frames. ], batch size: 165, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:10:38,481 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.020e+02 2.608e+02 3.095e+02 3.566e+02 5.691e+02, threshold=6.191e+02, percent-clipped=0.0 2023-05-01 21:10:39,273 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.58 vs. limit=2.0 2023-05-01 21:10:55,478 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241342.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:11:11,149 INFO [train.py:904] (7/8) Epoch 24, batch 7900, loss[loss=0.2051, simple_loss=0.2994, pruned_loss=0.05546, over 16208.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2894, pruned_loss=0.05873, over 3065960.24 frames. ], batch size: 165, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:11:15,850 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6309, 3.7221, 2.4652, 4.3516, 2.8719, 4.2671, 2.4308, 3.0284], device='cuda:7'), covar=tensor([0.0309, 0.0410, 0.1561, 0.0255, 0.0849, 0.0717, 0.1607, 0.0850], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0179, 0.0196, 0.0167, 0.0179, 0.0219, 0.0204, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 21:12:29,005 INFO [train.py:904] (7/8) Epoch 24, batch 7950, loss[loss=0.1978, simple_loss=0.2859, pruned_loss=0.05482, over 16781.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2895, pruned_loss=0.05895, over 3058950.84 frames. ], batch size: 83, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:12:35,019 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241406.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:13:12,619 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.937e+02 2.663e+02 3.306e+02 3.929e+02 7.099e+02, threshold=6.611e+02, percent-clipped=2.0 2023-05-01 21:13:46,688 INFO [train.py:904] (7/8) Epoch 24, batch 8000, loss[loss=0.176, simple_loss=0.269, pruned_loss=0.04148, over 16548.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2894, pruned_loss=0.05833, over 3089158.84 frames. ], batch size: 68, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:14:10,148 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241467.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 21:15:04,299 INFO [train.py:904] (7/8) Epoch 24, batch 8050, loss[loss=0.1848, simple_loss=0.2758, pruned_loss=0.04689, over 16129.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2894, pruned_loss=0.05799, over 3100211.71 frames. ], batch size: 165, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:15:47,626 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.925e+02 2.671e+02 3.058e+02 3.888e+02 6.821e+02, threshold=6.115e+02, percent-clipped=1.0 2023-05-01 21:16:22,148 INFO [train.py:904] (7/8) Epoch 24, batch 8100, loss[loss=0.2124, simple_loss=0.296, pruned_loss=0.06439, over 16720.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.289, pruned_loss=0.0574, over 3095119.58 frames. ], batch size: 134, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:17:14,382 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=241588.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:17:20,718 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-01 21:17:38,332 INFO [train.py:904] (7/8) Epoch 24, batch 8150, loss[loss=0.1671, simple_loss=0.2535, pruned_loss=0.0404, over 17041.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2868, pruned_loss=0.05731, over 3079881.60 frames. ], batch size: 50, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:18:06,904 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241621.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:18:22,061 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.026e+02 2.691e+02 3.345e+02 4.318e+02 1.089e+03, threshold=6.691e+02, percent-clipped=5.0 2023-05-01 21:18:30,850 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=241636.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:18:37,199 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8504, 2.1995, 2.4260, 3.1071, 2.2479, 2.3992, 2.3503, 2.3129], device='cuda:7'), covar=tensor([0.1423, 0.3278, 0.2433, 0.0722, 0.3986, 0.2288, 0.3302, 0.3172], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0459, 0.0374, 0.0331, 0.0441, 0.0524, 0.0429, 0.0535], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:18:39,510 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=241642.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:18:55,846 INFO [train.py:904] (7/8) Epoch 24, batch 8200, loss[loss=0.2074, simple_loss=0.2829, pruned_loss=0.06593, over 11854.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.284, pruned_loss=0.05638, over 3082559.49 frames. ], batch size: 246, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:19:38,370 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7976, 3.6950, 3.8944, 3.9685, 4.0759, 3.6629, 4.0040, 4.1117], device='cuda:7'), covar=tensor([0.1741, 0.1210, 0.1327, 0.0731, 0.0657, 0.1904, 0.0945, 0.0801], device='cuda:7'), in_proj_covar=tensor([0.0638, 0.0789, 0.0904, 0.0793, 0.0609, 0.0627, 0.0661, 0.0767], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:19:43,535 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241682.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:19:53,222 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-01 21:19:56,098 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=241690.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:20:16,379 INFO [train.py:904] (7/8) Epoch 24, batch 8250, loss[loss=0.2097, simple_loss=0.3105, pruned_loss=0.0544, over 15287.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.283, pruned_loss=0.05372, over 3074836.81 frames. ], batch size: 191, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:21:03,234 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.735e+02 2.441e+02 2.851e+02 3.404e+02 7.117e+02, threshold=5.702e+02, percent-clipped=1.0 2023-05-01 21:21:05,726 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.30 vs. limit=5.0 2023-05-01 21:21:20,197 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9577, 2.7002, 2.6171, 2.0402, 2.5470, 2.7585, 2.6406, 1.9440], device='cuda:7'), covar=tensor([0.0413, 0.0102, 0.0100, 0.0345, 0.0137, 0.0125, 0.0114, 0.0458], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0085, 0.0086, 0.0133, 0.0098, 0.0110, 0.0095, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 21:21:39,007 INFO [train.py:904] (7/8) Epoch 24, batch 8300, loss[loss=0.1766, simple_loss=0.2838, pruned_loss=0.03467, over 16865.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2802, pruned_loss=0.05105, over 3055780.50 frames. ], batch size: 102, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:21:41,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2346, 4.2043, 4.1041, 3.2763, 4.1615, 1.6168, 3.9249, 3.6856], device='cuda:7'), covar=tensor([0.0115, 0.0114, 0.0205, 0.0346, 0.0104, 0.3090, 0.0153, 0.0307], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0162, 0.0203, 0.0179, 0.0178, 0.0208, 0.0191, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:21:55,103 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241762.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:23:02,691 INFO [train.py:904] (7/8) Epoch 24, batch 8350, loss[loss=0.1825, simple_loss=0.2855, pruned_loss=0.03977, over 16652.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2788, pruned_loss=0.04911, over 3035528.62 frames. ], batch size: 134, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:23:03,892 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241803.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:23:37,830 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4063, 3.0084, 3.0638, 1.9383, 3.3011, 3.3146, 2.8335, 2.7904], device='cuda:7'), covar=tensor([0.0672, 0.0264, 0.0270, 0.1124, 0.0094, 0.0223, 0.0459, 0.0429], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0107, 0.0097, 0.0135, 0.0080, 0.0125, 0.0126, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 21:23:48,200 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.114e+02 2.559e+02 3.113e+02 5.727e+02, threshold=5.118e+02, percent-clipped=1.0 2023-05-01 21:23:48,852 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=241831.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 21:24:23,529 INFO [train.py:904] (7/8) Epoch 24, batch 8400, loss[loss=0.204, simple_loss=0.2976, pruned_loss=0.05517, over 16701.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2763, pruned_loss=0.04729, over 3023933.59 frames. ], batch size: 134, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:24:42,791 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241864.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:25:14,974 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1302, 2.2036, 2.1697, 3.7340, 2.0933, 2.4830, 2.2894, 2.2956], device='cuda:7'), covar=tensor([0.1316, 0.3711, 0.3273, 0.0601, 0.4578, 0.2660, 0.3860, 0.3496], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0450, 0.0368, 0.0324, 0.0433, 0.0513, 0.0421, 0.0525], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:25:28,870 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=241892.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:25:45,954 INFO [train.py:904] (7/8) Epoch 24, batch 8450, loss[loss=0.1729, simple_loss=0.2583, pruned_loss=0.04378, over 12467.00 frames. ], tot_loss[loss=0.1829, simple_loss=0.2745, pruned_loss=0.04558, over 3028172.63 frames. ], batch size: 247, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:26:31,272 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.077e+02 2.419e+02 2.958e+02 5.415e+02, threshold=4.839e+02, percent-clipped=2.0 2023-05-01 21:27:07,289 INFO [train.py:904] (7/8) Epoch 24, batch 8500, loss[loss=0.1561, simple_loss=0.2532, pruned_loss=0.02948, over 16718.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2704, pruned_loss=0.04305, over 3025894.35 frames. ], batch size: 89, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:27:47,689 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=241977.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:28:00,058 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5675, 4.8420, 4.6635, 4.6680, 4.4240, 4.4356, 4.3640, 4.9162], device='cuda:7'), covar=tensor([0.1360, 0.1088, 0.1088, 0.0872, 0.0886, 0.1234, 0.1108, 0.0929], device='cuda:7'), in_proj_covar=tensor([0.0684, 0.0821, 0.0682, 0.0638, 0.0522, 0.0530, 0.0692, 0.0644], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:28:33,832 INFO [train.py:904] (7/8) Epoch 24, batch 8550, loss[loss=0.1728, simple_loss=0.2624, pruned_loss=0.04162, over 16705.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2684, pruned_loss=0.042, over 3032144.27 frames. ], batch size: 57, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:29:05,833 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 21:29:26,918 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.227e+02 2.590e+02 3.113e+02 4.507e+02, threshold=5.180e+02, percent-clipped=0.0 2023-05-01 21:30:13,219 INFO [train.py:904] (7/8) Epoch 24, batch 8600, loss[loss=0.1755, simple_loss=0.2622, pruned_loss=0.04443, over 12515.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2692, pruned_loss=0.04135, over 3038728.05 frames. ], batch size: 248, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:30:31,818 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242062.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:31:51,615 INFO [train.py:904] (7/8) Epoch 24, batch 8650, loss[loss=0.1664, simple_loss=0.2666, pruned_loss=0.0331, over 16615.00 frames. ], tot_loss[loss=0.1739, simple_loss=0.2677, pruned_loss=0.04008, over 3040843.33 frames. ], batch size: 62, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:32:10,579 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242110.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:32:56,844 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.159e+02 2.524e+02 3.190e+02 5.272e+02, threshold=5.049e+02, percent-clipped=1.0 2023-05-01 21:33:30,894 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-01 21:33:37,198 INFO [train.py:904] (7/8) Epoch 24, batch 8700, loss[loss=0.1664, simple_loss=0.2542, pruned_loss=0.03923, over 12314.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2644, pruned_loss=0.03868, over 3028770.33 frames. ], batch size: 248, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:33:38,290 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242153.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:33:38,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1794, 2.1704, 2.2633, 3.7898, 2.1299, 2.5166, 2.3192, 2.3280], device='cuda:7'), covar=tensor([0.1313, 0.3783, 0.3138, 0.0547, 0.4250, 0.2627, 0.3707, 0.3430], device='cuda:7'), in_proj_covar=tensor([0.0399, 0.0448, 0.0367, 0.0322, 0.0431, 0.0510, 0.0419, 0.0522], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:33:50,501 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242159.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:34:42,480 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242187.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 21:35:13,454 INFO [train.py:904] (7/8) Epoch 24, batch 8750, loss[loss=0.16, simple_loss=0.2495, pruned_loss=0.03522, over 11916.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2637, pruned_loss=0.03806, over 3037474.84 frames. ], batch size: 247, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:35:42,400 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242214.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:36:10,682 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242226.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:36:21,486 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.077e+02 2.590e+02 3.155e+02 6.830e+02, threshold=5.181e+02, percent-clipped=2.0 2023-05-01 21:37:05,088 INFO [train.py:904] (7/8) Epoch 24, batch 8800, loss[loss=0.159, simple_loss=0.2573, pruned_loss=0.03037, over 15461.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2616, pruned_loss=0.03669, over 3032540.43 frames. ], batch size: 191, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:37:56,253 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242277.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:38:18,007 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242287.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:38:50,071 INFO [train.py:904] (7/8) Epoch 24, batch 8850, loss[loss=0.1803, simple_loss=0.2823, pruned_loss=0.03913, over 16937.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2653, pruned_loss=0.03696, over 3037269.40 frames. ], batch size: 116, lr: 2.79e-03, grad_scale: 8.0 2023-05-01 21:39:03,950 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242309.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:39:38,583 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242325.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:39:52,558 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.137e+02 2.486e+02 3.124e+02 5.230e+02, threshold=4.972e+02, percent-clipped=1.0 2023-05-01 21:40:38,022 INFO [train.py:904] (7/8) Epoch 24, batch 8900, loss[loss=0.1752, simple_loss=0.2738, pruned_loss=0.03824, over 16305.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2665, pruned_loss=0.03633, over 3064517.13 frames. ], batch size: 166, lr: 2.79e-03, grad_scale: 4.0 2023-05-01 21:41:12,538 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242370.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:42:44,447 INFO [train.py:904] (7/8) Epoch 24, batch 8950, loss[loss=0.1566, simple_loss=0.2506, pruned_loss=0.03129, over 16990.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2661, pruned_loss=0.03706, over 3052079.99 frames. ], batch size: 116, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:43:49,727 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.299e+02 2.560e+02 2.923e+02 8.049e+02, threshold=5.120e+02, percent-clipped=2.0 2023-05-01 21:44:35,826 INFO [train.py:904] (7/8) Epoch 24, batch 9000, loss[loss=0.1541, simple_loss=0.2508, pruned_loss=0.02869, over 16275.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2627, pruned_loss=0.03585, over 3055389.55 frames. ], batch size: 166, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:44:35,826 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 21:44:42,676 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6297, 3.2177, 3.3534, 2.1016, 3.5748, 3.6071, 3.0034, 2.9958], device='cuda:7'), covar=tensor([0.0643, 0.0258, 0.0217, 0.1053, 0.0078, 0.0155, 0.0391, 0.0385], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0105, 0.0095, 0.0133, 0.0079, 0.0122, 0.0123, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-01 21:44:45,529 INFO [train.py:938] (7/8) Epoch 24, validation: loss=0.1445, simple_loss=0.2484, pruned_loss=0.02026, over 944034.00 frames. 2023-05-01 21:44:45,530 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 21:44:59,714 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242459.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:45:49,147 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242483.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:45:58,863 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242487.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 21:46:28,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8367, 4.2662, 3.1141, 2.3998, 2.6542, 2.6243, 4.5766, 3.4728], device='cuda:7'), covar=tensor([0.2939, 0.0474, 0.1784, 0.2923, 0.2738, 0.2063, 0.0311, 0.1360], device='cuda:7'), in_proj_covar=tensor([0.0323, 0.0265, 0.0301, 0.0312, 0.0291, 0.0262, 0.0292, 0.0334], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 21:46:30,346 INFO [train.py:904] (7/8) Epoch 24, batch 9050, loss[loss=0.1808, simple_loss=0.263, pruned_loss=0.04936, over 12834.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2633, pruned_loss=0.03618, over 3058478.52 frames. ], batch size: 248, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:46:40,250 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242507.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:46:45,346 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242509.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:47:02,627 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8683, 1.4331, 1.7501, 1.8033, 1.9290, 1.9849, 1.7406, 1.8630], device='cuda:7'), covar=tensor([0.0259, 0.0458, 0.0276, 0.0323, 0.0321, 0.0197, 0.0444, 0.0153], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0191, 0.0179, 0.0181, 0.0196, 0.0155, 0.0195, 0.0153], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:47:30,080 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.227e+02 2.575e+02 3.137e+02 5.273e+02, threshold=5.150e+02, percent-clipped=1.0 2023-05-01 21:47:35,758 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242535.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 21:47:38,006 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-01 21:48:00,808 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242544.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:48:17,903 INFO [train.py:904] (7/8) Epoch 24, batch 9100, loss[loss=0.1695, simple_loss=0.2718, pruned_loss=0.03356, over 16176.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2627, pruned_loss=0.03639, over 3060163.60 frames. ], batch size: 165, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:48:23,527 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242555.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:49:27,701 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242582.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:50:16,318 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242602.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:50:16,543 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 21:50:16,962 INFO [train.py:904] (7/8) Epoch 24, batch 9150, loss[loss=0.1513, simple_loss=0.2476, pruned_loss=0.02749, over 15234.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.263, pruned_loss=0.03604, over 3052743.40 frames. ], batch size: 191, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 21:50:47,229 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242616.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:51:18,343 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7723, 4.6032, 4.8358, 4.9919, 5.1250, 4.6061, 5.1408, 5.1459], device='cuda:7'), covar=tensor([0.1923, 0.1243, 0.1498, 0.0665, 0.0597, 0.0930, 0.0671, 0.0978], device='cuda:7'), in_proj_covar=tensor([0.0624, 0.0771, 0.0885, 0.0779, 0.0597, 0.0615, 0.0647, 0.0749], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:51:21,274 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.667e+02 2.227e+02 2.578e+02 3.148e+02 6.071e+02, threshold=5.155e+02, percent-clipped=3.0 2023-05-01 21:52:01,746 INFO [train.py:904] (7/8) Epoch 24, batch 9200, loss[loss=0.1684, simple_loss=0.2607, pruned_loss=0.038, over 16707.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2588, pruned_loss=0.03538, over 3051046.76 frames. ], batch size: 134, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:52:22,372 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242663.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:52:24,946 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242665.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:52:34,081 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-01 21:53:35,781 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-05-01 21:53:38,109 INFO [train.py:904] (7/8) Epoch 24, batch 9250, loss[loss=0.1444, simple_loss=0.2291, pruned_loss=0.02991, over 12495.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2585, pruned_loss=0.03535, over 3040725.92 frames. ], batch size: 247, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:54:40,871 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.148e+02 2.469e+02 2.996e+02 5.681e+02, threshold=4.938e+02, percent-clipped=3.0 2023-05-01 21:55:27,533 INFO [train.py:904] (7/8) Epoch 24, batch 9300, loss[loss=0.1596, simple_loss=0.2549, pruned_loss=0.03219, over 15161.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2582, pruned_loss=0.03528, over 3042948.76 frames. ], batch size: 190, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:56:13,909 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242772.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:56:44,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6028, 1.7560, 2.1353, 2.5615, 2.4780, 2.8606, 2.0738, 2.8210], device='cuda:7'), covar=tensor([0.0247, 0.0620, 0.0422, 0.0374, 0.0366, 0.0244, 0.0534, 0.0212], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0189, 0.0177, 0.0179, 0.0195, 0.0155, 0.0194, 0.0152], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:57:14,583 INFO [train.py:904] (7/8) Epoch 24, batch 9350, loss[loss=0.1733, simple_loss=0.2569, pruned_loss=0.04483, over 12407.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2577, pruned_loss=0.03509, over 3042689.51 frames. ], batch size: 247, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:57:27,669 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242809.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:57:54,773 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9460, 2.2194, 2.4644, 3.2272, 2.2592, 2.3816, 2.3825, 2.3314], device='cuda:7'), covar=tensor([0.1315, 0.3574, 0.2777, 0.0715, 0.4531, 0.2709, 0.3552, 0.3516], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0447, 0.0369, 0.0322, 0.0432, 0.0509, 0.0419, 0.0522], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 21:58:13,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 1.952e+02 2.311e+02 3.086e+02 6.417e+02, threshold=4.623e+02, percent-clipped=1.0 2023-05-01 21:58:14,261 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242833.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:58:25,513 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242839.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:58:54,554 INFO [train.py:904] (7/8) Epoch 24, batch 9400, loss[loss=0.1369, simple_loss=0.2268, pruned_loss=0.02346, over 12631.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2574, pruned_loss=0.03485, over 3035561.35 frames. ], batch size: 247, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 21:59:03,732 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242857.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 21:59:54,998 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242882.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:00:35,988 INFO [train.py:904] (7/8) Epoch 24, batch 9450, loss[loss=0.1798, simple_loss=0.2688, pruned_loss=0.04536, over 12290.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2596, pruned_loss=0.03516, over 3041297.81 frames. ], batch size: 250, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:00:51,613 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242911.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:01:14,128 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=242921.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:01:32,437 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=242930.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:01:37,834 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 2.255e+02 2.567e+02 3.168e+02 7.490e+02, threshold=5.133e+02, percent-clipped=6.0 2023-05-01 22:02:16,749 INFO [train.py:904] (7/8) Epoch 24, batch 9500, loss[loss=0.1628, simple_loss=0.2562, pruned_loss=0.03468, over 12566.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2592, pruned_loss=0.03493, over 3043703.50 frames. ], batch size: 246, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:02:30,117 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=242958.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:02:43,772 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=242965.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:02:45,768 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-01 22:02:59,237 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7475, 4.9278, 5.0967, 4.8277, 4.8976, 5.4490, 4.9930, 4.6135], device='cuda:7'), covar=tensor([0.0952, 0.1746, 0.2142, 0.2048, 0.2394, 0.0816, 0.1547, 0.2523], device='cuda:7'), in_proj_covar=tensor([0.0393, 0.0586, 0.0645, 0.0480, 0.0634, 0.0673, 0.0501, 0.0637], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:03:16,286 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=242982.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 22:04:01,165 INFO [train.py:904] (7/8) Epoch 24, batch 9550, loss[loss=0.1968, simple_loss=0.2923, pruned_loss=0.05062, over 16681.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2589, pruned_loss=0.03512, over 3046344.08 frames. ], batch size: 134, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:04:24,735 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243013.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:05:06,081 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.055e+02 2.389e+02 2.872e+02 5.971e+02, threshold=4.777e+02, percent-clipped=1.0 2023-05-01 22:05:43,252 INFO [train.py:904] (7/8) Epoch 24, batch 9600, loss[loss=0.1714, simple_loss=0.2741, pruned_loss=0.03435, over 16902.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2602, pruned_loss=0.03585, over 3050555.74 frames. ], batch size: 96, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:05:54,302 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1031, 4.0975, 4.4243, 4.3885, 4.4288, 4.1998, 4.1907, 4.1682], device='cuda:7'), covar=tensor([0.0419, 0.0765, 0.0449, 0.0481, 0.0498, 0.0435, 0.0876, 0.0464], device='cuda:7'), in_proj_covar=tensor([0.0402, 0.0448, 0.0438, 0.0403, 0.0480, 0.0456, 0.0532, 0.0367], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 22:06:01,515 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.31 vs. limit=5.0 2023-05-01 22:07:32,101 INFO [train.py:904] (7/8) Epoch 24, batch 9650, loss[loss=0.1651, simple_loss=0.263, pruned_loss=0.03363, over 16212.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2617, pruned_loss=0.03627, over 3033132.13 frames. ], batch size: 165, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:08:29,868 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=243128.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:08:37,736 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.193e+02 2.679e+02 3.134e+02 6.217e+02, threshold=5.359e+02, percent-clipped=4.0 2023-05-01 22:08:42,597 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4582, 3.0901, 2.7204, 2.2837, 2.1965, 2.3119, 3.0017, 2.8691], device='cuda:7'), covar=tensor([0.2555, 0.0682, 0.1713, 0.2815, 0.2695, 0.2334, 0.0447, 0.1400], device='cuda:7'), in_proj_covar=tensor([0.0321, 0.0263, 0.0300, 0.0310, 0.0288, 0.0260, 0.0290, 0.0331], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:08:50,201 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243139.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:09:19,550 INFO [train.py:904] (7/8) Epoch 24, batch 9700, loss[loss=0.1571, simple_loss=0.2486, pruned_loss=0.03284, over 12470.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.261, pruned_loss=0.03568, over 3063843.44 frames. ], batch size: 248, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:10:30,939 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243187.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:11:01,239 INFO [train.py:904] (7/8) Epoch 24, batch 9750, loss[loss=0.1583, simple_loss=0.2566, pruned_loss=0.03, over 16795.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2603, pruned_loss=0.03615, over 3073524.74 frames. ], batch size: 83, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:11:11,007 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9775, 1.8400, 1.6642, 1.4961, 1.9954, 1.6398, 1.6014, 1.9176], device='cuda:7'), covar=tensor([0.0198, 0.0331, 0.0453, 0.0406, 0.0256, 0.0313, 0.0199, 0.0233], device='cuda:7'), in_proj_covar=tensor([0.0208, 0.0232, 0.0223, 0.0224, 0.0233, 0.0232, 0.0229, 0.0226], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:11:17,001 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243211.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:11:38,075 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-05-01 22:11:49,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7157, 2.7012, 2.4382, 2.5982, 3.0610, 2.8391, 3.2252, 3.3099], device='cuda:7'), covar=tensor([0.0147, 0.0505, 0.0582, 0.0526, 0.0410, 0.0439, 0.0303, 0.0308], device='cuda:7'), in_proj_covar=tensor([0.0209, 0.0233, 0.0224, 0.0225, 0.0234, 0.0233, 0.0229, 0.0228], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:12:03,364 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 1.977e+02 2.392e+02 3.038e+02 5.323e+02, threshold=4.783e+02, percent-clipped=0.0 2023-05-01 22:12:37,395 INFO [train.py:904] (7/8) Epoch 24, batch 9800, loss[loss=0.1576, simple_loss=0.2421, pruned_loss=0.03654, over 12151.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2597, pruned_loss=0.03518, over 3061797.45 frames. ], batch size: 246, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:12:49,025 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243258.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:12:50,827 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243259.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:13:17,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5378, 2.9936, 3.1923, 1.9575, 2.7766, 2.1890, 2.9747, 3.1804], device='cuda:7'), covar=tensor([0.0422, 0.0967, 0.0608, 0.2201, 0.0982, 0.1051, 0.0950, 0.1123], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0158, 0.0164, 0.0151, 0.0142, 0.0127, 0.0140, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:13:22,237 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=243277.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 22:13:58,694 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=243293.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:14:16,676 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-01 22:14:21,696 INFO [train.py:904] (7/8) Epoch 24, batch 9850, loss[loss=0.1472, simple_loss=0.2368, pruned_loss=0.02881, over 12473.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2604, pruned_loss=0.0349, over 3059953.84 frames. ], batch size: 246, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:14:28,946 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243306.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:14:29,114 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0154, 2.8039, 2.6726, 2.0479, 2.5921, 2.8229, 2.6777, 1.9839], device='cuda:7'), covar=tensor([0.0418, 0.0079, 0.0082, 0.0332, 0.0152, 0.0103, 0.0108, 0.0445], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0084, 0.0084, 0.0132, 0.0097, 0.0107, 0.0093, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:15:22,942 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 1.939e+02 2.274e+02 2.876e+02 6.181e+02, threshold=4.547e+02, percent-clipped=2.0 2023-05-01 22:16:11,636 INFO [train.py:904] (7/8) Epoch 24, batch 9900, loss[loss=0.1496, simple_loss=0.2401, pruned_loss=0.02955, over 12579.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2605, pruned_loss=0.03487, over 3051256.75 frames. ], batch size: 248, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:16:15,546 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=243354.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:18:09,543 INFO [train.py:904] (7/8) Epoch 24, batch 9950, loss[loss=0.1716, simple_loss=0.2679, pruned_loss=0.03759, over 16659.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.264, pruned_loss=0.03545, over 3081493.20 frames. ], batch size: 134, lr: 2.78e-03, grad_scale: 4.0 2023-05-01 22:19:12,944 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243428.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:19:24,339 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-01 22:19:24,683 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.110e+02 2.470e+02 3.179e+02 5.008e+02, threshold=4.941e+02, percent-clipped=4.0 2023-05-01 22:19:47,301 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2545, 4.3673, 4.5024, 4.2750, 4.3578, 4.8476, 4.4296, 4.0822], device='cuda:7'), covar=tensor([0.1561, 0.1966, 0.2116, 0.1990, 0.2454, 0.1040, 0.1593, 0.2510], device='cuda:7'), in_proj_covar=tensor([0.0386, 0.0576, 0.0634, 0.0472, 0.0623, 0.0658, 0.0493, 0.0625], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:20:10,958 INFO [train.py:904] (7/8) Epoch 24, batch 10000, loss[loss=0.165, simple_loss=0.262, pruned_loss=0.03402, over 16899.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2625, pruned_loss=0.03514, over 3082716.34 frames. ], batch size: 116, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:20:25,915 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0062, 3.1987, 3.1674, 2.2081, 2.9845, 3.2298, 3.0960, 1.9332], device='cuda:7'), covar=tensor([0.0566, 0.0056, 0.0070, 0.0391, 0.0124, 0.0095, 0.0080, 0.0558], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0084, 0.0084, 0.0132, 0.0097, 0.0107, 0.0093, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:20:56,064 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243476.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:21:51,126 INFO [train.py:904] (7/8) Epoch 24, batch 10050, loss[loss=0.1793, simple_loss=0.2699, pruned_loss=0.04437, over 16912.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2624, pruned_loss=0.03504, over 3074903.65 frames. ], batch size: 109, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:22:52,106 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.109e+02 2.585e+02 3.087e+02 5.366e+02, threshold=5.170e+02, percent-clipped=3.0 2023-05-01 22:23:25,159 INFO [train.py:904] (7/8) Epoch 24, batch 10100, loss[loss=0.1642, simple_loss=0.2528, pruned_loss=0.03785, over 16180.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2628, pruned_loss=0.03549, over 3052041.34 frames. ], batch size: 165, lr: 2.78e-03, grad_scale: 8.0 2023-05-01 22:23:44,512 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-01 22:24:07,947 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6844, 3.9022, 2.8853, 2.3235, 2.4589, 2.4573, 4.2094, 3.3413], device='cuda:7'), covar=tensor([0.2934, 0.0583, 0.1792, 0.2947, 0.2911, 0.2212, 0.0361, 0.1419], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0261, 0.0297, 0.0307, 0.0284, 0.0257, 0.0287, 0.0328], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:24:16,227 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243577.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 22:24:31,120 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-05-01 22:25:09,211 INFO [train.py:904] (7/8) Epoch 25, batch 0, loss[loss=0.2349, simple_loss=0.3035, pruned_loss=0.08317, over 16871.00 frames. ], tot_loss[loss=0.2349, simple_loss=0.3035, pruned_loss=0.08317, over 16871.00 frames. ], batch size: 116, lr: 2.72e-03, grad_scale: 8.0 2023-05-01 22:25:09,211 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 22:25:16,826 INFO [train.py:938] (7/8) Epoch 25, validation: loss=0.1443, simple_loss=0.2477, pruned_loss=0.02048, over 944034.00 frames. 2023-05-01 22:25:16,827 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 22:25:48,983 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243625.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:25:53,176 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1286, 5.1007, 5.5323, 5.5196, 5.5761, 5.2331, 5.1696, 4.8889], device='cuda:7'), covar=tensor([0.0392, 0.0511, 0.0454, 0.0480, 0.0514, 0.0401, 0.0980, 0.0512], device='cuda:7'), in_proj_covar=tensor([0.0397, 0.0443, 0.0435, 0.0400, 0.0476, 0.0454, 0.0528, 0.0364], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 22:26:03,176 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.517e+02 2.877e+02 3.521e+02 6.755e+02, threshold=5.755e+02, percent-clipped=6.0 2023-05-01 22:26:21,451 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=243649.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:26:26,916 INFO [train.py:904] (7/8) Epoch 25, batch 50, loss[loss=0.1732, simple_loss=0.2595, pruned_loss=0.0435, over 17158.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2645, pruned_loss=0.04652, over 751698.37 frames. ], batch size: 48, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:26:30,754 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-01 22:26:58,532 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4386, 4.2413, 4.4574, 4.6098, 4.7502, 4.3431, 4.5650, 4.7064], device='cuda:7'), covar=tensor([0.1943, 0.1381, 0.1609, 0.0872, 0.0750, 0.1128, 0.2071, 0.1207], device='cuda:7'), in_proj_covar=tensor([0.0621, 0.0764, 0.0878, 0.0774, 0.0594, 0.0609, 0.0642, 0.0743], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:27:35,810 INFO [train.py:904] (7/8) Epoch 25, batch 100, loss[loss=0.1735, simple_loss=0.2521, pruned_loss=0.04748, over 16815.00 frames. ], tot_loss[loss=0.1752, simple_loss=0.2614, pruned_loss=0.0445, over 1321796.09 frames. ], batch size: 102, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:28:00,463 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-01 22:28:12,773 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-01 22:28:22,085 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.579e+02 2.266e+02 2.864e+02 3.484e+02 1.313e+03, threshold=5.728e+02, percent-clipped=5.0 2023-05-01 22:28:37,071 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9397, 2.5437, 2.0448, 2.2635, 2.9177, 2.6568, 2.9833, 2.9796], device='cuda:7'), covar=tensor([0.0225, 0.0410, 0.0574, 0.0531, 0.0268, 0.0387, 0.0220, 0.0319], device='cuda:7'), in_proj_covar=tensor([0.0215, 0.0238, 0.0228, 0.0228, 0.0238, 0.0236, 0.0234, 0.0232], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:28:45,157 INFO [train.py:904] (7/8) Epoch 25, batch 150, loss[loss=0.1648, simple_loss=0.2635, pruned_loss=0.03299, over 17253.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.26, pruned_loss=0.04329, over 1769966.04 frames. ], batch size: 52, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:29:16,197 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2085, 5.1379, 5.0001, 4.5162, 4.6361, 5.0843, 5.0281, 4.6578], device='cuda:7'), covar=tensor([0.0599, 0.0637, 0.0409, 0.0418, 0.1246, 0.0527, 0.0331, 0.0880], device='cuda:7'), in_proj_covar=tensor([0.0288, 0.0430, 0.0335, 0.0336, 0.0337, 0.0387, 0.0230, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:29:55,583 INFO [train.py:904] (7/8) Epoch 25, batch 200, loss[loss=0.1621, simple_loss=0.2607, pruned_loss=0.03178, over 17050.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2604, pruned_loss=0.04314, over 2115017.26 frames. ], batch size: 53, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:30:40,583 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.064e+02 2.455e+02 3.172e+02 7.590e+02, threshold=4.909e+02, percent-clipped=1.0 2023-05-01 22:31:04,170 INFO [train.py:904] (7/8) Epoch 25, batch 250, loss[loss=0.1636, simple_loss=0.2591, pruned_loss=0.03401, over 17256.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2581, pruned_loss=0.04257, over 2378265.39 frames. ], batch size: 45, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:31:12,996 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3236, 3.4477, 3.3551, 1.9912, 2.8488, 2.1681, 3.6800, 3.7594], device='cuda:7'), covar=tensor([0.0234, 0.0883, 0.0744, 0.2445, 0.1046, 0.1236, 0.0567, 0.0919], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0161, 0.0166, 0.0152, 0.0143, 0.0129, 0.0141, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:31:45,752 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-01 22:32:13,889 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6352, 6.0014, 5.7237, 5.7498, 5.3574, 5.3712, 5.3366, 6.1200], device='cuda:7'), covar=tensor([0.1611, 0.0959, 0.1268, 0.0962, 0.0887, 0.0738, 0.1382, 0.0873], device='cuda:7'), in_proj_covar=tensor([0.0687, 0.0828, 0.0681, 0.0641, 0.0525, 0.0532, 0.0696, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:32:14,699 INFO [train.py:904] (7/8) Epoch 25, batch 300, loss[loss=0.1607, simple_loss=0.2475, pruned_loss=0.03696, over 17222.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2562, pruned_loss=0.0416, over 2585544.27 frames. ], batch size: 45, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:32:37,478 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0505, 5.6257, 5.7484, 5.4767, 5.5681, 6.1510, 5.5708, 5.3531], device='cuda:7'), covar=tensor([0.0944, 0.1854, 0.2733, 0.2105, 0.2640, 0.0943, 0.1716, 0.2368], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0603, 0.0665, 0.0493, 0.0650, 0.0685, 0.0514, 0.0652], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:33:00,866 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.231e+02 2.519e+02 2.986e+02 7.202e+02, threshold=5.038e+02, percent-clipped=2.0 2023-05-01 22:33:03,752 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7387, 3.7833, 2.4806, 4.2880, 3.0024, 4.2454, 2.5545, 3.1661], device='cuda:7'), covar=tensor([0.0317, 0.0406, 0.1537, 0.0364, 0.0848, 0.0533, 0.1521, 0.0776], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0177, 0.0193, 0.0166, 0.0177, 0.0215, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:33:19,471 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=243949.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:33:24,548 INFO [train.py:904] (7/8) Epoch 25, batch 350, loss[loss=0.1551, simple_loss=0.2539, pruned_loss=0.02818, over 16842.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2543, pruned_loss=0.04003, over 2757944.96 frames. ], batch size: 42, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:34:25,842 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=243997.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:34:38,355 INFO [train.py:904] (7/8) Epoch 25, batch 400, loss[loss=0.1481, simple_loss=0.2406, pruned_loss=0.02785, over 17039.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2527, pruned_loss=0.03945, over 2882436.87 frames. ], batch size: 50, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:34:50,870 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9501, 2.5154, 2.4489, 3.8795, 3.0901, 3.8147, 1.6448, 2.7203], device='cuda:7'), covar=tensor([0.1256, 0.0721, 0.1195, 0.0231, 0.0165, 0.0557, 0.1505, 0.0889], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0176, 0.0195, 0.0191, 0.0200, 0.0213, 0.0204, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:34:54,363 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8797, 4.4016, 3.0931, 2.3511, 2.6250, 2.6258, 4.7402, 3.5217], device='cuda:7'), covar=tensor([0.2910, 0.0552, 0.1866, 0.3005, 0.3078, 0.2227, 0.0327, 0.1580], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0268, 0.0305, 0.0315, 0.0294, 0.0265, 0.0295, 0.0339], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:35:02,387 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.51 vs. limit=2.0 2023-05-01 22:35:22,994 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.463e+02 2.193e+02 2.643e+02 3.297e+02 6.329e+02, threshold=5.285e+02, percent-clipped=2.0 2023-05-01 22:35:42,222 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.33 vs. limit=5.0 2023-05-01 22:35:47,035 INFO [train.py:904] (7/8) Epoch 25, batch 450, loss[loss=0.1567, simple_loss=0.2412, pruned_loss=0.03611, over 16734.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2515, pruned_loss=0.03918, over 2984329.33 frames. ], batch size: 89, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:36:50,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0114, 5.0040, 5.4791, 5.4565, 5.4516, 5.1641, 5.0755, 4.8755], device='cuda:7'), covar=tensor([0.0343, 0.0638, 0.0384, 0.0387, 0.0466, 0.0401, 0.0935, 0.0459], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0459, 0.0447, 0.0412, 0.0491, 0.0470, 0.0546, 0.0375], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 22:36:55,216 INFO [train.py:904] (7/8) Epoch 25, batch 500, loss[loss=0.1402, simple_loss=0.2245, pruned_loss=0.02793, over 16274.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2499, pruned_loss=0.0386, over 3055021.77 frames. ], batch size: 36, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:37:18,523 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8706, 4.2150, 4.1973, 3.0603, 3.7063, 4.2507, 3.8833, 2.6595], device='cuda:7'), covar=tensor([0.0536, 0.0086, 0.0061, 0.0385, 0.0122, 0.0105, 0.0096, 0.0481], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0088, 0.0087, 0.0136, 0.0100, 0.0111, 0.0096, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 22:37:37,571 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6659, 4.7750, 4.9271, 4.7224, 4.7417, 5.3963, 4.9054, 4.6100], device='cuda:7'), covar=tensor([0.1463, 0.2132, 0.2534, 0.2340, 0.2672, 0.1039, 0.1869, 0.2602], device='cuda:7'), in_proj_covar=tensor([0.0409, 0.0610, 0.0672, 0.0499, 0.0659, 0.0693, 0.0520, 0.0659], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:37:42,043 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 2.128e+02 2.486e+02 2.973e+02 4.659e+02, threshold=4.972e+02, percent-clipped=0.0 2023-05-01 22:38:05,736 INFO [train.py:904] (7/8) Epoch 25, batch 550, loss[loss=0.1376, simple_loss=0.2283, pruned_loss=0.02349, over 17007.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2491, pruned_loss=0.03783, over 3106140.23 frames. ], batch size: 41, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:39:15,783 INFO [train.py:904] (7/8) Epoch 25, batch 600, loss[loss=0.1351, simple_loss=0.2215, pruned_loss=0.0243, over 15915.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.2486, pruned_loss=0.03783, over 3160529.88 frames. ], batch size: 35, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:39:48,700 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8264, 4.2302, 3.0851, 2.3348, 2.6813, 2.5875, 4.5256, 3.4936], device='cuda:7'), covar=tensor([0.2873, 0.0569, 0.1782, 0.2962, 0.2836, 0.2208, 0.0372, 0.1478], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0268, 0.0306, 0.0317, 0.0296, 0.0266, 0.0297, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 22:40:02,987 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.116e+02 2.478e+02 2.900e+02 1.597e+03, threshold=4.957e+02, percent-clipped=3.0 2023-05-01 22:40:24,984 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-05-01 22:40:25,349 INFO [train.py:904] (7/8) Epoch 25, batch 650, loss[loss=0.1626, simple_loss=0.2514, pruned_loss=0.03689, over 16758.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2477, pruned_loss=0.03841, over 3198570.27 frames. ], batch size: 57, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:41:27,821 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6518, 2.5072, 2.6395, 4.4391, 2.4700, 2.8884, 2.5691, 2.7113], device='cuda:7'), covar=tensor([0.1294, 0.3746, 0.3055, 0.0532, 0.4122, 0.2637, 0.3720, 0.3587], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0459, 0.0379, 0.0333, 0.0444, 0.0526, 0.0433, 0.0538], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:41:33,627 INFO [train.py:904] (7/8) Epoch 25, batch 700, loss[loss=0.1548, simple_loss=0.2471, pruned_loss=0.03127, over 16997.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2472, pruned_loss=0.03755, over 3227701.87 frames. ], batch size: 50, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:42:20,954 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.146e+02 2.447e+02 2.875e+02 6.652e+02, threshold=4.894e+02, percent-clipped=4.0 2023-05-01 22:42:42,416 INFO [train.py:904] (7/8) Epoch 25, batch 750, loss[loss=0.1474, simple_loss=0.2377, pruned_loss=0.02854, over 17239.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2477, pruned_loss=0.0379, over 3256403.12 frames. ], batch size: 45, lr: 2.72e-03, grad_scale: 2.0 2023-05-01 22:42:42,861 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244353.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:42:58,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9830, 2.1376, 2.5500, 2.9264, 2.7622, 3.3659, 2.4245, 3.3353], device='cuda:7'), covar=tensor([0.0260, 0.0518, 0.0373, 0.0376, 0.0386, 0.0210, 0.0485, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0194, 0.0181, 0.0185, 0.0200, 0.0159, 0.0196, 0.0157], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:43:19,336 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-05-01 22:43:35,000 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8592, 2.8514, 2.7424, 4.9657, 3.9100, 4.3284, 1.8092, 3.1056], device='cuda:7'), covar=tensor([0.1402, 0.0842, 0.1259, 0.0258, 0.0300, 0.0487, 0.1640, 0.0863], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0176, 0.0194, 0.0192, 0.0200, 0.0213, 0.0204, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:43:52,155 INFO [train.py:904] (7/8) Epoch 25, batch 800, loss[loss=0.135, simple_loss=0.2229, pruned_loss=0.02356, over 15879.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2475, pruned_loss=0.03755, over 3280686.94 frames. ], batch size: 35, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:44:00,593 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0181, 2.2213, 2.3967, 2.7471, 2.0434, 3.2305, 1.8615, 2.7453], device='cuda:7'), covar=tensor([0.1171, 0.0803, 0.1071, 0.0226, 0.0136, 0.0409, 0.1447, 0.0741], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0176, 0.0194, 0.0192, 0.0199, 0.0213, 0.0203, 0.0193], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:44:08,223 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=244414.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:44:39,332 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.216e+02 2.518e+02 3.114e+02 7.971e+02, threshold=5.036e+02, percent-clipped=2.0 2023-05-01 22:44:53,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3107, 4.4087, 4.3057, 3.3732, 3.8278, 4.4338, 4.0320, 2.3709], device='cuda:7'), covar=tensor([0.0488, 0.0084, 0.0077, 0.0411, 0.0171, 0.0134, 0.0129, 0.0664], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0088, 0.0088, 0.0136, 0.0101, 0.0111, 0.0097, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 22:45:03,362 INFO [train.py:904] (7/8) Epoch 25, batch 850, loss[loss=0.1508, simple_loss=0.2278, pruned_loss=0.03692, over 16442.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.247, pruned_loss=0.03715, over 3285459.18 frames. ], batch size: 146, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:45:18,287 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-01 22:45:40,112 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8175, 4.1350, 4.1795, 3.0132, 3.6225, 4.2045, 3.8091, 2.7217], device='cuda:7'), covar=tensor([0.0495, 0.0105, 0.0062, 0.0391, 0.0129, 0.0117, 0.0108, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0088, 0.0088, 0.0136, 0.0101, 0.0111, 0.0097, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 22:46:12,670 INFO [train.py:904] (7/8) Epoch 25, batch 900, loss[loss=0.1719, simple_loss=0.2421, pruned_loss=0.05087, over 16882.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2466, pruned_loss=0.03661, over 3293902.22 frames. ], batch size: 90, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:47:00,725 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.397e+02 1.999e+02 2.403e+02 2.874e+02 4.869e+02, threshold=4.806e+02, percent-clipped=0.0 2023-05-01 22:47:23,402 INFO [train.py:904] (7/8) Epoch 25, batch 950, loss[loss=0.1812, simple_loss=0.2532, pruned_loss=0.05457, over 16847.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.247, pruned_loss=0.03717, over 3293082.42 frames. ], batch size: 116, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:48:33,908 INFO [train.py:904] (7/8) Epoch 25, batch 1000, loss[loss=0.1686, simple_loss=0.2649, pruned_loss=0.03613, over 17117.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.2459, pruned_loss=0.03735, over 3287756.49 frames. ], batch size: 49, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:49:12,279 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 22:49:21,004 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.078e+02 2.468e+02 2.868e+02 8.684e+02, threshold=4.936e+02, percent-clipped=5.0 2023-05-01 22:49:42,541 INFO [train.py:904] (7/8) Epoch 25, batch 1050, loss[loss=0.1424, simple_loss=0.2274, pruned_loss=0.02873, over 16807.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2464, pruned_loss=0.03795, over 3296494.26 frames. ], batch size: 42, lr: 2.72e-03, grad_scale: 4.0 2023-05-01 22:50:25,726 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2173, 3.9920, 4.1844, 4.3785, 4.4539, 4.0991, 4.2956, 4.4688], device='cuda:7'), covar=tensor([0.1868, 0.1475, 0.1791, 0.0913, 0.0816, 0.1346, 0.3340, 0.0911], device='cuda:7'), in_proj_covar=tensor([0.0668, 0.0822, 0.0946, 0.0833, 0.0635, 0.0652, 0.0685, 0.0796], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 22:50:53,107 INFO [train.py:904] (7/8) Epoch 25, batch 1100, loss[loss=0.1521, simple_loss=0.2438, pruned_loss=0.03018, over 17208.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2464, pruned_loss=0.03828, over 3294408.56 frames. ], batch size: 46, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 22:51:01,268 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=244709.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:51:13,215 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244717.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 22:51:40,296 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.068e+02 2.379e+02 2.713e+02 6.650e+02, threshold=4.759e+02, percent-clipped=1.0 2023-05-01 22:52:02,016 INFO [train.py:904] (7/8) Epoch 25, batch 1150, loss[loss=0.173, simple_loss=0.2668, pruned_loss=0.03957, over 16725.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.246, pruned_loss=0.03731, over 3294152.62 frames. ], batch size: 57, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 22:52:37,191 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=244778.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 22:52:58,081 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-01 22:53:11,755 INFO [train.py:904] (7/8) Epoch 25, batch 1200, loss[loss=0.1693, simple_loss=0.2478, pruned_loss=0.0454, over 16584.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.2447, pruned_loss=0.03706, over 3300703.61 frames. ], batch size: 68, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:53:57,424 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.058e+02 2.429e+02 2.907e+02 5.779e+02, threshold=4.859e+02, percent-clipped=1.0 2023-05-01 22:54:19,241 INFO [train.py:904] (7/8) Epoch 25, batch 1250, loss[loss=0.1684, simple_loss=0.2636, pruned_loss=0.03664, over 17284.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2448, pruned_loss=0.03749, over 3303207.99 frames. ], batch size: 52, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:55:06,341 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244887.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:55:26,312 INFO [train.py:904] (7/8) Epoch 25, batch 1300, loss[loss=0.1808, simple_loss=0.2698, pruned_loss=0.04584, over 17082.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2452, pruned_loss=0.03735, over 3306280.15 frames. ], batch size: 53, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:55:51,386 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-01 22:56:12,138 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.186e+02 2.539e+02 3.062e+02 9.189e+02, threshold=5.078e+02, percent-clipped=4.0 2023-05-01 22:56:28,309 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=244948.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:56:35,080 INFO [train.py:904] (7/8) Epoch 25, batch 1350, loss[loss=0.1707, simple_loss=0.2716, pruned_loss=0.03489, over 17116.00 frames. ], tot_loss[loss=0.16, simple_loss=0.2456, pruned_loss=0.03716, over 3311506.10 frames. ], batch size: 49, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:56:36,716 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=244954.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:57:43,146 INFO [train.py:904] (7/8) Epoch 25, batch 1400, loss[loss=0.1497, simple_loss=0.2297, pruned_loss=0.03479, over 16313.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2459, pruned_loss=0.03725, over 3309453.88 frames. ], batch size: 36, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:57:52,483 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245009.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:57:59,981 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245015.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:58:28,651 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.075e+02 2.428e+02 3.013e+02 6.709e+02, threshold=4.856e+02, percent-clipped=2.0 2023-05-01 22:58:51,291 INFO [train.py:904] (7/8) Epoch 25, batch 1450, loss[loss=0.1711, simple_loss=0.2449, pruned_loss=0.04869, over 12130.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2453, pruned_loss=0.03703, over 3309838.21 frames. ], batch size: 246, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 22:58:56,918 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245057.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 22:59:01,379 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7000, 3.3263, 3.8021, 2.1180, 3.8100, 3.8774, 3.2163, 2.9048], device='cuda:7'), covar=tensor([0.0757, 0.0297, 0.0183, 0.1099, 0.0134, 0.0203, 0.0394, 0.0440], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0139, 0.0084, 0.0130, 0.0129, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 22:59:01,734 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-01 22:59:18,780 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245073.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 22:59:20,011 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5251, 5.9090, 5.6816, 5.7087, 5.3079, 5.3697, 5.3202, 6.0184], device='cuda:7'), covar=tensor([0.1505, 0.1075, 0.1253, 0.0983, 0.1065, 0.0688, 0.1286, 0.0984], device='cuda:7'), in_proj_covar=tensor([0.0708, 0.0859, 0.0702, 0.0663, 0.0544, 0.0548, 0.0721, 0.0671], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:00:00,082 INFO [train.py:904] (7/8) Epoch 25, batch 1500, loss[loss=0.1547, simple_loss=0.2479, pruned_loss=0.03076, over 17175.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2451, pruned_loss=0.03669, over 3316418.51 frames. ], batch size: 46, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:00:46,573 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.188e+02 2.473e+02 2.927e+02 4.606e+02, threshold=4.945e+02, percent-clipped=0.0 2023-05-01 23:01:08,936 INFO [train.py:904] (7/8) Epoch 25, batch 1550, loss[loss=0.1538, simple_loss=0.2488, pruned_loss=0.02942, over 17024.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2468, pruned_loss=0.0376, over 3320180.81 frames. ], batch size: 50, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:01:12,342 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1518, 4.5387, 4.4828, 3.2920, 3.7752, 4.4947, 4.1152, 2.7554], device='cuda:7'), covar=tensor([0.0460, 0.0061, 0.0048, 0.0380, 0.0132, 0.0092, 0.0081, 0.0457], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0088, 0.0135, 0.0100, 0.0112, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 23:01:34,959 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 23:02:19,892 INFO [train.py:904] (7/8) Epoch 25, batch 1600, loss[loss=0.1771, simple_loss=0.2493, pruned_loss=0.05244, over 16909.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2488, pruned_loss=0.03855, over 3319386.06 frames. ], batch size: 109, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:03:06,949 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.282e+02 2.634e+02 3.263e+02 7.681e+02, threshold=5.268e+02, percent-clipped=4.0 2023-05-01 23:03:09,153 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1240, 4.8506, 5.1630, 5.3603, 5.5555, 4.8601, 5.5055, 5.5280], device='cuda:7'), covar=tensor([0.1889, 0.1412, 0.1708, 0.0762, 0.0549, 0.0876, 0.0594, 0.0653], device='cuda:7'), in_proj_covar=tensor([0.0678, 0.0836, 0.0963, 0.0846, 0.0644, 0.0662, 0.0694, 0.0812], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:03:16,801 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245243.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:03:28,986 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7348, 4.3796, 4.3732, 4.8823, 5.0732, 4.5744, 5.0360, 5.0408], device='cuda:7'), covar=tensor([0.1959, 0.1818, 0.2992, 0.1260, 0.0875, 0.1546, 0.1130, 0.1145], device='cuda:7'), in_proj_covar=tensor([0.0679, 0.0837, 0.0964, 0.0846, 0.0645, 0.0663, 0.0694, 0.0812], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:03:29,754 INFO [train.py:904] (7/8) Epoch 25, batch 1650, loss[loss=0.1742, simple_loss=0.2709, pruned_loss=0.03875, over 16801.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2503, pruned_loss=0.03856, over 3329869.82 frames. ], batch size: 57, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:04:28,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8659, 2.5172, 2.0306, 2.2487, 2.9063, 2.6629, 2.8855, 3.0259], device='cuda:7'), covar=tensor([0.0231, 0.0448, 0.0600, 0.0516, 0.0258, 0.0359, 0.0263, 0.0269], device='cuda:7'), in_proj_covar=tensor([0.0228, 0.0246, 0.0235, 0.0236, 0.0248, 0.0246, 0.0247, 0.0243], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:04:41,938 INFO [train.py:904] (7/8) Epoch 25, batch 1700, loss[loss=0.1797, simple_loss=0.2682, pruned_loss=0.04558, over 16681.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2527, pruned_loss=0.03962, over 3325267.18 frames. ], batch size: 62, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:04:51,859 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245310.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:04:53,451 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-05-01 23:04:59,324 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4064, 3.4897, 3.9142, 2.2284, 3.1576, 2.6299, 3.8083, 3.7468], device='cuda:7'), covar=tensor([0.0256, 0.0983, 0.0519, 0.2060, 0.0863, 0.0942, 0.0594, 0.1130], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0169, 0.0155, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:05:08,580 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9219, 2.1864, 2.5442, 2.9232, 2.7735, 3.4257, 2.4574, 3.4256], device='cuda:7'), covar=tensor([0.0314, 0.0572, 0.0378, 0.0361, 0.0402, 0.0217, 0.0529, 0.0203], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0187, 0.0203, 0.0161, 0.0199, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:05:26,298 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3937, 5.7328, 5.4489, 5.5584, 5.1782, 5.1873, 5.1375, 5.8624], device='cuda:7'), covar=tensor([0.1440, 0.1031, 0.1277, 0.0905, 0.0945, 0.0756, 0.1262, 0.1026], device='cuda:7'), in_proj_covar=tensor([0.0710, 0.0861, 0.0704, 0.0665, 0.0546, 0.0548, 0.0723, 0.0672], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:05:30,605 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.070e+02 2.579e+02 3.059e+02 5.844e+02, threshold=5.158e+02, percent-clipped=3.0 2023-05-01 23:05:52,660 INFO [train.py:904] (7/8) Epoch 25, batch 1750, loss[loss=0.1656, simple_loss=0.2676, pruned_loss=0.03181, over 17109.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.254, pruned_loss=0.03966, over 3319383.50 frames. ], batch size: 49, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:06:20,991 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245373.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:07:03,957 INFO [train.py:904] (7/8) Epoch 25, batch 1800, loss[loss=0.166, simple_loss=0.248, pruned_loss=0.04202, over 16717.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2556, pruned_loss=0.04012, over 3311861.22 frames. ], batch size: 89, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:07:29,413 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245421.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:07:44,371 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9123, 4.0794, 4.3554, 4.3206, 4.3806, 4.1047, 4.0183, 4.0579], device='cuda:7'), covar=tensor([0.0682, 0.0817, 0.0587, 0.0654, 0.0728, 0.0674, 0.1217, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0433, 0.0481, 0.0468, 0.0431, 0.0515, 0.0495, 0.0571, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 23:07:51,079 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.188e+02 2.630e+02 2.978e+02 6.380e+02, threshold=5.260e+02, percent-clipped=2.0 2023-05-01 23:08:12,923 INFO [train.py:904] (7/8) Epoch 25, batch 1850, loss[loss=0.1328, simple_loss=0.2156, pruned_loss=0.02493, over 16961.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2558, pruned_loss=0.04047, over 3318810.74 frames. ], batch size: 41, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:08:24,179 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245460.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:08:29,459 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7222, 3.3909, 3.9258, 2.1069, 3.9482, 3.9761, 3.2399, 2.9154], device='cuda:7'), covar=tensor([0.0748, 0.0280, 0.0163, 0.1134, 0.0106, 0.0218, 0.0395, 0.0489], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0139, 0.0084, 0.0130, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:08:36,935 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-01 23:09:23,472 INFO [train.py:904] (7/8) Epoch 25, batch 1900, loss[loss=0.1682, simple_loss=0.2553, pruned_loss=0.04055, over 16549.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2546, pruned_loss=0.03983, over 3324567.50 frames. ], batch size: 68, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:09:49,760 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245521.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:10:11,813 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.101e+02 2.496e+02 2.952e+02 1.304e+03, threshold=4.992e+02, percent-clipped=2.0 2023-05-01 23:10:20,947 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245543.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:10:22,675 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3449, 5.6938, 5.1690, 5.5786, 5.1719, 5.0115, 5.1337, 5.7186], device='cuda:7'), covar=tensor([0.2460, 0.1493, 0.2534, 0.1588, 0.1638, 0.1305, 0.2534, 0.1864], device='cuda:7'), in_proj_covar=tensor([0.0717, 0.0869, 0.0711, 0.0673, 0.0552, 0.0553, 0.0731, 0.0679], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:10:25,706 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245546.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:10:33,953 INFO [train.py:904] (7/8) Epoch 25, batch 1950, loss[loss=0.1512, simple_loss=0.2468, pruned_loss=0.02781, over 17095.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2543, pruned_loss=0.03949, over 3325946.58 frames. ], batch size: 49, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:11:26,899 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245591.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:11:41,397 INFO [train.py:904] (7/8) Epoch 25, batch 2000, loss[loss=0.1742, simple_loss=0.2475, pruned_loss=0.05043, over 16726.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2533, pruned_loss=0.03954, over 3325704.13 frames. ], batch size: 134, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:11:49,109 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245607.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:11:53,140 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=245610.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:12:08,138 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-01 23:12:31,551 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.177e+02 2.547e+02 3.013e+02 5.081e+02, threshold=5.093e+02, percent-clipped=1.0 2023-05-01 23:12:50,395 INFO [train.py:904] (7/8) Epoch 25, batch 2050, loss[loss=0.1704, simple_loss=0.2555, pruned_loss=0.04262, over 16852.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2537, pruned_loss=0.04006, over 3321445.54 frames. ], batch size: 83, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:12:57,318 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=245658.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:13:12,539 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3240, 1.5078, 2.1138, 2.2155, 2.2954, 2.4304, 1.6045, 2.4023], device='cuda:7'), covar=tensor([0.0243, 0.0614, 0.0299, 0.0311, 0.0303, 0.0263, 0.0710, 0.0180], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0197, 0.0184, 0.0188, 0.0204, 0.0162, 0.0200, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:13:34,852 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245686.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:13:58,100 INFO [train.py:904] (7/8) Epoch 25, batch 2100, loss[loss=0.164, simple_loss=0.2496, pruned_loss=0.03921, over 16904.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2542, pruned_loss=0.04005, over 3314104.92 frames. ], batch size: 90, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:14:38,470 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2786, 3.3141, 3.8103, 2.0341, 3.1527, 2.4526, 3.6952, 3.5919], device='cuda:7'), covar=tensor([0.0238, 0.0980, 0.0518, 0.2160, 0.0821, 0.0959, 0.0594, 0.1071], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0168, 0.0171, 0.0156, 0.0147, 0.0132, 0.0146, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:14:49,017 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.201e+02 2.533e+02 2.997e+02 6.005e+02, threshold=5.066e+02, percent-clipped=2.0 2023-05-01 23:14:59,054 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245747.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:15:08,426 INFO [train.py:904] (7/8) Epoch 25, batch 2150, loss[loss=0.1826, simple_loss=0.2636, pruned_loss=0.05082, over 16875.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2556, pruned_loss=0.04082, over 3314584.95 frames. ], batch size: 109, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:15:38,562 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-01 23:15:55,291 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-01 23:16:00,620 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245792.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:16:07,364 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0425, 4.9967, 4.9060, 4.4867, 4.5796, 4.9642, 4.7971, 4.5969], device='cuda:7'), covar=tensor([0.0614, 0.0760, 0.0297, 0.0343, 0.0965, 0.0534, 0.0434, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0314, 0.0468, 0.0363, 0.0367, 0.0368, 0.0422, 0.0249, 0.0440], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:16:15,996 INFO [train.py:904] (7/8) Epoch 25, batch 2200, loss[loss=0.1812, simple_loss=0.2532, pruned_loss=0.05457, over 16664.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2558, pruned_loss=0.04093, over 3326451.48 frames. ], batch size: 134, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:16:24,644 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1207, 2.2285, 2.2776, 3.8150, 2.1675, 2.5579, 2.2645, 2.3916], device='cuda:7'), covar=tensor([0.1538, 0.3706, 0.3086, 0.0672, 0.3964, 0.2517, 0.4046, 0.3052], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0466, 0.0384, 0.0338, 0.0445, 0.0533, 0.0438, 0.0545], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:16:34,852 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245816.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:17:06,937 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.642e+02 2.208e+02 2.555e+02 2.978e+02 8.197e+02, threshold=5.110e+02, percent-clipped=1.0 2023-05-01 23:17:23,665 INFO [train.py:904] (7/8) Epoch 25, batch 2250, loss[loss=0.1553, simple_loss=0.2383, pruned_loss=0.03613, over 16768.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2557, pruned_loss=0.0405, over 3327444.82 frames. ], batch size: 83, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:17:24,766 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245853.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:17:42,820 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2648, 2.3401, 2.5042, 3.9900, 2.2264, 2.6427, 2.4688, 2.4864], device='cuda:7'), covar=tensor([0.1600, 0.3940, 0.2966, 0.0721, 0.4399, 0.2696, 0.3898, 0.3676], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0466, 0.0384, 0.0338, 0.0445, 0.0534, 0.0439, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:18:16,674 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=245891.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:18:32,174 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=245902.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 23:18:32,942 INFO [train.py:904] (7/8) Epoch 25, batch 2300, loss[loss=0.174, simple_loss=0.2636, pruned_loss=0.04216, over 16370.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2559, pruned_loss=0.04064, over 3324518.88 frames. ], batch size: 68, lr: 2.71e-03, grad_scale: 2.0 2023-05-01 23:18:45,424 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5851, 2.4339, 2.4431, 4.3260, 2.4274, 2.8387, 2.5227, 2.6244], device='cuda:7'), covar=tensor([0.1338, 0.3832, 0.3329, 0.0612, 0.4293, 0.2753, 0.3710, 0.3878], device='cuda:7'), in_proj_covar=tensor([0.0418, 0.0466, 0.0384, 0.0338, 0.0445, 0.0534, 0.0438, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:18:47,806 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3632, 2.3737, 2.3701, 4.0239, 2.3093, 2.7137, 2.4543, 2.5148], device='cuda:7'), covar=tensor([0.1420, 0.3722, 0.3267, 0.0651, 0.4274, 0.2722, 0.3612, 0.3794], device='cuda:7'), in_proj_covar=tensor([0.0418, 0.0466, 0.0384, 0.0338, 0.0445, 0.0534, 0.0439, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:19:14,293 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8004, 4.8275, 5.2313, 5.2315, 5.2649, 4.9233, 4.8819, 4.7660], device='cuda:7'), covar=tensor([0.0350, 0.0605, 0.0456, 0.0417, 0.0574, 0.0434, 0.1083, 0.0468], device='cuda:7'), in_proj_covar=tensor([0.0433, 0.0485, 0.0474, 0.0434, 0.0519, 0.0498, 0.0577, 0.0393], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 23:19:24,139 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.182e+02 2.492e+02 2.872e+02 5.312e+02, threshold=4.984e+02, percent-clipped=2.0 2023-05-01 23:19:42,197 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=245952.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:19:42,924 INFO [train.py:904] (7/8) Epoch 25, batch 2350, loss[loss=0.1388, simple_loss=0.2236, pruned_loss=0.02697, over 17035.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2556, pruned_loss=0.04058, over 3328861.86 frames. ], batch size: 41, lr: 2.71e-03, grad_scale: 2.0 2023-05-01 23:20:38,450 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4653, 4.5507, 4.7120, 4.4184, 4.4997, 5.1266, 4.5848, 4.2390], device='cuda:7'), covar=tensor([0.1795, 0.2314, 0.2547, 0.2286, 0.2893, 0.1213, 0.1939, 0.2640], device='cuda:7'), in_proj_covar=tensor([0.0429, 0.0638, 0.0697, 0.0517, 0.0691, 0.0720, 0.0540, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:20:54,773 INFO [train.py:904] (7/8) Epoch 25, batch 2400, loss[loss=0.1546, simple_loss=0.25, pruned_loss=0.02956, over 17147.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2572, pruned_loss=0.04071, over 3319419.07 frames. ], batch size: 47, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:21:08,573 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7922, 3.8760, 2.6205, 4.6034, 3.0752, 4.4755, 2.7863, 3.2459], device='cuda:7'), covar=tensor([0.0295, 0.0384, 0.1435, 0.0281, 0.0847, 0.0510, 0.1335, 0.0712], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0181, 0.0195, 0.0172, 0.0180, 0.0222, 0.0204, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:21:46,456 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.068e+02 2.493e+02 3.230e+02 1.350e+03, threshold=4.987e+02, percent-clipped=5.0 2023-05-01 23:21:49,244 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246042.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:22:04,213 INFO [train.py:904] (7/8) Epoch 25, batch 2450, loss[loss=0.166, simple_loss=0.2449, pruned_loss=0.04358, over 16880.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2571, pruned_loss=0.03989, over 3328607.43 frames. ], batch size: 90, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:23:07,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2733, 3.4999, 3.7853, 2.1868, 3.1587, 2.4600, 3.6716, 3.6902], device='cuda:7'), covar=tensor([0.0339, 0.0929, 0.0554, 0.2070, 0.0828, 0.1015, 0.0648, 0.1040], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0169, 0.0171, 0.0157, 0.0148, 0.0132, 0.0147, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-01 23:23:10,963 INFO [train.py:904] (7/8) Epoch 25, batch 2500, loss[loss=0.1301, simple_loss=0.2237, pruned_loss=0.01826, over 16968.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2567, pruned_loss=0.0399, over 3321210.87 frames. ], batch size: 41, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:23:12,645 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5124, 4.3971, 4.2722, 2.8201, 3.6215, 4.2787, 3.8072, 2.3160], device='cuda:7'), covar=tensor([0.0665, 0.0071, 0.0073, 0.0512, 0.0158, 0.0122, 0.0123, 0.0626], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0088, 0.0089, 0.0136, 0.0101, 0.0112, 0.0098, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 23:23:12,977 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-01 23:23:21,534 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246111.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:23:28,059 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246116.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:24:01,584 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.119e+02 2.579e+02 3.197e+02 6.008e+02, threshold=5.158e+02, percent-clipped=3.0 2023-05-01 23:24:12,435 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246148.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:24:18,472 INFO [train.py:904] (7/8) Epoch 25, batch 2550, loss[loss=0.1643, simple_loss=0.2635, pruned_loss=0.03259, over 17111.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2572, pruned_loss=0.04021, over 3316618.09 frames. ], batch size: 48, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:24:35,215 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246164.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:24:45,435 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246172.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:25:17,899 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246196.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:25:26,289 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246202.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:25:27,017 INFO [train.py:904] (7/8) Epoch 25, batch 2600, loss[loss=0.174, simple_loss=0.2707, pruned_loss=0.0386, over 17137.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2574, pruned_loss=0.04004, over 3316664.15 frames. ], batch size: 48, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:25:41,672 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-01 23:25:45,792 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6879, 4.6464, 4.5865, 4.2364, 4.3193, 4.6462, 4.4349, 4.3811], device='cuda:7'), covar=tensor([0.0664, 0.0932, 0.0302, 0.0347, 0.0806, 0.0570, 0.0488, 0.0726], device='cuda:7'), in_proj_covar=tensor([0.0315, 0.0471, 0.0366, 0.0370, 0.0370, 0.0424, 0.0250, 0.0443], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:25:47,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9150, 2.8113, 2.6958, 4.4371, 3.4954, 4.1150, 1.7379, 3.0736], device='cuda:7'), covar=tensor([0.1355, 0.0781, 0.1205, 0.0196, 0.0253, 0.0440, 0.1610, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0178, 0.0198, 0.0197, 0.0205, 0.0218, 0.0205, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:25:52,191 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246221.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:26:11,040 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5315, 4.5609, 4.8901, 4.8722, 4.9001, 4.6251, 4.6239, 4.4748], device='cuda:7'), covar=tensor([0.0405, 0.0621, 0.0400, 0.0401, 0.0567, 0.0420, 0.0884, 0.0592], device='cuda:7'), in_proj_covar=tensor([0.0431, 0.0482, 0.0471, 0.0432, 0.0516, 0.0495, 0.0574, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 23:26:18,785 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.097e+02 2.510e+02 3.006e+02 6.140e+02, threshold=5.021e+02, percent-clipped=2.0 2023-05-01 23:26:22,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0401, 5.5883, 5.6721, 5.4110, 5.4115, 6.0759, 5.4963, 5.2264], device='cuda:7'), covar=tensor([0.1043, 0.2115, 0.2331, 0.1886, 0.2908, 0.0974, 0.1539, 0.2366], device='cuda:7'), in_proj_covar=tensor([0.0429, 0.0638, 0.0699, 0.0517, 0.0690, 0.0721, 0.0541, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:26:29,202 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246247.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:26:33,158 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246250.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:26:37,203 INFO [train.py:904] (7/8) Epoch 25, batch 2650, loss[loss=0.174, simple_loss=0.2609, pruned_loss=0.04356, over 16904.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2568, pruned_loss=0.03941, over 3326124.73 frames. ], batch size: 102, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:26:42,364 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246257.0, num_to_drop=1, layers_to_drop={3} 2023-05-01 23:26:43,609 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8460, 4.3866, 3.0709, 2.3870, 2.7783, 2.6994, 4.7787, 3.6472], device='cuda:7'), covar=tensor([0.2940, 0.0555, 0.1865, 0.2946, 0.2831, 0.2049, 0.0360, 0.1336], device='cuda:7'), in_proj_covar=tensor([0.0330, 0.0273, 0.0311, 0.0320, 0.0302, 0.0271, 0.0301, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:27:17,716 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246282.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:27:44,603 INFO [train.py:904] (7/8) Epoch 25, batch 2700, loss[loss=0.2033, simple_loss=0.2798, pruned_loss=0.06341, over 16874.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2573, pruned_loss=0.03937, over 3326373.23 frames. ], batch size: 116, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:28:34,094 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 1.968e+02 2.327e+02 2.625e+02 5.550e+02, threshold=4.654e+02, percent-clipped=1.0 2023-05-01 23:28:36,931 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246342.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:28:52,090 INFO [train.py:904] (7/8) Epoch 25, batch 2750, loss[loss=0.1608, simple_loss=0.2481, pruned_loss=0.03675, over 16817.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2581, pruned_loss=0.03946, over 3320938.03 frames. ], batch size: 39, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:29:44,292 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246390.0, num_to_drop=1, layers_to_drop={1} 2023-05-01 23:30:01,773 INFO [train.py:904] (7/8) Epoch 25, batch 2800, loss[loss=0.1887, simple_loss=0.2898, pruned_loss=0.04381, over 16595.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2574, pruned_loss=0.03955, over 3316622.66 frames. ], batch size: 57, lr: 2.71e-03, grad_scale: 8.0 2023-05-01 23:30:16,912 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7300, 4.1914, 2.9455, 2.3138, 2.6558, 2.6391, 4.5016, 3.5226], device='cuda:7'), covar=tensor([0.3124, 0.0608, 0.1922, 0.3078, 0.2866, 0.2161, 0.0432, 0.1421], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0275, 0.0312, 0.0323, 0.0304, 0.0273, 0.0303, 0.0350], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:30:54,343 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.989e+02 2.330e+02 2.808e+02 5.234e+02, threshold=4.661e+02, percent-clipped=4.0 2023-05-01 23:31:04,249 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246448.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:31:10,744 INFO [train.py:904] (7/8) Epoch 25, batch 2850, loss[loss=0.189, simple_loss=0.2923, pruned_loss=0.04284, over 16764.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2567, pruned_loss=0.0393, over 3321279.42 frames. ], batch size: 57, lr: 2.71e-03, grad_scale: 4.0 2023-05-01 23:31:26,680 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2196, 3.9017, 4.3783, 2.2532, 4.5925, 4.6990, 3.3474, 3.6364], device='cuda:7'), covar=tensor([0.0693, 0.0271, 0.0250, 0.1183, 0.0084, 0.0177, 0.0490, 0.0395], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0101, 0.0141, 0.0084, 0.0132, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:31:31,943 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246467.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:31:47,090 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7351, 3.8059, 2.4986, 4.1710, 3.0533, 4.0978, 2.5366, 3.0918], device='cuda:7'), covar=tensor([0.0314, 0.0440, 0.1429, 0.0412, 0.0708, 0.0752, 0.1439, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0182, 0.0197, 0.0175, 0.0180, 0.0225, 0.0206, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-01 23:31:50,603 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8265, 4.2022, 4.2620, 3.1138, 3.5471, 4.1959, 3.8211, 2.3935], device='cuda:7'), covar=tensor([0.0497, 0.0096, 0.0054, 0.0368, 0.0161, 0.0120, 0.0105, 0.0532], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0088, 0.0135, 0.0101, 0.0112, 0.0098, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-01 23:32:10,939 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246496.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:32:20,912 INFO [train.py:904] (7/8) Epoch 25, batch 2900, loss[loss=0.1906, simple_loss=0.2667, pruned_loss=0.05718, over 16188.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2561, pruned_loss=0.04012, over 3316120.65 frames. ], batch size: 165, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:33:14,958 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.258e+02 2.678e+02 3.242e+02 4.768e+02, threshold=5.357e+02, percent-clipped=1.0 2023-05-01 23:33:25,339 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246547.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:33:32,682 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246552.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:33:33,514 INFO [train.py:904] (7/8) Epoch 25, batch 2950, loss[loss=0.1783, simple_loss=0.2543, pruned_loss=0.05115, over 16252.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2553, pruned_loss=0.04029, over 3318976.79 frames. ], batch size: 165, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:34:06,267 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=246577.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:34:06,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0604, 2.5960, 2.1297, 2.4241, 2.9700, 2.7307, 2.9948, 3.0678], device='cuda:7'), covar=tensor([0.0242, 0.0427, 0.0575, 0.0485, 0.0256, 0.0360, 0.0251, 0.0303], device='cuda:7'), in_proj_covar=tensor([0.0228, 0.0246, 0.0233, 0.0236, 0.0248, 0.0245, 0.0249, 0.0244], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:34:32,323 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246595.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:34:42,718 INFO [train.py:904] (7/8) Epoch 25, batch 3000, loss[loss=0.1415, simple_loss=0.23, pruned_loss=0.02646, over 16838.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2552, pruned_loss=0.0407, over 3315152.07 frames. ], batch size: 42, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:34:42,718 INFO [train.py:929] (7/8) Computing validation loss 2023-05-01 23:34:52,581 INFO [train.py:938] (7/8) Epoch 25, validation: loss=0.1341, simple_loss=0.239, pruned_loss=0.01457, over 944034.00 frames. 2023-05-01 23:34:52,582 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-01 23:35:46,636 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.065e+02 2.466e+02 2.875e+02 5.514e+02, threshold=4.932e+02, percent-clipped=1.0 2023-05-01 23:36:05,628 INFO [train.py:904] (7/8) Epoch 25, batch 3050, loss[loss=0.194, simple_loss=0.275, pruned_loss=0.05652, over 12369.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.255, pruned_loss=0.04028, over 3314967.38 frames. ], batch size: 246, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:36:38,420 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246677.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:37:13,560 INFO [train.py:904] (7/8) Epoch 25, batch 3100, loss[loss=0.1622, simple_loss=0.2383, pruned_loss=0.04301, over 16954.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2542, pruned_loss=0.04035, over 3314593.40 frames. ], batch size: 116, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:37:47,024 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1788, 5.6728, 5.8399, 5.5512, 5.6209, 6.1877, 5.6718, 5.3510], device='cuda:7'), covar=tensor([0.0922, 0.2012, 0.2274, 0.1918, 0.2513, 0.0944, 0.1630, 0.2277], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0638, 0.0698, 0.0518, 0.0694, 0.0723, 0.0541, 0.0691], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:38:05,085 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246738.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:38:08,781 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.159e+02 2.563e+02 3.074e+02 4.728e+02, threshold=5.126e+02, percent-clipped=0.0 2023-05-01 23:38:25,023 INFO [train.py:904] (7/8) Epoch 25, batch 3150, loss[loss=0.2151, simple_loss=0.2863, pruned_loss=0.07188, over 12495.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2536, pruned_loss=0.04028, over 3312080.97 frames. ], batch size: 247, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:38:44,373 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246767.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:39:34,229 INFO [train.py:904] (7/8) Epoch 25, batch 3200, loss[loss=0.1803, simple_loss=0.2548, pruned_loss=0.05286, over 16762.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2532, pruned_loss=0.0395, over 3323817.54 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:39:51,751 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246815.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:40:27,531 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.068e+02 2.379e+02 2.953e+02 6.702e+02, threshold=4.759e+02, percent-clipped=1.0 2023-05-01 23:40:42,120 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246852.0, num_to_drop=1, layers_to_drop={2} 2023-05-01 23:40:42,964 INFO [train.py:904] (7/8) Epoch 25, batch 3250, loss[loss=0.186, simple_loss=0.2825, pruned_loss=0.04473, over 16736.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2535, pruned_loss=0.03964, over 3322909.24 frames. ], batch size: 57, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:41:16,922 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=246877.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:41:49,851 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246900.0, num_to_drop=1, layers_to_drop={0} 2023-05-01 23:41:53,760 INFO [train.py:904] (7/8) Epoch 25, batch 3300, loss[loss=0.1349, simple_loss=0.2279, pruned_loss=0.02101, over 17210.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2538, pruned_loss=0.0396, over 3324386.07 frames. ], batch size: 44, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:42:24,635 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=246925.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:42:36,872 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=246934.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:42:46,643 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.501e+02 2.089e+02 2.401e+02 2.822e+02 3.775e+02, threshold=4.802e+02, percent-clipped=0.0 2023-05-01 23:42:57,566 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1094, 3.2401, 3.3730, 2.2116, 2.9838, 2.3515, 3.6675, 3.6637], device='cuda:7'), covar=tensor([0.0218, 0.0873, 0.0618, 0.1856, 0.0847, 0.1055, 0.0455, 0.0805], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0169, 0.0171, 0.0156, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-01 23:43:02,688 INFO [train.py:904] (7/8) Epoch 25, batch 3350, loss[loss=0.1562, simple_loss=0.2556, pruned_loss=0.02839, over 17138.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2539, pruned_loss=0.03962, over 3318760.09 frames. ], batch size: 49, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:43:50,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1453, 4.1891, 4.5075, 4.4954, 4.5346, 4.2359, 4.2758, 4.2184], device='cuda:7'), covar=tensor([0.0433, 0.0721, 0.0446, 0.0423, 0.0529, 0.0453, 0.0803, 0.0636], device='cuda:7'), in_proj_covar=tensor([0.0436, 0.0491, 0.0476, 0.0437, 0.0520, 0.0501, 0.0580, 0.0397], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 23:44:01,894 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=246995.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:44:13,556 INFO [train.py:904] (7/8) Epoch 25, batch 3400, loss[loss=0.1889, simple_loss=0.2772, pruned_loss=0.05026, over 16708.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2541, pruned_loss=0.03926, over 3329256.71 frames. ], batch size: 62, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:44:56,958 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247033.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:45:08,923 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.118e+02 2.456e+02 2.904e+02 5.980e+02, threshold=4.912e+02, percent-clipped=3.0 2023-05-01 23:45:26,163 INFO [train.py:904] (7/8) Epoch 25, batch 3450, loss[loss=0.1678, simple_loss=0.2445, pruned_loss=0.04552, over 16875.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2542, pruned_loss=0.0394, over 3326623.99 frames. ], batch size: 96, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:46:35,600 INFO [train.py:904] (7/8) Epoch 25, batch 3500, loss[loss=0.1742, simple_loss=0.2572, pruned_loss=0.04557, over 16751.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2526, pruned_loss=0.03928, over 3318080.15 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:47:30,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.124e+02 2.460e+02 2.748e+02 9.578e+02, threshold=4.920e+02, percent-clipped=1.0 2023-05-01 23:47:45,781 INFO [train.py:904] (7/8) Epoch 25, batch 3550, loss[loss=0.1769, simple_loss=0.271, pruned_loss=0.04136, over 17281.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2524, pruned_loss=0.0396, over 3311553.69 frames. ], batch size: 52, lr: 2.70e-03, grad_scale: 4.0 2023-05-01 23:48:08,575 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-01 23:48:55,175 INFO [train.py:904] (7/8) Epoch 25, batch 3600, loss[loss=0.1703, simple_loss=0.252, pruned_loss=0.04429, over 11540.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2516, pruned_loss=0.03924, over 3306407.52 frames. ], batch size: 246, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:48:58,935 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-01 23:49:49,112 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.664e+02 2.145e+02 2.569e+02 3.333e+02 7.279e+02, threshold=5.139e+02, percent-clipped=3.0 2023-05-01 23:50:05,473 INFO [train.py:904] (7/8) Epoch 25, batch 3650, loss[loss=0.1574, simple_loss=0.2311, pruned_loss=0.04187, over 16482.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2503, pruned_loss=0.03949, over 3288705.01 frames. ], batch size: 146, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:50:59,839 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247290.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:51:02,331 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-01 23:51:17,363 INFO [train.py:904] (7/8) Epoch 25, batch 3700, loss[loss=0.1676, simple_loss=0.2446, pruned_loss=0.04523, over 16829.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2486, pruned_loss=0.04064, over 3295393.79 frames. ], batch size: 102, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:52:01,347 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=247333.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:52:12,948 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.806e+02 2.292e+02 2.615e+02 3.085e+02 5.082e+02, threshold=5.229e+02, percent-clipped=0.0 2023-05-01 23:52:28,378 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3707, 4.3572, 4.3126, 4.0301, 4.1087, 4.3611, 4.0048, 4.1743], device='cuda:7'), covar=tensor([0.0630, 0.0803, 0.0291, 0.0266, 0.0610, 0.0541, 0.0827, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0475, 0.0369, 0.0374, 0.0374, 0.0430, 0.0252, 0.0447], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:52:30,274 INFO [train.py:904] (7/8) Epoch 25, batch 3750, loss[loss=0.1814, simple_loss=0.2574, pruned_loss=0.05271, over 16755.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2494, pruned_loss=0.04194, over 3278427.22 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:52:39,068 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9601, 4.0707, 4.2966, 4.2738, 4.3073, 4.0213, 4.0753, 4.0147], device='cuda:7'), covar=tensor([0.0479, 0.0665, 0.0453, 0.0453, 0.0632, 0.0558, 0.0851, 0.0666], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0489, 0.0475, 0.0437, 0.0520, 0.0500, 0.0579, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-01 23:52:52,262 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8217, 1.9864, 2.4113, 2.6838, 2.7671, 2.7784, 1.9796, 2.9496], device='cuda:7'), covar=tensor([0.0190, 0.0491, 0.0352, 0.0293, 0.0321, 0.0289, 0.0576, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0196, 0.0184, 0.0189, 0.0205, 0.0163, 0.0200, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:53:07,490 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247379.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:53:11,037 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=247381.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:53:14,449 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.62 vs. limit=5.0 2023-05-01 23:53:42,911 INFO [train.py:904] (7/8) Epoch 25, batch 3800, loss[loss=0.1711, simple_loss=0.2489, pruned_loss=0.04668, over 16433.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2511, pruned_loss=0.04372, over 3266507.13 frames. ], batch size: 146, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:54:16,101 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247426.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:54:36,422 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247440.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:54:38,774 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.614e+02 2.181e+02 2.707e+02 3.148e+02 5.686e+02, threshold=5.414e+02, percent-clipped=1.0 2023-05-01 23:54:54,729 INFO [train.py:904] (7/8) Epoch 25, batch 3850, loss[loss=0.1645, simple_loss=0.2398, pruned_loss=0.04457, over 16776.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2513, pruned_loss=0.04412, over 3274912.27 frames. ], batch size: 83, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:55:42,338 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2778, 3.3840, 3.6586, 2.2433, 3.2244, 2.4190, 3.8896, 3.8445], device='cuda:7'), covar=tensor([0.0211, 0.0782, 0.0579, 0.1961, 0.0782, 0.1073, 0.0386, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0168, 0.0170, 0.0155, 0.0147, 0.0132, 0.0145, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-01 23:55:43,354 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247487.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:56:02,922 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247500.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:56:06,078 INFO [train.py:904] (7/8) Epoch 25, batch 3900, loss[loss=0.1464, simple_loss=0.2251, pruned_loss=0.03386, over 16866.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.251, pruned_loss=0.0446, over 3273724.41 frames. ], batch size: 96, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:56:39,905 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247525.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:57:03,109 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.630e+02 2.139e+02 2.490e+02 2.933e+02 5.052e+02, threshold=4.980e+02, percent-clipped=0.0 2023-05-01 23:57:11,437 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1650, 2.2088, 2.3192, 3.8683, 2.1312, 2.5348, 2.2960, 2.4024], device='cuda:7'), covar=tensor([0.1572, 0.3873, 0.3144, 0.0676, 0.4137, 0.2760, 0.4224, 0.3102], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0467, 0.0385, 0.0338, 0.0445, 0.0535, 0.0439, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:57:17,847 INFO [train.py:904] (7/8) Epoch 25, batch 3950, loss[loss=0.176, simple_loss=0.2417, pruned_loss=0.05511, over 16896.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2499, pruned_loss=0.045, over 3284456.82 frames. ], batch size: 109, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:57:20,028 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247554.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:57:30,937 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247561.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:57:35,502 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8447, 3.9585, 3.0304, 2.3452, 2.5371, 2.5045, 3.9660, 3.4422], device='cuda:7'), covar=tensor([0.2668, 0.0571, 0.1693, 0.3152, 0.2659, 0.2142, 0.0592, 0.1331], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0274, 0.0312, 0.0321, 0.0305, 0.0271, 0.0303, 0.0349], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-01 23:58:06,037 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247586.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:58:12,177 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=247590.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:58:12,421 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1231, 2.1769, 2.3190, 3.7746, 2.1535, 2.4637, 2.2537, 2.3536], device='cuda:7'), covar=tensor([0.1625, 0.3815, 0.3119, 0.0681, 0.4031, 0.2754, 0.4013, 0.3133], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0467, 0.0385, 0.0338, 0.0446, 0.0535, 0.0439, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:58:30,190 INFO [train.py:904] (7/8) Epoch 25, batch 4000, loss[loss=0.1742, simple_loss=0.258, pruned_loss=0.0452, over 16742.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2501, pruned_loss=0.04508, over 3290241.59 frames. ], batch size: 83, lr: 2.70e-03, grad_scale: 8.0 2023-05-01 23:58:48,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7224, 4.6023, 4.8031, 4.9534, 5.0973, 4.5857, 5.0624, 5.1101], device='cuda:7'), covar=tensor([0.1738, 0.1185, 0.1378, 0.0675, 0.0511, 0.0907, 0.0668, 0.0562], device='cuda:7'), in_proj_covar=tensor([0.0687, 0.0844, 0.0971, 0.0852, 0.0652, 0.0672, 0.0699, 0.0820], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-01 23:58:49,610 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247615.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:59:17,798 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-01 23:59:21,888 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=247638.0, num_to_drop=0, layers_to_drop=set() 2023-05-01 23:59:27,472 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 1.992e+02 2.395e+02 2.922e+02 5.946e+02, threshold=4.790e+02, percent-clipped=1.0 2023-05-01 23:59:44,184 INFO [train.py:904] (7/8) Epoch 25, batch 4050, loss[loss=0.159, simple_loss=0.2498, pruned_loss=0.03415, over 16404.00 frames. ], tot_loss[loss=0.1696, simple_loss=0.2507, pruned_loss=0.04429, over 3293982.46 frames. ], batch size: 146, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:00:32,808 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 00:00:59,193 INFO [train.py:904] (7/8) Epoch 25, batch 4100, loss[loss=0.1842, simple_loss=0.2705, pruned_loss=0.04899, over 16745.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2522, pruned_loss=0.04404, over 3299790.16 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:01:48,425 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247735.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:01:57,929 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.293e+02 2.110e+02 2.475e+02 2.947e+02 8.069e+02, threshold=4.949e+02, percent-clipped=1.0 2023-05-02 00:02:14,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.94 vs. limit=5.0 2023-05-02 00:02:14,957 INFO [train.py:904] (7/8) Epoch 25, batch 4150, loss[loss=0.1762, simple_loss=0.2674, pruned_loss=0.04247, over 16454.00 frames. ], tot_loss[loss=0.1759, simple_loss=0.2592, pruned_loss=0.04631, over 3258643.39 frames. ], batch size: 68, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:03:01,852 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:03:05,022 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247784.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:03:15,679 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-02 00:03:32,732 INFO [train.py:904] (7/8) Epoch 25, batch 4200, loss[loss=0.2011, simple_loss=0.2981, pruned_loss=0.05208, over 17117.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2663, pruned_loss=0.04784, over 3222974.43 frames. ], batch size: 48, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:04:30,295 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.155e+02 2.612e+02 3.147e+02 5.112e+02, threshold=5.224e+02, percent-clipped=1.0 2023-05-02 00:04:35,346 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=247845.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:04:46,261 INFO [train.py:904] (7/8) Epoch 25, batch 4250, loss[loss=0.1739, simple_loss=0.269, pruned_loss=0.03938, over 15410.00 frames. ], tot_loss[loss=0.1824, simple_loss=0.2698, pruned_loss=0.04749, over 3213898.32 frames. ], batch size: 191, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:04:52,109 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247856.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:05:28,725 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:05:30,596 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7908, 4.9918, 5.1565, 4.8614, 4.9603, 5.5403, 5.0106, 4.7095], device='cuda:7'), covar=tensor([0.1027, 0.1701, 0.1889, 0.2046, 0.2316, 0.0881, 0.1481, 0.2472], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0625, 0.0680, 0.0508, 0.0675, 0.0709, 0.0531, 0.0679], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:05:40,984 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-02 00:05:42,000 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 00:05:51,440 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 00:06:02,544 INFO [train.py:904] (7/8) Epoch 25, batch 4300, loss[loss=0.1818, simple_loss=0.2781, pruned_loss=0.04276, over 16304.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2712, pruned_loss=0.04646, over 3210945.37 frames. ], batch size: 165, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:06:13,453 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=247910.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:06:24,417 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.63 vs. limit=2.0 2023-05-02 00:07:01,391 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.557e+02 2.163e+02 2.477e+02 3.039e+02 5.024e+02, threshold=4.955e+02, percent-clipped=0.0 2023-05-02 00:07:17,410 INFO [train.py:904] (7/8) Epoch 25, batch 4350, loss[loss=0.1834, simple_loss=0.2737, pruned_loss=0.04656, over 16762.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.2747, pruned_loss=0.04747, over 3201891.80 frames. ], batch size: 83, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:07:19,813 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5480, 3.6405, 3.3397, 2.9483, 3.2493, 3.5170, 3.3298, 3.3325], device='cuda:7'), covar=tensor([0.0538, 0.0528, 0.0283, 0.0267, 0.0498, 0.0409, 0.1239, 0.0474], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0460, 0.0359, 0.0364, 0.0363, 0.0418, 0.0246, 0.0433], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:07:51,443 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0648, 2.3821, 1.9333, 2.1798, 2.6953, 2.3270, 2.6646, 2.8880], device='cuda:7'), covar=tensor([0.0156, 0.0451, 0.0631, 0.0539, 0.0285, 0.0441, 0.0243, 0.0274], device='cuda:7'), in_proj_covar=tensor([0.0224, 0.0243, 0.0230, 0.0233, 0.0244, 0.0243, 0.0244, 0.0241], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:08:05,889 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247984.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:08:24,025 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=247996.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:08:36,588 INFO [train.py:904] (7/8) Epoch 25, batch 4400, loss[loss=0.2139, simple_loss=0.3006, pruned_loss=0.06355, over 16425.00 frames. ], tot_loss[loss=0.1871, simple_loss=0.2769, pruned_loss=0.04867, over 3195603.52 frames. ], batch size: 68, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:09:07,375 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8038, 3.8117, 2.4701, 4.7296, 3.0052, 4.5980, 2.6686, 3.1174], device='cuda:7'), covar=tensor([0.0308, 0.0384, 0.1600, 0.0133, 0.0863, 0.0360, 0.1321, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0181, 0.0196, 0.0171, 0.0178, 0.0222, 0.0204, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:09:24,795 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248035.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:09:28,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5135, 4.2415, 4.0994, 2.6904, 3.7758, 4.2227, 3.7074, 2.5911], device='cuda:7'), covar=tensor([0.0552, 0.0031, 0.0052, 0.0430, 0.0088, 0.0102, 0.0089, 0.0415], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0088, 0.0134, 0.0100, 0.0111, 0.0097, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 00:09:34,458 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 1.997e+02 2.337e+02 2.699e+02 4.708e+02, threshold=4.674e+02, percent-clipped=0.0 2023-05-02 00:09:39,340 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248045.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:09:50,317 INFO [train.py:904] (7/8) Epoch 25, batch 4450, loss[loss=0.2192, simple_loss=0.3085, pruned_loss=0.06488, over 16610.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2805, pruned_loss=0.05006, over 3206727.50 frames. ], batch size: 57, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:09:56,481 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248057.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:10:01,001 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3338, 5.9567, 6.1770, 5.8148, 5.9802, 6.4320, 5.9713, 5.6063], device='cuda:7'), covar=tensor([0.0782, 0.1327, 0.1462, 0.1485, 0.1714, 0.0591, 0.1081, 0.2024], device='cuda:7'), in_proj_covar=tensor([0.0419, 0.0619, 0.0673, 0.0502, 0.0667, 0.0704, 0.0525, 0.0671], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:10:33,151 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:10:33,275 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:10:35,245 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248083.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:11:04,506 INFO [train.py:904] (7/8) Epoch 25, batch 4500, loss[loss=0.1864, simple_loss=0.279, pruned_loss=0.04691, over 16580.00 frames. ], tot_loss[loss=0.1914, simple_loss=0.2808, pruned_loss=0.05101, over 3207383.36 frames. ], batch size: 75, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:11:44,571 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:11:44,724 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1491, 5.2072, 5.0373, 4.6477, 4.7417, 5.1276, 4.9157, 4.8088], device='cuda:7'), covar=tensor([0.0496, 0.0311, 0.0216, 0.0257, 0.0796, 0.0282, 0.0279, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0456, 0.0356, 0.0361, 0.0360, 0.0414, 0.0244, 0.0429], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:12:00,024 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248140.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:12:01,972 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 1.873e+02 2.071e+02 2.399e+02 5.095e+02, threshold=4.142e+02, percent-clipped=1.0 2023-05-02 00:12:04,328 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248143.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:12:16,042 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 00:12:17,519 INFO [train.py:904] (7/8) Epoch 25, batch 4550, loss[loss=0.2028, simple_loss=0.2748, pruned_loss=0.0654, over 11699.00 frames. ], tot_loss[loss=0.1929, simple_loss=0.2816, pruned_loss=0.0521, over 3207396.79 frames. ], batch size: 246, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:12:22,802 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248156.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:13:00,133 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248181.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:13:19,584 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-02 00:13:23,225 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 00:13:32,548 INFO [train.py:904] (7/8) Epoch 25, batch 4600, loss[loss=0.1802, simple_loss=0.2704, pruned_loss=0.04503, over 16798.00 frames. ], tot_loss[loss=0.1932, simple_loss=0.2822, pruned_loss=0.05211, over 3224955.12 frames. ], batch size: 124, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:13:34,234 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248204.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:13:43,382 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248210.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:14:11,936 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248229.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:14:29,188 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 1.740e+02 1.903e+02 2.276e+02 6.899e+02, threshold=3.806e+02, percent-clipped=0.0 2023-05-02 00:14:36,332 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2003, 4.9223, 4.6370, 3.4939, 4.1595, 4.7861, 4.0852, 3.0381], device='cuda:7'), covar=tensor([0.0435, 0.0020, 0.0037, 0.0326, 0.0087, 0.0064, 0.0083, 0.0359], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0088, 0.0135, 0.0100, 0.0111, 0.0097, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 00:14:46,950 INFO [train.py:904] (7/8) Epoch 25, batch 4650, loss[loss=0.1862, simple_loss=0.2748, pruned_loss=0.04877, over 16381.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2815, pruned_loss=0.05224, over 3232173.39 frames. ], batch size: 146, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:14:54,073 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248258.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:16:01,804 INFO [train.py:904] (7/8) Epoch 25, batch 4700, loss[loss=0.1756, simple_loss=0.2642, pruned_loss=0.04343, over 17011.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2789, pruned_loss=0.05126, over 3218904.80 frames. ], batch size: 53, lr: 2.70e-03, grad_scale: 8.0 2023-05-02 00:16:44,339 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248332.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:16:50,512 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7553, 3.8002, 3.9166, 3.6771, 3.8703, 4.2452, 3.8892, 3.5634], device='cuda:7'), covar=tensor([0.2334, 0.2191, 0.2306, 0.2573, 0.2563, 0.1915, 0.1579, 0.2678], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0622, 0.0678, 0.0506, 0.0672, 0.0709, 0.0527, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:16:56,011 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248340.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:16:57,980 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 1.897e+02 2.230e+02 2.532e+02 4.236e+02, threshold=4.460e+02, percent-clipped=2.0 2023-05-02 00:17:13,841 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248352.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:17:14,765 INFO [train.py:904] (7/8) Epoch 25, batch 4750, loss[loss=0.1943, simple_loss=0.2741, pruned_loss=0.0572, over 12038.00 frames. ], tot_loss[loss=0.1873, simple_loss=0.2753, pruned_loss=0.04961, over 3195162.74 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:17:59,868 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248383.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:18:13,875 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248393.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:18:29,689 INFO [train.py:904] (7/8) Epoch 25, batch 4800, loss[loss=0.1691, simple_loss=0.2666, pruned_loss=0.03581, over 16479.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2719, pruned_loss=0.04793, over 3183430.39 frames. ], batch size: 75, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:19:20,382 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4196, 4.4854, 4.7416, 4.7207, 4.7141, 4.4748, 4.4599, 4.3748], device='cuda:7'), covar=tensor([0.0299, 0.0419, 0.0352, 0.0327, 0.0400, 0.0329, 0.0748, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0463, 0.0454, 0.0415, 0.0497, 0.0474, 0.0552, 0.0377], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 00:19:22,248 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248438.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:19:25,581 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248440.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:19:27,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 1.792e+02 2.090e+02 2.399e+02 3.533e+02, threshold=4.179e+02, percent-clipped=0.0 2023-05-02 00:19:31,563 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248444.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:19:43,851 INFO [train.py:904] (7/8) Epoch 25, batch 4850, loss[loss=0.2109, simple_loss=0.3079, pruned_loss=0.05702, over 15437.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2726, pruned_loss=0.04705, over 3175732.33 frames. ], batch size: 190, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:19:50,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3390, 4.4011, 4.6854, 4.6528, 4.6518, 4.4124, 4.3996, 4.3301], device='cuda:7'), covar=tensor([0.0312, 0.0476, 0.0357, 0.0352, 0.0394, 0.0342, 0.0776, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0461, 0.0452, 0.0413, 0.0495, 0.0473, 0.0551, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 00:20:37,517 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248488.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:20:58,439 INFO [train.py:904] (7/8) Epoch 25, batch 4900, loss[loss=0.1688, simple_loss=0.2582, pruned_loss=0.03976, over 16358.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2713, pruned_loss=0.04568, over 3171812.81 frames. ], batch size: 35, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:22:03,448 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.924e+02 2.287e+02 2.776e+02 8.047e+02, threshold=4.573e+02, percent-clipped=3.0 2023-05-02 00:22:20,183 INFO [train.py:904] (7/8) Epoch 25, batch 4950, loss[loss=0.1704, simple_loss=0.2649, pruned_loss=0.03796, over 17111.00 frames. ], tot_loss[loss=0.1798, simple_loss=0.2703, pruned_loss=0.04468, over 3181691.60 frames. ], batch size: 48, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:22:23,932 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 00:23:31,273 INFO [train.py:904] (7/8) Epoch 25, batch 5000, loss[loss=0.1845, simple_loss=0.2815, pruned_loss=0.04374, over 16291.00 frames. ], tot_loss[loss=0.181, simple_loss=0.2722, pruned_loss=0.04487, over 3184315.17 frames. ], batch size: 165, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:24:25,422 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248640.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:24:27,930 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.126e+02 2.416e+02 2.928e+02 6.487e+02, threshold=4.832e+02, percent-clipped=1.0 2023-05-02 00:24:42,946 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248652.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:24:43,832 INFO [train.py:904] (7/8) Epoch 25, batch 5050, loss[loss=0.1808, simple_loss=0.2738, pruned_loss=0.04392, over 16521.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2726, pruned_loss=0.04481, over 3184421.47 frames. ], batch size: 68, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:25:35,138 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248688.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:35,154 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248688.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:40,400 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248691.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:53,429 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248700.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:25:57,290 INFO [train.py:904] (7/8) Epoch 25, batch 5100, loss[loss=0.1669, simple_loss=0.2483, pruned_loss=0.0428, over 16350.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2703, pruned_loss=0.04391, over 3202469.00 frames. ], batch size: 35, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:26:15,831 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=248715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:26:46,180 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8033, 1.4548, 1.6980, 1.7864, 1.8789, 1.9797, 1.6490, 1.8143], device='cuda:7'), covar=tensor([0.0267, 0.0433, 0.0246, 0.0332, 0.0295, 0.0200, 0.0462, 0.0164], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0194, 0.0182, 0.0187, 0.0201, 0.0159, 0.0198, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:26:50,046 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248738.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:26:52,086 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=248739.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:26:56,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8078, 4.2392, 3.0912, 2.4411, 2.8396, 2.7457, 4.4742, 3.6871], device='cuda:7'), covar=tensor([0.2754, 0.0508, 0.1788, 0.2700, 0.2338, 0.1743, 0.0430, 0.1096], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0272, 0.0310, 0.0319, 0.0302, 0.0269, 0.0301, 0.0346], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:26:56,587 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 1.974e+02 2.293e+02 2.738e+02 8.128e+02, threshold=4.585e+02, percent-clipped=1.0 2023-05-02 00:27:12,691 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:27:13,425 INFO [train.py:904] (7/8) Epoch 25, batch 5150, loss[loss=0.1887, simple_loss=0.2847, pruned_loss=0.04642, over 16745.00 frames. ], tot_loss[loss=0.1796, simple_loss=0.2709, pruned_loss=0.04412, over 3168425.94 frames. ], batch size: 124, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:27:47,509 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=248776.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:27:51,238 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2621, 3.1355, 3.5542, 1.7695, 3.6495, 3.7136, 2.8450, 2.7345], device='cuda:7'), covar=tensor([0.0883, 0.0315, 0.0153, 0.1273, 0.0085, 0.0150, 0.0419, 0.0517], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0110, 0.0100, 0.0138, 0.0084, 0.0128, 0.0129, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:28:03,458 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=248786.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:28:28,606 INFO [train.py:904] (7/8) Epoch 25, batch 5200, loss[loss=0.1816, simple_loss=0.2666, pruned_loss=0.04835, over 12098.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2697, pruned_loss=0.0437, over 3171799.45 frames. ], batch size: 246, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:28:31,486 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4424, 4.6844, 4.4397, 4.5057, 4.2518, 4.1885, 4.1578, 4.7093], device='cuda:7'), covar=tensor([0.1266, 0.0876, 0.1106, 0.0860, 0.0816, 0.1668, 0.1216, 0.0977], device='cuda:7'), in_proj_covar=tensor([0.0696, 0.0837, 0.0693, 0.0647, 0.0534, 0.0534, 0.0705, 0.0656], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:28:37,592 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8222, 2.9942, 2.9124, 5.1172, 4.0004, 4.2554, 1.9327, 3.0751], device='cuda:7'), covar=tensor([0.1339, 0.0754, 0.1140, 0.0131, 0.0335, 0.0404, 0.1511, 0.0872], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0178, 0.0196, 0.0194, 0.0204, 0.0215, 0.0206, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:29:15,502 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-02 00:29:25,204 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.065e+02 2.411e+02 2.935e+02 6.100e+02, threshold=4.822e+02, percent-clipped=2.0 2023-05-02 00:29:41,117 INFO [train.py:904] (7/8) Epoch 25, batch 5250, loss[loss=0.1717, simple_loss=0.262, pruned_loss=0.04065, over 16500.00 frames. ], tot_loss[loss=0.1772, simple_loss=0.2674, pruned_loss=0.04347, over 3175683.03 frames. ], batch size: 75, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:30:26,654 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-05-02 00:30:53,692 INFO [train.py:904] (7/8) Epoch 25, batch 5300, loss[loss=0.1575, simple_loss=0.2391, pruned_loss=0.03791, over 17116.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2637, pruned_loss=0.0423, over 3188297.13 frames. ], batch size: 49, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:31:51,346 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 1.862e+02 2.129e+02 2.555e+02 4.739e+02, threshold=4.259e+02, percent-clipped=0.0 2023-05-02 00:32:02,579 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0532, 2.8627, 2.1910, 2.4569, 3.1965, 2.8912, 3.6088, 3.5853], device='cuda:7'), covar=tensor([0.0101, 0.0504, 0.0726, 0.0573, 0.0298, 0.0461, 0.0254, 0.0281], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0239, 0.0228, 0.0230, 0.0240, 0.0239, 0.0239, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:32:08,004 INFO [train.py:904] (7/8) Epoch 25, batch 5350, loss[loss=0.1732, simple_loss=0.2686, pruned_loss=0.03885, over 16476.00 frames. ], tot_loss[loss=0.1728, simple_loss=0.2623, pruned_loss=0.04168, over 3191858.13 frames. ], batch size: 75, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:33:01,214 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=248988.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:33:12,860 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3113, 4.3157, 4.2395, 3.3803, 4.2719, 1.6435, 4.0396, 3.8492], device='cuda:7'), covar=tensor([0.0111, 0.0117, 0.0169, 0.0373, 0.0099, 0.3010, 0.0141, 0.0294], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0167, 0.0206, 0.0184, 0.0184, 0.0213, 0.0196, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:33:19,858 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.95 vs. limit=5.0 2023-05-02 00:33:22,813 INFO [train.py:904] (7/8) Epoch 25, batch 5400, loss[loss=0.2086, simple_loss=0.2926, pruned_loss=0.06233, over 12038.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2647, pruned_loss=0.04218, over 3195516.77 frames. ], batch size: 246, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:33:47,472 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 00:34:12,059 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249036.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:34:17,285 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249039.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:34:20,473 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 1.925e+02 2.196e+02 2.578e+02 3.732e+02, threshold=4.392e+02, percent-clipped=0.0 2023-05-02 00:34:29,147 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=249047.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:34:40,337 INFO [train.py:904] (7/8) Epoch 25, batch 5450, loss[loss=0.2042, simple_loss=0.2959, pruned_loss=0.05626, over 16669.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2672, pruned_loss=0.04326, over 3204796.98 frames. ], batch size: 134, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:35:09,493 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=249071.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:35:34,932 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249087.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:35:55,702 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7151, 1.8389, 1.6253, 1.5670, 1.9804, 1.6191, 1.6488, 1.9461], device='cuda:7'), covar=tensor([0.0217, 0.0286, 0.0424, 0.0338, 0.0219, 0.0264, 0.0187, 0.0209], device='cuda:7'), in_proj_covar=tensor([0.0220, 0.0240, 0.0228, 0.0231, 0.0241, 0.0239, 0.0240, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:35:58,374 INFO [train.py:904] (7/8) Epoch 25, batch 5500, loss[loss=0.1796, simple_loss=0.2768, pruned_loss=0.04118, over 17110.00 frames. ], tot_loss[loss=0.1836, simple_loss=0.2737, pruned_loss=0.04677, over 3176017.35 frames. ], batch size: 47, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:36:23,637 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3267, 5.2842, 5.1308, 3.7939, 5.2881, 1.8441, 4.9171, 4.8477], device='cuda:7'), covar=tensor([0.0164, 0.0126, 0.0272, 0.0795, 0.0141, 0.3620, 0.0226, 0.0373], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0168, 0.0207, 0.0185, 0.0185, 0.0214, 0.0197, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:36:26,332 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0423, 2.8135, 2.6018, 4.8575, 3.3134, 4.2177, 1.8297, 2.9496], device='cuda:7'), covar=tensor([0.1223, 0.0879, 0.1324, 0.0222, 0.0421, 0.0402, 0.1632, 0.0949], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0178, 0.0195, 0.0194, 0.0204, 0.0215, 0.0205, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:37:00,874 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.871e+02 3.023e+02 3.674e+02 4.693e+02 9.094e+02, threshold=7.348e+02, percent-clipped=33.0 2023-05-02 00:37:17,264 INFO [train.py:904] (7/8) Epoch 25, batch 5550, loss[loss=0.2261, simple_loss=0.3055, pruned_loss=0.07335, over 16722.00 frames. ], tot_loss[loss=0.1915, simple_loss=0.2804, pruned_loss=0.05131, over 3160945.78 frames. ], batch size: 134, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:37:59,482 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7740, 3.8126, 3.9354, 3.7419, 3.8785, 4.2156, 3.8841, 3.6205], device='cuda:7'), covar=tensor([0.2220, 0.2123, 0.2459, 0.2438, 0.2386, 0.1812, 0.1778, 0.2722], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0611, 0.0667, 0.0497, 0.0663, 0.0699, 0.0519, 0.0667], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:38:41,405 INFO [train.py:904] (7/8) Epoch 25, batch 5600, loss[loss=0.2088, simple_loss=0.2904, pruned_loss=0.06365, over 16779.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2857, pruned_loss=0.05591, over 3116849.64 frames. ], batch size: 124, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:38:43,597 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.78 vs. limit=5.0 2023-05-02 00:39:02,290 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:39:19,977 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6149, 2.5222, 1.7803, 2.6092, 2.0465, 2.7103, 2.0555, 2.2981], device='cuda:7'), covar=tensor([0.0294, 0.0323, 0.1209, 0.0255, 0.0590, 0.0427, 0.1112, 0.0549], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0179, 0.0194, 0.0168, 0.0177, 0.0219, 0.0202, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:39:32,441 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8049, 4.6618, 4.8417, 5.0140, 5.1911, 4.6062, 5.1869, 5.1680], device='cuda:7'), covar=tensor([0.1932, 0.1271, 0.1594, 0.0652, 0.0488, 0.0951, 0.0516, 0.0621], device='cuda:7'), in_proj_covar=tensor([0.0659, 0.0810, 0.0934, 0.0817, 0.0626, 0.0644, 0.0673, 0.0786], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:39:47,613 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 3.336e+02 4.056e+02 4.975e+02 8.511e+02, threshold=8.112e+02, percent-clipped=2.0 2023-05-02 00:40:04,596 INFO [train.py:904] (7/8) Epoch 25, batch 5650, loss[loss=0.2278, simple_loss=0.3125, pruned_loss=0.07151, over 16329.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2907, pruned_loss=0.05974, over 3081221.87 frames. ], batch size: 146, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:40:41,258 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=249276.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:40:41,397 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-05-02 00:41:22,858 INFO [train.py:904] (7/8) Epoch 25, batch 5700, loss[loss=0.2173, simple_loss=0.316, pruned_loss=0.05928, over 16851.00 frames. ], tot_loss[loss=0.2071, simple_loss=0.2921, pruned_loss=0.06111, over 3076187.93 frames. ], batch size: 116, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:42:25,340 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.887e+02 2.916e+02 3.392e+02 3.948e+02 9.464e+02, threshold=6.785e+02, percent-clipped=1.0 2023-05-02 00:42:35,246 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249347.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:42:43,873 INFO [train.py:904] (7/8) Epoch 25, batch 5750, loss[loss=0.204, simple_loss=0.2853, pruned_loss=0.06137, over 17029.00 frames. ], tot_loss[loss=0.2103, simple_loss=0.295, pruned_loss=0.06281, over 3051482.97 frames. ], batch size: 53, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:43:13,850 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249371.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:43:53,992 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249395.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:44:07,475 INFO [train.py:904] (7/8) Epoch 25, batch 5800, loss[loss=0.1684, simple_loss=0.2649, pruned_loss=0.03592, over 16876.00 frames. ], tot_loss[loss=0.209, simple_loss=0.2945, pruned_loss=0.06178, over 3050793.62 frames. ], batch size: 96, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:44:32,819 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249419.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:45:09,432 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.976e+02 3.356e+02 4.382e+02 8.536e+02, threshold=6.712e+02, percent-clipped=3.0 2023-05-02 00:45:26,299 INFO [train.py:904] (7/8) Epoch 25, batch 5850, loss[loss=0.1995, simple_loss=0.2868, pruned_loss=0.05614, over 16526.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2921, pruned_loss=0.06009, over 3063074.80 frames. ], batch size: 68, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:46:41,335 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1721, 3.1201, 1.7128, 3.3629, 2.2631, 3.4331, 1.9478, 2.5016], device='cuda:7'), covar=tensor([0.0299, 0.0412, 0.1957, 0.0248, 0.0926, 0.0579, 0.1802, 0.0887], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0179, 0.0194, 0.0168, 0.0177, 0.0218, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:46:46,945 INFO [train.py:904] (7/8) Epoch 25, batch 5900, loss[loss=0.2302, simple_loss=0.2963, pruned_loss=0.08206, over 11639.00 frames. ], tot_loss[loss=0.2055, simple_loss=0.2914, pruned_loss=0.05978, over 3049169.48 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:47:19,242 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 00:47:52,232 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.624e+02 3.247e+02 4.027e+02 8.250e+02, threshold=6.495e+02, percent-clipped=2.0 2023-05-02 00:48:08,080 INFO [train.py:904] (7/8) Epoch 25, batch 5950, loss[loss=0.2075, simple_loss=0.2945, pruned_loss=0.06028, over 15343.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2924, pruned_loss=0.05838, over 3072949.62 frames. ], batch size: 191, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:48:37,010 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=249571.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:49:28,152 INFO [train.py:904] (7/8) Epoch 25, batch 6000, loss[loss=0.2245, simple_loss=0.3095, pruned_loss=0.06978, over 11610.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2908, pruned_loss=0.0575, over 3084398.91 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:49:28,153 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 00:49:38,595 INFO [train.py:938] (7/8) Epoch 25, validation: loss=0.1487, simple_loss=0.2613, pruned_loss=0.01811, over 944034.00 frames. 2023-05-02 00:49:38,596 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 00:50:28,968 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 00:50:36,541 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.971e+02 2.903e+02 3.391e+02 3.955e+02 5.249e+02, threshold=6.782e+02, percent-clipped=0.0 2023-05-02 00:50:54,736 INFO [train.py:904] (7/8) Epoch 25, batch 6050, loss[loss=0.1875, simple_loss=0.28, pruned_loss=0.04751, over 16601.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2894, pruned_loss=0.05677, over 3105608.25 frames. ], batch size: 68, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:51:51,466 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8035, 4.1955, 3.0388, 2.3795, 2.7874, 2.6107, 4.5057, 3.6678], device='cuda:7'), covar=tensor([0.2974, 0.0616, 0.1899, 0.2999, 0.2712, 0.2004, 0.0485, 0.1284], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0272, 0.0310, 0.0320, 0.0302, 0.0268, 0.0301, 0.0345], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:51:59,314 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 00:52:12,424 INFO [train.py:904] (7/8) Epoch 25, batch 6100, loss[loss=0.1944, simple_loss=0.29, pruned_loss=0.04937, over 16413.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2893, pruned_loss=0.05559, over 3109185.36 frames. ], batch size: 146, lr: 2.69e-03, grad_scale: 16.0 2023-05-02 00:52:34,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0215, 2.1370, 2.5850, 2.9453, 2.9049, 3.4234, 2.3367, 3.3376], device='cuda:7'), covar=tensor([0.0222, 0.0495, 0.0341, 0.0298, 0.0283, 0.0169, 0.0491, 0.0130], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0194, 0.0182, 0.0186, 0.0201, 0.0160, 0.0199, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:52:35,772 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1758, 3.2415, 1.9445, 3.4731, 2.4351, 3.5210, 2.0942, 2.6374], device='cuda:7'), covar=tensor([0.0327, 0.0443, 0.1630, 0.0232, 0.0852, 0.0575, 0.1522, 0.0817], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0180, 0.0195, 0.0168, 0.0177, 0.0219, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 00:53:15,563 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.704e+02 2.711e+02 3.338e+02 3.699e+02 9.166e+02, threshold=6.676e+02, percent-clipped=2.0 2023-05-02 00:53:17,784 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 00:53:29,951 INFO [train.py:904] (7/8) Epoch 25, batch 6150, loss[loss=0.1849, simple_loss=0.2805, pruned_loss=0.04467, over 16831.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2875, pruned_loss=0.0554, over 3105113.38 frames. ], batch size: 102, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:54:06,372 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9071, 2.2082, 2.4662, 3.1200, 2.2290, 2.4060, 2.3862, 2.3251], device='cuda:7'), covar=tensor([0.1372, 0.3086, 0.2343, 0.0690, 0.3860, 0.2176, 0.3061, 0.3004], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0463, 0.0379, 0.0332, 0.0442, 0.0529, 0.0433, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:54:33,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5462, 3.7431, 2.6576, 2.2693, 2.4481, 2.3873, 3.9378, 3.3248], device='cuda:7'), covar=tensor([0.3017, 0.0599, 0.2020, 0.2730, 0.2579, 0.2152, 0.0450, 0.1280], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0272, 0.0311, 0.0320, 0.0302, 0.0268, 0.0301, 0.0345], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:54:42,852 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3958, 2.2226, 1.8960, 2.0191, 2.4767, 2.1781, 2.1778, 2.6306], device='cuda:7'), covar=tensor([0.0246, 0.0431, 0.0593, 0.0499, 0.0267, 0.0410, 0.0240, 0.0281], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0240, 0.0228, 0.0231, 0.0240, 0.0239, 0.0240, 0.0237], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:54:51,561 INFO [train.py:904] (7/8) Epoch 25, batch 6200, loss[loss=0.2201, simple_loss=0.29, pruned_loss=0.07509, over 11710.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2853, pruned_loss=0.05505, over 3098754.32 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:54:52,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3261, 3.4357, 3.6049, 3.5887, 3.6068, 3.4143, 3.4785, 3.5025], device='cuda:7'), covar=tensor([0.0417, 0.0714, 0.0453, 0.0459, 0.0546, 0.0525, 0.0782, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0469, 0.0455, 0.0416, 0.0498, 0.0475, 0.0554, 0.0381], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 00:55:01,880 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4112, 3.2719, 2.6222, 2.1686, 2.2593, 2.3075, 3.3857, 3.0449], device='cuda:7'), covar=tensor([0.2922, 0.0661, 0.1884, 0.2739, 0.2457, 0.2170, 0.0526, 0.1266], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0273, 0.0311, 0.0321, 0.0303, 0.0269, 0.0302, 0.0346], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 00:55:10,468 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249814.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:55:13,580 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1057, 2.2223, 2.2425, 3.7471, 2.1512, 2.5413, 2.3072, 2.3780], device='cuda:7'), covar=tensor([0.1488, 0.3608, 0.3104, 0.0600, 0.4146, 0.2589, 0.3627, 0.3380], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0463, 0.0380, 0.0332, 0.0442, 0.0530, 0.0434, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:55:55,234 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.970e+02 2.750e+02 3.284e+02 4.072e+02 6.772e+02, threshold=6.569e+02, percent-clipped=1.0 2023-05-02 00:56:10,151 INFO [train.py:904] (7/8) Epoch 25, batch 6250, loss[loss=0.208, simple_loss=0.2886, pruned_loss=0.06368, over 15289.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2854, pruned_loss=0.05538, over 3100292.82 frames. ], batch size: 190, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:56:37,222 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2712, 4.1335, 4.3387, 4.4750, 4.6368, 4.1988, 4.5713, 4.6418], device='cuda:7'), covar=tensor([0.1883, 0.1344, 0.1761, 0.0832, 0.0676, 0.1295, 0.0959, 0.0750], device='cuda:7'), in_proj_covar=tensor([0.0657, 0.0808, 0.0933, 0.0817, 0.0626, 0.0645, 0.0675, 0.0785], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 00:56:38,896 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=249871.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:56:44,523 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=249875.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:57:26,277 INFO [train.py:904] (7/8) Epoch 25, batch 6300, loss[loss=0.2129, simple_loss=0.3006, pruned_loss=0.0626, over 16586.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.285, pruned_loss=0.05418, over 3129746.28 frames. ], batch size: 62, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:57:49,328 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 00:57:52,644 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=249919.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 00:58:29,097 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.822e+02 2.746e+02 3.246e+02 3.923e+02 7.422e+02, threshold=6.492e+02, percent-clipped=1.0 2023-05-02 00:58:45,048 INFO [train.py:904] (7/8) Epoch 25, batch 6350, loss[loss=0.2437, simple_loss=0.3099, pruned_loss=0.08875, over 11728.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2858, pruned_loss=0.05525, over 3131110.33 frames. ], batch size: 248, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 00:58:48,539 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249955.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 00:59:20,607 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=249976.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:00:04,132 INFO [train.py:904] (7/8) Epoch 25, batch 6400, loss[loss=0.1728, simple_loss=0.2602, pruned_loss=0.04269, over 16553.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.286, pruned_loss=0.05637, over 3113206.21 frames. ], batch size: 62, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 01:00:24,526 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250016.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 01:00:56,560 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250037.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:01:05,940 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.084e+02 2.850e+02 3.461e+02 4.247e+02 7.668e+02, threshold=6.921e+02, percent-clipped=1.0 2023-05-02 01:01:20,206 INFO [train.py:904] (7/8) Epoch 25, batch 6450, loss[loss=0.1901, simple_loss=0.2844, pruned_loss=0.04786, over 16333.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2861, pruned_loss=0.056, over 3098067.11 frames. ], batch size: 165, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 01:02:01,881 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-02 01:02:04,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3628, 2.8652, 3.0478, 1.9058, 2.8107, 2.0648, 3.0638, 3.1269], device='cuda:7'), covar=tensor([0.0286, 0.0796, 0.0575, 0.2127, 0.0829, 0.1008, 0.0640, 0.0847], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0167, 0.0169, 0.0155, 0.0147, 0.0131, 0.0145, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:02:36,780 INFO [train.py:904] (7/8) Epoch 25, batch 6500, loss[loss=0.2134, simple_loss=0.3103, pruned_loss=0.05829, over 16929.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2846, pruned_loss=0.05533, over 3124644.16 frames. ], batch size: 109, lr: 2.69e-03, grad_scale: 8.0 2023-05-02 01:03:25,913 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7906, 2.7202, 2.8688, 2.1622, 2.7488, 2.1701, 2.7118, 2.9008], device='cuda:7'), covar=tensor([0.0259, 0.0753, 0.0487, 0.1731, 0.0761, 0.0893, 0.0521, 0.0657], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0169, 0.0154, 0.0146, 0.0131, 0.0144, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:03:28,322 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7492, 4.7327, 4.6417, 3.7430, 4.5560, 1.8893, 4.3199, 4.2196], device='cuda:7'), covar=tensor([0.0160, 0.0153, 0.0219, 0.0441, 0.0138, 0.2775, 0.0194, 0.0274], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0166, 0.0206, 0.0183, 0.0183, 0.0212, 0.0195, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:03:40,562 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.643e+02 3.230e+02 3.830e+02 1.038e+03, threshold=6.461e+02, percent-clipped=2.0 2023-05-02 01:03:43,547 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6840, 3.7591, 2.3039, 4.2719, 2.8020, 4.1513, 2.3780, 3.0465], device='cuda:7'), covar=tensor([0.0300, 0.0371, 0.1744, 0.0269, 0.0945, 0.0772, 0.1724, 0.0839], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0195, 0.0168, 0.0177, 0.0219, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:03:52,681 INFO [train.py:904] (7/8) Epoch 25, batch 6550, loss[loss=0.1966, simple_loss=0.3013, pruned_loss=0.04597, over 16308.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.287, pruned_loss=0.05586, over 3118409.16 frames. ], batch size: 165, lr: 2.69e-03, grad_scale: 4.0 2023-05-02 01:04:19,066 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=250170.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:04:35,665 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6862, 4.7149, 5.0234, 5.0217, 5.0578, 4.7938, 4.7382, 4.5579], device='cuda:7'), covar=tensor([0.0319, 0.0609, 0.0487, 0.0455, 0.0405, 0.0483, 0.0881, 0.0541], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0470, 0.0457, 0.0417, 0.0500, 0.0478, 0.0556, 0.0382], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 01:05:05,613 INFO [train.py:904] (7/8) Epoch 25, batch 6600, loss[loss=0.2249, simple_loss=0.2948, pruned_loss=0.07745, over 11257.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.29, pruned_loss=0.05724, over 3107984.99 frames. ], batch size: 247, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:05:06,404 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.64 vs. limit=5.0 2023-05-02 01:06:08,500 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.914e+02 2.823e+02 3.325e+02 3.947e+02 6.947e+02, threshold=6.650e+02, percent-clipped=1.0 2023-05-02 01:06:21,904 INFO [train.py:904] (7/8) Epoch 25, batch 6650, loss[loss=0.2463, simple_loss=0.309, pruned_loss=0.09185, over 11276.00 frames. ], tot_loss[loss=0.2037, simple_loss=0.2906, pruned_loss=0.05836, over 3090739.91 frames. ], batch size: 247, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:07:22,046 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6641, 3.3424, 3.9112, 1.9210, 4.0954, 4.1076, 3.0563, 2.9872], device='cuda:7'), covar=tensor([0.0778, 0.0347, 0.0200, 0.1288, 0.0080, 0.0176, 0.0465, 0.0514], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0101, 0.0140, 0.0085, 0.0131, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:07:37,555 INFO [train.py:904] (7/8) Epoch 25, batch 6700, loss[loss=0.2112, simple_loss=0.3009, pruned_loss=0.06075, over 15206.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2891, pruned_loss=0.05831, over 3099220.25 frames. ], batch size: 190, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:07:50,666 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=250311.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 01:08:19,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8866, 1.9641, 2.4828, 2.9035, 2.7238, 3.2763, 2.1425, 3.2667], device='cuda:7'), covar=tensor([0.0224, 0.0551, 0.0345, 0.0300, 0.0333, 0.0213, 0.0562, 0.0145], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0194, 0.0181, 0.0185, 0.0200, 0.0160, 0.0198, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:08:23,202 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=250332.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:08:41,138 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 2.721e+02 3.128e+02 3.707e+02 8.776e+02, threshold=6.256e+02, percent-clipped=1.0 2023-05-02 01:08:54,003 INFO [train.py:904] (7/8) Epoch 25, batch 6750, loss[loss=0.1976, simple_loss=0.2824, pruned_loss=0.0564, over 15397.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2879, pruned_loss=0.05806, over 3096844.34 frames. ], batch size: 191, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:09:48,817 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 01:09:49,874 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7706, 3.7086, 3.9236, 3.6939, 3.8905, 4.2375, 3.9358, 3.6284], device='cuda:7'), covar=tensor([0.2160, 0.2517, 0.2495, 0.2374, 0.2453, 0.1929, 0.1552, 0.2618], device='cuda:7'), in_proj_covar=tensor([0.0418, 0.0612, 0.0673, 0.0501, 0.0664, 0.0700, 0.0519, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 01:10:10,572 INFO [train.py:904] (7/8) Epoch 25, batch 6800, loss[loss=0.1885, simple_loss=0.2869, pruned_loss=0.04507, over 16824.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2874, pruned_loss=0.05779, over 3096093.49 frames. ], batch size: 96, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:11:16,514 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.847e+02 3.369e+02 4.013e+02 7.021e+02, threshold=6.738e+02, percent-clipped=2.0 2023-05-02 01:11:27,502 INFO [train.py:904] (7/8) Epoch 25, batch 6850, loss[loss=0.205, simple_loss=0.3018, pruned_loss=0.05411, over 16530.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2889, pruned_loss=0.05834, over 3089859.44 frames. ], batch size: 35, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:11:46,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5290, 3.5501, 3.5035, 2.7415, 3.3794, 2.0578, 3.1635, 2.8801], device='cuda:7'), covar=tensor([0.0205, 0.0159, 0.0220, 0.0261, 0.0119, 0.2487, 0.0165, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0167, 0.0206, 0.0183, 0.0183, 0.0213, 0.0195, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:11:52,830 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=250470.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:12:08,725 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0558, 2.3344, 2.2728, 3.9163, 2.1550, 2.6430, 2.2942, 2.4504], device='cuda:7'), covar=tensor([0.1602, 0.3881, 0.3178, 0.0576, 0.4274, 0.2682, 0.4170, 0.3232], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0460, 0.0377, 0.0331, 0.0440, 0.0528, 0.0431, 0.0537], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:12:33,223 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 01:12:43,358 INFO [train.py:904] (7/8) Epoch 25, batch 6900, loss[loss=0.1762, simple_loss=0.2749, pruned_loss=0.03876, over 16733.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2912, pruned_loss=0.05797, over 3104512.21 frames. ], batch size: 83, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:13:06,194 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=250518.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:13:13,914 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4863, 3.3498, 2.5674, 2.0981, 2.4130, 2.1071, 3.4583, 3.1281], device='cuda:7'), covar=tensor([0.3001, 0.0628, 0.1942, 0.2685, 0.2492, 0.2392, 0.0579, 0.1297], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0272, 0.0311, 0.0320, 0.0303, 0.0269, 0.0301, 0.0345], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 01:13:47,070 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.918e+02 2.786e+02 3.277e+02 4.153e+02 1.055e+03, threshold=6.554e+02, percent-clipped=5.0 2023-05-02 01:14:00,201 INFO [train.py:904] (7/8) Epoch 25, batch 6950, loss[loss=0.2054, simple_loss=0.2943, pruned_loss=0.05825, over 16732.00 frames. ], tot_loss[loss=0.2069, simple_loss=0.2935, pruned_loss=0.06009, over 3089735.65 frames. ], batch size: 124, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:14:00,759 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5650, 1.7979, 2.2385, 2.5326, 2.5139, 2.8366, 1.8613, 2.8245], device='cuda:7'), covar=tensor([0.0252, 0.0586, 0.0363, 0.0362, 0.0372, 0.0227, 0.0651, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0195, 0.0183, 0.0186, 0.0202, 0.0161, 0.0199, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:15:18,390 INFO [train.py:904] (7/8) Epoch 25, batch 7000, loss[loss=0.1997, simple_loss=0.3001, pruned_loss=0.04964, over 16463.00 frames. ], tot_loss[loss=0.2057, simple_loss=0.2933, pruned_loss=0.05908, over 3085594.99 frames. ], batch size: 68, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:15:32,213 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=250611.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 01:16:03,584 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=250632.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:16:18,360 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9298, 3.1991, 3.3978, 2.0698, 2.9158, 2.1872, 3.4023, 3.4515], device='cuda:7'), covar=tensor([0.0241, 0.0811, 0.0609, 0.2050, 0.0841, 0.1043, 0.0624, 0.0984], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0168, 0.0170, 0.0156, 0.0148, 0.0132, 0.0146, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:16:22,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.716e+02 3.192e+02 3.909e+02 7.863e+02, threshold=6.384e+02, percent-clipped=2.0 2023-05-02 01:16:35,092 INFO [train.py:904] (7/8) Epoch 25, batch 7050, loss[loss=0.2008, simple_loss=0.2894, pruned_loss=0.05608, over 16629.00 frames. ], tot_loss[loss=0.2059, simple_loss=0.2941, pruned_loss=0.05887, over 3087256.41 frames. ], batch size: 57, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:16:44,699 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=250659.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 01:16:49,672 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=250662.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:17:17,494 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=250680.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:17:44,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3799, 2.5620, 2.4696, 4.1684, 2.2742, 2.8065, 2.5873, 2.7379], device='cuda:7'), covar=tensor([0.1497, 0.3476, 0.2940, 0.0568, 0.4363, 0.2600, 0.3369, 0.3429], device='cuda:7'), in_proj_covar=tensor([0.0410, 0.0460, 0.0377, 0.0331, 0.0440, 0.0527, 0.0432, 0.0538], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:17:51,098 INFO [train.py:904] (7/8) Epoch 25, batch 7100, loss[loss=0.1991, simple_loss=0.2836, pruned_loss=0.0573, over 16719.00 frames. ], tot_loss[loss=0.2051, simple_loss=0.2926, pruned_loss=0.0588, over 3070386.02 frames. ], batch size: 124, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:17:59,072 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3716, 3.3728, 3.4407, 3.5184, 3.5440, 3.2969, 3.5089, 3.5881], device='cuda:7'), covar=tensor([0.1364, 0.0969, 0.1044, 0.0705, 0.0724, 0.2382, 0.1105, 0.0901], device='cuda:7'), in_proj_covar=tensor([0.0651, 0.0800, 0.0924, 0.0810, 0.0622, 0.0642, 0.0673, 0.0782], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:17:59,329 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-05-02 01:18:23,475 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250723.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:18:56,842 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 2.742e+02 3.283e+02 4.161e+02 6.949e+02, threshold=6.565e+02, percent-clipped=2.0 2023-05-02 01:19:09,296 INFO [train.py:904] (7/8) Epoch 25, batch 7150, loss[loss=0.1876, simple_loss=0.2764, pruned_loss=0.04937, over 16638.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2907, pruned_loss=0.05878, over 3061385.71 frames. ], batch size: 57, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:20:22,682 INFO [train.py:904] (7/8) Epoch 25, batch 7200, loss[loss=0.1906, simple_loss=0.2909, pruned_loss=0.04513, over 16693.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2887, pruned_loss=0.05704, over 3053066.34 frames. ], batch size: 134, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:20:23,976 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6991, 1.7872, 1.6464, 1.5033, 1.8864, 1.5487, 1.5536, 1.8694], device='cuda:7'), covar=tensor([0.0197, 0.0338, 0.0456, 0.0363, 0.0231, 0.0297, 0.0197, 0.0227], device='cuda:7'), in_proj_covar=tensor([0.0217, 0.0237, 0.0227, 0.0228, 0.0238, 0.0236, 0.0236, 0.0235], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:21:08,885 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=250833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:21:11,271 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 01:21:28,108 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=250845.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:21:28,840 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.397e+02 2.821e+02 3.422e+02 6.552e+02, threshold=5.642e+02, percent-clipped=0.0 2023-05-02 01:21:32,789 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4968, 4.6782, 4.8556, 4.6335, 4.7334, 5.1949, 4.7347, 4.5090], device='cuda:7'), covar=tensor([0.1432, 0.1906, 0.2146, 0.1912, 0.2283, 0.0945, 0.1562, 0.2305], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0617, 0.0678, 0.0506, 0.0671, 0.0704, 0.0527, 0.0676], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 01:21:32,995 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7060, 2.3280, 2.3879, 3.6238, 2.5210, 3.7401, 1.4940, 2.7689], device='cuda:7'), covar=tensor([0.1401, 0.0908, 0.1328, 0.0201, 0.0214, 0.0383, 0.1786, 0.0890], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0182, 0.0200, 0.0198, 0.0209, 0.0219, 0.0209, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:21:41,048 INFO [train.py:904] (7/8) Epoch 25, batch 7250, loss[loss=0.1687, simple_loss=0.2573, pruned_loss=0.04009, over 16941.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2861, pruned_loss=0.05599, over 3049188.51 frames. ], batch size: 109, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:22:20,244 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6191, 2.5663, 1.9381, 2.6734, 2.1238, 2.7623, 2.1670, 2.4140], device='cuda:7'), covar=tensor([0.0316, 0.0387, 0.1252, 0.0282, 0.0668, 0.0539, 0.1229, 0.0581], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0195, 0.0168, 0.0178, 0.0219, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:22:42,122 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:22:55,045 INFO [train.py:904] (7/8) Epoch 25, batch 7300, loss[loss=0.206, simple_loss=0.2953, pruned_loss=0.05833, over 16666.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2851, pruned_loss=0.05545, over 3052766.91 frames. ], batch size: 62, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:23:00,871 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=250906.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 01:23:11,188 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.84 vs. limit=5.0 2023-05-02 01:23:59,654 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.893e+02 3.388e+02 3.918e+02 7.629e+02, threshold=6.776e+02, percent-clipped=7.0 2023-05-02 01:24:09,712 INFO [train.py:904] (7/8) Epoch 25, batch 7350, loss[loss=0.1919, simple_loss=0.2738, pruned_loss=0.05501, over 17052.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2859, pruned_loss=0.05593, over 3062156.40 frames. ], batch size: 55, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:24:38,543 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 01:25:03,076 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5480, 2.6744, 2.1678, 2.4388, 3.0479, 2.6608, 3.1538, 3.2910], device='cuda:7'), covar=tensor([0.0138, 0.0450, 0.0586, 0.0470, 0.0270, 0.0416, 0.0220, 0.0246], device='cuda:7'), in_proj_covar=tensor([0.0215, 0.0235, 0.0225, 0.0226, 0.0235, 0.0235, 0.0234, 0.0234], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:25:27,071 INFO [train.py:904] (7/8) Epoch 25, batch 7400, loss[loss=0.224, simple_loss=0.3099, pruned_loss=0.06908, over 16509.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2873, pruned_loss=0.05655, over 3068874.95 frames. ], batch size: 68, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:25:38,919 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7248, 4.9517, 4.7706, 4.7482, 4.5177, 4.4793, 4.4618, 5.0476], device='cuda:7'), covar=tensor([0.1187, 0.0873, 0.0938, 0.0869, 0.0772, 0.1123, 0.1126, 0.0805], device='cuda:7'), in_proj_covar=tensor([0.0688, 0.0833, 0.0687, 0.0641, 0.0528, 0.0534, 0.0698, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:25:50,975 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251018.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:26:02,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6682, 3.7081, 2.2058, 4.2377, 2.8460, 4.1809, 2.4699, 3.0013], device='cuda:7'), covar=tensor([0.0317, 0.0481, 0.1889, 0.0286, 0.0937, 0.0623, 0.1619, 0.0901], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0181, 0.0197, 0.0169, 0.0179, 0.0220, 0.0205, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:26:35,984 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.726e+02 3.311e+02 3.902e+02 7.761e+02, threshold=6.622e+02, percent-clipped=1.0 2023-05-02 01:26:46,740 INFO [train.py:904] (7/8) Epoch 25, batch 7450, loss[loss=0.2326, simple_loss=0.3031, pruned_loss=0.08105, over 11519.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2885, pruned_loss=0.05741, over 3081744.79 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:28:06,276 INFO [train.py:904] (7/8) Epoch 25, batch 7500, loss[loss=0.2082, simple_loss=0.2931, pruned_loss=0.06161, over 17134.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2887, pruned_loss=0.05691, over 3064121.30 frames. ], batch size: 49, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:28:38,257 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3761, 3.3299, 3.4165, 3.4823, 3.5240, 3.2893, 3.5143, 3.5585], device='cuda:7'), covar=tensor([0.1197, 0.0994, 0.1032, 0.0684, 0.0667, 0.2052, 0.1009, 0.0892], device='cuda:7'), in_proj_covar=tensor([0.0648, 0.0800, 0.0921, 0.0808, 0.0620, 0.0638, 0.0670, 0.0778], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:29:12,779 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.911e+02 3.474e+02 4.178e+02 7.715e+02, threshold=6.948e+02, percent-clipped=3.0 2023-05-02 01:29:23,961 INFO [train.py:904] (7/8) Epoch 25, batch 7550, loss[loss=0.1958, simple_loss=0.2757, pruned_loss=0.05797, over 16607.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2882, pruned_loss=0.05751, over 3054238.36 frames. ], batch size: 62, lr: 2.68e-03, grad_scale: 4.0 2023-05-02 01:30:19,206 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251189.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:30:32,681 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:30:37,926 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251201.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 01:30:39,958 INFO [train.py:904] (7/8) Epoch 25, batch 7600, loss[loss=0.2031, simple_loss=0.2967, pruned_loss=0.05478, over 16744.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2871, pruned_loss=0.05769, over 3061739.65 frames. ], batch size: 83, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:31:43,661 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.965e+02 3.524e+02 4.371e+02 1.071e+03, threshold=7.047e+02, percent-clipped=4.0 2023-05-02 01:31:53,116 INFO [train.py:904] (7/8) Epoch 25, batch 7650, loss[loss=0.2005, simple_loss=0.2844, pruned_loss=0.05833, over 17038.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2879, pruned_loss=0.05823, over 3065682.97 frames. ], batch size: 55, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:32:03,136 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251259.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:33:07,033 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251301.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:33:08,936 INFO [train.py:904] (7/8) Epoch 25, batch 7700, loss[loss=0.1863, simple_loss=0.2772, pruned_loss=0.04769, over 16254.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2879, pruned_loss=0.05877, over 3051761.73 frames. ], batch size: 165, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:33:32,436 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:34:14,485 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.002e+02 2.953e+02 3.490e+02 4.215e+02 8.631e+02, threshold=6.980e+02, percent-clipped=3.0 2023-05-02 01:34:25,533 INFO [train.py:904] (7/8) Epoch 25, batch 7750, loss[loss=0.1912, simple_loss=0.281, pruned_loss=0.05071, over 16634.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2879, pruned_loss=0.05877, over 3056587.49 frames. ], batch size: 62, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:34:39,763 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8715, 2.6682, 2.5661, 1.9477, 2.5308, 2.6876, 2.5537, 1.9576], device='cuda:7'), covar=tensor([0.0440, 0.0106, 0.0097, 0.0380, 0.0154, 0.0145, 0.0143, 0.0403], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0087, 0.0087, 0.0134, 0.0099, 0.0111, 0.0096, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 01:34:39,797 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251362.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:34:45,643 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251366.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:35:40,752 INFO [train.py:904] (7/8) Epoch 25, batch 7800, loss[loss=0.1666, simple_loss=0.2639, pruned_loss=0.03464, over 16848.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2886, pruned_loss=0.05909, over 3066632.05 frames. ], batch size: 96, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:35:42,502 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251404.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:36:44,883 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 01:36:45,154 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.753e+02 2.864e+02 3.394e+02 4.001e+02 5.962e+02, threshold=6.789e+02, percent-clipped=0.0 2023-05-02 01:36:55,313 INFO [train.py:904] (7/8) Epoch 25, batch 7850, loss[loss=0.1769, simple_loss=0.2788, pruned_loss=0.03748, over 16761.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.289, pruned_loss=0.05799, over 3079064.88 frames. ], batch size: 83, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:37:09,148 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 01:37:14,502 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251465.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:37:41,132 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 01:37:50,414 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251489.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:38:07,476 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251501.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 01:38:09,247 INFO [train.py:904] (7/8) Epoch 25, batch 7900, loss[loss=0.2206, simple_loss=0.303, pruned_loss=0.06914, over 16469.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2887, pruned_loss=0.05814, over 3065369.11 frames. ], batch size: 35, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:38:47,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3611, 3.1874, 3.5979, 1.8024, 3.6877, 3.7651, 2.9013, 2.7531], device='cuda:7'), covar=tensor([0.0837, 0.0315, 0.0193, 0.1291, 0.0089, 0.0208, 0.0443, 0.0510], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0139, 0.0084, 0.0130, 0.0129, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:38:57,946 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 01:39:04,227 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251537.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:39:17,278 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.008e+02 2.635e+02 3.162e+02 3.936e+02 7.872e+02, threshold=6.324e+02, percent-clipped=1.0 2023-05-02 01:39:22,146 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251549.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:39:28,557 INFO [train.py:904] (7/8) Epoch 25, batch 7950, loss[loss=0.2467, simple_loss=0.3089, pruned_loss=0.09227, over 11377.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2889, pruned_loss=0.05846, over 3071310.63 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:39:30,203 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251554.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:40:41,083 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3137, 2.8290, 3.0102, 2.0204, 2.7391, 2.0339, 3.0035, 3.1353], device='cuda:7'), covar=tensor([0.0312, 0.0880, 0.0646, 0.2136, 0.0932, 0.1133, 0.0721, 0.0941], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0167, 0.0170, 0.0155, 0.0147, 0.0131, 0.0145, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:40:46,600 INFO [train.py:904] (7/8) Epoch 25, batch 8000, loss[loss=0.1818, simple_loss=0.2783, pruned_loss=0.04263, over 16514.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2898, pruned_loss=0.05924, over 3057602.50 frames. ], batch size: 68, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:41:44,789 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251641.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:41:51,618 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.813e+02 2.777e+02 3.202e+02 3.994e+02 6.409e+02, threshold=6.404e+02, percent-clipped=2.0 2023-05-02 01:42:01,901 INFO [train.py:904] (7/8) Epoch 25, batch 8050, loss[loss=0.1886, simple_loss=0.2706, pruned_loss=0.05324, over 16677.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2896, pruned_loss=0.0586, over 3078135.14 frames. ], batch size: 57, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:42:09,018 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251657.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:42:44,803 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 01:42:53,435 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3488, 3.4711, 3.6145, 3.5998, 3.6086, 3.4214, 3.4719, 3.4929], device='cuda:7'), covar=tensor([0.0397, 0.0689, 0.0482, 0.0460, 0.0537, 0.0591, 0.0827, 0.0573], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0476, 0.0460, 0.0424, 0.0507, 0.0486, 0.0561, 0.0387], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 01:43:16,979 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:43:17,643 INFO [train.py:904] (7/8) Epoch 25, batch 8100, loss[loss=0.2142, simple_loss=0.299, pruned_loss=0.0647, over 15516.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2893, pruned_loss=0.05825, over 3075068.04 frames. ], batch size: 191, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:43:27,319 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=251709.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:43:56,137 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8711, 2.1685, 2.4507, 3.1035, 2.2357, 2.3339, 2.3509, 2.2682], device='cuda:7'), covar=tensor([0.1391, 0.3321, 0.2500, 0.0762, 0.4008, 0.2433, 0.3182, 0.3234], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0462, 0.0378, 0.0332, 0.0441, 0.0528, 0.0433, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:44:22,962 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.808e+02 3.404e+02 4.395e+02 1.324e+03, threshold=6.808e+02, percent-clipped=5.0 2023-05-02 01:44:33,804 INFO [train.py:904] (7/8) Epoch 25, batch 8150, loss[loss=0.2304, simple_loss=0.3004, pruned_loss=0.08026, over 11731.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2867, pruned_loss=0.05716, over 3065633.51 frames. ], batch size: 248, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:44:40,467 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1441, 2.1835, 2.6808, 3.0961, 2.9950, 3.6440, 2.2923, 3.5993], device='cuda:7'), covar=tensor([0.0238, 0.0545, 0.0402, 0.0312, 0.0331, 0.0156, 0.0588, 0.0135], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0195, 0.0183, 0.0185, 0.0201, 0.0161, 0.0199, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:44:45,088 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251760.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:45:00,575 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=251770.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:45:42,047 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9774, 4.2124, 4.0443, 4.0609, 3.8130, 3.8092, 3.8775, 4.2028], device='cuda:7'), covar=tensor([0.1056, 0.0856, 0.0960, 0.0857, 0.0712, 0.1761, 0.0966, 0.0973], device='cuda:7'), in_proj_covar=tensor([0.0692, 0.0836, 0.0691, 0.0647, 0.0529, 0.0537, 0.0701, 0.0655], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 01:45:50,918 INFO [train.py:904] (7/8) Epoch 25, batch 8200, loss[loss=0.1977, simple_loss=0.2884, pruned_loss=0.05353, over 16942.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2837, pruned_loss=0.05617, over 3075125.49 frames. ], batch size: 90, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:46:59,193 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.400e+02 2.618e+02 3.043e+02 3.668e+02 8.118e+02, threshold=6.086e+02, percent-clipped=2.0 2023-05-02 01:47:10,666 INFO [train.py:904] (7/8) Epoch 25, batch 8250, loss[loss=0.1704, simple_loss=0.2711, pruned_loss=0.03488, over 16938.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2826, pruned_loss=0.05361, over 3052289.40 frames. ], batch size: 116, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:47:12,498 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251854.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:47:22,072 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7634, 2.9701, 2.6056, 4.3118, 2.8174, 4.1103, 1.5594, 3.1289], device='cuda:7'), covar=tensor([0.1435, 0.0677, 0.1140, 0.0192, 0.0144, 0.0389, 0.1700, 0.0662], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0199, 0.0197, 0.0208, 0.0218, 0.0209, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:48:21,814 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.97 vs. limit=5.0 2023-05-02 01:48:29,887 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=251902.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:48:30,720 INFO [train.py:904] (7/8) Epoch 25, batch 8300, loss[loss=0.1838, simple_loss=0.2841, pruned_loss=0.04177, over 16895.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2801, pruned_loss=0.05046, over 3074928.48 frames. ], batch size: 109, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:49:41,164 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.151e+02 2.560e+02 3.132e+02 5.384e+02, threshold=5.121e+02, percent-clipped=0.0 2023-05-02 01:49:52,631 INFO [train.py:904] (7/8) Epoch 25, batch 8350, loss[loss=0.1718, simple_loss=0.2713, pruned_loss=0.03616, over 16323.00 frames. ], tot_loss[loss=0.1877, simple_loss=0.2783, pruned_loss=0.04849, over 3051453.67 frames. ], batch size: 146, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:49:59,581 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=251957.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:51:05,922 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=251997.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:51:17,315 INFO [train.py:904] (7/8) Epoch 25, batch 8400, loss[loss=0.1571, simple_loss=0.2524, pruned_loss=0.03085, over 15301.00 frames. ], tot_loss[loss=0.1848, simple_loss=0.276, pruned_loss=0.0468, over 3045539.05 frames. ], batch size: 191, lr: 2.68e-03, grad_scale: 8.0 2023-05-02 01:51:22,059 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252005.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:52:11,978 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9929, 3.7205, 4.0912, 2.3540, 4.2938, 4.3018, 3.4037, 3.4648], device='cuda:7'), covar=tensor([0.0627, 0.0228, 0.0201, 0.0975, 0.0069, 0.0161, 0.0339, 0.0345], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0109, 0.0098, 0.0137, 0.0083, 0.0127, 0.0127, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 01:52:27,820 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.253e+02 2.668e+02 3.204e+02 6.662e+02, threshold=5.337e+02, percent-clipped=3.0 2023-05-02 01:52:40,186 INFO [train.py:904] (7/8) Epoch 25, batch 8450, loss[loss=0.1621, simple_loss=0.2617, pruned_loss=0.03128, over 16191.00 frames. ], tot_loss[loss=0.1819, simple_loss=0.2739, pruned_loss=0.04493, over 3048786.97 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:52:49,130 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.26 vs. limit=5.0 2023-05-02 01:52:51,861 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=252060.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:52:59,913 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252065.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:53:00,300 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 01:53:51,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9707, 2.8216, 2.6706, 2.0006, 2.5440, 2.8096, 2.6769, 2.0185], device='cuda:7'), covar=tensor([0.0417, 0.0090, 0.0085, 0.0367, 0.0147, 0.0116, 0.0111, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0085, 0.0086, 0.0133, 0.0098, 0.0110, 0.0095, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 01:54:01,928 INFO [train.py:904] (7/8) Epoch 25, batch 8500, loss[loss=0.1595, simple_loss=0.2568, pruned_loss=0.03109, over 16711.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2705, pruned_loss=0.0428, over 3055815.11 frames. ], batch size: 83, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:54:11,581 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252108.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 01:54:33,528 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3989, 2.9723, 3.3305, 2.0526, 2.9780, 2.1523, 3.1650, 3.1975], device='cuda:7'), covar=tensor([0.0317, 0.0850, 0.0423, 0.2112, 0.0743, 0.1096, 0.0609, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0154, 0.0163, 0.0166, 0.0152, 0.0144, 0.0128, 0.0142, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:55:15,324 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.299e+02 2.680e+02 3.253e+02 6.153e+02, threshold=5.360e+02, percent-clipped=1.0 2023-05-02 01:55:28,848 INFO [train.py:904] (7/8) Epoch 25, batch 8550, loss[loss=0.1651, simple_loss=0.2542, pruned_loss=0.03801, over 11951.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2686, pruned_loss=0.04212, over 3040276.07 frames. ], batch size: 246, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:56:12,711 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6641, 2.6879, 1.9320, 2.7990, 2.1180, 2.8492, 2.1271, 2.4006], device='cuda:7'), covar=tensor([0.0296, 0.0350, 0.1268, 0.0320, 0.0661, 0.0472, 0.1209, 0.0576], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0176, 0.0193, 0.0164, 0.0174, 0.0214, 0.0201, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:56:24,208 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4973, 3.5584, 2.7226, 2.1125, 2.1835, 2.3206, 3.7370, 3.0808], device='cuda:7'), covar=tensor([0.3276, 0.0592, 0.2001, 0.3195, 0.3162, 0.2340, 0.0440, 0.1441], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0267, 0.0305, 0.0315, 0.0298, 0.0266, 0.0295, 0.0338], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 01:57:09,091 INFO [train.py:904] (7/8) Epoch 25, batch 8600, loss[loss=0.1884, simple_loss=0.2859, pruned_loss=0.04542, over 15282.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.269, pruned_loss=0.04128, over 3036785.33 frames. ], batch size: 191, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:58:34,360 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.200e+02 2.678e+02 3.278e+02 9.763e+02, threshold=5.356e+02, percent-clipped=4.0 2023-05-02 01:58:37,234 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0107, 3.0574, 1.7420, 3.2832, 2.2173, 3.2802, 1.9988, 2.5161], device='cuda:7'), covar=tensor([0.0334, 0.0445, 0.1918, 0.0281, 0.0974, 0.0601, 0.1672, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0176, 0.0193, 0.0164, 0.0175, 0.0214, 0.0201, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 01:58:48,813 INFO [train.py:904] (7/8) Epoch 25, batch 8650, loss[loss=0.157, simple_loss=0.2519, pruned_loss=0.031, over 12125.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2673, pruned_loss=0.04013, over 3020939.21 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 01:59:39,888 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1033, 3.2926, 3.1736, 2.2098, 3.0028, 3.2649, 3.1219, 1.9366], device='cuda:7'), covar=tensor([0.0554, 0.0063, 0.0066, 0.0426, 0.0111, 0.0092, 0.0084, 0.0608], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0085, 0.0086, 0.0133, 0.0098, 0.0109, 0.0095, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 02:00:24,136 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=252297.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:00:27,558 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-02 02:00:33,693 INFO [train.py:904] (7/8) Epoch 25, batch 8700, loss[loss=0.1767, simple_loss=0.272, pruned_loss=0.04063, over 16331.00 frames. ], tot_loss[loss=0.1715, simple_loss=0.2648, pruned_loss=0.03914, over 3029495.95 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:01:53,640 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252345.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:01:54,400 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.271e+02 2.533e+02 3.010e+02 6.763e+02, threshold=5.067e+02, percent-clipped=2.0 2023-05-02 02:02:09,363 INFO [train.py:904] (7/8) Epoch 25, batch 8750, loss[loss=0.1785, simple_loss=0.2833, pruned_loss=0.03688, over 16538.00 frames. ], tot_loss[loss=0.1708, simple_loss=0.2648, pruned_loss=0.03837, over 3041602.85 frames. ], batch size: 68, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:02:41,194 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=252365.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:03:39,611 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 02:04:01,520 INFO [train.py:904] (7/8) Epoch 25, batch 8800, loss[loss=0.1883, simple_loss=0.278, pruned_loss=0.04934, over 12781.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2636, pruned_loss=0.03727, over 3061539.85 frames. ], batch size: 246, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:04:22,162 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=252413.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:04:30,859 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252417.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:04:50,730 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-02 02:05:07,242 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:05:31,486 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.474e+02 2.215e+02 2.676e+02 3.061e+02 6.379e+02, threshold=5.352e+02, percent-clipped=4.0 2023-05-02 02:05:35,913 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0568, 3.4157, 3.7768, 2.0320, 3.0521, 2.3473, 3.5179, 3.6453], device='cuda:7'), covar=tensor([0.0285, 0.0834, 0.0518, 0.2192, 0.0822, 0.1054, 0.0658, 0.0932], device='cuda:7'), in_proj_covar=tensor([0.0153, 0.0162, 0.0165, 0.0151, 0.0143, 0.0128, 0.0141, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 02:05:46,051 INFO [train.py:904] (7/8) Epoch 25, batch 8850, loss[loss=0.1471, simple_loss=0.2413, pruned_loss=0.0265, over 12492.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2654, pruned_loss=0.03682, over 3038859.69 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:06:39,854 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252478.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:07:16,692 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252495.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:07:31,617 INFO [train.py:904] (7/8) Epoch 25, batch 8900, loss[loss=0.1811, simple_loss=0.2745, pruned_loss=0.04385, over 15337.00 frames. ], tot_loss[loss=0.1691, simple_loss=0.2658, pruned_loss=0.03621, over 3049767.91 frames. ], batch size: 191, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:07:57,296 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9974, 2.1061, 2.5222, 2.9681, 2.7493, 3.3563, 2.2476, 3.3163], device='cuda:7'), covar=tensor([0.0182, 0.0559, 0.0395, 0.0295, 0.0357, 0.0168, 0.0558, 0.0163], device='cuda:7'), in_proj_covar=tensor([0.0188, 0.0191, 0.0180, 0.0181, 0.0197, 0.0156, 0.0194, 0.0156], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:09:18,935 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.583e+02 2.220e+02 2.582e+02 3.143e+02 6.266e+02, threshold=5.163e+02, percent-clipped=2.0 2023-05-02 02:09:34,575 INFO [train.py:904] (7/8) Epoch 25, batch 8950, loss[loss=0.1578, simple_loss=0.253, pruned_loss=0.0313, over 16360.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2652, pruned_loss=0.03619, over 3070885.15 frames. ], batch size: 146, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:10:59,623 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252592.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:11:21,913 INFO [train.py:904] (7/8) Epoch 25, batch 9000, loss[loss=0.1555, simple_loss=0.2488, pruned_loss=0.03111, over 12201.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2624, pruned_loss=0.03526, over 3055661.85 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:11:21,913 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 02:11:29,871 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4240, 3.4614, 2.1651, 3.8243, 2.7237, 3.7444, 2.4285, 2.9207], device='cuda:7'), covar=tensor([0.0335, 0.0414, 0.1667, 0.0260, 0.0863, 0.0646, 0.1536, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0173, 0.0190, 0.0160, 0.0172, 0.0209, 0.0198, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 02:11:31,594 INFO [train.py:938] (7/8) Epoch 25, validation: loss=0.1442, simple_loss=0.248, pruned_loss=0.02014, over 944034.00 frames. 2023-05-02 02:11:31,595 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 02:11:39,443 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 02:11:47,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8773, 3.1086, 2.7175, 4.9193, 3.6680, 4.4631, 1.7055, 3.3190], device='cuda:7'), covar=tensor([0.1354, 0.0744, 0.1150, 0.0130, 0.0197, 0.0328, 0.1676, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0176, 0.0194, 0.0192, 0.0201, 0.0213, 0.0205, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 02:11:53,559 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:12:36,791 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6935, 4.8379, 4.9945, 4.8400, 4.8537, 5.3852, 4.9406, 4.6346], device='cuda:7'), covar=tensor([0.1138, 0.1940, 0.2038, 0.2035, 0.2469, 0.0999, 0.1398, 0.2442], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0592, 0.0656, 0.0488, 0.0645, 0.0683, 0.0510, 0.0652], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 02:13:01,733 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.125e+02 2.594e+02 3.127e+02 6.449e+02, threshold=5.187e+02, percent-clipped=3.0 2023-05-02 02:13:14,919 INFO [train.py:904] (7/8) Epoch 25, batch 9050, loss[loss=0.1569, simple_loss=0.2467, pruned_loss=0.03351, over 16945.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2632, pruned_loss=0.03573, over 3073375.48 frames. ], batch size: 109, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:13:16,708 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252653.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:13:47,205 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252668.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:13:56,995 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:14:58,391 INFO [train.py:904] (7/8) Epoch 25, batch 9100, loss[loss=0.1905, simple_loss=0.2883, pruned_loss=0.04632, over 16366.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2629, pruned_loss=0.0362, over 3082466.72 frames. ], batch size: 146, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:16:01,700 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:16:41,928 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.125e+02 2.555e+02 2.905e+02 5.383e+02, threshold=5.110e+02, percent-clipped=1.0 2023-05-02 02:16:57,786 INFO [train.py:904] (7/8) Epoch 25, batch 9150, loss[loss=0.1744, simple_loss=0.2624, pruned_loss=0.04321, over 12094.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2635, pruned_loss=0.03588, over 3073364.13 frames. ], batch size: 247, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:17:42,777 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:18:20,264 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252790.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:18:33,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3812, 4.3615, 4.1913, 3.5627, 4.3145, 1.7752, 4.0764, 3.9119], device='cuda:7'), covar=tensor([0.0113, 0.0096, 0.0218, 0.0243, 0.0100, 0.2768, 0.0134, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0162, 0.0200, 0.0176, 0.0178, 0.0208, 0.0189, 0.0170], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:18:43,286 INFO [train.py:904] (7/8) Epoch 25, batch 9200, loss[loss=0.1622, simple_loss=0.2578, pruned_loss=0.03326, over 16657.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2596, pruned_loss=0.03526, over 3067355.92 frames. ], batch size: 134, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:19:03,512 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.71 vs. limit=5.0 2023-05-02 02:20:05,204 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.337e+02 2.810e+02 3.410e+02 9.767e+02, threshold=5.620e+02, percent-clipped=4.0 2023-05-02 02:20:19,108 INFO [train.py:904] (7/8) Epoch 25, batch 9250, loss[loss=0.1352, simple_loss=0.2247, pruned_loss=0.02284, over 12576.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2585, pruned_loss=0.03491, over 3045802.49 frames. ], batch size: 247, lr: 2.67e-03, grad_scale: 16.0 2023-05-02 02:20:28,468 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252857.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:21:18,633 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5947, 3.6732, 3.4454, 3.0871, 3.2836, 3.5730, 3.3641, 3.3908], device='cuda:7'), covar=tensor([0.0608, 0.0628, 0.0295, 0.0263, 0.0502, 0.0475, 0.1226, 0.0474], device='cuda:7'), in_proj_covar=tensor([0.0292, 0.0436, 0.0339, 0.0342, 0.0341, 0.0393, 0.0234, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:22:11,615 INFO [train.py:904] (7/8) Epoch 25, batch 9300, loss[loss=0.1665, simple_loss=0.2581, pruned_loss=0.03741, over 15226.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2567, pruned_loss=0.03437, over 3018811.59 frames. ], batch size: 190, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:22:47,912 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=252918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:23:45,080 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 1.953e+02 2.173e+02 2.787e+02 5.118e+02, threshold=4.346e+02, percent-clipped=0.0 2023-05-02 02:23:47,894 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252948.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:23:55,325 INFO [train.py:904] (7/8) Epoch 25, batch 9350, loss[loss=0.1845, simple_loss=0.2624, pruned_loss=0.05331, over 12605.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2573, pruned_loss=0.03458, over 3023832.12 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:24:16,594 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-05-02 02:24:28,694 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-05-02 02:24:29,996 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=252969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:25:17,121 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=252993.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:25:25,273 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7386, 4.8289, 4.6157, 4.1953, 4.2698, 4.7001, 4.5527, 4.4368], device='cuda:7'), covar=tensor([0.0628, 0.0754, 0.0363, 0.0378, 0.1000, 0.0678, 0.0460, 0.0726], device='cuda:7'), in_proj_covar=tensor([0.0290, 0.0433, 0.0339, 0.0341, 0.0340, 0.0392, 0.0234, 0.0405], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:25:36,668 INFO [train.py:904] (7/8) Epoch 25, batch 9400, loss[loss=0.1645, simple_loss=0.2465, pruned_loss=0.04123, over 12898.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2573, pruned_loss=0.03416, over 3029208.67 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:26:19,777 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253024.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:27:05,792 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.088e+02 2.613e+02 3.190e+02 8.435e+02, threshold=5.225e+02, percent-clipped=5.0 2023-05-02 02:27:17,039 INFO [train.py:904] (7/8) Epoch 25, batch 9450, loss[loss=0.1665, simple_loss=0.2568, pruned_loss=0.03804, over 12527.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2585, pruned_loss=0.03417, over 3029048.17 frames. ], batch size: 250, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:27:19,793 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253054.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:27:57,793 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253073.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:28:31,241 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253090.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:28:54,666 INFO [train.py:904] (7/8) Epoch 25, batch 9500, loss[loss=0.1447, simple_loss=0.2369, pruned_loss=0.02625, over 12789.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2577, pruned_loss=0.0338, over 3040006.11 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:29:33,443 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253121.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:30:06,530 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253138.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:30:23,927 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6277, 3.6052, 3.5792, 2.8968, 3.4837, 2.0091, 3.3152, 2.9853], device='cuda:7'), covar=tensor([0.0138, 0.0142, 0.0179, 0.0196, 0.0097, 0.2461, 0.0121, 0.0272], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0161, 0.0199, 0.0174, 0.0176, 0.0206, 0.0187, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:30:26,754 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.108e+02 2.493e+02 2.917e+02 7.333e+02, threshold=4.986e+02, percent-clipped=1.0 2023-05-02 02:30:40,909 INFO [train.py:904] (7/8) Epoch 25, batch 9550, loss[loss=0.1783, simple_loss=0.2776, pruned_loss=0.03951, over 16757.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2571, pruned_loss=0.03373, over 3055418.09 frames. ], batch size: 124, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:31:00,477 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5534, 3.7248, 2.7713, 2.2059, 2.3162, 2.3619, 3.9867, 3.2210], device='cuda:7'), covar=tensor([0.3086, 0.0586, 0.1934, 0.3033, 0.2902, 0.2243, 0.0372, 0.1475], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0265, 0.0303, 0.0311, 0.0292, 0.0263, 0.0292, 0.0335], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 02:31:33,174 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9858, 2.2047, 2.3952, 3.2050, 2.2296, 2.4071, 2.3521, 2.2933], device='cuda:7'), covar=tensor([0.1228, 0.3708, 0.2823, 0.0767, 0.4564, 0.2706, 0.3609, 0.3899], device='cuda:7'), in_proj_covar=tensor([0.0401, 0.0451, 0.0371, 0.0322, 0.0432, 0.0513, 0.0423, 0.0526], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:32:18,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8157, 4.2971, 4.2361, 3.0487, 3.7531, 4.2711, 3.8555, 2.6525], device='cuda:7'), covar=tensor([0.0509, 0.0044, 0.0048, 0.0379, 0.0111, 0.0094, 0.0072, 0.0483], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0084, 0.0085, 0.0132, 0.0098, 0.0107, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 02:32:22,182 INFO [train.py:904] (7/8) Epoch 25, batch 9600, loss[loss=0.1431, simple_loss=0.2383, pruned_loss=0.02391, over 17146.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2586, pruned_loss=0.0344, over 3067662.68 frames. ], batch size: 49, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:32:43,096 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253213.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:33:05,787 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 02:33:29,390 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-02 02:33:55,686 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.388e+02 2.773e+02 3.482e+02 6.416e+02, threshold=5.546e+02, percent-clipped=8.0 2023-05-02 02:33:59,811 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253248.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:34:07,803 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0743, 3.2584, 3.1951, 2.2576, 2.9659, 3.2915, 3.1040, 1.9159], device='cuda:7'), covar=tensor([0.0535, 0.0055, 0.0069, 0.0411, 0.0117, 0.0079, 0.0099, 0.0563], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0084, 0.0085, 0.0132, 0.0098, 0.0108, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 02:34:10,809 INFO [train.py:904] (7/8) Epoch 25, batch 9650, loss[loss=0.1584, simple_loss=0.258, pruned_loss=0.02942, over 16867.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2608, pruned_loss=0.03447, over 3084968.39 frames. ], batch size: 90, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:34:52,672 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:35:46,111 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253296.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:36:00,322 INFO [train.py:904] (7/8) Epoch 25, batch 9700, loss[loss=0.1552, simple_loss=0.2419, pruned_loss=0.03429, over 12331.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2593, pruned_loss=0.03431, over 3064640.41 frames. ], batch size: 250, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:36:27,592 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:36:42,978 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253324.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:36:56,162 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253329.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:37:34,481 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.159e+02 2.493e+02 2.931e+02 5.551e+02, threshold=4.985e+02, percent-clipped=1.0 2023-05-02 02:37:38,189 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253349.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:37:44,443 INFO [train.py:904] (7/8) Epoch 25, batch 9750, loss[loss=0.1603, simple_loss=0.2644, pruned_loss=0.02811, over 16352.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2584, pruned_loss=0.03472, over 3050610.02 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:38:22,158 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253372.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:38:38,997 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253380.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:39:01,198 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253390.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:39:23,677 INFO [train.py:904] (7/8) Epoch 25, batch 9800, loss[loss=0.1744, simple_loss=0.2823, pruned_loss=0.03322, over 16235.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2599, pruned_loss=0.03398, over 3074866.55 frames. ], batch size: 165, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:39:32,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3770, 3.8544, 3.7827, 2.6503, 3.4479, 3.8541, 3.5131, 2.1947], device='cuda:7'), covar=tensor([0.0586, 0.0047, 0.0056, 0.0413, 0.0117, 0.0084, 0.0091, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0132, 0.0084, 0.0085, 0.0132, 0.0098, 0.0107, 0.0094, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 02:39:53,073 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253419.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:39:58,972 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1487, 3.2352, 1.8782, 3.4848, 2.3494, 3.4791, 2.1844, 2.6604], device='cuda:7'), covar=tensor([0.0337, 0.0435, 0.1889, 0.0300, 0.0960, 0.0622, 0.1584, 0.0843], device='cuda:7'), in_proj_covar=tensor([0.0167, 0.0172, 0.0191, 0.0160, 0.0172, 0.0208, 0.0198, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 02:40:07,759 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 02:40:38,480 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253441.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:40:55,716 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 1.999e+02 2.298e+02 2.745e+02 1.319e+03, threshold=4.597e+02, percent-clipped=1.0 2023-05-02 02:41:06,308 INFO [train.py:904] (7/8) Epoch 25, batch 9850, loss[loss=0.16, simple_loss=0.2543, pruned_loss=0.03287, over 15218.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2607, pruned_loss=0.0338, over 3072041.60 frames. ], batch size: 190, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:41:46,408 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.59 vs. limit=5.0 2023-05-02 02:42:04,006 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253480.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:43:00,293 INFO [train.py:904] (7/8) Epoch 25, batch 9900, loss[loss=0.1659, simple_loss=0.2649, pruned_loss=0.03343, over 15303.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2604, pruned_loss=0.03383, over 3047424.76 frames. ], batch size: 190, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:43:25,547 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253513.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:43:42,128 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253521.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:44:46,259 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.031e+02 2.358e+02 2.986e+02 6.654e+02, threshold=4.715e+02, percent-clipped=4.0 2023-05-02 02:44:59,743 INFO [train.py:904] (7/8) Epoch 25, batch 9950, loss[loss=0.1576, simple_loss=0.26, pruned_loss=0.02765, over 15443.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2627, pruned_loss=0.03437, over 3055512.69 frames. ], batch size: 190, lr: 2.67e-03, grad_scale: 4.0 2023-05-02 02:45:18,770 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253561.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:45:34,691 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253567.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:46:14,341 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253582.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:47:02,501 INFO [train.py:904] (7/8) Epoch 25, batch 10000, loss[loss=0.1572, simple_loss=0.2501, pruned_loss=0.0322, over 12672.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2611, pruned_loss=0.03389, over 3076223.83 frames. ], batch size: 248, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:47:48,929 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5254, 3.1391, 3.4026, 1.6896, 3.5164, 3.6722, 2.9736, 2.8442], device='cuda:7'), covar=tensor([0.0727, 0.0293, 0.0218, 0.1400, 0.0106, 0.0172, 0.0444, 0.0471], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0107, 0.0094, 0.0135, 0.0080, 0.0122, 0.0124, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 02:47:53,004 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253628.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:48:36,793 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.008e+02 2.354e+02 2.749e+02 5.133e+02, threshold=4.708e+02, percent-clipped=1.0 2023-05-02 02:48:39,828 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253649.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:48:41,815 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=253650.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:48:46,088 INFO [train.py:904] (7/8) Epoch 25, batch 10050, loss[loss=0.1756, simple_loss=0.2681, pruned_loss=0.04153, over 16924.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2615, pruned_loss=0.03401, over 3059982.06 frames. ], batch size: 109, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:49:48,569 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253685.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:50:11,422 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=253697.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:50:20,883 INFO [train.py:904] (7/8) Epoch 25, batch 10100, loss[loss=0.1527, simple_loss=0.247, pruned_loss=0.02922, over 16588.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2625, pruned_loss=0.03444, over 3076922.99 frames. ], batch size: 57, lr: 2.67e-03, grad_scale: 8.0 2023-05-02 02:50:35,704 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=253711.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:51:21,589 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253736.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:51:33,772 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.192e+02 2.668e+02 3.131e+02 7.050e+02, threshold=5.336e+02, percent-clipped=1.0 2023-05-02 02:52:07,294 INFO [train.py:904] (7/8) Epoch 26, batch 0, loss[loss=0.1746, simple_loss=0.2612, pruned_loss=0.04399, over 16308.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2612, pruned_loss=0.04399, over 16308.00 frames. ], batch size: 36, lr: 2.61e-03, grad_scale: 8.0 2023-05-02 02:52:07,294 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 02:52:14,683 INFO [train.py:938] (7/8) Epoch 26, validation: loss=0.1437, simple_loss=0.2472, pruned_loss=0.02009, over 944034.00 frames. 2023-05-02 02:52:14,684 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 02:52:45,671 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253775.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:53:23,661 INFO [train.py:904] (7/8) Epoch 26, batch 50, loss[loss=0.1735, simple_loss=0.2698, pruned_loss=0.03863, over 17069.00 frames. ], tot_loss[loss=0.1782, simple_loss=0.2647, pruned_loss=0.04586, over 756435.97 frames. ], batch size: 53, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:54:28,066 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.375e+02 2.827e+02 3.414e+02 5.395e+02, threshold=5.654e+02, percent-clipped=1.0 2023-05-02 02:54:30,988 INFO [train.py:904] (7/8) Epoch 26, batch 100, loss[loss=0.1728, simple_loss=0.2533, pruned_loss=0.0461, over 16783.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2638, pruned_loss=0.04605, over 1316062.66 frames. ], batch size: 102, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:54:51,349 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-02 02:55:03,774 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253877.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:55:39,325 INFO [train.py:904] (7/8) Epoch 26, batch 150, loss[loss=0.1666, simple_loss=0.2444, pruned_loss=0.04437, over 16686.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2609, pruned_loss=0.04425, over 1756702.42 frames. ], batch size: 89, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:56:08,346 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=253923.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:56:11,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7320, 3.8911, 2.5398, 4.3727, 3.0205, 4.2825, 2.5357, 3.1769], device='cuda:7'), covar=tensor([0.0324, 0.0429, 0.1651, 0.0346, 0.0888, 0.0692, 0.1635, 0.0849], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0175, 0.0194, 0.0165, 0.0176, 0.0213, 0.0202, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 02:56:46,628 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.138e+02 2.556e+02 3.055e+02 5.694e+02, threshold=5.112e+02, percent-clipped=1.0 2023-05-02 02:56:49,074 INFO [train.py:904] (7/8) Epoch 26, batch 200, loss[loss=0.1532, simple_loss=0.2426, pruned_loss=0.03192, over 16835.00 frames. ], tot_loss[loss=0.1735, simple_loss=0.26, pruned_loss=0.04351, over 2111363.40 frames. ], batch size: 42, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:57:00,882 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.42 vs. limit=5.0 2023-05-02 02:57:02,003 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8064, 4.2244, 3.0108, 2.3442, 2.6100, 2.4599, 4.5834, 3.4813], device='cuda:7'), covar=tensor([0.3072, 0.0592, 0.1933, 0.3075, 0.3075, 0.2250, 0.0403, 0.1580], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0267, 0.0307, 0.0315, 0.0295, 0.0266, 0.0296, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 02:57:34,441 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=253985.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:58:03,917 INFO [train.py:904] (7/8) Epoch 26, batch 250, loss[loss=0.1759, simple_loss=0.2534, pruned_loss=0.04917, over 16911.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2581, pruned_loss=0.04272, over 2389572.88 frames. ], batch size: 109, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:58:07,804 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=254006.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:58:42,531 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3614, 4.1464, 4.4228, 4.5438, 4.6625, 4.2507, 4.4936, 4.6384], device='cuda:7'), covar=tensor([0.1694, 0.1259, 0.1335, 0.0694, 0.0600, 0.1131, 0.2184, 0.0745], device='cuda:7'), in_proj_covar=tensor([0.0637, 0.0779, 0.0898, 0.0793, 0.0606, 0.0628, 0.0662, 0.0763], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 02:58:47,214 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254033.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:58:50,828 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254036.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:59:10,749 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.700e+02 2.097e+02 2.455e+02 3.037e+02 5.274e+02, threshold=4.910e+02, percent-clipped=1.0 2023-05-02 02:59:14,073 INFO [train.py:904] (7/8) Epoch 26, batch 300, loss[loss=0.1686, simple_loss=0.25, pruned_loss=0.04358, over 16289.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2562, pruned_loss=0.04139, over 2587584.28 frames. ], batch size: 165, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 02:59:45,548 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254075.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 02:59:57,804 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254084.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:00:23,594 INFO [train.py:904] (7/8) Epoch 26, batch 350, loss[loss=0.1284, simple_loss=0.2121, pruned_loss=0.0224, over 16939.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2541, pruned_loss=0.04015, over 2751605.30 frames. ], batch size: 41, lr: 2.61e-03, grad_scale: 1.0 2023-05-02 03:00:24,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1458, 2.1612, 2.7300, 3.1325, 2.8182, 3.5682, 2.3418, 3.5368], device='cuda:7'), covar=tensor([0.0300, 0.0656, 0.0417, 0.0379, 0.0431, 0.0240, 0.0613, 0.0232], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0195, 0.0183, 0.0185, 0.0202, 0.0159, 0.0198, 0.0159], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:00:52,069 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254123.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:01:30,802 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.013e+02 2.457e+02 3.049e+02 5.935e+02, threshold=4.915e+02, percent-clipped=1.0 2023-05-02 03:01:33,894 INFO [train.py:904] (7/8) Epoch 26, batch 400, loss[loss=0.173, simple_loss=0.2521, pruned_loss=0.0469, over 16402.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2528, pruned_loss=0.04009, over 2870771.71 frames. ], batch size: 146, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:02:08,518 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254177.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:02:14,792 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7989, 4.6042, 4.7083, 4.3535, 4.3585, 4.7895, 4.5541, 4.5279], device='cuda:7'), covar=tensor([0.0778, 0.0946, 0.0434, 0.0422, 0.1066, 0.0526, 0.0526, 0.0721], device='cuda:7'), in_proj_covar=tensor([0.0299, 0.0446, 0.0349, 0.0352, 0.0350, 0.0403, 0.0239, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:02:44,717 INFO [train.py:904] (7/8) Epoch 26, batch 450, loss[loss=0.1541, simple_loss=0.2296, pruned_loss=0.03929, over 16834.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2511, pruned_loss=0.03993, over 2976719.30 frames. ], batch size: 96, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:02:50,959 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 03:03:12,812 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254223.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:03:14,575 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 03:03:15,095 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254225.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:03:39,169 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 03:03:50,703 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.473e+02 2.133e+02 2.439e+02 2.881e+02 4.883e+02, threshold=4.879e+02, percent-clipped=0.0 2023-05-02 03:03:52,997 INFO [train.py:904] (7/8) Epoch 26, batch 500, loss[loss=0.133, simple_loss=0.2217, pruned_loss=0.02214, over 16784.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2497, pruned_loss=0.03875, over 3061532.15 frames. ], batch size: 39, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:03:53,472 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9155, 2.0041, 2.6094, 2.9562, 2.6428, 3.4396, 2.3250, 3.4351], device='cuda:7'), covar=tensor([0.0304, 0.0631, 0.0393, 0.0369, 0.0430, 0.0189, 0.0573, 0.0195], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0195, 0.0183, 0.0186, 0.0202, 0.0160, 0.0199, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:04:19,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254271.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:04:23,832 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3055, 4.3086, 4.6526, 4.6345, 4.6708, 4.3744, 4.3934, 4.2758], device='cuda:7'), covar=tensor([0.0409, 0.0723, 0.0429, 0.0440, 0.0540, 0.0474, 0.0807, 0.0669], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0475, 0.0463, 0.0424, 0.0509, 0.0487, 0.0560, 0.0388], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 03:04:41,864 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7660, 3.9483, 2.6652, 4.5649, 3.2064, 4.5188, 2.7501, 3.3635], device='cuda:7'), covar=tensor([0.0358, 0.0425, 0.1582, 0.0292, 0.0838, 0.0489, 0.1431, 0.0714], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0179, 0.0197, 0.0169, 0.0178, 0.0217, 0.0204, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:05:01,751 INFO [train.py:904] (7/8) Epoch 26, batch 550, loss[loss=0.1743, simple_loss=0.2548, pruned_loss=0.04691, over 16510.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2496, pruned_loss=0.03835, over 3130164.34 frames. ], batch size: 146, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:05:05,482 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=254306.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:06:08,402 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.164e+02 2.429e+02 2.784e+02 6.302e+02, threshold=4.858e+02, percent-clipped=1.0 2023-05-02 03:06:11,644 INFO [train.py:904] (7/8) Epoch 26, batch 600, loss[loss=0.1635, simple_loss=0.2552, pruned_loss=0.03592, over 16712.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2493, pruned_loss=0.03839, over 3181317.91 frames. ], batch size: 62, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:06:13,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=254354.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:06:56,069 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2023-05-02 03:07:21,739 INFO [train.py:904] (7/8) Epoch 26, batch 650, loss[loss=0.181, simple_loss=0.2738, pruned_loss=0.04406, over 16802.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2479, pruned_loss=0.03828, over 3203277.10 frames. ], batch size: 57, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:08:28,770 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.176e+02 2.565e+02 3.138e+02 6.256e+02, threshold=5.131e+02, percent-clipped=2.0 2023-05-02 03:08:30,997 INFO [train.py:904] (7/8) Epoch 26, batch 700, loss[loss=0.1724, simple_loss=0.2526, pruned_loss=0.04605, over 16835.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2483, pruned_loss=0.03846, over 3232468.35 frames. ], batch size: 96, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:09:41,472 INFO [train.py:904] (7/8) Epoch 26, batch 750, loss[loss=0.1745, simple_loss=0.2465, pruned_loss=0.05127, over 16680.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2479, pruned_loss=0.03846, over 3253060.75 frames. ], batch size: 89, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:10:27,474 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254536.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:10:29,904 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4477, 2.4019, 2.3248, 4.1273, 2.3231, 2.7803, 2.4312, 2.5545], device='cuda:7'), covar=tensor([0.1352, 0.3786, 0.3323, 0.0615, 0.4203, 0.2607, 0.3786, 0.3697], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0466, 0.0384, 0.0335, 0.0446, 0.0532, 0.0438, 0.0545], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:10:46,996 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.386e+02 2.194e+02 2.558e+02 3.127e+02 5.314e+02, threshold=5.115e+02, percent-clipped=1.0 2023-05-02 03:10:50,626 INFO [train.py:904] (7/8) Epoch 26, batch 800, loss[loss=0.1587, simple_loss=0.2566, pruned_loss=0.03038, over 16767.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2485, pruned_loss=0.03837, over 3270919.48 frames. ], batch size: 57, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:10:50,952 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2389, 5.7400, 5.8117, 5.4940, 5.6905, 6.2079, 5.7112, 5.4272], device='cuda:7'), covar=tensor([0.0921, 0.1963, 0.2663, 0.2124, 0.2427, 0.0904, 0.1488, 0.2166], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0622, 0.0691, 0.0512, 0.0677, 0.0713, 0.0535, 0.0681], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 03:11:20,522 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254574.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:11:51,475 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:11:59,179 INFO [train.py:904] (7/8) Epoch 26, batch 850, loss[loss=0.1549, simple_loss=0.2403, pruned_loss=0.03479, over 15485.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2478, pruned_loss=0.03768, over 3288352.30 frames. ], batch size: 191, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:12:45,978 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254635.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:13:00,357 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4886, 3.5651, 3.8048, 2.6637, 3.4896, 3.9290, 3.5898, 2.4451], device='cuda:7'), covar=tensor([0.0530, 0.0199, 0.0066, 0.0411, 0.0131, 0.0101, 0.0113, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0088, 0.0135, 0.0101, 0.0112, 0.0097, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 03:13:08,054 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.023e+02 2.316e+02 2.771e+02 4.972e+02, threshold=4.632e+02, percent-clipped=1.0 2023-05-02 03:13:10,212 INFO [train.py:904] (7/8) Epoch 26, batch 900, loss[loss=0.1936, simple_loss=0.2659, pruned_loss=0.06063, over 16847.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2465, pruned_loss=0.03721, over 3296052.64 frames. ], batch size: 90, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:13:49,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7610, 4.8640, 4.9948, 4.8416, 4.8353, 5.4739, 4.9825, 4.6439], device='cuda:7'), covar=tensor([0.1385, 0.2209, 0.2655, 0.2485, 0.2950, 0.1070, 0.1874, 0.2756], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0623, 0.0694, 0.0514, 0.0680, 0.0714, 0.0538, 0.0683], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 03:14:19,298 INFO [train.py:904] (7/8) Epoch 26, batch 950, loss[loss=0.1587, simple_loss=0.2389, pruned_loss=0.03923, over 16698.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.2471, pruned_loss=0.03729, over 3312496.36 frames. ], batch size: 134, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:14:24,875 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 03:14:46,163 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5541, 1.8494, 2.2377, 2.4096, 2.5296, 2.5178, 1.8891, 2.6609], device='cuda:7'), covar=tensor([0.0238, 0.0573, 0.0377, 0.0368, 0.0397, 0.0416, 0.0592, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0199, 0.0186, 0.0188, 0.0205, 0.0163, 0.0201, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:15:24,093 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.119e+02 2.483e+02 2.968e+02 5.109e+02, threshold=4.965e+02, percent-clipped=3.0 2023-05-02 03:15:27,114 INFO [train.py:904] (7/8) Epoch 26, batch 1000, loss[loss=0.138, simple_loss=0.2274, pruned_loss=0.02429, over 17221.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.2461, pruned_loss=0.03725, over 3311976.11 frames. ], batch size: 45, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:15:31,153 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254755.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:15:48,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6046, 3.7535, 4.3486, 2.4356, 3.4012, 2.6955, 3.9819, 3.9437], device='cuda:7'), covar=tensor([0.0280, 0.0979, 0.0439, 0.2016, 0.0822, 0.1007, 0.0651, 0.1021], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0167, 0.0169, 0.0156, 0.0148, 0.0131, 0.0146, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:16:06,352 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-02 03:16:28,403 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254797.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:16:35,910 INFO [train.py:904] (7/8) Epoch 26, batch 1050, loss[loss=0.1717, simple_loss=0.259, pruned_loss=0.04217, over 16573.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.245, pruned_loss=0.03701, over 3314468.20 frames. ], batch size: 146, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:16:55,237 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254816.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:17:21,651 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 03:17:43,019 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=254851.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:17:43,595 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.171e+02 2.721e+02 3.139e+02 5.089e+02, threshold=5.442e+02, percent-clipped=1.0 2023-05-02 03:17:44,778 INFO [train.py:904] (7/8) Epoch 26, batch 1100, loss[loss=0.1604, simple_loss=0.2401, pruned_loss=0.04032, over 16898.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2443, pruned_loss=0.03708, over 3314000.92 frames. ], batch size: 96, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:17:53,573 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254858.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:18:37,663 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=254892.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:18:53,040 INFO [train.py:904] (7/8) Epoch 26, batch 1150, loss[loss=0.1645, simple_loss=0.2624, pruned_loss=0.03327, over 17052.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2454, pruned_loss=0.03697, over 3317231.96 frames. ], batch size: 50, lr: 2.61e-03, grad_scale: 2.0 2023-05-02 03:19:05,049 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=254912.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:19:29,866 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=254930.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:19:42,424 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0716, 1.9897, 2.6798, 2.9813, 2.9070, 3.3195, 2.0836, 3.3767], device='cuda:7'), covar=tensor([0.0223, 0.0644, 0.0365, 0.0313, 0.0323, 0.0215, 0.0713, 0.0168], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0198, 0.0186, 0.0188, 0.0205, 0.0163, 0.0201, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:19:59,618 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.804e+02 2.093e+02 2.455e+02 3.027e+02 6.541e+02, threshold=4.911e+02, percent-clipped=2.0 2023-05-02 03:20:00,668 INFO [train.py:904] (7/8) Epoch 26, batch 1200, loss[loss=0.1715, simple_loss=0.2461, pruned_loss=0.04842, over 16932.00 frames. ], tot_loss[loss=0.1593, simple_loss=0.245, pruned_loss=0.03676, over 3316191.58 frames. ], batch size: 109, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:21:10,677 INFO [train.py:904] (7/8) Epoch 26, batch 1250, loss[loss=0.1392, simple_loss=0.2266, pruned_loss=0.02592, over 16847.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2448, pruned_loss=0.03678, over 3324649.23 frames. ], batch size: 42, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:21:13,212 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9092, 2.7342, 2.6501, 1.9637, 2.5687, 2.7404, 2.6195, 1.9887], device='cuda:7'), covar=tensor([0.0473, 0.0119, 0.0107, 0.0388, 0.0147, 0.0159, 0.0154, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0102, 0.0113, 0.0098, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 03:21:25,557 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8387, 2.0430, 2.4645, 2.6943, 2.7658, 2.7843, 2.1247, 2.9967], device='cuda:7'), covar=tensor([0.0218, 0.0545, 0.0366, 0.0332, 0.0360, 0.0319, 0.0550, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0199, 0.0187, 0.0189, 0.0205, 0.0163, 0.0201, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:22:19,976 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.021e+02 2.372e+02 2.873e+02 4.328e+02, threshold=4.744e+02, percent-clipped=0.0 2023-05-02 03:22:21,145 INFO [train.py:904] (7/8) Epoch 26, batch 1300, loss[loss=0.1518, simple_loss=0.2355, pruned_loss=0.03402, over 15445.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.245, pruned_loss=0.03712, over 3323380.42 frames. ], batch size: 190, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:23:30,846 INFO [train.py:904] (7/8) Epoch 26, batch 1350, loss[loss=0.1482, simple_loss=0.2546, pruned_loss=0.02094, over 17035.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.2462, pruned_loss=0.0367, over 3324393.89 frames. ], batch size: 50, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:23:42,891 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255111.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:24:38,901 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.137e+02 2.466e+02 2.927e+02 7.701e+02, threshold=4.932e+02, percent-clipped=1.0 2023-05-02 03:24:40,852 INFO [train.py:904] (7/8) Epoch 26, batch 1400, loss[loss=0.1559, simple_loss=0.2387, pruned_loss=0.03658, over 16396.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.246, pruned_loss=0.03707, over 3311814.69 frames. ], batch size: 36, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:24:41,768 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255153.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:25:34,111 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:25:48,429 INFO [train.py:904] (7/8) Epoch 26, batch 1450, loss[loss=0.16, simple_loss=0.2363, pruned_loss=0.04185, over 16757.00 frames. ], tot_loss[loss=0.1589, simple_loss=0.2446, pruned_loss=0.03662, over 3318650.70 frames. ], batch size: 124, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:25:54,194 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255207.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:26:14,515 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-02 03:26:25,723 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9261, 1.9814, 2.5185, 2.8837, 2.8162, 2.9592, 2.1782, 3.1023], device='cuda:7'), covar=tensor([0.0240, 0.0577, 0.0395, 0.0314, 0.0357, 0.0308, 0.0609, 0.0205], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0200, 0.0188, 0.0190, 0.0206, 0.0164, 0.0202, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:26:26,865 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255230.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:26:40,773 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255240.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:26:56,776 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.211e+02 2.586e+02 2.983e+02 9.031e+02, threshold=5.172e+02, percent-clipped=4.0 2023-05-02 03:26:57,923 INFO [train.py:904] (7/8) Epoch 26, batch 1500, loss[loss=0.178, simple_loss=0.2704, pruned_loss=0.04283, over 16990.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2445, pruned_loss=0.03691, over 3315512.35 frames. ], batch size: 55, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:27:31,539 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255278.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:27:36,902 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1649, 4.9163, 5.1390, 5.3454, 5.6152, 4.8650, 5.5550, 5.5766], device='cuda:7'), covar=tensor([0.1886, 0.1586, 0.2167, 0.0972, 0.0581, 0.0832, 0.0561, 0.0688], device='cuda:7'), in_proj_covar=tensor([0.0683, 0.0835, 0.0964, 0.0851, 0.0645, 0.0670, 0.0704, 0.0816], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:28:04,984 INFO [train.py:904] (7/8) Epoch 26, batch 1550, loss[loss=0.1447, simple_loss=0.2296, pruned_loss=0.02994, over 16990.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2453, pruned_loss=0.03804, over 3323350.89 frames. ], batch size: 41, lr: 2.61e-03, grad_scale: 4.0 2023-05-02 03:29:12,929 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.260e+02 2.709e+02 3.380e+02 1.198e+03, threshold=5.419e+02, percent-clipped=4.0 2023-05-02 03:29:14,111 INFO [train.py:904] (7/8) Epoch 26, batch 1600, loss[loss=0.1548, simple_loss=0.2538, pruned_loss=0.02794, over 17017.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2469, pruned_loss=0.03853, over 3320925.25 frames. ], batch size: 50, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:30:03,518 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=255388.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:30:12,121 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.63 vs. limit=5.0 2023-05-02 03:30:23,428 INFO [train.py:904] (7/8) Epoch 26, batch 1650, loss[loss=0.1658, simple_loss=0.2461, pruned_loss=0.04269, over 16510.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2485, pruned_loss=0.03893, over 3324284.08 frames. ], batch size: 75, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:30:35,805 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255411.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:31:10,129 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2039, 5.6947, 5.8752, 5.5461, 5.6903, 6.2486, 5.7179, 5.4391], device='cuda:7'), covar=tensor([0.0935, 0.1950, 0.2544, 0.2050, 0.2614, 0.0897, 0.1569, 0.2289], device='cuda:7'), in_proj_covar=tensor([0.0428, 0.0633, 0.0701, 0.0519, 0.0686, 0.0726, 0.0545, 0.0690], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 03:31:29,046 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=255449.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:31:32,751 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.657e+02 2.212e+02 2.578e+02 3.028e+02 5.571e+02, threshold=5.156e+02, percent-clipped=1.0 2023-05-02 03:31:33,342 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8871, 4.0456, 2.8345, 4.7293, 3.2531, 4.6621, 2.8155, 3.3328], device='cuda:7'), covar=tensor([0.0349, 0.0403, 0.1515, 0.0261, 0.0797, 0.0472, 0.1482, 0.0793], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0183, 0.0201, 0.0175, 0.0181, 0.0223, 0.0208, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:31:34,040 INFO [train.py:904] (7/8) Epoch 26, batch 1700, loss[loss=0.1513, simple_loss=0.2331, pruned_loss=0.03479, over 16966.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2491, pruned_loss=0.03866, over 3329693.46 frames. ], batch size: 41, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:31:34,422 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255453.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:31:43,230 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255459.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:32:13,981 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9157, 2.5934, 1.9972, 2.3261, 2.9339, 2.6746, 2.8904, 3.0510], device='cuda:7'), covar=tensor([0.0236, 0.0477, 0.0639, 0.0561, 0.0295, 0.0395, 0.0260, 0.0315], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0249, 0.0236, 0.0237, 0.0248, 0.0247, 0.0246, 0.0246], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:32:40,155 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255501.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:32:42,747 INFO [train.py:904] (7/8) Epoch 26, batch 1750, loss[loss=0.1412, simple_loss=0.2327, pruned_loss=0.02481, over 17024.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.251, pruned_loss=0.03894, over 3323180.28 frames. ], batch size: 41, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:32:47,964 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=255507.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:33:40,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2972, 2.8011, 3.1693, 1.9345, 3.2369, 3.2630, 2.7040, 2.5668], device='cuda:7'), covar=tensor([0.0791, 0.0336, 0.0236, 0.1107, 0.0149, 0.0269, 0.0505, 0.0468], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0113, 0.0101, 0.0141, 0.0086, 0.0132, 0.0131, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:33:49,127 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 2.153e+02 2.652e+02 3.222e+02 7.615e+02, threshold=5.303e+02, percent-clipped=5.0 2023-05-02 03:33:51,390 INFO [train.py:904] (7/8) Epoch 26, batch 1800, loss[loss=0.1848, simple_loss=0.2669, pruned_loss=0.05135, over 15473.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.253, pruned_loss=0.03974, over 3310805.41 frames. ], batch size: 191, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:33:54,880 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=255555.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:34:18,265 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2388, 5.2040, 4.9634, 4.4339, 5.0264, 1.9339, 4.7674, 4.7381], device='cuda:7'), covar=tensor([0.0104, 0.0085, 0.0236, 0.0422, 0.0113, 0.2992, 0.0162, 0.0287], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0172, 0.0209, 0.0185, 0.0187, 0.0218, 0.0199, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:34:59,035 INFO [train.py:904] (7/8) Epoch 26, batch 1850, loss[loss=0.1995, simple_loss=0.2876, pruned_loss=0.05573, over 12552.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.254, pruned_loss=0.03964, over 3316160.03 frames. ], batch size: 246, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:35:26,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1077, 3.1933, 3.4167, 2.2429, 2.9958, 2.3879, 3.5733, 3.5763], device='cuda:7'), covar=tensor([0.0232, 0.0870, 0.0626, 0.1906, 0.0824, 0.1057, 0.0531, 0.0870], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0169, 0.0171, 0.0158, 0.0148, 0.0133, 0.0147, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 03:35:46,045 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9772, 4.1019, 2.8608, 4.6752, 3.3877, 4.6202, 2.9858, 3.4286], device='cuda:7'), covar=tensor([0.0330, 0.0408, 0.1485, 0.0324, 0.0789, 0.0606, 0.1392, 0.0777], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0184, 0.0201, 0.0176, 0.0182, 0.0223, 0.0208, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:36:05,021 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.481e+02 2.077e+02 2.353e+02 2.960e+02 6.152e+02, threshold=4.707e+02, percent-clipped=2.0 2023-05-02 03:36:06,160 INFO [train.py:904] (7/8) Epoch 26, batch 1900, loss[loss=0.1661, simple_loss=0.2628, pruned_loss=0.03464, over 17052.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.253, pruned_loss=0.03905, over 3318103.60 frames. ], batch size: 50, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:36:46,203 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.33 vs. limit=5.0 2023-05-02 03:37:16,461 INFO [train.py:904] (7/8) Epoch 26, batch 1950, loss[loss=0.1841, simple_loss=0.2653, pruned_loss=0.0514, over 16687.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2514, pruned_loss=0.03845, over 3327368.88 frames. ], batch size: 124, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:37:59,941 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2488, 3.3173, 3.2424, 5.2552, 4.3988, 4.5515, 2.1632, 3.6189], device='cuda:7'), covar=tensor([0.1213, 0.0732, 0.1006, 0.0234, 0.0230, 0.0394, 0.1464, 0.0681], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0199, 0.0206, 0.0220, 0.0210, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:38:12,958 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=255744.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:38:24,266 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.052e+02 2.408e+02 2.894e+02 5.578e+02, threshold=4.816e+02, percent-clipped=3.0 2023-05-02 03:38:25,468 INFO [train.py:904] (7/8) Epoch 26, batch 2000, loss[loss=0.1733, simple_loss=0.2536, pruned_loss=0.04652, over 16496.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2513, pruned_loss=0.03856, over 3328368.91 frames. ], batch size: 146, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:38:31,697 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9181, 4.0374, 2.7837, 4.7757, 3.2862, 4.6737, 2.9337, 3.4371], device='cuda:7'), covar=tensor([0.0326, 0.0423, 0.1464, 0.0252, 0.0738, 0.0472, 0.1317, 0.0723], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0184, 0.0201, 0.0176, 0.0182, 0.0223, 0.0208, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 03:39:35,300 INFO [train.py:904] (7/8) Epoch 26, batch 2050, loss[loss=0.1572, simple_loss=0.2591, pruned_loss=0.02762, over 17299.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2509, pruned_loss=0.03828, over 3326087.67 frames. ], batch size: 52, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:39:53,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3001, 2.4405, 2.5114, 4.0734, 2.3320, 2.7647, 2.5005, 2.5336], device='cuda:7'), covar=tensor([0.1516, 0.3488, 0.3067, 0.0645, 0.4189, 0.2583, 0.3573, 0.3314], device='cuda:7'), in_proj_covar=tensor([0.0418, 0.0469, 0.0384, 0.0338, 0.0445, 0.0535, 0.0439, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:40:17,356 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9092, 4.3862, 3.2137, 2.3690, 2.7757, 2.6804, 4.8154, 3.6632], device='cuda:7'), covar=tensor([0.3046, 0.0650, 0.1825, 0.3053, 0.3053, 0.2232, 0.0357, 0.1519], device='cuda:7'), in_proj_covar=tensor([0.0333, 0.0275, 0.0312, 0.0323, 0.0305, 0.0273, 0.0303, 0.0350], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 03:40:44,484 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.146e+02 2.478e+02 2.962e+02 5.196e+02, threshold=4.956e+02, percent-clipped=1.0 2023-05-02 03:40:45,701 INFO [train.py:904] (7/8) Epoch 26, batch 2100, loss[loss=0.1325, simple_loss=0.2194, pruned_loss=0.0228, over 17207.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2517, pruned_loss=0.0385, over 3333684.59 frames. ], batch size: 44, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:41:54,474 INFO [train.py:904] (7/8) Epoch 26, batch 2150, loss[loss=0.1706, simple_loss=0.2651, pruned_loss=0.03803, over 16754.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2526, pruned_loss=0.03906, over 3334293.83 frames. ], batch size: 57, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:42:06,122 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 03:42:52,709 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9003, 4.8297, 4.7980, 4.4786, 4.5232, 4.8431, 4.6688, 4.5681], device='cuda:7'), covar=tensor([0.0641, 0.0844, 0.0319, 0.0325, 0.0850, 0.0442, 0.0475, 0.0662], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0474, 0.0369, 0.0374, 0.0372, 0.0427, 0.0254, 0.0446], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 03:43:04,643 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.196e+02 2.655e+02 3.354e+02 7.063e+02, threshold=5.310e+02, percent-clipped=7.0 2023-05-02 03:43:05,939 INFO [train.py:904] (7/8) Epoch 26, batch 2200, loss[loss=0.1405, simple_loss=0.2338, pruned_loss=0.02359, over 17209.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2531, pruned_loss=0.03896, over 3342738.47 frames. ], batch size: 45, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:43:11,234 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=255956.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:44:19,347 INFO [train.py:904] (7/8) Epoch 26, batch 2250, loss[loss=0.181, simple_loss=0.2784, pruned_loss=0.04177, over 17069.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2548, pruned_loss=0.03967, over 3326700.98 frames. ], batch size: 53, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:44:40,012 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256017.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:44:51,110 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-02 03:45:09,544 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256038.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:45:17,850 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256044.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:45:27,440 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.041e+02 2.398e+02 2.796e+02 4.429e+02, threshold=4.795e+02, percent-clipped=0.0 2023-05-02 03:45:29,159 INFO [train.py:904] (7/8) Epoch 26, batch 2300, loss[loss=0.1752, simple_loss=0.2539, pruned_loss=0.04828, over 16728.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2554, pruned_loss=0.04019, over 3323462.04 frames. ], batch size: 83, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:46:09,903 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256081.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:14,211 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256084.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:24,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256092.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:34,636 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256099.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:46:39,732 INFO [train.py:904] (7/8) Epoch 26, batch 2350, loss[loss=0.178, simple_loss=0.2652, pruned_loss=0.04541, over 16848.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.256, pruned_loss=0.04054, over 3321671.67 frames. ], batch size: 102, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:47:13,971 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9390, 2.8372, 2.5527, 2.7754, 3.1401, 2.9579, 3.4554, 3.4076], device='cuda:7'), covar=tensor([0.0164, 0.0473, 0.0526, 0.0471, 0.0323, 0.0451, 0.0308, 0.0314], device='cuda:7'), in_proj_covar=tensor([0.0232, 0.0249, 0.0237, 0.0238, 0.0250, 0.0248, 0.0248, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:47:19,449 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256132.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:47:32,107 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256142.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:47:37,182 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256145.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:47:42,501 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1248, 2.1500, 2.2718, 3.8809, 2.1586, 2.4930, 2.2652, 2.3338], device='cuda:7'), covar=tensor([0.1641, 0.4377, 0.3108, 0.0670, 0.4422, 0.2797, 0.4123, 0.3400], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0470, 0.0386, 0.0338, 0.0447, 0.0536, 0.0441, 0.0549], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:47:45,357 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.073e+02 2.391e+02 2.953e+02 7.650e+02, threshold=4.783e+02, percent-clipped=2.0 2023-05-02 03:47:46,531 INFO [train.py:904] (7/8) Epoch 26, batch 2400, loss[loss=0.1765, simple_loss=0.2741, pruned_loss=0.03946, over 17081.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.258, pruned_loss=0.04089, over 3329768.26 frames. ], batch size: 53, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:47:54,426 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 03:48:42,865 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256193.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 03:48:55,694 INFO [train.py:904] (7/8) Epoch 26, batch 2450, loss[loss=0.189, simple_loss=0.2731, pruned_loss=0.05245, over 16526.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2585, pruned_loss=0.04098, over 3329918.37 frames. ], batch size: 68, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:50:01,732 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.193e+02 2.630e+02 3.165e+02 5.747e+02, threshold=5.261e+02, percent-clipped=2.0 2023-05-02 03:50:03,665 INFO [train.py:904] (7/8) Epoch 26, batch 2500, loss[loss=0.2085, simple_loss=0.2875, pruned_loss=0.06477, over 16349.00 frames. ], tot_loss[loss=0.1702, simple_loss=0.2582, pruned_loss=0.04106, over 3320802.22 frames. ], batch size: 146, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:50:48,281 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256285.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:51:13,216 INFO [train.py:904] (7/8) Epoch 26, batch 2550, loss[loss=0.1998, simple_loss=0.2909, pruned_loss=0.05431, over 16679.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2579, pruned_loss=0.04093, over 3320918.64 frames. ], batch size: 57, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:51:26,256 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256312.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:52:14,110 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256346.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:52:22,558 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.565e+02 2.118e+02 2.541e+02 2.890e+02 5.508e+02, threshold=5.081e+02, percent-clipped=1.0 2023-05-02 03:52:23,668 INFO [train.py:904] (7/8) Epoch 26, batch 2600, loss[loss=0.1731, simple_loss=0.2585, pruned_loss=0.04391, over 16440.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2574, pruned_loss=0.04028, over 3312799.75 frames. ], batch size: 146, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:52:51,383 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256372.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:53:05,478 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1282, 5.7352, 5.8975, 5.5496, 5.6553, 6.2274, 5.7893, 5.5289], device='cuda:7'), covar=tensor([0.0997, 0.1952, 0.2339, 0.2015, 0.2647, 0.1024, 0.1266, 0.2095], device='cuda:7'), in_proj_covar=tensor([0.0434, 0.0638, 0.0705, 0.0522, 0.0692, 0.0727, 0.0547, 0.0697], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 03:53:21,510 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256394.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:53:33,353 INFO [train.py:904] (7/8) Epoch 26, batch 2650, loss[loss=0.1693, simple_loss=0.2538, pruned_loss=0.04235, over 16862.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2582, pruned_loss=0.04006, over 3316371.18 frames. ], batch size: 102, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:54:14,561 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256433.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:54:20,824 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256437.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:54:24,140 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256440.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:54:40,364 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.530e+02 2.157e+02 2.407e+02 2.834e+02 4.989e+02, threshold=4.815e+02, percent-clipped=0.0 2023-05-02 03:54:41,418 INFO [train.py:904] (7/8) Epoch 26, batch 2700, loss[loss=0.1662, simple_loss=0.2496, pruned_loss=0.0414, over 16358.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2584, pruned_loss=0.03996, over 3317048.15 frames. ], batch size: 165, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:55:06,555 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256471.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:55:28,874 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256488.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 03:55:48,629 INFO [train.py:904] (7/8) Epoch 26, batch 2750, loss[loss=0.159, simple_loss=0.2557, pruned_loss=0.03118, over 17114.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2578, pruned_loss=0.03917, over 3318397.21 frames. ], batch size: 47, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:56:30,073 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256532.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:56:33,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5157, 2.4433, 2.4011, 4.4128, 2.4022, 2.7708, 2.5249, 2.5731], device='cuda:7'), covar=tensor([0.1331, 0.3654, 0.3241, 0.0564, 0.4032, 0.2682, 0.3565, 0.3859], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0471, 0.0386, 0.0339, 0.0447, 0.0538, 0.0442, 0.0551], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:56:56,720 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.155e+02 2.522e+02 3.116e+02 8.730e+02, threshold=5.045e+02, percent-clipped=2.0 2023-05-02 03:56:58,714 INFO [train.py:904] (7/8) Epoch 26, batch 2800, loss[loss=0.1842, simple_loss=0.2625, pruned_loss=0.05298, over 16876.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2576, pruned_loss=0.03936, over 3311363.19 frames. ], batch size: 109, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:57:02,641 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0890, 4.6050, 4.5885, 3.3448, 3.9213, 4.5881, 4.1089, 2.8346], device='cuda:7'), covar=tensor([0.0500, 0.0079, 0.0049, 0.0379, 0.0148, 0.0094, 0.0095, 0.0478], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0101, 0.0113, 0.0098, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 03:58:07,642 INFO [train.py:904] (7/8) Epoch 26, batch 2850, loss[loss=0.1539, simple_loss=0.2435, pruned_loss=0.03216, over 17242.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.257, pruned_loss=0.03901, over 3316822.81 frames. ], batch size: 45, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:58:21,269 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256612.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:59:00,665 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256641.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:59:09,191 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0427, 2.2932, 2.7575, 3.0794, 2.9665, 3.5624, 2.5013, 3.5798], device='cuda:7'), covar=tensor([0.0267, 0.0538, 0.0372, 0.0357, 0.0345, 0.0223, 0.0502, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0199, 0.0188, 0.0191, 0.0206, 0.0166, 0.0203, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 03:59:15,040 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.116e+02 2.591e+02 2.981e+02 5.904e+02, threshold=5.182e+02, percent-clipped=1.0 2023-05-02 03:59:16,873 INFO [train.py:904] (7/8) Epoch 26, batch 2900, loss[loss=0.1511, simple_loss=0.2463, pruned_loss=0.02799, over 17042.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2562, pruned_loss=0.03946, over 3324127.09 frames. ], batch size: 53, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 03:59:27,994 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256660.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 03:59:29,232 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0994, 5.4823, 5.2187, 5.2552, 4.9740, 4.9229, 4.9134, 5.5634], device='cuda:7'), covar=tensor([0.1412, 0.0891, 0.1183, 0.0872, 0.0903, 0.0987, 0.1352, 0.0981], device='cuda:7'), in_proj_covar=tensor([0.0716, 0.0870, 0.0713, 0.0671, 0.0553, 0.0551, 0.0730, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:00:14,006 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256694.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:00:25,267 INFO [train.py:904] (7/8) Epoch 26, batch 2950, loss[loss=0.1271, simple_loss=0.2178, pruned_loss=0.01815, over 16990.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2559, pruned_loss=0.04023, over 3313390.89 frames. ], batch size: 41, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:00:33,226 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8718, 3.1894, 3.0630, 5.1534, 4.2436, 4.4839, 1.9545, 3.2181], device='cuda:7'), covar=tensor([0.1334, 0.0710, 0.1043, 0.0185, 0.0215, 0.0360, 0.1492, 0.0768], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0180, 0.0199, 0.0200, 0.0206, 0.0219, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 04:01:01,181 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256728.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:14,688 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256737.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:17,992 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256740.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:21,587 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256742.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:01:35,078 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.243e+02 2.622e+02 3.035e+02 4.998e+02, threshold=5.244e+02, percent-clipped=0.0 2023-05-02 04:01:35,093 INFO [train.py:904] (7/8) Epoch 26, batch 3000, loss[loss=0.1845, simple_loss=0.2645, pruned_loss=0.05228, over 16377.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.256, pruned_loss=0.04065, over 3319713.14 frames. ], batch size: 146, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:01:35,093 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 04:01:44,003 INFO [train.py:938] (7/8) Epoch 26, validation: loss=0.1339, simple_loss=0.2392, pruned_loss=0.01435, over 944034.00 frames. 2023-05-02 04:01:44,004 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 04:02:27,415 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256785.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:02:31,547 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256788.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:02:31,653 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256788.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:02:46,243 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 04:02:53,051 INFO [train.py:904] (7/8) Epoch 26, batch 3050, loss[loss=0.1752, simple_loss=0.273, pruned_loss=0.03872, over 16803.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.2553, pruned_loss=0.04078, over 3307642.43 frames. ], batch size: 57, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:03:18,754 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7920, 2.6585, 2.2805, 2.5024, 2.9673, 2.7753, 3.2713, 3.2060], device='cuda:7'), covar=tensor([0.0167, 0.0483, 0.0597, 0.0570, 0.0363, 0.0442, 0.0344, 0.0324], device='cuda:7'), in_proj_covar=tensor([0.0234, 0.0249, 0.0237, 0.0238, 0.0249, 0.0248, 0.0248, 0.0248], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:03:27,122 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=256827.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:03:27,220 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256827.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:03:38,905 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256836.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:03:39,438 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 04:04:02,823 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.111e+02 2.504e+02 3.018e+02 1.623e+03, threshold=5.009e+02, percent-clipped=3.0 2023-05-02 04:04:02,845 INFO [train.py:904] (7/8) Epoch 26, batch 3100, loss[loss=0.1731, simple_loss=0.2464, pruned_loss=0.04986, over 16699.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2556, pruned_loss=0.04092, over 3306152.94 frames. ], batch size: 134, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:04:47,248 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=256883.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:04:53,677 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256888.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:05:13,359 INFO [train.py:904] (7/8) Epoch 26, batch 3150, loss[loss=0.2061, simple_loss=0.2895, pruned_loss=0.06133, over 12384.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2548, pruned_loss=0.04079, over 3301803.06 frames. ], batch size: 246, lr: 2.60e-03, grad_scale: 4.0 2023-05-02 04:06:05,176 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=256941.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:06:09,390 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=256944.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:06:21,403 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.162e+02 2.494e+02 3.116e+02 1.108e+03, threshold=4.988e+02, percent-clipped=4.0 2023-05-02 04:06:21,418 INFO [train.py:904] (7/8) Epoch 26, batch 3200, loss[loss=0.1482, simple_loss=0.2373, pruned_loss=0.02959, over 16906.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2539, pruned_loss=0.04047, over 3307464.38 frames. ], batch size: 96, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:07:11,963 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=256989.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:07:12,421 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 04:07:29,906 INFO [train.py:904] (7/8) Epoch 26, batch 3250, loss[loss=0.1921, simple_loss=0.2753, pruned_loss=0.05441, over 15435.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2536, pruned_loss=0.04074, over 3306048.71 frames. ], batch size: 190, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:07:43,800 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257013.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:08:03,569 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257028.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:08:38,571 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.275e+02 2.633e+02 3.156e+02 7.653e+02, threshold=5.267e+02, percent-clipped=3.0 2023-05-02 04:08:38,591 INFO [train.py:904] (7/8) Epoch 26, batch 3300, loss[loss=0.1675, simple_loss=0.2569, pruned_loss=0.03909, over 16767.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2549, pruned_loss=0.04105, over 3305412.53 frames. ], batch size: 83, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:09:07,474 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257074.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:09:09,567 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257076.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:09:48,084 INFO [train.py:904] (7/8) Epoch 26, batch 3350, loss[loss=0.1485, simple_loss=0.2332, pruned_loss=0.03185, over 16991.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2549, pruned_loss=0.04057, over 3305336.96 frames. ], batch size: 41, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:10:20,802 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257127.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:10:27,637 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 04:10:46,890 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7906, 3.0026, 3.2308, 2.0942, 2.7852, 2.1798, 3.3656, 3.3266], device='cuda:7'), covar=tensor([0.0281, 0.0956, 0.0579, 0.2003, 0.0889, 0.1084, 0.0591, 0.0874], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0171, 0.0171, 0.0157, 0.0148, 0.0133, 0.0147, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 04:10:56,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.179e+02 2.494e+02 3.258e+02 7.595e+02, threshold=4.988e+02, percent-clipped=3.0 2023-05-02 04:10:56,602 INFO [train.py:904] (7/8) Epoch 26, batch 3400, loss[loss=0.1528, simple_loss=0.2566, pruned_loss=0.02451, over 17032.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2548, pruned_loss=0.04029, over 3314255.92 frames. ], batch size: 50, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:11:13,612 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257165.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:11:27,561 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257175.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:11:39,307 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257183.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 04:12:06,428 INFO [train.py:904] (7/8) Epoch 26, batch 3450, loss[loss=0.1575, simple_loss=0.2551, pruned_loss=0.02996, over 17127.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2537, pruned_loss=0.03953, over 3321885.79 frames. ], batch size: 47, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:12:33,997 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:12:39,896 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257226.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:12:57,445 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257239.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:12:59,809 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257241.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:13:16,460 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.059e+02 2.361e+02 2.793e+02 7.239e+02, threshold=4.722e+02, percent-clipped=1.0 2023-05-02 04:13:16,482 INFO [train.py:904] (7/8) Epoch 26, batch 3500, loss[loss=0.1537, simple_loss=0.2503, pruned_loss=0.02851, over 17124.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2527, pruned_loss=0.039, over 3309834.86 frames. ], batch size: 47, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:13:58,934 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257283.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:14:25,776 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257302.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:14:26,389 INFO [train.py:904] (7/8) Epoch 26, batch 3550, loss[loss=0.1711, simple_loss=0.2705, pruned_loss=0.03585, over 17009.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2509, pruned_loss=0.03817, over 3312146.29 frames. ], batch size: 55, lr: 2.60e-03, grad_scale: 8.0 2023-05-02 04:15:13,209 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9294, 5.3171, 5.0555, 5.0703, 4.8005, 4.8300, 4.7350, 5.4190], device='cuda:7'), covar=tensor([0.1417, 0.0933, 0.1070, 0.0933, 0.0891, 0.0952, 0.1329, 0.0851], device='cuda:7'), in_proj_covar=tensor([0.0725, 0.0883, 0.0720, 0.0681, 0.0559, 0.0558, 0.0739, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:15:34,872 INFO [train.py:904] (7/8) Epoch 26, batch 3600, loss[loss=0.1711, simple_loss=0.2436, pruned_loss=0.04928, over 16879.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2496, pruned_loss=0.03824, over 3302800.52 frames. ], batch size: 116, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:15:35,978 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.009e+02 2.326e+02 2.677e+02 5.900e+02, threshold=4.651e+02, percent-clipped=2.0 2023-05-02 04:15:58,539 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257369.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:16:47,824 INFO [train.py:904] (7/8) Epoch 26, batch 3650, loss[loss=0.189, simple_loss=0.2536, pruned_loss=0.06224, over 15645.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2485, pruned_loss=0.03868, over 3294274.78 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:17:30,094 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 04:17:36,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9620, 2.6895, 2.5442, 1.9990, 2.5663, 2.7708, 2.5864, 1.9626], device='cuda:7'), covar=tensor([0.0421, 0.0099, 0.0095, 0.0381, 0.0146, 0.0149, 0.0140, 0.0416], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0089, 0.0136, 0.0102, 0.0113, 0.0098, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 04:17:49,574 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 04:17:58,063 INFO [train.py:904] (7/8) Epoch 26, batch 3700, loss[loss=0.1622, simple_loss=0.2436, pruned_loss=0.04038, over 11324.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2473, pruned_loss=0.04015, over 3266832.78 frames. ], batch size: 248, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:17:59,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 2.076e+02 2.563e+02 3.144e+02 6.860e+02, threshold=5.126e+02, percent-clipped=4.0 2023-05-02 04:18:11,254 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-02 04:18:19,096 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257467.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:18:41,378 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257483.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:19:09,135 INFO [train.py:904] (7/8) Epoch 26, batch 3750, loss[loss=0.1673, simple_loss=0.2421, pruned_loss=0.04622, over 16956.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2476, pruned_loss=0.04133, over 3271134.05 frames. ], batch size: 109, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:19:34,090 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257521.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:19:44,963 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257528.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:19:49,559 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257531.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:20:02,012 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257539.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:20:20,038 INFO [train.py:904] (7/8) Epoch 26, batch 3800, loss[loss=0.2173, simple_loss=0.2906, pruned_loss=0.07199, over 12225.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2496, pruned_loss=0.04282, over 3273124.10 frames. ], batch size: 246, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:20:22,166 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.842e+02 2.272e+02 2.465e+02 2.848e+02 4.858e+02, threshold=4.931e+02, percent-clipped=0.0 2023-05-02 04:20:55,731 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257578.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:20:55,800 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257578.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:21:09,667 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257587.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:21:13,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3074, 2.5365, 2.4316, 4.1859, 2.3372, 2.8876, 2.5503, 2.6696], device='cuda:7'), covar=tensor([0.1335, 0.3177, 0.2753, 0.0458, 0.3705, 0.2164, 0.3159, 0.2763], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0472, 0.0386, 0.0339, 0.0447, 0.0539, 0.0443, 0.0552], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:21:22,679 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:21:30,549 INFO [train.py:904] (7/8) Epoch 26, batch 3850, loss[loss=0.1574, simple_loss=0.241, pruned_loss=0.03691, over 16873.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2496, pruned_loss=0.04343, over 3276643.24 frames. ], batch size: 116, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:22:06,862 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8727, 2.0160, 2.4154, 2.7479, 2.7995, 2.7844, 1.9705, 3.0622], device='cuda:7'), covar=tensor([0.0206, 0.0536, 0.0413, 0.0287, 0.0346, 0.0323, 0.0641, 0.0183], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0199, 0.0188, 0.0192, 0.0207, 0.0166, 0.0203, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:22:12,204 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 04:22:20,389 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257639.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:22:41,241 INFO [train.py:904] (7/8) Epoch 26, batch 3900, loss[loss=0.1595, simple_loss=0.2433, pruned_loss=0.03781, over 16885.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2498, pruned_loss=0.04389, over 3272983.48 frames. ], batch size: 116, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:22:42,472 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.694e+02 2.236e+02 2.511e+02 3.000e+02 4.952e+02, threshold=5.021e+02, percent-clipped=1.0 2023-05-02 04:23:04,621 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257669.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:23:51,424 INFO [train.py:904] (7/8) Epoch 26, batch 3950, loss[loss=0.17, simple_loss=0.2444, pruned_loss=0.04776, over 16488.00 frames. ], tot_loss[loss=0.1684, simple_loss=0.249, pruned_loss=0.04394, over 3285388.83 frames. ], batch size: 68, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:24:12,553 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257717.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:25:02,912 INFO [train.py:904] (7/8) Epoch 26, batch 4000, loss[loss=0.1776, simple_loss=0.2622, pruned_loss=0.04651, over 17111.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2492, pruned_loss=0.04444, over 3286378.54 frames. ], batch size: 47, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:25:03,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.551e+02 2.101e+02 2.446e+02 2.886e+02 5.630e+02, threshold=4.893e+02, percent-clipped=0.0 2023-05-02 04:26:13,228 INFO [train.py:904] (7/8) Epoch 26, batch 4050, loss[loss=0.1914, simple_loss=0.2731, pruned_loss=0.05483, over 11980.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2503, pruned_loss=0.04383, over 3280083.29 frames. ], batch size: 247, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:26:40,019 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257821.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:26:43,272 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257823.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 04:27:25,381 INFO [train.py:904] (7/8) Epoch 26, batch 4100, loss[loss=0.1829, simple_loss=0.2655, pruned_loss=0.05015, over 17005.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2529, pruned_loss=0.04395, over 3275786.61 frames. ], batch size: 50, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:27:26,544 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.240e+02 1.750e+02 2.045e+02 2.393e+02 4.321e+02, threshold=4.090e+02, percent-clipped=1.0 2023-05-02 04:27:35,064 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257859.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:27:49,184 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257869.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:02,722 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9227, 4.1780, 4.0189, 4.0580, 3.6964, 3.8324, 3.8035, 4.1726], device='cuda:7'), covar=tensor([0.1101, 0.0902, 0.0978, 0.0757, 0.0812, 0.1375, 0.0939, 0.0888], device='cuda:7'), in_proj_covar=tensor([0.0718, 0.0873, 0.0712, 0.0674, 0.0555, 0.0554, 0.0735, 0.0682], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:28:02,791 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257878.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:10,617 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257883.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:30,804 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=257897.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:28:39,717 INFO [train.py:904] (7/8) Epoch 26, batch 4150, loss[loss=0.1925, simple_loss=0.2905, pruned_loss=0.04719, over 16234.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2595, pruned_loss=0.04587, over 3246629.04 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:28:41,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0294, 5.4920, 5.6833, 5.3535, 5.4861, 6.0615, 5.4827, 5.1411], device='cuda:7'), covar=tensor([0.0886, 0.1719, 0.1816, 0.1805, 0.2246, 0.0823, 0.1334, 0.2155], device='cuda:7'), in_proj_covar=tensor([0.0427, 0.0631, 0.0696, 0.0516, 0.0683, 0.0722, 0.0541, 0.0688], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 04:29:07,253 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257920.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:15,921 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257926.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:28,859 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=257934.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 04:29:44,218 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=257944.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:29:45,356 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=257945.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:53,588 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=257950.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:29:56,815 INFO [train.py:904] (7/8) Epoch 26, batch 4200, loss[loss=0.2084, simple_loss=0.3034, pruned_loss=0.05668, over 16547.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2661, pruned_loss=0.04726, over 3222403.42 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:29:58,475 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.624e+02 2.083e+02 2.573e+02 2.968e+02 5.903e+02, threshold=5.147e+02, percent-clipped=5.0 2023-05-02 04:30:58,176 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 04:31:15,318 INFO [train.py:904] (7/8) Epoch 26, batch 4250, loss[loss=0.1664, simple_loss=0.2621, pruned_loss=0.03541, over 16252.00 frames. ], tot_loss[loss=0.1821, simple_loss=0.2696, pruned_loss=0.04735, over 3186001.67 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:31:29,225 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=258011.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:32:29,606 INFO [train.py:904] (7/8) Epoch 26, batch 4300, loss[loss=0.1855, simple_loss=0.2799, pruned_loss=0.04554, over 16808.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.271, pruned_loss=0.04705, over 3169072.03 frames. ], batch size: 124, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:32:31,426 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.134e+02 2.518e+02 2.964e+02 4.474e+02, threshold=5.035e+02, percent-clipped=0.0 2023-05-02 04:33:45,784 INFO [train.py:904] (7/8) Epoch 26, batch 4350, loss[loss=0.2134, simple_loss=0.2891, pruned_loss=0.06887, over 11705.00 frames. ], tot_loss[loss=0.1842, simple_loss=0.2733, pruned_loss=0.0475, over 3162879.91 frames. ], batch size: 246, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:34:15,502 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258123.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:34:58,990 INFO [train.py:904] (7/8) Epoch 26, batch 4400, loss[loss=0.1792, simple_loss=0.273, pruned_loss=0.04273, over 15498.00 frames. ], tot_loss[loss=0.1864, simple_loss=0.2755, pruned_loss=0.04862, over 3157060.23 frames. ], batch size: 191, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:35:00,103 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.313e+02 2.752e+02 3.228e+02 5.813e+02, threshold=5.503e+02, percent-clipped=1.0 2023-05-02 04:35:25,259 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258171.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:35:56,703 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1240, 2.3934, 2.3966, 3.9694, 2.1026, 2.6955, 2.4730, 2.4444], device='cuda:7'), covar=tensor([0.1483, 0.3453, 0.2850, 0.0554, 0.4353, 0.2332, 0.3176, 0.3379], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0471, 0.0384, 0.0337, 0.0447, 0.0539, 0.0441, 0.0551], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:36:09,586 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1306, 2.0690, 2.5791, 2.9970, 2.9472, 3.5202, 2.1543, 3.5365], device='cuda:7'), covar=tensor([0.0246, 0.0594, 0.0379, 0.0326, 0.0332, 0.0179, 0.0635, 0.0160], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0199, 0.0188, 0.0192, 0.0207, 0.0165, 0.0204, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:36:11,040 INFO [train.py:904] (7/8) Epoch 26, batch 4450, loss[loss=0.1955, simple_loss=0.2961, pruned_loss=0.04749, over 17256.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.279, pruned_loss=0.04999, over 3160165.00 frames. ], batch size: 52, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:36:24,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7645, 2.8459, 2.6281, 4.4306, 3.3864, 4.0167, 1.6847, 3.1355], device='cuda:7'), covar=tensor([0.1282, 0.0751, 0.1141, 0.0148, 0.0228, 0.0355, 0.1581, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0180, 0.0198, 0.0199, 0.0207, 0.0218, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 04:36:28,237 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:36:47,130 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8671, 5.1137, 4.9601, 4.9725, 4.7020, 4.6110, 4.5428, 5.2247], device='cuda:7'), covar=tensor([0.1148, 0.0779, 0.0901, 0.0761, 0.0735, 0.0930, 0.1135, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0706, 0.0856, 0.0699, 0.0661, 0.0545, 0.0544, 0.0718, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:36:55,567 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258234.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:37:03,964 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258239.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 04:37:05,323 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2371, 5.2637, 5.1241, 4.7945, 4.8366, 5.1832, 4.9751, 4.8471], device='cuda:7'), covar=tensor([0.0465, 0.0248, 0.0196, 0.0233, 0.0674, 0.0265, 0.0292, 0.0468], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0459, 0.0360, 0.0363, 0.0362, 0.0416, 0.0246, 0.0434], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:37:22,666 INFO [train.py:904] (7/8) Epoch 26, batch 4500, loss[loss=0.1872, simple_loss=0.2704, pruned_loss=0.05195, over 16982.00 frames. ], tot_loss[loss=0.1905, simple_loss=0.2795, pruned_loss=0.0507, over 3177554.97 frames. ], batch size: 109, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:37:23,852 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 1.900e+02 2.188e+02 2.602e+02 4.588e+02, threshold=4.376e+02, percent-clipped=0.0 2023-05-02 04:38:05,648 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258282.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 04:38:35,280 INFO [train.py:904] (7/8) Epoch 26, batch 4550, loss[loss=0.2285, simple_loss=0.3137, pruned_loss=0.07166, over 16406.00 frames. ], tot_loss[loss=0.1922, simple_loss=0.2806, pruned_loss=0.05194, over 3182330.35 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:38:39,795 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258306.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:39:48,245 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-02 04:39:48,627 INFO [train.py:904] (7/8) Epoch 26, batch 4600, loss[loss=0.1878, simple_loss=0.2746, pruned_loss=0.0505, over 16249.00 frames. ], tot_loss[loss=0.1936, simple_loss=0.2821, pruned_loss=0.0525, over 3186559.84 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:39:50,256 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 1.782e+02 2.088e+02 2.418e+02 3.840e+02, threshold=4.176e+02, percent-clipped=0.0 2023-05-02 04:39:51,498 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0512, 2.2887, 2.2365, 3.6192, 2.1132, 2.5980, 2.3754, 2.3876], device='cuda:7'), covar=tensor([0.1494, 0.3355, 0.3079, 0.0672, 0.4291, 0.2412, 0.3265, 0.3438], device='cuda:7'), in_proj_covar=tensor([0.0419, 0.0469, 0.0382, 0.0336, 0.0445, 0.0537, 0.0440, 0.0548], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:41:03,102 INFO [train.py:904] (7/8) Epoch 26, batch 4650, loss[loss=0.1825, simple_loss=0.2702, pruned_loss=0.04737, over 16888.00 frames. ], tot_loss[loss=0.193, simple_loss=0.2813, pruned_loss=0.05235, over 3206571.07 frames. ], batch size: 102, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:41:39,277 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9691, 5.4345, 5.6678, 5.3215, 5.4429, 6.0138, 5.4590, 5.1683], device='cuda:7'), covar=tensor([0.0998, 0.1635, 0.2033, 0.1760, 0.2189, 0.0756, 0.1303, 0.2127], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0624, 0.0685, 0.0509, 0.0674, 0.0714, 0.0532, 0.0679], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 04:42:14,228 INFO [train.py:904] (7/8) Epoch 26, batch 4700, loss[loss=0.1763, simple_loss=0.2653, pruned_loss=0.04368, over 16830.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2778, pruned_loss=0.0506, over 3211534.48 frames. ], batch size: 116, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:42:16,008 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.737e+02 2.039e+02 2.431e+02 7.214e+02, threshold=4.078e+02, percent-clipped=1.0 2023-05-02 04:42:17,173 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=258454.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:42:39,344 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4144, 2.0723, 1.7293, 1.9046, 2.3732, 2.0651, 1.9564, 2.4759], device='cuda:7'), covar=tensor([0.0210, 0.0499, 0.0661, 0.0568, 0.0290, 0.0473, 0.0202, 0.0340], device='cuda:7'), in_proj_covar=tensor([0.0230, 0.0245, 0.0234, 0.0235, 0.0245, 0.0245, 0.0244, 0.0245], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:43:06,446 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4190, 3.6080, 3.2774, 3.0065, 2.9705, 3.4816, 3.2549, 3.2635], device='cuda:7'), covar=tensor([0.0742, 0.0824, 0.0452, 0.0452, 0.0972, 0.0564, 0.1851, 0.0585], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0457, 0.0358, 0.0361, 0.0359, 0.0414, 0.0244, 0.0431], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:43:26,801 INFO [train.py:904] (7/8) Epoch 26, batch 4750, loss[loss=0.1942, simple_loss=0.2753, pruned_loss=0.0565, over 11890.00 frames. ], tot_loss[loss=0.1856, simple_loss=0.274, pruned_loss=0.04858, over 3207402.07 frames. ], batch size: 247, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:43:39,673 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=258511.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:43:45,133 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258515.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:43:45,177 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=258515.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:44:21,187 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258539.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:44:41,665 INFO [train.py:904] (7/8) Epoch 26, batch 4800, loss[loss=0.1941, simple_loss=0.295, pruned_loss=0.04661, over 16372.00 frames. ], tot_loss[loss=0.1825, simple_loss=0.2708, pruned_loss=0.04706, over 3192869.74 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:44:43,329 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 1.748e+02 2.158e+02 2.549e+02 7.061e+02, threshold=4.316e+02, percent-clipped=3.0 2023-05-02 04:44:57,187 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258563.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:45:11,429 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=258572.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:45:34,406 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258587.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:45:45,033 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6163, 4.6445, 4.9602, 4.9376, 4.9434, 4.6529, 4.6157, 4.4637], device='cuda:7'), covar=tensor([0.0288, 0.0501, 0.0336, 0.0359, 0.0467, 0.0355, 0.0997, 0.0492], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0474, 0.0461, 0.0423, 0.0508, 0.0485, 0.0564, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 04:45:57,768 INFO [train.py:904] (7/8) Epoch 26, batch 4850, loss[loss=0.187, simple_loss=0.2849, pruned_loss=0.04452, over 16335.00 frames. ], tot_loss[loss=0.1809, simple_loss=0.2707, pruned_loss=0.04556, over 3201898.49 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:46:03,347 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=258606.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:47:14,837 INFO [train.py:904] (7/8) Epoch 26, batch 4900, loss[loss=0.1605, simple_loss=0.2505, pruned_loss=0.0352, over 17121.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2699, pruned_loss=0.04454, over 3183307.41 frames. ], batch size: 49, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:47:16,710 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 1.994e+02 2.195e+02 2.536e+02 4.973e+02, threshold=4.389e+02, percent-clipped=2.0 2023-05-02 04:47:17,122 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=258654.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:47:28,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1205, 3.1605, 1.8050, 3.4299, 2.2975, 3.4696, 2.0215, 2.6012], device='cuda:7'), covar=tensor([0.0317, 0.0380, 0.1788, 0.0196, 0.0910, 0.0516, 0.1662, 0.0818], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0180, 0.0196, 0.0170, 0.0179, 0.0219, 0.0204, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 04:47:44,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0275, 3.3688, 3.5196, 1.9901, 2.9627, 2.2915, 3.5220, 3.6074], device='cuda:7'), covar=tensor([0.0246, 0.0743, 0.0603, 0.2169, 0.0888, 0.0958, 0.0629, 0.0805], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0169, 0.0170, 0.0156, 0.0147, 0.0131, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 04:48:29,888 INFO [train.py:904] (7/8) Epoch 26, batch 4950, loss[loss=0.1983, simple_loss=0.2909, pruned_loss=0.0529, over 16219.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2695, pruned_loss=0.04416, over 3181539.87 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:49:41,075 INFO [train.py:904] (7/8) Epoch 26, batch 5000, loss[loss=0.1776, simple_loss=0.2627, pruned_loss=0.04624, over 16436.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2716, pruned_loss=0.04433, over 3193280.80 frames. ], batch size: 35, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:49:42,192 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.044e+02 2.396e+02 2.809e+02 5.347e+02, threshold=4.792e+02, percent-clipped=3.0 2023-05-02 04:50:54,499 INFO [train.py:904] (7/8) Epoch 26, batch 5050, loss[loss=0.1862, simple_loss=0.282, pruned_loss=0.04517, over 16396.00 frames. ], tot_loss[loss=0.1808, simple_loss=0.2727, pruned_loss=0.0445, over 3192325.02 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:51:05,559 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258810.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:52:01,349 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6615, 2.9313, 3.2003, 1.9481, 2.8002, 2.1438, 3.2127, 3.1653], device='cuda:7'), covar=tensor([0.0247, 0.0833, 0.0653, 0.2107, 0.0882, 0.0998, 0.0633, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0170, 0.0171, 0.0157, 0.0148, 0.0132, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 04:52:07,192 INFO [train.py:904] (7/8) Epoch 26, batch 5100, loss[loss=0.1566, simple_loss=0.2392, pruned_loss=0.03695, over 17002.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2707, pruned_loss=0.0436, over 3197784.64 frames. ], batch size: 55, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:52:08,932 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.999e+02 2.336e+02 2.714e+02 4.372e+02, threshold=4.672e+02, percent-clipped=0.0 2023-05-02 04:52:27,837 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=258867.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:53:21,031 INFO [train.py:904] (7/8) Epoch 26, batch 5150, loss[loss=0.1677, simple_loss=0.2585, pruned_loss=0.03842, over 16998.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.2706, pruned_loss=0.04282, over 3201613.64 frames. ], batch size: 55, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:53:43,896 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2132, 2.2380, 2.7006, 3.1576, 3.0568, 3.6780, 2.3290, 3.6285], device='cuda:7'), covar=tensor([0.0228, 0.0568, 0.0365, 0.0335, 0.0336, 0.0152, 0.0610, 0.0154], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0199, 0.0187, 0.0192, 0.0207, 0.0164, 0.0204, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 04:54:35,100 INFO [train.py:904] (7/8) Epoch 26, batch 5200, loss[loss=0.1461, simple_loss=0.2402, pruned_loss=0.02603, over 16748.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2693, pruned_loss=0.04235, over 3194183.80 frames. ], batch size: 83, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:54:36,752 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 1.846e+02 2.126e+02 2.536e+02 3.966e+02, threshold=4.251e+02, percent-clipped=0.0 2023-05-02 04:54:53,673 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=258965.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:55:48,696 INFO [train.py:904] (7/8) Epoch 26, batch 5250, loss[loss=0.1875, simple_loss=0.277, pruned_loss=0.049, over 12186.00 frames. ], tot_loss[loss=0.1756, simple_loss=0.2669, pruned_loss=0.0422, over 3192329.13 frames. ], batch size: 248, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:56:23,075 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259026.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:56:46,156 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.61 vs. limit=5.0 2023-05-02 04:57:03,188 INFO [train.py:904] (7/8) Epoch 26, batch 5300, loss[loss=0.1742, simple_loss=0.2593, pruned_loss=0.04449, over 15392.00 frames. ], tot_loss[loss=0.1733, simple_loss=0.2637, pruned_loss=0.04139, over 3197561.81 frames. ], batch size: 190, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:57:04,413 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 1.887e+02 2.180e+02 2.580e+02 4.921e+02, threshold=4.360e+02, percent-clipped=4.0 2023-05-02 04:58:18,027 INFO [train.py:904] (7/8) Epoch 26, batch 5350, loss[loss=0.1813, simple_loss=0.2696, pruned_loss=0.04648, over 16741.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2622, pruned_loss=0.04091, over 3208365.35 frames. ], batch size: 76, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:58:27,891 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259110.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:58:34,304 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.86 vs. limit=5.0 2023-05-02 04:58:47,725 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4817, 3.6496, 2.3048, 4.1732, 2.7816, 4.1079, 2.4124, 2.9763], device='cuda:7'), covar=tensor([0.0351, 0.0356, 0.1688, 0.0172, 0.0911, 0.0520, 0.1615, 0.0804], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0179, 0.0195, 0.0169, 0.0178, 0.0217, 0.0204, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 04:59:10,699 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 04:59:31,435 INFO [train.py:904] (7/8) Epoch 26, batch 5400, loss[loss=0.1864, simple_loss=0.2813, pruned_loss=0.04578, over 16437.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2647, pruned_loss=0.04131, over 3199399.03 frames. ], batch size: 146, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 04:59:32,577 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 1.944e+02 2.298e+02 2.661e+02 4.475e+02, threshold=4.595e+02, percent-clipped=1.0 2023-05-02 04:59:38,449 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259158.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 04:59:53,862 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259167.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:00:48,653 INFO [train.py:904] (7/8) Epoch 26, batch 5450, loss[loss=0.202, simple_loss=0.2912, pruned_loss=0.05641, over 16272.00 frames. ], tot_loss[loss=0.1766, simple_loss=0.2676, pruned_loss=0.04283, over 3206733.72 frames. ], batch size: 35, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 05:01:08,445 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:01:25,094 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259226.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:02:05,217 INFO [train.py:904] (7/8) Epoch 26, batch 5500, loss[loss=0.2439, simple_loss=0.317, pruned_loss=0.08538, over 11695.00 frames. ], tot_loss[loss=0.1837, simple_loss=0.2744, pruned_loss=0.0465, over 3196182.52 frames. ], batch size: 248, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 05:02:06,405 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7853, 3.2095, 3.2361, 1.9866, 2.8438, 2.1206, 3.3051, 3.4723], device='cuda:7'), covar=tensor([0.0294, 0.0769, 0.0654, 0.2203, 0.0886, 0.1061, 0.0674, 0.0885], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0170, 0.0172, 0.0157, 0.0149, 0.0132, 0.0147, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 05:02:07,103 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.367e+02 2.972e+02 3.909e+02 6.600e+02, threshold=5.944e+02, percent-clipped=13.0 2023-05-02 05:02:23,201 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259264.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:02:39,754 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 05:02:58,280 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259287.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:03:22,303 INFO [train.py:904] (7/8) Epoch 26, batch 5550, loss[loss=0.2036, simple_loss=0.286, pruned_loss=0.06064, over 16207.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2813, pruned_loss=0.05193, over 3143423.09 frames. ], batch size: 165, lr: 2.59e-03, grad_scale: 8.0 2023-05-02 05:03:50,600 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259321.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:03:57,854 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259325.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:04:40,422 INFO [train.py:904] (7/8) Epoch 26, batch 5600, loss[loss=0.2623, simple_loss=0.3286, pruned_loss=0.09795, over 11408.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2863, pruned_loss=0.05646, over 3099027.65 frames. ], batch size: 247, lr: 2.58e-03, grad_scale: 16.0 2023-05-02 05:04:41,809 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.957e+02 3.012e+02 3.705e+02 4.647e+02 1.087e+03, threshold=7.410e+02, percent-clipped=8.0 2023-05-02 05:06:04,636 INFO [train.py:904] (7/8) Epoch 26, batch 5650, loss[loss=0.1907, simple_loss=0.2803, pruned_loss=0.05056, over 16700.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2909, pruned_loss=0.06014, over 3087455.90 frames. ], batch size: 89, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:06:22,292 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259414.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:07:22,189 INFO [train.py:904] (7/8) Epoch 26, batch 5700, loss[loss=0.2645, simple_loss=0.3175, pruned_loss=0.1058, over 11316.00 frames. ], tot_loss[loss=0.2065, simple_loss=0.2911, pruned_loss=0.06093, over 3091961.28 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:07:25,095 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.978e+02 3.137e+02 3.722e+02 4.714e+02 8.155e+02, threshold=7.444e+02, percent-clipped=2.0 2023-05-02 05:07:56,976 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259475.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:08:12,062 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1793, 5.4279, 5.2121, 5.2293, 4.9784, 4.8346, 4.8836, 5.5747], device='cuda:7'), covar=tensor([0.1179, 0.0833, 0.1062, 0.0966, 0.0768, 0.0915, 0.1275, 0.0801], device='cuda:7'), in_proj_covar=tensor([0.0695, 0.0839, 0.0690, 0.0647, 0.0532, 0.0532, 0.0707, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:08:39,974 INFO [train.py:904] (7/8) Epoch 26, batch 5750, loss[loss=0.2585, simple_loss=0.3165, pruned_loss=0.1003, over 11232.00 frames. ], tot_loss[loss=0.2097, simple_loss=0.2941, pruned_loss=0.06265, over 3058884.90 frames. ], batch size: 247, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:08:49,177 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6592, 4.0303, 2.6031, 2.3745, 2.6358, 2.4132, 4.1662, 3.3657], device='cuda:7'), covar=tensor([0.3171, 0.0668, 0.2245, 0.2738, 0.2739, 0.2266, 0.0591, 0.1378], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0271, 0.0308, 0.0320, 0.0301, 0.0269, 0.0300, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 05:08:54,685 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0431, 3.3398, 3.3507, 2.1304, 3.1475, 3.4107, 3.1844, 2.0018], device='cuda:7'), covar=tensor([0.0666, 0.0073, 0.0084, 0.0526, 0.0128, 0.0123, 0.0115, 0.0539], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0089, 0.0134, 0.0101, 0.0113, 0.0097, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 05:10:02,533 INFO [train.py:904] (7/8) Epoch 26, batch 5800, loss[loss=0.1727, simple_loss=0.2675, pruned_loss=0.03897, over 16261.00 frames. ], tot_loss[loss=0.2084, simple_loss=0.2941, pruned_loss=0.06135, over 3067074.22 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:10:05,680 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 2.780e+02 3.397e+02 4.110e+02 5.922e+02, threshold=6.793e+02, percent-clipped=0.0 2023-05-02 05:10:47,067 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259582.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:11:19,040 INFO [train.py:904] (7/8) Epoch 26, batch 5850, loss[loss=0.218, simple_loss=0.3082, pruned_loss=0.06393, over 15328.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2918, pruned_loss=0.05943, over 3080864.98 frames. ], batch size: 190, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:11:43,707 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 05:11:44,971 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259620.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 05:11:47,021 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259621.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:12:39,967 INFO [train.py:904] (7/8) Epoch 26, batch 5900, loss[loss=0.2233, simple_loss=0.2903, pruned_loss=0.07816, over 11807.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2912, pruned_loss=0.059, over 3082342.20 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:12:43,696 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.729e+02 2.720e+02 3.451e+02 4.224e+02 7.867e+02, threshold=6.903e+02, percent-clipped=3.0 2023-05-02 05:13:09,228 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259669.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:14:01,060 INFO [train.py:904] (7/8) Epoch 26, batch 5950, loss[loss=0.2226, simple_loss=0.3096, pruned_loss=0.06785, over 16924.00 frames. ], tot_loss[loss=0.2041, simple_loss=0.2926, pruned_loss=0.05784, over 3094754.44 frames. ], batch size: 109, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:14:46,948 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-02 05:15:12,614 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259749.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:15:18,035 INFO [train.py:904] (7/8) Epoch 26, batch 6000, loss[loss=0.1786, simple_loss=0.2692, pruned_loss=0.04394, over 16782.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2917, pruned_loss=0.0575, over 3088643.81 frames. ], batch size: 39, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:15:18,036 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 05:15:28,182 INFO [train.py:938] (7/8) Epoch 26, validation: loss=0.1485, simple_loss=0.2607, pruned_loss=0.01818, over 944034.00 frames. 2023-05-02 05:15:28,183 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 05:15:30,552 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.937e+02 2.681e+02 3.578e+02 4.333e+02 8.656e+02, threshold=7.155e+02, percent-clipped=3.0 2023-05-02 05:15:46,800 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.45 vs. limit=2.0 2023-05-02 05:15:55,090 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=259770.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:16:13,653 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:16:46,436 INFO [train.py:904] (7/8) Epoch 26, batch 6050, loss[loss=0.2166, simple_loss=0.2834, pruned_loss=0.07496, over 11903.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2906, pruned_loss=0.05734, over 3081757.86 frames. ], batch size: 247, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:16:59,713 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259810.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 05:17:04,426 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0849, 3.1796, 3.1826, 2.1785, 3.0522, 3.2272, 3.0629, 1.9117], device='cuda:7'), covar=tensor([0.0573, 0.0080, 0.0091, 0.0449, 0.0121, 0.0144, 0.0119, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0088, 0.0090, 0.0135, 0.0102, 0.0114, 0.0098, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 05:17:51,168 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259843.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:18:05,784 INFO [train.py:904] (7/8) Epoch 26, batch 6100, loss[loss=0.1984, simple_loss=0.2837, pruned_loss=0.05659, over 17027.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2896, pruned_loss=0.05599, over 3095889.41 frames. ], batch size: 55, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:18:09,304 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.693e+02 2.757e+02 3.100e+02 3.601e+02 8.920e+02, threshold=6.201e+02, percent-clipped=2.0 2023-05-02 05:18:27,928 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=259866.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:18:44,781 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-02 05:18:52,997 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259882.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:19:23,804 INFO [train.py:904] (7/8) Epoch 26, batch 6150, loss[loss=0.2078, simple_loss=0.2906, pruned_loss=0.0625, over 15548.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2875, pruned_loss=0.0555, over 3095219.66 frames. ], batch size: 191, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:19:49,536 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=259920.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:19:49,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4688, 3.6628, 2.7196, 2.1809, 2.4416, 2.4685, 3.9108, 3.3058], device='cuda:7'), covar=tensor([0.3181, 0.0652, 0.1923, 0.2893, 0.2665, 0.2130, 0.0497, 0.1277], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0274, 0.0310, 0.0322, 0.0303, 0.0271, 0.0302, 0.0347], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 05:20:00,789 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=259927.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:20:05,847 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259930.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:20:14,253 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2023-05-02 05:20:42,391 INFO [train.py:904] (7/8) Epoch 26, batch 6200, loss[loss=0.2207, simple_loss=0.2926, pruned_loss=0.07442, over 11357.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2849, pruned_loss=0.05449, over 3109140.53 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:20:44,636 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.064e+02 2.750e+02 3.295e+02 3.920e+02 8.789e+02, threshold=6.590e+02, percent-clipped=2.0 2023-05-02 05:21:05,648 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=259968.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:22:00,549 INFO [train.py:904] (7/8) Epoch 26, batch 6250, loss[loss=0.197, simple_loss=0.2823, pruned_loss=0.05583, over 15329.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2851, pruned_loss=0.0546, over 3111600.54 frames. ], batch size: 190, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:22:20,618 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2547, 2.3800, 2.3314, 3.9748, 2.2630, 2.7041, 2.3903, 2.4873], device='cuda:7'), covar=tensor([0.1432, 0.3483, 0.3007, 0.0552, 0.4039, 0.2439, 0.3688, 0.3245], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0467, 0.0381, 0.0334, 0.0444, 0.0535, 0.0439, 0.0546], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:22:27,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1750, 2.4071, 2.6287, 1.9760, 2.7445, 2.8078, 2.4534, 2.4013], device='cuda:7'), covar=tensor([0.0671, 0.0275, 0.0247, 0.0959, 0.0140, 0.0301, 0.0469, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0112, 0.0100, 0.0139, 0.0086, 0.0130, 0.0129, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:22:41,272 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260030.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:23:15,625 INFO [train.py:904] (7/8) Epoch 26, batch 6300, loss[loss=0.1792, simple_loss=0.2767, pruned_loss=0.04087, over 16905.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2851, pruned_loss=0.05409, over 3105661.74 frames. ], batch size: 96, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:23:19,603 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.817e+02 3.510e+02 4.108e+02 7.742e+02, threshold=7.020e+02, percent-clipped=4.0 2023-05-02 05:23:44,857 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260070.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:24:08,728 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4069, 2.9787, 2.7258, 2.3053, 2.2892, 2.3396, 2.9330, 2.9226], device='cuda:7'), covar=tensor([0.2442, 0.0708, 0.1617, 0.2805, 0.2427, 0.2145, 0.0611, 0.1502], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0273, 0.0309, 0.0322, 0.0303, 0.0271, 0.0300, 0.0346], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 05:24:16,361 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0172, 2.2072, 2.3111, 3.4713, 2.1158, 2.4893, 2.2827, 2.3239], device='cuda:7'), covar=tensor([0.1450, 0.3367, 0.2856, 0.0653, 0.4107, 0.2351, 0.3546, 0.3162], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0467, 0.0381, 0.0334, 0.0444, 0.0534, 0.0438, 0.0545], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:24:17,964 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260091.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:24:35,608 INFO [train.py:904] (7/8) Epoch 26, batch 6350, loss[loss=0.2027, simple_loss=0.2783, pruned_loss=0.06351, over 16647.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.2856, pruned_loss=0.05506, over 3111774.40 frames. ], batch size: 62, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:24:39,477 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260105.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 05:24:59,118 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260118.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:25:23,083 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9912, 2.1904, 2.2879, 3.6004, 2.0330, 2.5132, 2.2725, 2.3290], device='cuda:7'), covar=tensor([0.1515, 0.3611, 0.3012, 0.0630, 0.4406, 0.2502, 0.3695, 0.3188], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0465, 0.0380, 0.0332, 0.0443, 0.0532, 0.0436, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:25:30,303 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260138.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 05:25:52,114 INFO [train.py:904] (7/8) Epoch 26, batch 6400, loss[loss=0.1922, simple_loss=0.2749, pruned_loss=0.05471, over 16697.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2856, pruned_loss=0.05581, over 3117012.42 frames. ], batch size: 134, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:25:54,621 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 2.921e+02 3.424e+02 4.125e+02 9.297e+02, threshold=6.848e+02, percent-clipped=3.0 2023-05-02 05:25:56,182 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 05:27:08,959 INFO [train.py:904] (7/8) Epoch 26, batch 6450, loss[loss=0.2014, simple_loss=0.2941, pruned_loss=0.05435, over 17247.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2859, pruned_loss=0.05522, over 3117740.20 frames. ], batch size: 52, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:27:35,896 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.02 vs. limit=5.0 2023-05-02 05:27:40,351 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:27:43,906 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260224.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:27:53,560 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 05:28:28,518 INFO [train.py:904] (7/8) Epoch 26, batch 6500, loss[loss=0.1859, simple_loss=0.2748, pruned_loss=0.04854, over 16225.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2836, pruned_loss=0.05426, over 3123575.38 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:28:31,541 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.633e+02 2.830e+02 3.427e+02 3.864e+02 5.518e+02, threshold=6.855e+02, percent-clipped=0.0 2023-05-02 05:29:18,524 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.61 vs. limit=5.0 2023-05-02 05:29:20,124 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260285.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 05:29:49,650 INFO [train.py:904] (7/8) Epoch 26, batch 6550, loss[loss=0.2008, simple_loss=0.3054, pruned_loss=0.04814, over 16175.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2866, pruned_loss=0.05534, over 3120136.57 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:30:12,836 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8355, 2.6791, 2.9054, 2.1700, 2.6963, 2.1651, 2.7544, 2.8774], device='cuda:7'), covar=tensor([0.0278, 0.0847, 0.0493, 0.1770, 0.0745, 0.0941, 0.0588, 0.0672], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0168, 0.0170, 0.0156, 0.0147, 0.0131, 0.0145, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:30:22,247 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-02 05:30:30,419 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3212, 2.4521, 2.4222, 4.2208, 2.2682, 2.7914, 2.4660, 2.5947], device='cuda:7'), covar=tensor([0.1396, 0.3473, 0.2875, 0.0484, 0.3975, 0.2467, 0.3545, 0.3076], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0465, 0.0379, 0.0332, 0.0442, 0.0532, 0.0435, 0.0543], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:31:07,560 INFO [train.py:904] (7/8) Epoch 26, batch 6600, loss[loss=0.2049, simple_loss=0.2788, pruned_loss=0.06552, over 11551.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2885, pruned_loss=0.05631, over 3098811.79 frames. ], batch size: 246, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:31:09,948 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.723e+02 2.646e+02 3.224e+02 4.276e+02 9.076e+02, threshold=6.447e+02, percent-clipped=2.0 2023-05-02 05:31:21,652 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-02 05:31:59,521 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260386.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:32:00,212 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-05-02 05:32:26,389 INFO [train.py:904] (7/8) Epoch 26, batch 6650, loss[loss=0.1885, simple_loss=0.2766, pruned_loss=0.05018, over 16195.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2885, pruned_loss=0.05676, over 3109911.51 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:32:30,402 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260405.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:33:21,111 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260438.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:33:43,220 INFO [train.py:904] (7/8) Epoch 26, batch 6700, loss[loss=0.2005, simple_loss=0.292, pruned_loss=0.05443, over 16826.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2875, pruned_loss=0.05704, over 3103223.24 frames. ], batch size: 102, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:33:43,550 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260453.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:33:45,911 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.665e+02 3.203e+02 3.728e+02 7.997e+02, threshold=6.406e+02, percent-clipped=3.0 2023-05-02 05:34:35,489 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260486.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:35:00,736 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-05-02 05:35:01,140 INFO [train.py:904] (7/8) Epoch 26, batch 6750, loss[loss=0.18, simple_loss=0.2684, pruned_loss=0.04579, over 16795.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.287, pruned_loss=0.05772, over 3087194.36 frames. ], batch size: 102, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:35:31,772 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260522.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:35:35,526 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2430, 4.2358, 4.1150, 3.3272, 4.1918, 1.6551, 3.9729, 3.6818], device='cuda:7'), covar=tensor([0.0107, 0.0099, 0.0201, 0.0351, 0.0090, 0.3145, 0.0133, 0.0298], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0172, 0.0210, 0.0185, 0.0186, 0.0217, 0.0198, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:36:19,268 INFO [train.py:904] (7/8) Epoch 26, batch 6800, loss[loss=0.2069, simple_loss=0.2913, pruned_loss=0.06124, over 15356.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2874, pruned_loss=0.05791, over 3081511.91 frames. ], batch size: 191, lr: 2.58e-03, grad_scale: 8.0 2023-05-02 05:36:21,476 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.811e+02 2.923e+02 3.478e+02 4.093e+02 6.667e+02, threshold=6.957e+02, percent-clipped=2.0 2023-05-02 05:36:45,311 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:36:47,202 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 05:36:51,982 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3632, 2.9345, 3.0623, 1.9047, 2.7186, 2.0712, 3.0032, 3.1671], device='cuda:7'), covar=tensor([0.0315, 0.0877, 0.0624, 0.2181, 0.0936, 0.1076, 0.0696, 0.0923], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0170, 0.0172, 0.0157, 0.0149, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 05:37:01,730 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=260580.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 05:37:13,661 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-02 05:37:27,578 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0424, 3.5291, 3.6057, 2.0365, 2.9938, 2.3006, 3.5639, 3.7674], device='cuda:7'), covar=tensor([0.0276, 0.0742, 0.0721, 0.2232, 0.0898, 0.1149, 0.0572, 0.0698], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0170, 0.0172, 0.0157, 0.0149, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 05:37:35,992 INFO [train.py:904] (7/8) Epoch 26, batch 6850, loss[loss=0.1963, simple_loss=0.3033, pruned_loss=0.04467, over 16731.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2885, pruned_loss=0.05788, over 3097705.28 frames. ], batch size: 39, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:37:58,863 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-05-02 05:38:50,109 INFO [train.py:904] (7/8) Epoch 26, batch 6900, loss[loss=0.2068, simple_loss=0.2952, pruned_loss=0.05925, over 16979.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2905, pruned_loss=0.05704, over 3115657.29 frames. ], batch size: 55, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:38:53,861 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.801e+02 2.543e+02 3.098e+02 3.712e+02 7.299e+02, threshold=6.197e+02, percent-clipped=1.0 2023-05-02 05:38:55,104 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1857, 2.0669, 1.7679, 1.8115, 2.3007, 1.9930, 1.9742, 2.3943], device='cuda:7'), covar=tensor([0.0221, 0.0409, 0.0525, 0.0448, 0.0267, 0.0367, 0.0228, 0.0270], device='cuda:7'), in_proj_covar=tensor([0.0224, 0.0240, 0.0231, 0.0231, 0.0241, 0.0239, 0.0239, 0.0239], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:39:09,595 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8199, 1.3914, 1.7188, 1.6602, 1.7690, 1.9201, 1.6424, 1.8043], device='cuda:7'), covar=tensor([0.0253, 0.0413, 0.0235, 0.0308, 0.0293, 0.0208, 0.0452, 0.0163], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0195, 0.0183, 0.0188, 0.0203, 0.0161, 0.0200, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:39:27,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 05:39:40,178 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260686.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:40:05,864 INFO [train.py:904] (7/8) Epoch 26, batch 6950, loss[loss=0.263, simple_loss=0.3274, pruned_loss=0.09928, over 10952.00 frames. ], tot_loss[loss=0.2046, simple_loss=0.2918, pruned_loss=0.05872, over 3081497.56 frames. ], batch size: 249, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:40:07,006 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5783, 4.5867, 4.4552, 3.6790, 4.5347, 1.7298, 4.2867, 4.1041], device='cuda:7'), covar=tensor([0.0124, 0.0102, 0.0193, 0.0344, 0.0098, 0.2947, 0.0137, 0.0269], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0171, 0.0209, 0.0184, 0.0184, 0.0215, 0.0197, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:40:48,879 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260730.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:40:54,448 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260734.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:41:21,373 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260751.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:41:23,890 INFO [train.py:904] (7/8) Epoch 26, batch 7000, loss[loss=0.2257, simple_loss=0.3171, pruned_loss=0.06717, over 16777.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2909, pruned_loss=0.0571, over 3097814.21 frames. ], batch size: 124, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:41:29,423 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.843e+02 3.539e+02 4.501e+02 8.180e+02, threshold=7.079e+02, percent-clipped=4.0 2023-05-02 05:41:48,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2021, 2.8387, 3.1165, 1.7811, 3.2265, 3.2938, 2.6932, 2.5481], device='cuda:7'), covar=tensor([0.0828, 0.0284, 0.0224, 0.1213, 0.0119, 0.0207, 0.0473, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0111, 0.0100, 0.0138, 0.0086, 0.0129, 0.0129, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:42:24,285 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260791.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:42:41,497 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1094, 2.2170, 2.3203, 3.8772, 2.0737, 2.5012, 2.3078, 2.4040], device='cuda:7'), covar=tensor([0.1516, 0.3655, 0.3048, 0.0618, 0.4365, 0.2628, 0.3735, 0.3528], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0465, 0.0379, 0.0332, 0.0443, 0.0532, 0.0437, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:42:42,027 INFO [train.py:904] (7/8) Epoch 26, batch 7050, loss[loss=0.206, simple_loss=0.2969, pruned_loss=0.05753, over 16676.00 frames. ], tot_loss[loss=0.2035, simple_loss=0.2918, pruned_loss=0.05763, over 3078314.70 frames. ], batch size: 76, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:42:56,394 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260812.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:43:00,778 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2948, 3.4365, 3.5909, 3.5672, 3.5757, 3.3964, 3.4280, 3.4717], device='cuda:7'), covar=tensor([0.0435, 0.0704, 0.0468, 0.0470, 0.0588, 0.0571, 0.0830, 0.0570], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0477, 0.0463, 0.0425, 0.0511, 0.0487, 0.0564, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 05:43:07,628 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7810, 2.5136, 2.3353, 3.5718, 2.5122, 3.7121, 1.3755, 2.6903], device='cuda:7'), covar=tensor([0.1389, 0.0839, 0.1326, 0.0192, 0.0188, 0.0411, 0.1827, 0.0884], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0180, 0.0199, 0.0198, 0.0208, 0.0218, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:43:48,112 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9998, 3.4019, 3.1954, 1.7710, 2.6874, 1.8991, 3.4264, 3.6692], device='cuda:7'), covar=tensor([0.0246, 0.0803, 0.0788, 0.2700, 0.1142, 0.1331, 0.0703, 0.0987], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0170, 0.0172, 0.0157, 0.0149, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 05:43:52,624 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8132, 1.3996, 1.7265, 1.6650, 1.7584, 1.9179, 1.6764, 1.7430], device='cuda:7'), covar=tensor([0.0257, 0.0376, 0.0227, 0.0292, 0.0260, 0.0190, 0.0419, 0.0143], device='cuda:7'), in_proj_covar=tensor([0.0191, 0.0193, 0.0181, 0.0187, 0.0201, 0.0160, 0.0198, 0.0160], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:43:59,440 INFO [train.py:904] (7/8) Epoch 26, batch 7100, loss[loss=0.2339, simple_loss=0.307, pruned_loss=0.0804, over 11090.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2903, pruned_loss=0.05734, over 3087669.84 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:44:05,370 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.003e+02 2.616e+02 2.972e+02 3.634e+02 7.546e+02, threshold=5.943e+02, percent-clipped=1.0 2023-05-02 05:44:35,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4226, 3.3778, 2.7138, 2.1756, 2.2423, 2.3113, 3.5290, 3.0759], device='cuda:7'), covar=tensor([0.3219, 0.0701, 0.1871, 0.2833, 0.2718, 0.2313, 0.0508, 0.1493], device='cuda:7'), in_proj_covar=tensor([0.0332, 0.0274, 0.0309, 0.0322, 0.0303, 0.0271, 0.0301, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 05:44:43,259 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=260880.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:44:44,471 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=260881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:45:17,438 INFO [train.py:904] (7/8) Epoch 26, batch 7150, loss[loss=0.1899, simple_loss=0.2755, pruned_loss=0.05217, over 16793.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.289, pruned_loss=0.05814, over 3055080.91 frames. ], batch size: 124, lr: 2.58e-03, grad_scale: 2.0 2023-05-02 05:45:53,522 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=260928.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:46:13,631 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=260942.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:46:29,377 INFO [train.py:904] (7/8) Epoch 26, batch 7200, loss[loss=0.1801, simple_loss=0.2826, pruned_loss=0.03882, over 16709.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2871, pruned_loss=0.05667, over 3044024.80 frames. ], batch size: 83, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:46:32,011 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4839, 3.2977, 3.6290, 1.8573, 3.8057, 3.8446, 2.9255, 2.8631], device='cuda:7'), covar=tensor([0.0811, 0.0300, 0.0212, 0.1223, 0.0081, 0.0151, 0.0459, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0111, 0.0101, 0.0139, 0.0086, 0.0130, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:46:35,557 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.782e+02 3.336e+02 4.472e+02 6.403e+02, threshold=6.673e+02, percent-clipped=2.0 2023-05-02 05:47:18,926 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8291, 2.7891, 2.3068, 2.6443, 3.1386, 2.7978, 3.2892, 3.3351], device='cuda:7'), covar=tensor([0.0113, 0.0450, 0.0601, 0.0485, 0.0300, 0.0429, 0.0265, 0.0291], device='cuda:7'), in_proj_covar=tensor([0.0222, 0.0238, 0.0229, 0.0230, 0.0240, 0.0237, 0.0237, 0.0237], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:47:21,148 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9042, 4.1060, 4.3564, 4.3055, 4.3133, 4.0852, 3.9002, 4.0704], device='cuda:7'), covar=tensor([0.0553, 0.0787, 0.0525, 0.0636, 0.0661, 0.0645, 0.1408, 0.0681], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0477, 0.0462, 0.0425, 0.0511, 0.0486, 0.0564, 0.0391], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 05:47:53,029 INFO [train.py:904] (7/8) Epoch 26, batch 7250, loss[loss=0.1931, simple_loss=0.2793, pruned_loss=0.05345, over 15352.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2851, pruned_loss=0.05564, over 3054806.30 frames. ], batch size: 191, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:49:09,273 INFO [train.py:904] (7/8) Epoch 26, batch 7300, loss[loss=0.2075, simple_loss=0.3053, pruned_loss=0.05485, over 16465.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2844, pruned_loss=0.05528, over 3053958.43 frames. ], batch size: 68, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:49:15,977 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.944e+02 2.610e+02 3.123e+02 3.831e+02 8.496e+02, threshold=6.245e+02, percent-clipped=1.0 2023-05-02 05:49:34,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5029, 2.0678, 1.6286, 1.7497, 2.3466, 1.9417, 2.1258, 2.4981], device='cuda:7'), covar=tensor([0.0204, 0.0445, 0.0646, 0.0564, 0.0304, 0.0451, 0.0249, 0.0278], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0237, 0.0229, 0.0229, 0.0239, 0.0237, 0.0236, 0.0236], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:50:00,910 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261086.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:50:26,315 INFO [train.py:904] (7/8) Epoch 26, batch 7350, loss[loss=0.2089, simple_loss=0.2931, pruned_loss=0.06236, over 16943.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2857, pruned_loss=0.05594, over 3056159.16 frames. ], batch size: 109, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:50:33,370 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261107.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:51:18,029 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261136.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:51:44,989 INFO [train.py:904] (7/8) Epoch 26, batch 7400, loss[loss=0.2047, simple_loss=0.2948, pruned_loss=0.05734, over 16706.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2871, pruned_loss=0.05668, over 3063738.28 frames. ], batch size: 134, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:51:50,757 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.938e+02 2.670e+02 3.232e+02 3.696e+02 7.635e+02, threshold=6.463e+02, percent-clipped=2.0 2023-05-02 05:52:35,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8388, 3.2686, 3.2358, 2.1207, 2.9093, 2.2308, 3.3523, 3.5509], device='cuda:7'), covar=tensor([0.0319, 0.0866, 0.0755, 0.2230, 0.0972, 0.1104, 0.0735, 0.0963], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0171, 0.0156, 0.0148, 0.0132, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 05:52:43,268 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6466, 2.5422, 1.9180, 2.6601, 2.1282, 2.7746, 2.1615, 2.3580], device='cuda:7'), covar=tensor([0.0323, 0.0424, 0.1227, 0.0314, 0.0670, 0.0609, 0.1169, 0.0619], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0180, 0.0198, 0.0171, 0.0180, 0.0220, 0.0206, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:52:55,461 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261197.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 05:53:04,390 INFO [train.py:904] (7/8) Epoch 26, batch 7450, loss[loss=0.2236, simple_loss=0.3168, pruned_loss=0.06524, over 16777.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.288, pruned_loss=0.0577, over 3055212.06 frames. ], batch size: 124, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:53:15,780 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5753, 2.6419, 2.3324, 3.8039, 2.5917, 3.8898, 1.4110, 2.7892], device='cuda:7'), covar=tensor([0.1567, 0.0844, 0.1399, 0.0241, 0.0225, 0.0465, 0.1866, 0.0855], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0180, 0.0200, 0.0198, 0.0209, 0.0218, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 05:54:01,695 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261237.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:54:21,629 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-02 05:54:27,106 INFO [train.py:904] (7/8) Epoch 26, batch 7500, loss[loss=0.2237, simple_loss=0.2938, pruned_loss=0.07679, over 11505.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.288, pruned_loss=0.05718, over 3046726.39 frames. ], batch size: 248, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:54:33,504 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 2.815e+02 3.421e+02 3.855e+02 8.023e+02, threshold=6.841e+02, percent-clipped=2.0 2023-05-02 05:55:45,658 INFO [train.py:904] (7/8) Epoch 26, batch 7550, loss[loss=0.2078, simple_loss=0.2877, pruned_loss=0.0639, over 16276.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2872, pruned_loss=0.05732, over 3040110.42 frames. ], batch size: 165, lr: 2.58e-03, grad_scale: 4.0 2023-05-02 05:56:07,675 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 05:57:01,731 INFO [train.py:904] (7/8) Epoch 26, batch 7600, loss[loss=0.1961, simple_loss=0.2838, pruned_loss=0.05419, over 16427.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2866, pruned_loss=0.05718, over 3057919.55 frames. ], batch size: 146, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 05:57:07,521 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.730e+02 2.718e+02 3.314e+02 3.844e+02 7.163e+02, threshold=6.628e+02, percent-clipped=1.0 2023-05-02 05:57:32,346 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7775, 5.0533, 5.2440, 4.9835, 5.0776, 5.6108, 5.1005, 4.8718], device='cuda:7'), covar=tensor([0.1050, 0.1782, 0.2594, 0.1812, 0.2159, 0.0884, 0.1615, 0.2251], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0629, 0.0691, 0.0511, 0.0678, 0.0718, 0.0540, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 05:57:54,606 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261386.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:58:16,420 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1377, 2.2795, 2.1997, 3.8161, 2.2011, 2.5750, 2.3199, 2.3944], device='cuda:7'), covar=tensor([0.1492, 0.3502, 0.3232, 0.0630, 0.4262, 0.2528, 0.3561, 0.3412], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0465, 0.0378, 0.0331, 0.0442, 0.0531, 0.0436, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 05:58:20,648 INFO [train.py:904] (7/8) Epoch 26, batch 7650, loss[loss=0.1947, simple_loss=0.284, pruned_loss=0.05269, over 16447.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.287, pruned_loss=0.05702, over 3084428.30 frames. ], batch size: 68, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 05:58:26,602 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261407.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:58:33,510 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 05:59:08,102 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:36,567 INFO [train.py:904] (7/8) Epoch 26, batch 7700, loss[loss=0.1926, simple_loss=0.2864, pruned_loss=0.04939, over 16796.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.287, pruned_loss=0.05743, over 3080881.50 frames. ], batch size: 39, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 05:59:40,112 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261455.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:42,124 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261456.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 05:59:46,571 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 2.911e+02 3.600e+02 4.177e+02 7.928e+02, threshold=7.200e+02, percent-clipped=2.0 2023-05-02 06:00:36,511 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261492.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 06:00:52,216 INFO [train.py:904] (7/8) Epoch 26, batch 7750, loss[loss=0.2272, simple_loss=0.2953, pruned_loss=0.07953, over 11214.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2868, pruned_loss=0.05717, over 3084661.32 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:01:14,467 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261517.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:01:19,005 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7340, 4.3231, 4.2297, 2.8575, 3.8045, 4.3256, 3.7333, 2.5249], device='cuda:7'), covar=tensor([0.0479, 0.0047, 0.0059, 0.0393, 0.0099, 0.0118, 0.0100, 0.0424], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0087, 0.0088, 0.0134, 0.0100, 0.0113, 0.0097, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 06:01:44,997 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261537.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:02:09,782 INFO [train.py:904] (7/8) Epoch 26, batch 7800, loss[loss=0.263, simple_loss=0.3206, pruned_loss=0.1027, over 11456.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.288, pruned_loss=0.05834, over 3063971.35 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:02:19,359 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.114e+02 2.737e+02 3.262e+02 3.975e+02 6.664e+02, threshold=6.524e+02, percent-clipped=0.0 2023-05-02 06:03:00,430 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261585.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:03:28,327 INFO [train.py:904] (7/8) Epoch 26, batch 7850, loss[loss=0.1941, simple_loss=0.288, pruned_loss=0.05009, over 16205.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2887, pruned_loss=0.0585, over 3042920.42 frames. ], batch size: 165, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:04:46,521 INFO [train.py:904] (7/8) Epoch 26, batch 7900, loss[loss=0.1908, simple_loss=0.2863, pruned_loss=0.04765, over 15532.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2879, pruned_loss=0.05799, over 3038016.94 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:04:52,550 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-05-02 06:04:55,567 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.825e+02 2.587e+02 3.177e+02 3.708e+02 5.885e+02, threshold=6.353e+02, percent-clipped=0.0 2023-05-02 06:04:58,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6854, 3.0396, 3.2281, 2.0692, 2.9595, 2.1769, 3.2400, 3.3442], device='cuda:7'), covar=tensor([0.0273, 0.0838, 0.0582, 0.2138, 0.0820, 0.1039, 0.0666, 0.0859], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0168, 0.0169, 0.0155, 0.0146, 0.0131, 0.0146, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 06:06:05,692 INFO [train.py:904] (7/8) Epoch 26, batch 7950, loss[loss=0.2663, simple_loss=0.321, pruned_loss=0.1058, over 11662.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2885, pruned_loss=0.05845, over 3035806.90 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 2.0 2023-05-02 06:06:36,487 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 06:06:53,755 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2023-05-02 06:07:05,464 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2231, 1.5913, 1.9467, 2.0921, 2.1738, 2.4178, 1.7522, 2.3503], device='cuda:7'), covar=tensor([0.0243, 0.0523, 0.0333, 0.0363, 0.0357, 0.0229, 0.0560, 0.0168], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0196, 0.0184, 0.0189, 0.0204, 0.0162, 0.0201, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:07:23,914 INFO [train.py:904] (7/8) Epoch 26, batch 8000, loss[loss=0.205, simple_loss=0.2953, pruned_loss=0.05731, over 16662.00 frames. ], tot_loss[loss=0.2044, simple_loss=0.2899, pruned_loss=0.05944, over 3030820.82 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:07:32,670 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.840e+02 2.545e+02 3.130e+02 3.720e+02 6.505e+02, threshold=6.260e+02, percent-clipped=1.0 2023-05-02 06:08:14,985 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-05-02 06:08:24,318 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=261792.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 06:08:40,304 INFO [train.py:904] (7/8) Epoch 26, batch 8050, loss[loss=0.1887, simple_loss=0.2796, pruned_loss=0.04891, over 16723.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2896, pruned_loss=0.05919, over 3017559.95 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:08:54,780 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=261812.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:09:13,008 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261824.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:09:13,268 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.62 vs. limit=2.0 2023-05-02 06:09:27,663 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:09:37,847 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=261840.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 06:09:57,237 INFO [train.py:904] (7/8) Epoch 26, batch 8100, loss[loss=0.1931, simple_loss=0.2863, pruned_loss=0.04997, over 16312.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2882, pruned_loss=0.05782, over 3042946.25 frames. ], batch size: 35, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:10:06,903 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.718e+02 2.493e+02 2.989e+02 3.536e+02 5.601e+02, threshold=5.979e+02, percent-clipped=0.0 2023-05-02 06:10:46,796 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261885.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:11:00,950 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=261894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:11:14,762 INFO [train.py:904] (7/8) Epoch 26, batch 8150, loss[loss=0.2145, simple_loss=0.2857, pruned_loss=0.07159, over 11278.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2855, pruned_loss=0.05647, over 3051035.56 frames. ], batch size: 246, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:12:18,445 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9124, 4.2207, 3.1012, 2.5601, 2.8752, 2.7166, 4.6032, 3.6631], device='cuda:7'), covar=tensor([0.3035, 0.0664, 0.1948, 0.2693, 0.2761, 0.2078, 0.0433, 0.1464], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0274, 0.0310, 0.0323, 0.0305, 0.0273, 0.0302, 0.0348], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:12:25,681 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-05-02 06:12:33,072 INFO [train.py:904] (7/8) Epoch 26, batch 8200, loss[loss=0.1807, simple_loss=0.2748, pruned_loss=0.04328, over 16709.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2826, pruned_loss=0.05554, over 3082810.76 frames. ], batch size: 83, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:12:43,209 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 2.684e+02 3.208e+02 4.137e+02 6.714e+02, threshold=6.416e+02, percent-clipped=2.0 2023-05-02 06:13:12,989 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=261977.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:13:58,693 INFO [train.py:904] (7/8) Epoch 26, batch 8250, loss[loss=0.2012, simple_loss=0.3023, pruned_loss=0.05003, over 15159.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2817, pruned_loss=0.05323, over 3053868.77 frames. ], batch size: 190, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:14:57,705 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=262038.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:15:13,042 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9814, 5.2694, 5.0563, 5.0705, 4.7769, 4.8141, 4.6365, 5.3638], device='cuda:7'), covar=tensor([0.1156, 0.0877, 0.0932, 0.0874, 0.0827, 0.0813, 0.1238, 0.0794], device='cuda:7'), in_proj_covar=tensor([0.0690, 0.0829, 0.0683, 0.0638, 0.0529, 0.0530, 0.0697, 0.0649], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:15:16,413 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9710, 2.3032, 1.9083, 2.0105, 2.5675, 2.2777, 2.3947, 2.8040], device='cuda:7'), covar=tensor([0.0242, 0.0479, 0.0649, 0.0573, 0.0392, 0.0449, 0.0278, 0.0314], device='cuda:7'), in_proj_covar=tensor([0.0221, 0.0237, 0.0229, 0.0229, 0.0239, 0.0237, 0.0237, 0.0235], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:15:22,706 INFO [train.py:904] (7/8) Epoch 26, batch 8300, loss[loss=0.1636, simple_loss=0.2654, pruned_loss=0.0309, over 15506.00 frames. ], tot_loss[loss=0.1891, simple_loss=0.2784, pruned_loss=0.04992, over 3045220.44 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:15:32,445 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-05-02 06:15:32,831 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.264e+02 2.806e+02 3.280e+02 5.175e+02, threshold=5.611e+02, percent-clipped=0.0 2023-05-02 06:16:44,237 INFO [train.py:904] (7/8) Epoch 26, batch 8350, loss[loss=0.196, simple_loss=0.2914, pruned_loss=0.05033, over 16687.00 frames. ], tot_loss[loss=0.1883, simple_loss=0.2791, pruned_loss=0.04874, over 3049974.86 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 4.0 2023-05-02 06:16:48,366 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5235, 3.5005, 3.4642, 2.6876, 3.3517, 2.1415, 3.1589, 2.7972], device='cuda:7'), covar=tensor([0.0177, 0.0178, 0.0194, 0.0253, 0.0127, 0.2428, 0.0146, 0.0260], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0166, 0.0204, 0.0179, 0.0180, 0.0210, 0.0192, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:16:59,146 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262112.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:17:17,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8284, 5.0962, 5.2258, 5.0164, 5.1033, 5.6203, 5.1287, 4.8173], device='cuda:7'), covar=tensor([0.1011, 0.1997, 0.2055, 0.2044, 0.2301, 0.0955, 0.1381, 0.2435], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0613, 0.0676, 0.0502, 0.0664, 0.0704, 0.0530, 0.0674], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:17:47,067 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-02 06:17:58,219 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-02 06:18:01,768 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9841, 1.9607, 2.1416, 3.4251, 1.9103, 2.1882, 2.1101, 2.0728], device='cuda:7'), covar=tensor([0.1645, 0.4540, 0.3511, 0.0786, 0.5779, 0.3478, 0.4371, 0.4690], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0459, 0.0374, 0.0326, 0.0435, 0.0524, 0.0430, 0.0536], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:18:04,037 INFO [train.py:904] (7/8) Epoch 26, batch 8400, loss[loss=0.1744, simple_loss=0.2825, pruned_loss=0.03312, over 16749.00 frames. ], tot_loss[loss=0.185, simple_loss=0.2767, pruned_loss=0.04662, over 3060868.28 frames. ], batch size: 89, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:18:13,167 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.393e+02 2.671e+02 3.015e+02 4.313e+02, threshold=5.342e+02, percent-clipped=0.0 2023-05-02 06:18:15,413 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262160.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:18:47,991 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=262180.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:19:02,305 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=262189.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:19:24,263 INFO [train.py:904] (7/8) Epoch 26, batch 8450, loss[loss=0.1932, simple_loss=0.2694, pruned_loss=0.05854, over 12432.00 frames. ], tot_loss[loss=0.1828, simple_loss=0.2751, pruned_loss=0.04523, over 3047312.52 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:19:40,610 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7177, 2.6853, 1.7082, 2.8353, 2.0982, 2.8552, 1.9644, 2.3941], device='cuda:7'), covar=tensor([0.0301, 0.0366, 0.1580, 0.0362, 0.0779, 0.0503, 0.1615, 0.0637], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0176, 0.0193, 0.0167, 0.0176, 0.0215, 0.0202, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 06:20:14,871 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8503, 3.7067, 3.8986, 4.0013, 4.0902, 3.6729, 4.0450, 4.1099], device='cuda:7'), covar=tensor([0.1627, 0.1241, 0.1375, 0.0762, 0.0659, 0.1840, 0.0839, 0.0898], device='cuda:7'), in_proj_covar=tensor([0.0644, 0.0798, 0.0917, 0.0804, 0.0615, 0.0639, 0.0666, 0.0784], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:20:48,812 INFO [train.py:904] (7/8) Epoch 26, batch 8500, loss[loss=0.1553, simple_loss=0.2517, pruned_loss=0.0294, over 16497.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2707, pruned_loss=0.04265, over 3047578.74 frames. ], batch size: 68, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:20:57,924 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.087e+02 2.550e+02 3.186e+02 5.088e+02, threshold=5.100e+02, percent-clipped=0.0 2023-05-02 06:21:26,240 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6846, 3.8469, 2.9136, 2.1773, 2.3709, 2.4953, 4.0865, 3.3258], device='cuda:7'), covar=tensor([0.3087, 0.0604, 0.1893, 0.3343, 0.3190, 0.2311, 0.0370, 0.1396], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0271, 0.0306, 0.0319, 0.0300, 0.0269, 0.0298, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:21:45,369 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 06:21:51,751 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4046, 3.3082, 3.4442, 3.5136, 3.5684, 3.3245, 3.5316, 3.6039], device='cuda:7'), covar=tensor([0.1286, 0.1089, 0.1101, 0.0707, 0.0681, 0.1859, 0.0932, 0.1009], device='cuda:7'), in_proj_covar=tensor([0.0641, 0.0795, 0.0913, 0.0802, 0.0613, 0.0636, 0.0664, 0.0780], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:22:09,486 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 06:22:09,828 INFO [train.py:904] (7/8) Epoch 26, batch 8550, loss[loss=0.1896, simple_loss=0.2818, pruned_loss=0.04867, over 16546.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2687, pruned_loss=0.04187, over 3036105.17 frames. ], batch size: 62, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:23:07,973 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=262333.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:23:47,968 INFO [train.py:904] (7/8) Epoch 26, batch 8600, loss[loss=0.1861, simple_loss=0.2823, pruned_loss=0.04489, over 16660.00 frames. ], tot_loss[loss=0.1762, simple_loss=0.2695, pruned_loss=0.04149, over 3041303.82 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:23:59,713 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.426e+02 2.293e+02 2.800e+02 3.467e+02 6.859e+02, threshold=5.600e+02, percent-clipped=2.0 2023-05-02 06:24:32,762 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2023-05-02 06:24:57,596 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.90 vs. limit=5.0 2023-05-02 06:25:12,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5454, 3.5123, 3.4871, 2.6977, 3.4087, 2.0918, 3.1833, 2.8031], device='cuda:7'), covar=tensor([0.0132, 0.0105, 0.0167, 0.0199, 0.0096, 0.2418, 0.0124, 0.0263], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0166, 0.0203, 0.0178, 0.0179, 0.0210, 0.0192, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:25:27,334 INFO [train.py:904] (7/8) Epoch 26, batch 8650, loss[loss=0.1678, simple_loss=0.2659, pruned_loss=0.03482, over 15293.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2676, pruned_loss=0.04017, over 3036958.02 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:27:13,555 INFO [train.py:904] (7/8) Epoch 26, batch 8700, loss[loss=0.1624, simple_loss=0.2607, pruned_loss=0.03211, over 16909.00 frames. ], tot_loss[loss=0.1716, simple_loss=0.2652, pruned_loss=0.03898, over 3041044.50 frames. ], batch size: 96, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:27:25,320 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 2.080e+02 2.583e+02 3.221e+02 6.009e+02, threshold=5.165e+02, percent-clipped=1.0 2023-05-02 06:27:26,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7230, 2.7602, 2.5674, 4.2938, 2.6993, 4.0930, 1.5809, 3.0789], device='cuda:7'), covar=tensor([0.1494, 0.0794, 0.1211, 0.0161, 0.0109, 0.0343, 0.1812, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0176, 0.0196, 0.0193, 0.0203, 0.0214, 0.0205, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 06:28:02,909 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262480.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:28:05,676 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 06:28:20,425 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262489.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:28:48,936 INFO [train.py:904] (7/8) Epoch 26, batch 8750, loss[loss=0.1627, simple_loss=0.2522, pruned_loss=0.03662, over 12249.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2648, pruned_loss=0.03833, over 3040838.93 frames. ], batch size: 248, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:29:49,351 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262528.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:30:09,754 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262537.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:30:36,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6946, 2.6394, 1.8523, 2.8289, 2.1080, 2.8370, 2.1499, 2.4383], device='cuda:7'), covar=tensor([0.0314, 0.0336, 0.1285, 0.0265, 0.0681, 0.0454, 0.1300, 0.0618], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0175, 0.0192, 0.0165, 0.0175, 0.0212, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 06:30:41,541 INFO [train.py:904] (7/8) Epoch 26, batch 8800, loss[loss=0.1683, simple_loss=0.2628, pruned_loss=0.03691, over 16645.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.2628, pruned_loss=0.03712, over 3046855.25 frames. ], batch size: 62, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:30:52,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.093e+02 2.466e+02 2.812e+02 4.660e+02, threshold=4.932e+02, percent-clipped=0.0 2023-05-02 06:32:25,988 INFO [train.py:904] (7/8) Epoch 26, batch 8850, loss[loss=0.1611, simple_loss=0.268, pruned_loss=0.02705, over 15277.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2657, pruned_loss=0.03655, over 3044897.34 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:33:32,266 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=262633.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:34:13,873 INFO [train.py:904] (7/8) Epoch 26, batch 8900, loss[loss=0.1686, simple_loss=0.2595, pruned_loss=0.03889, over 12510.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2658, pruned_loss=0.03585, over 3049784.66 frames. ], batch size: 247, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:34:26,823 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.190e+02 2.649e+02 3.221e+02 8.286e+02, threshold=5.298e+02, percent-clipped=4.0 2023-05-02 06:35:20,125 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=262681.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:35:27,235 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 06:36:18,600 INFO [train.py:904] (7/8) Epoch 26, batch 8950, loss[loss=0.148, simple_loss=0.2455, pruned_loss=0.02532, over 16678.00 frames. ], tot_loss[loss=0.1685, simple_loss=0.265, pruned_loss=0.03596, over 3068511.74 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:38:00,846 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=262749.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:38:08,099 INFO [train.py:904] (7/8) Epoch 26, batch 9000, loss[loss=0.1365, simple_loss=0.2346, pruned_loss=0.01925, over 16869.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2619, pruned_loss=0.03457, over 3084187.28 frames. ], batch size: 90, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:38:08,099 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 06:38:18,519 INFO [train.py:938] (7/8) Epoch 26, validation: loss=0.1435, simple_loss=0.2475, pruned_loss=0.01976, over 944034.00 frames. 2023-05-02 06:38:18,520 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 06:38:30,748 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.305e+02 2.045e+02 2.519e+02 3.077e+02 5.140e+02, threshold=5.037e+02, percent-clipped=0.0 2023-05-02 06:38:48,457 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-05-02 06:40:03,325 INFO [train.py:904] (7/8) Epoch 26, batch 9050, loss[loss=0.1639, simple_loss=0.2531, pruned_loss=0.03736, over 16125.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2634, pruned_loss=0.0356, over 3057885.45 frames. ], batch size: 165, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:40:20,428 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=262810.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:41:48,720 INFO [train.py:904] (7/8) Epoch 26, batch 9100, loss[loss=0.1685, simple_loss=0.2686, pruned_loss=0.03424, over 15259.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2626, pruned_loss=0.03583, over 3074660.05 frames. ], batch size: 190, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:42:01,227 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.262e+02 2.654e+02 3.227e+02 6.240e+02, threshold=5.309e+02, percent-clipped=4.0 2023-05-02 06:43:45,869 INFO [train.py:904] (7/8) Epoch 26, batch 9150, loss[loss=0.1572, simple_loss=0.2535, pruned_loss=0.0304, over 16419.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2624, pruned_loss=0.03542, over 3067832.95 frames. ], batch size: 166, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:44:00,914 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5054, 3.5392, 2.6022, 2.1919, 2.1580, 2.1570, 3.5802, 2.9365], device='cuda:7'), covar=tensor([0.3202, 0.0711, 0.2132, 0.2950, 0.3135, 0.2591, 0.0539, 0.1641], device='cuda:7'), in_proj_covar=tensor([0.0327, 0.0269, 0.0304, 0.0317, 0.0296, 0.0268, 0.0296, 0.0341], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:44:35,419 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1587, 4.4930, 4.4822, 3.0613, 3.8658, 4.4318, 4.0670, 2.5794], device='cuda:7'), covar=tensor([0.0412, 0.0046, 0.0040, 0.0370, 0.0112, 0.0093, 0.0069, 0.0470], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0085, 0.0086, 0.0130, 0.0098, 0.0109, 0.0094, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:44:50,313 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4345, 3.6828, 3.7063, 2.4553, 3.3284, 3.6925, 3.5025, 2.1605], device='cuda:7'), covar=tensor([0.0488, 0.0058, 0.0056, 0.0422, 0.0127, 0.0107, 0.0083, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0085, 0.0086, 0.0130, 0.0098, 0.0109, 0.0094, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:45:27,081 INFO [train.py:904] (7/8) Epoch 26, batch 9200, loss[loss=0.1633, simple_loss=0.2585, pruned_loss=0.03406, over 16369.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2584, pruned_loss=0.03474, over 3062176.10 frames. ], batch size: 146, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:45:36,603 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.229e+02 2.531e+02 3.050e+02 5.061e+02, threshold=5.062e+02, percent-clipped=0.0 2023-05-02 06:47:01,956 INFO [train.py:904] (7/8) Epoch 26, batch 9250, loss[loss=0.1527, simple_loss=0.2468, pruned_loss=0.0293, over 16169.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.258, pruned_loss=0.03442, over 3062978.15 frames. ], batch size: 165, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:48:49,720 INFO [train.py:904] (7/8) Epoch 26, batch 9300, loss[loss=0.1525, simple_loss=0.2468, pruned_loss=0.02906, over 16669.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2574, pruned_loss=0.03403, over 3083367.74 frames. ], batch size: 134, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:49:02,027 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.028e+02 2.420e+02 3.075e+02 6.199e+02, threshold=4.840e+02, percent-clipped=1.0 2023-05-02 06:50:20,375 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263095.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:50:33,219 INFO [train.py:904] (7/8) Epoch 26, batch 9350, loss[loss=0.1762, simple_loss=0.2662, pruned_loss=0.04308, over 15345.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2578, pruned_loss=0.03447, over 3088518.04 frames. ], batch size: 191, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:50:38,637 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263105.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:50:51,216 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263111.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:51:13,459 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1783, 3.2329, 3.2406, 2.2725, 2.9800, 3.2668, 3.1156, 1.9895], device='cuda:7'), covar=tensor([0.0511, 0.0061, 0.0067, 0.0413, 0.0126, 0.0101, 0.0092, 0.0533], device='cuda:7'), in_proj_covar=tensor([0.0133, 0.0084, 0.0086, 0.0131, 0.0098, 0.0109, 0.0094, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 06:51:48,388 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263140.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:52:14,172 INFO [train.py:904] (7/8) Epoch 26, batch 9400, loss[loss=0.1487, simple_loss=0.2371, pruned_loss=0.0302, over 12231.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2576, pruned_loss=0.03412, over 3078200.41 frames. ], batch size: 248, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:52:19,959 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263156.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:52:25,031 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.115e+02 2.425e+02 3.007e+02 4.996e+02, threshold=4.851e+02, percent-clipped=1.0 2023-05-02 06:52:52,633 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263172.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:52:54,907 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9487, 4.2428, 4.1009, 4.0778, 3.7278, 3.7991, 3.8682, 4.2558], device='cuda:7'), covar=tensor([0.1240, 0.1024, 0.0934, 0.0837, 0.0876, 0.1994, 0.0973, 0.1024], device='cuda:7'), in_proj_covar=tensor([0.0683, 0.0821, 0.0673, 0.0631, 0.0524, 0.0525, 0.0691, 0.0644], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 06:53:50,929 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263201.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:53:53,501 INFO [train.py:904] (7/8) Epoch 26, batch 9450, loss[loss=0.1541, simple_loss=0.2535, pruned_loss=0.02738, over 16662.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2588, pruned_loss=0.03415, over 3065335.97 frames. ], batch size: 89, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:54:14,443 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263214.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:55:15,595 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 06:55:19,978 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263246.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:55:33,115 INFO [train.py:904] (7/8) Epoch 26, batch 9500, loss[loss=0.1689, simple_loss=0.2618, pruned_loss=0.03799, over 16871.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2584, pruned_loss=0.03388, over 3086428.85 frames. ], batch size: 96, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:55:47,502 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 2.048e+02 2.318e+02 3.052e+02 5.561e+02, threshold=4.636e+02, percent-clipped=2.0 2023-05-02 06:56:19,778 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:57:17,624 INFO [train.py:904] (7/8) Epoch 26, batch 9550, loss[loss=0.1895, simple_loss=0.2837, pruned_loss=0.04767, over 16817.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2579, pruned_loss=0.03399, over 3081854.03 frames. ], batch size: 124, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:57:27,349 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263307.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 06:57:40,788 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3407, 4.3406, 4.7018, 4.6669, 4.6720, 4.4297, 4.3805, 4.3682], device='cuda:7'), covar=tensor([0.0377, 0.0852, 0.0448, 0.0439, 0.0491, 0.0435, 0.0891, 0.0465], device='cuda:7'), in_proj_covar=tensor([0.0408, 0.0461, 0.0448, 0.0412, 0.0497, 0.0473, 0.0543, 0.0376], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 06:58:11,680 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2515, 3.0634, 3.2321, 1.7215, 3.4220, 3.5484, 2.8240, 2.6715], device='cuda:7'), covar=tensor([0.0870, 0.0335, 0.0270, 0.1407, 0.0122, 0.0218, 0.0502, 0.0544], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0107, 0.0095, 0.0135, 0.0082, 0.0123, 0.0125, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 06:58:58,575 INFO [train.py:904] (7/8) Epoch 26, batch 9600, loss[loss=0.1609, simple_loss=0.2601, pruned_loss=0.03092, over 16596.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2595, pruned_loss=0.03448, over 3095706.04 frames. ], batch size: 68, lr: 2.57e-03, grad_scale: 8.0 2023-05-02 06:59:09,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.689e+02 2.249e+02 2.699e+02 3.019e+02 5.664e+02, threshold=5.399e+02, percent-clipped=4.0 2023-05-02 06:59:28,664 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 07:00:45,653 INFO [train.py:904] (7/8) Epoch 26, batch 9650, loss[loss=0.1568, simple_loss=0.2462, pruned_loss=0.03368, over 12093.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.261, pruned_loss=0.03481, over 3081881.94 frames. ], batch size: 248, lr: 2.56e-03, grad_scale: 8.0 2023-05-02 07:00:52,322 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263405.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:02:29,094 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263451.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:02:31,841 INFO [train.py:904] (7/8) Epoch 26, batch 9700, loss[loss=0.1723, simple_loss=0.2635, pruned_loss=0.04056, over 16968.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2602, pruned_loss=0.03447, over 3099705.70 frames. ], batch size: 109, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:02:33,076 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263453.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:02:42,019 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.181e+02 2.600e+02 3.178e+02 8.821e+02, threshold=5.200e+02, percent-clipped=1.0 2023-05-02 07:02:58,819 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263467.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:04:02,103 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263496.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:04:13,670 INFO [train.py:904] (7/8) Epoch 26, batch 9750, loss[loss=0.163, simple_loss=0.2594, pruned_loss=0.03335, over 16364.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.259, pruned_loss=0.03456, over 3092996.57 frames. ], batch size: 146, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:05:51,445 INFO [train.py:904] (7/8) Epoch 26, batch 9800, loss[loss=0.1553, simple_loss=0.2635, pruned_loss=0.0236, over 16860.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2586, pruned_loss=0.03345, over 3103536.51 frames. ], batch size: 96, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:06:03,263 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 2.162e+02 2.515e+02 2.948e+02 4.356e+02, threshold=5.031e+02, percent-clipped=0.0 2023-05-02 07:06:23,921 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:07:23,219 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3497, 2.9675, 3.0927, 1.7504, 3.2487, 3.3760, 2.7498, 2.6411], device='cuda:7'), covar=tensor([0.0723, 0.0303, 0.0248, 0.1292, 0.0129, 0.0195, 0.0502, 0.0475], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0106, 0.0094, 0.0134, 0.0082, 0.0123, 0.0125, 0.0125], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 07:07:23,400 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.50 vs. limit=2.0 2023-05-02 07:07:33,456 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=263602.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:07:34,243 INFO [train.py:904] (7/8) Epoch 26, batch 9850, loss[loss=0.1549, simple_loss=0.2528, pruned_loss=0.02854, over 15376.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2598, pruned_loss=0.03341, over 3100188.95 frames. ], batch size: 191, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:09:01,336 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3666, 4.2172, 4.4115, 4.5176, 4.6816, 4.2469, 4.6615, 4.7070], device='cuda:7'), covar=tensor([0.2022, 0.1320, 0.1632, 0.0900, 0.0646, 0.1217, 0.0826, 0.0697], device='cuda:7'), in_proj_covar=tensor([0.0633, 0.0781, 0.0892, 0.0789, 0.0601, 0.0625, 0.0656, 0.0763], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:09:24,095 INFO [train.py:904] (7/8) Epoch 26, batch 9900, loss[loss=0.1609, simple_loss=0.2479, pruned_loss=0.03691, over 12251.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2606, pruned_loss=0.03352, over 3092943.23 frames. ], batch size: 248, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:09:36,803 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 1.934e+02 2.301e+02 2.936e+02 6.933e+02, threshold=4.602e+02, percent-clipped=3.0 2023-05-02 07:11:20,746 INFO [train.py:904] (7/8) Epoch 26, batch 9950, loss[loss=0.1398, simple_loss=0.2425, pruned_loss=0.01852, over 16613.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2627, pruned_loss=0.0339, over 3075104.21 frames. ], batch size: 75, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:11:28,176 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 07:12:48,098 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4588, 4.5684, 4.6971, 4.4991, 4.6238, 5.0684, 4.6000, 4.2366], device='cuda:7'), covar=tensor([0.1317, 0.1855, 0.2025, 0.2013, 0.2127, 0.0876, 0.1492, 0.2337], device='cuda:7'), in_proj_covar=tensor([0.0398, 0.0590, 0.0653, 0.0483, 0.0637, 0.0680, 0.0511, 0.0643], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 07:13:19,240 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263751.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:13:21,146 INFO [train.py:904] (7/8) Epoch 26, batch 10000, loss[loss=0.1598, simple_loss=0.2501, pruned_loss=0.03479, over 12745.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2611, pruned_loss=0.03369, over 3092024.81 frames. ], batch size: 248, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:13:34,914 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.135e+02 2.379e+02 2.761e+02 5.378e+02, threshold=4.758e+02, percent-clipped=3.0 2023-05-02 07:13:51,446 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263767.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:14:45,216 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6642, 2.6782, 1.9278, 2.8138, 2.1680, 2.8122, 2.1217, 2.4006], device='cuda:7'), covar=tensor([0.0308, 0.0396, 0.1259, 0.0300, 0.0680, 0.0528, 0.1306, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0168, 0.0172, 0.0188, 0.0161, 0.0172, 0.0208, 0.0198, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:14:48,918 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263796.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:14:53,809 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263799.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:14:59,218 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9569, 5.2470, 5.0580, 5.0588, 4.7226, 4.8119, 4.5817, 5.3438], device='cuda:7'), covar=tensor([0.1258, 0.0906, 0.0924, 0.0787, 0.0891, 0.0921, 0.1372, 0.0926], device='cuda:7'), in_proj_covar=tensor([0.0680, 0.0819, 0.0670, 0.0629, 0.0523, 0.0522, 0.0691, 0.0644], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:15:00,039 INFO [train.py:904] (7/8) Epoch 26, batch 10050, loss[loss=0.17, simple_loss=0.2646, pruned_loss=0.03769, over 16946.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2613, pruned_loss=0.03393, over 3076603.30 frames. ], batch size: 109, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:15:23,691 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263815.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:15:57,708 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263833.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:16:17,296 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263844.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:16:30,288 INFO [train.py:904] (7/8) Epoch 26, batch 10100, loss[loss=0.1566, simple_loss=0.2513, pruned_loss=0.03089, over 15382.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2612, pruned_loss=0.03378, over 3088213.62 frames. ], batch size: 191, lr: 2.56e-03, grad_scale: 16.0 2023-05-02 07:16:39,584 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.128e+02 2.539e+02 3.188e+02 8.403e+02, threshold=5.079e+02, percent-clipped=3.0 2023-05-02 07:17:02,807 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263870.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:17:36,333 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=263894.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 07:18:08,993 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=263902.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:18:09,862 INFO [train.py:904] (7/8) Epoch 27, batch 0, loss[loss=0.2022, simple_loss=0.2809, pruned_loss=0.06176, over 17037.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2809, pruned_loss=0.06176, over 17037.00 frames. ], batch size: 53, lr: 2.51e-03, grad_scale: 16.0 2023-05-02 07:18:09,863 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 07:18:17,112 INFO [train.py:938] (7/8) Epoch 27, validation: loss=0.1434, simple_loss=0.2467, pruned_loss=0.02006, over 944034.00 frames. 2023-05-02 07:18:17,113 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 07:18:39,056 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:19:22,226 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=263950.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:19:26,178 INFO [train.py:904] (7/8) Epoch 27, batch 50, loss[loss=0.1676, simple_loss=0.2534, pruned_loss=0.04088, over 16839.00 frames. ], tot_loss[loss=0.1785, simple_loss=0.2664, pruned_loss=0.0453, over 750474.44 frames. ], batch size: 42, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:19:36,251 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=263960.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:19:39,891 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.543e+02 2.464e+02 3.043e+02 3.721e+02 7.050e+02, threshold=6.086e+02, percent-clipped=6.0 2023-05-02 07:19:50,817 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7389, 4.1753, 2.9776, 2.3270, 2.6518, 2.5620, 4.4410, 3.4127], device='cuda:7'), covar=tensor([0.3027, 0.0609, 0.1884, 0.3001, 0.2845, 0.2192, 0.0409, 0.1602], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0268, 0.0306, 0.0317, 0.0293, 0.0268, 0.0297, 0.0340], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 07:20:36,417 INFO [train.py:904] (7/8) Epoch 27, batch 100, loss[loss=0.1796, simple_loss=0.2776, pruned_loss=0.04082, over 16681.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2628, pruned_loss=0.04328, over 1321484.25 frames. ], batch size: 62, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:21:01,015 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264021.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:21:10,299 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-05-02 07:21:36,117 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1761, 5.8946, 6.0105, 5.6822, 5.9258, 6.3676, 5.8509, 5.5132], device='cuda:7'), covar=tensor([0.0952, 0.1752, 0.2411, 0.2054, 0.2034, 0.0778, 0.1507, 0.2217], device='cuda:7'), in_proj_covar=tensor([0.0407, 0.0602, 0.0669, 0.0493, 0.0652, 0.0693, 0.0522, 0.0657], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 07:21:44,757 INFO [train.py:904] (7/8) Epoch 27, batch 150, loss[loss=0.1826, simple_loss=0.2638, pruned_loss=0.05074, over 16531.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.26, pruned_loss=0.04185, over 1766228.97 frames. ], batch size: 146, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:21:57,551 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.626e+02 2.228e+02 2.541e+02 3.123e+02 6.802e+02, threshold=5.083e+02, percent-clipped=1.0 2023-05-02 07:22:06,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3005, 4.1133, 4.3913, 4.5153, 4.6088, 4.2007, 4.4203, 4.6090], device='cuda:7'), covar=tensor([0.1871, 0.1380, 0.1426, 0.0755, 0.0683, 0.1242, 0.2512, 0.0764], device='cuda:7'), in_proj_covar=tensor([0.0645, 0.0797, 0.0913, 0.0804, 0.0613, 0.0637, 0.0670, 0.0777], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:22:38,364 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-02 07:22:53,635 INFO [train.py:904] (7/8) Epoch 27, batch 200, loss[loss=0.1691, simple_loss=0.2656, pruned_loss=0.03625, over 16990.00 frames. ], tot_loss[loss=0.1707, simple_loss=0.2587, pruned_loss=0.04132, over 2108803.67 frames. ], batch size: 55, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:24:00,836 INFO [train.py:904] (7/8) Epoch 27, batch 250, loss[loss=0.1619, simple_loss=0.2411, pruned_loss=0.04131, over 16272.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2575, pruned_loss=0.04129, over 2382415.32 frames. ], batch size: 165, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:24:14,013 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.137e+02 2.566e+02 3.283e+02 9.180e+02, threshold=5.132e+02, percent-clipped=3.0 2023-05-02 07:24:50,579 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264189.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:24:50,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6886, 4.6482, 4.5583, 4.0373, 4.6222, 1.8267, 4.3615, 4.2228], device='cuda:7'), covar=tensor([0.0168, 0.0142, 0.0229, 0.0366, 0.0127, 0.2890, 0.0173, 0.0305], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0167, 0.0203, 0.0176, 0.0180, 0.0212, 0.0193, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:25:10,476 INFO [train.py:904] (7/8) Epoch 27, batch 300, loss[loss=0.1757, simple_loss=0.2537, pruned_loss=0.04885, over 16809.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2555, pruned_loss=0.04045, over 2576636.09 frames. ], batch size: 124, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:25:42,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0242, 3.0879, 3.3547, 2.1500, 2.9489, 2.2742, 3.4743, 3.4750], device='cuda:7'), covar=tensor([0.0235, 0.0961, 0.0620, 0.2086, 0.0881, 0.1085, 0.0577, 0.0892], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0167, 0.0154, 0.0145, 0.0130, 0.0144, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:25:53,804 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264234.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:26:19,425 INFO [train.py:904] (7/8) Epoch 27, batch 350, loss[loss=0.1625, simple_loss=0.2499, pruned_loss=0.03752, over 17173.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2536, pruned_loss=0.03971, over 2739293.13 frames. ], batch size: 46, lr: 2.51e-03, grad_scale: 1.0 2023-05-02 07:26:34,331 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 2.070e+02 2.475e+02 2.951e+02 5.112e+02, threshold=4.951e+02, percent-clipped=0.0 2023-05-02 07:26:42,184 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6126, 3.6702, 3.4567, 3.1592, 3.2890, 3.5701, 3.3572, 3.4563], device='cuda:7'), covar=tensor([0.0545, 0.0598, 0.0334, 0.0306, 0.0559, 0.0454, 0.1489, 0.0499], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0452, 0.0353, 0.0355, 0.0352, 0.0408, 0.0243, 0.0423], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:27:17,919 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264295.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:27:27,703 INFO [train.py:904] (7/8) Epoch 27, batch 400, loss[loss=0.1991, simple_loss=0.2749, pruned_loss=0.06164, over 16894.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2521, pruned_loss=0.03965, over 2854236.83 frames. ], batch size: 109, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:27:33,118 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-05-02 07:27:45,296 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264316.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:28:28,215 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2343, 3.1414, 3.4575, 2.1942, 2.9748, 2.2892, 3.6485, 3.5881], device='cuda:7'), covar=tensor([0.0250, 0.1086, 0.0693, 0.2136, 0.0949, 0.1136, 0.0573, 0.1003], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0166, 0.0167, 0.0155, 0.0146, 0.0130, 0.0144, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:28:33,912 INFO [train.py:904] (7/8) Epoch 27, batch 450, loss[loss=0.1764, simple_loss=0.2515, pruned_loss=0.05067, over 16528.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2509, pruned_loss=0.03901, over 2958986.33 frames. ], batch size: 146, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:28:37,488 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3428, 5.3183, 5.2381, 4.6614, 4.8144, 5.2062, 5.2007, 4.8533], device='cuda:7'), covar=tensor([0.0581, 0.0497, 0.0299, 0.0381, 0.1096, 0.0490, 0.0318, 0.0852], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0453, 0.0353, 0.0355, 0.0353, 0.0410, 0.0244, 0.0425], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:28:48,386 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.223e+02 2.597e+02 3.088e+02 6.420e+02, threshold=5.195e+02, percent-clipped=1.0 2023-05-02 07:29:34,770 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3454, 3.9985, 4.4956, 2.5070, 4.7344, 4.8692, 3.5446, 3.6236], device='cuda:7'), covar=tensor([0.0677, 0.0302, 0.0224, 0.1168, 0.0084, 0.0140, 0.0431, 0.0437], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0111, 0.0099, 0.0139, 0.0085, 0.0129, 0.0130, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:29:44,922 INFO [train.py:904] (7/8) Epoch 27, batch 500, loss[loss=0.1536, simple_loss=0.2324, pruned_loss=0.03744, over 16865.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2488, pruned_loss=0.03804, over 3039376.89 frames. ], batch size: 109, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:30:29,931 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 07:30:51,116 INFO [train.py:904] (7/8) Epoch 27, batch 550, loss[loss=0.1972, simple_loss=0.266, pruned_loss=0.06421, over 16897.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2493, pruned_loss=0.03814, over 3100756.73 frames. ], batch size: 116, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:30:55,032 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 07:31:04,231 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.059e+02 2.347e+02 2.787e+02 4.742e+02, threshold=4.693e+02, percent-clipped=0.0 2023-05-02 07:31:26,693 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264479.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:31:41,334 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=264489.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:31:55,571 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264500.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 07:31:59,348 INFO [train.py:904] (7/8) Epoch 27, batch 600, loss[loss=0.1495, simple_loss=0.2256, pruned_loss=0.03675, over 16487.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2479, pruned_loss=0.03804, over 3132756.53 frames. ], batch size: 75, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:32:46,462 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=264537.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:32:50,490 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264540.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:32:54,089 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2023-05-02 07:33:08,808 INFO [train.py:904] (7/8) Epoch 27, batch 650, loss[loss=0.1657, simple_loss=0.261, pruned_loss=0.03519, over 17122.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2474, pruned_loss=0.03821, over 3178719.80 frames. ], batch size: 48, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:33:19,002 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264561.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 07:33:20,705 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.185e+02 2.561e+02 3.175e+02 5.971e+02, threshold=5.123e+02, percent-clipped=4.0 2023-05-02 07:33:58,969 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264590.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:34:15,993 INFO [train.py:904] (7/8) Epoch 27, batch 700, loss[loss=0.1496, simple_loss=0.2343, pruned_loss=0.03245, over 16763.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2481, pruned_loss=0.03822, over 3211044.92 frames. ], batch size: 83, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:34:34,307 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=264616.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:35:22,126 INFO [train.py:904] (7/8) Epoch 27, batch 750, loss[loss=0.137, simple_loss=0.2292, pruned_loss=0.02237, over 16811.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2482, pruned_loss=0.03792, over 3242773.26 frames. ], batch size: 42, lr: 2.51e-03, grad_scale: 2.0 2023-05-02 07:35:35,427 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.184e+02 2.458e+02 2.836e+02 5.785e+02, threshold=4.916e+02, percent-clipped=2.0 2023-05-02 07:35:36,356 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7160, 5.0331, 4.8358, 4.8368, 4.5657, 4.5680, 4.4896, 5.1195], device='cuda:7'), covar=tensor([0.1273, 0.0940, 0.1041, 0.0873, 0.0931, 0.1158, 0.1324, 0.0905], device='cuda:7'), in_proj_covar=tensor([0.0713, 0.0857, 0.0702, 0.0661, 0.0548, 0.0546, 0.0727, 0.0672], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:35:36,728 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 07:35:37,451 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=264664.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:36:16,854 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-02 07:36:28,539 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0294, 2.9899, 2.9506, 5.2186, 4.2761, 4.4983, 1.8741, 3.3415], device='cuda:7'), covar=tensor([0.1317, 0.0789, 0.1160, 0.0230, 0.0242, 0.0433, 0.1559, 0.0755], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0179, 0.0198, 0.0198, 0.0204, 0.0217, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:36:29,182 INFO [train.py:904] (7/8) Epoch 27, batch 800, loss[loss=0.1393, simple_loss=0.2275, pruned_loss=0.02557, over 16960.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2476, pruned_loss=0.03764, over 3255013.64 frames. ], batch size: 41, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:36:44,380 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9986, 4.7630, 5.0559, 5.2349, 5.4217, 4.8019, 5.3745, 5.4242], device='cuda:7'), covar=tensor([0.2220, 0.1588, 0.1945, 0.0831, 0.0619, 0.0914, 0.0663, 0.0739], device='cuda:7'), in_proj_covar=tensor([0.0670, 0.0829, 0.0951, 0.0836, 0.0635, 0.0660, 0.0695, 0.0807], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:36:54,967 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-05-02 07:37:36,924 INFO [train.py:904] (7/8) Epoch 27, batch 850, loss[loss=0.1547, simple_loss=0.2451, pruned_loss=0.03217, over 16826.00 frames. ], tot_loss[loss=0.1609, simple_loss=0.247, pruned_loss=0.03742, over 3272675.08 frames. ], batch size: 39, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:37:51,809 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.032e+02 2.277e+02 2.694e+02 3.998e+02, threshold=4.553e+02, percent-clipped=0.0 2023-05-02 07:38:44,878 INFO [train.py:904] (7/8) Epoch 27, batch 900, loss[loss=0.1628, simple_loss=0.2407, pruned_loss=0.04242, over 12500.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2467, pruned_loss=0.03682, over 3276910.83 frames. ], batch size: 247, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:38:47,136 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=264804.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:39:19,820 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0636, 2.2172, 2.3702, 3.6890, 2.2060, 2.4717, 2.3039, 2.3526], device='cuda:7'), covar=tensor([0.1629, 0.3790, 0.3249, 0.0759, 0.4030, 0.2701, 0.3965, 0.3232], device='cuda:7'), in_proj_covar=tensor([0.0419, 0.0470, 0.0385, 0.0334, 0.0445, 0.0536, 0.0441, 0.0549], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:39:28,087 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264835.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:39:51,357 INFO [train.py:904] (7/8) Epoch 27, batch 950, loss[loss=0.1297, simple_loss=0.2153, pruned_loss=0.02205, over 17003.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2468, pruned_loss=0.03715, over 3272083.63 frames. ], batch size: 41, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:39:53,164 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 07:39:55,667 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=264856.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 07:40:04,152 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.346e+02 2.026e+02 2.339e+02 3.072e+02 8.203e+02, threshold=4.678e+02, percent-clipped=3.0 2023-05-02 07:40:07,771 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=264865.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:40:41,785 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=264890.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:40:57,192 INFO [train.py:904] (7/8) Epoch 27, batch 1000, loss[loss=0.1668, simple_loss=0.2591, pruned_loss=0.03728, over 17081.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.2456, pruned_loss=0.03673, over 3288092.52 frames. ], batch size: 53, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:41:47,416 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=264938.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:41:52,379 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5574, 2.5538, 2.5388, 4.3573, 2.5846, 2.8648, 2.6052, 2.7185], device='cuda:7'), covar=tensor([0.1363, 0.3700, 0.3294, 0.0585, 0.4033, 0.2757, 0.3629, 0.3878], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0473, 0.0387, 0.0337, 0.0448, 0.0539, 0.0443, 0.0552], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:41:52,678 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-02 07:42:07,416 INFO [train.py:904] (7/8) Epoch 27, batch 1050, loss[loss=0.1597, simple_loss=0.2333, pruned_loss=0.04301, over 16357.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.2454, pruned_loss=0.03672, over 3294550.03 frames. ], batch size: 165, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:42:19,712 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.065e+02 2.389e+02 2.901e+02 6.221e+02, threshold=4.777e+02, percent-clipped=3.0 2023-05-02 07:43:16,452 INFO [train.py:904] (7/8) Epoch 27, batch 1100, loss[loss=0.1611, simple_loss=0.238, pruned_loss=0.04215, over 16672.00 frames. ], tot_loss[loss=0.1587, simple_loss=0.245, pruned_loss=0.03624, over 3292033.53 frames. ], batch size: 134, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:43:19,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0113, 2.1969, 2.6926, 2.9623, 2.7564, 3.4736, 2.4725, 3.4569], device='cuda:7'), covar=tensor([0.0288, 0.0546, 0.0372, 0.0379, 0.0411, 0.0213, 0.0498, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0196, 0.0184, 0.0189, 0.0205, 0.0163, 0.0202, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:44:25,477 INFO [train.py:904] (7/8) Epoch 27, batch 1150, loss[loss=0.1485, simple_loss=0.2457, pruned_loss=0.02562, over 17015.00 frames. ], tot_loss[loss=0.1582, simple_loss=0.2447, pruned_loss=0.03583, over 3299984.42 frames. ], batch size: 50, lr: 2.51e-03, grad_scale: 4.0 2023-05-02 07:44:39,252 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.949e+02 2.374e+02 2.990e+02 6.257e+02, threshold=4.748e+02, percent-clipped=3.0 2023-05-02 07:45:10,930 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 07:45:34,346 INFO [train.py:904] (7/8) Epoch 27, batch 1200, loss[loss=0.1504, simple_loss=0.2375, pruned_loss=0.03169, over 17211.00 frames. ], tot_loss[loss=0.1575, simple_loss=0.244, pruned_loss=0.03552, over 3296282.87 frames. ], batch size: 44, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:46:19,464 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265135.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:46:29,409 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7621, 4.0199, 2.8367, 4.5956, 3.3247, 4.5797, 2.7657, 3.4453], device='cuda:7'), covar=tensor([0.0363, 0.0411, 0.1443, 0.0342, 0.0767, 0.0446, 0.1556, 0.0687], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0182, 0.0198, 0.0175, 0.0182, 0.0223, 0.0207, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:46:42,969 INFO [train.py:904] (7/8) Epoch 27, batch 1250, loss[loss=0.1382, simple_loss=0.2256, pruned_loss=0.02536, over 16820.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.2448, pruned_loss=0.03666, over 3306452.04 frames. ], batch size: 39, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:46:48,397 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265156.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 07:46:53,532 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265160.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:46:57,313 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.048e+02 2.480e+02 3.047e+02 6.939e+02, threshold=4.960e+02, percent-clipped=4.0 2023-05-02 07:47:06,674 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6168, 3.6977, 3.4433, 3.1125, 3.2809, 3.5758, 3.3509, 3.4440], device='cuda:7'), covar=tensor([0.0623, 0.0786, 0.0341, 0.0310, 0.0540, 0.0545, 0.1458, 0.0531], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0474, 0.0369, 0.0372, 0.0368, 0.0430, 0.0254, 0.0444], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 07:47:25,264 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265183.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:47:31,536 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7856, 2.3796, 2.3875, 3.6416, 2.8809, 3.7842, 1.6056, 2.8171], device='cuda:7'), covar=tensor([0.1515, 0.0836, 0.1375, 0.0239, 0.0161, 0.0435, 0.1730, 0.0904], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0199, 0.0199, 0.0205, 0.0219, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 07:47:53,013 INFO [train.py:904] (7/8) Epoch 27, batch 1300, loss[loss=0.1516, simple_loss=0.2325, pruned_loss=0.03539, over 16733.00 frames. ], tot_loss[loss=0.1589, simple_loss=0.2443, pruned_loss=0.03671, over 3300518.56 frames. ], batch size: 134, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:47:54,332 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265204.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:48:41,240 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0617, 2.2670, 2.6195, 3.0424, 2.8480, 3.6165, 2.3469, 3.5375], device='cuda:7'), covar=tensor([0.0296, 0.0534, 0.0419, 0.0362, 0.0394, 0.0187, 0.0555, 0.0171], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0197, 0.0185, 0.0191, 0.0206, 0.0165, 0.0204, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:48:51,484 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-02 07:49:00,606 INFO [train.py:904] (7/8) Epoch 27, batch 1350, loss[loss=0.1556, simple_loss=0.2474, pruned_loss=0.03188, over 17038.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.2448, pruned_loss=0.03636, over 3302725.09 frames. ], batch size: 55, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:49:14,444 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.174e+02 2.461e+02 3.019e+02 8.065e+02, threshold=4.923e+02, percent-clipped=2.0 2023-05-02 07:49:35,228 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265278.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:49:37,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7972, 6.1634, 5.8699, 5.9953, 5.5272, 5.5665, 5.5463, 6.3003], device='cuda:7'), covar=tensor([0.1408, 0.0974, 0.1118, 0.1002, 0.0977, 0.0709, 0.1336, 0.0957], device='cuda:7'), in_proj_covar=tensor([0.0722, 0.0867, 0.0711, 0.0670, 0.0554, 0.0549, 0.0736, 0.0681], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:50:07,775 INFO [train.py:904] (7/8) Epoch 27, batch 1400, loss[loss=0.131, simple_loss=0.2167, pruned_loss=0.0226, over 17180.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2457, pruned_loss=0.03683, over 3305905.61 frames. ], batch size: 40, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:50:56,877 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265339.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:51:15,559 INFO [train.py:904] (7/8) Epoch 27, batch 1450, loss[loss=0.1608, simple_loss=0.2349, pruned_loss=0.04337, over 16645.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.2448, pruned_loss=0.03696, over 3309692.33 frames. ], batch size: 134, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:51:29,669 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.640e+02 2.111e+02 2.470e+02 3.011e+02 5.399e+02, threshold=4.940e+02, percent-clipped=1.0 2023-05-02 07:51:32,784 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-05-02 07:52:24,305 INFO [train.py:904] (7/8) Epoch 27, batch 1500, loss[loss=0.1652, simple_loss=0.2571, pruned_loss=0.03668, over 17231.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2447, pruned_loss=0.03679, over 3312463.34 frames. ], batch size: 46, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:52:41,855 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-05-02 07:52:48,072 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 07:53:36,980 INFO [train.py:904] (7/8) Epoch 27, batch 1550, loss[loss=0.1916, simple_loss=0.2621, pruned_loss=0.06057, over 16779.00 frames. ], tot_loss[loss=0.1604, simple_loss=0.246, pruned_loss=0.03736, over 3302010.02 frames. ], batch size: 83, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:53:45,992 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265460.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:53:49,835 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.243e+02 2.540e+02 3.021e+02 6.885e+02, threshold=5.080e+02, percent-clipped=3.0 2023-05-02 07:54:26,342 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8901, 4.6618, 4.9330, 5.0929, 5.2836, 4.6910, 5.2996, 5.2999], device='cuda:7'), covar=tensor([0.1972, 0.1445, 0.1719, 0.0773, 0.0597, 0.0987, 0.0536, 0.0661], device='cuda:7'), in_proj_covar=tensor([0.0689, 0.0848, 0.0979, 0.0859, 0.0652, 0.0677, 0.0713, 0.0829], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:54:43,804 INFO [train.py:904] (7/8) Epoch 27, batch 1600, loss[loss=0.1815, simple_loss=0.2576, pruned_loss=0.05275, over 16777.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2479, pruned_loss=0.03866, over 3307255.32 frames. ], batch size: 124, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:54:51,468 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265508.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:55:21,272 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265530.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:55:32,003 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1880, 2.4324, 2.8209, 3.1605, 2.9882, 3.6590, 2.6436, 3.5949], device='cuda:7'), covar=tensor([0.0274, 0.0482, 0.0350, 0.0342, 0.0357, 0.0216, 0.0477, 0.0210], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0196, 0.0185, 0.0191, 0.0206, 0.0165, 0.0203, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 07:55:51,088 INFO [train.py:904] (7/8) Epoch 27, batch 1650, loss[loss=0.2008, simple_loss=0.2979, pruned_loss=0.05187, over 17075.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2504, pruned_loss=0.03958, over 3310734.59 frames. ], batch size: 53, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:56:00,205 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-02 07:56:04,415 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.682e+02 2.309e+02 2.667e+02 3.273e+02 6.043e+02, threshold=5.334e+02, percent-clipped=4.0 2023-05-02 07:56:44,676 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265591.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:57:00,236 INFO [train.py:904] (7/8) Epoch 27, batch 1700, loss[loss=0.1621, simple_loss=0.2419, pruned_loss=0.0411, over 16987.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2517, pruned_loss=0.0395, over 3319186.64 frames. ], batch size: 41, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:57:16,073 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265614.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:57:42,669 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265634.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:57:43,952 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0338, 4.9643, 4.9088, 4.4004, 4.5488, 4.9304, 4.8870, 4.5833], device='cuda:7'), covar=tensor([0.0623, 0.0713, 0.0349, 0.0424, 0.1189, 0.0528, 0.0414, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0476, 0.0370, 0.0374, 0.0370, 0.0431, 0.0254, 0.0446], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 07:58:07,542 INFO [train.py:904] (7/8) Epoch 27, batch 1750, loss[loss=0.1577, simple_loss=0.2552, pruned_loss=0.03003, over 17031.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2518, pruned_loss=0.039, over 3315522.26 frames. ], batch size: 50, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:58:16,242 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0801, 5.1277, 5.5439, 5.4980, 5.5422, 5.1795, 5.1058, 4.9514], device='cuda:7'), covar=tensor([0.0348, 0.0618, 0.0352, 0.0428, 0.0490, 0.0420, 0.1078, 0.0485], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0492, 0.0473, 0.0436, 0.0525, 0.0503, 0.0577, 0.0401], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 07:58:22,126 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.457e+02 2.193e+02 2.494e+02 2.973e+02 6.444e+02, threshold=4.988e+02, percent-clipped=1.0 2023-05-02 07:58:38,525 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265675.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 07:59:15,662 INFO [train.py:904] (7/8) Epoch 27, batch 1800, loss[loss=0.1471, simple_loss=0.243, pruned_loss=0.02558, over 17117.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2523, pruned_loss=0.03866, over 3319809.00 frames. ], batch size: 49, lr: 2.51e-03, grad_scale: 8.0 2023-05-02 07:59:32,502 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 07:59:46,635 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3284, 3.3457, 3.6410, 2.4768, 3.2919, 3.7393, 3.4233, 2.0664], device='cuda:7'), covar=tensor([0.0548, 0.0161, 0.0066, 0.0421, 0.0128, 0.0098, 0.0113, 0.0546], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0089, 0.0090, 0.0135, 0.0102, 0.0114, 0.0098, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 07:59:58,479 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-05-02 08:00:23,857 INFO [train.py:904] (7/8) Epoch 27, batch 1850, loss[loss=0.1525, simple_loss=0.2471, pruned_loss=0.0289, over 17113.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2534, pruned_loss=0.03921, over 3309663.86 frames. ], batch size: 47, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:00:37,845 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.122e+02 2.552e+02 2.913e+02 6.429e+02, threshold=5.105e+02, percent-clipped=1.0 2023-05-02 08:00:55,788 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265776.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:01:02,940 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4499, 3.5897, 4.0113, 2.1533, 3.2765, 2.4217, 3.9002, 3.7484], device='cuda:7'), covar=tensor([0.0253, 0.0938, 0.0450, 0.2151, 0.0769, 0.1048, 0.0594, 0.1119], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0171, 0.0157, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:01:21,001 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9642, 3.0833, 3.2980, 2.1000, 2.9352, 2.2314, 3.5146, 3.4352], device='cuda:7'), covar=tensor([0.0230, 0.0965, 0.0652, 0.2013, 0.0898, 0.1071, 0.0546, 0.0956], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0171, 0.0157, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:01:33,282 INFO [train.py:904] (7/8) Epoch 27, batch 1900, loss[loss=0.1487, simple_loss=0.237, pruned_loss=0.03016, over 15895.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2524, pruned_loss=0.03866, over 3315522.87 frames. ], batch size: 35, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:02:42,107 INFO [train.py:904] (7/8) Epoch 27, batch 1950, loss[loss=0.1562, simple_loss=0.2443, pruned_loss=0.03404, over 16711.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2529, pruned_loss=0.03872, over 3312231.71 frames. ], batch size: 89, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:02:54,910 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.085e+02 2.462e+02 3.209e+02 6.071e+02, threshold=4.924e+02, percent-clipped=6.0 2023-05-02 08:03:08,246 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265872.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:03:27,064 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265886.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:03:45,835 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.79 vs. limit=5.0 2023-05-02 08:03:50,718 INFO [train.py:904] (7/8) Epoch 27, batch 2000, loss[loss=0.1844, simple_loss=0.2649, pruned_loss=0.05194, over 16301.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2526, pruned_loss=0.03912, over 3307927.42 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:04:10,481 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=265917.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:04:33,637 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265933.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:04:34,771 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=265934.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:05:01,218 INFO [train.py:904] (7/8) Epoch 27, batch 2050, loss[loss=0.1458, simple_loss=0.229, pruned_loss=0.03127, over 17037.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2519, pruned_loss=0.03886, over 3308946.20 frames. ], batch size: 41, lr: 2.50e-03, grad_scale: 16.0 2023-05-02 08:05:14,206 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.179e+02 2.495e+02 3.022e+02 6.334e+02, threshold=4.990e+02, percent-clipped=2.0 2023-05-02 08:05:25,400 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=265970.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 08:05:36,731 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=265978.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:05:41,863 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=265982.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:06:04,923 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6248, 3.5307, 4.0998, 2.2508, 3.2812, 2.5421, 3.9896, 3.7785], device='cuda:7'), covar=tensor([0.0237, 0.1076, 0.0476, 0.2144, 0.0844, 0.1006, 0.0593, 0.1210], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0170, 0.0157, 0.0147, 0.0132, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:06:14,061 INFO [train.py:904] (7/8) Epoch 27, batch 2100, loss[loss=0.1737, simple_loss=0.2505, pruned_loss=0.04841, over 16738.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.252, pruned_loss=0.03867, over 3298342.12 frames. ], batch size: 124, lr: 2.50e-03, grad_scale: 16.0 2023-05-02 08:06:57,149 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 08:07:06,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5944, 2.2870, 1.8261, 2.1333, 2.6296, 2.3839, 2.5183, 2.7229], device='cuda:7'), covar=tensor([0.0271, 0.0480, 0.0636, 0.0498, 0.0262, 0.0394, 0.0237, 0.0329], device='cuda:7'), in_proj_covar=tensor([0.0233, 0.0248, 0.0237, 0.0237, 0.0249, 0.0248, 0.0247, 0.0246], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:07:22,093 INFO [train.py:904] (7/8) Epoch 27, batch 2150, loss[loss=0.1647, simple_loss=0.2545, pruned_loss=0.03747, over 16811.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.253, pruned_loss=0.03933, over 3301838.27 frames. ], batch size: 96, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:07:37,031 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.561e+02 2.274e+02 2.676e+02 3.152e+02 5.314e+02, threshold=5.352e+02, percent-clipped=2.0 2023-05-02 08:07:47,123 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266071.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:08:32,979 INFO [train.py:904] (7/8) Epoch 27, batch 2200, loss[loss=0.1576, simple_loss=0.2434, pruned_loss=0.03588, over 15826.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2541, pruned_loss=0.03967, over 3305004.58 frames. ], batch size: 35, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:09:34,236 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3412, 2.2510, 2.2379, 4.0493, 2.2151, 2.5712, 2.3445, 2.3692], device='cuda:7'), covar=tensor([0.1575, 0.4352, 0.3651, 0.0683, 0.4750, 0.3115, 0.4132, 0.4387], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0474, 0.0387, 0.0338, 0.0446, 0.0542, 0.0444, 0.0553], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:09:41,316 INFO [train.py:904] (7/8) Epoch 27, batch 2250, loss[loss=0.2217, simple_loss=0.303, pruned_loss=0.07019, over 11906.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2548, pruned_loss=0.03987, over 3305455.75 frames. ], batch size: 246, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:09:56,595 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.643e+02 2.173e+02 2.503e+02 2.978e+02 4.988e+02, threshold=5.005e+02, percent-clipped=0.0 2023-05-02 08:10:28,081 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266186.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:10:48,217 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3990, 5.7631, 5.5200, 5.6495, 5.2571, 5.2154, 5.1990, 5.8962], device='cuda:7'), covar=tensor([0.1540, 0.1078, 0.1136, 0.0831, 0.0859, 0.0747, 0.1349, 0.0869], device='cuda:7'), in_proj_covar=tensor([0.0722, 0.0872, 0.0712, 0.0672, 0.0554, 0.0550, 0.0737, 0.0684], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:10:51,424 INFO [train.py:904] (7/8) Epoch 27, batch 2300, loss[loss=0.1598, simple_loss=0.2465, pruned_loss=0.03655, over 16797.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2547, pruned_loss=0.0398, over 3313698.68 frames. ], batch size: 102, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:11:06,871 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266214.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:11:23,559 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266226.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 08:11:26,476 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266228.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:11:34,365 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266234.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:11:59,658 INFO [train.py:904] (7/8) Epoch 27, batch 2350, loss[loss=0.1812, simple_loss=0.265, pruned_loss=0.04871, over 16536.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.255, pruned_loss=0.03957, over 3324898.24 frames. ], batch size: 68, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:12:14,161 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.140e+02 2.442e+02 2.999e+02 5.112e+02, threshold=4.884e+02, percent-clipped=1.0 2023-05-02 08:12:23,002 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266270.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 08:12:28,002 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266273.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:12:28,142 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9807, 2.9040, 2.5426, 4.5988, 3.3788, 4.0831, 1.7420, 2.9758], device='cuda:7'), covar=tensor([0.1365, 0.0851, 0.1356, 0.0217, 0.0278, 0.0508, 0.1697, 0.0936], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0199, 0.0200, 0.0206, 0.0219, 0.0208, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:12:30,461 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:12:40,494 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-02 08:12:46,483 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266287.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 08:13:07,775 INFO [train.py:904] (7/8) Epoch 27, batch 2400, loss[loss=0.1543, simple_loss=0.2446, pruned_loss=0.03197, over 16801.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2557, pruned_loss=0.03947, over 3314806.79 frames. ], batch size: 83, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:13:18,747 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8702, 4.5239, 3.0899, 2.3809, 2.6221, 2.7313, 4.8315, 3.5316], device='cuda:7'), covar=tensor([0.3058, 0.0473, 0.1897, 0.3035, 0.3012, 0.2086, 0.0349, 0.1531], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0276, 0.0313, 0.0326, 0.0305, 0.0276, 0.0305, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 08:13:28,959 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:14:07,855 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266346.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:14:16,222 INFO [train.py:904] (7/8) Epoch 27, batch 2450, loss[loss=0.1829, simple_loss=0.275, pruned_loss=0.04545, over 16737.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2565, pruned_loss=0.03953, over 3315830.82 frames. ], batch size: 57, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:14:18,398 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8657, 4.8519, 4.7054, 4.1446, 4.8088, 2.0065, 4.5386, 4.4642], device='cuda:7'), covar=tensor([0.0152, 0.0133, 0.0234, 0.0399, 0.0138, 0.2781, 0.0199, 0.0257], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0174, 0.0211, 0.0185, 0.0188, 0.0219, 0.0202, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:14:24,462 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2186, 5.2145, 4.9620, 4.4280, 5.1193, 1.9482, 4.8753, 4.8615], device='cuda:7'), covar=tensor([0.0140, 0.0118, 0.0255, 0.0444, 0.0112, 0.2922, 0.0177, 0.0284], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0174, 0.0211, 0.0185, 0.0188, 0.0219, 0.0202, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:14:31,086 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.203e+02 2.440e+02 2.935e+02 4.080e+02, threshold=4.880e+02, percent-clipped=0.0 2023-05-02 08:14:41,926 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266371.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:15:26,457 INFO [train.py:904] (7/8) Epoch 27, batch 2500, loss[loss=0.1586, simple_loss=0.2378, pruned_loss=0.03964, over 16814.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2567, pruned_loss=0.0395, over 3319515.61 frames. ], batch size: 102, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:15:31,596 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266407.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:15:31,652 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8360, 2.6556, 2.4565, 4.0609, 3.3140, 3.9833, 1.5465, 2.8657], device='cuda:7'), covar=tensor([0.1399, 0.0725, 0.1281, 0.0188, 0.0140, 0.0394, 0.1662, 0.0858], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0200, 0.0206, 0.0219, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:15:48,892 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266419.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:16:26,102 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-05-02 08:16:35,711 INFO [train.py:904] (7/8) Epoch 27, batch 2550, loss[loss=0.1405, simple_loss=0.2312, pruned_loss=0.02487, over 17203.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2572, pruned_loss=0.03933, over 3318226.51 frames. ], batch size: 45, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:16:51,127 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.440e+02 2.045e+02 2.342e+02 2.686e+02 5.745e+02, threshold=4.685e+02, percent-clipped=2.0 2023-05-02 08:16:52,119 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266464.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:17:45,262 INFO [train.py:904] (7/8) Epoch 27, batch 2600, loss[loss=0.1672, simple_loss=0.2537, pruned_loss=0.04037, over 16203.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2565, pruned_loss=0.03899, over 3326969.11 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:18:17,168 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266525.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:18:21,054 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266528.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:18:40,412 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-02 08:18:55,438 INFO [train.py:904] (7/8) Epoch 27, batch 2650, loss[loss=0.1773, simple_loss=0.2757, pruned_loss=0.03948, over 16667.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2561, pruned_loss=0.03806, over 3326766.60 frames. ], batch size: 62, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:19:11,570 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 2.008e+02 2.392e+02 2.897e+02 5.210e+02, threshold=4.785e+02, percent-clipped=1.0 2023-05-02 08:19:21,007 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:19:24,634 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266573.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:19:29,455 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266576.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:19:37,437 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266582.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 08:20:05,809 INFO [train.py:904] (7/8) Epoch 27, batch 2700, loss[loss=0.1483, simple_loss=0.2372, pruned_loss=0.02973, over 17007.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2559, pruned_loss=0.03756, over 3331928.13 frames. ], batch size: 41, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:20:07,074 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7173, 2.9002, 3.0982, 2.0382, 2.7323, 2.0909, 3.3496, 3.2923], device='cuda:7'), covar=tensor([0.0279, 0.0999, 0.0664, 0.2048, 0.0974, 0.1148, 0.0559, 0.0848], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0171, 0.0171, 0.0157, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:20:31,929 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266621.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:21:15,406 INFO [train.py:904] (7/8) Epoch 27, batch 2750, loss[loss=0.1669, simple_loss=0.2628, pruned_loss=0.03553, over 16453.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2561, pruned_loss=0.03712, over 3338449.96 frames. ], batch size: 75, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:21:29,201 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.109e+02 2.416e+02 2.772e+02 5.063e+02, threshold=4.832e+02, percent-clipped=2.0 2023-05-02 08:21:51,232 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266679.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:22:22,675 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:22:23,528 INFO [train.py:904] (7/8) Epoch 27, batch 2800, loss[loss=0.1473, simple_loss=0.2433, pruned_loss=0.02569, over 16990.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2557, pruned_loss=0.03685, over 3343762.57 frames. ], batch size: 41, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:22:25,175 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266704.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:23:15,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266740.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:23:31,360 INFO [train.py:904] (7/8) Epoch 27, batch 2850, loss[loss=0.1446, simple_loss=0.2322, pruned_loss=0.02845, over 16785.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2547, pruned_loss=0.0368, over 3343083.01 frames. ], batch size: 39, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:23:41,876 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 08:23:48,199 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.557e+02 2.101e+02 2.361e+02 2.749e+02 4.923e+02, threshold=4.721e+02, percent-clipped=1.0 2023-05-02 08:23:49,846 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266765.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:24:17,740 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 08:24:41,245 INFO [train.py:904] (7/8) Epoch 27, batch 2900, loss[loss=0.1539, simple_loss=0.2274, pruned_loss=0.04023, over 16736.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.253, pruned_loss=0.03662, over 3347018.27 frames. ], batch size: 83, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:25:04,250 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=266820.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:25:49,403 INFO [train.py:904] (7/8) Epoch 27, batch 2950, loss[loss=0.1758, simple_loss=0.2518, pruned_loss=0.04992, over 16719.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2534, pruned_loss=0.03766, over 3350145.47 frames. ], batch size: 134, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:26:04,414 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.609e+02 2.163e+02 2.599e+02 3.216e+02 6.007e+02, threshold=5.198e+02, percent-clipped=1.0 2023-05-02 08:26:14,835 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266870.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:26:29,636 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=266882.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 08:26:58,275 INFO [train.py:904] (7/8) Epoch 27, batch 3000, loss[loss=0.164, simple_loss=0.246, pruned_loss=0.04099, over 16643.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2537, pruned_loss=0.03794, over 3355662.42 frames. ], batch size: 89, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:26:58,275 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 08:27:03,965 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6766, 4.9489, 4.8531, 4.8907, 4.6777, 4.7057, 4.3393, 5.0086], device='cuda:7'), covar=tensor([0.1148, 0.0823, 0.0658, 0.0632, 0.0636, 0.0374, 0.1204, 0.0645], device='cuda:7'), in_proj_covar=tensor([0.0724, 0.0878, 0.0716, 0.0677, 0.0558, 0.0551, 0.0742, 0.0688], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:27:07,050 INFO [train.py:938] (7/8) Epoch 27, validation: loss=0.1336, simple_loss=0.2386, pruned_loss=0.01429, over 944034.00 frames. 2023-05-02 08:27:07,050 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 08:27:27,843 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:27:30,394 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=266920.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:27:45,320 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=266930.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 08:28:16,690 INFO [train.py:904] (7/8) Epoch 27, batch 3050, loss[loss=0.1642, simple_loss=0.2521, pruned_loss=0.03819, over 16815.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2542, pruned_loss=0.03857, over 3350832.07 frames. ], batch size: 102, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:28:30,547 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.121e+02 2.455e+02 2.792e+02 5.832e+02, threshold=4.910e+02, percent-clipped=2.0 2023-05-02 08:28:44,102 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7281, 3.8891, 2.5783, 4.4634, 2.9682, 4.3696, 2.4598, 3.1582], device='cuda:7'), covar=tensor([0.0352, 0.0416, 0.1496, 0.0365, 0.0835, 0.0514, 0.1592, 0.0799], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0184, 0.0199, 0.0177, 0.0183, 0.0225, 0.0207, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:28:55,887 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=266981.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:29:24,853 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267002.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:29:25,705 INFO [train.py:904] (7/8) Epoch 27, batch 3100, loss[loss=0.1834, simple_loss=0.2623, pruned_loss=0.05224, over 12215.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2543, pruned_loss=0.03904, over 3344361.06 frames. ], batch size: 247, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:29:54,270 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2813, 2.3761, 2.3100, 4.1620, 2.3110, 2.6672, 2.4037, 2.4622], device='cuda:7'), covar=tensor([0.1491, 0.3759, 0.3413, 0.0603, 0.4302, 0.2843, 0.3898, 0.3966], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0475, 0.0388, 0.0340, 0.0447, 0.0543, 0.0446, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:30:11,881 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267035.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:30:33,231 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267050.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:30:36,918 INFO [train.py:904] (7/8) Epoch 27, batch 3150, loss[loss=0.1515, simple_loss=0.248, pruned_loss=0.02752, over 17108.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2535, pruned_loss=0.03916, over 3340377.48 frames. ], batch size: 47, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:30:45,788 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267060.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:30:49,231 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 08:30:50,836 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.137e+02 2.489e+02 2.954e+02 4.893e+02, threshold=4.979e+02, percent-clipped=1.0 2023-05-02 08:31:15,407 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=267082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:31:29,860 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4827, 3.5918, 3.7723, 2.6808, 3.4218, 3.8308, 3.5307, 2.2597], device='cuda:7'), covar=tensor([0.0526, 0.0156, 0.0070, 0.0390, 0.0148, 0.0117, 0.0123, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0090, 0.0092, 0.0137, 0.0103, 0.0117, 0.0100, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 08:31:37,723 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=267098.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:31:43,514 INFO [train.py:904] (7/8) Epoch 27, batch 3200, loss[loss=0.1465, simple_loss=0.2392, pruned_loss=0.02694, over 17179.00 frames. ], tot_loss[loss=0.1656, simple_loss=0.2527, pruned_loss=0.03925, over 3327827.37 frames. ], batch size: 46, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:32:08,052 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267120.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:32:39,054 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=267143.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:32:50,800 INFO [train.py:904] (7/8) Epoch 27, batch 3250, loss[loss=0.1477, simple_loss=0.246, pruned_loss=0.02465, over 17101.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.252, pruned_loss=0.03877, over 3334910.52 frames. ], batch size: 49, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:33:01,502 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=267159.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:33:07,330 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.439e+02 2.100e+02 2.558e+02 2.992e+02 5.304e+02, threshold=5.115e+02, percent-clipped=1.0 2023-05-02 08:33:12,826 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267168.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:33:29,430 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-02 08:33:52,579 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7282, 4.7022, 4.6574, 4.3272, 4.3613, 4.6902, 4.5228, 4.4711], device='cuda:7'), covar=tensor([0.0712, 0.0816, 0.0346, 0.0349, 0.0895, 0.0551, 0.0500, 0.0732], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0483, 0.0376, 0.0381, 0.0376, 0.0438, 0.0258, 0.0453], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 08:34:00,067 INFO [train.py:904] (7/8) Epoch 27, batch 3300, loss[loss=0.1414, simple_loss=0.2365, pruned_loss=0.02315, over 17168.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2531, pruned_loss=0.03899, over 3334611.50 frames. ], batch size: 46, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:34:37,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6718, 2.5636, 1.9318, 2.6997, 2.1391, 2.7990, 2.1308, 2.4086], device='cuda:7'), covar=tensor([0.0381, 0.0417, 0.1375, 0.0340, 0.0686, 0.0557, 0.1342, 0.0711], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0183, 0.0198, 0.0177, 0.0182, 0.0224, 0.0206, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:35:07,435 INFO [train.py:904] (7/8) Epoch 27, batch 3350, loss[loss=0.1832, simple_loss=0.2606, pruned_loss=0.05288, over 16721.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2531, pruned_loss=0.03866, over 3328063.05 frames. ], batch size: 124, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:35:22,682 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 1.912e+02 2.198e+02 2.588e+02 3.700e+02, threshold=4.395e+02, percent-clipped=0.0 2023-05-02 08:35:38,224 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267276.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:36:05,048 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9693, 5.0704, 5.3931, 5.3757, 5.4198, 5.1062, 5.0355, 4.8998], device='cuda:7'), covar=tensor([0.0374, 0.0522, 0.0419, 0.0440, 0.0416, 0.0373, 0.0843, 0.0451], device='cuda:7'), in_proj_covar=tensor([0.0445, 0.0500, 0.0482, 0.0445, 0.0530, 0.0510, 0.0587, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 08:36:14,894 INFO [train.py:904] (7/8) Epoch 27, batch 3400, loss[loss=0.1756, simple_loss=0.2522, pruned_loss=0.04947, over 16881.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2529, pruned_loss=0.03848, over 3333443.24 frames. ], batch size: 116, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:36:58,123 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267335.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:37:22,629 INFO [train.py:904] (7/8) Epoch 27, batch 3450, loss[loss=0.1635, simple_loss=0.2612, pruned_loss=0.03285, over 17116.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2519, pruned_loss=0.03793, over 3329467.78 frames. ], batch size: 49, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:37:31,897 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267360.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:37:36,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.041e+02 2.356e+02 2.660e+02 5.077e+02, threshold=4.713e+02, percent-clipped=1.0 2023-05-02 08:38:03,486 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267383.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:38:30,924 INFO [train.py:904] (7/8) Epoch 27, batch 3500, loss[loss=0.1772, simple_loss=0.2632, pruned_loss=0.04564, over 17019.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2504, pruned_loss=0.0375, over 3321537.60 frames. ], batch size: 53, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:38:39,648 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267408.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:39:07,334 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.63 vs. limit=5.0 2023-05-02 08:39:12,327 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0919, 3.1575, 3.3225, 2.1526, 2.9136, 2.2886, 3.5649, 3.5244], device='cuda:7'), covar=tensor([0.0248, 0.1005, 0.0661, 0.2073, 0.0891, 0.1086, 0.0532, 0.0898], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0174, 0.0171, 0.0158, 0.0149, 0.0133, 0.0148, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:39:18,473 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.08 vs. limit=2.0 2023-05-02 08:39:19,832 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267438.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:39:40,534 INFO [train.py:904] (7/8) Epoch 27, batch 3550, loss[loss=0.1419, simple_loss=0.2261, pruned_loss=0.02883, over 16837.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2499, pruned_loss=0.03773, over 3322181.57 frames. ], batch size: 42, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:39:41,967 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267454.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:39:53,968 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.045e+02 2.366e+02 2.928e+02 5.405e+02, threshold=4.732e+02, percent-clipped=1.0 2023-05-02 08:40:08,333 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-02 08:40:37,293 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=267494.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:40:41,018 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 08:40:48,164 INFO [train.py:904] (7/8) Epoch 27, batch 3600, loss[loss=0.1603, simple_loss=0.2572, pruned_loss=0.03173, over 17272.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2489, pruned_loss=0.03757, over 3331112.63 frames. ], batch size: 52, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:41:11,820 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8863, 4.1256, 2.7597, 4.6717, 3.2367, 4.6337, 2.7852, 3.4308], device='cuda:7'), covar=tensor([0.0356, 0.0409, 0.1539, 0.0290, 0.0861, 0.0442, 0.1566, 0.0787], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0185, 0.0200, 0.0178, 0.0183, 0.0226, 0.0207, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:41:53,981 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3453, 2.3731, 2.5685, 4.1225, 2.3905, 2.7267, 2.4441, 2.5617], device='cuda:7'), covar=tensor([0.1521, 0.3779, 0.3108, 0.0618, 0.4152, 0.2637, 0.4076, 0.3211], device='cuda:7'), in_proj_covar=tensor([0.0425, 0.0477, 0.0389, 0.0341, 0.0449, 0.0545, 0.0448, 0.0557], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:42:00,666 INFO [train.py:904] (7/8) Epoch 27, batch 3650, loss[loss=0.1572, simple_loss=0.2366, pruned_loss=0.03891, over 16454.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2475, pruned_loss=0.03792, over 3316622.14 frames. ], batch size: 146, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:42:03,561 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=267555.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:42:16,700 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 2.009e+02 2.392e+02 2.845e+02 4.698e+02, threshold=4.785e+02, percent-clipped=1.0 2023-05-02 08:42:36,714 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267576.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:43:14,281 INFO [train.py:904] (7/8) Epoch 27, batch 3700, loss[loss=0.1683, simple_loss=0.2473, pruned_loss=0.04467, over 16184.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2463, pruned_loss=0.0392, over 3311369.74 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:43:45,317 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267624.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:43:47,603 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-05-02 08:44:20,881 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8167, 2.7772, 2.5571, 4.2109, 3.4981, 4.1256, 1.6132, 2.9871], device='cuda:7'), covar=tensor([0.1383, 0.0710, 0.1237, 0.0192, 0.0195, 0.0387, 0.1649, 0.0844], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0180, 0.0199, 0.0202, 0.0208, 0.0220, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:44:27,024 INFO [train.py:904] (7/8) Epoch 27, batch 3750, loss[loss=0.1794, simple_loss=0.2581, pruned_loss=0.05034, over 16218.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2479, pruned_loss=0.04098, over 3292193.25 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:44:42,772 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.122e+02 2.445e+02 2.969e+02 5.138e+02, threshold=4.890e+02, percent-clipped=1.0 2023-05-02 08:45:40,200 INFO [train.py:904] (7/8) Epoch 27, batch 3800, loss[loss=0.1732, simple_loss=0.2517, pruned_loss=0.04732, over 16870.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2498, pruned_loss=0.04187, over 3287837.78 frames. ], batch size: 116, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:45:41,619 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 08:46:21,654 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8485, 2.4258, 2.4396, 3.2776, 2.5689, 3.5934, 1.6454, 2.7969], device='cuda:7'), covar=tensor([0.1375, 0.0782, 0.1201, 0.0256, 0.0139, 0.0380, 0.1628, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0180, 0.0198, 0.0202, 0.0207, 0.0219, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:46:31,928 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267738.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:46:49,353 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7206, 3.5257, 3.8358, 2.1776, 3.9254, 3.9502, 3.3200, 3.0669], device='cuda:7'), covar=tensor([0.0797, 0.0275, 0.0219, 0.1185, 0.0118, 0.0234, 0.0380, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0102, 0.0139, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:46:52,912 INFO [train.py:904] (7/8) Epoch 27, batch 3850, loss[loss=0.1639, simple_loss=0.243, pruned_loss=0.04243, over 16290.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2495, pruned_loss=0.04269, over 3289994.70 frames. ], batch size: 165, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:46:54,318 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=267754.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:47:08,878 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.727e+02 2.090e+02 2.451e+02 2.947e+02 6.738e+02, threshold=4.901e+02, percent-clipped=4.0 2023-05-02 08:47:13,755 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 08:47:39,706 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267786.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:47:44,090 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0480, 5.5977, 5.7747, 5.4050, 5.5545, 6.1092, 5.6013, 5.3033], device='cuda:7'), covar=tensor([0.0926, 0.1832, 0.1772, 0.1897, 0.2145, 0.0795, 0.1365, 0.2349], device='cuda:7'), in_proj_covar=tensor([0.0437, 0.0646, 0.0710, 0.0527, 0.0703, 0.0737, 0.0552, 0.0701], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 08:48:02,761 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=267802.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:48:03,556 INFO [train.py:904] (7/8) Epoch 27, batch 3900, loss[loss=0.1447, simple_loss=0.2228, pruned_loss=0.03325, over 16840.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2493, pruned_loss=0.04321, over 3299140.92 frames. ], batch size: 96, lr: 2.50e-03, grad_scale: 8.0 2023-05-02 08:48:10,246 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8575, 2.9345, 3.2486, 2.0896, 2.8938, 2.2428, 3.4194, 3.3188], device='cuda:7'), covar=tensor([0.0257, 0.0986, 0.0621, 0.2017, 0.0868, 0.1040, 0.0524, 0.0885], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0173, 0.0171, 0.0157, 0.0149, 0.0133, 0.0147, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:49:11,726 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=267850.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:49:14,866 INFO [train.py:904] (7/8) Epoch 27, batch 3950, loss[loss=0.1652, simple_loss=0.2437, pruned_loss=0.04333, over 16391.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2486, pruned_loss=0.04363, over 3292045.89 frames. ], batch size: 146, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:49:23,523 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1852, 3.9874, 4.0365, 4.3997, 4.4481, 4.1089, 4.2605, 4.4484], device='cuda:7'), covar=tensor([0.1765, 0.1540, 0.1944, 0.0873, 0.0976, 0.1755, 0.2641, 0.1302], device='cuda:7'), in_proj_covar=tensor([0.0699, 0.0860, 0.0999, 0.0870, 0.0662, 0.0691, 0.0722, 0.0841], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:49:26,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4980, 3.4721, 3.5138, 3.5894, 3.6454, 3.3615, 3.5885, 3.7143], device='cuda:7'), covar=tensor([0.1267, 0.0918, 0.1056, 0.0602, 0.0665, 0.2482, 0.1458, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0698, 0.0860, 0.0999, 0.0870, 0.0662, 0.0691, 0.0722, 0.0841], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:49:32,331 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.233e+02 2.530e+02 3.152e+02 5.510e+02, threshold=5.059e+02, percent-clipped=1.0 2023-05-02 08:49:56,351 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5785, 3.6447, 2.3754, 3.8166, 2.9267, 3.7804, 2.4016, 2.9855], device='cuda:7'), covar=tensor([0.0264, 0.0373, 0.1449, 0.0328, 0.0746, 0.0825, 0.1399, 0.0679], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0183, 0.0198, 0.0176, 0.0181, 0.0223, 0.0205, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:50:12,374 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.72 vs. limit=5.0 2023-05-02 08:50:28,773 INFO [train.py:904] (7/8) Epoch 27, batch 4000, loss[loss=0.1697, simple_loss=0.2546, pruned_loss=0.0424, over 16314.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.25, pruned_loss=0.04436, over 3286602.03 frames. ], batch size: 165, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:51:42,557 INFO [train.py:904] (7/8) Epoch 27, batch 4050, loss[loss=0.1767, simple_loss=0.2571, pruned_loss=0.04818, over 16722.00 frames. ], tot_loss[loss=0.1692, simple_loss=0.2508, pruned_loss=0.04377, over 3287051.20 frames. ], batch size: 124, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:51:48,666 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4218, 2.8046, 2.9961, 1.9921, 2.7091, 2.0847, 3.0995, 3.1170], device='cuda:7'), covar=tensor([0.0303, 0.0861, 0.0740, 0.2169, 0.0982, 0.1102, 0.0591, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0173, 0.0171, 0.0157, 0.0149, 0.0133, 0.0147, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 08:51:58,304 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.841e+02 2.143e+02 2.539e+02 4.386e+02, threshold=4.286e+02, percent-clipped=0.0 2023-05-02 08:52:42,160 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7807, 3.8671, 2.5902, 4.4979, 3.1312, 4.4706, 2.6916, 3.2109], device='cuda:7'), covar=tensor([0.0290, 0.0350, 0.1541, 0.0137, 0.0690, 0.0365, 0.1390, 0.0718], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0182, 0.0197, 0.0176, 0.0181, 0.0223, 0.0205, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:52:59,895 INFO [train.py:904] (7/8) Epoch 27, batch 4100, loss[loss=0.1827, simple_loss=0.2693, pruned_loss=0.04804, over 16769.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2522, pruned_loss=0.04318, over 3281882.08 frames. ], batch size: 124, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 08:53:29,081 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0135, 2.4258, 1.9427, 2.1483, 2.7565, 2.3429, 2.6623, 2.8924], device='cuda:7'), covar=tensor([0.0193, 0.0480, 0.0657, 0.0558, 0.0317, 0.0447, 0.0309, 0.0310], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0247, 0.0237, 0.0237, 0.0250, 0.0248, 0.0249, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 08:54:16,176 INFO [train.py:904] (7/8) Epoch 27, batch 4150, loss[loss=0.2027, simple_loss=0.2913, pruned_loss=0.05712, over 16879.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2583, pruned_loss=0.04497, over 3257993.53 frames. ], batch size: 116, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:54:33,172 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.031e+02 2.466e+02 3.151e+02 6.245e+02, threshold=4.932e+02, percent-clipped=8.0 2023-05-02 08:54:57,701 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5882, 3.6984, 2.7479, 2.3155, 2.4608, 2.4959, 4.0122, 3.2965], device='cuda:7'), covar=tensor([0.3063, 0.0655, 0.1916, 0.2984, 0.2782, 0.2076, 0.0465, 0.1405], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0278, 0.0315, 0.0329, 0.0309, 0.0277, 0.0306, 0.0354], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 08:55:32,122 INFO [train.py:904] (7/8) Epoch 27, batch 4200, loss[loss=0.1995, simple_loss=0.2874, pruned_loss=0.05576, over 16806.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.265, pruned_loss=0.04622, over 3230902.68 frames. ], batch size: 39, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:55:59,953 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 08:56:41,009 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=268150.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:56:44,673 INFO [train.py:904] (7/8) Epoch 27, batch 4250, loss[loss=0.1768, simple_loss=0.2682, pruned_loss=0.04268, over 16770.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2684, pruned_loss=0.04616, over 3198918.79 frames. ], batch size: 39, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:57:00,875 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.686e+02 2.148e+02 2.458e+02 3.113e+02 4.874e+02, threshold=4.916e+02, percent-clipped=0.0 2023-05-02 08:57:50,362 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=268198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:57:57,376 INFO [train.py:904] (7/8) Epoch 27, batch 4300, loss[loss=0.1942, simple_loss=0.2968, pruned_loss=0.04579, over 16638.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2694, pruned_loss=0.04519, over 3198024.41 frames. ], batch size: 57, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:58:10,949 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268212.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 08:58:30,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6445, 3.5554, 3.8927, 2.0057, 4.0630, 4.0564, 3.0451, 3.2173], device='cuda:7'), covar=tensor([0.0839, 0.0278, 0.0235, 0.1249, 0.0089, 0.0153, 0.0513, 0.0429], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0111, 0.0101, 0.0139, 0.0087, 0.0131, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 08:59:12,227 INFO [train.py:904] (7/8) Epoch 27, batch 4350, loss[loss=0.2016, simple_loss=0.2903, pruned_loss=0.0565, over 16681.00 frames. ], tot_loss[loss=0.1834, simple_loss=0.2734, pruned_loss=0.04672, over 3200616.43 frames. ], batch size: 57, lr: 2.49e-03, grad_scale: 16.0 2023-05-02 08:59:27,937 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.120e+02 2.432e+02 2.957e+02 3.904e+02, threshold=4.863e+02, percent-clipped=0.0 2023-05-02 08:59:43,211 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268273.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 08:59:45,598 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:00:27,006 INFO [train.py:904] (7/8) Epoch 27, batch 4400, loss[loss=0.1989, simple_loss=0.2898, pruned_loss=0.05398, over 16900.00 frames. ], tot_loss[loss=0.1854, simple_loss=0.2753, pruned_loss=0.04772, over 3200709.32 frames. ], batch size: 109, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:01:14,948 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268336.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:01:37,973 INFO [train.py:904] (7/8) Epoch 27, batch 4450, loss[loss=0.1877, simple_loss=0.2875, pruned_loss=0.04394, over 16799.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2791, pruned_loss=0.0495, over 3193916.17 frames. ], batch size: 102, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:01:55,125 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.500e+02 1.898e+02 2.329e+02 2.751e+02 5.089e+02, threshold=4.658e+02, percent-clipped=1.0 2023-05-02 09:02:36,957 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6334, 4.9738, 5.2691, 5.2020, 5.2782, 4.8978, 4.6082, 4.5958], device='cuda:7'), covar=tensor([0.0576, 0.0624, 0.0481, 0.0569, 0.0620, 0.0573, 0.1484, 0.0637], device='cuda:7'), in_proj_covar=tensor([0.0432, 0.0487, 0.0469, 0.0432, 0.0517, 0.0496, 0.0571, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 09:02:50,825 INFO [train.py:904] (7/8) Epoch 27, batch 4500, loss[loss=0.1807, simple_loss=0.2724, pruned_loss=0.04451, over 16962.00 frames. ], tot_loss[loss=0.1902, simple_loss=0.2798, pruned_loss=0.05032, over 3214831.72 frames. ], batch size: 41, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:04:03,669 INFO [train.py:904] (7/8) Epoch 27, batch 4550, loss[loss=0.1917, simple_loss=0.2787, pruned_loss=0.05239, over 16660.00 frames. ], tot_loss[loss=0.1917, simple_loss=0.2804, pruned_loss=0.05149, over 3218649.18 frames. ], batch size: 57, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:04:20,728 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 1.908e+02 2.230e+02 2.489e+02 1.268e+03, threshold=4.461e+02, percent-clipped=3.0 2023-05-02 09:05:04,936 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7172, 2.8545, 2.8327, 4.7132, 3.6066, 4.0842, 1.7827, 3.0825], device='cuda:7'), covar=tensor([0.1359, 0.0754, 0.1040, 0.0108, 0.0240, 0.0340, 0.1568, 0.0761], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0179, 0.0198, 0.0200, 0.0207, 0.0218, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 09:05:15,501 INFO [train.py:904] (7/8) Epoch 27, batch 4600, loss[loss=0.1872, simple_loss=0.2748, pruned_loss=0.04983, over 16748.00 frames. ], tot_loss[loss=0.192, simple_loss=0.2811, pruned_loss=0.05149, over 3217977.35 frames. ], batch size: 124, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:06:23,488 INFO [train.py:904] (7/8) Epoch 27, batch 4650, loss[loss=0.1846, simple_loss=0.2799, pruned_loss=0.04461, over 16714.00 frames. ], tot_loss[loss=0.1926, simple_loss=0.2811, pruned_loss=0.05205, over 3209466.69 frames. ], batch size: 124, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:06:40,835 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.843e+02 2.064e+02 2.500e+02 6.008e+02, threshold=4.128e+02, percent-clipped=2.0 2023-05-02 09:06:45,259 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=268568.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 09:07:32,883 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268601.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:07:34,898 INFO [train.py:904] (7/8) Epoch 27, batch 4700, loss[loss=0.192, simple_loss=0.2881, pruned_loss=0.04795, over 15378.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2783, pruned_loss=0.05085, over 3213271.16 frames. ], batch size: 190, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:08:15,786 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=268631.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:08:17,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5554, 4.6421, 4.4460, 4.0997, 4.1099, 4.5299, 4.2844, 4.2542], device='cuda:7'), covar=tensor([0.0611, 0.0587, 0.0304, 0.0321, 0.1002, 0.0632, 0.0559, 0.0659], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0465, 0.0362, 0.0368, 0.0364, 0.0420, 0.0248, 0.0434], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:08:45,868 INFO [train.py:904] (7/8) Epoch 27, batch 4750, loss[loss=0.2079, simple_loss=0.2854, pruned_loss=0.0652, over 12112.00 frames. ], tot_loss[loss=0.1863, simple_loss=0.2745, pruned_loss=0.04903, over 3195493.58 frames. ], batch size: 249, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:09:00,258 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268662.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:09:04,569 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 1.855e+02 2.157e+02 2.478e+02 7.312e+02, threshold=4.313e+02, percent-clipped=3.0 2023-05-02 09:09:21,980 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7165, 4.6832, 4.5663, 3.4649, 4.6589, 1.5226, 4.3423, 4.1066], device='cuda:7'), covar=tensor([0.0142, 0.0166, 0.0212, 0.0719, 0.0128, 0.3608, 0.0184, 0.0408], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0175, 0.0212, 0.0187, 0.0189, 0.0219, 0.0202, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:09:59,310 INFO [train.py:904] (7/8) Epoch 27, batch 4800, loss[loss=0.1567, simple_loss=0.2496, pruned_loss=0.03192, over 16745.00 frames. ], tot_loss[loss=0.1818, simple_loss=0.2704, pruned_loss=0.04665, over 3202814.90 frames. ], batch size: 76, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:10:29,565 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.07 vs. limit=5.0 2023-05-02 09:11:14,198 INFO [train.py:904] (7/8) Epoch 27, batch 4850, loss[loss=0.2086, simple_loss=0.2928, pruned_loss=0.06216, over 11612.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2708, pruned_loss=0.04566, over 3201770.98 frames. ], batch size: 246, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:11:31,502 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 1.812e+02 2.075e+02 2.444e+02 6.308e+02, threshold=4.151e+02, percent-clipped=1.0 2023-05-02 09:12:27,703 INFO [train.py:904] (7/8) Epoch 27, batch 4900, loss[loss=0.1693, simple_loss=0.2575, pruned_loss=0.04051, over 17043.00 frames. ], tot_loss[loss=0.179, simple_loss=0.2696, pruned_loss=0.0442, over 3202547.36 frames. ], batch size: 55, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:13:20,093 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.88 vs. limit=5.0 2023-05-02 09:13:37,970 INFO [train.py:904] (7/8) Epoch 27, batch 4950, loss[loss=0.1818, simple_loss=0.2774, pruned_loss=0.04306, over 15310.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2689, pruned_loss=0.04354, over 3196113.31 frames. ], batch size: 190, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:13:54,420 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.970e+02 2.304e+02 2.682e+02 4.795e+02, threshold=4.609e+02, percent-clipped=2.0 2023-05-02 09:13:59,134 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=268868.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 09:14:36,936 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:14:50,673 INFO [train.py:904] (7/8) Epoch 27, batch 5000, loss[loss=0.1962, simple_loss=0.2875, pruned_loss=0.05245, over 16943.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2703, pruned_loss=0.04351, over 3213203.20 frames. ], batch size: 109, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:15:09,314 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=268916.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:15:30,384 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=268931.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:00,161 INFO [train.py:904] (7/8) Epoch 27, batch 5050, loss[loss=0.1644, simple_loss=0.2613, pruned_loss=0.03375, over 15447.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2714, pruned_loss=0.04349, over 3216510.55 frames. ], batch size: 190, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:16:03,835 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=268955.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:05,793 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=268957.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:17,236 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 1.970e+02 2.360e+02 2.844e+02 4.584e+02, threshold=4.719e+02, percent-clipped=0.0 2023-05-02 09:16:37,103 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=268979.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:16:54,215 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=268991.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:17:11,200 INFO [train.py:904] (7/8) Epoch 27, batch 5100, loss[loss=0.177, simple_loss=0.2679, pruned_loss=0.04301, over 16451.00 frames. ], tot_loss[loss=0.1781, simple_loss=0.27, pruned_loss=0.04306, over 3219514.01 frames. ], batch size: 146, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:17:26,923 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0273, 2.2801, 2.2867, 3.7002, 2.1219, 2.5985, 2.3378, 2.3959], device='cuda:7'), covar=tensor([0.1556, 0.3468, 0.2988, 0.0611, 0.4115, 0.2528, 0.3590, 0.3255], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0474, 0.0384, 0.0337, 0.0446, 0.0541, 0.0444, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:17:40,245 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 09:18:12,953 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2023-05-02 09:18:22,951 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269052.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:18:23,596 INFO [train.py:904] (7/8) Epoch 27, batch 5150, loss[loss=0.188, simple_loss=0.2754, pruned_loss=0.05033, over 16468.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2695, pruned_loss=0.04218, over 3221076.93 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:18:41,483 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 1.963e+02 2.152e+02 2.508e+02 3.926e+02, threshold=4.304e+02, percent-clipped=0.0 2023-05-02 09:19:07,828 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9391, 3.8542, 4.3706, 2.0840, 4.5660, 4.5424, 3.2846, 3.2670], device='cuda:7'), covar=tensor([0.0807, 0.0255, 0.0157, 0.1301, 0.0063, 0.0126, 0.0390, 0.0500], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0101, 0.0141, 0.0087, 0.0131, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 09:19:36,087 INFO [train.py:904] (7/8) Epoch 27, batch 5200, loss[loss=0.1494, simple_loss=0.2452, pruned_loss=0.02679, over 16862.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2678, pruned_loss=0.04157, over 3218360.26 frames. ], batch size: 90, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:19:40,736 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2727, 5.5697, 5.2908, 5.3767, 5.1304, 5.0727, 4.9261, 5.6687], device='cuda:7'), covar=tensor([0.1378, 0.0902, 0.1006, 0.0821, 0.0831, 0.0767, 0.1276, 0.0908], device='cuda:7'), in_proj_covar=tensor([0.0710, 0.0862, 0.0703, 0.0662, 0.0547, 0.0543, 0.0726, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:20:05,710 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9239, 2.9231, 2.5536, 4.9456, 3.6099, 4.2555, 1.7356, 3.0083], device='cuda:7'), covar=tensor([0.1331, 0.0816, 0.1314, 0.0128, 0.0306, 0.0378, 0.1677, 0.0872], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0179, 0.0199, 0.0200, 0.0206, 0.0217, 0.0208, 0.0197], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 09:20:47,203 INFO [train.py:904] (7/8) Epoch 27, batch 5250, loss[loss=0.1576, simple_loss=0.2518, pruned_loss=0.03172, over 16527.00 frames. ], tot_loss[loss=0.1745, simple_loss=0.2658, pruned_loss=0.04162, over 3213235.24 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:21:04,388 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 1.922e+02 2.287e+02 2.667e+02 4.356e+02, threshold=4.574e+02, percent-clipped=2.0 2023-05-02 09:21:19,263 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 09:22:00,569 INFO [train.py:904] (7/8) Epoch 27, batch 5300, loss[loss=0.1444, simple_loss=0.2328, pruned_loss=0.02799, over 16544.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2623, pruned_loss=0.04071, over 3215915.05 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:22:08,779 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0223, 4.4234, 3.3205, 2.6515, 2.9786, 2.8517, 4.8785, 3.7829], device='cuda:7'), covar=tensor([0.2738, 0.0580, 0.1703, 0.2682, 0.2698, 0.1886, 0.0355, 0.1309], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0276, 0.0313, 0.0326, 0.0305, 0.0275, 0.0305, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 09:22:55,481 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269241.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:23:08,886 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269250.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:23:12,969 INFO [train.py:904] (7/8) Epoch 27, batch 5350, loss[loss=0.1796, simple_loss=0.2771, pruned_loss=0.04104, over 15378.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2615, pruned_loss=0.04039, over 3221954.04 frames. ], batch size: 190, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:23:18,878 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269257.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:23:29,781 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.050e+02 2.380e+02 2.740e+02 5.067e+02, threshold=4.761e+02, percent-clipped=1.0 2023-05-02 09:24:09,718 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1973, 4.2825, 4.1079, 3.8090, 3.8080, 4.2010, 3.8880, 3.9934], device='cuda:7'), covar=tensor([0.0636, 0.0622, 0.0304, 0.0308, 0.0782, 0.0520, 0.0932, 0.0553], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0465, 0.0361, 0.0366, 0.0361, 0.0421, 0.0247, 0.0432], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:24:24,476 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269302.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:24:25,142 INFO [train.py:904] (7/8) Epoch 27, batch 5400, loss[loss=0.1975, simple_loss=0.2815, pruned_loss=0.05672, over 11965.00 frames. ], tot_loss[loss=0.1725, simple_loss=0.2637, pruned_loss=0.04064, over 3209810.53 frames. ], batch size: 247, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:24:28,475 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269305.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:25:31,347 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269347.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:25:41,703 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269352.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:25:42,381 INFO [train.py:904] (7/8) Epoch 27, batch 5450, loss[loss=0.1783, simple_loss=0.2732, pruned_loss=0.04168, over 17189.00 frames. ], tot_loss[loss=0.1758, simple_loss=0.2669, pruned_loss=0.04233, over 3195291.33 frames. ], batch size: 44, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:26:01,151 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.011e+02 2.409e+02 3.005e+02 5.920e+02, threshold=4.819e+02, percent-clipped=1.0 2023-05-02 09:26:04,991 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2854, 4.2890, 4.1901, 3.4058, 4.2598, 1.7119, 4.0167, 3.7715], device='cuda:7'), covar=tensor([0.0140, 0.0129, 0.0217, 0.0421, 0.0113, 0.3050, 0.0163, 0.0361], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0173, 0.0210, 0.0185, 0.0186, 0.0216, 0.0199, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:27:00,966 INFO [train.py:904] (7/8) Epoch 27, batch 5500, loss[loss=0.248, simple_loss=0.3201, pruned_loss=0.08792, over 11800.00 frames. ], tot_loss[loss=0.1833, simple_loss=0.2738, pruned_loss=0.04638, over 3172124.29 frames. ], batch size: 247, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:27:16,724 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269413.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:28:00,691 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269441.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:28:19,429 INFO [train.py:904] (7/8) Epoch 27, batch 5550, loss[loss=0.2516, simple_loss=0.32, pruned_loss=0.09166, over 10861.00 frames. ], tot_loss[loss=0.1908, simple_loss=0.2804, pruned_loss=0.05062, over 3145622.50 frames. ], batch size: 247, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:28:38,501 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.260e+02 3.053e+02 3.491e+02 4.207e+02 9.161e+02, threshold=6.983e+02, percent-clipped=6.0 2023-05-02 09:29:16,026 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-05-02 09:29:37,456 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269502.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:29:38,101 INFO [train.py:904] (7/8) Epoch 27, batch 5600, loss[loss=0.2097, simple_loss=0.2951, pruned_loss=0.06212, over 16626.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2853, pruned_loss=0.05495, over 3112456.63 frames. ], batch size: 57, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:30:56,816 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269550.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:31:01,873 INFO [train.py:904] (7/8) Epoch 27, batch 5650, loss[loss=0.2572, simple_loss=0.3234, pruned_loss=0.09553, over 11377.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2905, pruned_loss=0.05902, over 3075672.17 frames. ], batch size: 249, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:31:03,722 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8760, 2.2348, 2.4688, 3.1263, 2.2545, 2.4595, 2.3906, 2.3491], device='cuda:7'), covar=tensor([0.1486, 0.3160, 0.2345, 0.0778, 0.3873, 0.2109, 0.3049, 0.3022], device='cuda:7'), in_proj_covar=tensor([0.0418, 0.0468, 0.0380, 0.0333, 0.0442, 0.0536, 0.0439, 0.0548], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:31:20,026 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.426e+02 3.478e+02 4.295e+02 5.160e+02 1.255e+03, threshold=8.591e+02, percent-clipped=5.0 2023-05-02 09:31:54,319 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3643, 2.5780, 2.1059, 2.3377, 2.9305, 2.5681, 2.9640, 3.1004], device='cuda:7'), covar=tensor([0.0151, 0.0415, 0.0607, 0.0488, 0.0291, 0.0432, 0.0282, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0225, 0.0240, 0.0230, 0.0231, 0.0241, 0.0240, 0.0238, 0.0239], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:32:02,375 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6969, 2.6808, 2.3470, 3.9299, 2.7579, 3.8964, 1.5604, 2.8852], device='cuda:7'), covar=tensor([0.1420, 0.0793, 0.1352, 0.0189, 0.0224, 0.0435, 0.1751, 0.0826], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0179, 0.0199, 0.0200, 0.0207, 0.0218, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 09:32:11,865 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:32:13,073 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269598.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:32:18,908 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 09:32:20,979 INFO [train.py:904] (7/8) Epoch 27, batch 5700, loss[loss=0.2843, simple_loss=0.3477, pruned_loss=0.1104, over 10819.00 frames. ], tot_loss[loss=0.2075, simple_loss=0.2925, pruned_loss=0.06118, over 3057039.65 frames. ], batch size: 248, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:33:32,065 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269647.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:33:41,285 INFO [train.py:904] (7/8) Epoch 27, batch 5750, loss[loss=0.209, simple_loss=0.2968, pruned_loss=0.06059, over 16242.00 frames. ], tot_loss[loss=0.2106, simple_loss=0.2953, pruned_loss=0.06291, over 3051653.10 frames. ], batch size: 165, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:33:58,768 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 09:33:59,196 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 2.911e+02 3.444e+02 4.109e+02 8.393e+02, threshold=6.889e+02, percent-clipped=0.0 2023-05-02 09:34:35,888 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.87 vs. limit=5.0 2023-05-02 09:34:50,304 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269695.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:35:02,102 INFO [train.py:904] (7/8) Epoch 27, batch 5800, loss[loss=0.1876, simple_loss=0.2816, pruned_loss=0.04681, over 16305.00 frames. ], tot_loss[loss=0.2083, simple_loss=0.2942, pruned_loss=0.0612, over 3057671.95 frames. ], batch size: 165, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:35:11,570 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269708.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:35:24,211 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 09:36:21,260 INFO [train.py:904] (7/8) Epoch 27, batch 5850, loss[loss=0.2261, simple_loss=0.2899, pruned_loss=0.08114, over 11410.00 frames. ], tot_loss[loss=0.2062, simple_loss=0.2923, pruned_loss=0.06006, over 3045249.01 frames. ], batch size: 246, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:36:36,757 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9225, 2.1672, 2.3191, 3.3754, 2.1170, 2.4050, 2.2839, 2.2984], device='cuda:7'), covar=tensor([0.1541, 0.3468, 0.2875, 0.0715, 0.4199, 0.2556, 0.3527, 0.3316], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0468, 0.0379, 0.0333, 0.0441, 0.0536, 0.0438, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:36:40,943 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.085e+02 3.568e+02 4.510e+02 7.349e+02, threshold=7.136e+02, percent-clipped=2.0 2023-05-02 09:36:45,404 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0898, 2.2806, 2.2795, 3.7239, 2.1728, 2.5925, 2.3290, 2.3912], device='cuda:7'), covar=tensor([0.1525, 0.3569, 0.3108, 0.0612, 0.4134, 0.2577, 0.3705, 0.3198], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0468, 0.0379, 0.0333, 0.0441, 0.0536, 0.0438, 0.0548], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:36:53,627 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:37:35,938 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=269797.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 09:37:45,800 INFO [train.py:904] (7/8) Epoch 27, batch 5900, loss[loss=0.2164, simple_loss=0.2931, pruned_loss=0.06983, over 11991.00 frames. ], tot_loss[loss=0.2061, simple_loss=0.2918, pruned_loss=0.06021, over 3033411.48 frames. ], batch size: 246, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:37:59,509 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8794, 3.1853, 3.4089, 2.0599, 2.9955, 2.2330, 3.4283, 3.4049], device='cuda:7'), covar=tensor([0.0255, 0.0832, 0.0569, 0.2073, 0.0832, 0.1023, 0.0571, 0.0830], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0171, 0.0170, 0.0157, 0.0149, 0.0133, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 09:38:38,104 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269834.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:38:40,728 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=269836.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:39:00,118 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0512, 2.4115, 2.5721, 1.9290, 2.6753, 2.7460, 2.4786, 2.4135], device='cuda:7'), covar=tensor([0.0702, 0.0283, 0.0260, 0.0970, 0.0152, 0.0348, 0.0480, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0110, 0.0101, 0.0139, 0.0086, 0.0131, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 09:39:06,834 INFO [train.py:904] (7/8) Epoch 27, batch 5950, loss[loss=0.1918, simple_loss=0.2885, pruned_loss=0.0475, over 16522.00 frames. ], tot_loss[loss=0.2049, simple_loss=0.2926, pruned_loss=0.05862, over 3059938.57 frames. ], batch size: 75, lr: 2.49e-03, grad_scale: 4.0 2023-05-02 09:39:27,601 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.704e+02 3.273e+02 4.077e+02 6.463e+02, threshold=6.547e+02, percent-clipped=0.0 2023-05-02 09:39:31,955 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3714, 3.2847, 3.3772, 3.4915, 3.5177, 3.3094, 3.4806, 3.5636], device='cuda:7'), covar=tensor([0.1246, 0.1114, 0.1154, 0.0700, 0.0797, 0.2422, 0.1283, 0.1026], device='cuda:7'), in_proj_covar=tensor([0.0664, 0.0815, 0.0945, 0.0825, 0.0627, 0.0659, 0.0685, 0.0800], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:40:17,765 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=269897.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:40:17,842 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=269897.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:40:24,916 INFO [train.py:904] (7/8) Epoch 27, batch 6000, loss[loss=0.1972, simple_loss=0.2905, pruned_loss=0.05191, over 17050.00 frames. ], tot_loss[loss=0.2031, simple_loss=0.2909, pruned_loss=0.05771, over 3082004.66 frames. ], batch size: 55, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:40:24,917 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 09:40:35,107 INFO [train.py:938] (7/8) Epoch 27, validation: loss=0.148, simple_loss=0.2603, pruned_loss=0.01783, over 944034.00 frames. 2023-05-02 09:40:35,108 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 09:41:30,000 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4864, 3.4516, 2.7062, 2.1893, 2.3061, 2.3034, 3.6209, 3.1452], device='cuda:7'), covar=tensor([0.3134, 0.0772, 0.2003, 0.3035, 0.2683, 0.2351, 0.0585, 0.1353], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0276, 0.0313, 0.0327, 0.0305, 0.0275, 0.0305, 0.0351], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 09:41:37,595 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=269945.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:41:52,122 INFO [train.py:904] (7/8) Epoch 27, batch 6050, loss[loss=0.1972, simple_loss=0.2943, pruned_loss=0.05007, over 16491.00 frames. ], tot_loss[loss=0.2029, simple_loss=0.2907, pruned_loss=0.05754, over 3090642.22 frames. ], batch size: 68, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:42:12,241 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.929e+02 2.707e+02 3.068e+02 3.739e+02 6.756e+02, threshold=6.136e+02, percent-clipped=2.0 2023-05-02 09:43:13,694 INFO [train.py:904] (7/8) Epoch 27, batch 6100, loss[loss=0.1652, simple_loss=0.2604, pruned_loss=0.03503, over 16856.00 frames. ], tot_loss[loss=0.2012, simple_loss=0.2895, pruned_loss=0.05647, over 3084118.56 frames. ], batch size: 42, lr: 2.49e-03, grad_scale: 8.0 2023-05-02 09:43:23,125 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270008.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:43:42,432 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.61 vs. limit=2.0 2023-05-02 09:43:43,777 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5585, 4.5838, 4.9049, 4.8767, 4.8942, 4.6055, 4.5757, 4.4609], device='cuda:7'), covar=tensor([0.0338, 0.0624, 0.0403, 0.0402, 0.0511, 0.0433, 0.0913, 0.0535], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0482, 0.0467, 0.0428, 0.0516, 0.0493, 0.0569, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 09:44:33,205 INFO [train.py:904] (7/8) Epoch 27, batch 6150, loss[loss=0.185, simple_loss=0.2674, pruned_loss=0.05129, over 11482.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.2875, pruned_loss=0.05589, over 3090727.02 frames. ], batch size: 246, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:44:37,805 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270056.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:44:53,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.916e+02 2.602e+02 3.189e+02 4.066e+02 6.857e+02, threshold=6.378e+02, percent-clipped=4.0 2023-05-02 09:45:07,879 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-05-02 09:45:42,287 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270097.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 09:45:51,125 INFO [train.py:904] (7/8) Epoch 27, batch 6200, loss[loss=0.2057, simple_loss=0.2973, pruned_loss=0.0571, over 16716.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.286, pruned_loss=0.05554, over 3087852.08 frames. ], batch size: 134, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:46:33,700 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270129.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 09:46:58,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270145.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:47:09,694 INFO [train.py:904] (7/8) Epoch 27, batch 6250, loss[loss=0.1867, simple_loss=0.289, pruned_loss=0.04217, over 16907.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2861, pruned_loss=0.05606, over 3074602.75 frames. ], batch size: 96, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:47:29,204 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.849e+02 2.710e+02 3.440e+02 4.218e+02 8.566e+02, threshold=6.880e+02, percent-clipped=4.0 2023-05-02 09:48:08,667 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:48:18,296 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0894, 4.1662, 4.4276, 4.4122, 4.4247, 4.1773, 4.1711, 4.1531], device='cuda:7'), covar=tensor([0.0372, 0.0676, 0.0425, 0.0438, 0.0433, 0.0477, 0.0846, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0481, 0.0468, 0.0429, 0.0516, 0.0493, 0.0567, 0.0395], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 09:48:25,154 INFO [train.py:904] (7/8) Epoch 27, batch 6300, loss[loss=0.1903, simple_loss=0.2822, pruned_loss=0.0492, over 16778.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2858, pruned_loss=0.0555, over 3068490.97 frames. ], batch size: 83, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:49:04,635 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3919, 3.3413, 3.4263, 3.4893, 3.5245, 3.2713, 3.5064, 3.5767], device='cuda:7'), covar=tensor([0.1287, 0.0932, 0.0952, 0.0661, 0.0661, 0.2402, 0.1025, 0.0821], device='cuda:7'), in_proj_covar=tensor([0.0668, 0.0817, 0.0949, 0.0830, 0.0630, 0.0662, 0.0688, 0.0804], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:49:33,388 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0454, 5.3402, 5.1610, 5.1139, 4.8649, 4.8643, 4.7578, 5.4610], device='cuda:7'), covar=tensor([0.1307, 0.0946, 0.0998, 0.0907, 0.0863, 0.0878, 0.1211, 0.0975], device='cuda:7'), in_proj_covar=tensor([0.0706, 0.0855, 0.0696, 0.0658, 0.0539, 0.0538, 0.0718, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:49:45,985 INFO [train.py:904] (7/8) Epoch 27, batch 6350, loss[loss=0.2301, simple_loss=0.3001, pruned_loss=0.08006, over 11776.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2863, pruned_loss=0.0566, over 3044204.51 frames. ], batch size: 246, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:50:05,609 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.880e+02 2.767e+02 3.274e+02 3.776e+02 6.685e+02, threshold=6.549e+02, percent-clipped=0.0 2023-05-02 09:50:28,890 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270280.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 09:51:02,484 INFO [train.py:904] (7/8) Epoch 27, batch 6400, loss[loss=0.2034, simple_loss=0.2893, pruned_loss=0.05874, over 16744.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2863, pruned_loss=0.05743, over 3039119.95 frames. ], batch size: 89, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:52:01,752 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270341.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 09:52:19,101 INFO [train.py:904] (7/8) Epoch 27, batch 6450, loss[loss=0.1764, simple_loss=0.2823, pruned_loss=0.03525, over 16867.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.286, pruned_loss=0.05642, over 3059206.28 frames. ], batch size: 102, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:52:25,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4618, 4.4392, 4.3364, 3.5310, 4.4110, 1.8075, 4.1521, 3.8612], device='cuda:7'), covar=tensor([0.0119, 0.0108, 0.0185, 0.0335, 0.0092, 0.2898, 0.0146, 0.0300], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0173, 0.0211, 0.0186, 0.0186, 0.0216, 0.0200, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:52:39,050 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 2.688e+02 3.236e+02 4.371e+02 9.664e+02, threshold=6.472e+02, percent-clipped=9.0 2023-05-02 09:53:27,621 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270396.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:53:37,254 INFO [train.py:904] (7/8) Epoch 27, batch 6500, loss[loss=0.2101, simple_loss=0.2944, pruned_loss=0.06291, over 16634.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2838, pruned_loss=0.0558, over 3064108.79 frames. ], batch size: 57, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:53:44,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9905, 3.2752, 3.4688, 2.0988, 2.9438, 2.3189, 3.5344, 3.5727], device='cuda:7'), covar=tensor([0.0281, 0.0832, 0.0627, 0.2141, 0.0903, 0.1046, 0.0594, 0.0915], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0171, 0.0170, 0.0156, 0.0148, 0.0133, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 09:54:17,016 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270429.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 09:54:33,509 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270439.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:54:35,965 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7731, 4.8733, 4.6412, 4.2372, 4.2653, 4.7260, 4.5528, 4.4724], device='cuda:7'), covar=tensor([0.0864, 0.1161, 0.0305, 0.0449, 0.0995, 0.0653, 0.0731, 0.0814], device='cuda:7'), in_proj_covar=tensor([0.0306, 0.0462, 0.0358, 0.0362, 0.0358, 0.0415, 0.0245, 0.0430], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:54:58,128 INFO [train.py:904] (7/8) Epoch 27, batch 6550, loss[loss=0.2343, simple_loss=0.3066, pruned_loss=0.08098, over 11391.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2863, pruned_loss=0.05618, over 3079439.44 frames. ], batch size: 248, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:55:04,753 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270457.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:55:17,873 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.942e+02 2.679e+02 3.213e+02 3.936e+02 7.742e+02, threshold=6.427e+02, percent-clipped=6.0 2023-05-02 09:55:34,671 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270477.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:55:37,842 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9249, 4.9652, 4.8049, 4.4080, 4.4565, 4.8724, 4.7560, 4.5700], device='cuda:7'), covar=tensor([0.0749, 0.0769, 0.0348, 0.0405, 0.1071, 0.0585, 0.0482, 0.0783], device='cuda:7'), in_proj_covar=tensor([0.0305, 0.0461, 0.0357, 0.0361, 0.0357, 0.0414, 0.0245, 0.0429], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:55:47,693 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.5844, 2.7089, 2.6862, 4.5797, 3.5161, 3.9972, 1.4961, 3.0577], device='cuda:7'), covar=tensor([0.1471, 0.0805, 0.1202, 0.0160, 0.0288, 0.0417, 0.1754, 0.0822], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0179, 0.0200, 0.0201, 0.0207, 0.0219, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 09:55:58,268 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270492.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:56:00,822 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2680, 5.5493, 5.2818, 5.2864, 5.0657, 4.9706, 4.8901, 5.6336], device='cuda:7'), covar=tensor([0.1372, 0.0887, 0.1114, 0.0963, 0.0858, 0.0892, 0.1375, 0.0947], device='cuda:7'), in_proj_covar=tensor([0.0705, 0.0853, 0.0696, 0.0657, 0.0538, 0.0537, 0.0717, 0.0666], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 09:56:10,604 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270500.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:56:14,143 INFO [train.py:904] (7/8) Epoch 27, batch 6600, loss[loss=0.2039, simple_loss=0.2986, pruned_loss=0.05464, over 16694.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2879, pruned_loss=0.05553, over 3118073.76 frames. ], batch size: 134, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:57:09,774 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270540.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:57:30,479 INFO [train.py:904] (7/8) Epoch 27, batch 6650, loss[loss=0.1958, simple_loss=0.2846, pruned_loss=0.05347, over 16849.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2883, pruned_loss=0.05666, over 3105729.68 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:57:50,327 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.739e+02 3.273e+02 3.972e+02 7.700e+02, threshold=6.545e+02, percent-clipped=3.0 2023-05-02 09:58:46,098 INFO [train.py:904] (7/8) Epoch 27, batch 6700, loss[loss=0.2385, simple_loss=0.3065, pruned_loss=0.0852, over 11313.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2872, pruned_loss=0.05694, over 3106323.25 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 09:59:01,635 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 09:59:01,834 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.59 vs. limit=2.0 2023-05-02 09:59:36,009 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270636.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 10:00:01,100 INFO [train.py:904] (7/8) Epoch 27, batch 6750, loss[loss=0.1976, simple_loss=0.2882, pruned_loss=0.05353, over 16495.00 frames. ], tot_loss[loss=0.1999, simple_loss=0.2863, pruned_loss=0.05669, over 3124217.69 frames. ], batch size: 68, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:00:20,169 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.179e+02 2.756e+02 3.273e+02 4.054e+02 1.247e+03, threshold=6.545e+02, percent-clipped=2.0 2023-05-02 10:00:32,928 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:01:05,717 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0373, 3.2434, 3.5640, 1.9723, 2.9686, 2.2715, 3.5542, 3.5411], device='cuda:7'), covar=tensor([0.0241, 0.0828, 0.0550, 0.2190, 0.0840, 0.1005, 0.0600, 0.0922], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0170, 0.0156, 0.0148, 0.0132, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 10:01:14,954 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.19 vs. limit=5.0 2023-05-02 10:01:15,222 INFO [train.py:904] (7/8) Epoch 27, batch 6800, loss[loss=0.2354, simple_loss=0.3191, pruned_loss=0.07586, over 16338.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2866, pruned_loss=0.05685, over 3117920.83 frames. ], batch size: 165, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:01:20,732 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5776, 2.5541, 1.8010, 2.7041, 2.1031, 2.7719, 2.0271, 2.3196], device='cuda:7'), covar=tensor([0.0317, 0.0361, 0.1326, 0.0262, 0.0626, 0.0505, 0.1324, 0.0668], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0180, 0.0195, 0.0171, 0.0179, 0.0218, 0.0202, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 10:01:40,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0995, 2.1345, 2.6878, 3.0613, 2.9567, 3.5713, 2.2382, 3.5301], device='cuda:7'), covar=tensor([0.0253, 0.0578, 0.0377, 0.0364, 0.0326, 0.0166, 0.0621, 0.0153], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0197, 0.0184, 0.0190, 0.0205, 0.0164, 0.0202, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:01:53,367 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270726.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:02:33,314 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:02:34,170 INFO [train.py:904] (7/8) Epoch 27, batch 6850, loss[loss=0.1939, simple_loss=0.2925, pruned_loss=0.04769, over 16873.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2883, pruned_loss=0.05785, over 3101960.04 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:02:53,222 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.704e+02 3.220e+02 3.695e+02 5.610e+02, threshold=6.440e+02, percent-clipped=0.0 2023-05-02 10:03:03,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8687, 4.8353, 4.7506, 3.8347, 4.7332, 1.8025, 4.4571, 4.2411], device='cuda:7'), covar=tensor([0.0164, 0.0162, 0.0220, 0.0424, 0.0143, 0.3141, 0.0255, 0.0323], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0174, 0.0212, 0.0186, 0.0187, 0.0217, 0.0200, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:03:25,245 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270787.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:03:37,708 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270795.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:03:49,494 INFO [train.py:904] (7/8) Epoch 27, batch 6900, loss[loss=0.2247, simple_loss=0.3007, pruned_loss=0.07433, over 11748.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.2912, pruned_loss=0.05774, over 3094581.97 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:04:01,355 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4260, 1.7119, 2.1200, 2.3069, 2.4037, 2.6195, 1.8748, 2.5783], device='cuda:7'), covar=tensor([0.0232, 0.0546, 0.0368, 0.0383, 0.0375, 0.0248, 0.0598, 0.0169], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0197, 0.0185, 0.0190, 0.0205, 0.0165, 0.0202, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:04:06,040 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4032, 4.5630, 4.6991, 4.4905, 4.5762, 5.0542, 4.6221, 4.3731], device='cuda:7'), covar=tensor([0.1499, 0.2012, 0.2517, 0.2165, 0.2430, 0.1093, 0.1657, 0.2528], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0629, 0.0692, 0.0511, 0.0682, 0.0719, 0.0540, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:05:00,959 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4198, 3.3760, 3.4430, 3.5166, 3.5744, 3.3214, 3.5389, 3.6063], device='cuda:7'), covar=tensor([0.1217, 0.0941, 0.1065, 0.0613, 0.0651, 0.2263, 0.1095, 0.0846], device='cuda:7'), in_proj_covar=tensor([0.0661, 0.0808, 0.0940, 0.0819, 0.0625, 0.0655, 0.0684, 0.0795], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:05:07,710 INFO [train.py:904] (7/8) Epoch 27, batch 6950, loss[loss=0.2083, simple_loss=0.2941, pruned_loss=0.06118, over 16900.00 frames. ], tot_loss[loss=0.2053, simple_loss=0.2924, pruned_loss=0.05915, over 3079495.63 frames. ], batch size: 109, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:05:28,453 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.915e+02 3.513e+02 4.432e+02 8.204e+02, threshold=7.026e+02, percent-clipped=5.0 2023-05-02 10:05:33,993 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-02 10:06:06,645 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5228, 3.5063, 3.4731, 2.6256, 3.3509, 2.0819, 3.1181, 2.7685], device='cuda:7'), covar=tensor([0.0169, 0.0149, 0.0178, 0.0233, 0.0107, 0.2437, 0.0156, 0.0270], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0173, 0.0211, 0.0185, 0.0186, 0.0216, 0.0199, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:06:23,856 INFO [train.py:904] (7/8) Epoch 27, batch 7000, loss[loss=0.1957, simple_loss=0.2923, pruned_loss=0.04953, over 16704.00 frames. ], tot_loss[loss=0.2043, simple_loss=0.2917, pruned_loss=0.05848, over 3067024.01 frames. ], batch size: 134, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:06:46,526 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=270917.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:07:15,575 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=270936.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 10:07:40,704 INFO [train.py:904] (7/8) Epoch 27, batch 7050, loss[loss=0.207, simple_loss=0.2933, pruned_loss=0.0604, over 15331.00 frames. ], tot_loss[loss=0.2036, simple_loss=0.2921, pruned_loss=0.05757, over 3086733.94 frames. ], batch size: 191, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:07:42,436 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0470, 4.1314, 3.9375, 3.6563, 3.7081, 4.0549, 3.7145, 3.8524], device='cuda:7'), covar=tensor([0.0581, 0.0561, 0.0293, 0.0304, 0.0698, 0.0500, 0.1103, 0.0556], device='cuda:7'), in_proj_covar=tensor([0.0304, 0.0460, 0.0357, 0.0360, 0.0356, 0.0414, 0.0245, 0.0428], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:08:01,041 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.024e+02 2.738e+02 3.222e+02 3.975e+02 6.307e+02, threshold=6.443e+02, percent-clipped=0.0 2023-05-02 10:08:05,128 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=270969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:08:18,594 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=270978.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:08:28,176 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=270984.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 10:08:55,789 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3548, 2.8770, 3.0467, 1.9104, 2.7089, 2.1402, 2.9345, 3.1704], device='cuda:7'), covar=tensor([0.0373, 0.0893, 0.0648, 0.2276, 0.0986, 0.1006, 0.0852, 0.0897], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0172, 0.0171, 0.0157, 0.0149, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 10:08:58,246 INFO [train.py:904] (7/8) Epoch 27, batch 7100, loss[loss=0.202, simple_loss=0.2907, pruned_loss=0.05661, over 16236.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2905, pruned_loss=0.05715, over 3098121.23 frames. ], batch size: 165, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:09:34,074 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-05-02 10:10:12,330 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0079, 2.1753, 2.2375, 3.4593, 2.1128, 2.4935, 2.2755, 2.2903], device='cuda:7'), covar=tensor([0.1456, 0.3418, 0.3085, 0.0691, 0.4247, 0.2437, 0.3453, 0.3630], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0469, 0.0381, 0.0332, 0.0443, 0.0537, 0.0441, 0.0548], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:10:15,860 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271052.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:10:16,508 INFO [train.py:904] (7/8) Epoch 27, batch 7150, loss[loss=0.1785, simple_loss=0.2708, pruned_loss=0.04313, over 16908.00 frames. ], tot_loss[loss=0.202, simple_loss=0.2892, pruned_loss=0.05742, over 3076601.88 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:10:37,965 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 2.616e+02 3.372e+02 3.961e+02 8.125e+02, threshold=6.744e+02, percent-clipped=3.0 2023-05-02 10:11:00,511 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:11:19,946 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271095.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:11:20,097 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4984, 1.7888, 2.1208, 2.4867, 2.4980, 2.8416, 1.8567, 2.7360], device='cuda:7'), covar=tensor([0.0299, 0.0606, 0.0424, 0.0390, 0.0376, 0.0237, 0.0667, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0196, 0.0183, 0.0189, 0.0204, 0.0164, 0.0201, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:11:28,069 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271100.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:11:31,641 INFO [train.py:904] (7/8) Epoch 27, batch 7200, loss[loss=0.183, simple_loss=0.2874, pruned_loss=0.03933, over 16833.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.287, pruned_loss=0.0553, over 3096312.98 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:11:53,636 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4007, 2.5206, 2.4368, 4.2072, 2.4071, 2.9142, 2.5632, 2.6399], device='cuda:7'), covar=tensor([0.1337, 0.3352, 0.2972, 0.0520, 0.3963, 0.2343, 0.3509, 0.3235], device='cuda:7'), in_proj_covar=tensor([0.0417, 0.0470, 0.0381, 0.0333, 0.0444, 0.0537, 0.0441, 0.0549], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:11:54,843 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2012, 1.5745, 1.9441, 2.1622, 2.2291, 2.3846, 1.7425, 2.3199], device='cuda:7'), covar=tensor([0.0256, 0.0513, 0.0309, 0.0328, 0.0331, 0.0232, 0.0585, 0.0160], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0196, 0.0183, 0.0188, 0.0204, 0.0163, 0.0201, 0.0162], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:12:19,323 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6105, 3.5549, 2.8173, 2.2586, 2.4383, 2.4574, 3.9357, 3.2687], device='cuda:7'), covar=tensor([0.3097, 0.0860, 0.1867, 0.2856, 0.2694, 0.2191, 0.0466, 0.1464], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0276, 0.0313, 0.0328, 0.0306, 0.0276, 0.0305, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:12:37,998 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271143.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:12:53,508 INFO [train.py:904] (7/8) Epoch 27, batch 7250, loss[loss=0.1733, simple_loss=0.2606, pruned_loss=0.04296, over 16600.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2841, pruned_loss=0.05391, over 3102608.18 frames. ], batch size: 62, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:13:13,958 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0951, 3.5922, 3.5766, 2.2736, 3.2924, 3.5923, 3.3288, 2.0467], device='cuda:7'), covar=tensor([0.0624, 0.0075, 0.0076, 0.0478, 0.0128, 0.0131, 0.0124, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0089, 0.0090, 0.0136, 0.0101, 0.0115, 0.0098, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 10:13:16,041 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.855e+02 2.666e+02 3.085e+02 3.699e+02 8.603e+02, threshold=6.169e+02, percent-clipped=3.0 2023-05-02 10:13:28,571 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1023, 3.6870, 3.6936, 2.3632, 3.3848, 3.7171, 3.4267, 2.0747], device='cuda:7'), covar=tensor([0.0652, 0.0079, 0.0079, 0.0500, 0.0124, 0.0150, 0.0121, 0.0539], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0089, 0.0090, 0.0136, 0.0101, 0.0115, 0.0098, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 10:13:30,261 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8329, 3.8192, 3.9558, 3.7399, 3.9247, 4.2892, 3.9116, 3.6610], device='cuda:7'), covar=tensor([0.1946, 0.2221, 0.2302, 0.2288, 0.2404, 0.1593, 0.1712, 0.2519], device='cuda:7'), in_proj_covar=tensor([0.0426, 0.0633, 0.0694, 0.0515, 0.0686, 0.0724, 0.0545, 0.0692], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:14:00,104 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 10:14:10,782 INFO [train.py:904] (7/8) Epoch 27, batch 7300, loss[loss=0.1845, simple_loss=0.2736, pruned_loss=0.0477, over 16557.00 frames. ], tot_loss[loss=0.195, simple_loss=0.283, pruned_loss=0.05353, over 3082592.64 frames. ], batch size: 62, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:15:29,752 INFO [train.py:904] (7/8) Epoch 27, batch 7350, loss[loss=0.2329, simple_loss=0.3003, pruned_loss=0.08273, over 11193.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2849, pruned_loss=0.0552, over 3056630.10 frames. ], batch size: 247, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:15:30,289 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9826, 2.8063, 2.5423, 4.6052, 3.4515, 4.0646, 1.6723, 3.0323], device='cuda:7'), covar=tensor([0.1238, 0.0785, 0.1261, 0.0180, 0.0290, 0.0378, 0.1648, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0179, 0.0200, 0.0199, 0.0207, 0.0218, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 10:15:42,242 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271261.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:15:50,539 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 2.875e+02 3.303e+02 4.103e+02 7.171e+02, threshold=6.606e+02, percent-clipped=3.0 2023-05-02 10:15:53,280 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 10:15:54,816 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:16:01,231 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271273.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:16:48,733 INFO [train.py:904] (7/8) Epoch 27, batch 7400, loss[loss=0.2013, simple_loss=0.2902, pruned_loss=0.05625, over 16314.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2863, pruned_loss=0.056, over 3069443.35 frames. ], batch size: 35, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:17:09,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8025, 4.8611, 4.6790, 4.3678, 4.3699, 4.7778, 4.5870, 4.5026], device='cuda:7'), covar=tensor([0.0626, 0.0676, 0.0329, 0.0331, 0.0924, 0.0465, 0.0507, 0.0697], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0455, 0.0354, 0.0356, 0.0352, 0.0409, 0.0243, 0.0423], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:17:11,161 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:17:18,907 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271322.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:17:33,048 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9611, 4.1394, 3.0644, 2.4662, 2.8320, 2.6710, 4.5066, 3.6678], device='cuda:7'), covar=tensor([0.2696, 0.0575, 0.1818, 0.2843, 0.2573, 0.2003, 0.0414, 0.1247], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0276, 0.0314, 0.0329, 0.0307, 0.0277, 0.0306, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:18:06,496 INFO [train.py:904] (7/8) Epoch 27, batch 7450, loss[loss=0.22, simple_loss=0.3066, pruned_loss=0.06675, over 16989.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2875, pruned_loss=0.05721, over 3054959.78 frames. ], batch size: 109, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:18:11,074 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4487, 3.4218, 3.8885, 2.0284, 3.9524, 4.0823, 2.9826, 2.9297], device='cuda:7'), covar=tensor([0.0857, 0.0304, 0.0216, 0.1171, 0.0108, 0.0176, 0.0458, 0.0496], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0110, 0.0102, 0.0139, 0.0086, 0.0131, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 10:18:30,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.940e+02 2.819e+02 3.560e+02 4.384e+02 9.484e+02, threshold=7.119e+02, percent-clipped=1.0 2023-05-02 10:18:45,259 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271375.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:18:55,972 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271382.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:19:30,459 INFO [train.py:904] (7/8) Epoch 27, batch 7500, loss[loss=0.1846, simple_loss=0.2769, pruned_loss=0.04614, over 16688.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2873, pruned_loss=0.05632, over 3049545.26 frames. ], batch size: 134, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:20:08,159 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271426.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:20:14,066 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271430.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:20:23,995 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271436.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:20:49,667 INFO [train.py:904] (7/8) Epoch 27, batch 7550, loss[loss=0.2337, simple_loss=0.3004, pruned_loss=0.08351, over 11600.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2866, pruned_loss=0.05667, over 3042214.71 frames. ], batch size: 248, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:21:11,195 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.886e+02 2.650e+02 3.300e+02 4.210e+02 7.541e+02, threshold=6.599e+02, percent-clipped=2.0 2023-05-02 10:21:41,660 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:22:05,496 INFO [train.py:904] (7/8) Epoch 27, batch 7600, loss[loss=0.19, simple_loss=0.2894, pruned_loss=0.04532, over 16865.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2856, pruned_loss=0.05614, over 3058418.33 frames. ], batch size: 96, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:23:22,938 INFO [train.py:904] (7/8) Epoch 27, batch 7650, loss[loss=0.1833, simple_loss=0.2761, pruned_loss=0.04524, over 16727.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2851, pruned_loss=0.05533, over 3086580.22 frames. ], batch size: 124, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:23:24,251 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.74 vs. limit=2.0 2023-05-02 10:23:45,489 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.699e+02 2.765e+02 3.286e+02 4.189e+02 6.927e+02, threshold=6.573e+02, percent-clipped=1.0 2023-05-02 10:23:55,787 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271573.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:24:43,700 INFO [train.py:904] (7/8) Epoch 27, batch 7700, loss[loss=0.2229, simple_loss=0.2923, pruned_loss=0.07677, over 11923.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.285, pruned_loss=0.05577, over 3076022.72 frames. ], batch size: 248, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:25:06,105 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271617.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:25:12,495 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271621.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:25:17,025 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8882, 2.1953, 2.4729, 3.1208, 2.2593, 2.3687, 2.3439, 2.3086], device='cuda:7'), covar=tensor([0.1519, 0.3535, 0.2573, 0.0816, 0.4276, 0.2445, 0.3252, 0.3449], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0470, 0.0381, 0.0333, 0.0443, 0.0536, 0.0441, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:26:00,563 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5866, 2.5698, 1.9133, 2.6604, 2.1454, 2.7714, 2.1403, 2.3561], device='cuda:7'), covar=tensor([0.0327, 0.0401, 0.1353, 0.0273, 0.0639, 0.0522, 0.1166, 0.0644], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0182, 0.0198, 0.0173, 0.0181, 0.0221, 0.0205, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 10:26:02,454 INFO [train.py:904] (7/8) Epoch 27, batch 7750, loss[loss=0.186, simple_loss=0.2783, pruned_loss=0.04684, over 16587.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2854, pruned_loss=0.05619, over 3065553.82 frames. ], batch size: 62, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:26:24,500 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.915e+02 2.896e+02 3.348e+02 4.032e+02 8.871e+02, threshold=6.696e+02, percent-clipped=2.0 2023-05-02 10:27:19,547 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9239, 3.2105, 3.3666, 1.9763, 2.9334, 2.1899, 3.4255, 3.4478], device='cuda:7'), covar=tensor([0.0248, 0.0810, 0.0560, 0.2186, 0.0831, 0.1050, 0.0595, 0.0972], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0171, 0.0170, 0.0156, 0.0149, 0.0133, 0.0146, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 10:27:20,093 INFO [train.py:904] (7/8) Epoch 27, batch 7800, loss[loss=0.1876, simple_loss=0.2767, pruned_loss=0.04925, over 16870.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2867, pruned_loss=0.05714, over 3062601.55 frames. ], batch size: 116, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:27:59,428 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 10:28:04,772 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271731.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:28:36,666 INFO [train.py:904] (7/8) Epoch 27, batch 7850, loss[loss=0.1965, simple_loss=0.2932, pruned_loss=0.04987, over 16485.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2877, pruned_loss=0.05696, over 3067971.56 frames. ], batch size: 68, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:28:57,993 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.838e+02 3.379e+02 4.179e+02 9.334e+02, threshold=6.757e+02, percent-clipped=5.0 2023-05-02 10:29:04,122 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2023-05-02 10:29:12,010 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8926, 2.1806, 2.4593, 3.1255, 2.1904, 2.3690, 2.3505, 2.2723], device='cuda:7'), covar=tensor([0.1464, 0.3335, 0.2508, 0.0802, 0.4275, 0.2411, 0.3382, 0.3402], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0470, 0.0381, 0.0333, 0.0445, 0.0537, 0.0441, 0.0549], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:29:20,881 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=271782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:29:52,547 INFO [train.py:904] (7/8) Epoch 27, batch 7900, loss[loss=0.2328, simple_loss=0.299, pruned_loss=0.08328, over 11487.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2873, pruned_loss=0.05696, over 3055580.28 frames. ], batch size: 248, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:30:37,445 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3061, 4.3832, 4.2045, 3.9179, 3.9351, 4.2952, 4.0549, 4.0199], device='cuda:7'), covar=tensor([0.0641, 0.0687, 0.0322, 0.0330, 0.0826, 0.0573, 0.0733, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0300, 0.0452, 0.0352, 0.0354, 0.0351, 0.0407, 0.0242, 0.0421], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:30:59,276 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3984, 4.6946, 4.4893, 4.5345, 4.2453, 4.1929, 4.1947, 4.7322], device='cuda:7'), covar=tensor([0.1294, 0.0883, 0.1020, 0.0876, 0.0809, 0.1653, 0.1277, 0.0847], device='cuda:7'), in_proj_covar=tensor([0.0703, 0.0849, 0.0699, 0.0655, 0.0536, 0.0539, 0.0714, 0.0666], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:31:00,736 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.06 vs. limit=5.0 2023-05-02 10:31:12,169 INFO [train.py:904] (7/8) Epoch 27, batch 7950, loss[loss=0.2013, simple_loss=0.2873, pruned_loss=0.05763, over 15458.00 frames. ], tot_loss[loss=0.2007, simple_loss=0.2872, pruned_loss=0.05705, over 3063166.45 frames. ], batch size: 191, lr: 2.48e-03, grad_scale: 4.0 2023-05-02 10:31:15,318 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271854.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:31:36,736 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 2.764e+02 3.233e+02 3.778e+02 5.723e+02, threshold=6.466e+02, percent-clipped=0.0 2023-05-02 10:31:49,692 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9976, 4.9777, 4.7868, 4.0691, 4.9130, 1.9248, 4.6205, 4.4724], device='cuda:7'), covar=tensor([0.0120, 0.0099, 0.0218, 0.0458, 0.0102, 0.2981, 0.0163, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0171, 0.0210, 0.0183, 0.0185, 0.0215, 0.0197, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:32:29,934 INFO [train.py:904] (7/8) Epoch 27, batch 8000, loss[loss=0.1895, simple_loss=0.2788, pruned_loss=0.05011, over 17082.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2872, pruned_loss=0.05733, over 3060350.02 frames. ], batch size: 49, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:32:49,708 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271915.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:32:52,896 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=271917.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:33:19,495 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=271934.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:33:26,621 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3942, 4.5040, 4.6775, 4.3856, 4.5333, 5.0303, 4.5344, 4.2463], device='cuda:7'), covar=tensor([0.1532, 0.2072, 0.2221, 0.2139, 0.2455, 0.1064, 0.1597, 0.2509], device='cuda:7'), in_proj_covar=tensor([0.0426, 0.0631, 0.0693, 0.0514, 0.0685, 0.0723, 0.0544, 0.0690], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:33:29,176 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0832, 3.9497, 4.1350, 4.2546, 4.3811, 3.9996, 4.3241, 4.4083], device='cuda:7'), covar=tensor([0.1786, 0.1180, 0.1449, 0.0757, 0.0674, 0.1362, 0.0898, 0.0832], device='cuda:7'), in_proj_covar=tensor([0.0656, 0.0798, 0.0931, 0.0809, 0.0623, 0.0648, 0.0680, 0.0792], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:33:31,141 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4182, 2.7792, 3.0822, 1.9460, 2.8172, 2.0656, 3.0482, 3.1141], device='cuda:7'), covar=tensor([0.0305, 0.0933, 0.0623, 0.2170, 0.0876, 0.1106, 0.0656, 0.0909], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0170, 0.0156, 0.0148, 0.0133, 0.0146, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 10:33:47,194 INFO [train.py:904] (7/8) Epoch 27, batch 8050, loss[loss=0.2051, simple_loss=0.301, pruned_loss=0.05458, over 16360.00 frames. ], tot_loss[loss=0.2017, simple_loss=0.2879, pruned_loss=0.05775, over 3054398.51 frames. ], batch size: 146, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:34:05,659 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=271965.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:34:09,139 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.787e+02 2.577e+02 3.144e+02 3.701e+02 6.468e+02, threshold=6.287e+02, percent-clipped=1.0 2023-05-02 10:34:49,622 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=271995.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:35:04,687 INFO [train.py:904] (7/8) Epoch 27, batch 8100, loss[loss=0.2011, simple_loss=0.2772, pruned_loss=0.06248, over 17021.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2874, pruned_loss=0.05727, over 3058075.05 frames. ], batch size: 50, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:35:45,691 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272031.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:36:17,771 INFO [train.py:904] (7/8) Epoch 27, batch 8150, loss[loss=0.2235, simple_loss=0.3025, pruned_loss=0.07222, over 11862.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2849, pruned_loss=0.05613, over 3067576.04 frames. ], batch size: 246, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:36:24,054 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2444, 3.8669, 3.8545, 2.5383, 3.5071, 3.8651, 3.4738, 2.2132], device='cuda:7'), covar=tensor([0.0595, 0.0063, 0.0062, 0.0423, 0.0118, 0.0127, 0.0105, 0.0497], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0089, 0.0089, 0.0135, 0.0100, 0.0114, 0.0097, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 10:36:39,769 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.743e+02 3.279e+02 3.939e+02 6.192e+02, threshold=6.559e+02, percent-clipped=0.0 2023-05-02 10:36:57,283 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272079.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:36:57,389 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9820, 3.8510, 4.0422, 4.1551, 4.2486, 3.8815, 4.1846, 4.2702], device='cuda:7'), covar=tensor([0.1682, 0.1256, 0.1371, 0.0710, 0.0668, 0.1558, 0.0896, 0.0813], device='cuda:7'), in_proj_covar=tensor([0.0659, 0.0801, 0.0934, 0.0812, 0.0625, 0.0650, 0.0682, 0.0794], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:37:01,634 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:37:32,137 INFO [train.py:904] (7/8) Epoch 27, batch 8200, loss[loss=0.2277, simple_loss=0.2925, pruned_loss=0.0815, over 11343.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2827, pruned_loss=0.05561, over 3065553.35 frames. ], batch size: 246, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:38:12,539 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272127.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:38:16,463 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:38:53,589 INFO [train.py:904] (7/8) Epoch 27, batch 8250, loss[loss=0.1785, simple_loss=0.2781, pruned_loss=0.0395, over 16366.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2817, pruned_loss=0.05302, over 3071688.71 frames. ], batch size: 146, lr: 2.48e-03, grad_scale: 8.0 2023-05-02 10:39:19,039 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.616e+02 2.668e+02 3.172e+02 3.719e+02 6.913e+02, threshold=6.344e+02, percent-clipped=1.0 2023-05-02 10:39:40,219 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272181.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 10:39:50,742 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272188.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:39:52,141 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0150, 2.4016, 2.5967, 1.9707, 2.6679, 2.7441, 2.4889, 2.5338], device='cuda:7'), covar=tensor([0.0764, 0.0273, 0.0295, 0.1057, 0.0175, 0.0291, 0.0486, 0.0447], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0110, 0.0101, 0.0139, 0.0086, 0.0130, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 10:40:14,331 INFO [train.py:904] (7/8) Epoch 27, batch 8300, loss[loss=0.1819, simple_loss=0.2756, pruned_loss=0.04409, over 16329.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2784, pruned_loss=0.05003, over 3055952.25 frames. ], batch size: 146, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:40:26,391 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272210.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:40:38,674 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-05-02 10:40:46,060 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272223.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:41:15,186 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272242.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 10:41:23,032 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272247.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:41:28,375 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1180, 1.5865, 1.9634, 2.1004, 2.2288, 2.3300, 1.8067, 2.2435], device='cuda:7'), covar=tensor([0.0296, 0.0528, 0.0304, 0.0382, 0.0328, 0.0230, 0.0547, 0.0175], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0197, 0.0184, 0.0189, 0.0204, 0.0163, 0.0200, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:41:31,413 INFO [train.py:904] (7/8) Epoch 27, batch 8350, loss[loss=0.1671, simple_loss=0.2751, pruned_loss=0.02953, over 16816.00 frames. ], tot_loss[loss=0.1867, simple_loss=0.2774, pruned_loss=0.04794, over 3046995.86 frames. ], batch size: 102, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:41:54,865 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.144e+02 2.517e+02 3.037e+02 5.148e+02, threshold=5.034e+02, percent-clipped=0.0 2023-05-02 10:42:21,257 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272284.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:42:30,970 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272290.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:42:50,443 INFO [train.py:904] (7/8) Epoch 27, batch 8400, loss[loss=0.1615, simple_loss=0.2573, pruned_loss=0.0329, over 15401.00 frames. ], tot_loss[loss=0.183, simple_loss=0.2747, pruned_loss=0.04566, over 3050121.31 frames. ], batch size: 191, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:42:59,358 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272308.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:43:06,017 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 10:44:09,899 INFO [train.py:904] (7/8) Epoch 27, batch 8450, loss[loss=0.174, simple_loss=0.2706, pruned_loss=0.03864, over 16918.00 frames. ], tot_loss[loss=0.1804, simple_loss=0.2729, pruned_loss=0.0439, over 3055191.13 frames. ], batch size: 109, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:44:34,141 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.487e+02 2.148e+02 2.573e+02 3.182e+02 7.232e+02, threshold=5.146e+02, percent-clipped=1.0 2023-05-02 10:44:42,811 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272373.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:45:27,630 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-02 10:45:32,414 INFO [train.py:904] (7/8) Epoch 27, batch 8500, loss[loss=0.162, simple_loss=0.2463, pruned_loss=0.03879, over 11718.00 frames. ], tot_loss[loss=0.177, simple_loss=0.2698, pruned_loss=0.04214, over 3031462.58 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:45:33,979 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-02 10:45:50,680 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 10:45:58,830 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6007, 3.5611, 2.8397, 2.2011, 2.2414, 2.4514, 3.8258, 3.1723], device='cuda:7'), covar=tensor([0.2892, 0.0622, 0.1865, 0.3321, 0.3244, 0.2301, 0.0392, 0.1505], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0269, 0.0306, 0.0320, 0.0299, 0.0270, 0.0298, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:46:24,356 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:46:57,066 INFO [train.py:904] (7/8) Epoch 27, batch 8550, loss[loss=0.1825, simple_loss=0.2806, pruned_loss=0.04217, over 16279.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2672, pruned_loss=0.04083, over 3028643.22 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:47:25,005 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.466e+02 2.247e+02 2.591e+02 3.219e+02 6.393e+02, threshold=5.181e+02, percent-clipped=2.0 2023-05-02 10:47:55,111 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272483.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:48:35,741 INFO [train.py:904] (7/8) Epoch 27, batch 8600, loss[loss=0.1784, simple_loss=0.2767, pruned_loss=0.04007, over 15414.00 frames. ], tot_loss[loss=0.1742, simple_loss=0.268, pruned_loss=0.04021, over 3026754.88 frames. ], batch size: 192, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:48:50,722 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272510.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:49:43,704 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272537.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 10:50:13,945 INFO [train.py:904] (7/8) Epoch 27, batch 8650, loss[loss=0.1634, simple_loss=0.267, pruned_loss=0.0299, over 16798.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2666, pruned_loss=0.03891, over 3048161.49 frames. ], batch size: 124, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:50:27,063 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272558.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:50:48,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.181e+02 2.564e+02 3.046e+02 5.835e+02, threshold=5.129e+02, percent-clipped=3.0 2023-05-02 10:51:15,237 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272579.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:51:36,721 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272590.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:52:00,243 INFO [train.py:904] (7/8) Epoch 27, batch 8700, loss[loss=0.164, simple_loss=0.25, pruned_loss=0.03898, over 12355.00 frames. ], tot_loss[loss=0.1703, simple_loss=0.2643, pruned_loss=0.03811, over 3051795.17 frames. ], batch size: 247, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:52:01,254 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272603.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:52:21,645 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=272613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:52:33,849 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-05-02 10:53:06,659 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272638.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:53:32,844 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.85 vs. limit=5.0 2023-05-02 10:53:35,418 INFO [train.py:904] (7/8) Epoch 27, batch 8750, loss[loss=0.1596, simple_loss=0.2496, pruned_loss=0.03482, over 12232.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2636, pruned_loss=0.0377, over 3029742.42 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 10:54:12,788 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8168, 3.8039, 3.9147, 3.6706, 3.9088, 4.2543, 3.9006, 3.5868], device='cuda:7'), covar=tensor([0.2064, 0.2360, 0.2544, 0.2773, 0.2564, 0.1697, 0.1643, 0.2548], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0612, 0.0671, 0.0499, 0.0666, 0.0705, 0.0529, 0.0668], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 10:54:15,883 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.360e+02 2.765e+02 3.360e+02 6.939e+02, threshold=5.529e+02, percent-clipped=4.0 2023-05-02 10:54:28,960 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=272674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:54:45,154 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 10:55:17,455 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0546, 4.1309, 3.9697, 3.7216, 3.7400, 4.0603, 3.7601, 3.8765], device='cuda:7'), covar=tensor([0.0551, 0.0658, 0.0303, 0.0281, 0.0674, 0.0513, 0.0970, 0.0603], device='cuda:7'), in_proj_covar=tensor([0.0297, 0.0448, 0.0348, 0.0351, 0.0346, 0.0403, 0.0240, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:55:27,075 INFO [train.py:904] (7/8) Epoch 27, batch 8800, loss[loss=0.1638, simple_loss=0.2657, pruned_loss=0.03098, over 15271.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2628, pruned_loss=0.03686, over 3044994.23 frames. ], batch size: 190, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:56:23,113 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:57:01,896 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-05-02 10:57:12,476 INFO [train.py:904] (7/8) Epoch 27, batch 8850, loss[loss=0.1791, simple_loss=0.2824, pruned_loss=0.0379, over 16334.00 frames. ], tot_loss[loss=0.1686, simple_loss=0.2651, pruned_loss=0.03609, over 3055079.27 frames. ], batch size: 146, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:57:46,551 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.085e+02 2.509e+02 2.907e+02 5.151e+02, threshold=5.018e+02, percent-clipped=0.0 2023-05-02 10:57:57,013 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5128, 3.4661, 3.4986, 2.4587, 3.4193, 1.9419, 3.2172, 2.7324], device='cuda:7'), covar=tensor([0.0196, 0.0164, 0.0223, 0.0393, 0.0149, 0.3330, 0.0176, 0.0381], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0169, 0.0206, 0.0179, 0.0182, 0.0212, 0.0194, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 10:58:17,428 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272783.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 10:58:41,490 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-02 10:58:57,770 INFO [train.py:904] (7/8) Epoch 27, batch 8900, loss[loss=0.153, simple_loss=0.2464, pruned_loss=0.02976, over 12888.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2645, pruned_loss=0.03537, over 3034008.43 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 10:59:56,076 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4979, 3.4090, 3.5418, 3.6066, 3.6592, 3.3435, 3.6230, 3.6915], device='cuda:7'), covar=tensor([0.1218, 0.1017, 0.1042, 0.0665, 0.0569, 0.2301, 0.0869, 0.0854], device='cuda:7'), in_proj_covar=tensor([0.0636, 0.0779, 0.0904, 0.0791, 0.0605, 0.0633, 0.0661, 0.0773], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:00:01,542 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272831.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:00:08,098 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9787, 4.1117, 3.8923, 3.6713, 3.5308, 4.0171, 3.6740, 3.7245], device='cuda:7'), covar=tensor([0.0687, 0.0773, 0.0417, 0.0386, 0.0931, 0.0617, 0.1233, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0296, 0.0446, 0.0347, 0.0349, 0.0344, 0.0401, 0.0239, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:00:20,725 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272837.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 11:00:52,884 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8490, 4.6263, 4.8181, 4.9847, 5.1936, 4.6052, 5.1739, 5.1829], device='cuda:7'), covar=tensor([0.1785, 0.1386, 0.1782, 0.0830, 0.0538, 0.1040, 0.0621, 0.0733], device='cuda:7'), in_proj_covar=tensor([0.0636, 0.0779, 0.0904, 0.0791, 0.0605, 0.0633, 0.0662, 0.0773], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:01:00,401 INFO [train.py:904] (7/8) Epoch 27, batch 8950, loss[loss=0.148, simple_loss=0.2424, pruned_loss=0.02679, over 15213.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2639, pruned_loss=0.03538, over 3053784.06 frames. ], batch size: 191, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:01:35,341 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 1.963e+02 2.469e+02 3.081e+02 5.293e+02, threshold=4.938e+02, percent-clipped=1.0 2023-05-02 11:01:56,988 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272879.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:01:57,319 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 11:02:12,417 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272885.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 11:02:40,478 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-02 11:02:49,722 INFO [train.py:904] (7/8) Epoch 27, batch 9000, loss[loss=0.1564, simple_loss=0.2449, pruned_loss=0.03392, over 12032.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.261, pruned_loss=0.03423, over 3065921.29 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:02:49,723 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 11:02:57,708 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8651, 4.9951, 4.9522, 4.7400, 4.8102, 5.3294, 4.8707, 4.6128], device='cuda:7'), covar=tensor([0.0807, 0.2047, 0.2719, 0.1511, 0.2203, 0.0747, 0.1484, 0.2126], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0613, 0.0672, 0.0500, 0.0665, 0.0706, 0.0530, 0.0668], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 11:02:59,740 INFO [train.py:938] (7/8) Epoch 27, validation: loss=0.1436, simple_loss=0.2474, pruned_loss=0.01989, over 944034.00 frames. 2023-05-02 11:02:59,741 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 11:03:01,103 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=272903.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:03:17,427 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3203, 2.3651, 2.2766, 3.9256, 2.2435, 2.6456, 2.3922, 2.4771], device='cuda:7'), covar=tensor([0.1332, 0.3624, 0.3272, 0.0564, 0.4071, 0.2592, 0.3884, 0.3278], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0463, 0.0378, 0.0326, 0.0437, 0.0528, 0.0436, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:03:26,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4803, 2.4799, 2.3833, 4.1727, 2.3196, 2.7611, 2.4703, 2.5767], device='cuda:7'), covar=tensor([0.1242, 0.3558, 0.3217, 0.0504, 0.4126, 0.2600, 0.3873, 0.3224], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0463, 0.0378, 0.0327, 0.0438, 0.0528, 0.0437, 0.0540], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:03:51,519 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272927.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:04:42,536 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=272951.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:04:44,601 INFO [train.py:904] (7/8) Epoch 27, batch 9050, loss[loss=0.1613, simple_loss=0.2525, pruned_loss=0.03507, over 15407.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2613, pruned_loss=0.03431, over 3077962.12 frames. ], batch size: 192, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:04:56,339 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4797, 3.2110, 3.4621, 1.9103, 3.5985, 3.6621, 2.9956, 2.9079], device='cuda:7'), covar=tensor([0.0735, 0.0261, 0.0183, 0.1203, 0.0093, 0.0178, 0.0431, 0.0449], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0106, 0.0097, 0.0134, 0.0082, 0.0124, 0.0125, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 11:05:18,777 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.208e+02 2.579e+02 3.072e+02 5.940e+02, threshold=5.159e+02, percent-clipped=5.0 2023-05-02 11:05:19,948 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=272969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:05:31,772 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3539, 4.6819, 4.5238, 4.5237, 4.1978, 4.1828, 4.2150, 4.7247], device='cuda:7'), covar=tensor([0.1266, 0.0902, 0.0937, 0.0809, 0.0845, 0.1431, 0.1125, 0.0855], device='cuda:7'), in_proj_covar=tensor([0.0694, 0.0840, 0.0689, 0.0645, 0.0529, 0.0532, 0.0702, 0.0658], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:06:32,231 INFO [train.py:904] (7/8) Epoch 27, batch 9100, loss[loss=0.1683, simple_loss=0.2707, pruned_loss=0.03294, over 16706.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.261, pruned_loss=0.0346, over 3087838.72 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:07:36,908 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8642, 3.8196, 3.9636, 3.7016, 3.9189, 4.2984, 3.9556, 3.6216], device='cuda:7'), covar=tensor([0.1980, 0.2665, 0.2586, 0.2670, 0.2813, 0.1657, 0.1722, 0.2847], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0615, 0.0672, 0.0501, 0.0666, 0.0706, 0.0531, 0.0668], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 11:07:37,006 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=273029.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:07:39,921 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 11:08:32,214 INFO [train.py:904] (7/8) Epoch 27, batch 9150, loss[loss=0.1568, simple_loss=0.2502, pruned_loss=0.03169, over 15503.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2612, pruned_loss=0.03451, over 3078536.06 frames. ], batch size: 192, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:09:08,130 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.260e+02 2.634e+02 3.413e+02 5.711e+02, threshold=5.268e+02, percent-clipped=2.0 2023-05-02 11:09:27,810 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=273077.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:10:18,358 INFO [train.py:904] (7/8) Epoch 27, batch 9200, loss[loss=0.1723, simple_loss=0.2663, pruned_loss=0.03921, over 16646.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2575, pruned_loss=0.0339, over 3056140.30 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:11:53,375 INFO [train.py:904] (7/8) Epoch 27, batch 9250, loss[loss=0.1476, simple_loss=0.2469, pruned_loss=0.02417, over 16547.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2574, pruned_loss=0.03422, over 3039566.41 frames. ], batch size: 68, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:12:17,005 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 11:12:25,534 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 2.225e+02 2.595e+02 3.229e+02 5.341e+02, threshold=5.191e+02, percent-clipped=1.0 2023-05-02 11:12:53,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 11:13:23,182 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.11 vs. limit=5.0 2023-05-02 11:13:43,829 INFO [train.py:904] (7/8) Epoch 27, batch 9300, loss[loss=0.1491, simple_loss=0.2435, pruned_loss=0.02733, over 15314.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2558, pruned_loss=0.03366, over 3032883.05 frames. ], batch size: 190, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:15:28,953 INFO [train.py:904] (7/8) Epoch 27, batch 9350, loss[loss=0.1578, simple_loss=0.2553, pruned_loss=0.03012, over 16675.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2562, pruned_loss=0.03412, over 3047949.91 frames. ], batch size: 89, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:15:56,828 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 11:16:02,981 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.192e+02 2.614e+02 3.426e+02 6.584e+02, threshold=5.229e+02, percent-clipped=2.0 2023-05-02 11:16:03,992 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=273269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:17:10,298 INFO [train.py:904] (7/8) Epoch 27, batch 9400, loss[loss=0.1499, simple_loss=0.2362, pruned_loss=0.03183, over 12338.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2567, pruned_loss=0.03394, over 3059762.30 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:17:39,666 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=273317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:18:52,056 INFO [train.py:904] (7/8) Epoch 27, batch 9450, loss[loss=0.1693, simple_loss=0.2508, pruned_loss=0.04387, over 12429.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2581, pruned_loss=0.03376, over 3057460.91 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:19:19,216 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8498, 3.2593, 3.5338, 2.0042, 3.0196, 2.1121, 3.4571, 3.3247], device='cuda:7'), covar=tensor([0.0245, 0.0832, 0.0485, 0.2215, 0.0765, 0.1149, 0.0514, 0.0952], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0164, 0.0165, 0.0152, 0.0144, 0.0129, 0.0141, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:19:21,971 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.197e+02 2.626e+02 3.206e+02 9.861e+02, threshold=5.251e+02, percent-clipped=1.0 2023-05-02 11:20:32,364 INFO [train.py:904] (7/8) Epoch 27, batch 9500, loss[loss=0.1359, simple_loss=0.2274, pruned_loss=0.02221, over 12634.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2581, pruned_loss=0.03398, over 3070042.44 frames. ], batch size: 246, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:20:49,200 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=273409.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:21:38,348 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=273435.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:22:18,628 INFO [train.py:904] (7/8) Epoch 27, batch 9550, loss[loss=0.1799, simple_loss=0.2833, pruned_loss=0.03826, over 15276.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2579, pruned_loss=0.03376, over 3059078.15 frames. ], batch size: 190, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:22:51,236 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.160e+02 2.156e+02 2.505e+02 2.980e+02 5.260e+02, threshold=5.009e+02, percent-clipped=1.0 2023-05-02 11:22:54,304 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=273470.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:23:46,485 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=273496.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:23:57,855 INFO [train.py:904] (7/8) Epoch 27, batch 9600, loss[loss=0.1861, simple_loss=0.2858, pruned_loss=0.04318, over 16350.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2589, pruned_loss=0.03452, over 3042519.66 frames. ], batch size: 146, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:25:44,218 INFO [train.py:904] (7/8) Epoch 27, batch 9650, loss[loss=0.1588, simple_loss=0.2671, pruned_loss=0.02523, over 16873.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.261, pruned_loss=0.03494, over 3039358.32 frames. ], batch size: 102, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:26:23,246 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.167e+02 2.602e+02 2.993e+02 6.217e+02, threshold=5.204e+02, percent-clipped=3.0 2023-05-02 11:26:51,041 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1221, 2.5208, 2.6523, 1.9233, 2.7301, 2.8677, 2.5116, 2.4790], device='cuda:7'), covar=tensor([0.0673, 0.0297, 0.0250, 0.1039, 0.0139, 0.0235, 0.0508, 0.0453], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0106, 0.0097, 0.0134, 0.0082, 0.0124, 0.0125, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 11:27:19,721 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8007, 5.0800, 4.8734, 4.9057, 4.6169, 4.5510, 4.4753, 5.1651], device='cuda:7'), covar=tensor([0.1160, 0.0875, 0.0957, 0.0785, 0.0728, 0.1090, 0.1229, 0.0808], device='cuda:7'), in_proj_covar=tensor([0.0684, 0.0829, 0.0678, 0.0638, 0.0522, 0.0524, 0.0694, 0.0647], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:27:30,808 INFO [train.py:904] (7/8) Epoch 27, batch 9700, loss[loss=0.1943, simple_loss=0.2773, pruned_loss=0.05561, over 12315.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2598, pruned_loss=0.03464, over 3052780.96 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:27:35,952 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0253, 3.9028, 4.0831, 4.1903, 4.3023, 3.8927, 4.2874, 4.3265], device='cuda:7'), covar=tensor([0.1753, 0.1181, 0.1486, 0.0813, 0.0618, 0.1462, 0.0687, 0.0798], device='cuda:7'), in_proj_covar=tensor([0.0639, 0.0780, 0.0903, 0.0794, 0.0608, 0.0633, 0.0665, 0.0774], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:29:12,479 INFO [train.py:904] (7/8) Epoch 27, batch 9750, loss[loss=0.1713, simple_loss=0.2645, pruned_loss=0.03899, over 16670.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2588, pruned_loss=0.0346, over 3069040.47 frames. ], batch size: 134, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:29:19,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9832, 5.3037, 5.4789, 5.2535, 5.3440, 5.8404, 5.3285, 5.0761], device='cuda:7'), covar=tensor([0.0891, 0.1838, 0.2327, 0.1839, 0.2228, 0.0902, 0.1574, 0.2155], device='cuda:7'), in_proj_covar=tensor([0.0405, 0.0605, 0.0663, 0.0491, 0.0653, 0.0696, 0.0522, 0.0652], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 11:29:42,117 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 2.115e+02 2.683e+02 3.309e+02 9.068e+02, threshold=5.365e+02, percent-clipped=4.0 2023-05-02 11:30:51,298 INFO [train.py:904] (7/8) Epoch 27, batch 9800, loss[loss=0.1536, simple_loss=0.2597, pruned_loss=0.02372, over 16874.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2592, pruned_loss=0.03348, over 3098734.75 frames. ], batch size: 90, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:31:18,940 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.71 vs. limit=2.0 2023-05-02 11:31:20,968 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-05-02 11:32:36,075 INFO [train.py:904] (7/8) Epoch 27, batch 9850, loss[loss=0.1728, simple_loss=0.2667, pruned_loss=0.03947, over 16147.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2599, pruned_loss=0.03298, over 3099217.00 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:33:00,293 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=273765.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:33:11,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.029e+02 2.351e+02 3.028e+02 6.435e+02, threshold=4.703e+02, percent-clipped=1.0 2023-05-02 11:34:02,984 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=273791.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:34:25,795 INFO [train.py:904] (7/8) Epoch 27, batch 9900, loss[loss=0.1468, simple_loss=0.2395, pruned_loss=0.02704, over 12529.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2592, pruned_loss=0.03295, over 3058035.47 frames. ], batch size: 248, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:34:34,167 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 11:36:22,575 INFO [train.py:904] (7/8) Epoch 27, batch 9950, loss[loss=0.1608, simple_loss=0.2626, pruned_loss=0.02943, over 16281.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.262, pruned_loss=0.03348, over 3059399.04 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 4.0 2023-05-02 11:37:03,262 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 1.975e+02 2.538e+02 3.032e+02 5.048e+02, threshold=5.077e+02, percent-clipped=2.0 2023-05-02 11:38:25,125 INFO [train.py:904] (7/8) Epoch 27, batch 10000, loss[loss=0.1784, simple_loss=0.2768, pruned_loss=0.03998, over 16294.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2606, pruned_loss=0.03301, over 3069298.19 frames. ], batch size: 146, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:39:55,053 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6472, 2.4685, 2.2865, 3.8314, 2.1577, 3.7533, 1.4173, 2.8362], device='cuda:7'), covar=tensor([0.1521, 0.0910, 0.1395, 0.0158, 0.0124, 0.0362, 0.1944, 0.0788], device='cuda:7'), in_proj_covar=tensor([0.0169, 0.0175, 0.0196, 0.0193, 0.0198, 0.0212, 0.0205, 0.0194], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:40:06,633 INFO [train.py:904] (7/8) Epoch 27, batch 10050, loss[loss=0.1755, simple_loss=0.2754, pruned_loss=0.03782, over 16149.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2607, pruned_loss=0.03324, over 3064608.77 frames. ], batch size: 165, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:40:39,177 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.137e+02 2.431e+02 3.038e+02 5.006e+02, threshold=4.862e+02, percent-clipped=0.0 2023-05-02 11:41:41,042 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274001.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:41:43,727 INFO [train.py:904] (7/8) Epoch 27, batch 10100, loss[loss=0.1723, simple_loss=0.2619, pruned_loss=0.04131, over 16953.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2609, pruned_loss=0.03334, over 3077945.17 frames. ], batch size: 109, lr: 2.47e-03, grad_scale: 8.0 2023-05-02 11:42:08,787 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4907, 3.5071, 2.7278, 2.1527, 2.2282, 2.3492, 3.7223, 3.0732], device='cuda:7'), covar=tensor([0.3023, 0.0625, 0.1845, 0.3315, 0.3050, 0.2252, 0.0442, 0.1520], device='cuda:7'), in_proj_covar=tensor([0.0326, 0.0266, 0.0303, 0.0316, 0.0293, 0.0268, 0.0295, 0.0338], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 11:42:47,949 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1054, 2.5310, 2.6104, 1.8931, 2.6949, 2.8358, 2.5731, 2.4908], device='cuda:7'), covar=tensor([0.0728, 0.0263, 0.0292, 0.1104, 0.0155, 0.0261, 0.0488, 0.0467], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0105, 0.0096, 0.0134, 0.0082, 0.0124, 0.0124, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 11:42:51,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5652, 2.8623, 3.2682, 1.9546, 2.7328, 2.0585, 3.0999, 3.0497], device='cuda:7'), covar=tensor([0.0322, 0.0995, 0.0548, 0.2312, 0.0910, 0.1126, 0.0765, 0.1226], device='cuda:7'), in_proj_covar=tensor([0.0155, 0.0162, 0.0163, 0.0152, 0.0143, 0.0128, 0.0140, 0.0173], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:43:27,176 INFO [train.py:904] (7/8) Epoch 28, batch 0, loss[loss=0.212, simple_loss=0.29, pruned_loss=0.06701, over 16260.00 frames. ], tot_loss[loss=0.212, simple_loss=0.29, pruned_loss=0.06701, over 16260.00 frames. ], batch size: 165, lr: 2.42e-03, grad_scale: 8.0 2023-05-02 11:43:27,176 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 11:43:34,597 INFO [train.py:938] (7/8) Epoch 28, validation: loss=0.1434, simple_loss=0.2464, pruned_loss=0.02022, over 944034.00 frames. 2023-05-02 11:43:34,598 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 11:43:48,118 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274062.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:43:52,205 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274065.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:44:01,204 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 2.305e+02 2.657e+02 3.293e+02 5.505e+02, threshold=5.314e+02, percent-clipped=2.0 2023-05-02 11:44:13,587 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8216, 2.4903, 2.4763, 3.8379, 2.8042, 3.7735, 1.6181, 2.8912], device='cuda:7'), covar=tensor([0.1450, 0.0773, 0.1246, 0.0169, 0.0121, 0.0415, 0.1679, 0.0834], device='cuda:7'), in_proj_covar=tensor([0.0170, 0.0176, 0.0197, 0.0194, 0.0199, 0.0213, 0.0206, 0.0195], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:44:27,638 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274091.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:44:43,862 INFO [train.py:904] (7/8) Epoch 28, batch 50, loss[loss=0.236, simple_loss=0.3106, pruned_loss=0.08073, over 11796.00 frames. ], tot_loss[loss=0.1794, simple_loss=0.2659, pruned_loss=0.04642, over 739940.46 frames. ], batch size: 246, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:44:59,070 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274113.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:44:59,214 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0108, 4.9225, 4.8907, 4.4798, 4.6535, 4.9298, 4.8761, 4.6177], device='cuda:7'), covar=tensor([0.0571, 0.0613, 0.0319, 0.0343, 0.0944, 0.0450, 0.0387, 0.0740], device='cuda:7'), in_proj_covar=tensor([0.0294, 0.0439, 0.0343, 0.0345, 0.0340, 0.0395, 0.0236, 0.0409], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:45:02,848 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5273, 2.5347, 2.0405, 2.4473, 2.8856, 2.6824, 3.1451, 3.1531], device='cuda:7'), covar=tensor([0.0227, 0.0573, 0.0791, 0.0552, 0.0378, 0.0526, 0.0306, 0.0367], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0238, 0.0227, 0.0228, 0.0239, 0.0237, 0.0232, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:45:24,147 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274131.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:45:34,074 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274139.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:45:54,002 INFO [train.py:904] (7/8) Epoch 28, batch 100, loss[loss=0.1595, simple_loss=0.2468, pruned_loss=0.03608, over 16854.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2617, pruned_loss=0.04214, over 1322490.46 frames. ], batch size: 42, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:46:20,303 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.302e+02 2.829e+02 3.561e+02 6.259e+02, threshold=5.657e+02, percent-clipped=1.0 2023-05-02 11:46:23,024 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5482, 5.9572, 5.6599, 5.7218, 5.3552, 5.2387, 5.3752, 6.0877], device='cuda:7'), covar=tensor([0.1564, 0.1063, 0.1441, 0.0973, 0.0956, 0.0761, 0.1443, 0.0984], device='cuda:7'), in_proj_covar=tensor([0.0698, 0.0846, 0.0689, 0.0650, 0.0533, 0.0533, 0.0708, 0.0661], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:46:40,669 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-05-02 11:46:48,689 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:46:52,430 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-05-02 11:47:02,022 INFO [train.py:904] (7/8) Epoch 28, batch 150, loss[loss=0.2052, simple_loss=0.29, pruned_loss=0.06024, over 12091.00 frames. ], tot_loss[loss=0.1722, simple_loss=0.2608, pruned_loss=0.04179, over 1771179.66 frames. ], batch size: 246, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:48:08,599 INFO [train.py:904] (7/8) Epoch 28, batch 200, loss[loss=0.1905, simple_loss=0.2669, pruned_loss=0.05709, over 16878.00 frames. ], tot_loss[loss=0.173, simple_loss=0.2607, pruned_loss=0.04264, over 2123726.52 frames. ], batch size: 116, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:48:30,346 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:48:33,862 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 2.228e+02 2.594e+02 3.067e+02 5.865e+02, threshold=5.188e+02, percent-clipped=1.0 2023-05-02 11:49:16,137 INFO [train.py:904] (7/8) Epoch 28, batch 250, loss[loss=0.16, simple_loss=0.2481, pruned_loss=0.03594, over 16661.00 frames. ], tot_loss[loss=0.1704, simple_loss=0.2582, pruned_loss=0.04129, over 2394036.93 frames. ], batch size: 68, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:49:16,632 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274303.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:49:26,637 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8317, 4.0115, 2.6971, 4.6271, 3.2773, 4.5002, 2.7103, 3.3367], device='cuda:7'), covar=tensor([0.0338, 0.0393, 0.1581, 0.0319, 0.0799, 0.0559, 0.1542, 0.0769], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0177, 0.0194, 0.0168, 0.0177, 0.0215, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:49:51,992 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274330.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:50:23,026 INFO [train.py:904] (7/8) Epoch 28, batch 300, loss[loss=0.1582, simple_loss=0.2378, pruned_loss=0.03932, over 16800.00 frames. ], tot_loss[loss=0.169, simple_loss=0.2564, pruned_loss=0.04083, over 2598736.88 frames. ], batch size: 102, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:50:30,573 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274357.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:50:39,344 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274364.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:50:49,960 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.173e+02 2.525e+02 3.031e+02 5.896e+02, threshold=5.050e+02, percent-clipped=1.0 2023-05-02 11:51:31,016 INFO [train.py:904] (7/8) Epoch 28, batch 350, loss[loss=0.1551, simple_loss=0.2406, pruned_loss=0.03479, over 12482.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.253, pruned_loss=0.03921, over 2755193.31 frames. ], batch size: 246, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:52:37,186 INFO [train.py:904] (7/8) Epoch 28, batch 400, loss[loss=0.2121, simple_loss=0.2922, pruned_loss=0.06605, over 16900.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2526, pruned_loss=0.03959, over 2868999.30 frames. ], batch size: 109, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 11:52:48,279 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5466, 4.6842, 4.8012, 4.6312, 4.6807, 5.2447, 4.7200, 4.4210], device='cuda:7'), covar=tensor([0.1560, 0.2149, 0.2452, 0.2281, 0.2641, 0.1187, 0.1814, 0.2650], device='cuda:7'), in_proj_covar=tensor([0.0416, 0.0623, 0.0682, 0.0505, 0.0674, 0.0714, 0.0536, 0.0670], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 11:53:03,893 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.265e+02 2.649e+02 3.069e+02 5.270e+02, threshold=5.298e+02, percent-clipped=3.0 2023-05-02 11:53:24,168 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:53:44,160 INFO [train.py:904] (7/8) Epoch 28, batch 450, loss[loss=0.1648, simple_loss=0.2477, pruned_loss=0.04093, over 12421.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2507, pruned_loss=0.03872, over 2965427.73 frames. ], batch size: 246, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:54:02,310 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-02 11:54:35,087 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274540.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:54:52,730 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9445, 2.6095, 2.0467, 2.3519, 2.9209, 2.7055, 2.8921, 3.0083], device='cuda:7'), covar=tensor([0.0341, 0.0460, 0.0643, 0.0558, 0.0310, 0.0423, 0.0321, 0.0347], device='cuda:7'), in_proj_covar=tensor([0.0229, 0.0246, 0.0235, 0.0235, 0.0247, 0.0245, 0.0242, 0.0242], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 11:54:53,427 INFO [train.py:904] (7/8) Epoch 28, batch 500, loss[loss=0.1744, simple_loss=0.2526, pruned_loss=0.04807, over 16863.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2493, pruned_loss=0.03792, over 3047231.27 frames. ], batch size: 96, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:55:01,577 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3690, 3.3456, 2.1961, 3.5527, 2.7266, 3.5239, 2.2585, 2.8477], device='cuda:7'), covar=tensor([0.0310, 0.0488, 0.1486, 0.0400, 0.0752, 0.0835, 0.1456, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0179, 0.0195, 0.0170, 0.0178, 0.0217, 0.0204, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:55:21,740 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.414e+02 2.207e+02 2.697e+02 3.239e+02 1.435e+03, threshold=5.394e+02, percent-clipped=3.0 2023-05-02 11:55:58,982 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274601.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:56:00,846 INFO [train.py:904] (7/8) Epoch 28, batch 550, loss[loss=0.1459, simple_loss=0.2342, pruned_loss=0.02883, over 17232.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2484, pruned_loss=0.03719, over 3110361.81 frames. ], batch size: 45, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:56:21,122 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 11:56:31,532 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274625.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:56:41,932 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4450, 3.4941, 2.2868, 3.7033, 2.8769, 3.6645, 2.3465, 2.9180], device='cuda:7'), covar=tensor([0.0317, 0.0461, 0.1548, 0.0432, 0.0727, 0.0841, 0.1448, 0.0723], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0180, 0.0197, 0.0172, 0.0179, 0.0219, 0.0206, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 11:56:56,817 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 11:57:09,875 INFO [train.py:904] (7/8) Epoch 28, batch 600, loss[loss=0.1531, simple_loss=0.2387, pruned_loss=0.03371, over 17216.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2481, pruned_loss=0.03727, over 3155224.07 frames. ], batch size: 45, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:57:15,385 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274657.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:57:17,524 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274659.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:57:36,635 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.130e+02 2.496e+02 2.954e+02 2.616e+03, threshold=4.992e+02, percent-clipped=3.0 2023-05-02 11:58:16,074 INFO [train.py:904] (7/8) Epoch 28, batch 650, loss[loss=0.1692, simple_loss=0.2434, pruned_loss=0.04754, over 16919.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2471, pruned_loss=0.0372, over 3198135.53 frames. ], batch size: 109, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:58:18,660 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274705.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 11:58:57,876 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9956, 2.1049, 2.4833, 2.8535, 2.8614, 2.8791, 2.0926, 3.0867], device='cuda:7'), covar=tensor([0.0213, 0.0531, 0.0388, 0.0318, 0.0347, 0.0329, 0.0638, 0.0206], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0200, 0.0188, 0.0193, 0.0208, 0.0167, 0.0206, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 11:59:25,796 INFO [train.py:904] (7/8) Epoch 28, batch 700, loss[loss=0.176, simple_loss=0.2453, pruned_loss=0.05334, over 16707.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2471, pruned_loss=0.03773, over 3218119.94 frames. ], batch size: 134, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 11:59:54,346 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.008e+02 2.338e+02 2.859e+02 5.050e+02, threshold=4.676e+02, percent-clipped=1.0 2023-05-02 12:00:13,335 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274787.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:00:34,720 INFO [train.py:904] (7/8) Epoch 28, batch 750, loss[loss=0.1843, simple_loss=0.2587, pruned_loss=0.05496, over 16673.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2477, pruned_loss=0.0379, over 3241130.39 frames. ], batch size: 89, lr: 2.42e-03, grad_scale: 1.0 2023-05-02 12:01:19,982 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274835.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:01:22,670 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8881, 2.7100, 2.7917, 5.0697, 4.0125, 4.3902, 1.6865, 3.2726], device='cuda:7'), covar=tensor([0.1362, 0.0888, 0.1188, 0.0175, 0.0212, 0.0405, 0.1641, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0179, 0.0200, 0.0200, 0.0204, 0.0218, 0.0209, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:01:45,416 INFO [train.py:904] (7/8) Epoch 28, batch 800, loss[loss=0.1661, simple_loss=0.2518, pruned_loss=0.04022, over 17168.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2467, pruned_loss=0.03746, over 3264232.22 frames. ], batch size: 46, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:02:15,004 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.216e+02 2.000e+02 2.423e+02 2.829e+02 5.216e+02, threshold=4.846e+02, percent-clipped=1.0 2023-05-02 12:02:43,459 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=274894.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:02:46,942 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=274896.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:02:57,024 INFO [train.py:904] (7/8) Epoch 28, batch 850, loss[loss=0.1555, simple_loss=0.2359, pruned_loss=0.03759, over 16717.00 frames. ], tot_loss[loss=0.1595, simple_loss=0.2457, pruned_loss=0.03669, over 3283403.58 frames. ], batch size: 89, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:03:27,477 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274925.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:05,753 INFO [train.py:904] (7/8) Epoch 28, batch 900, loss[loss=0.1941, simple_loss=0.2874, pruned_loss=0.05039, over 17069.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.2453, pruned_loss=0.03614, over 3290450.61 frames. ], batch size: 53, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:04:09,248 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=274955.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:14,367 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=274959.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:19,891 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2569, 4.3355, 4.3644, 4.2132, 4.3133, 4.8479, 4.3212, 4.0177], device='cuda:7'), covar=tensor([0.1865, 0.2295, 0.2967, 0.2533, 0.2956, 0.1294, 0.1905, 0.2859], device='cuda:7'), in_proj_covar=tensor([0.0425, 0.0638, 0.0700, 0.0517, 0.0689, 0.0727, 0.0549, 0.0686], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 12:04:33,653 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=274973.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:04:34,542 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.075e+02 2.378e+02 2.870e+02 6.521e+02, threshold=4.756e+02, percent-clipped=4.0 2023-05-02 12:05:14,561 INFO [train.py:904] (7/8) Epoch 28, batch 950, loss[loss=0.1645, simple_loss=0.2625, pruned_loss=0.03326, over 17142.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.2455, pruned_loss=0.03604, over 3302181.09 frames. ], batch size: 47, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:05:21,080 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275007.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:05:56,312 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3160, 5.3599, 5.7229, 5.7029, 5.7429, 5.4265, 5.3271, 5.1663], device='cuda:7'), covar=tensor([0.0354, 0.0561, 0.0400, 0.0403, 0.0478, 0.0382, 0.0985, 0.0467], device='cuda:7'), in_proj_covar=tensor([0.0439, 0.0493, 0.0477, 0.0438, 0.0524, 0.0502, 0.0575, 0.0402], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 12:06:23,175 INFO [train.py:904] (7/8) Epoch 28, batch 1000, loss[loss=0.1598, simple_loss=0.2503, pruned_loss=0.03461, over 17123.00 frames. ], tot_loss[loss=0.1581, simple_loss=0.2444, pruned_loss=0.0359, over 3310240.09 frames. ], batch size: 49, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:06:52,348 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.067e+02 2.369e+02 2.767e+02 9.232e+02, threshold=4.738e+02, percent-clipped=2.0 2023-05-02 12:07:31,791 INFO [train.py:904] (7/8) Epoch 28, batch 1050, loss[loss=0.165, simple_loss=0.2665, pruned_loss=0.03178, over 17115.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2444, pruned_loss=0.0355, over 3319140.29 frames. ], batch size: 48, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:08:09,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2466, 3.2765, 3.6642, 2.2896, 3.0844, 2.4768, 3.7148, 3.6414], device='cuda:7'), covar=tensor([0.0259, 0.1003, 0.0606, 0.2022, 0.0867, 0.1038, 0.0527, 0.0954], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0170, 0.0170, 0.0157, 0.0148, 0.0133, 0.0147, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 12:08:39,623 INFO [train.py:904] (7/8) Epoch 28, batch 1100, loss[loss=0.1421, simple_loss=0.2303, pruned_loss=0.02697, over 17229.00 frames. ], tot_loss[loss=0.1574, simple_loss=0.2436, pruned_loss=0.03561, over 3308857.02 frames. ], batch size: 43, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:08:47,471 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275158.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:09:07,966 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 1.984e+02 2.393e+02 2.792e+02 5.361e+02, threshold=4.787e+02, percent-clipped=2.0 2023-05-02 12:09:39,589 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275196.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:09:49,916 INFO [train.py:904] (7/8) Epoch 28, batch 1150, loss[loss=0.1731, simple_loss=0.2629, pruned_loss=0.04169, over 17073.00 frames. ], tot_loss[loss=0.1571, simple_loss=0.2438, pruned_loss=0.03519, over 3314275.24 frames. ], batch size: 53, lr: 2.42e-03, grad_scale: 2.0 2023-05-02 12:09:52,367 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 12:10:10,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1716, 5.1676, 4.8689, 4.3441, 5.0077, 1.8187, 4.7288, 4.5861], device='cuda:7'), covar=tensor([0.0105, 0.0091, 0.0217, 0.0399, 0.0105, 0.2976, 0.0148, 0.0291], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0174, 0.0210, 0.0182, 0.0187, 0.0217, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:10:12,446 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275219.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:10:30,475 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4968, 2.2940, 1.8549, 2.1232, 2.6362, 2.3798, 2.4803, 2.6975], device='cuda:7'), covar=tensor([0.0293, 0.0529, 0.0614, 0.0543, 0.0305, 0.0419, 0.0236, 0.0366], device='cuda:7'), in_proj_covar=tensor([0.0234, 0.0250, 0.0238, 0.0238, 0.0251, 0.0248, 0.0247, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:10:44,959 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275244.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:10:53,464 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275250.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:10:57,386 INFO [train.py:904] (7/8) Epoch 28, batch 1200, loss[loss=0.1597, simple_loss=0.2363, pruned_loss=0.0415, over 16890.00 frames. ], tot_loss[loss=0.157, simple_loss=0.2437, pruned_loss=0.03518, over 3317364.09 frames. ], batch size: 109, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:11:26,952 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.036e+02 2.270e+02 2.715e+02 4.480e+02, threshold=4.540e+02, percent-clipped=0.0 2023-05-02 12:11:28,842 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 12:12:03,269 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2755, 5.0764, 5.3066, 5.4726, 5.6608, 4.9998, 5.5909, 5.6303], device='cuda:7'), covar=tensor([0.1799, 0.1281, 0.1610, 0.0715, 0.0568, 0.0864, 0.0609, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0683, 0.0832, 0.0965, 0.0844, 0.0644, 0.0673, 0.0709, 0.0821], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:12:06,867 INFO [train.py:904] (7/8) Epoch 28, batch 1250, loss[loss=0.1756, simple_loss=0.2529, pruned_loss=0.04913, over 16847.00 frames. ], tot_loss[loss=0.1578, simple_loss=0.2442, pruned_loss=0.03569, over 3321122.05 frames. ], batch size: 102, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:12:12,718 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275307.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:13:09,291 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275347.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 12:13:17,200 INFO [train.py:904] (7/8) Epoch 28, batch 1300, loss[loss=0.1556, simple_loss=0.2404, pruned_loss=0.03542, over 16880.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2444, pruned_loss=0.03555, over 3319898.20 frames. ], batch size: 90, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:13:39,161 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275368.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:13:46,873 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.360e+02 1.989e+02 2.411e+02 2.818e+02 5.022e+02, threshold=4.822e+02, percent-clipped=1.0 2023-05-02 12:14:02,899 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275385.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:14:26,885 INFO [train.py:904] (7/8) Epoch 28, batch 1350, loss[loss=0.177, simple_loss=0.2545, pruned_loss=0.04973, over 15512.00 frames. ], tot_loss[loss=0.1573, simple_loss=0.2442, pruned_loss=0.03524, over 3321983.53 frames. ], batch size: 190, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:14:35,349 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275408.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 12:15:28,687 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275446.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:15:37,728 INFO [train.py:904] (7/8) Epoch 28, batch 1400, loss[loss=0.1481, simple_loss=0.2259, pruned_loss=0.03519, over 15318.00 frames. ], tot_loss[loss=0.1572, simple_loss=0.2441, pruned_loss=0.03517, over 3327589.09 frames. ], batch size: 190, lr: 2.42e-03, grad_scale: 4.0 2023-05-02 12:16:06,678 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.102e+02 2.484e+02 3.057e+02 6.401e+02, threshold=4.968e+02, percent-clipped=1.0 2023-05-02 12:16:46,822 INFO [train.py:904] (7/8) Epoch 28, batch 1450, loss[loss=0.1867, simple_loss=0.2764, pruned_loss=0.04854, over 17088.00 frames. ], tot_loss[loss=0.1573, simple_loss=0.2435, pruned_loss=0.03555, over 3327887.99 frames. ], batch size: 53, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:17:03,276 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275514.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:17:06,829 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8184, 2.0860, 2.3997, 2.6596, 2.6961, 2.6612, 1.9933, 2.9140], device='cuda:7'), covar=tensor([0.0193, 0.0506, 0.0358, 0.0325, 0.0373, 0.0403, 0.0599, 0.0220], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0200, 0.0189, 0.0194, 0.0209, 0.0168, 0.0205, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 12:17:53,972 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275550.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:17:57,213 INFO [train.py:904] (7/8) Epoch 28, batch 1500, loss[loss=0.1822, simple_loss=0.2524, pruned_loss=0.05601, over 16874.00 frames. ], tot_loss[loss=0.1576, simple_loss=0.2431, pruned_loss=0.03601, over 3325520.76 frames. ], batch size: 116, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:18:26,334 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.124e+02 2.433e+02 3.002e+02 9.704e+02, threshold=4.866e+02, percent-clipped=1.0 2023-05-02 12:19:01,079 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275598.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:19:07,367 INFO [train.py:904] (7/8) Epoch 28, batch 1550, loss[loss=0.1479, simple_loss=0.2412, pruned_loss=0.02727, over 16994.00 frames. ], tot_loss[loss=0.1594, simple_loss=0.2443, pruned_loss=0.03726, over 3319157.90 frames. ], batch size: 41, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:19:18,166 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 12:19:31,508 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7490, 3.5492, 3.9674, 2.1769, 4.0016, 4.0537, 3.3415, 2.9771], device='cuda:7'), covar=tensor([0.0789, 0.0273, 0.0165, 0.1206, 0.0107, 0.0228, 0.0369, 0.0494], device='cuda:7'), in_proj_covar=tensor([0.0151, 0.0111, 0.0103, 0.0141, 0.0087, 0.0133, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:20:18,407 INFO [train.py:904] (7/8) Epoch 28, batch 1600, loss[loss=0.1595, simple_loss=0.2338, pruned_loss=0.04263, over 16929.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2464, pruned_loss=0.03751, over 3322757.01 frames. ], batch size: 109, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:20:33,170 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275663.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:20:48,276 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.650e+02 2.228e+02 2.653e+02 3.158e+02 7.607e+02, threshold=5.306e+02, percent-clipped=1.0 2023-05-02 12:21:10,199 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9369, 4.5142, 4.4969, 3.2269, 3.6761, 4.4860, 4.0777, 2.8138], device='cuda:7'), covar=tensor([0.0532, 0.0071, 0.0055, 0.0396, 0.0162, 0.0092, 0.0087, 0.0491], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0091, 0.0092, 0.0136, 0.0103, 0.0115, 0.0099, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 12:21:29,006 INFO [train.py:904] (7/8) Epoch 28, batch 1650, loss[loss=0.1478, simple_loss=0.2369, pruned_loss=0.02937, over 16784.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.249, pruned_loss=0.0379, over 3319692.33 frames. ], batch size: 39, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:21:29,415 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275703.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 12:22:21,953 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=275741.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:22:37,929 INFO [train.py:904] (7/8) Epoch 28, batch 1700, loss[loss=0.1702, simple_loss=0.2644, pruned_loss=0.03804, over 17003.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2501, pruned_loss=0.03824, over 3327695.74 frames. ], batch size: 55, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:23:08,712 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.542e+02 2.133e+02 2.503e+02 2.990e+02 4.977e+02, threshold=5.005e+02, percent-clipped=0.0 2023-05-02 12:23:15,885 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6712, 4.4759, 4.7092, 4.8596, 5.0221, 4.5246, 4.9918, 5.0432], device='cuda:7'), covar=tensor([0.1788, 0.1275, 0.1560, 0.0781, 0.0573, 0.1087, 0.1097, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0692, 0.0842, 0.0978, 0.0855, 0.0652, 0.0683, 0.0718, 0.0831], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:23:49,171 INFO [train.py:904] (7/8) Epoch 28, batch 1750, loss[loss=0.1659, simple_loss=0.2407, pruned_loss=0.04558, over 16719.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2512, pruned_loss=0.03823, over 3328054.87 frames. ], batch size: 124, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:24:04,521 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3009, 5.2394, 5.1498, 4.6045, 4.7666, 5.1827, 5.1279, 4.7597], device='cuda:7'), covar=tensor([0.0644, 0.0510, 0.0352, 0.0403, 0.1175, 0.0538, 0.0332, 0.0977], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0479, 0.0373, 0.0376, 0.0370, 0.0430, 0.0256, 0.0447], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 12:24:05,549 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275814.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:24:59,133 INFO [train.py:904] (7/8) Epoch 28, batch 1800, loss[loss=0.1839, simple_loss=0.2693, pruned_loss=0.04929, over 15497.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2518, pruned_loss=0.03815, over 3330854.92 frames. ], batch size: 191, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:25:08,997 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275860.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:25:11,868 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=275862.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:25:30,539 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.316e+02 2.618e+02 3.157e+02 1.143e+03, threshold=5.235e+02, percent-clipped=6.0 2023-05-02 12:26:08,323 INFO [train.py:904] (7/8) Epoch 28, batch 1850, loss[loss=0.1638, simple_loss=0.2553, pruned_loss=0.03613, over 17124.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2526, pruned_loss=0.03794, over 3322124.07 frames. ], batch size: 55, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:26:33,825 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=275921.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:26:49,184 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1856, 3.8769, 4.4740, 2.1957, 4.6063, 4.7019, 3.4635, 3.5628], device='cuda:7'), covar=tensor([0.0703, 0.0299, 0.0210, 0.1260, 0.0086, 0.0195, 0.0429, 0.0445], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0111, 0.0102, 0.0139, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:27:05,327 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3349, 5.6910, 5.4469, 5.5465, 5.1908, 5.1010, 5.1054, 5.8330], device='cuda:7'), covar=tensor([0.1438, 0.0890, 0.1077, 0.0845, 0.0879, 0.0823, 0.1300, 0.0855], device='cuda:7'), in_proj_covar=tensor([0.0724, 0.0880, 0.0716, 0.0678, 0.0556, 0.0555, 0.0740, 0.0686], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:27:16,676 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=275952.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:27:17,459 INFO [train.py:904] (7/8) Epoch 28, batch 1900, loss[loss=0.1649, simple_loss=0.2618, pruned_loss=0.034, over 17050.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2513, pruned_loss=0.03688, over 3323787.17 frames. ], batch size: 53, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:27:22,933 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-02 12:27:32,182 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=275963.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:27:39,846 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2276, 3.9178, 4.4103, 2.3422, 4.6036, 4.7139, 3.5647, 3.6386], device='cuda:7'), covar=tensor([0.0727, 0.0323, 0.0304, 0.1224, 0.0099, 0.0188, 0.0419, 0.0427], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0110, 0.0101, 0.0139, 0.0086, 0.0131, 0.0129, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:27:44,514 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5214, 4.3267, 4.5772, 4.7270, 4.8708, 4.4111, 4.7844, 4.8629], device='cuda:7'), covar=tensor([0.1959, 0.1358, 0.1599, 0.0827, 0.0618, 0.1214, 0.1656, 0.1070], device='cuda:7'), in_proj_covar=tensor([0.0693, 0.0843, 0.0981, 0.0858, 0.0652, 0.0685, 0.0719, 0.0832], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:27:47,689 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.028e+02 2.382e+02 2.719e+02 5.605e+02, threshold=4.765e+02, percent-clipped=1.0 2023-05-02 12:28:30,506 INFO [train.py:904] (7/8) Epoch 28, batch 1950, loss[loss=0.1563, simple_loss=0.257, pruned_loss=0.02777, over 17143.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2514, pruned_loss=0.03657, over 3324792.44 frames. ], batch size: 47, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:28:30,834 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276003.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 12:28:40,998 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276011.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:28:44,272 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276013.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:29:24,699 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276041.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:29:37,551 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276051.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 12:29:40,570 INFO [train.py:904] (7/8) Epoch 28, batch 2000, loss[loss=0.1445, simple_loss=0.2332, pruned_loss=0.02789, over 15761.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2515, pruned_loss=0.03686, over 3310404.70 frames. ], batch size: 35, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:30:11,366 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.574e+02 2.104e+02 2.511e+02 2.919e+02 7.567e+02, threshold=5.023e+02, percent-clipped=3.0 2023-05-02 12:30:31,278 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276089.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:30:50,230 INFO [train.py:904] (7/8) Epoch 28, batch 2050, loss[loss=0.166, simple_loss=0.2535, pruned_loss=0.03921, over 17200.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2516, pruned_loss=0.03751, over 3318326.51 frames. ], batch size: 44, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:30:54,958 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 12:31:09,043 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-05-02 12:31:21,953 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7373, 6.0785, 5.8322, 5.9211, 5.4742, 5.4948, 5.5117, 6.2259], device='cuda:7'), covar=tensor([0.1467, 0.0945, 0.1091, 0.0916, 0.0935, 0.0622, 0.1204, 0.0840], device='cuda:7'), in_proj_covar=tensor([0.0727, 0.0889, 0.0723, 0.0683, 0.0560, 0.0557, 0.0745, 0.0691], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:31:26,617 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0714, 2.2332, 2.2553, 3.7504, 2.1611, 2.5331, 2.2788, 2.3550], device='cuda:7'), covar=tensor([0.1607, 0.3680, 0.3205, 0.0714, 0.4034, 0.2614, 0.4092, 0.3073], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0475, 0.0388, 0.0338, 0.0447, 0.0545, 0.0447, 0.0556], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:31:31,887 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0916, 4.6239, 4.5732, 3.3181, 3.8306, 4.5138, 4.0883, 2.8290], device='cuda:7'), covar=tensor([0.0485, 0.0068, 0.0048, 0.0374, 0.0157, 0.0105, 0.0101, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0092, 0.0092, 0.0137, 0.0103, 0.0116, 0.0100, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 12:32:00,360 INFO [train.py:904] (7/8) Epoch 28, batch 2100, loss[loss=0.1476, simple_loss=0.2346, pruned_loss=0.03036, over 15987.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2527, pruned_loss=0.03804, over 3325683.80 frames. ], batch size: 35, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:32:15,434 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3124, 4.3041, 4.6373, 4.6163, 4.6631, 4.4008, 4.3824, 4.2977], device='cuda:7'), covar=tensor([0.0391, 0.0787, 0.0442, 0.0460, 0.0521, 0.0473, 0.0833, 0.0749], device='cuda:7'), in_proj_covar=tensor([0.0443, 0.0498, 0.0481, 0.0444, 0.0529, 0.0508, 0.0583, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 12:32:30,700 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.091e+02 2.490e+02 2.967e+02 8.308e+02, threshold=4.979e+02, percent-clipped=2.0 2023-05-02 12:33:09,379 INFO [train.py:904] (7/8) Epoch 28, batch 2150, loss[loss=0.1694, simple_loss=0.2493, pruned_loss=0.04473, over 16338.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2547, pruned_loss=0.03949, over 3311779.58 frames. ], batch size: 165, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:33:27,928 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276216.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:34:18,071 INFO [train.py:904] (7/8) Epoch 28, batch 2200, loss[loss=0.1567, simple_loss=0.262, pruned_loss=0.02573, over 17038.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.255, pruned_loss=0.0396, over 3310389.90 frames. ], batch size: 50, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:34:23,933 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.4308, 5.7920, 5.5516, 5.6375, 5.1840, 5.2531, 5.2190, 5.9466], device='cuda:7'), covar=tensor([0.1402, 0.1058, 0.1129, 0.0968, 0.1013, 0.0776, 0.1373, 0.0890], device='cuda:7'), in_proj_covar=tensor([0.0728, 0.0889, 0.0721, 0.0683, 0.0559, 0.0558, 0.0745, 0.0691], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:34:50,533 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 2.258e+02 2.667e+02 3.359e+02 8.724e+02, threshold=5.333e+02, percent-clipped=6.0 2023-05-02 12:35:27,987 INFO [train.py:904] (7/8) Epoch 28, batch 2250, loss[loss=0.1583, simple_loss=0.2501, pruned_loss=0.03329, over 17093.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2549, pruned_loss=0.03961, over 3318999.56 frames. ], batch size: 49, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:35:35,479 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276308.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:35:35,538 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0199, 4.7958, 5.0464, 5.2700, 5.4628, 4.7995, 5.4579, 5.4619], device='cuda:7'), covar=tensor([0.2054, 0.1394, 0.1775, 0.0805, 0.0557, 0.0957, 0.0554, 0.0654], device='cuda:7'), in_proj_covar=tensor([0.0700, 0.0852, 0.0990, 0.0867, 0.0659, 0.0691, 0.0725, 0.0840], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:35:48,037 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:35:58,343 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-02 12:36:32,855 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8452, 4.5230, 4.5215, 5.0430, 5.2192, 4.6713, 5.1856, 5.2445], device='cuda:7'), covar=tensor([0.1931, 0.1531, 0.2837, 0.1141, 0.0874, 0.1485, 0.1079, 0.1087], device='cuda:7'), in_proj_covar=tensor([0.0699, 0.0851, 0.0988, 0.0865, 0.0658, 0.0689, 0.0724, 0.0839], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:36:37,088 INFO [train.py:904] (7/8) Epoch 28, batch 2300, loss[loss=0.1693, simple_loss=0.2504, pruned_loss=0.04412, over 16435.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2548, pruned_loss=0.0396, over 3314090.69 frames. ], batch size: 146, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:36:52,507 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3548, 5.2908, 5.1801, 4.6396, 4.8038, 5.2403, 5.2135, 4.8098], device='cuda:7'), covar=tensor([0.0594, 0.0534, 0.0344, 0.0392, 0.1158, 0.0503, 0.0365, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0482, 0.0376, 0.0378, 0.0371, 0.0433, 0.0257, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 12:37:08,708 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.645e+02 2.043e+02 2.409e+02 2.878e+02 4.638e+02, threshold=4.818e+02, percent-clipped=0.0 2023-05-02 12:37:11,205 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4146, 2.6577, 2.1726, 2.4246, 2.9712, 2.7118, 3.1160, 3.1520], device='cuda:7'), covar=tensor([0.0254, 0.0474, 0.0636, 0.0510, 0.0339, 0.0416, 0.0296, 0.0340], device='cuda:7'), in_proj_covar=tensor([0.0238, 0.0253, 0.0240, 0.0241, 0.0254, 0.0252, 0.0251, 0.0250], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:37:12,891 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276378.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:37:46,505 INFO [train.py:904] (7/8) Epoch 28, batch 2350, loss[loss=0.1822, simple_loss=0.2903, pruned_loss=0.03708, over 16602.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2555, pruned_loss=0.03972, over 3321871.10 frames. ], batch size: 62, lr: 2.41e-03, grad_scale: 4.0 2023-05-02 12:38:10,941 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2946, 3.3383, 3.7677, 2.2199, 3.0791, 2.3612, 3.6363, 3.6042], device='cuda:7'), covar=tensor([0.0317, 0.1124, 0.0524, 0.2128, 0.0942, 0.1118, 0.0696, 0.1291], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0172, 0.0170, 0.0157, 0.0149, 0.0133, 0.0147, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 12:38:54,364 INFO [train.py:904] (7/8) Epoch 28, batch 2400, loss[loss=0.1871, simple_loss=0.2678, pruned_loss=0.05318, over 16647.00 frames. ], tot_loss[loss=0.1678, simple_loss=0.2559, pruned_loss=0.03982, over 3324714.82 frames. ], batch size: 134, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:39:26,389 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.588e+02 2.076e+02 2.394e+02 2.729e+02 1.080e+03, threshold=4.789e+02, percent-clipped=2.0 2023-05-02 12:39:39,841 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9406, 4.5046, 4.4846, 3.3039, 3.6535, 4.4697, 3.9578, 2.7579], device='cuda:7'), covar=tensor([0.0502, 0.0074, 0.0046, 0.0351, 0.0168, 0.0090, 0.0105, 0.0470], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0136, 0.0103, 0.0116, 0.0100, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 12:40:04,311 INFO [train.py:904] (7/8) Epoch 28, batch 2450, loss[loss=0.1858, simple_loss=0.2862, pruned_loss=0.04267, over 17109.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2559, pruned_loss=0.03961, over 3322373.07 frames. ], batch size: 55, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:40:23,639 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276516.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:40:23,824 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7385, 2.6135, 2.3352, 2.4524, 2.9792, 2.7172, 3.1899, 3.1712], device='cuda:7'), covar=tensor([0.0174, 0.0513, 0.0602, 0.0596, 0.0375, 0.0499, 0.0310, 0.0359], device='cuda:7'), in_proj_covar=tensor([0.0239, 0.0253, 0.0241, 0.0242, 0.0254, 0.0252, 0.0251, 0.0251], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:40:38,022 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7479, 3.3754, 3.7869, 1.9950, 3.8559, 3.9070, 3.2266, 2.9490], device='cuda:7'), covar=tensor([0.0700, 0.0266, 0.0181, 0.1163, 0.0098, 0.0202, 0.0376, 0.0442], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0110, 0.0101, 0.0138, 0.0086, 0.0131, 0.0129, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:41:13,997 INFO [train.py:904] (7/8) Epoch 28, batch 2500, loss[loss=0.1742, simple_loss=0.2669, pruned_loss=0.04076, over 17267.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2561, pruned_loss=0.03894, over 3321580.58 frames. ], batch size: 52, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:41:23,415 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9506, 4.1383, 3.1378, 2.4763, 2.6885, 2.7037, 4.4137, 3.4947], device='cuda:7'), covar=tensor([0.2695, 0.0567, 0.1772, 0.2941, 0.2844, 0.2099, 0.0448, 0.1465], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0280, 0.0317, 0.0331, 0.0309, 0.0281, 0.0309, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 12:41:30,329 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276564.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:41:45,572 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.004e+02 2.344e+02 2.885e+02 4.375e+02, threshold=4.688e+02, percent-clipped=0.0 2023-05-02 12:42:23,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9596, 2.1081, 2.6050, 2.9638, 2.7915, 3.4418, 2.4219, 3.4658], device='cuda:7'), covar=tensor([0.0324, 0.0624, 0.0402, 0.0408, 0.0435, 0.0238, 0.0573, 0.0188], device='cuda:7'), in_proj_covar=tensor([0.0201, 0.0201, 0.0189, 0.0196, 0.0210, 0.0169, 0.0205, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 12:42:24,131 INFO [train.py:904] (7/8) Epoch 28, batch 2550, loss[loss=0.1619, simple_loss=0.2564, pruned_loss=0.03373, over 17131.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2556, pruned_loss=0.03868, over 3312106.89 frames. ], batch size: 49, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:42:31,855 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276608.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:42:36,028 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276611.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:43:28,879 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8106, 3.7478, 3.8561, 3.6032, 3.7622, 4.2299, 3.8544, 3.5240], device='cuda:7'), covar=tensor([0.2104, 0.2550, 0.2679, 0.2703, 0.3116, 0.2089, 0.1735, 0.2936], device='cuda:7'), in_proj_covar=tensor([0.0437, 0.0654, 0.0718, 0.0532, 0.0707, 0.0745, 0.0559, 0.0708], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 12:43:33,182 INFO [train.py:904] (7/8) Epoch 28, batch 2600, loss[loss=0.1707, simple_loss=0.2606, pruned_loss=0.04043, over 15638.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2557, pruned_loss=0.03878, over 3306596.26 frames. ], batch size: 190, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:43:38,051 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=276656.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:43:55,781 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276668.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:44:01,096 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276672.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:44:02,628 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276673.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:44:05,672 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.592e+02 2.267e+02 2.519e+02 3.133e+02 7.416e+02, threshold=5.037e+02, percent-clipped=2.0 2023-05-02 12:44:41,588 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 12:44:43,682 INFO [train.py:904] (7/8) Epoch 28, batch 2650, loss[loss=0.1604, simple_loss=0.2573, pruned_loss=0.03172, over 16050.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2564, pruned_loss=0.03883, over 3313419.57 frames. ], batch size: 35, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:45:19,793 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:45:50,753 INFO [train.py:904] (7/8) Epoch 28, batch 2700, loss[loss=0.1662, simple_loss=0.2539, pruned_loss=0.03918, over 16763.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2567, pruned_loss=0.0387, over 3319147.17 frames. ], batch size: 83, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:46:23,524 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.276e+02 2.127e+02 2.476e+02 3.244e+02 1.015e+03, threshold=4.951e+02, percent-clipped=5.0 2023-05-02 12:46:42,631 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276790.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:46:56,333 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7055, 2.5444, 2.4808, 4.0023, 3.2755, 3.9753, 1.5589, 2.8759], device='cuda:7'), covar=tensor([0.1486, 0.0773, 0.1258, 0.0187, 0.0146, 0.0367, 0.1693, 0.0868], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0205, 0.0207, 0.0220, 0.0210, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:47:00,579 INFO [train.py:904] (7/8) Epoch 28, batch 2750, loss[loss=0.1725, simple_loss=0.2518, pruned_loss=0.04657, over 16702.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2568, pruned_loss=0.03863, over 3314484.44 frames. ], batch size: 134, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:48:07,266 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276851.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:48:09,141 INFO [train.py:904] (7/8) Epoch 28, batch 2800, loss[loss=0.1646, simple_loss=0.2656, pruned_loss=0.03181, over 17179.00 frames. ], tot_loss[loss=0.167, simple_loss=0.2567, pruned_loss=0.03865, over 3311576.36 frames. ], batch size: 46, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:48:25,019 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276863.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:48:25,287 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 12:48:41,263 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.121e+02 2.382e+02 2.775e+02 4.197e+02, threshold=4.764e+02, percent-clipped=0.0 2023-05-02 12:49:20,050 INFO [train.py:904] (7/8) Epoch 28, batch 2850, loss[loss=0.1505, simple_loss=0.2491, pruned_loss=0.02591, over 17124.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.2557, pruned_loss=0.03785, over 3315185.51 frames. ], batch size: 47, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:49:50,148 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276924.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:49:57,686 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=276931.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:50:03,793 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9106, 2.8664, 2.4015, 2.7293, 3.1501, 2.9340, 3.4628, 3.3672], device='cuda:7'), covar=tensor([0.0182, 0.0560, 0.0653, 0.0485, 0.0381, 0.0449, 0.0338, 0.0334], device='cuda:7'), in_proj_covar=tensor([0.0239, 0.0253, 0.0240, 0.0241, 0.0254, 0.0251, 0.0252, 0.0250], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:50:27,310 INFO [train.py:904] (7/8) Epoch 28, batch 2900, loss[loss=0.1851, simple_loss=0.2723, pruned_loss=0.04893, over 16675.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2546, pruned_loss=0.03813, over 3320282.14 frames. ], batch size: 57, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:50:46,679 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=276967.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:50:54,984 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=276973.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:50:58,052 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 2.124e+02 2.391e+02 2.943e+02 6.733e+02, threshold=4.783e+02, percent-clipped=5.0 2023-05-02 12:51:22,003 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=276992.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 12:51:27,157 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2494, 5.6992, 5.8452, 5.6058, 5.6850, 6.1884, 5.7161, 5.4543], device='cuda:7'), covar=tensor([0.0790, 0.1931, 0.2383, 0.1785, 0.2390, 0.0784, 0.1380, 0.2189], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0655, 0.0718, 0.0532, 0.0706, 0.0742, 0.0558, 0.0708], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 12:51:36,312 INFO [train.py:904] (7/8) Epoch 28, batch 2950, loss[loss=0.1975, simple_loss=0.2847, pruned_loss=0.05516, over 17044.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2542, pruned_loss=0.03878, over 3309011.27 frames. ], batch size: 55, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:51:48,553 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277011.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:52:01,178 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277021.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:52:05,385 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277024.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:52:45,413 INFO [train.py:904] (7/8) Epoch 28, batch 3000, loss[loss=0.159, simple_loss=0.2544, pruned_loss=0.03181, over 16676.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2544, pruned_loss=0.03923, over 3301993.24 frames. ], batch size: 57, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:52:45,414 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 12:52:54,790 INFO [train.py:938] (7/8) Epoch 28, validation: loss=0.1335, simple_loss=0.2385, pruned_loss=0.01427, over 944034.00 frames. 2023-05-02 12:52:54,791 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 12:52:56,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5915, 2.7162, 2.3965, 2.4405, 3.0007, 2.6500, 3.1499, 3.1626], device='cuda:7'), covar=tensor([0.0207, 0.0471, 0.0559, 0.0523, 0.0354, 0.0455, 0.0329, 0.0351], device='cuda:7'), in_proj_covar=tensor([0.0239, 0.0253, 0.0240, 0.0240, 0.0254, 0.0251, 0.0251, 0.0250], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:53:21,077 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277072.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:53:25,667 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 2.148e+02 2.585e+02 3.132e+02 1.139e+03, threshold=5.170e+02, percent-clipped=2.0 2023-05-02 12:53:34,202 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2131, 2.7800, 3.0435, 1.7124, 3.1877, 3.1968, 2.7388, 2.5425], device='cuda:7'), covar=tensor([0.0909, 0.0363, 0.0311, 0.1328, 0.0172, 0.0278, 0.0524, 0.0490], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0103, 0.0141, 0.0088, 0.0133, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:54:02,144 INFO [train.py:904] (7/8) Epoch 28, batch 3050, loss[loss=0.175, simple_loss=0.2691, pruned_loss=0.04045, over 16755.00 frames. ], tot_loss[loss=0.1669, simple_loss=0.2543, pruned_loss=0.03977, over 3309501.88 frames. ], batch size: 62, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:55:01,474 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277146.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:55:11,562 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.70 vs. limit=2.0 2023-05-02 12:55:11,785 INFO [train.py:904] (7/8) Epoch 28, batch 3100, loss[loss=0.1669, simple_loss=0.2476, pruned_loss=0.04306, over 16420.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2536, pruned_loss=0.03942, over 3317277.45 frames. ], batch size: 75, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:55:43,286 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 2.064e+02 2.462e+02 2.812e+02 4.738e+02, threshold=4.925e+02, percent-clipped=0.0 2023-05-02 12:55:45,448 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277177.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:56:00,516 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 12:56:12,079 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0874, 3.8668, 4.3665, 2.2057, 4.5723, 4.6609, 3.3991, 3.5013], device='cuda:7'), covar=tensor([0.0726, 0.0285, 0.0250, 0.1197, 0.0085, 0.0194, 0.0424, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0103, 0.0140, 0.0088, 0.0133, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 12:56:21,091 INFO [train.py:904] (7/8) Epoch 28, batch 3150, loss[loss=0.1554, simple_loss=0.254, pruned_loss=0.0284, over 17259.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2517, pruned_loss=0.03876, over 3321326.71 frames. ], batch size: 52, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:56:44,566 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277219.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:56:44,694 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277219.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:57:10,803 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7567, 4.7028, 4.5903, 4.0114, 4.6660, 1.7095, 4.4220, 4.2612], device='cuda:7'), covar=tensor([0.0172, 0.0154, 0.0232, 0.0407, 0.0147, 0.3173, 0.0182, 0.0286], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0179, 0.0217, 0.0190, 0.0193, 0.0222, 0.0206, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 12:57:10,861 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277238.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:57:30,728 INFO [train.py:904] (7/8) Epoch 28, batch 3200, loss[loss=0.1807, simple_loss=0.2771, pruned_loss=0.04211, over 16661.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.251, pruned_loss=0.03814, over 3322208.11 frames. ], batch size: 57, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:57:44,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7967, 1.9834, 2.3822, 2.7118, 2.7280, 2.7293, 2.0471, 2.9120], device='cuda:7'), covar=tensor([0.0221, 0.0518, 0.0382, 0.0304, 0.0341, 0.0311, 0.0545, 0.0212], device='cuda:7'), in_proj_covar=tensor([0.0202, 0.0201, 0.0190, 0.0196, 0.0211, 0.0169, 0.0206, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 12:57:50,722 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277267.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:58:01,818 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.445e+02 1.963e+02 2.375e+02 2.950e+02 8.294e+02, threshold=4.751e+02, percent-clipped=4.0 2023-05-02 12:58:07,821 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277280.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:58:17,849 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277287.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 12:58:39,068 INFO [train.py:904] (7/8) Epoch 28, batch 3250, loss[loss=0.1399, simple_loss=0.2333, pruned_loss=0.02329, over 17220.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2504, pruned_loss=0.03784, over 3329959.54 frames. ], batch size: 44, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 12:58:42,504 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277305.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:58:55,897 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277315.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:59:09,115 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277324.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 12:59:21,775 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 12:59:47,707 INFO [train.py:904] (7/8) Epoch 28, batch 3300, loss[loss=0.1824, simple_loss=0.2581, pruned_loss=0.05338, over 16469.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2515, pruned_loss=0.03782, over 3331899.27 frames. ], batch size: 75, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:00:05,293 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277366.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:00:06,292 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277367.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:00:10,381 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1457, 4.0143, 4.2134, 4.3301, 4.4036, 3.9876, 4.1961, 4.4111], device='cuda:7'), covar=tensor([0.1668, 0.1156, 0.1303, 0.0708, 0.0664, 0.1397, 0.2390, 0.0761], device='cuda:7'), in_proj_covar=tensor([0.0709, 0.0862, 0.1002, 0.0879, 0.0667, 0.0701, 0.0734, 0.0847], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:00:12,569 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277372.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:00:18,088 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.125e+02 2.555e+02 3.019e+02 5.327e+02, threshold=5.111e+02, percent-clipped=1.0 2023-05-02 13:00:55,913 INFO [train.py:904] (7/8) Epoch 28, batch 3350, loss[loss=0.1679, simple_loss=0.2641, pruned_loss=0.03588, over 16667.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2527, pruned_loss=0.03824, over 3326264.98 frames. ], batch size: 57, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:01:24,854 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7944, 3.9534, 2.9835, 2.3608, 2.5133, 2.5208, 4.0281, 3.3962], device='cuda:7'), covar=tensor([0.2771, 0.0584, 0.1716, 0.3139, 0.2923, 0.2125, 0.0569, 0.1560], device='cuda:7'), in_proj_covar=tensor([0.0337, 0.0279, 0.0315, 0.0329, 0.0308, 0.0279, 0.0307, 0.0357], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 13:01:50,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2420, 5.1722, 4.9524, 4.3727, 5.0747, 1.9814, 4.8179, 4.8607], device='cuda:7'), covar=tensor([0.0099, 0.0103, 0.0239, 0.0481, 0.0101, 0.2991, 0.0156, 0.0251], device='cuda:7'), in_proj_covar=tensor([0.0186, 0.0181, 0.0219, 0.0192, 0.0195, 0.0224, 0.0208, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:01:55,673 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277446.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:02:03,950 INFO [train.py:904] (7/8) Epoch 28, batch 3400, loss[loss=0.1567, simple_loss=0.2446, pruned_loss=0.03436, over 17159.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2531, pruned_loss=0.03796, over 3320998.22 frames. ], batch size: 46, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:02:34,599 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.063e+02 2.423e+02 2.740e+02 5.592e+02, threshold=4.846e+02, percent-clipped=2.0 2023-05-02 13:03:01,427 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277494.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:03:13,663 INFO [train.py:904] (7/8) Epoch 28, batch 3450, loss[loss=0.1483, simple_loss=0.2344, pruned_loss=0.03112, over 17222.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2518, pruned_loss=0.03766, over 3317190.50 frames. ], batch size: 45, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:03:19,473 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 13:03:36,031 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277519.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:03:56,528 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277533.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:23,515 INFO [train.py:904] (7/8) Epoch 28, batch 3500, loss[loss=0.1651, simple_loss=0.2597, pruned_loss=0.03523, over 16759.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2501, pruned_loss=0.03727, over 3313377.08 frames. ], batch size: 57, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:04:25,821 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277554.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:42,186 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277567.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:47,762 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277570.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:54,620 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277575.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:04:55,306 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.480e+02 2.163e+02 2.511e+02 2.754e+02 5.260e+02, threshold=5.022e+02, percent-clipped=2.0 2023-05-02 13:05:10,716 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277587.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 13:05:32,631 INFO [train.py:904] (7/8) Epoch 28, batch 3550, loss[loss=0.1839, simple_loss=0.2588, pruned_loss=0.05448, over 16892.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2491, pruned_loss=0.03721, over 3314883.29 frames. ], batch size: 109, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:05:35,877 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-02 13:05:49,405 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277615.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:06:11,703 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277631.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:06:17,722 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277635.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:06:35,288 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7151, 3.8730, 2.5199, 4.4986, 2.9813, 4.3763, 2.4964, 3.1939], device='cuda:7'), covar=tensor([0.0371, 0.0413, 0.1673, 0.0303, 0.0932, 0.0562, 0.1703, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0186, 0.0199, 0.0179, 0.0183, 0.0227, 0.0207, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 13:06:42,336 INFO [train.py:904] (7/8) Epoch 28, batch 3600, loss[loss=0.1655, simple_loss=0.2683, pruned_loss=0.03138, over 17273.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2483, pruned_loss=0.03649, over 3320420.81 frames. ], batch size: 52, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:06:53,442 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277661.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:07:02,053 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277667.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:07:14,468 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.038e+02 2.278e+02 2.834e+02 5.503e+02, threshold=4.555e+02, percent-clipped=2.0 2023-05-02 13:07:37,679 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277691.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:07:53,650 INFO [train.py:904] (7/8) Epoch 28, batch 3650, loss[loss=0.1545, simple_loss=0.2282, pruned_loss=0.04039, over 16518.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2478, pruned_loss=0.03763, over 3302909.81 frames. ], batch size: 75, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:08:12,450 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:08:38,687 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8883, 3.8304, 3.9601, 4.0342, 4.1066, 3.7445, 3.9675, 4.1390], device='cuda:7'), covar=tensor([0.1577, 0.1125, 0.1127, 0.0690, 0.0703, 0.1988, 0.2348, 0.0795], device='cuda:7'), in_proj_covar=tensor([0.0710, 0.0864, 0.1006, 0.0881, 0.0669, 0.0702, 0.0737, 0.0850], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:09:06,685 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=277752.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 13:09:07,238 INFO [train.py:904] (7/8) Epoch 28, batch 3700, loss[loss=0.164, simple_loss=0.24, pruned_loss=0.04404, over 16836.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2469, pruned_loss=0.03924, over 3284507.57 frames. ], batch size: 96, lr: 2.41e-03, grad_scale: 8.0 2023-05-02 13:09:41,449 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.185e+02 2.658e+02 3.107e+02 6.146e+02, threshold=5.316e+02, percent-clipped=3.0 2023-05-02 13:10:08,971 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9932, 4.0928, 4.3089, 4.2935, 4.3244, 4.0936, 4.1210, 4.0529], device='cuda:7'), covar=tensor([0.0387, 0.0646, 0.0422, 0.0412, 0.0491, 0.0456, 0.0742, 0.0567], device='cuda:7'), in_proj_covar=tensor([0.0448, 0.0507, 0.0487, 0.0449, 0.0536, 0.0514, 0.0592, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 13:10:22,506 INFO [train.py:904] (7/8) Epoch 28, batch 3750, loss[loss=0.1686, simple_loss=0.2478, pruned_loss=0.04469, over 16482.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2471, pruned_loss=0.04038, over 3274966.82 frames. ], batch size: 68, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:11:06,985 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:11:35,743 INFO [train.py:904] (7/8) Epoch 28, batch 3800, loss[loss=0.1657, simple_loss=0.2447, pruned_loss=0.04338, over 16389.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2482, pruned_loss=0.0414, over 3271847.63 frames. ], batch size: 68, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:12:08,798 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277875.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:12:11,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.049e+02 2.321e+02 2.888e+02 4.993e+02, threshold=4.641e+02, percent-clipped=0.0 2023-05-02 13:12:17,664 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:12:41,426 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-05-02 13:12:48,160 INFO [train.py:904] (7/8) Epoch 28, batch 3850, loss[loss=0.1559, simple_loss=0.2458, pruned_loss=0.03294, over 16593.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2489, pruned_loss=0.04234, over 3272143.02 frames. ], batch size: 62, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:12:59,662 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277910.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:13:19,028 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=277923.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:13:23,069 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=277926.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:14:00,817 INFO [train.py:904] (7/8) Epoch 28, batch 3900, loss[loss=0.1812, simple_loss=0.2647, pruned_loss=0.04884, over 16636.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2486, pruned_loss=0.04308, over 3280928.16 frames. ], batch size: 62, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:14:02,442 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-05-02 13:14:13,161 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=277961.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:14:28,367 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=277971.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:14:36,146 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.147e+02 2.438e+02 2.765e+02 5.958e+02, threshold=4.876e+02, percent-clipped=2.0 2023-05-02 13:15:16,810 INFO [train.py:904] (7/8) Epoch 28, batch 3950, loss[loss=0.1617, simple_loss=0.2457, pruned_loss=0.03886, over 17062.00 frames. ], tot_loss[loss=0.1676, simple_loss=0.2481, pruned_loss=0.04355, over 3273548.18 frames. ], batch size: 50, lr: 2.40e-03, grad_scale: 4.0 2023-05-02 13:15:25,692 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278009.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:15:58,285 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278032.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:16:19,862 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278047.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 13:16:26,871 INFO [train.py:904] (7/8) Epoch 28, batch 4000, loss[loss=0.1898, simple_loss=0.273, pruned_loss=0.05329, over 11768.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2483, pruned_loss=0.04383, over 3272281.03 frames. ], batch size: 246, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:16:28,148 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3378, 2.2237, 1.8888, 1.9988, 2.5051, 2.1694, 2.1530, 2.5886], device='cuda:7'), covar=tensor([0.0271, 0.0478, 0.0603, 0.0524, 0.0301, 0.0414, 0.0280, 0.0283], device='cuda:7'), in_proj_covar=tensor([0.0238, 0.0250, 0.0237, 0.0238, 0.0251, 0.0248, 0.0250, 0.0249], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:17:01,537 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 1.968e+02 2.308e+02 2.954e+02 7.046e+02, threshold=4.616e+02, percent-clipped=2.0 2023-05-02 13:17:11,108 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-02 13:17:38,528 INFO [train.py:904] (7/8) Epoch 28, batch 4050, loss[loss=0.1687, simple_loss=0.2519, pruned_loss=0.0427, over 16670.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2484, pruned_loss=0.04286, over 3283704.89 frames. ], batch size: 57, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:18:00,123 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-05-02 13:18:20,320 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5302, 2.3262, 1.9094, 2.0872, 2.5837, 2.2543, 2.3182, 2.7076], device='cuda:7'), covar=tensor([0.0183, 0.0397, 0.0578, 0.0459, 0.0256, 0.0363, 0.0208, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0238, 0.0250, 0.0237, 0.0238, 0.0251, 0.0248, 0.0250, 0.0249], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:18:52,740 INFO [train.py:904] (7/8) Epoch 28, batch 4100, loss[loss=0.1779, simple_loss=0.2603, pruned_loss=0.04775, over 16668.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2505, pruned_loss=0.04272, over 3280197.89 frames. ], batch size: 62, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:19:24,740 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6416, 2.9825, 3.3111, 2.0077, 2.8718, 2.1296, 3.1943, 3.2494], device='cuda:7'), covar=tensor([0.0262, 0.0915, 0.0539, 0.2143, 0.0854, 0.1024, 0.0610, 0.1042], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0174, 0.0171, 0.0158, 0.0150, 0.0133, 0.0148, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 13:19:29,062 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 1.827e+02 2.078e+02 2.388e+02 3.753e+02, threshold=4.156e+02, percent-clipped=0.0 2023-05-02 13:20:09,088 INFO [train.py:904] (7/8) Epoch 28, batch 4150, loss[loss=0.2001, simple_loss=0.285, pruned_loss=0.05756, over 16832.00 frames. ], tot_loss[loss=0.1741, simple_loss=0.2578, pruned_loss=0.0452, over 3220362.93 frames. ], batch size: 116, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:20:19,342 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278210.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:20:42,380 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278226.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:20:43,990 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 13:21:22,143 INFO [train.py:904] (7/8) Epoch 28, batch 4200, loss[loss=0.1858, simple_loss=0.2816, pruned_loss=0.04498, over 16685.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.2652, pruned_loss=0.04709, over 3203790.24 frames. ], batch size: 124, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:21:30,724 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278258.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:21:47,445 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 13:21:51,760 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4944, 2.6221, 2.2047, 2.4197, 2.9719, 2.5007, 3.0317, 3.1240], device='cuda:7'), covar=tensor([0.0179, 0.0470, 0.0634, 0.0522, 0.0337, 0.0515, 0.0269, 0.0336], device='cuda:7'), in_proj_covar=tensor([0.0238, 0.0250, 0.0237, 0.0238, 0.0251, 0.0248, 0.0249, 0.0249], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:21:54,357 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278274.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:21:57,452 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8607, 4.9124, 4.7226, 4.3857, 4.4141, 4.8664, 4.6317, 4.5026], device='cuda:7'), covar=tensor([0.0604, 0.0555, 0.0322, 0.0353, 0.1016, 0.0402, 0.0453, 0.0675], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0485, 0.0377, 0.0379, 0.0375, 0.0435, 0.0257, 0.0451], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 13:21:58,172 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.712e+02 2.281e+02 2.720e+02 3.199e+02 7.364e+02, threshold=5.441e+02, percent-clipped=7.0 2023-05-02 13:22:37,407 INFO [train.py:904] (7/8) Epoch 28, batch 4250, loss[loss=0.1741, simple_loss=0.2731, pruned_loss=0.03756, over 16828.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.269, pruned_loss=0.04677, over 3196787.45 frames. ], batch size: 116, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:22:52,351 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 13:22:55,543 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7867, 3.9494, 3.1266, 2.4311, 2.8512, 2.7865, 4.3033, 3.6096], device='cuda:7'), covar=tensor([0.2909, 0.0701, 0.1761, 0.3160, 0.2619, 0.1858, 0.0589, 0.1283], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0277, 0.0313, 0.0328, 0.0308, 0.0278, 0.0306, 0.0355], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 13:23:07,802 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=278323.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:23:13,437 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:23:44,733 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278347.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 13:23:52,570 INFO [train.py:904] (7/8) Epoch 28, batch 4300, loss[loss=0.1816, simple_loss=0.2718, pruned_loss=0.0457, over 11847.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2704, pruned_loss=0.04611, over 3186179.60 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:24:25,538 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=278374.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:24:29,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.019e+02 2.401e+02 2.918e+02 4.618e+02, threshold=4.803e+02, percent-clipped=0.0 2023-05-02 13:24:40,745 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278384.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:24:56,297 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278395.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:25:07,431 INFO [train.py:904] (7/8) Epoch 28, batch 4350, loss[loss=0.18, simple_loss=0.2708, pruned_loss=0.04458, over 16593.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2735, pruned_loss=0.04715, over 3189706.54 frames. ], batch size: 62, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:25:47,111 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7913, 1.4162, 1.7161, 1.6690, 1.7580, 1.9427, 1.6522, 1.8048], device='cuda:7'), covar=tensor([0.0286, 0.0423, 0.0252, 0.0322, 0.0310, 0.0198, 0.0448, 0.0134], device='cuda:7'), in_proj_covar=tensor([0.0201, 0.0200, 0.0189, 0.0194, 0.0210, 0.0168, 0.0204, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:25:57,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2102, 3.8306, 3.7532, 2.2122, 3.5149, 3.8736, 3.5321, 1.7594], device='cuda:7'), covar=tensor([0.0667, 0.0058, 0.0089, 0.0591, 0.0120, 0.0123, 0.0122, 0.0663], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0102, 0.0115, 0.0099, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 13:25:57,127 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278435.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:26:22,394 INFO [train.py:904] (7/8) Epoch 28, batch 4400, loss[loss=0.1825, simple_loss=0.2767, pruned_loss=0.04414, over 17140.00 frames. ], tot_loss[loss=0.1861, simple_loss=0.2755, pruned_loss=0.04837, over 3178199.01 frames. ], batch size: 49, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:26:31,331 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 13:26:34,027 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8727, 3.7670, 3.9306, 4.0156, 4.0854, 3.7122, 4.0267, 4.1364], device='cuda:7'), covar=tensor([0.1353, 0.1112, 0.1090, 0.0624, 0.0538, 0.1953, 0.0895, 0.0668], device='cuda:7'), in_proj_covar=tensor([0.0686, 0.0839, 0.0971, 0.0854, 0.0649, 0.0679, 0.0711, 0.0823], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:26:58,178 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.247e+02 2.621e+02 3.049e+02 4.749e+02, threshold=5.243e+02, percent-clipped=0.0 2023-05-02 13:27:35,972 INFO [train.py:904] (7/8) Epoch 28, batch 4450, loss[loss=0.1761, simple_loss=0.2762, pruned_loss=0.03795, over 16877.00 frames. ], tot_loss[loss=0.1889, simple_loss=0.2789, pruned_loss=0.04947, over 3187755.05 frames. ], batch size: 96, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:27:48,025 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=278511.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:28:49,081 INFO [train.py:904] (7/8) Epoch 28, batch 4500, loss[loss=0.1983, simple_loss=0.2832, pruned_loss=0.05667, over 16913.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2788, pruned_loss=0.05003, over 3197185.64 frames. ], batch size: 109, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:28:59,020 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2464, 2.1668, 2.7377, 3.1602, 3.0119, 3.6818, 2.4380, 3.6305], device='cuda:7'), covar=tensor([0.0213, 0.0588, 0.0336, 0.0315, 0.0321, 0.0162, 0.0583, 0.0129], device='cuda:7'), in_proj_covar=tensor([0.0200, 0.0199, 0.0188, 0.0194, 0.0210, 0.0168, 0.0203, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:29:17,582 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=278572.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:29:23,508 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 1.834e+02 2.148e+02 2.381e+02 3.720e+02, threshold=4.296e+02, percent-clipped=0.0 2023-05-02 13:30:01,305 INFO [train.py:904] (7/8) Epoch 28, batch 4550, loss[loss=0.1966, simple_loss=0.2862, pruned_loss=0.05348, over 17243.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2796, pruned_loss=0.0509, over 3198847.56 frames. ], batch size: 52, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:30:36,997 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278627.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:31:12,175 INFO [train.py:904] (7/8) Epoch 28, batch 4600, loss[loss=0.2012, simple_loss=0.2944, pruned_loss=0.05402, over 16590.00 frames. ], tot_loss[loss=0.192, simple_loss=0.281, pruned_loss=0.05156, over 3198085.67 frames. ], batch size: 75, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:31:43,900 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=278675.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:31:46,045 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 1.855e+02 2.019e+02 2.344e+02 4.106e+02, threshold=4.037e+02, percent-clipped=0.0 2023-05-02 13:31:49,272 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278679.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:31:51,622 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 13:32:22,031 INFO [train.py:904] (7/8) Epoch 28, batch 4650, loss[loss=0.1738, simple_loss=0.264, pruned_loss=0.04179, over 16789.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2803, pruned_loss=0.05178, over 3197815.16 frames. ], batch size: 83, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:32:24,746 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.65 vs. limit=5.0 2023-05-02 13:33:01,593 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278730.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:33:06,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-02 13:33:33,722 INFO [train.py:904] (7/8) Epoch 28, batch 4700, loss[loss=0.1804, simple_loss=0.2641, pruned_loss=0.04837, over 16661.00 frames. ], tot_loss[loss=0.1894, simple_loss=0.2775, pruned_loss=0.05067, over 3196383.78 frames. ], batch size: 57, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:34:07,039 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 1.813e+02 1.994e+02 2.322e+02 3.636e+02, threshold=3.989e+02, percent-clipped=0.0 2023-05-02 13:34:45,689 INFO [train.py:904] (7/8) Epoch 28, batch 4750, loss[loss=0.1819, simple_loss=0.2693, pruned_loss=0.04724, over 16260.00 frames. ], tot_loss[loss=0.1851, simple_loss=0.2734, pruned_loss=0.0484, over 3204848.46 frames. ], batch size: 165, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:35:24,092 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-05-02 13:35:47,882 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3824, 3.3845, 2.6179, 2.1505, 2.2483, 2.3445, 3.6247, 3.0390], device='cuda:7'), covar=tensor([0.3348, 0.0756, 0.2193, 0.3287, 0.2536, 0.2287, 0.0632, 0.1512], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0275, 0.0312, 0.0326, 0.0306, 0.0276, 0.0304, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 13:35:53,418 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9634, 2.1857, 2.1902, 3.5292, 2.1379, 2.5251, 2.2800, 2.3536], device='cuda:7'), covar=tensor([0.1732, 0.4016, 0.3286, 0.0732, 0.4332, 0.2777, 0.4062, 0.3399], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0476, 0.0384, 0.0338, 0.0446, 0.0545, 0.0447, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:35:57,564 INFO [train.py:904] (7/8) Epoch 28, batch 4800, loss[loss=0.213, simple_loss=0.2926, pruned_loss=0.0667, over 12023.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2701, pruned_loss=0.04661, over 3205832.64 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:36:18,867 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=278867.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:36:20,667 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-05-02 13:36:31,926 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9671, 5.2758, 4.8341, 5.1855, 4.8870, 4.5940, 4.8039, 5.3620], device='cuda:7'), covar=tensor([0.2176, 0.1330, 0.1870, 0.1245, 0.1344, 0.1531, 0.1926, 0.1625], device='cuda:7'), in_proj_covar=tensor([0.0719, 0.0879, 0.0715, 0.0675, 0.0553, 0.0549, 0.0733, 0.0683], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:36:32,587 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 1.887e+02 2.183e+02 2.468e+02 4.688e+02, threshold=4.367e+02, percent-clipped=1.0 2023-05-02 13:37:13,024 INFO [train.py:904] (7/8) Epoch 28, batch 4850, loss[loss=0.1828, simple_loss=0.2785, pruned_loss=0.04356, over 16448.00 frames. ], tot_loss[loss=0.1811, simple_loss=0.2707, pruned_loss=0.04571, over 3183543.06 frames. ], batch size: 146, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:37:26,837 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1919, 2.8427, 3.0805, 1.8063, 3.2114, 3.2555, 2.7547, 2.5574], device='cuda:7'), covar=tensor([0.0851, 0.0314, 0.0210, 0.1225, 0.0118, 0.0201, 0.0451, 0.0523], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0111, 0.0103, 0.0139, 0.0087, 0.0131, 0.0131, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 13:38:26,579 INFO [train.py:904] (7/8) Epoch 28, batch 4900, loss[loss=0.1801, simple_loss=0.2678, pruned_loss=0.04616, over 16609.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2696, pruned_loss=0.04429, over 3180655.17 frames. ], batch size: 62, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:38:48,862 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1002, 5.1489, 4.9718, 4.6012, 4.6044, 5.0910, 4.8591, 4.7553], device='cuda:7'), covar=tensor([0.0608, 0.0511, 0.0329, 0.0315, 0.1006, 0.0516, 0.0357, 0.0585], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0469, 0.0366, 0.0368, 0.0362, 0.0420, 0.0249, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:39:00,728 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 1.895e+02 2.133e+02 2.360e+02 3.975e+02, threshold=4.266e+02, percent-clipped=0.0 2023-05-02 13:39:04,606 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7474, 4.8289, 5.1618, 5.1284, 5.1368, 4.8319, 4.7776, 4.6573], device='cuda:7'), covar=tensor([0.0323, 0.0501, 0.0324, 0.0361, 0.0433, 0.0367, 0.0926, 0.0469], device='cuda:7'), in_proj_covar=tensor([0.0434, 0.0490, 0.0472, 0.0436, 0.0519, 0.0497, 0.0572, 0.0401], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 13:39:04,636 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=278979.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:39:39,809 INFO [train.py:904] (7/8) Epoch 28, batch 4950, loss[loss=0.1741, simple_loss=0.2726, pruned_loss=0.03783, over 16497.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2686, pruned_loss=0.04316, over 3192439.82 frames. ], batch size: 68, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:40:14,417 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:40:19,498 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=279030.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:40:51,498 INFO [train.py:904] (7/8) Epoch 28, batch 5000, loss[loss=0.1855, simple_loss=0.2764, pruned_loss=0.04736, over 12000.00 frames. ], tot_loss[loss=0.1783, simple_loss=0.2702, pruned_loss=0.04318, over 3199005.92 frames. ], batch size: 247, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:41:21,646 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 13:41:26,571 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.098e+02 2.415e+02 2.843e+02 4.357e+02, threshold=4.831e+02, percent-clipped=1.0 2023-05-02 13:41:28,686 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279078.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:42:04,202 INFO [train.py:904] (7/8) Epoch 28, batch 5050, loss[loss=0.1833, simple_loss=0.275, pruned_loss=0.0458, over 17123.00 frames. ], tot_loss[loss=0.1789, simple_loss=0.2709, pruned_loss=0.04342, over 3201452.43 frames. ], batch size: 47, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:42:18,013 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0153, 3.2925, 3.4937, 1.9829, 2.8790, 2.2313, 3.4403, 3.5406], device='cuda:7'), covar=tensor([0.0264, 0.0856, 0.0586, 0.2168, 0.0919, 0.1011, 0.0675, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0172, 0.0171, 0.0157, 0.0148, 0.0132, 0.0146, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 13:42:45,784 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=279131.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:43:17,185 INFO [train.py:904] (7/8) Epoch 28, batch 5100, loss[loss=0.1755, simple_loss=0.2622, pruned_loss=0.04444, over 16833.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2694, pruned_loss=0.04277, over 3199520.60 frames. ], batch size: 109, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:43:37,512 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=279167.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:43:52,645 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 1.882e+02 2.136e+02 2.468e+02 5.149e+02, threshold=4.272e+02, percent-clipped=1.0 2023-05-02 13:44:15,274 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=279192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:44:30,685 INFO [train.py:904] (7/8) Epoch 28, batch 5150, loss[loss=0.1825, simple_loss=0.2851, pruned_loss=0.03998, over 15423.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2702, pruned_loss=0.04242, over 3182275.33 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:44:48,014 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8417, 3.7630, 3.8883, 4.0100, 4.0951, 3.6703, 4.0329, 4.1275], device='cuda:7'), covar=tensor([0.1624, 0.1039, 0.1328, 0.0689, 0.0545, 0.1899, 0.0850, 0.0707], device='cuda:7'), in_proj_covar=tensor([0.0677, 0.0824, 0.0956, 0.0843, 0.0639, 0.0669, 0.0700, 0.0811], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:44:49,081 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279215.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:45:42,236 INFO [train.py:904] (7/8) Epoch 28, batch 5200, loss[loss=0.1728, simple_loss=0.2632, pruned_loss=0.04115, over 16789.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2686, pruned_loss=0.04174, over 3185051.01 frames. ], batch size: 124, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:45:52,249 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.49 vs. limit=5.0 2023-05-02 13:46:17,350 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.964e+02 2.228e+02 2.622e+02 8.480e+02, threshold=4.456e+02, percent-clipped=3.0 2023-05-02 13:46:53,559 INFO [train.py:904] (7/8) Epoch 28, batch 5250, loss[loss=0.1527, simple_loss=0.2471, pruned_loss=0.02917, over 16739.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2653, pruned_loss=0.04107, over 3201849.15 frames. ], batch size: 89, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:47:03,657 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9738, 2.8998, 2.4836, 2.7275, 3.2947, 2.9074, 3.5026, 3.5063], device='cuda:7'), covar=tensor([0.0098, 0.0502, 0.0617, 0.0474, 0.0306, 0.0445, 0.0285, 0.0320], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0246, 0.0234, 0.0234, 0.0246, 0.0244, 0.0243, 0.0245], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:47:43,287 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.58 vs. limit=2.0 2023-05-02 13:48:06,872 INFO [train.py:904] (7/8) Epoch 28, batch 5300, loss[loss=0.1436, simple_loss=0.232, pruned_loss=0.02763, over 16681.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2618, pruned_loss=0.04014, over 3198988.83 frames. ], batch size: 89, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:48:41,222 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 1.963e+02 2.333e+02 2.759e+02 6.747e+02, threshold=4.665e+02, percent-clipped=2.0 2023-05-02 13:49:20,984 INFO [train.py:904] (7/8) Epoch 28, batch 5350, loss[loss=0.1646, simple_loss=0.2572, pruned_loss=0.03597, over 16222.00 frames. ], tot_loss[loss=0.1699, simple_loss=0.2607, pruned_loss=0.03958, over 3205993.13 frames. ], batch size: 165, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:50:32,515 INFO [train.py:904] (7/8) Epoch 28, batch 5400, loss[loss=0.2033, simple_loss=0.2904, pruned_loss=0.05812, over 12274.00 frames. ], tot_loss[loss=0.172, simple_loss=0.2633, pruned_loss=0.04032, over 3181428.12 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:51:08,323 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.450e+02 1.928e+02 2.246e+02 2.506e+02 7.644e+02, threshold=4.493e+02, percent-clipped=1.0 2023-05-02 13:51:19,582 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9106, 4.8851, 4.7258, 3.9710, 4.8320, 1.6785, 4.5744, 4.4326], device='cuda:7'), covar=tensor([0.0088, 0.0101, 0.0176, 0.0467, 0.0100, 0.3163, 0.0120, 0.0289], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0174, 0.0212, 0.0186, 0.0189, 0.0217, 0.0201, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 13:51:24,277 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=279487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:51:50,353 INFO [train.py:904] (7/8) Epoch 28, batch 5450, loss[loss=0.2394, simple_loss=0.3281, pruned_loss=0.0753, over 15460.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2661, pruned_loss=0.04151, over 3185344.76 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:52:19,184 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6277, 2.4741, 2.3112, 3.8092, 2.4375, 3.7494, 1.4960, 2.7353], device='cuda:7'), covar=tensor([0.1442, 0.0880, 0.1381, 0.0245, 0.0258, 0.0445, 0.1782, 0.0876], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0202, 0.0204, 0.0208, 0.0219, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 13:52:32,324 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=279529.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:52:53,242 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=279542.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:53:09,374 INFO [train.py:904] (7/8) Epoch 28, batch 5500, loss[loss=0.1901, simple_loss=0.2855, pruned_loss=0.04735, over 16869.00 frames. ], tot_loss[loss=0.1823, simple_loss=0.2731, pruned_loss=0.04573, over 3149092.43 frames. ], batch size: 96, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:53:47,385 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.670e+02 2.730e+02 3.219e+02 3.961e+02 9.894e+02, threshold=6.439e+02, percent-clipped=12.0 2023-05-02 13:54:02,897 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8468, 2.9383, 2.6229, 4.8241, 3.5474, 4.1838, 1.6619, 3.1458], device='cuda:7'), covar=tensor([0.1347, 0.0759, 0.1246, 0.0164, 0.0325, 0.0415, 0.1728, 0.0812], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0204, 0.0208, 0.0219, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 13:54:09,453 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=279590.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:54:24,898 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7511, 2.5080, 2.3364, 3.8789, 2.5658, 3.8482, 1.4867, 2.8378], device='cuda:7'), covar=tensor([0.1451, 0.0932, 0.1411, 0.0219, 0.0258, 0.0432, 0.1934, 0.0883], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0202, 0.0205, 0.0208, 0.0219, 0.0212, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 13:54:28,780 INFO [train.py:904] (7/8) Epoch 28, batch 5550, loss[loss=0.207, simple_loss=0.2851, pruned_loss=0.06442, over 16659.00 frames. ], tot_loss[loss=0.1895, simple_loss=0.2795, pruned_loss=0.04973, over 3152626.52 frames. ], batch size: 57, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:54:29,434 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=279603.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:54:32,640 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3716, 2.9286, 2.7160, 2.3549, 2.3519, 2.3719, 2.9363, 2.9276], device='cuda:7'), covar=tensor([0.2212, 0.0677, 0.1423, 0.2364, 0.2062, 0.1942, 0.0520, 0.1193], device='cuda:7'), in_proj_covar=tensor([0.0335, 0.0275, 0.0312, 0.0327, 0.0305, 0.0276, 0.0304, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 13:54:54,805 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8837, 2.6682, 2.5402, 1.9009, 2.5358, 2.6592, 2.5149, 1.9473], device='cuda:7'), covar=tensor([0.0452, 0.0111, 0.0106, 0.0393, 0.0160, 0.0147, 0.0141, 0.0414], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0089, 0.0090, 0.0134, 0.0102, 0.0114, 0.0098, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 13:55:27,127 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3485, 3.2008, 3.6276, 1.8522, 3.7674, 3.7775, 2.8465, 2.8028], device='cuda:7'), covar=tensor([0.0879, 0.0313, 0.0217, 0.1174, 0.0087, 0.0185, 0.0496, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0147, 0.0110, 0.0101, 0.0137, 0.0086, 0.0129, 0.0128, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 13:55:48,655 INFO [train.py:904] (7/8) Epoch 28, batch 5600, loss[loss=0.1923, simple_loss=0.2792, pruned_loss=0.05273, over 16498.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2841, pruned_loss=0.0536, over 3137371.00 frames. ], batch size: 68, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:55:59,451 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.72 vs. limit=2.0 2023-05-02 13:56:28,895 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.204e+02 3.162e+02 3.859e+02 4.792e+02 1.536e+03, threshold=7.717e+02, percent-clipped=11.0 2023-05-02 13:57:11,937 INFO [train.py:904] (7/8) Epoch 28, batch 5650, loss[loss=0.2029, simple_loss=0.2897, pruned_loss=0.05804, over 16706.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2881, pruned_loss=0.05615, over 3142601.59 frames. ], batch size: 134, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:57:15,028 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2023-05-02 13:58:29,312 INFO [train.py:904] (7/8) Epoch 28, batch 5700, loss[loss=0.241, simple_loss=0.3082, pruned_loss=0.08692, over 11227.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2896, pruned_loss=0.05791, over 3128221.26 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 13:59:05,448 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.265e+02 3.251e+02 3.778e+02 4.612e+02 7.364e+02, threshold=7.556e+02, percent-clipped=0.0 2023-05-02 13:59:22,622 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=279787.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 13:59:24,497 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-05-02 13:59:46,956 INFO [train.py:904] (7/8) Epoch 28, batch 5750, loss[loss=0.1912, simple_loss=0.28, pruned_loss=0.05125, over 17071.00 frames. ], tot_loss[loss=0.2056, simple_loss=0.2922, pruned_loss=0.05952, over 3103730.21 frames. ], batch size: 55, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:00:39,833 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=279835.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:01:00,341 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 14:01:07,315 INFO [train.py:904] (7/8) Epoch 28, batch 5800, loss[loss=0.2181, simple_loss=0.2915, pruned_loss=0.07237, over 11781.00 frames. ], tot_loss[loss=0.204, simple_loss=0.2917, pruned_loss=0.05814, over 3114583.38 frames. ], batch size: 248, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:01:42,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6845, 4.3624, 4.3084, 2.8793, 3.8113, 4.3494, 3.7621, 2.5897], device='cuda:7'), covar=tensor([0.0529, 0.0058, 0.0062, 0.0423, 0.0120, 0.0112, 0.0108, 0.0447], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0089, 0.0090, 0.0134, 0.0102, 0.0115, 0.0098, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 14:01:46,121 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.097e+02 2.839e+02 3.367e+02 4.183e+02 8.215e+02, threshold=6.734e+02, percent-clipped=2.0 2023-05-02 14:01:57,436 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=279885.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:02:16,990 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=279898.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:02:25,216 INFO [train.py:904] (7/8) Epoch 28, batch 5850, loss[loss=0.1988, simple_loss=0.2952, pruned_loss=0.05117, over 16818.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2897, pruned_loss=0.05679, over 3116712.71 frames. ], batch size: 102, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:02:51,855 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 14:03:44,224 INFO [train.py:904] (7/8) Epoch 28, batch 5900, loss[loss=0.1966, simple_loss=0.2855, pruned_loss=0.05384, over 17126.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2896, pruned_loss=0.05736, over 3099630.67 frames. ], batch size: 48, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:04:26,172 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.117e+02 2.561e+02 3.002e+02 3.588e+02 6.021e+02, threshold=6.005e+02, percent-clipped=0.0 2023-05-02 14:05:07,656 INFO [train.py:904] (7/8) Epoch 28, batch 5950, loss[loss=0.2028, simple_loss=0.2892, pruned_loss=0.05826, over 15247.00 frames. ], tot_loss[loss=0.2016, simple_loss=0.2904, pruned_loss=0.05643, over 3094060.92 frames. ], batch size: 190, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:06:20,938 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280051.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:06:23,846 INFO [train.py:904] (7/8) Epoch 28, batch 6000, loss[loss=0.2058, simple_loss=0.296, pruned_loss=0.05782, over 16922.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2897, pruned_loss=0.05624, over 3109544.77 frames. ], batch size: 109, lr: 2.40e-03, grad_scale: 8.0 2023-05-02 14:06:23,846 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 14:06:34,278 INFO [train.py:938] (7/8) Epoch 28, validation: loss=0.148, simple_loss=0.2602, pruned_loss=0.0179, over 944034.00 frames. 2023-05-02 14:06:34,278 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 14:07:11,172 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.966e+02 2.746e+02 3.426e+02 4.098e+02 7.990e+02, threshold=6.852e+02, percent-clipped=5.0 2023-05-02 14:07:25,929 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280088.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:07:51,285 INFO [train.py:904] (7/8) Epoch 28, batch 6050, loss[loss=0.2026, simple_loss=0.2971, pruned_loss=0.05402, over 16924.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2888, pruned_loss=0.05611, over 3100275.10 frames. ], batch size: 109, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:08:06,534 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280112.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:08:55,238 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 14:09:02,701 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280149.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:09:08,315 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1812, 1.5658, 1.9694, 2.1559, 2.2409, 2.4038, 1.7790, 2.3171], device='cuda:7'), covar=tensor([0.0290, 0.0568, 0.0302, 0.0400, 0.0367, 0.0243, 0.0572, 0.0161], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0200, 0.0188, 0.0194, 0.0210, 0.0167, 0.0204, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:09:08,880 INFO [train.py:904] (7/8) Epoch 28, batch 6100, loss[loss=0.2033, simple_loss=0.2925, pruned_loss=0.05701, over 15254.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2881, pruned_loss=0.05474, over 3115980.74 frames. ], batch size: 190, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:09:51,399 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.727e+02 2.432e+02 2.900e+02 3.731e+02 6.785e+02, threshold=5.800e+02, percent-clipped=0.0 2023-05-02 14:10:01,957 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280185.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:10:12,385 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-02 14:10:22,462 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:10:29,430 INFO [train.py:904] (7/8) Epoch 28, batch 6150, loss[loss=0.2075, simple_loss=0.2901, pruned_loss=0.06245, over 15387.00 frames. ], tot_loss[loss=0.1981, simple_loss=0.2867, pruned_loss=0.05479, over 3093873.19 frames. ], batch size: 190, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:10:42,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2469, 4.1337, 4.2921, 4.4293, 4.5453, 4.1609, 4.4982, 4.5737], device='cuda:7'), covar=tensor([0.1750, 0.1134, 0.1447, 0.0722, 0.0595, 0.1207, 0.0827, 0.0698], device='cuda:7'), in_proj_covar=tensor([0.0670, 0.0818, 0.0949, 0.0834, 0.0638, 0.0665, 0.0697, 0.0805], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:11:17,169 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280233.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:11:36,302 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280246.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:11:45,502 INFO [train.py:904] (7/8) Epoch 28, batch 6200, loss[loss=0.1715, simple_loss=0.2669, pruned_loss=0.03802, over 16764.00 frames. ], tot_loss[loss=0.1957, simple_loss=0.2841, pruned_loss=0.05364, over 3100330.02 frames. ], batch size: 83, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:11:50,185 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0378, 3.0643, 1.9587, 3.2450, 2.4163, 3.3293, 2.1215, 2.6083], device='cuda:7'), covar=tensor([0.0328, 0.0404, 0.1633, 0.0244, 0.0841, 0.0624, 0.1510, 0.0745], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0194, 0.0171, 0.0179, 0.0218, 0.0202, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:12:24,030 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.646e+02 3.232e+02 3.925e+02 8.814e+02, threshold=6.464e+02, percent-clipped=6.0 2023-05-02 14:12:56,060 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-02 14:13:00,721 INFO [train.py:904] (7/8) Epoch 28, batch 6250, loss[loss=0.2035, simple_loss=0.2894, pruned_loss=0.05883, over 16906.00 frames. ], tot_loss[loss=0.1951, simple_loss=0.2835, pruned_loss=0.05338, over 3099736.11 frames. ], batch size: 109, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:13:36,610 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:14:14,440 INFO [train.py:904] (7/8) Epoch 28, batch 6300, loss[loss=0.1801, simple_loss=0.2758, pruned_loss=0.04219, over 16623.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2837, pruned_loss=0.05315, over 3094169.51 frames. ], batch size: 76, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:14:26,254 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-05-02 14:14:34,449 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1246, 5.1469, 5.0012, 4.6295, 4.6326, 5.0799, 4.8836, 4.7887], device='cuda:7'), covar=tensor([0.0626, 0.0668, 0.0316, 0.0334, 0.1067, 0.0529, 0.0404, 0.0704], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0471, 0.0364, 0.0367, 0.0362, 0.0421, 0.0249, 0.0435], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:14:53,565 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.897e+02 2.662e+02 3.040e+02 3.867e+02 7.586e+02, threshold=6.080e+02, percent-clipped=1.0 2023-05-02 14:15:09,960 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280388.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:15:31,106 INFO [train.py:904] (7/8) Epoch 28, batch 6350, loss[loss=0.1863, simple_loss=0.2765, pruned_loss=0.04804, over 16535.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2849, pruned_loss=0.05475, over 3081534.97 frames. ], batch size: 68, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:15:36,792 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280407.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:16:19,886 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 14:16:31,410 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280444.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:16:43,563 INFO [train.py:904] (7/8) Epoch 28, batch 6400, loss[loss=0.1629, simple_loss=0.2515, pruned_loss=0.03709, over 16867.00 frames. ], tot_loss[loss=0.1974, simple_loss=0.2843, pruned_loss=0.05531, over 3089905.05 frames. ], batch size: 102, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:17:19,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.958e+02 3.396e+02 3.952e+02 7.468e+02, threshold=6.793e+02, percent-clipped=6.0 2023-05-02 14:17:21,117 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 14:17:56,105 INFO [train.py:904] (7/8) Epoch 28, batch 6450, loss[loss=0.1881, simple_loss=0.2601, pruned_loss=0.05808, over 11585.00 frames. ], tot_loss[loss=0.1969, simple_loss=0.2842, pruned_loss=0.05477, over 3097768.50 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:18:19,083 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280518.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:18:42,177 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280533.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:19:03,612 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280547.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:19:10,979 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7447, 1.8298, 1.6592, 1.4776, 1.9989, 1.5998, 1.6488, 1.9266], device='cuda:7'), covar=tensor([0.0246, 0.0333, 0.0466, 0.0449, 0.0278, 0.0325, 0.0242, 0.0286], device='cuda:7'), in_proj_covar=tensor([0.0227, 0.0241, 0.0231, 0.0231, 0.0243, 0.0240, 0.0238, 0.0240], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:19:13,568 INFO [train.py:904] (7/8) Epoch 28, batch 6500, loss[loss=0.1852, simple_loss=0.2674, pruned_loss=0.05157, over 17250.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2831, pruned_loss=0.05444, over 3112631.04 frames. ], batch size: 45, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:19:49,979 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.728e+02 2.616e+02 3.245e+02 4.274e+02 6.880e+02, threshold=6.490e+02, percent-clipped=1.0 2023-05-02 14:19:52,130 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280579.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:20:04,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9135, 4.1692, 4.0259, 4.0562, 3.7481, 3.7886, 3.8466, 4.1701], device='cuda:7'), covar=tensor([0.1162, 0.0930, 0.1004, 0.0887, 0.0789, 0.1595, 0.0876, 0.1021], device='cuda:7'), in_proj_covar=tensor([0.0717, 0.0873, 0.0714, 0.0672, 0.0549, 0.0547, 0.0726, 0.0677], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:20:15,893 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280594.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:20:28,532 INFO [train.py:904] (7/8) Epoch 28, batch 6550, loss[loss=0.2212, simple_loss=0.2943, pruned_loss=0.0741, over 11590.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2858, pruned_loss=0.05507, over 3111001.90 frames. ], batch size: 246, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:20:37,423 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280608.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:20:45,862 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280614.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:21:44,263 INFO [train.py:904] (7/8) Epoch 28, batch 6600, loss[loss=0.192, simple_loss=0.2891, pruned_loss=0.04745, over 16858.00 frames. ], tot_loss[loss=0.199, simple_loss=0.288, pruned_loss=0.05499, over 3132001.80 frames. ], batch size: 116, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:21:50,833 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7417, 4.5335, 4.4480, 4.9212, 5.0758, 4.6066, 5.0458, 5.1139], device='cuda:7'), covar=tensor([0.2173, 0.1457, 0.2755, 0.1115, 0.0924, 0.1384, 0.1191, 0.1029], device='cuda:7'), in_proj_covar=tensor([0.0669, 0.0817, 0.0948, 0.0833, 0.0638, 0.0665, 0.0699, 0.0803], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:22:17,321 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.02 vs. limit=5.0 2023-05-02 14:22:18,711 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280675.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:22:21,699 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.864e+02 2.836e+02 3.315e+02 4.097e+02 1.046e+03, threshold=6.629e+02, percent-clipped=1.0 2023-05-02 14:22:29,467 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280683.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:23:00,603 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=280702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:23:01,183 INFO [train.py:904] (7/8) Epoch 28, batch 6650, loss[loss=0.1824, simple_loss=0.2751, pruned_loss=0.04482, over 16720.00 frames. ], tot_loss[loss=0.1996, simple_loss=0.288, pruned_loss=0.05563, over 3122491.68 frames. ], batch size: 76, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:23:07,306 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280707.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:02,772 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280744.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:15,398 INFO [train.py:904] (7/8) Epoch 28, batch 6700, loss[loss=0.2043, simple_loss=0.2912, pruned_loss=0.05876, over 16315.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2873, pruned_loss=0.0562, over 3111306.00 frames. ], batch size: 146, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:24:18,765 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280755.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:30,586 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=280763.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:24:51,655 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.083e+02 2.827e+02 3.469e+02 4.301e+02 6.613e+02, threshold=6.939e+02, percent-clipped=0.0 2023-05-02 14:25:13,161 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=280792.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:25:20,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2694, 3.3512, 2.0405, 3.7014, 2.5912, 3.7004, 2.1651, 2.7709], device='cuda:7'), covar=tensor([0.0352, 0.0454, 0.1837, 0.0242, 0.0871, 0.0604, 0.1771, 0.0842], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0181, 0.0196, 0.0172, 0.0180, 0.0220, 0.0204, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:25:29,259 INFO [train.py:904] (7/8) Epoch 28, batch 6750, loss[loss=0.2288, simple_loss=0.3054, pruned_loss=0.07605, over 11689.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2867, pruned_loss=0.05633, over 3118298.68 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:25:39,281 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7167, 2.6625, 2.3980, 4.4163, 3.1852, 3.9073, 1.4992, 2.8991], device='cuda:7'), covar=tensor([0.1467, 0.0889, 0.1485, 0.0187, 0.0294, 0.0460, 0.1885, 0.0901], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0201, 0.0203, 0.0208, 0.0219, 0.0210, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:26:11,266 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 14:26:41,172 INFO [train.py:904] (7/8) Epoch 28, batch 6800, loss[loss=0.2076, simple_loss=0.2867, pruned_loss=0.06423, over 11608.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2868, pruned_loss=0.05594, over 3124033.13 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:27:13,228 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280874.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:27:18,679 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.178e+02 2.733e+02 3.177e+02 3.942e+02 7.279e+02, threshold=6.355e+02, percent-clipped=1.0 2023-05-02 14:27:36,563 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280889.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:27:55,356 INFO [train.py:904] (7/8) Epoch 28, batch 6850, loss[loss=0.1832, simple_loss=0.2935, pruned_loss=0.03649, over 16820.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2885, pruned_loss=0.0571, over 3105031.44 frames. ], batch size: 102, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:27:56,256 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280903.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:28:32,825 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.47 vs. limit=5.0 2023-05-02 14:29:06,761 INFO [train.py:904] (7/8) Epoch 28, batch 6900, loss[loss=0.2506, simple_loss=0.3164, pruned_loss=0.09237, over 11561.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2902, pruned_loss=0.05632, over 3113683.43 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:29:34,344 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=280970.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:29:38,607 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-05-02 14:29:45,361 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.637e+02 2.748e+02 3.135e+02 3.737e+02 6.416e+02, threshold=6.270e+02, percent-clipped=1.0 2023-05-02 14:29:54,006 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=280983.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:30:23,411 INFO [train.py:904] (7/8) Epoch 28, batch 6950, loss[loss=0.1747, simple_loss=0.2667, pruned_loss=0.04136, over 16448.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2911, pruned_loss=0.05748, over 3103075.27 frames. ], batch size: 146, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:31:03,954 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281031.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:31:30,281 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281049.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:31:35,813 INFO [train.py:904] (7/8) Epoch 28, batch 7000, loss[loss=0.1842, simple_loss=0.29, pruned_loss=0.03914, over 16582.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2912, pruned_loss=0.05701, over 3101126.73 frames. ], batch size: 75, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:31:44,081 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281058.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:32:12,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 2.882e+02 3.440e+02 4.031e+02 6.341e+02, threshold=6.881e+02, percent-clipped=1.0 2023-05-02 14:32:50,411 INFO [train.py:904] (7/8) Epoch 28, batch 7050, loss[loss=0.197, simple_loss=0.2936, pruned_loss=0.05013, over 16690.00 frames. ], tot_loss[loss=0.2033, simple_loss=0.292, pruned_loss=0.05732, over 3086978.82 frames. ], batch size: 134, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:33:01,807 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281110.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 14:34:04,188 INFO [train.py:904] (7/8) Epoch 28, batch 7100, loss[loss=0.1732, simple_loss=0.2667, pruned_loss=0.03987, over 16854.00 frames. ], tot_loss[loss=0.2025, simple_loss=0.2907, pruned_loss=0.05713, over 3078628.20 frames. ], batch size: 96, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:34:29,470 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281169.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:34:37,867 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281174.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:34:43,394 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.809e+02 2.877e+02 3.420e+02 4.394e+02 9.887e+02, threshold=6.841e+02, percent-clipped=1.0 2023-05-02 14:35:00,257 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281189.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:35:20,728 INFO [train.py:904] (7/8) Epoch 28, batch 7150, loss[loss=0.201, simple_loss=0.293, pruned_loss=0.05447, over 16194.00 frames. ], tot_loss[loss=0.2023, simple_loss=0.2895, pruned_loss=0.05752, over 3066241.29 frames. ], batch size: 165, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:35:21,854 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281203.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:35:23,771 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 14:35:47,784 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:36:00,668 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281230.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 14:36:09,136 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281237.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:36:19,518 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8958, 3.6904, 4.2532, 2.1927, 4.4856, 4.4465, 3.2869, 3.2918], device='cuda:7'), covar=tensor([0.0771, 0.0293, 0.0209, 0.1190, 0.0070, 0.0158, 0.0387, 0.0465], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0104, 0.0139, 0.0087, 0.0131, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:36:29,668 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281251.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:36:30,192 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 14:36:31,502 INFO [train.py:904] (7/8) Epoch 28, batch 7200, loss[loss=0.1883, simple_loss=0.2816, pruned_loss=0.04747, over 16243.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2869, pruned_loss=0.05537, over 3084687.53 frames. ], batch size: 165, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:36:56,527 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281270.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:37:05,531 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 14:37:07,762 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.572e+02 3.176e+02 3.806e+02 6.231e+02, threshold=6.351e+02, percent-clipped=0.0 2023-05-02 14:37:46,532 INFO [train.py:904] (7/8) Epoch 28, batch 7250, loss[loss=0.1876, simple_loss=0.2701, pruned_loss=0.05253, over 16898.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2849, pruned_loss=0.05459, over 3085976.45 frames. ], batch size: 109, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:37:54,329 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9999, 3.3797, 3.3635, 2.1964, 3.1203, 3.3909, 3.1829, 1.9721], device='cuda:7'), covar=tensor([0.0639, 0.0070, 0.0078, 0.0499, 0.0139, 0.0144, 0.0123, 0.0537], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0102, 0.0115, 0.0098, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 14:38:09,207 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:38:59,578 INFO [train.py:904] (7/8) Epoch 28, batch 7300, loss[loss=0.192, simple_loss=0.29, pruned_loss=0.04705, over 16899.00 frames. ], tot_loss[loss=0.196, simple_loss=0.2837, pruned_loss=0.05418, over 3082504.32 frames. ], batch size: 96, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:39:08,168 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281358.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:39:20,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6792, 2.3571, 2.2335, 3.4939, 2.4601, 3.5081, 1.4067, 2.7046], device='cuda:7'), covar=tensor([0.1460, 0.0922, 0.1415, 0.0204, 0.0219, 0.0424, 0.1849, 0.0914], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0202, 0.0203, 0.0208, 0.0219, 0.0210, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:39:39,520 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.942e+02 3.043e+02 3.727e+02 4.632e+02 1.394e+03, threshold=7.454e+02, percent-clipped=7.0 2023-05-02 14:40:13,658 INFO [train.py:904] (7/8) Epoch 28, batch 7350, loss[loss=0.2083, simple_loss=0.2934, pruned_loss=0.06167, over 16219.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2857, pruned_loss=0.05544, over 3070716.97 frames. ], batch size: 165, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:40:16,315 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281405.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 14:40:17,486 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281406.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:40:50,139 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281426.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:41:28,058 INFO [train.py:904] (7/8) Epoch 28, batch 7400, loss[loss=0.1914, simple_loss=0.2804, pruned_loss=0.05116, over 16625.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.287, pruned_loss=0.05621, over 3059098.02 frames. ], batch size: 57, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:41:33,710 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281456.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:42:08,141 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.701e+02 3.025e+02 3.387e+02 6.725e+02, threshold=6.049e+02, percent-clipped=0.0 2023-05-02 14:42:18,980 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281487.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:42:42,682 INFO [train.py:904] (7/8) Epoch 28, batch 7450, loss[loss=0.1787, simple_loss=0.2688, pruned_loss=0.04428, over 16792.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2872, pruned_loss=0.05618, over 3091168.14 frames. ], batch size: 39, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:43:06,505 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281517.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:43:17,787 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281525.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 14:43:58,017 INFO [train.py:904] (7/8) Epoch 28, batch 7500, loss[loss=0.2003, simple_loss=0.2851, pruned_loss=0.05769, over 16759.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2872, pruned_loss=0.05613, over 3063532.47 frames. ], batch size: 124, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:44:36,768 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.668e+02 2.813e+02 3.506e+02 4.192e+02 1.275e+03, threshold=7.011e+02, percent-clipped=6.0 2023-05-02 14:45:11,679 INFO [train.py:904] (7/8) Epoch 28, batch 7550, loss[loss=0.1772, simple_loss=0.2726, pruned_loss=0.04086, over 16706.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2861, pruned_loss=0.0558, over 3061022.80 frames. ], batch size: 89, lr: 2.39e-03, grad_scale: 2.0 2023-05-02 14:46:25,957 INFO [train.py:904] (7/8) Epoch 28, batch 7600, loss[loss=0.1882, simple_loss=0.278, pruned_loss=0.04917, over 16405.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.286, pruned_loss=0.05645, over 3049991.20 frames. ], batch size: 75, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:47:04,824 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 2.608e+02 3.017e+02 3.545e+02 6.203e+02, threshold=6.033e+02, percent-clipped=0.0 2023-05-02 14:47:40,204 INFO [train.py:904] (7/8) Epoch 28, batch 7650, loss[loss=0.1992, simple_loss=0.2922, pruned_loss=0.05305, over 16787.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2865, pruned_loss=0.05677, over 3051218.81 frames. ], batch size: 83, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:47:43,590 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281705.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 14:48:51,546 INFO [train.py:904] (7/8) Epoch 28, batch 7700, loss[loss=0.1996, simple_loss=0.2832, pruned_loss=0.05799, over 16750.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2858, pruned_loss=0.05651, over 3075015.28 frames. ], batch size: 124, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:48:52,556 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281753.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:49:03,467 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2505, 4.2457, 4.1325, 3.3072, 4.1912, 1.6323, 3.9809, 3.6766], device='cuda:7'), covar=tensor([0.0139, 0.0124, 0.0215, 0.0312, 0.0095, 0.3186, 0.0150, 0.0307], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0173, 0.0212, 0.0184, 0.0187, 0.0216, 0.0199, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:49:31,079 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.035e+02 3.039e+02 3.444e+02 4.107e+02 7.486e+02, threshold=6.887e+02, percent-clipped=3.0 2023-05-02 14:49:35,832 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:50:07,531 INFO [train.py:904] (7/8) Epoch 28, batch 7750, loss[loss=0.1948, simple_loss=0.2855, pruned_loss=0.05207, over 16569.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2857, pruned_loss=0.05645, over 3079291.85 frames. ], batch size: 68, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:50:22,073 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=281812.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:50:40,472 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=281825.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:51:03,145 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0948, 4.9094, 5.1289, 5.2873, 5.4664, 4.8546, 5.5144, 5.4757], device='cuda:7'), covar=tensor([0.2039, 0.1310, 0.1713, 0.0765, 0.0699, 0.0944, 0.0645, 0.0657], device='cuda:7'), in_proj_covar=tensor([0.0664, 0.0811, 0.0939, 0.0825, 0.0631, 0.0657, 0.0692, 0.0798], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:51:21,178 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281852.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:51:21,940 INFO [train.py:904] (7/8) Epoch 28, batch 7800, loss[loss=0.1801, simple_loss=0.2693, pruned_loss=0.04543, over 16698.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2866, pruned_loss=0.05667, over 3092878.07 frames. ], batch size: 134, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:51:44,302 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-02 14:51:52,696 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=281873.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:52:03,261 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.973e+02 2.715e+02 3.377e+02 4.149e+02 7.003e+02, threshold=6.754e+02, percent-clipped=1.0 2023-05-02 14:52:22,301 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6089, 3.5624, 2.8126, 2.2611, 2.3783, 2.4098, 3.7992, 3.2335], device='cuda:7'), covar=tensor([0.2897, 0.0708, 0.1895, 0.3004, 0.2560, 0.2308, 0.0489, 0.1383], device='cuda:7'), in_proj_covar=tensor([0.0337, 0.0273, 0.0314, 0.0328, 0.0306, 0.0278, 0.0305, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 14:52:37,336 INFO [train.py:904] (7/8) Epoch 28, batch 7850, loss[loss=0.2149, simple_loss=0.3087, pruned_loss=0.06049, over 16908.00 frames. ], tot_loss[loss=0.2013, simple_loss=0.2879, pruned_loss=0.05736, over 3057331.37 frames. ], batch size: 116, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:52:47,404 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281910.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 14:52:51,832 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281913.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:53:38,078 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0111, 3.0336, 1.9743, 3.2281, 2.3690, 3.3173, 2.2611, 2.5830], device='cuda:7'), covar=tensor([0.0342, 0.0454, 0.1703, 0.0300, 0.0878, 0.0659, 0.1447, 0.0837], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0182, 0.0197, 0.0173, 0.0181, 0.0221, 0.0205, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:53:38,082 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=281945.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:53:48,943 INFO [train.py:904] (7/8) Epoch 28, batch 7900, loss[loss=0.1968, simple_loss=0.2867, pruned_loss=0.05349, over 16704.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2876, pruned_loss=0.0576, over 3027954.61 frames. ], batch size: 124, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:54:17,010 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=281971.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 14:54:29,911 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.540e+02 2.510e+02 2.970e+02 3.651e+02 5.631e+02, threshold=5.940e+02, percent-clipped=0.0 2023-05-02 14:55:09,798 INFO [train.py:904] (7/8) Epoch 28, batch 7950, loss[loss=0.1967, simple_loss=0.282, pruned_loss=0.0557, over 16627.00 frames. ], tot_loss[loss=0.2019, simple_loss=0.2877, pruned_loss=0.05807, over 3027337.16 frames. ], batch size: 57, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 14:55:14,866 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282006.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:56:27,152 INFO [train.py:904] (7/8) Epoch 28, batch 8000, loss[loss=0.1954, simple_loss=0.2875, pruned_loss=0.05163, over 16854.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2884, pruned_loss=0.05863, over 3036584.80 frames. ], batch size: 96, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:57:07,769 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.806e+02 3.354e+02 4.135e+02 1.069e+03, threshold=6.709e+02, percent-clipped=7.0 2023-05-02 14:57:11,224 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282082.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:57:24,964 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8611, 3.7331, 4.2777, 1.9090, 4.4323, 4.4722, 3.3906, 3.3099], device='cuda:7'), covar=tensor([0.0874, 0.0318, 0.0235, 0.1470, 0.0096, 0.0189, 0.0459, 0.0515], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0140, 0.0087, 0.0132, 0.0130, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:57:33,386 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282096.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:57:42,453 INFO [train.py:904] (7/8) Epoch 28, batch 8050, loss[loss=0.2017, simple_loss=0.2864, pruned_loss=0.05855, over 16716.00 frames. ], tot_loss[loss=0.201, simple_loss=0.2872, pruned_loss=0.05743, over 3059635.00 frames. ], batch size: 124, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:57:56,112 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282112.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:58:17,591 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0948, 2.4688, 2.3604, 2.9055, 1.8748, 3.1616, 1.8600, 2.7288], device='cuda:7'), covar=tensor([0.1132, 0.0614, 0.1086, 0.0222, 0.0116, 0.0358, 0.1482, 0.0695], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0179, 0.0200, 0.0202, 0.0207, 0.0218, 0.0210, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 14:58:22,816 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:58:57,924 INFO [train.py:904] (7/8) Epoch 28, batch 8100, loss[loss=0.2097, simple_loss=0.2938, pruned_loss=0.06275, over 16849.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2876, pruned_loss=0.05764, over 3051835.15 frames. ], batch size: 102, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 14:59:04,421 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282157.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:59:09,073 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282160.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 14:59:15,111 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2247, 2.3960, 2.3915, 3.9038, 2.2763, 2.7132, 2.4401, 2.5214], device='cuda:7'), covar=tensor([0.1463, 0.3381, 0.2965, 0.0580, 0.4146, 0.2386, 0.3504, 0.3258], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0475, 0.0385, 0.0337, 0.0447, 0.0543, 0.0448, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:59:21,098 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2205, 1.5987, 1.9712, 2.1537, 2.2666, 2.4333, 1.8696, 2.3444], device='cuda:7'), covar=tensor([0.0264, 0.0563, 0.0328, 0.0411, 0.0350, 0.0230, 0.0568, 0.0181], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0198, 0.0185, 0.0191, 0.0208, 0.0164, 0.0202, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 14:59:38,417 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.025e+02 2.655e+02 3.116e+02 4.195e+02 8.434e+02, threshold=6.231e+02, percent-clipped=4.0 2023-05-02 15:00:13,682 INFO [train.py:904] (7/8) Epoch 28, batch 8150, loss[loss=0.1763, simple_loss=0.2591, pruned_loss=0.04677, over 16532.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2863, pruned_loss=0.05773, over 3022435.38 frames. ], batch size: 75, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 15:00:15,406 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0202, 3.0823, 1.9508, 3.2460, 2.4030, 3.3330, 2.2062, 2.6196], device='cuda:7'), covar=tensor([0.0356, 0.0444, 0.1709, 0.0264, 0.0855, 0.0600, 0.1529, 0.0807], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0180, 0.0196, 0.0172, 0.0180, 0.0220, 0.0205, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 15:00:17,549 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-02 15:00:21,797 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282208.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:01:29,259 INFO [train.py:904] (7/8) Epoch 28, batch 8200, loss[loss=0.1718, simple_loss=0.2655, pruned_loss=0.03902, over 16490.00 frames. ], tot_loss[loss=0.1978, simple_loss=0.283, pruned_loss=0.05627, over 3031944.36 frames. ], batch size: 68, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 15:01:50,787 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282266.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 15:02:13,185 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.791e+02 2.646e+02 3.224e+02 3.830e+02 8.155e+02, threshold=6.449e+02, percent-clipped=2.0 2023-05-02 15:02:47,662 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282301.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:02:50,332 INFO [train.py:904] (7/8) Epoch 28, batch 8250, loss[loss=0.1617, simple_loss=0.2637, pruned_loss=0.02987, over 16837.00 frames. ], tot_loss[loss=0.1934, simple_loss=0.2811, pruned_loss=0.05287, over 3040935.95 frames. ], batch size: 96, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 15:03:39,994 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7321, 1.9927, 2.3974, 2.7172, 2.6430, 3.1713, 2.1876, 3.1407], device='cuda:7'), covar=tensor([0.0267, 0.0619, 0.0400, 0.0364, 0.0388, 0.0212, 0.0584, 0.0189], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0197, 0.0184, 0.0189, 0.0206, 0.0164, 0.0201, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:04:07,374 INFO [train.py:904] (7/8) Epoch 28, batch 8300, loss[loss=0.1966, simple_loss=0.2814, pruned_loss=0.05592, over 12275.00 frames. ], tot_loss[loss=0.1892, simple_loss=0.2781, pruned_loss=0.05015, over 3020472.02 frames. ], batch size: 248, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 15:04:24,061 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6158, 2.6429, 2.3805, 2.4262, 3.0033, 2.6659, 3.0501, 3.2154], device='cuda:7'), covar=tensor([0.0148, 0.0452, 0.0533, 0.0501, 0.0309, 0.0419, 0.0293, 0.0298], device='cuda:7'), in_proj_covar=tensor([0.0223, 0.0238, 0.0229, 0.0229, 0.0239, 0.0238, 0.0235, 0.0238], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:04:50,785 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.126e+02 2.473e+02 2.954e+02 5.391e+02, threshold=4.946e+02, percent-clipped=0.0 2023-05-02 15:05:26,438 INFO [train.py:904] (7/8) Epoch 28, batch 8350, loss[loss=0.1687, simple_loss=0.2713, pruned_loss=0.03301, over 16256.00 frames. ], tot_loss[loss=0.1874, simple_loss=0.2775, pruned_loss=0.04862, over 3001784.67 frames. ], batch size: 165, lr: 2.39e-03, grad_scale: 4.0 2023-05-02 15:05:29,608 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0177, 3.1053, 1.9579, 3.2922, 2.3271, 3.2965, 2.2120, 2.6727], device='cuda:7'), covar=tensor([0.0353, 0.0413, 0.1699, 0.0328, 0.0900, 0.0557, 0.1573, 0.0780], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0179, 0.0195, 0.0170, 0.0178, 0.0218, 0.0203, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 15:06:43,075 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282452.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:06:43,835 INFO [train.py:904] (7/8) Epoch 28, batch 8400, loss[loss=0.1785, simple_loss=0.26, pruned_loss=0.04848, over 12236.00 frames. ], tot_loss[loss=0.184, simple_loss=0.2751, pruned_loss=0.04642, over 3006315.21 frames. ], batch size: 247, lr: 2.39e-03, grad_scale: 8.0 2023-05-02 15:07:26,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.221e+02 2.637e+02 3.332e+02 6.864e+02, threshold=5.273e+02, percent-clipped=5.0 2023-05-02 15:07:29,732 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 15:07:31,290 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9828, 4.0205, 4.3191, 4.2856, 4.3010, 4.0614, 4.0406, 4.0939], device='cuda:7'), covar=tensor([0.0392, 0.0783, 0.0478, 0.0503, 0.0451, 0.0556, 0.1000, 0.0520], device='cuda:7'), in_proj_covar=tensor([0.0434, 0.0491, 0.0473, 0.0438, 0.0520, 0.0498, 0.0574, 0.0402], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:07:40,632 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282489.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:08:02,091 INFO [train.py:904] (7/8) Epoch 28, batch 8450, loss[loss=0.1883, simple_loss=0.2861, pruned_loss=0.04522, over 15239.00 frames. ], tot_loss[loss=0.1813, simple_loss=0.2734, pruned_loss=0.04458, over 3027648.29 frames. ], batch size: 190, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:08:11,406 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282508.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:09:17,963 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282550.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:09:21,860 INFO [train.py:904] (7/8) Epoch 28, batch 8500, loss[loss=0.1559, simple_loss=0.2523, pruned_loss=0.02969, over 16425.00 frames. ], tot_loss[loss=0.1768, simple_loss=0.2695, pruned_loss=0.04209, over 3033457.87 frames. ], batch size: 68, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:09:28,042 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282556.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:09:43,405 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282566.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:10:07,073 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.128e+02 2.662e+02 3.202e+02 5.740e+02, threshold=5.324e+02, percent-clipped=2.0 2023-05-02 15:10:25,217 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5473, 3.5177, 3.5107, 2.7132, 3.3841, 2.1409, 3.2450, 2.8388], device='cuda:7'), covar=tensor([0.0176, 0.0167, 0.0199, 0.0247, 0.0132, 0.2366, 0.0152, 0.0283], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0171, 0.0209, 0.0181, 0.0184, 0.0213, 0.0196, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:10:42,322 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282601.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:10:45,545 INFO [train.py:904] (7/8) Epoch 28, batch 8550, loss[loss=0.1713, simple_loss=0.2702, pruned_loss=0.03614, over 16776.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2669, pruned_loss=0.04098, over 3041581.23 frames. ], batch size: 83, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:11:06,548 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282614.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 15:12:14,507 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282649.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:12:21,569 INFO [train.py:904] (7/8) Epoch 28, batch 8600, loss[loss=0.1615, simple_loss=0.263, pruned_loss=0.03004, over 16887.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2678, pruned_loss=0.03997, over 3058456.18 frames. ], batch size: 90, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:12:32,839 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282658.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:13:12,539 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282677.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:13:15,388 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2011, 2.5121, 2.6453, 1.9781, 2.7845, 2.8455, 2.5919, 2.5068], device='cuda:7'), covar=tensor([0.0599, 0.0267, 0.0244, 0.0953, 0.0110, 0.0244, 0.0414, 0.0447], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0108, 0.0100, 0.0135, 0.0084, 0.0127, 0.0126, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:13:17,334 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.534e+02 2.220e+02 2.463e+02 3.072e+02 6.042e+02, threshold=4.927e+02, percent-clipped=1.0 2023-05-02 15:13:50,800 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282698.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 15:13:58,793 INFO [train.py:904] (7/8) Epoch 28, batch 8650, loss[loss=0.1572, simple_loss=0.2694, pruned_loss=0.02255, over 15381.00 frames. ], tot_loss[loss=0.1718, simple_loss=0.2662, pruned_loss=0.03863, over 3048261.43 frames. ], batch size: 192, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:14:28,945 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4072, 3.2357, 3.4362, 1.9450, 3.6211, 3.6613, 3.0059, 2.8257], device='cuda:7'), covar=tensor([0.0793, 0.0280, 0.0222, 0.1180, 0.0082, 0.0179, 0.0409, 0.0502], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0108, 0.0100, 0.0135, 0.0084, 0.0127, 0.0126, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:14:36,875 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282719.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:15:14,440 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282738.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:15:40,186 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=282752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:15:40,926 INFO [train.py:904] (7/8) Epoch 28, batch 8700, loss[loss=0.1726, simple_loss=0.2597, pruned_loss=0.04271, over 16958.00 frames. ], tot_loss[loss=0.1697, simple_loss=0.2638, pruned_loss=0.03781, over 3047262.64 frames. ], batch size: 109, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:15:48,513 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.10 vs. limit=2.0 2023-05-02 15:15:53,191 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282759.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 15:16:29,664 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.456e+02 2.234e+02 2.610e+02 3.054e+02 5.139e+02, threshold=5.220e+02, percent-clipped=1.0 2023-05-02 15:17:06,890 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=282800.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:17:12,017 INFO [train.py:904] (7/8) Epoch 28, batch 8750, loss[loss=0.179, simple_loss=0.2835, pruned_loss=0.03726, over 16444.00 frames. ], tot_loss[loss=0.1693, simple_loss=0.2638, pruned_loss=0.03742, over 3051553.43 frames. ], batch size: 146, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:17:18,913 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=282805.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:18:47,227 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=282845.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:19:02,841 INFO [train.py:904] (7/8) Epoch 28, batch 8800, loss[loss=0.1633, simple_loss=0.2597, pruned_loss=0.0335, over 16673.00 frames. ], tot_loss[loss=0.1672, simple_loss=0.2618, pruned_loss=0.03629, over 3037787.22 frames. ], batch size: 76, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:19:29,243 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=282866.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:20:00,725 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.123e+02 2.465e+02 3.098e+02 6.856e+02, threshold=4.929e+02, percent-clipped=4.0 2023-05-02 15:20:47,913 INFO [train.py:904] (7/8) Epoch 28, batch 8850, loss[loss=0.1679, simple_loss=0.271, pruned_loss=0.03235, over 16771.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2642, pruned_loss=0.0358, over 3033061.92 frames. ], batch size: 124, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:20:55,421 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4561, 2.0442, 1.7801, 1.7424, 2.2680, 1.8736, 1.8018, 2.3246], device='cuda:7'), covar=tensor([0.0209, 0.0475, 0.0597, 0.0518, 0.0346, 0.0422, 0.0200, 0.0331], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0235, 0.0226, 0.0227, 0.0237, 0.0235, 0.0232, 0.0234], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:22:15,653 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1387, 5.1258, 4.8419, 4.3280, 4.9856, 1.9738, 4.7310, 4.6262], device='cuda:7'), covar=tensor([0.0078, 0.0079, 0.0224, 0.0305, 0.0076, 0.2646, 0.0120, 0.0247], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0169, 0.0207, 0.0180, 0.0183, 0.0213, 0.0196, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:22:35,079 INFO [train.py:904] (7/8) Epoch 28, batch 8900, loss[loss=0.1528, simple_loss=0.2419, pruned_loss=0.03182, over 12586.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2647, pruned_loss=0.03519, over 3038065.82 frames. ], batch size: 246, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:22:57,453 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0338, 3.7293, 4.0091, 2.0944, 4.2785, 4.3504, 3.4166, 3.3233], device='cuda:7'), covar=tensor([0.0623, 0.0274, 0.0267, 0.1223, 0.0074, 0.0146, 0.0363, 0.0430], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0107, 0.0098, 0.0134, 0.0083, 0.0126, 0.0125, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:23:38,032 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.086e+02 2.432e+02 2.867e+02 4.868e+02, threshold=4.864e+02, percent-clipped=0.0 2023-05-02 15:24:39,877 INFO [train.py:904] (7/8) Epoch 28, batch 8950, loss[loss=0.1575, simple_loss=0.2497, pruned_loss=0.03258, over 13001.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2644, pruned_loss=0.03548, over 3047915.49 frames. ], batch size: 248, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:25:04,719 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283014.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:25:45,376 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283033.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:26:27,369 INFO [train.py:904] (7/8) Epoch 28, batch 9000, loss[loss=0.1364, simple_loss=0.2321, pruned_loss=0.02036, over 16888.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2605, pruned_loss=0.03403, over 3054940.18 frames. ], batch size: 83, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:26:27,370 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 15:26:38,045 INFO [train.py:938] (7/8) Epoch 28, validation: loss=0.1436, simple_loss=0.2472, pruned_loss=0.02006, over 944034.00 frames. 2023-05-02 15:26:38,046 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 15:26:41,529 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283054.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 15:27:10,366 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5263, 3.3269, 3.6726, 1.9274, 3.8629, 3.9095, 2.9966, 2.9230], device='cuda:7'), covar=tensor([0.0745, 0.0281, 0.0224, 0.1213, 0.0075, 0.0167, 0.0449, 0.0474], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0107, 0.0098, 0.0135, 0.0083, 0.0126, 0.0125, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:27:36,921 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.008e+02 2.381e+02 2.804e+02 4.985e+02, threshold=4.761e+02, percent-clipped=1.0 2023-05-02 15:28:21,096 INFO [train.py:904] (7/8) Epoch 28, batch 9050, loss[loss=0.1643, simple_loss=0.2559, pruned_loss=0.03639, over 16227.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2616, pruned_loss=0.03442, over 3073797.38 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:28:48,247 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283115.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:29:11,109 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8408, 3.8668, 3.9739, 3.7765, 3.9632, 4.3490, 4.0251, 3.6874], device='cuda:7'), covar=tensor([0.2090, 0.2219, 0.2378, 0.2346, 0.2620, 0.1694, 0.1497, 0.2520], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0611, 0.0674, 0.0495, 0.0664, 0.0702, 0.0527, 0.0666], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 15:29:46,514 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283144.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:29:48,366 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283145.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:30:04,163 INFO [train.py:904] (7/8) Epoch 28, batch 9100, loss[loss=0.1722, simple_loss=0.2685, pruned_loss=0.03792, over 16942.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2618, pruned_loss=0.03514, over 3087018.95 frames. ], batch size: 116, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:30:13,014 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.96 vs. limit=5.0 2023-05-02 15:30:20,778 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283161.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:30:58,179 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283176.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:31:08,549 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.173e+02 2.551e+02 2.852e+02 5.102e+02, threshold=5.103e+02, percent-clipped=2.0 2023-05-02 15:31:21,528 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8471, 3.9028, 4.0063, 3.7201, 3.9463, 4.3250, 3.9701, 3.6155], device='cuda:7'), covar=tensor([0.2136, 0.2315, 0.2365, 0.2537, 0.2597, 0.1681, 0.1494, 0.2664], device='cuda:7'), in_proj_covar=tensor([0.0414, 0.0614, 0.0677, 0.0498, 0.0667, 0.0706, 0.0529, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 15:31:30,541 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3323, 3.4789, 2.1163, 3.8385, 2.5134, 3.7545, 2.1844, 2.8098], device='cuda:7'), covar=tensor([0.0341, 0.0380, 0.1708, 0.0210, 0.0912, 0.0534, 0.1759, 0.0767], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0175, 0.0192, 0.0166, 0.0175, 0.0213, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 15:31:39,558 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283193.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:31:49,961 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4014, 4.6797, 4.5185, 4.5341, 4.2476, 4.2822, 4.1265, 4.7467], device='cuda:7'), covar=tensor([0.1228, 0.0964, 0.0997, 0.0834, 0.0823, 0.1371, 0.1326, 0.0927], device='cuda:7'), in_proj_covar=tensor([0.0699, 0.0846, 0.0693, 0.0654, 0.0534, 0.0536, 0.0703, 0.0658], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:31:59,124 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3721, 3.4725, 2.0859, 3.8331, 2.5415, 3.7744, 2.1558, 2.8400], device='cuda:7'), covar=tensor([0.0337, 0.0380, 0.1758, 0.0225, 0.0921, 0.0531, 0.1784, 0.0796], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0175, 0.0192, 0.0166, 0.0175, 0.0213, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 15:32:01,018 INFO [train.py:904] (7/8) Epoch 28, batch 9150, loss[loss=0.1587, simple_loss=0.2551, pruned_loss=0.03121, over 15394.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2622, pruned_loss=0.03504, over 3088360.74 frames. ], batch size: 191, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:32:07,132 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283205.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:33:44,282 INFO [train.py:904] (7/8) Epoch 28, batch 9200, loss[loss=0.1738, simple_loss=0.2719, pruned_loss=0.03782, over 15237.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2574, pruned_loss=0.03397, over 3084759.04 frames. ], batch size: 190, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:34:05,294 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8513, 2.7515, 2.5320, 1.8886, 2.4978, 2.7678, 2.6217, 1.9287], device='cuda:7'), covar=tensor([0.0416, 0.0092, 0.0087, 0.0399, 0.0169, 0.0128, 0.0118, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0134, 0.0086, 0.0088, 0.0131, 0.0099, 0.0111, 0.0095, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 15:34:34,271 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.118e+02 2.676e+02 3.077e+02 8.618e+02, threshold=5.351e+02, percent-clipped=1.0 2023-05-02 15:34:39,387 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283283.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:35:20,280 INFO [train.py:904] (7/8) Epoch 28, batch 9250, loss[loss=0.1641, simple_loss=0.2568, pruned_loss=0.03568, over 16780.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2575, pruned_loss=0.03421, over 3083824.08 frames. ], batch size: 124, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:35:44,207 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283314.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:36:28,967 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283333.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:36:55,359 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283344.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:36:55,454 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7167, 3.0875, 3.4310, 2.0197, 2.9442, 2.1463, 3.1982, 3.2574], device='cuda:7'), covar=tensor([0.0344, 0.0980, 0.0490, 0.2252, 0.0822, 0.1141, 0.0752, 0.1110], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0166, 0.0167, 0.0155, 0.0145, 0.0130, 0.0143, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 15:37:14,589 INFO [train.py:904] (7/8) Epoch 28, batch 9300, loss[loss=0.1547, simple_loss=0.2488, pruned_loss=0.03037, over 16730.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2551, pruned_loss=0.03318, over 3079569.76 frames. ], batch size: 134, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:37:16,934 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283354.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 15:37:34,866 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283362.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:38:16,511 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 1.947e+02 2.341e+02 2.771e+02 4.646e+02, threshold=4.683e+02, percent-clipped=0.0 2023-05-02 15:38:18,081 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283381.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:38:20,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9901, 4.2608, 4.1145, 4.1209, 3.8087, 3.8549, 3.8864, 4.2705], device='cuda:7'), covar=tensor([0.1162, 0.0956, 0.1000, 0.0806, 0.0793, 0.1939, 0.0960, 0.0995], device='cuda:7'), in_proj_covar=tensor([0.0694, 0.0840, 0.0688, 0.0650, 0.0531, 0.0534, 0.0698, 0.0654], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:38:53,382 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8338, 1.4470, 1.7105, 1.7800, 1.8813, 1.9302, 1.7610, 1.9445], device='cuda:7'), covar=tensor([0.0288, 0.0495, 0.0265, 0.0371, 0.0370, 0.0268, 0.0461, 0.0189], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0195, 0.0183, 0.0187, 0.0205, 0.0162, 0.0199, 0.0161], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:38:58,957 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283402.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:38:59,602 INFO [train.py:904] (7/8) Epoch 28, batch 9350, loss[loss=0.1852, simple_loss=0.2725, pruned_loss=0.04894, over 16872.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2553, pruned_loss=0.03305, over 3093968.53 frames. ], batch size: 124, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:40:41,048 INFO [train.py:904] (7/8) Epoch 28, batch 9400, loss[loss=0.1376, simple_loss=0.2264, pruned_loss=0.0244, over 12293.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2554, pruned_loss=0.03301, over 3070724.11 frames. ], batch size: 248, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:40:58,121 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4313, 2.7653, 3.2340, 2.0087, 2.8267, 2.1117, 3.0740, 2.9730], device='cuda:7'), covar=tensor([0.0301, 0.1057, 0.0518, 0.2221, 0.0840, 0.1112, 0.0615, 0.1062], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0166, 0.0154, 0.0144, 0.0130, 0.0142, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 15:41:00,060 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283461.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:41:19,452 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283471.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:41:39,547 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.116e+02 2.396e+02 2.984e+02 4.135e+02, threshold=4.791e+02, percent-clipped=0.0 2023-05-02 15:42:20,565 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283500.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:42:24,710 INFO [train.py:904] (7/8) Epoch 28, batch 9450, loss[loss=0.1666, simple_loss=0.2622, pruned_loss=0.03554, over 16893.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.257, pruned_loss=0.03343, over 3052956.87 frames. ], batch size: 116, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:42:36,845 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283509.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:43:49,071 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283544.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:44:06,793 INFO [train.py:904] (7/8) Epoch 28, batch 9500, loss[loss=0.1502, simple_loss=0.2499, pruned_loss=0.02525, over 15369.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2563, pruned_loss=0.03316, over 3050303.15 frames. ], batch size: 191, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:45:03,874 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.031e+02 2.419e+02 2.973e+02 5.266e+02, threshold=4.838e+02, percent-clipped=2.0 2023-05-02 15:45:13,246 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 15:45:53,236 INFO [train.py:904] (7/8) Epoch 28, batch 9550, loss[loss=0.1695, simple_loss=0.276, pruned_loss=0.03155, over 16406.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2565, pruned_loss=0.03335, over 3069043.88 frames. ], batch size: 146, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:45:59,200 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283605.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 15:47:09,831 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283639.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:47:34,681 INFO [train.py:904] (7/8) Epoch 28, batch 9600, loss[loss=0.1662, simple_loss=0.2713, pruned_loss=0.03054, over 16211.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2577, pruned_loss=0.03361, over 3070919.07 frames. ], batch size: 166, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:47:53,064 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 15:48:29,443 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.192e+02 2.621e+02 3.097e+02 5.622e+02, threshold=5.243e+02, percent-clipped=5.0 2023-05-02 15:49:23,011 INFO [train.py:904] (7/8) Epoch 28, batch 9650, loss[loss=0.1736, simple_loss=0.2694, pruned_loss=0.03889, over 16272.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2593, pruned_loss=0.03371, over 3067451.43 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:50:39,458 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.03 vs. limit=5.0 2023-05-02 15:51:10,214 INFO [train.py:904] (7/8) Epoch 28, batch 9700, loss[loss=0.1893, simple_loss=0.284, pruned_loss=0.04727, over 16748.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2585, pruned_loss=0.03357, over 3069829.55 frames. ], batch size: 124, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:51:11,174 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2886, 3.0452, 3.1081, 1.9123, 3.2787, 3.3539, 2.8617, 2.7140], device='cuda:7'), covar=tensor([0.0776, 0.0273, 0.0256, 0.1157, 0.0107, 0.0210, 0.0450, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0143, 0.0107, 0.0098, 0.0134, 0.0083, 0.0126, 0.0125, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:51:47,402 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283771.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:52:08,991 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.180e+02 2.423e+02 2.886e+02 5.203e+02, threshold=4.846e+02, percent-clipped=0.0 2023-05-02 15:52:48,109 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283800.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:52:52,817 INFO [train.py:904] (7/8) Epoch 28, batch 9750, loss[loss=0.158, simple_loss=0.2425, pruned_loss=0.03681, over 12524.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2569, pruned_loss=0.03348, over 3061099.27 frames. ], batch size: 248, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:53:05,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4628, 3.4052, 3.4890, 3.5603, 3.5829, 3.3153, 3.5653, 3.6412], device='cuda:7'), covar=tensor([0.1336, 0.1040, 0.1126, 0.0739, 0.0709, 0.2402, 0.0958, 0.0877], device='cuda:7'), in_proj_covar=tensor([0.0640, 0.0783, 0.0904, 0.0799, 0.0607, 0.0633, 0.0668, 0.0773], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:53:24,792 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283819.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:54:13,349 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283842.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:54:24,071 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283848.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:54:29,958 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5844, 3.6694, 3.4205, 3.0903, 3.3057, 3.5571, 3.3439, 3.3749], device='cuda:7'), covar=tensor([0.0612, 0.0621, 0.0302, 0.0267, 0.0476, 0.0452, 0.1306, 0.0477], device='cuda:7'), in_proj_covar=tensor([0.0299, 0.0450, 0.0350, 0.0350, 0.0345, 0.0404, 0.0241, 0.0416], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 15:54:31,905 INFO [train.py:904] (7/8) Epoch 28, batch 9800, loss[loss=0.1483, simple_loss=0.2349, pruned_loss=0.03083, over 12467.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2573, pruned_loss=0.0327, over 3069924.11 frames. ], batch size: 248, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:54:55,272 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2923, 3.0068, 3.1344, 1.8733, 3.2816, 3.3566, 2.8181, 2.6906], device='cuda:7'), covar=tensor([0.0767, 0.0282, 0.0231, 0.1198, 0.0101, 0.0201, 0.0460, 0.0466], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0107, 0.0098, 0.0135, 0.0083, 0.0126, 0.0126, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 15:55:23,056 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.061e+02 2.416e+02 3.000e+02 5.773e+02, threshold=4.831e+02, percent-clipped=3.0 2023-05-02 15:56:11,010 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=283900.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 15:56:15,645 INFO [train.py:904] (7/8) Epoch 28, batch 9850, loss[loss=0.1651, simple_loss=0.2643, pruned_loss=0.03291, over 16393.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2586, pruned_loss=0.03291, over 3058772.41 frames. ], batch size: 146, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:56:17,474 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=283903.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:57:37,776 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=283939.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:58:00,280 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-02 15:58:06,907 INFO [train.py:904] (7/8) Epoch 28, batch 9900, loss[loss=0.1633, simple_loss=0.2684, pruned_loss=0.02912, over 16256.00 frames. ], tot_loss[loss=0.162, simple_loss=0.2586, pruned_loss=0.03273, over 3055983.17 frames. ], batch size: 165, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 15:58:38,493 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=283966.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 15:59:13,297 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 2.134e+02 2.412e+02 2.887e+02 8.278e+02, threshold=4.823e+02, percent-clipped=3.0 2023-05-02 15:59:30,176 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=283987.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:00:07,767 INFO [train.py:904] (7/8) Epoch 28, batch 9950, loss[loss=0.1461, simple_loss=0.2466, pruned_loss=0.02279, over 16726.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2615, pruned_loss=0.03312, over 3080632.18 frames. ], batch size: 83, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:01:08,225 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=284027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:02:08,468 INFO [train.py:904] (7/8) Epoch 28, batch 10000, loss[loss=0.1569, simple_loss=0.2597, pruned_loss=0.0271, over 16772.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2603, pruned_loss=0.03288, over 3093940.39 frames. ], batch size: 83, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:03:03,888 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.562e+02 2.120e+02 2.490e+02 3.242e+02 7.022e+02, threshold=4.980e+02, percent-clipped=2.0 2023-05-02 16:03:50,409 INFO [train.py:904] (7/8) Epoch 28, batch 10050, loss[loss=0.168, simple_loss=0.2594, pruned_loss=0.03832, over 12201.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2602, pruned_loss=0.03291, over 3079289.97 frames. ], batch size: 247, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:05:22,090 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 16:05:24,426 INFO [train.py:904] (7/8) Epoch 28, batch 10100, loss[loss=0.161, simple_loss=0.2507, pruned_loss=0.03571, over 16329.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.26, pruned_loss=0.03286, over 3081412.34 frames. ], batch size: 146, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:06:20,722 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.222e+02 2.734e+02 3.195e+02 6.018e+02, threshold=5.468e+02, percent-clipped=8.0 2023-05-02 16:06:39,455 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=284198.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:06:42,099 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=284200.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:06:44,470 INFO [train.py:904] (7/8) Epoch 28, batch 10150, loss[loss=0.143, simple_loss=0.2335, pruned_loss=0.02626, over 12617.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2587, pruned_loss=0.03284, over 3062689.35 frames. ], batch size: 250, lr: 2.38e-03, grad_scale: 8.0 2023-05-02 16:07:10,322 INFO [train.py:904] (7/8) Epoch 29, batch 0, loss[loss=0.2317, simple_loss=0.2867, pruned_loss=0.08832, over 16869.00 frames. ], tot_loss[loss=0.2317, simple_loss=0.2867, pruned_loss=0.08832, over 16869.00 frames. ], batch size: 109, lr: 2.34e-03, grad_scale: 8.0 2023-05-02 16:07:10,322 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 16:07:17,745 INFO [train.py:938] (7/8) Epoch 29, validation: loss=0.1427, simple_loss=0.246, pruned_loss=0.0197, over 944034.00 frames. 2023-05-02 16:07:17,746 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 16:07:53,559 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-05-02 16:08:07,727 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4227, 3.6516, 3.8643, 2.5940, 3.5461, 3.9578, 3.6084, 2.0551], device='cuda:7'), covar=tensor([0.0651, 0.0412, 0.0084, 0.0506, 0.0156, 0.0122, 0.0147, 0.0675], device='cuda:7'), in_proj_covar=tensor([0.0135, 0.0087, 0.0088, 0.0132, 0.0100, 0.0111, 0.0095, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 16:08:10,893 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1934, 3.2487, 3.4558, 2.3162, 3.0605, 2.4557, 3.7028, 3.7027], device='cuda:7'), covar=tensor([0.0267, 0.1076, 0.0785, 0.2173, 0.0954, 0.1236, 0.0572, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0163, 0.0165, 0.0153, 0.0143, 0.0129, 0.0141, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 16:08:19,053 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=284248.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:08:26,892 INFO [train.py:904] (7/8) Epoch 29, batch 50, loss[loss=0.1497, simple_loss=0.2405, pruned_loss=0.0295, over 17248.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2611, pruned_loss=0.04414, over 750738.66 frames. ], batch size: 45, lr: 2.34e-03, grad_scale: 2.0 2023-05-02 16:08:28,890 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 16:09:08,280 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.422e+02 3.076e+02 3.842e+02 2.178e+03, threshold=6.152e+02, percent-clipped=5.0 2023-05-02 16:09:37,386 INFO [train.py:904] (7/8) Epoch 29, batch 100, loss[loss=0.1647, simple_loss=0.2573, pruned_loss=0.03608, over 17105.00 frames. ], tot_loss[loss=0.1744, simple_loss=0.2605, pruned_loss=0.04416, over 1313888.11 frames. ], batch size: 47, lr: 2.34e-03, grad_scale: 2.0 2023-05-02 16:10:02,635 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=284322.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:10:46,114 INFO [train.py:904] (7/8) Epoch 29, batch 150, loss[loss=0.1838, simple_loss=0.2809, pruned_loss=0.04339, over 16616.00 frames. ], tot_loss[loss=0.1694, simple_loss=0.2573, pruned_loss=0.04076, over 1759728.35 frames. ], batch size: 62, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:11:25,631 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.248e+02 2.614e+02 3.025e+02 1.081e+03, threshold=5.228e+02, percent-clipped=2.0 2023-05-02 16:11:49,877 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 16:11:53,403 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4238, 4.2612, 4.3684, 4.5883, 4.6935, 4.3450, 4.5380, 4.6729], device='cuda:7'), covar=tensor([0.1834, 0.1412, 0.1912, 0.0967, 0.0835, 0.1165, 0.2327, 0.1014], device='cuda:7'), in_proj_covar=tensor([0.0652, 0.0797, 0.0918, 0.0811, 0.0617, 0.0642, 0.0681, 0.0787], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:11:55,152 INFO [train.py:904] (7/8) Epoch 29, batch 200, loss[loss=0.1748, simple_loss=0.2494, pruned_loss=0.05015, over 16877.00 frames. ], tot_loss[loss=0.1695, simple_loss=0.2573, pruned_loss=0.04084, over 2109911.15 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:12:51,474 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284443.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:13:04,661 INFO [train.py:904] (7/8) Epoch 29, batch 250, loss[loss=0.1755, simple_loss=0.2673, pruned_loss=0.04186, over 16770.00 frames. ], tot_loss[loss=0.1689, simple_loss=0.2567, pruned_loss=0.04059, over 2373375.01 frames. ], batch size: 57, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:13:47,581 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.177e+02 2.500e+02 3.084e+02 6.085e+02, threshold=5.001e+02, percent-clipped=1.0 2023-05-02 16:14:09,292 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=284498.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:14:12,103 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4218, 4.2342, 4.4426, 4.6041, 4.7187, 4.2711, 4.5583, 4.6998], device='cuda:7'), covar=tensor([0.1737, 0.1254, 0.1489, 0.0743, 0.0619, 0.1201, 0.2235, 0.0795], device='cuda:7'), in_proj_covar=tensor([0.0657, 0.0802, 0.0925, 0.0817, 0.0621, 0.0647, 0.0686, 0.0792], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:14:17,001 INFO [train.py:904] (7/8) Epoch 29, batch 300, loss[loss=0.1685, simple_loss=0.2507, pruned_loss=0.04317, over 16899.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2551, pruned_loss=0.04034, over 2576843.07 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:14:17,423 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=284504.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:14:48,815 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.68 vs. limit=2.0 2023-05-02 16:15:14,306 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=284546.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:15:23,226 INFO [train.py:904] (7/8) Epoch 29, batch 350, loss[loss=0.1779, simple_loss=0.2545, pruned_loss=0.05072, over 16884.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2534, pruned_loss=0.03938, over 2740112.90 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:15:25,833 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-05-02 16:16:02,937 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.083e+02 2.374e+02 2.884e+02 4.611e+02, threshold=4.748e+02, percent-clipped=0.0 2023-05-02 16:16:23,776 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-05-02 16:16:32,000 INFO [train.py:904] (7/8) Epoch 29, batch 400, loss[loss=0.1522, simple_loss=0.244, pruned_loss=0.03025, over 16672.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2516, pruned_loss=0.03848, over 2876369.69 frames. ], batch size: 57, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:16:57,117 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=284622.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:16:59,778 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-02 16:17:36,801 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-05-02 16:17:41,218 INFO [train.py:904] (7/8) Epoch 29, batch 450, loss[loss=0.1796, simple_loss=0.2556, pruned_loss=0.05177, over 12266.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2498, pruned_loss=0.03825, over 2967847.93 frames. ], batch size: 246, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:17:45,100 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.92 vs. limit=5.0 2023-05-02 16:17:46,129 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2438, 5.1929, 4.9687, 4.4957, 5.0540, 1.9385, 4.8451, 4.6696], device='cuda:7'), covar=tensor([0.0099, 0.0084, 0.0234, 0.0412, 0.0110, 0.3205, 0.0148, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0172, 0.0210, 0.0180, 0.0186, 0.0216, 0.0198, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:18:03,052 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=284670.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:18:18,522 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.454e+02 1.978e+02 2.312e+02 2.858e+02 4.933e+02, threshold=4.623e+02, percent-clipped=2.0 2023-05-02 16:18:47,411 INFO [train.py:904] (7/8) Epoch 29, batch 500, loss[loss=0.1476, simple_loss=0.242, pruned_loss=0.02659, over 16849.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2483, pruned_loss=0.03738, over 3049371.13 frames. ], batch size: 42, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:18:49,144 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6212, 3.6673, 4.3647, 2.5399, 3.5799, 2.8268, 4.0134, 3.9993], device='cuda:7'), covar=tensor([0.0235, 0.0951, 0.0406, 0.1926, 0.0735, 0.0910, 0.0557, 0.1051], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0168, 0.0169, 0.0156, 0.0147, 0.0132, 0.0145, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 16:19:48,329 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-05-02 16:19:56,203 INFO [train.py:904] (7/8) Epoch 29, batch 550, loss[loss=0.1263, simple_loss=0.211, pruned_loss=0.02082, over 16831.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2478, pruned_loss=0.03723, over 3104337.29 frames. ], batch size: 39, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:20:35,009 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1794, 5.1439, 4.9154, 4.3666, 4.9483, 1.9057, 4.7623, 4.6877], device='cuda:7'), covar=tensor([0.0104, 0.0096, 0.0252, 0.0451, 0.0121, 0.3009, 0.0157, 0.0290], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0173, 0.0211, 0.0181, 0.0186, 0.0217, 0.0198, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:20:35,718 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.399e+02 1.886e+02 2.233e+02 2.634e+02 5.354e+02, threshold=4.465e+02, percent-clipped=1.0 2023-05-02 16:20:52,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2832, 5.2783, 5.1613, 4.6328, 4.8067, 5.2286, 5.1454, 4.8240], device='cuda:7'), covar=tensor([0.0637, 0.0571, 0.0380, 0.0389, 0.1176, 0.0543, 0.0322, 0.0930], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0469, 0.0364, 0.0366, 0.0360, 0.0421, 0.0251, 0.0434], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:20:57,947 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=284799.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:21:04,187 INFO [train.py:904] (7/8) Epoch 29, batch 600, loss[loss=0.1648, simple_loss=0.2398, pruned_loss=0.04487, over 16118.00 frames. ], tot_loss[loss=0.1615, simple_loss=0.2476, pruned_loss=0.0377, over 3155827.43 frames. ], batch size: 164, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:21:21,958 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284816.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:22:12,351 INFO [train.py:904] (7/8) Epoch 29, batch 650, loss[loss=0.1691, simple_loss=0.2737, pruned_loss=0.03226, over 17278.00 frames. ], tot_loss[loss=0.1602, simple_loss=0.2461, pruned_loss=0.03715, over 3186802.49 frames. ], batch size: 52, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:22:46,737 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=284877.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:22:53,845 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.111e+02 2.518e+02 3.065e+02 7.660e+02, threshold=5.037e+02, percent-clipped=4.0 2023-05-02 16:23:22,622 INFO [train.py:904] (7/8) Epoch 29, batch 700, loss[loss=0.1745, simple_loss=0.2602, pruned_loss=0.04444, over 17083.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2455, pruned_loss=0.03645, over 3208500.53 frames. ], batch size: 53, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:24:05,451 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9241, 2.1899, 2.3456, 3.4083, 2.1780, 2.3899, 2.3200, 2.3201], device='cuda:7'), covar=tensor([0.1680, 0.3593, 0.3293, 0.0898, 0.4338, 0.2798, 0.3708, 0.3581], device='cuda:7'), in_proj_covar=tensor([0.0420, 0.0474, 0.0389, 0.0336, 0.0447, 0.0542, 0.0448, 0.0555], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:24:12,219 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284939.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:24:30,129 INFO [train.py:904] (7/8) Epoch 29, batch 750, loss[loss=0.1688, simple_loss=0.2468, pruned_loss=0.04538, over 12387.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2458, pruned_loss=0.03675, over 3227000.31 frames. ], batch size: 246, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:25:13,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 2.091e+02 2.322e+02 2.629e+02 4.322e+02, threshold=4.644e+02, percent-clipped=0.0 2023-05-02 16:25:30,698 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=284996.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:25:37,930 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285000.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:25:42,931 INFO [train.py:904] (7/8) Epoch 29, batch 800, loss[loss=0.1662, simple_loss=0.2512, pruned_loss=0.04055, over 16882.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.2462, pruned_loss=0.03705, over 3244260.32 frames. ], batch size: 109, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:26:08,939 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6058, 6.0010, 5.7526, 5.8176, 5.4073, 5.5218, 5.3532, 6.1383], device='cuda:7'), covar=tensor([0.1404, 0.0977, 0.1093, 0.0855, 0.0862, 0.0684, 0.1382, 0.0921], device='cuda:7'), in_proj_covar=tensor([0.0720, 0.0869, 0.0712, 0.0673, 0.0549, 0.0549, 0.0730, 0.0681], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:26:34,326 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=285042.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:26:49,083 INFO [train.py:904] (7/8) Epoch 29, batch 850, loss[loss=0.1609, simple_loss=0.2567, pruned_loss=0.03262, over 17059.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2456, pruned_loss=0.03686, over 3261166.69 frames. ], batch size: 50, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:26:53,089 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285057.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:26:54,228 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9894, 2.9860, 2.6137, 5.0749, 4.0550, 4.3538, 1.6303, 3.3032], device='cuda:7'), covar=tensor([0.1343, 0.0800, 0.1387, 0.0196, 0.0294, 0.0466, 0.1725, 0.0785], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0180, 0.0200, 0.0203, 0.0204, 0.0217, 0.0210, 0.0198], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 16:27:01,068 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8792, 2.5456, 2.0820, 2.3754, 2.8902, 2.6737, 2.8120, 2.9552], device='cuda:7'), covar=tensor([0.0306, 0.0471, 0.0646, 0.0507, 0.0300, 0.0394, 0.0283, 0.0360], device='cuda:7'), in_proj_covar=tensor([0.0232, 0.0246, 0.0236, 0.0236, 0.0247, 0.0246, 0.0241, 0.0245], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:27:09,260 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1292, 4.5215, 4.5535, 3.4195, 3.7032, 4.4773, 3.9946, 2.6514], device='cuda:7'), covar=tensor([0.0455, 0.0074, 0.0047, 0.0348, 0.0171, 0.0106, 0.0114, 0.0510], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0091, 0.0135, 0.0102, 0.0115, 0.0098, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 16:27:31,496 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.142e+02 2.515e+02 3.015e+02 5.883e+02, threshold=5.030e+02, percent-clipped=4.0 2023-05-02 16:27:47,570 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 16:27:49,548 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285099.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:27:54,972 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285103.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:27:55,653 INFO [train.py:904] (7/8) Epoch 29, batch 900, loss[loss=0.1682, simple_loss=0.2417, pruned_loss=0.04735, over 16743.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2454, pruned_loss=0.03651, over 3280320.83 frames. ], batch size: 124, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:28:20,483 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6927, 3.8719, 2.9300, 2.3370, 2.4366, 2.4466, 4.0051, 3.2647], device='cuda:7'), covar=tensor([0.2974, 0.0592, 0.1831, 0.3099, 0.2987, 0.2296, 0.0517, 0.1764], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0274, 0.0314, 0.0328, 0.0304, 0.0279, 0.0305, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:28:56,119 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285147.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:29:05,602 INFO [train.py:904] (7/8) Epoch 29, batch 950, loss[loss=0.1704, simple_loss=0.2466, pruned_loss=0.0471, over 16882.00 frames. ], tot_loss[loss=0.1592, simple_loss=0.2453, pruned_loss=0.03653, over 3282741.52 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:29:30,496 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285172.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:29:47,165 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.122e+02 2.524e+02 3.029e+02 7.170e+02, threshold=5.047e+02, percent-clipped=3.0 2023-05-02 16:30:14,197 INFO [train.py:904] (7/8) Epoch 29, batch 1000, loss[loss=0.1594, simple_loss=0.2382, pruned_loss=0.04032, over 15594.00 frames. ], tot_loss[loss=0.1586, simple_loss=0.2443, pruned_loss=0.03648, over 3295804.07 frames. ], batch size: 191, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:31:24,212 INFO [train.py:904] (7/8) Epoch 29, batch 1050, loss[loss=0.1581, simple_loss=0.2372, pruned_loss=0.03946, over 16871.00 frames. ], tot_loss[loss=0.1589, simple_loss=0.2445, pruned_loss=0.03661, over 3283716.29 frames. ], batch size: 109, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:32:05,304 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 1.938e+02 2.240e+02 2.554e+02 8.528e+02, threshold=4.479e+02, percent-clipped=1.0 2023-05-02 16:32:19,767 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285295.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:32:31,249 INFO [train.py:904] (7/8) Epoch 29, batch 1100, loss[loss=0.17, simple_loss=0.2573, pruned_loss=0.04131, over 17041.00 frames. ], tot_loss[loss=0.1588, simple_loss=0.2446, pruned_loss=0.03644, over 3284735.77 frames. ], batch size: 53, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:33:37,442 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285352.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:33:40,187 INFO [train.py:904] (7/8) Epoch 29, batch 1150, loss[loss=0.179, simple_loss=0.2499, pruned_loss=0.05404, over 16907.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.2448, pruned_loss=0.03585, over 3293447.39 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 2.0 2023-05-02 16:34:22,233 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.113e+02 2.491e+02 2.895e+02 1.258e+03, threshold=4.983e+02, percent-clipped=2.0 2023-05-02 16:34:40,322 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285398.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 16:34:47,470 INFO [train.py:904] (7/8) Epoch 29, batch 1200, loss[loss=0.1361, simple_loss=0.2289, pruned_loss=0.02164, over 17262.00 frames. ], tot_loss[loss=0.1574, simple_loss=0.2441, pruned_loss=0.0354, over 3297679.84 frames. ], batch size: 45, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:35:47,422 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8635, 5.0871, 5.2750, 5.0053, 5.0959, 5.7106, 5.1898, 4.9387], device='cuda:7'), covar=tensor([0.1367, 0.2251, 0.2944, 0.2460, 0.2698, 0.1092, 0.1926, 0.2505], device='cuda:7'), in_proj_covar=tensor([0.0432, 0.0645, 0.0717, 0.0524, 0.0703, 0.0737, 0.0555, 0.0701], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:35:56,652 INFO [train.py:904] (7/8) Epoch 29, batch 1250, loss[loss=0.1744, simple_loss=0.241, pruned_loss=0.05393, over 16764.00 frames. ], tot_loss[loss=0.1582, simple_loss=0.2444, pruned_loss=0.03597, over 3303295.27 frames. ], batch size: 83, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:36:21,897 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285472.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:36:36,011 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-05-02 16:36:38,704 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.107e+02 2.455e+02 3.038e+02 4.730e+02, threshold=4.911e+02, percent-clipped=0.0 2023-05-02 16:36:40,490 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7422, 4.0591, 2.9805, 2.3943, 2.6582, 2.6554, 4.2312, 3.4269], device='cuda:7'), covar=tensor([0.2967, 0.0612, 0.1985, 0.3232, 0.2849, 0.2216, 0.0547, 0.1554], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0275, 0.0315, 0.0329, 0.0306, 0.0281, 0.0306, 0.0355], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:37:05,021 INFO [train.py:904] (7/8) Epoch 29, batch 1300, loss[loss=0.1597, simple_loss=0.2446, pruned_loss=0.0374, over 16487.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.2436, pruned_loss=0.03612, over 3301270.48 frames. ], batch size: 146, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:37:28,090 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285520.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:38:04,245 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.02 vs. limit=5.0 2023-05-02 16:38:12,054 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1418, 5.2335, 5.6327, 5.5927, 5.6274, 5.2560, 5.2267, 5.0604], device='cuda:7'), covar=tensor([0.0382, 0.0538, 0.0397, 0.0442, 0.0486, 0.0449, 0.0948, 0.0441], device='cuda:7'), in_proj_covar=tensor([0.0444, 0.0502, 0.0483, 0.0445, 0.0532, 0.0507, 0.0582, 0.0407], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 16:38:13,889 INFO [train.py:904] (7/8) Epoch 29, batch 1350, loss[loss=0.1699, simple_loss=0.2502, pruned_loss=0.04484, over 16825.00 frames. ], tot_loss[loss=0.1568, simple_loss=0.2428, pruned_loss=0.03539, over 3302468.87 frames. ], batch size: 96, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:38:57,529 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3747, 6.0644, 6.1311, 5.7665, 5.9329, 6.4288, 5.9582, 5.6890], device='cuda:7'), covar=tensor([0.0874, 0.1724, 0.2194, 0.2158, 0.2365, 0.1004, 0.1489, 0.2201], device='cuda:7'), in_proj_covar=tensor([0.0431, 0.0643, 0.0716, 0.0524, 0.0701, 0.0734, 0.0554, 0.0699], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:38:58,362 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.007e+02 2.357e+02 2.705e+02 4.553e+02, threshold=4.714e+02, percent-clipped=0.0 2023-05-02 16:39:11,761 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285595.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:39:22,820 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7215, 5.0461, 4.8229, 4.8655, 4.5547, 4.5149, 4.5098, 5.1082], device='cuda:7'), covar=tensor([0.1274, 0.0866, 0.1024, 0.0857, 0.0873, 0.1368, 0.1220, 0.0957], device='cuda:7'), in_proj_covar=tensor([0.0731, 0.0881, 0.0722, 0.0685, 0.0558, 0.0557, 0.0742, 0.0692], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:39:24,824 INFO [train.py:904] (7/8) Epoch 29, batch 1400, loss[loss=0.1515, simple_loss=0.2362, pruned_loss=0.03336, over 16485.00 frames. ], tot_loss[loss=0.1567, simple_loss=0.2426, pruned_loss=0.03542, over 3304071.07 frames. ], batch size: 68, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:39:49,685 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=285622.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:40:14,003 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9515, 5.2563, 5.3990, 5.1538, 5.2238, 5.8091, 5.2623, 4.9683], device='cuda:7'), covar=tensor([0.1393, 0.2067, 0.2385, 0.2120, 0.2853, 0.1057, 0.1868, 0.2536], device='cuda:7'), in_proj_covar=tensor([0.0431, 0.0643, 0.0715, 0.0524, 0.0702, 0.0734, 0.0554, 0.0700], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:40:19,874 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285643.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:40:31,681 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285652.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:40:33,565 INFO [train.py:904] (7/8) Epoch 29, batch 1450, loss[loss=0.1577, simple_loss=0.2466, pruned_loss=0.03442, over 17209.00 frames. ], tot_loss[loss=0.1562, simple_loss=0.242, pruned_loss=0.03522, over 3307078.95 frames. ], batch size: 44, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:40:43,706 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8281, 4.9631, 5.1251, 4.9265, 4.9664, 5.5902, 5.0785, 4.7881], device='cuda:7'), covar=tensor([0.1364, 0.2132, 0.2564, 0.2182, 0.2561, 0.1023, 0.1806, 0.2506], device='cuda:7'), in_proj_covar=tensor([0.0431, 0.0643, 0.0715, 0.0524, 0.0702, 0.0734, 0.0553, 0.0700], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:41:13,955 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=285683.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:41:16,390 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.247e+02 2.721e+02 3.292e+02 6.741e+02, threshold=5.441e+02, percent-clipped=4.0 2023-05-02 16:41:34,547 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=285698.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 16:41:37,508 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285700.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:41:43,191 INFO [train.py:904] (7/8) Epoch 29, batch 1500, loss[loss=0.1321, simple_loss=0.2173, pruned_loss=0.02346, over 16849.00 frames. ], tot_loss[loss=0.1563, simple_loss=0.2418, pruned_loss=0.03542, over 3321079.35 frames. ], batch size: 42, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:42:19,599 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1851, 5.1800, 4.8990, 4.3489, 5.0022, 1.9754, 4.7642, 4.6635], device='cuda:7'), covar=tensor([0.0128, 0.0121, 0.0287, 0.0492, 0.0150, 0.3080, 0.0182, 0.0311], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0176, 0.0215, 0.0185, 0.0192, 0.0220, 0.0202, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:42:40,635 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=285746.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 16:42:51,736 INFO [train.py:904] (7/8) Epoch 29, batch 1550, loss[loss=0.1938, simple_loss=0.2653, pruned_loss=0.06114, over 16804.00 frames. ], tot_loss[loss=0.1568, simple_loss=0.2426, pruned_loss=0.03552, over 3331419.29 frames. ], batch size: 83, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:43:34,665 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.209e+02 2.639e+02 3.034e+02 6.721e+02, threshold=5.278e+02, percent-clipped=2.0 2023-05-02 16:44:00,614 INFO [train.py:904] (7/8) Epoch 29, batch 1600, loss[loss=0.1396, simple_loss=0.2292, pruned_loss=0.02496, over 17022.00 frames. ], tot_loss[loss=0.158, simple_loss=0.2445, pruned_loss=0.03578, over 3328100.09 frames. ], batch size: 41, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:44:17,242 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9549, 4.7371, 5.0144, 5.1939, 5.3610, 4.7389, 5.3562, 5.3834], device='cuda:7'), covar=tensor([0.2063, 0.1446, 0.1850, 0.0853, 0.0634, 0.1041, 0.0631, 0.0741], device='cuda:7'), in_proj_covar=tensor([0.0699, 0.0853, 0.0985, 0.0869, 0.0660, 0.0683, 0.0727, 0.0840], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:44:33,753 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3176, 2.3996, 2.3871, 4.1197, 2.2226, 2.7645, 2.4054, 2.4777], device='cuda:7'), covar=tensor([0.1507, 0.3672, 0.3264, 0.0641, 0.4442, 0.2637, 0.3722, 0.3774], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0478, 0.0391, 0.0339, 0.0449, 0.0548, 0.0451, 0.0562], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:44:53,654 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-05-02 16:45:09,524 INFO [train.py:904] (7/8) Epoch 29, batch 1650, loss[loss=0.166, simple_loss=0.2621, pruned_loss=0.03499, over 17049.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.2468, pruned_loss=0.03689, over 3324121.55 frames. ], batch size: 55, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:45:50,990 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.202e+02 2.447e+02 3.087e+02 4.596e+02, threshold=4.895e+02, percent-clipped=0.0 2023-05-02 16:46:16,879 INFO [train.py:904] (7/8) Epoch 29, batch 1700, loss[loss=0.1773, simple_loss=0.2595, pruned_loss=0.04757, over 16893.00 frames. ], tot_loss[loss=0.1601, simple_loss=0.247, pruned_loss=0.03662, over 3333232.88 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:47:12,967 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-05-02 16:47:24,310 INFO [train.py:904] (7/8) Epoch 29, batch 1750, loss[loss=0.1478, simple_loss=0.2435, pruned_loss=0.02601, over 17199.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2478, pruned_loss=0.03726, over 3329589.18 frames. ], batch size: 46, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:47:58,395 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=285978.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:48:07,053 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.447e+02 2.051e+02 2.355e+02 2.951e+02 5.834e+02, threshold=4.710e+02, percent-clipped=3.0 2023-05-02 16:48:36,742 INFO [train.py:904] (7/8) Epoch 29, batch 1800, loss[loss=0.1879, simple_loss=0.2652, pruned_loss=0.05532, over 16850.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2491, pruned_loss=0.03721, over 3333758.32 frames. ], batch size: 109, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:49:43,967 INFO [train.py:904] (7/8) Epoch 29, batch 1850, loss[loss=0.1403, simple_loss=0.2294, pruned_loss=0.02566, over 16827.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2505, pruned_loss=0.03748, over 3340206.84 frames. ], batch size: 42, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:50:16,630 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-05-02 16:50:28,096 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.081e+02 2.345e+02 2.768e+02 6.474e+02, threshold=4.689e+02, percent-clipped=2.0 2023-05-02 16:50:53,109 INFO [train.py:904] (7/8) Epoch 29, batch 1900, loss[loss=0.1832, simple_loss=0.2726, pruned_loss=0.0469, over 12012.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2508, pruned_loss=0.03706, over 3337477.97 frames. ], batch size: 248, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:51:20,389 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286123.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:51:29,260 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286129.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:51:56,552 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 16:51:58,704 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-02 16:51:59,686 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 16:52:04,253 INFO [train.py:904] (7/8) Epoch 29, batch 1950, loss[loss=0.1908, simple_loss=0.2768, pruned_loss=0.0524, over 16263.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2513, pruned_loss=0.0374, over 3329753.93 frames. ], batch size: 165, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:52:24,357 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8442, 4.4102, 3.1343, 2.4403, 2.7189, 2.6689, 4.8023, 3.5684], device='cuda:7'), covar=tensor([0.3159, 0.0592, 0.1937, 0.2966, 0.3056, 0.2188, 0.0360, 0.1563], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0277, 0.0316, 0.0330, 0.0308, 0.0281, 0.0307, 0.0357], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:52:45,846 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3958, 2.4025, 2.3365, 4.1907, 2.3432, 2.7552, 2.4700, 2.5503], device='cuda:7'), covar=tensor([0.1462, 0.3952, 0.3497, 0.0579, 0.4293, 0.2761, 0.3876, 0.3853], device='cuda:7'), in_proj_covar=tensor([0.0424, 0.0478, 0.0390, 0.0339, 0.0448, 0.0548, 0.0451, 0.0560], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:52:46,906 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286184.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:52:48,717 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.778e+02 2.194e+02 2.537e+02 3.082e+02 2.053e+03, threshold=5.073e+02, percent-clipped=1.0 2023-05-02 16:52:49,814 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5978, 2.2696, 1.8408, 2.0812, 2.6106, 2.4123, 2.4981, 2.7005], device='cuda:7'), covar=tensor([0.0271, 0.0486, 0.0638, 0.0524, 0.0288, 0.0376, 0.0262, 0.0363], device='cuda:7'), in_proj_covar=tensor([0.0239, 0.0253, 0.0240, 0.0241, 0.0253, 0.0251, 0.0249, 0.0251], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 16:52:55,399 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286190.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:53:10,094 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3472, 3.4058, 2.1897, 3.5992, 2.7031, 3.5562, 2.2774, 2.7573], device='cuda:7'), covar=tensor([0.0339, 0.0465, 0.1687, 0.0391, 0.0828, 0.0841, 0.1457, 0.0808], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0184, 0.0200, 0.0178, 0.0183, 0.0225, 0.0209, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 16:53:13,099 INFO [train.py:904] (7/8) Epoch 29, batch 2000, loss[loss=0.156, simple_loss=0.2379, pruned_loss=0.03707, over 16749.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2506, pruned_loss=0.03714, over 3322499.39 frames. ], batch size: 124, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:54:21,812 INFO [train.py:904] (7/8) Epoch 29, batch 2050, loss[loss=0.1529, simple_loss=0.2355, pruned_loss=0.0351, over 16836.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2505, pruned_loss=0.03731, over 3331450.48 frames. ], batch size: 102, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:54:54,821 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286278.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:55:04,679 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.605e+02 2.125e+02 2.436e+02 2.931e+02 6.185e+02, threshold=4.871e+02, percent-clipped=3.0 2023-05-02 16:55:05,387 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8465, 4.4011, 3.1118, 2.3780, 2.6111, 2.6456, 4.7440, 3.5513], device='cuda:7'), covar=tensor([0.3071, 0.0572, 0.1915, 0.3227, 0.3177, 0.2264, 0.0345, 0.1557], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0278, 0.0318, 0.0331, 0.0310, 0.0283, 0.0309, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 16:55:18,588 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286295.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:55:30,418 INFO [train.py:904] (7/8) Epoch 29, batch 2100, loss[loss=0.2065, simple_loss=0.2857, pruned_loss=0.06363, over 16866.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2524, pruned_loss=0.03842, over 3332647.37 frames. ], batch size: 116, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:55:54,401 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-02 16:56:00,354 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286326.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:56:26,293 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286344.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:56:40,456 INFO [train.py:904] (7/8) Epoch 29, batch 2150, loss[loss=0.1543, simple_loss=0.2447, pruned_loss=0.03192, over 16846.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2538, pruned_loss=0.03897, over 3330126.96 frames. ], batch size: 102, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 16:56:44,762 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286356.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:57:24,983 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.220e+02 2.758e+02 3.117e+02 6.077e+02, threshold=5.516e+02, percent-clipped=2.0 2023-05-02 16:57:50,676 INFO [train.py:904] (7/8) Epoch 29, batch 2200, loss[loss=0.1748, simple_loss=0.2754, pruned_loss=0.03712, over 16678.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2546, pruned_loss=0.03887, over 3340628.65 frames. ], batch size: 57, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:57:53,017 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286405.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:58:59,532 INFO [train.py:904] (7/8) Epoch 29, batch 2250, loss[loss=0.1739, simple_loss=0.253, pruned_loss=0.04743, over 16308.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.255, pruned_loss=0.03915, over 3333659.29 frames. ], batch size: 165, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 16:59:07,436 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-05-02 16:59:34,558 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286479.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 16:59:43,732 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286485.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 16:59:45,655 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.269e+02 2.585e+02 3.162e+02 6.397e+02, threshold=5.169e+02, percent-clipped=1.0 2023-05-02 16:59:48,968 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0370, 5.0647, 5.5047, 5.5024, 5.5318, 5.1569, 5.1470, 4.9358], device='cuda:7'), covar=tensor([0.0419, 0.0670, 0.0426, 0.0441, 0.0507, 0.0439, 0.0933, 0.0519], device='cuda:7'), in_proj_covar=tensor([0.0450, 0.0509, 0.0491, 0.0452, 0.0539, 0.0514, 0.0591, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 17:00:08,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7298, 4.2792, 3.0270, 2.3423, 2.6629, 2.5207, 4.6181, 3.5040], device='cuda:7'), covar=tensor([0.3112, 0.0511, 0.1979, 0.3173, 0.3080, 0.2370, 0.0408, 0.1468], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0278, 0.0318, 0.0331, 0.0309, 0.0283, 0.0309, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:00:08,684 INFO [train.py:904] (7/8) Epoch 29, batch 2300, loss[loss=0.1818, simple_loss=0.2702, pruned_loss=0.04666, over 16709.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.255, pruned_loss=0.03906, over 3338104.32 frames. ], batch size: 57, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:00:12,317 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2502, 3.3725, 3.7241, 2.2612, 3.1185, 2.5099, 3.6918, 3.7212], device='cuda:7'), covar=tensor([0.0296, 0.1032, 0.0607, 0.2115, 0.0879, 0.1093, 0.0576, 0.0918], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0173, 0.0172, 0.0160, 0.0150, 0.0134, 0.0148, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 17:00:28,075 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8316, 3.9592, 4.1407, 4.1351, 4.1561, 3.9363, 3.9657, 3.9140], device='cuda:7'), covar=tensor([0.0414, 0.0643, 0.0461, 0.0431, 0.0558, 0.0502, 0.0737, 0.0583], device='cuda:7'), in_proj_covar=tensor([0.0448, 0.0506, 0.0490, 0.0450, 0.0537, 0.0513, 0.0589, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 17:00:51,931 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9672, 2.9325, 2.6929, 4.4378, 3.6980, 4.1673, 1.7078, 3.1325], device='cuda:7'), covar=tensor([0.1372, 0.0683, 0.1145, 0.0190, 0.0170, 0.0396, 0.1595, 0.0817], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0206, 0.0207, 0.0221, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:01:13,724 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2765, 5.8403, 5.9593, 5.6325, 5.7295, 6.2850, 5.8421, 5.5416], device='cuda:7'), covar=tensor([0.0863, 0.1823, 0.2838, 0.1912, 0.2460, 0.0924, 0.1525, 0.2275], device='cuda:7'), in_proj_covar=tensor([0.0440, 0.0654, 0.0728, 0.0532, 0.0714, 0.0747, 0.0562, 0.0710], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:01:17,624 INFO [train.py:904] (7/8) Epoch 29, batch 2350, loss[loss=0.1443, simple_loss=0.2364, pruned_loss=0.02608, over 17210.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2555, pruned_loss=0.03966, over 3328045.05 frames. ], batch size: 44, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:01:22,270 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8059, 4.7106, 4.7183, 4.3719, 4.4646, 4.7385, 4.5892, 4.5015], device='cuda:7'), covar=tensor([0.0644, 0.0985, 0.0315, 0.0361, 0.0814, 0.0595, 0.0426, 0.0651], device='cuda:7'), in_proj_covar=tensor([0.0324, 0.0489, 0.0380, 0.0382, 0.0375, 0.0438, 0.0260, 0.0454], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:02:03,068 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.083e+02 2.366e+02 2.716e+02 6.133e+02, threshold=4.732e+02, percent-clipped=1.0 2023-05-02 17:02:05,584 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286588.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:02:27,440 INFO [train.py:904] (7/8) Epoch 29, batch 2400, loss[loss=0.185, simple_loss=0.2802, pruned_loss=0.04493, over 17106.00 frames. ], tot_loss[loss=0.1677, simple_loss=0.2556, pruned_loss=0.03989, over 3314344.70 frames. ], batch size: 55, lr: 2.33e-03, grad_scale: 8.0 2023-05-02 17:03:30,998 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286649.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:03:32,001 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9050, 4.9335, 5.3578, 5.3391, 5.3894, 5.0285, 4.9637, 4.7609], device='cuda:7'), covar=tensor([0.0376, 0.0658, 0.0385, 0.0451, 0.0518, 0.0395, 0.1042, 0.0518], device='cuda:7'), in_proj_covar=tensor([0.0449, 0.0509, 0.0491, 0.0452, 0.0539, 0.0515, 0.0592, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 17:03:34,197 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286651.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:03:37,428 INFO [train.py:904] (7/8) Epoch 29, batch 2450, loss[loss=0.1509, simple_loss=0.2496, pruned_loss=0.02607, over 17204.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2559, pruned_loss=0.03953, over 3310238.58 frames. ], batch size: 44, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:04:00,711 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286671.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:04:23,750 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.201e+02 2.590e+02 3.137e+02 8.541e+02, threshold=5.179e+02, percent-clipped=2.0 2023-05-02 17:04:41,148 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286700.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:04:46,161 INFO [train.py:904] (7/8) Epoch 29, batch 2500, loss[loss=0.1746, simple_loss=0.2565, pruned_loss=0.04634, over 16858.00 frames. ], tot_loss[loss=0.1665, simple_loss=0.2546, pruned_loss=0.03919, over 3313554.10 frames. ], batch size: 96, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:05:16,460 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2023-05-02 17:05:25,026 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286732.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:05:38,098 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8851, 2.5780, 2.0745, 2.3013, 2.9038, 2.6959, 2.8692, 2.9903], device='cuda:7'), covar=tensor([0.0246, 0.0404, 0.0574, 0.0485, 0.0252, 0.0370, 0.0247, 0.0304], device='cuda:7'), in_proj_covar=tensor([0.0241, 0.0254, 0.0242, 0.0242, 0.0254, 0.0253, 0.0250, 0.0252], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:05:51,677 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 17:05:55,671 INFO [train.py:904] (7/8) Epoch 29, batch 2550, loss[loss=0.165, simple_loss=0.2484, pruned_loss=0.0408, over 16805.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2558, pruned_loss=0.03994, over 3313966.98 frames. ], batch size: 102, lr: 2.33e-03, grad_scale: 4.0 2023-05-02 17:05:58,992 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.55 vs. limit=2.0 2023-05-02 17:06:19,635 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=286770.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:06:32,955 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286779.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:06:41,022 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286785.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:06:44,660 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.642e+02 2.042e+02 2.418e+02 2.906e+02 6.403e+02, threshold=4.836e+02, percent-clipped=2.0 2023-05-02 17:07:07,735 INFO [train.py:904] (7/8) Epoch 29, batch 2600, loss[loss=0.1532, simple_loss=0.2379, pruned_loss=0.03426, over 16810.00 frames. ], tot_loss[loss=0.1666, simple_loss=0.2549, pruned_loss=0.03916, over 3320744.31 frames. ], batch size: 102, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:07:39,072 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286827.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:07:40,534 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-05-02 17:07:45,126 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=286831.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:07:47,351 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286833.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:08:15,901 INFO [train.py:904] (7/8) Epoch 29, batch 2650, loss[loss=0.1826, simple_loss=0.2621, pruned_loss=0.05155, over 16875.00 frames. ], tot_loss[loss=0.1673, simple_loss=0.2562, pruned_loss=0.03914, over 3324860.38 frames. ], batch size: 116, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:08:31,105 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2023-05-02 17:09:00,293 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.057e+02 2.433e+02 2.859e+02 4.818e+02, threshold=4.867e+02, percent-clipped=0.0 2023-05-02 17:09:22,346 INFO [train.py:904] (7/8) Epoch 29, batch 2700, loss[loss=0.1586, simple_loss=0.2596, pruned_loss=0.02882, over 17115.00 frames. ], tot_loss[loss=0.1674, simple_loss=0.2568, pruned_loss=0.03893, over 3328925.70 frames. ], batch size: 49, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:09:51,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9668, 4.4770, 4.3681, 3.2994, 3.6860, 4.3782, 3.8951, 2.4629], device='cuda:7'), covar=tensor([0.0503, 0.0088, 0.0057, 0.0361, 0.0169, 0.0104, 0.0098, 0.0527], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0093, 0.0137, 0.0104, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 17:10:17,483 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=286944.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:10:26,956 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=286951.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:10:30,041 INFO [train.py:904] (7/8) Epoch 29, batch 2750, loss[loss=0.1795, simple_loss=0.2819, pruned_loss=0.03854, over 16732.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.257, pruned_loss=0.03861, over 3328949.44 frames. ], batch size: 57, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:11:17,994 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 1.930e+02 2.341e+02 2.772e+02 4.818e+02, threshold=4.681e+02, percent-clipped=0.0 2023-05-02 17:11:33,748 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=286999.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:11:35,360 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287000.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:11:40,753 INFO [train.py:904] (7/8) Epoch 29, batch 2800, loss[loss=0.159, simple_loss=0.2598, pruned_loss=0.02909, over 17273.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2565, pruned_loss=0.03775, over 3336030.35 frames. ], batch size: 52, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:11:41,105 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6527, 6.0518, 5.7499, 5.8344, 5.5003, 5.4256, 5.4904, 6.1915], device='cuda:7'), covar=tensor([0.1625, 0.1050, 0.1240, 0.1048, 0.0894, 0.0653, 0.1422, 0.1047], device='cuda:7'), in_proj_covar=tensor([0.0735, 0.0889, 0.0728, 0.0689, 0.0562, 0.0560, 0.0747, 0.0699], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:12:13,070 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:12:42,719 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287048.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:12:50,055 INFO [train.py:904] (7/8) Epoch 29, batch 2850, loss[loss=0.1782, simple_loss=0.2742, pruned_loss=0.04106, over 17068.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2555, pruned_loss=0.03749, over 3334509.92 frames. ], batch size: 53, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:13:39,986 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.623e+02 2.226e+02 2.685e+02 3.371e+02 7.491e+02, threshold=5.370e+02, percent-clipped=5.0 2023-05-02 17:14:00,587 INFO [train.py:904] (7/8) Epoch 29, batch 2900, loss[loss=0.1395, simple_loss=0.2333, pruned_loss=0.02282, over 17214.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2538, pruned_loss=0.03764, over 3331617.98 frames. ], batch size: 46, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:14:31,475 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287126.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:15:10,996 INFO [train.py:904] (7/8) Epoch 29, batch 2950, loss[loss=0.181, simple_loss=0.2565, pruned_loss=0.0528, over 16839.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2537, pruned_loss=0.03803, over 3327871.21 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:15:34,351 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7164, 4.1373, 2.9742, 2.4122, 2.6411, 2.6455, 4.3935, 3.4643], device='cuda:7'), covar=tensor([0.3136, 0.0617, 0.1927, 0.3162, 0.2990, 0.2269, 0.0459, 0.1597], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0278, 0.0317, 0.0331, 0.0309, 0.0282, 0.0309, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:15:59,395 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.092e+02 2.416e+02 3.171e+02 5.496e+02, threshold=4.831e+02, percent-clipped=1.0 2023-05-02 17:16:09,322 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9309, 5.2721, 5.4471, 5.0990, 5.2535, 5.8440, 5.3112, 4.9894], device='cuda:7'), covar=tensor([0.1234, 0.2098, 0.2711, 0.2407, 0.2495, 0.1059, 0.1663, 0.2618], device='cuda:7'), in_proj_covar=tensor([0.0442, 0.0658, 0.0731, 0.0537, 0.0718, 0.0751, 0.0565, 0.0712], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:16:20,172 INFO [train.py:904] (7/8) Epoch 29, batch 3000, loss[loss=0.1603, simple_loss=0.2494, pruned_loss=0.03564, over 16497.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2532, pruned_loss=0.03806, over 3327191.72 frames. ], batch size: 75, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:16:20,172 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 17:16:28,748 INFO [train.py:938] (7/8) Epoch 29, validation: loss=0.1336, simple_loss=0.2385, pruned_loss=0.01438, over 944034.00 frames. 2023-05-02 17:16:28,748 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 17:17:25,963 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287244.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:17:40,008 INFO [train.py:904] (7/8) Epoch 29, batch 3050, loss[loss=0.1605, simple_loss=0.2416, pruned_loss=0.03971, over 15915.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2535, pruned_loss=0.03826, over 3325337.59 frames. ], batch size: 35, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:18:29,495 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.465e+02 2.177e+02 2.556e+02 3.004e+02 5.914e+02, threshold=5.112e+02, percent-clipped=3.0 2023-05-02 17:18:34,412 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287292.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:18:49,727 INFO [train.py:904] (7/8) Epoch 29, batch 3100, loss[loss=0.1806, simple_loss=0.2644, pruned_loss=0.04843, over 16397.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.253, pruned_loss=0.03891, over 3311681.61 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:18:56,524 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 17:19:22,332 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:20:00,332 INFO [train.py:904] (7/8) Epoch 29, batch 3150, loss[loss=0.1841, simple_loss=0.2589, pruned_loss=0.05466, over 16763.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.252, pruned_loss=0.03867, over 3312962.36 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:20:30,290 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287375.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:20:49,325 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.202e+02 2.547e+02 3.057e+02 5.331e+02, threshold=5.094e+02, percent-clipped=1.0 2023-05-02 17:21:10,106 INFO [train.py:904] (7/8) Epoch 29, batch 3200, loss[loss=0.1473, simple_loss=0.2435, pruned_loss=0.02557, over 17098.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2506, pruned_loss=0.03785, over 3318579.51 frames. ], batch size: 47, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:21:20,141 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287411.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:21:39,839 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-02 17:21:40,302 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=287426.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:21:58,604 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1305, 5.0827, 4.8650, 3.9476, 4.9636, 1.7892, 4.6205, 4.5411], device='cuda:7'), covar=tensor([0.0143, 0.0133, 0.0302, 0.0655, 0.0177, 0.3450, 0.0216, 0.0386], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0180, 0.0219, 0.0190, 0.0197, 0.0223, 0.0207, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:22:06,661 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4435, 4.0418, 4.4881, 2.5746, 4.7065, 4.7508, 3.5826, 3.8103], device='cuda:7'), covar=tensor([0.0608, 0.0272, 0.0236, 0.1033, 0.0104, 0.0200, 0.0399, 0.0380], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0103, 0.0139, 0.0087, 0.0133, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:22:18,333 INFO [train.py:904] (7/8) Epoch 29, batch 3250, loss[loss=0.1728, simple_loss=0.2561, pruned_loss=0.04473, over 16723.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.252, pruned_loss=0.03786, over 3318011.95 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:22:23,006 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9037, 3.6067, 4.0558, 2.1936, 4.1792, 4.2168, 3.2274, 3.1987], device='cuda:7'), covar=tensor([0.0734, 0.0304, 0.0233, 0.1196, 0.0125, 0.0253, 0.0445, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0103, 0.0139, 0.0087, 0.0133, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:22:43,521 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287472.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:22:45,745 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=287474.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:23:05,923 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.139e+02 2.558e+02 3.030e+02 4.692e+02, threshold=5.116e+02, percent-clipped=0.0 2023-05-02 17:23:26,872 INFO [train.py:904] (7/8) Epoch 29, batch 3300, loss[loss=0.1604, simple_loss=0.2481, pruned_loss=0.03637, over 15890.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.253, pruned_loss=0.03823, over 3317577.53 frames. ], batch size: 35, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:24:11,423 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287536.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:24:16,924 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7742, 2.7884, 2.5883, 4.9025, 3.7763, 4.1836, 1.7059, 3.0433], device='cuda:7'), covar=tensor([0.1399, 0.0859, 0.1287, 0.0244, 0.0236, 0.0444, 0.1711, 0.0840], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0208, 0.0208, 0.0221, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:24:34,315 INFO [train.py:904] (7/8) Epoch 29, batch 3350, loss[loss=0.1487, simple_loss=0.2456, pruned_loss=0.02586, over 17137.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2534, pruned_loss=0.03802, over 3319045.61 frames. ], batch size: 49, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:24:46,636 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287562.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:25:14,583 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287583.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:25:21,671 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.115e+02 2.472e+02 3.004e+02 5.841e+02, threshold=4.943e+02, percent-clipped=1.0 2023-05-02 17:25:33,279 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3570, 5.3212, 5.1927, 4.6306, 4.8002, 5.2489, 5.1586, 4.8087], device='cuda:7'), covar=tensor([0.0586, 0.0483, 0.0332, 0.0417, 0.1145, 0.0512, 0.0373, 0.0871], device='cuda:7'), in_proj_covar=tensor([0.0328, 0.0495, 0.0384, 0.0386, 0.0379, 0.0443, 0.0263, 0.0459], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:25:33,339 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287597.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:25:43,315 INFO [train.py:904] (7/8) Epoch 29, batch 3400, loss[loss=0.1654, simple_loss=0.247, pruned_loss=0.04195, over 15508.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.254, pruned_loss=0.03812, over 3308767.19 frames. ], batch size: 190, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:26:10,464 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287623.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:26:40,317 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287644.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:26:53,543 INFO [train.py:904] (7/8) Epoch 29, batch 3450, loss[loss=0.1597, simple_loss=0.248, pruned_loss=0.03569, over 16743.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2526, pruned_loss=0.03794, over 3301559.26 frames. ], batch size: 124, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:27:43,292 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.405e+02 1.986e+02 2.391e+02 2.898e+02 4.066e+02, threshold=4.783e+02, percent-clipped=0.0 2023-05-02 17:28:04,238 INFO [train.py:904] (7/8) Epoch 29, batch 3500, loss[loss=0.1617, simple_loss=0.2533, pruned_loss=0.03501, over 16694.00 frames. ], tot_loss[loss=0.1637, simple_loss=0.2516, pruned_loss=0.03786, over 3303228.62 frames. ], batch size: 62, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:28:18,018 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-05-02 17:28:26,793 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0578, 4.9194, 4.9268, 4.5163, 4.6144, 4.9800, 4.7868, 4.6649], device='cuda:7'), covar=tensor([0.0766, 0.0936, 0.0385, 0.0409, 0.1048, 0.0584, 0.0504, 0.0791], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0498, 0.0386, 0.0388, 0.0382, 0.0446, 0.0265, 0.0461], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:28:26,825 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287720.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:29:14,573 INFO [train.py:904] (7/8) Epoch 29, batch 3550, loss[loss=0.1327, simple_loss=0.2206, pruned_loss=0.0224, over 16779.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2504, pruned_loss=0.03732, over 3317707.89 frames. ], batch size: 39, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:29:33,159 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287767.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:29:42,464 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=287773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:29:53,966 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287781.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 17:30:07,191 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 2.118e+02 2.537e+02 3.010e+02 5.172e+02, threshold=5.074e+02, percent-clipped=1.0 2023-05-02 17:30:26,894 INFO [train.py:904] (7/8) Epoch 29, batch 3600, loss[loss=0.1453, simple_loss=0.2436, pruned_loss=0.02351, over 17207.00 frames. ], tot_loss[loss=0.1614, simple_loss=0.2491, pruned_loss=0.03683, over 3314858.84 frames. ], batch size: 46, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:31:11,563 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=287834.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:31:40,155 INFO [train.py:904] (7/8) Epoch 29, batch 3650, loss[loss=0.1723, simple_loss=0.2465, pruned_loss=0.04907, over 16733.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2479, pruned_loss=0.03792, over 3312951.10 frames. ], batch size: 134, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:32:35,503 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.232e+02 2.562e+02 3.214e+02 7.635e+02, threshold=5.123e+02, percent-clipped=1.0 2023-05-02 17:32:38,236 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287892.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:32:47,999 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4694, 3.4984, 2.3513, 3.7411, 2.8654, 3.6768, 2.4103, 2.8678], device='cuda:7'), covar=tensor([0.0340, 0.0489, 0.1531, 0.0419, 0.0804, 0.0911, 0.1462, 0.0819], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0186, 0.0201, 0.0181, 0.0184, 0.0229, 0.0209, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:32:57,084 INFO [train.py:904] (7/8) Epoch 29, batch 3700, loss[loss=0.1707, simple_loss=0.2416, pruned_loss=0.04992, over 16761.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2463, pruned_loss=0.03929, over 3310508.67 frames. ], batch size: 83, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:33:18,788 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287918.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:33:46,006 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8386, 4.7441, 4.7285, 4.4113, 4.4354, 4.7573, 4.5619, 4.5442], device='cuda:7'), covar=tensor([0.0674, 0.0747, 0.0341, 0.0361, 0.0923, 0.0544, 0.0481, 0.0699], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0497, 0.0385, 0.0388, 0.0381, 0.0445, 0.0264, 0.0460], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:33:48,748 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=287939.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:33:58,071 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-05-02 17:34:09,993 INFO [train.py:904] (7/8) Epoch 29, batch 3750, loss[loss=0.1779, simple_loss=0.2518, pruned_loss=0.05201, over 16859.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2474, pruned_loss=0.04056, over 3308211.81 frames. ], batch size: 116, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:34:50,116 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3379, 3.1750, 2.8029, 5.2976, 4.1848, 4.4811, 2.0541, 3.3567], device='cuda:7'), covar=tensor([0.1150, 0.0764, 0.1312, 0.0117, 0.0397, 0.0379, 0.1536, 0.0901], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0182, 0.0202, 0.0209, 0.0208, 0.0221, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:35:05,549 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.186e+02 2.475e+02 3.056e+02 8.739e+02, threshold=4.949e+02, percent-clipped=4.0 2023-05-02 17:35:29,292 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288003.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:35:30,038 INFO [train.py:904] (7/8) Epoch 29, batch 3800, loss[loss=0.1902, simple_loss=0.2648, pruned_loss=0.05777, over 16945.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2486, pruned_loss=0.04161, over 3285675.25 frames. ], batch size: 109, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:35:37,627 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3479, 4.4067, 4.6861, 4.6635, 4.7397, 4.4233, 4.4537, 4.3127], device='cuda:7'), covar=tensor([0.0409, 0.0699, 0.0432, 0.0411, 0.0565, 0.0492, 0.0834, 0.0846], device='cuda:7'), in_proj_covar=tensor([0.0454, 0.0515, 0.0496, 0.0456, 0.0545, 0.0523, 0.0601, 0.0420], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 17:36:40,377 INFO [train.py:904] (7/8) Epoch 29, batch 3850, loss[loss=0.1565, simple_loss=0.2415, pruned_loss=0.03571, over 16466.00 frames. ], tot_loss[loss=0.1671, simple_loss=0.2496, pruned_loss=0.0423, over 3280706.30 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:36:55,460 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288064.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:36:59,626 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288067.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:37:12,699 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288076.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 17:37:15,300 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2672, 2.7362, 2.2464, 2.5243, 3.0401, 2.7911, 3.1081, 3.2081], device='cuda:7'), covar=tensor([0.0242, 0.0442, 0.0615, 0.0488, 0.0294, 0.0381, 0.0284, 0.0297], device='cuda:7'), in_proj_covar=tensor([0.0240, 0.0250, 0.0239, 0.0240, 0.0252, 0.0250, 0.0248, 0.0250], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:37:30,812 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.131e+02 2.477e+02 2.774e+02 5.510e+02, threshold=4.954e+02, percent-clipped=1.0 2023-05-02 17:37:50,144 INFO [train.py:904] (7/8) Epoch 29, batch 3900, loss[loss=0.1586, simple_loss=0.2405, pruned_loss=0.03838, over 15658.00 frames. ], tot_loss[loss=0.1675, simple_loss=0.2493, pruned_loss=0.04281, over 3269825.77 frames. ], batch size: 191, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:37:53,650 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288106.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:38:06,535 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288115.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:38:27,303 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288129.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:39:00,641 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288152.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:39:03,116 INFO [train.py:904] (7/8) Epoch 29, batch 3950, loss[loss=0.1548, simple_loss=0.233, pruned_loss=0.0383, over 16516.00 frames. ], tot_loss[loss=0.1681, simple_loss=0.2492, pruned_loss=0.04349, over 3268909.25 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:39:22,362 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288167.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 17:39:55,370 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.270e+02 2.585e+02 3.079e+02 6.589e+02, threshold=5.170e+02, percent-clipped=2.0 2023-05-02 17:39:59,736 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:40:09,542 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8035, 2.7703, 2.4882, 4.3352, 3.3797, 4.0530, 1.6532, 2.9082], device='cuda:7'), covar=tensor([0.1382, 0.0752, 0.1271, 0.0198, 0.0219, 0.0394, 0.1695, 0.0891], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0201, 0.0208, 0.0207, 0.0220, 0.0211, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:40:16,251 INFO [train.py:904] (7/8) Epoch 29, batch 4000, loss[loss=0.1643, simple_loss=0.2466, pruned_loss=0.04104, over 17264.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2493, pruned_loss=0.04359, over 3272313.81 frames. ], batch size: 52, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:40:30,265 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288213.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:40:37,283 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288218.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:41:07,408 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288239.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:41:09,026 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288240.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:41:24,429 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9891, 3.0895, 3.5035, 2.1543, 3.0770, 2.2899, 3.5524, 3.5085], device='cuda:7'), covar=tensor([0.0231, 0.0944, 0.0577, 0.2163, 0.0840, 0.1033, 0.0490, 0.0835], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0174, 0.0172, 0.0158, 0.0149, 0.0134, 0.0147, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 17:41:29,253 INFO [train.py:904] (7/8) Epoch 29, batch 4050, loss[loss=0.1543, simple_loss=0.2411, pruned_loss=0.03378, over 16900.00 frames. ], tot_loss[loss=0.1679, simple_loss=0.2499, pruned_loss=0.04299, over 3280210.84 frames. ], batch size: 96, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:41:47,213 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288266.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:42:17,276 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288287.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:42:21,065 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.858e+02 2.084e+02 2.415e+02 5.689e+02, threshold=4.167e+02, percent-clipped=1.0 2023-05-02 17:42:37,666 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4832, 1.8404, 2.2437, 2.4836, 2.5782, 2.8223, 1.9446, 2.7235], device='cuda:7'), covar=tensor([0.0284, 0.0596, 0.0369, 0.0390, 0.0345, 0.0217, 0.0600, 0.0170], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0203, 0.0191, 0.0198, 0.0215, 0.0171, 0.0207, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 17:42:42,064 INFO [train.py:904] (7/8) Epoch 29, batch 4100, loss[loss=0.1969, simple_loss=0.2885, pruned_loss=0.05261, over 16689.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2518, pruned_loss=0.04279, over 3285052.27 frames. ], batch size: 134, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:43:17,878 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288327.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:43:45,802 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288345.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:43:50,677 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1829, 5.4662, 5.2072, 5.2964, 5.0198, 4.8852, 4.9204, 5.5917], device='cuda:7'), covar=tensor([0.1225, 0.0855, 0.1111, 0.0836, 0.0775, 0.0937, 0.1249, 0.0879], device='cuda:7'), in_proj_covar=tensor([0.0737, 0.0886, 0.0728, 0.0690, 0.0566, 0.0562, 0.0746, 0.0697], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:43:59,221 INFO [train.py:904] (7/8) Epoch 29, batch 4150, loss[loss=0.2384, simple_loss=0.3096, pruned_loss=0.08367, over 11627.00 frames. ], tot_loss[loss=0.1743, simple_loss=0.2588, pruned_loss=0.04488, over 3247754.45 frames. ], batch size: 246, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:44:00,586 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5433, 3.6381, 2.6624, 2.2097, 2.3398, 2.4241, 3.7853, 3.2060], device='cuda:7'), covar=tensor([0.3122, 0.0623, 0.2088, 0.2946, 0.2800, 0.2245, 0.0534, 0.1407], device='cuda:7'), in_proj_covar=tensor([0.0337, 0.0277, 0.0315, 0.0330, 0.0310, 0.0281, 0.0308, 0.0357], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:44:07,918 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288359.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:44:29,397 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4419, 3.0012, 2.6958, 2.3121, 2.2840, 2.3592, 2.9836, 2.8689], device='cuda:7'), covar=tensor([0.2356, 0.0643, 0.1511, 0.2521, 0.2389, 0.2055, 0.0502, 0.1249], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0276, 0.0315, 0.0329, 0.0310, 0.0280, 0.0307, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:44:33,957 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288376.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:44:33,983 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1660, 3.8357, 3.7500, 2.3711, 3.3568, 3.8381, 3.4208, 2.2777], device='cuda:7'), covar=tensor([0.0622, 0.0058, 0.0075, 0.0493, 0.0140, 0.0120, 0.0129, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0137, 0.0104, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 17:44:40,232 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2023-05-02 17:44:51,091 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288388.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:44:52,909 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.550e+02 2.295e+02 2.546e+02 3.002e+02 5.285e+02, threshold=5.091e+02, percent-clipped=5.0 2023-05-02 17:45:02,570 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288396.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:45:13,939 INFO [train.py:904] (7/8) Epoch 29, batch 4200, loss[loss=0.1952, simple_loss=0.284, pruned_loss=0.0532, over 11510.00 frames. ], tot_loss[loss=0.1797, simple_loss=0.266, pruned_loss=0.04671, over 3220122.44 frames. ], batch size: 247, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:45:18,706 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288406.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:45:45,981 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288424.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:45:53,567 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288429.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:46:30,289 INFO [train.py:904] (7/8) Epoch 29, batch 4250, loss[loss=0.1749, simple_loss=0.2692, pruned_loss=0.04027, over 16330.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2692, pruned_loss=0.04662, over 3195005.46 frames. ], batch size: 35, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:46:36,726 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288457.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:46:43,224 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288462.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:46:47,869 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3796, 3.5211, 3.6500, 3.6358, 3.6616, 3.4738, 3.5006, 3.5297], device='cuda:7'), covar=tensor([0.0420, 0.0639, 0.0534, 0.0481, 0.0494, 0.0541, 0.0919, 0.0592], device='cuda:7'), in_proj_covar=tensor([0.0446, 0.0506, 0.0488, 0.0448, 0.0535, 0.0513, 0.0593, 0.0414], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 17:46:57,249 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5332, 3.3565, 3.7523, 1.9148, 3.9719, 3.9543, 2.9319, 2.8712], device='cuda:7'), covar=tensor([0.0902, 0.0322, 0.0245, 0.1291, 0.0099, 0.0182, 0.0473, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0112, 0.0102, 0.0139, 0.0087, 0.0132, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 17:47:05,785 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288477.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:47:24,433 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.453e+02 2.222e+02 2.492e+02 3.034e+02 5.623e+02, threshold=4.984e+02, percent-clipped=1.0 2023-05-02 17:47:45,989 INFO [train.py:904] (7/8) Epoch 29, batch 4300, loss[loss=0.1766, simple_loss=0.2734, pruned_loss=0.0399, over 16892.00 frames. ], tot_loss[loss=0.1802, simple_loss=0.2694, pruned_loss=0.04544, over 3192836.13 frames. ], batch size: 96, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:47:46,957 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 17:47:51,256 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288508.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:48:26,817 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-05-02 17:48:27,959 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=5.01 vs. limit=5.0 2023-05-02 17:48:59,354 INFO [train.py:904] (7/8) Epoch 29, batch 4350, loss[loss=0.1995, simple_loss=0.2968, pruned_loss=0.05106, over 16683.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2726, pruned_loss=0.04638, over 3203836.63 frames. ], batch size: 89, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:49:53,411 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.193e+02 2.447e+02 2.913e+02 5.150e+02, threshold=4.894e+02, percent-clipped=1.0 2023-05-02 17:50:13,850 INFO [train.py:904] (7/8) Epoch 29, batch 4400, loss[loss=0.1734, simple_loss=0.2761, pruned_loss=0.03532, over 16918.00 frames. ], tot_loss[loss=0.1846, simple_loss=0.2745, pruned_loss=0.04731, over 3205060.39 frames. ], batch size: 96, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:50:22,310 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288609.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:51:26,309 INFO [train.py:904] (7/8) Epoch 29, batch 4450, loss[loss=0.2035, simple_loss=0.2966, pruned_loss=0.05521, over 16988.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.2777, pruned_loss=0.04863, over 3208037.29 frames. ], batch size: 41, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:51:34,139 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288659.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:51:49,614 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288670.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:52:08,451 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288683.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:52:18,787 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.398e+02 1.892e+02 2.239e+02 2.604e+02 5.228e+02, threshold=4.478e+02, percent-clipped=1.0 2023-05-02 17:52:34,525 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288701.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:52:37,603 INFO [train.py:904] (7/8) Epoch 29, batch 4500, loss[loss=0.1834, simple_loss=0.2714, pruned_loss=0.04767, over 17024.00 frames. ], tot_loss[loss=0.1893, simple_loss=0.2788, pruned_loss=0.04989, over 3208076.25 frames. ], batch size: 55, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:52:43,746 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288707.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:53:49,136 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288752.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:53:51,865 INFO [train.py:904] (7/8) Epoch 29, batch 4550, loss[loss=0.1877, simple_loss=0.2805, pruned_loss=0.04741, over 17139.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2793, pruned_loss=0.05039, over 3225435.78 frames. ], batch size: 47, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:53:58,830 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5489, 4.2460, 4.1731, 2.8522, 3.7366, 4.2077, 3.7064, 2.6226], device='cuda:7'), covar=tensor([0.0557, 0.0038, 0.0053, 0.0422, 0.0105, 0.0091, 0.0102, 0.0413], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 17:54:03,084 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288762.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:54:16,467 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6445, 4.3871, 4.2750, 3.0673, 3.8138, 4.2744, 3.8014, 2.5785], device='cuda:7'), covar=tensor([0.0518, 0.0032, 0.0051, 0.0378, 0.0108, 0.0108, 0.0101, 0.0437], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0092, 0.0136, 0.0103, 0.0117, 0.0099, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 17:54:44,112 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 1.747e+02 2.008e+02 2.421e+02 6.234e+02, threshold=4.017e+02, percent-clipped=2.0 2023-05-02 17:54:54,487 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=288797.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:55:04,706 INFO [train.py:904] (7/8) Epoch 29, batch 4600, loss[loss=0.1973, simple_loss=0.2823, pruned_loss=0.05614, over 16396.00 frames. ], tot_loss[loss=0.1907, simple_loss=0.2801, pruned_loss=0.05067, over 3232759.25 frames. ], batch size: 146, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:55:12,238 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288808.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 17:55:14,478 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288810.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:56:18,392 INFO [train.py:904] (7/8) Epoch 29, batch 4650, loss[loss=0.2055, simple_loss=0.2949, pruned_loss=0.05803, over 16398.00 frames. ], tot_loss[loss=0.1906, simple_loss=0.2792, pruned_loss=0.05098, over 3218905.68 frames. ], batch size: 146, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 17:56:21,002 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=288856.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:56:24,881 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=288858.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:57:10,671 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.194e+02 1.769e+02 2.025e+02 2.417e+02 4.191e+02, threshold=4.050e+02, percent-clipped=1.0 2023-05-02 17:57:23,810 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0080, 2.2765, 2.3445, 3.6322, 2.1514, 2.5849, 2.3669, 2.4183], device='cuda:7'), covar=tensor([0.1672, 0.3562, 0.3189, 0.0702, 0.4338, 0.2595, 0.3602, 0.3509], device='cuda:7'), in_proj_covar=tensor([0.0425, 0.0480, 0.0389, 0.0340, 0.0449, 0.0552, 0.0452, 0.0562], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:57:29,738 INFO [train.py:904] (7/8) Epoch 29, batch 4700, loss[loss=0.1844, simple_loss=0.267, pruned_loss=0.05088, over 16365.00 frames. ], tot_loss[loss=0.1878, simple_loss=0.2761, pruned_loss=0.04972, over 3232991.35 frames. ], batch size: 35, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:57:42,527 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8714, 4.9708, 5.2166, 5.2007, 5.2417, 4.9548, 4.8999, 4.6701], device='cuda:7'), covar=tensor([0.0311, 0.0390, 0.0344, 0.0322, 0.0404, 0.0321, 0.0860, 0.0473], device='cuda:7'), in_proj_covar=tensor([0.0436, 0.0493, 0.0477, 0.0437, 0.0523, 0.0500, 0.0580, 0.0405], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 17:58:00,043 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7381, 4.9158, 5.0996, 4.8438, 4.8893, 5.4758, 4.9782, 4.6647], device='cuda:7'), covar=tensor([0.1069, 0.1710, 0.1957, 0.1971, 0.2489, 0.0891, 0.1414, 0.2240], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0638, 0.0704, 0.0517, 0.0693, 0.0727, 0.0547, 0.0689], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:58:31,323 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8717, 4.9671, 4.7991, 4.4074, 4.3856, 4.8598, 4.6782, 4.6048], device='cuda:7'), covar=tensor([0.0664, 0.0592, 0.0313, 0.0340, 0.1141, 0.0631, 0.0406, 0.0569], device='cuda:7'), in_proj_covar=tensor([0.0316, 0.0479, 0.0371, 0.0373, 0.0367, 0.0428, 0.0254, 0.0443], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 17:58:41,606 INFO [train.py:904] (7/8) Epoch 29, batch 4750, loss[loss=0.1832, simple_loss=0.2707, pruned_loss=0.04782, over 15411.00 frames. ], tot_loss[loss=0.1843, simple_loss=0.2727, pruned_loss=0.04797, over 3219048.89 frames. ], batch size: 190, lr: 2.32e-03, grad_scale: 4.0 2023-05-02 17:58:52,601 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5220, 3.5435, 2.6185, 2.2892, 2.3593, 2.4220, 3.7724, 3.0594], device='cuda:7'), covar=tensor([0.3126, 0.0708, 0.2157, 0.2831, 0.2766, 0.2302, 0.0516, 0.1609], device='cuda:7'), in_proj_covar=tensor([0.0334, 0.0274, 0.0313, 0.0327, 0.0307, 0.0278, 0.0305, 0.0354], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 17:58:57,735 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=288965.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 17:59:23,615 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=288983.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:59:35,280 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 1.780e+02 1.987e+02 2.347e+02 4.131e+02, threshold=3.973e+02, percent-clipped=1.0 2023-05-02 17:59:50,966 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289001.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 17:59:54,504 INFO [train.py:904] (7/8) Epoch 29, batch 4800, loss[loss=0.1783, simple_loss=0.2587, pruned_loss=0.04897, over 11742.00 frames. ], tot_loss[loss=0.1807, simple_loss=0.2692, pruned_loss=0.04607, over 3214453.05 frames. ], batch size: 248, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:00:13,079 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4703, 3.2974, 3.6237, 1.8217, 3.8148, 3.7919, 2.9217, 2.7609], device='cuda:7'), covar=tensor([0.0836, 0.0285, 0.0189, 0.1332, 0.0085, 0.0171, 0.0448, 0.0565], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0140, 0.0088, 0.0133, 0.0131, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:00:37,064 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289031.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:01:00,404 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289046.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:01:05,206 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289049.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:01:05,797 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-05-02 18:01:09,497 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289052.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:01:12,137 INFO [train.py:904] (7/8) Epoch 29, batch 4850, loss[loss=0.1965, simple_loss=0.2841, pruned_loss=0.05441, over 11891.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2693, pruned_loss=0.04525, over 3189842.71 frames. ], batch size: 247, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:02:08,682 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 1.829e+02 2.187e+02 2.631e+02 3.612e+02, threshold=4.373e+02, percent-clipped=0.0 2023-05-02 18:02:23,000 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289100.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:02:28,916 INFO [train.py:904] (7/8) Epoch 29, batch 4900, loss[loss=0.1606, simple_loss=0.2539, pruned_loss=0.03367, over 16508.00 frames. ], tot_loss[loss=0.1777, simple_loss=0.2682, pruned_loss=0.04358, over 3188743.23 frames. ], batch size: 68, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:02:33,585 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289107.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 18:03:16,475 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289136.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:03:24,720 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9055, 2.4592, 2.0734, 2.1894, 2.7663, 2.4180, 2.5294, 2.9186], device='cuda:7'), covar=tensor([0.0232, 0.0516, 0.0646, 0.0627, 0.0335, 0.0492, 0.0240, 0.0324], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0247, 0.0236, 0.0237, 0.0248, 0.0246, 0.0244, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:03:42,187 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289153.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:03:43,046 INFO [train.py:904] (7/8) Epoch 29, batch 4950, loss[loss=0.1787, simple_loss=0.2647, pruned_loss=0.04633, over 17219.00 frames. ], tot_loss[loss=0.1765, simple_loss=0.2675, pruned_loss=0.04277, over 3206704.55 frames. ], batch size: 45, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:04:38,082 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.471e+02 2.040e+02 2.385e+02 2.853e+02 5.286e+02, threshold=4.771e+02, percent-clipped=2.0 2023-05-02 18:04:46,926 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289197.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:04:56,968 INFO [train.py:904] (7/8) Epoch 29, batch 5000, loss[loss=0.1895, simple_loss=0.2858, pruned_loss=0.04655, over 16782.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2692, pruned_loss=0.04287, over 3211349.78 frames. ], batch size: 89, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:05:00,034 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4039, 4.5919, 4.4705, 2.7964, 3.9095, 4.4630, 3.8806, 2.2112], device='cuda:7'), covar=tensor([0.0683, 0.0039, 0.0051, 0.0497, 0.0118, 0.0089, 0.0114, 0.0579], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0092, 0.0136, 0.0103, 0.0116, 0.0099, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 18:05:29,448 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 18:06:09,978 INFO [train.py:904] (7/8) Epoch 29, batch 5050, loss[loss=0.1641, simple_loss=0.2544, pruned_loss=0.03688, over 17142.00 frames. ], tot_loss[loss=0.1775, simple_loss=0.2698, pruned_loss=0.04257, over 3230540.32 frames. ], batch size: 47, lr: 2.32e-03, grad_scale: 8.0 2023-05-02 18:06:26,928 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289265.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:07:03,095 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 1.941e+02 2.202e+02 2.540e+02 3.761e+02, threshold=4.404e+02, percent-clipped=0.0 2023-05-02 18:07:22,117 INFO [train.py:904] (7/8) Epoch 29, batch 5100, loss[loss=0.1775, simple_loss=0.2671, pruned_loss=0.04391, over 16749.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2682, pruned_loss=0.04195, over 3222265.56 frames. ], batch size: 124, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:07:34,913 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289313.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:07:57,022 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5881, 3.6721, 3.4549, 3.1575, 3.2702, 3.5440, 3.3546, 3.3892], device='cuda:7'), covar=tensor([0.0571, 0.0487, 0.0310, 0.0293, 0.0586, 0.0487, 0.1263, 0.0469], device='cuda:7'), in_proj_covar=tensor([0.0313, 0.0475, 0.0368, 0.0369, 0.0365, 0.0426, 0.0253, 0.0440], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:07:57,248 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 18:08:36,369 INFO [train.py:904] (7/8) Epoch 29, batch 5150, loss[loss=0.1547, simple_loss=0.2578, pruned_loss=0.02576, over 16827.00 frames. ], tot_loss[loss=0.1754, simple_loss=0.2681, pruned_loss=0.04131, over 3206783.56 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:09:29,043 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 1.894e+02 2.237e+02 2.654e+02 4.017e+02, threshold=4.473e+02, percent-clipped=0.0 2023-05-02 18:09:45,673 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289402.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 18:09:47,653 INFO [train.py:904] (7/8) Epoch 29, batch 5200, loss[loss=0.1607, simple_loss=0.2507, pruned_loss=0.03532, over 16841.00 frames. ], tot_loss[loss=0.1746, simple_loss=0.2672, pruned_loss=0.041, over 3192449.26 frames. ], batch size: 116, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:10:19,821 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.37 vs. limit=5.0 2023-05-02 18:10:38,466 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9979, 3.2610, 3.5001, 2.0651, 3.0114, 2.4124, 3.5131, 3.5206], device='cuda:7'), covar=tensor([0.0239, 0.0842, 0.0650, 0.2093, 0.0856, 0.0960, 0.0557, 0.0836], device='cuda:7'), in_proj_covar=tensor([0.0163, 0.0173, 0.0173, 0.0158, 0.0149, 0.0134, 0.0147, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 18:10:59,562 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289453.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:11:00,511 INFO [train.py:904] (7/8) Epoch 29, batch 5250, loss[loss=0.1711, simple_loss=0.2672, pruned_loss=0.03749, over 16243.00 frames. ], tot_loss[loss=0.1729, simple_loss=0.2646, pruned_loss=0.0406, over 3205894.90 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:11:55,945 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 1.894e+02 2.157e+02 2.543e+02 4.571e+02, threshold=4.315e+02, percent-clipped=1.0 2023-05-02 18:11:58,160 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289492.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:12:10,847 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289501.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:12:15,018 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289503.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:12:15,674 INFO [train.py:904] (7/8) Epoch 29, batch 5300, loss[loss=0.1789, simple_loss=0.2639, pruned_loss=0.047, over 12267.00 frames. ], tot_loss[loss=0.17, simple_loss=0.2607, pruned_loss=0.03962, over 3212728.65 frames. ], batch size: 247, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:13:16,863 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 18:13:23,553 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0772, 2.3026, 2.2883, 3.7748, 2.1240, 2.6534, 2.3830, 2.4327], device='cuda:7'), covar=tensor([0.1522, 0.3624, 0.3154, 0.0604, 0.4137, 0.2526, 0.3503, 0.3206], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0478, 0.0387, 0.0339, 0.0446, 0.0550, 0.0450, 0.0559], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:13:26,900 INFO [train.py:904] (7/8) Epoch 29, batch 5350, loss[loss=0.1855, simple_loss=0.272, pruned_loss=0.04956, over 16082.00 frames. ], tot_loss[loss=0.1687, simple_loss=0.2592, pruned_loss=0.03915, over 3209287.40 frames. ], batch size: 35, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:13:42,103 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289564.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:13:53,775 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 18:14:21,419 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 1.853e+02 2.169e+02 2.585e+02 3.843e+02, threshold=4.338e+02, percent-clipped=0.0 2023-05-02 18:14:41,104 INFO [train.py:904] (7/8) Epoch 29, batch 5400, loss[loss=0.1778, simple_loss=0.2701, pruned_loss=0.04272, over 16433.00 frames. ], tot_loss[loss=0.1711, simple_loss=0.2621, pruned_loss=0.04011, over 3200092.31 frames. ], batch size: 68, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:15:34,753 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.85 vs. limit=5.0 2023-05-02 18:15:42,268 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9955, 4.0094, 4.2848, 4.2569, 4.2984, 4.0648, 4.0511, 4.0479], device='cuda:7'), covar=tensor([0.0390, 0.0806, 0.0476, 0.0450, 0.0525, 0.0520, 0.0998, 0.0554], device='cuda:7'), in_proj_covar=tensor([0.0437, 0.0494, 0.0479, 0.0438, 0.0525, 0.0502, 0.0582, 0.0404], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 18:15:57,442 INFO [train.py:904] (7/8) Epoch 29, batch 5450, loss[loss=0.2307, simple_loss=0.3147, pruned_loss=0.07336, over 15267.00 frames. ], tot_loss[loss=0.1737, simple_loss=0.2649, pruned_loss=0.04124, over 3206928.33 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:16:07,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6726, 2.4571, 2.4066, 3.5829, 2.2966, 3.7080, 1.5443, 2.8133], device='cuda:7'), covar=tensor([0.1393, 0.0839, 0.1228, 0.0228, 0.0145, 0.0394, 0.1764, 0.0785], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0181, 0.0200, 0.0205, 0.0206, 0.0218, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:16:47,880 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5663, 3.5425, 3.5197, 2.7709, 3.4054, 2.0689, 3.2339, 2.7748], device='cuda:7'), covar=tensor([0.0204, 0.0205, 0.0220, 0.0271, 0.0145, 0.2686, 0.0166, 0.0315], device='cuda:7'), in_proj_covar=tensor([0.0185, 0.0179, 0.0218, 0.0190, 0.0195, 0.0222, 0.0206, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:16:53,638 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 2.272e+02 2.915e+02 3.734e+02 7.781e+02, threshold=5.831e+02, percent-clipped=14.0 2023-05-02 18:16:58,490 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8217, 3.6099, 4.2325, 2.0697, 4.4471, 4.3660, 3.0477, 3.3683], device='cuda:7'), covar=tensor([0.0803, 0.0318, 0.0211, 0.1207, 0.0067, 0.0187, 0.0498, 0.0431], device='cuda:7'), in_proj_covar=tensor([0.0149, 0.0111, 0.0103, 0.0139, 0.0087, 0.0131, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:17:12,129 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289702.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 18:17:14,649 INFO [train.py:904] (7/8) Epoch 29, batch 5500, loss[loss=0.2298, simple_loss=0.3321, pruned_loss=0.06372, over 16754.00 frames. ], tot_loss[loss=0.1812, simple_loss=0.2722, pruned_loss=0.04513, over 3195215.64 frames. ], batch size: 83, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:17:31,310 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9379, 4.2018, 4.0403, 4.0694, 3.7303, 3.8344, 3.8636, 4.2072], device='cuda:7'), covar=tensor([0.1233, 0.0936, 0.1035, 0.0890, 0.0839, 0.1809, 0.0943, 0.1042], device='cuda:7'), in_proj_covar=tensor([0.0721, 0.0866, 0.0714, 0.0671, 0.0553, 0.0550, 0.0726, 0.0682], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:18:26,454 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289750.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:18:31,719 INFO [train.py:904] (7/8) Epoch 29, batch 5550, loss[loss=0.2289, simple_loss=0.3117, pruned_loss=0.07308, over 15430.00 frames. ], tot_loss[loss=0.1885, simple_loss=0.2786, pruned_loss=0.04918, over 3180273.48 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:18:36,559 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 18:19:30,766 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.953e+02 3.616e+02 4.343e+02 9.960e+02, threshold=7.231e+02, percent-clipped=12.0 2023-05-02 18:19:34,073 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=289792.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:19:46,080 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 18:19:53,196 INFO [train.py:904] (7/8) Epoch 29, batch 5600, loss[loss=0.2717, simple_loss=0.33, pruned_loss=0.1067, over 11021.00 frames. ], tot_loss[loss=0.196, simple_loss=0.284, pruned_loss=0.054, over 3112181.95 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:20:54,481 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=289840.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:20:54,576 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3095, 3.4487, 3.5578, 3.5395, 3.5703, 3.4227, 3.4353, 3.4767], device='cuda:7'), covar=tensor([0.0479, 0.0903, 0.0613, 0.0579, 0.0630, 0.0780, 0.0898, 0.0725], device='cuda:7'), in_proj_covar=tensor([0.0439, 0.0497, 0.0482, 0.0441, 0.0527, 0.0504, 0.0585, 0.0407], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 18:21:16,891 INFO [train.py:904] (7/8) Epoch 29, batch 5650, loss[loss=0.2112, simple_loss=0.3019, pruned_loss=0.06021, over 16342.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2889, pruned_loss=0.05791, over 3084835.52 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:21:25,654 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=289859.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:22:15,728 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.920e+02 3.088e+02 3.696e+02 4.617e+02 9.164e+02, threshold=7.392e+02, percent-clipped=2.0 2023-05-02 18:22:35,926 INFO [train.py:904] (7/8) Epoch 29, batch 5700, loss[loss=0.2332, simple_loss=0.3028, pruned_loss=0.08178, over 11549.00 frames. ], tot_loss[loss=0.2034, simple_loss=0.2895, pruned_loss=0.05871, over 3066443.19 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:23:02,345 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289920.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:23:06,487 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3754, 2.9284, 2.6715, 2.3235, 2.2592, 2.3523, 3.0342, 2.8474], device='cuda:7'), covar=tensor([0.2558, 0.0795, 0.1681, 0.2539, 0.2289, 0.2220, 0.0502, 0.1409], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0275, 0.0314, 0.0329, 0.0307, 0.0278, 0.0306, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 18:23:55,708 INFO [train.py:904] (7/8) Epoch 29, batch 5750, loss[loss=0.1939, simple_loss=0.2933, pruned_loss=0.0472, over 16723.00 frames. ], tot_loss[loss=0.206, simple_loss=0.292, pruned_loss=0.05995, over 3066157.43 frames. ], batch size: 134, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:24:41,645 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=289981.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 18:24:56,850 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.660e+02 3.372e+02 4.165e+02 9.302e+02, threshold=6.745e+02, percent-clipped=2.0 2023-05-02 18:25:08,747 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=289998.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:25:20,974 INFO [train.py:904] (7/8) Epoch 29, batch 5800, loss[loss=0.1878, simple_loss=0.277, pruned_loss=0.04925, over 16523.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2916, pruned_loss=0.05897, over 3056574.46 frames. ], batch size: 75, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:26:34,376 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=290050.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:26:39,160 INFO [train.py:904] (7/8) Epoch 29, batch 5850, loss[loss=0.2248, simple_loss=0.2965, pruned_loss=0.07651, over 11809.00 frames. ], tot_loss[loss=0.203, simple_loss=0.2902, pruned_loss=0.05788, over 3054622.31 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:26:48,290 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=290059.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:27:38,221 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-02 18:27:39,838 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.628e+02 3.224e+02 3.685e+02 6.132e+02, threshold=6.447e+02, percent-clipped=0.0 2023-05-02 18:28:01,364 INFO [train.py:904] (7/8) Epoch 29, batch 5900, loss[loss=0.1839, simple_loss=0.2703, pruned_loss=0.04871, over 16706.00 frames. ], tot_loss[loss=0.2021, simple_loss=0.2894, pruned_loss=0.0574, over 3073623.65 frames. ], batch size: 134, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:28:15,126 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=290111.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:28:47,596 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1352, 3.5277, 3.6463, 2.3311, 3.2844, 3.6379, 3.4418, 2.0318], device='cuda:7'), covar=tensor([0.0631, 0.0099, 0.0076, 0.0504, 0.0141, 0.0140, 0.0113, 0.0559], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0137, 0.0104, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 18:29:20,974 INFO [train.py:904] (7/8) Epoch 29, batch 5950, loss[loss=0.1865, simple_loss=0.2816, pruned_loss=0.04572, over 16601.00 frames. ], tot_loss[loss=0.2014, simple_loss=0.2903, pruned_loss=0.05626, over 3085480.20 frames. ], batch size: 57, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:29:30,222 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290159.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:30:18,755 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 2.594e+02 3.044e+02 3.971e+02 8.061e+02, threshold=6.088e+02, percent-clipped=1.0 2023-05-02 18:30:32,124 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-05-02 18:30:40,233 INFO [train.py:904] (7/8) Epoch 29, batch 6000, loss[loss=0.1842, simple_loss=0.271, pruned_loss=0.0487, over 16824.00 frames. ], tot_loss[loss=0.2008, simple_loss=0.2896, pruned_loss=0.056, over 3077887.05 frames. ], batch size: 39, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:30:40,233 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 18:30:50,249 INFO [train.py:938] (7/8) Epoch 29, validation: loss=0.1475, simple_loss=0.2594, pruned_loss=0.01778, over 944034.00 frames. 2023-05-02 18:30:50,249 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 18:30:55,868 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290207.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:31:00,145 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-05-02 18:32:11,213 INFO [train.py:904] (7/8) Epoch 29, batch 6050, loss[loss=0.1774, simple_loss=0.2769, pruned_loss=0.03896, over 16831.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2872, pruned_loss=0.05435, over 3112051.32 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:32:20,097 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.24 vs. limit=5.0 2023-05-02 18:32:21,275 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9246, 2.2086, 2.5051, 3.1044, 2.2558, 2.4097, 2.4064, 2.3164], device='cuda:7'), covar=tensor([0.1445, 0.3434, 0.2470, 0.0763, 0.4130, 0.2369, 0.3079, 0.3378], device='cuda:7'), in_proj_covar=tensor([0.0422, 0.0476, 0.0386, 0.0337, 0.0446, 0.0546, 0.0448, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:32:41,980 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 18:32:42,701 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290276.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 18:32:45,350 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-02 18:33:04,446 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.589e+02 2.468e+02 2.905e+02 3.476e+02 8.072e+02, threshold=5.810e+02, percent-clipped=1.0 2023-05-02 18:33:29,601 INFO [train.py:904] (7/8) Epoch 29, batch 6100, loss[loss=0.2032, simple_loss=0.3006, pruned_loss=0.05286, over 16474.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.287, pruned_loss=0.05374, over 3121839.51 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:33:43,276 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1308, 4.3844, 4.0648, 3.8915, 3.5399, 4.3003, 3.9693, 3.9488], device='cuda:7'), covar=tensor([0.0922, 0.0618, 0.0467, 0.0431, 0.1539, 0.0547, 0.1053, 0.0728], device='cuda:7'), in_proj_covar=tensor([0.0311, 0.0471, 0.0365, 0.0366, 0.0361, 0.0422, 0.0251, 0.0437], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:33:53,070 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=290318.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:34:46,606 INFO [train.py:904] (7/8) Epoch 29, batch 6150, loss[loss=0.1692, simple_loss=0.265, pruned_loss=0.03675, over 16811.00 frames. ], tot_loss[loss=0.1965, simple_loss=0.2858, pruned_loss=0.05359, over 3122694.69 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:34:47,127 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290354.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:35:27,527 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=290379.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:35:44,982 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.708e+02 2.649e+02 3.165e+02 3.863e+02 6.871e+02, threshold=6.329e+02, percent-clipped=1.0 2023-05-02 18:36:04,353 INFO [train.py:904] (7/8) Epoch 29, batch 6200, loss[loss=0.2098, simple_loss=0.2809, pruned_loss=0.06932, over 11436.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2847, pruned_loss=0.05378, over 3102560.96 frames. ], batch size: 246, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:36:08,301 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290406.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:37:20,428 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2836, 3.0973, 3.4280, 1.8883, 3.5495, 3.5617, 2.8695, 2.7257], device='cuda:7'), covar=tensor([0.0883, 0.0304, 0.0201, 0.1215, 0.0100, 0.0224, 0.0439, 0.0546], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0139, 0.0088, 0.0133, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:37:21,100 INFO [train.py:904] (7/8) Epoch 29, batch 6250, loss[loss=0.1969, simple_loss=0.2914, pruned_loss=0.05118, over 16327.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2845, pruned_loss=0.05403, over 3111131.63 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:38:15,683 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.611e+02 2.626e+02 3.170e+02 3.673e+02 5.508e+02, threshold=6.341e+02, percent-clipped=0.0 2023-05-02 18:38:34,382 INFO [train.py:904] (7/8) Epoch 29, batch 6300, loss[loss=0.1961, simple_loss=0.2864, pruned_loss=0.05289, over 16933.00 frames. ], tot_loss[loss=0.195, simple_loss=0.2835, pruned_loss=0.05321, over 3122357.15 frames. ], batch size: 109, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:38:39,617 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6067, 2.5375, 1.9906, 2.6427, 2.1867, 2.7767, 2.1731, 2.3613], device='cuda:7'), covar=tensor([0.0316, 0.0379, 0.1233, 0.0295, 0.0641, 0.0560, 0.1153, 0.0579], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0180, 0.0195, 0.0173, 0.0179, 0.0220, 0.0203, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:39:45,205 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0697, 3.3678, 3.3763, 2.1269, 3.1455, 3.3938, 3.1984, 1.9001], device='cuda:7'), covar=tensor([0.0612, 0.0080, 0.0083, 0.0508, 0.0135, 0.0137, 0.0121, 0.0581], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0091, 0.0093, 0.0137, 0.0104, 0.0118, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 18:39:52,810 INFO [train.py:904] (7/8) Epoch 29, batch 6350, loss[loss=0.1859, simple_loss=0.2769, pruned_loss=0.04743, over 16809.00 frames. ], tot_loss[loss=0.1958, simple_loss=0.2838, pruned_loss=0.05393, over 3105765.91 frames. ], batch size: 83, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:40:28,375 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290576.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 18:40:37,355 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.78 vs. limit=5.0 2023-05-02 18:40:50,035 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.932e+02 2.799e+02 3.423e+02 4.274e+02 8.913e+02, threshold=6.847e+02, percent-clipped=6.0 2023-05-02 18:41:09,296 INFO [train.py:904] (7/8) Epoch 29, batch 6400, loss[loss=0.1789, simple_loss=0.2687, pruned_loss=0.04457, over 16852.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2844, pruned_loss=0.0548, over 3107927.17 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:41:39,270 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290624.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 18:42:23,302 INFO [train.py:904] (7/8) Epoch 29, batch 6450, loss[loss=0.2172, simple_loss=0.2853, pruned_loss=0.07454, over 11610.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2843, pruned_loss=0.05399, over 3114722.92 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:42:23,790 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290654.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:42:26,205 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0049, 4.7109, 4.5816, 3.1880, 4.0699, 4.6631, 4.1564, 2.5407], device='cuda:7'), covar=tensor([0.0497, 0.0073, 0.0064, 0.0399, 0.0117, 0.0137, 0.0100, 0.0487], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0091, 0.0092, 0.0137, 0.0104, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 18:42:53,369 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=290674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:43:03,184 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-02 18:43:19,913 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.946e+02 2.663e+02 2.912e+02 3.666e+02 7.754e+02, threshold=5.824e+02, percent-clipped=2.0 2023-05-02 18:43:37,902 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290702.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:43:40,047 INFO [train.py:904] (7/8) Epoch 29, batch 6500, loss[loss=0.1975, simple_loss=0.2779, pruned_loss=0.0585, over 16199.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2822, pruned_loss=0.05326, over 3100423.76 frames. ], batch size: 165, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:43:44,285 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290706.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:43:48,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9917, 3.8632, 4.0470, 4.1713, 4.2739, 3.8683, 4.2179, 4.3035], device='cuda:7'), covar=tensor([0.1789, 0.1285, 0.1362, 0.0688, 0.0604, 0.1611, 0.0923, 0.0810], device='cuda:7'), in_proj_covar=tensor([0.0685, 0.0834, 0.0962, 0.0844, 0.0643, 0.0666, 0.0706, 0.0821], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:44:29,792 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3454, 2.8565, 3.0345, 1.9641, 2.7440, 2.1235, 3.0794, 3.1670], device='cuda:7'), covar=tensor([0.0305, 0.0861, 0.0702, 0.2260, 0.0949, 0.1099, 0.0639, 0.0822], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0171, 0.0171, 0.0157, 0.0148, 0.0133, 0.0145, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 18:44:59,498 INFO [train.py:904] (7/8) Epoch 29, batch 6550, loss[loss=0.21, simple_loss=0.3051, pruned_loss=0.05745, over 17073.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2852, pruned_loss=0.05445, over 3100208.05 frames. ], batch size: 53, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:44:59,829 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=290754.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:45:07,994 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4276, 2.8542, 2.8508, 1.8678, 2.5689, 1.8416, 3.0996, 3.1546], device='cuda:7'), covar=tensor([0.0242, 0.0961, 0.0789, 0.2547, 0.1162, 0.1458, 0.0643, 0.0894], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0171, 0.0171, 0.0157, 0.0148, 0.0133, 0.0146, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 18:45:25,545 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 18:45:28,767 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7224, 4.5251, 4.7446, 4.9066, 5.0899, 4.6297, 5.1046, 5.1104], device='cuda:7'), covar=tensor([0.2025, 0.1357, 0.1655, 0.0792, 0.0818, 0.0996, 0.0741, 0.0720], device='cuda:7'), in_proj_covar=tensor([0.0682, 0.0831, 0.0958, 0.0842, 0.0642, 0.0663, 0.0703, 0.0819], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:45:58,158 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.498e+02 2.774e+02 3.264e+02 3.867e+02 7.263e+02, threshold=6.527e+02, percent-clipped=1.0 2023-05-02 18:46:19,476 INFO [train.py:904] (7/8) Epoch 29, batch 6600, loss[loss=0.2132, simple_loss=0.301, pruned_loss=0.06264, over 15328.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2873, pruned_loss=0.05491, over 3110633.39 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:47:20,757 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 18:47:38,823 INFO [train.py:904] (7/8) Epoch 29, batch 6650, loss[loss=0.197, simple_loss=0.2879, pruned_loss=0.05307, over 15420.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2875, pruned_loss=0.05601, over 3086155.72 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:48:35,967 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.668e+02 3.227e+02 3.787e+02 6.260e+02, threshold=6.453e+02, percent-clipped=0.0 2023-05-02 18:48:55,757 INFO [train.py:904] (7/8) Epoch 29, batch 6700, loss[loss=0.2059, simple_loss=0.2864, pruned_loss=0.06268, over 15309.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2868, pruned_loss=0.05628, over 3105375.98 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:49:14,085 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1282, 4.8855, 5.1175, 5.2937, 5.5034, 4.8147, 5.4778, 5.4834], device='cuda:7'), covar=tensor([0.2051, 0.1479, 0.1862, 0.0838, 0.0638, 0.1009, 0.0691, 0.0715], device='cuda:7'), in_proj_covar=tensor([0.0682, 0.0831, 0.0958, 0.0844, 0.0642, 0.0663, 0.0704, 0.0819], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:49:21,598 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4416, 3.4072, 2.6919, 2.1983, 2.2567, 2.3104, 3.5623, 3.0916], device='cuda:7'), covar=tensor([0.3015, 0.0678, 0.1895, 0.2917, 0.2708, 0.2298, 0.0541, 0.1329], device='cuda:7'), in_proj_covar=tensor([0.0336, 0.0276, 0.0314, 0.0329, 0.0307, 0.0279, 0.0306, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 18:50:12,980 INFO [train.py:904] (7/8) Epoch 29, batch 6750, loss[loss=0.1789, simple_loss=0.2711, pruned_loss=0.04336, over 16717.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.286, pruned_loss=0.05625, over 3106538.67 frames. ], batch size: 124, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:50:43,289 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=290974.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:51:10,932 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 2.623e+02 3.095e+02 3.882e+02 8.048e+02, threshold=6.190e+02, percent-clipped=2.0 2023-05-02 18:51:28,436 INFO [train.py:904] (7/8) Epoch 29, batch 6800, loss[loss=0.2059, simple_loss=0.2895, pruned_loss=0.06116, over 15317.00 frames. ], tot_loss[loss=0.2005, simple_loss=0.2868, pruned_loss=0.05703, over 3092529.15 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:51:35,754 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6219, 2.5746, 1.9462, 2.7010, 2.1852, 2.7831, 2.1348, 2.3305], device='cuda:7'), covar=tensor([0.0303, 0.0348, 0.1238, 0.0224, 0.0668, 0.0420, 0.1160, 0.0647], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0181, 0.0196, 0.0174, 0.0181, 0.0221, 0.0204, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:51:57,701 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=291022.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:52:40,594 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1043, 2.2082, 2.2499, 3.8590, 2.0673, 2.5004, 2.2943, 2.3868], device='cuda:7'), covar=tensor([0.1632, 0.4132, 0.3338, 0.0647, 0.4689, 0.2910, 0.4554, 0.3506], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0476, 0.0387, 0.0337, 0.0447, 0.0547, 0.0448, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:52:46,133 INFO [train.py:904] (7/8) Epoch 29, batch 6850, loss[loss=0.1775, simple_loss=0.2859, pruned_loss=0.03455, over 16881.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2882, pruned_loss=0.05739, over 3079304.22 frames. ], batch size: 96, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:52:58,627 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2419, 2.4787, 2.4873, 4.0763, 2.2990, 2.8012, 2.4858, 2.6297], device='cuda:7'), covar=tensor([0.1544, 0.3674, 0.2974, 0.0573, 0.4088, 0.2542, 0.3901, 0.2951], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0476, 0.0387, 0.0337, 0.0448, 0.0547, 0.0449, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:53:09,410 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=291069.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:53:42,658 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.921e+02 2.621e+02 3.195e+02 3.750e+02 8.276e+02, threshold=6.389e+02, percent-clipped=2.0 2023-05-02 18:54:02,738 INFO [train.py:904] (7/8) Epoch 29, batch 6900, loss[loss=0.2353, simple_loss=0.3074, pruned_loss=0.08164, over 11526.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2899, pruned_loss=0.05618, over 3104302.59 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:54:41,259 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5795, 4.7787, 4.9457, 4.6807, 4.7618, 5.2877, 4.7988, 4.5260], device='cuda:7'), covar=tensor([0.1358, 0.1872, 0.2580, 0.1952, 0.2171, 0.0955, 0.1686, 0.2511], device='cuda:7'), in_proj_covar=tensor([0.0434, 0.0645, 0.0714, 0.0522, 0.0699, 0.0733, 0.0554, 0.0699], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 18:54:45,320 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=291130.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 18:55:01,524 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1062, 2.4738, 2.0356, 2.2573, 2.8093, 2.4595, 2.6799, 2.9787], device='cuda:7'), covar=tensor([0.0217, 0.0501, 0.0678, 0.0539, 0.0310, 0.0446, 0.0249, 0.0316], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0243, 0.0234, 0.0234, 0.0244, 0.0242, 0.0238, 0.0242], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 18:55:20,073 INFO [train.py:904] (7/8) Epoch 29, batch 6950, loss[loss=0.2486, simple_loss=0.3172, pruned_loss=0.08998, over 11382.00 frames. ], tot_loss[loss=0.2028, simple_loss=0.2912, pruned_loss=0.0572, over 3111198.49 frames. ], batch size: 247, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:55:39,231 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0472, 3.0790, 1.9498, 3.2888, 2.4199, 3.3284, 2.1410, 2.5497], device='cuda:7'), covar=tensor([0.0347, 0.0455, 0.1787, 0.0269, 0.0895, 0.0627, 0.1552, 0.0765], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0182, 0.0197, 0.0175, 0.0181, 0.0222, 0.0205, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 18:56:18,594 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.816e+02 3.069e+02 3.631e+02 4.410e+02 8.633e+02, threshold=7.261e+02, percent-clipped=4.0 2023-05-02 18:56:27,319 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1961, 4.3465, 4.5208, 4.2472, 4.3374, 4.8439, 4.3924, 4.1175], device='cuda:7'), covar=tensor([0.1929, 0.1951, 0.2517, 0.1982, 0.2363, 0.1060, 0.1696, 0.2339], device='cuda:7'), in_proj_covar=tensor([0.0433, 0.0643, 0.0713, 0.0519, 0.0697, 0.0731, 0.0552, 0.0697], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 18:56:36,628 INFO [train.py:904] (7/8) Epoch 29, batch 7000, loss[loss=0.1811, simple_loss=0.2841, pruned_loss=0.03907, over 17027.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2906, pruned_loss=0.05618, over 3108381.17 frames. ], batch size: 50, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:57:50,294 INFO [train.py:904] (7/8) Epoch 29, batch 7050, loss[loss=0.1848, simple_loss=0.2726, pruned_loss=0.04844, over 16757.00 frames. ], tot_loss[loss=0.2026, simple_loss=0.2913, pruned_loss=0.05692, over 3076116.04 frames. ], batch size: 39, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 18:58:49,449 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.848e+02 2.753e+02 3.384e+02 4.060e+02 1.222e+03, threshold=6.768e+02, percent-clipped=4.0 2023-05-02 18:59:07,333 INFO [train.py:904] (7/8) Epoch 29, batch 7100, loss[loss=0.2048, simple_loss=0.2792, pruned_loss=0.06515, over 11641.00 frames. ], tot_loss[loss=0.2024, simple_loss=0.2902, pruned_loss=0.05726, over 3067856.09 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 19:00:05,752 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2382, 2.3945, 2.4414, 4.0206, 2.2940, 2.7215, 2.3991, 2.5705], device='cuda:7'), covar=tensor([0.1518, 0.3430, 0.2971, 0.0554, 0.4120, 0.2431, 0.3599, 0.3070], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0476, 0.0386, 0.0337, 0.0447, 0.0547, 0.0449, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:00:25,899 INFO [train.py:904] (7/8) Epoch 29, batch 7150, loss[loss=0.1985, simple_loss=0.2878, pruned_loss=0.05454, over 16333.00 frames. ], tot_loss[loss=0.2015, simple_loss=0.2885, pruned_loss=0.05721, over 3058915.65 frames. ], batch size: 146, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 19:00:28,570 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-05-02 19:01:23,568 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.798e+02 3.396e+02 4.128e+02 9.099e+02, threshold=6.792e+02, percent-clipped=1.0 2023-05-02 19:01:41,845 INFO [train.py:904] (7/8) Epoch 29, batch 7200, loss[loss=0.1641, simple_loss=0.2639, pruned_loss=0.03211, over 16863.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2859, pruned_loss=0.05543, over 3064606.50 frames. ], batch size: 96, lr: 2.31e-03, grad_scale: 8.0 2023-05-02 19:02:14,514 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=291425.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:02:29,100 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.57 vs. limit=5.0 2023-05-02 19:03:00,080 INFO [train.py:904] (7/8) Epoch 29, batch 7250, loss[loss=0.1949, simple_loss=0.2805, pruned_loss=0.0546, over 15473.00 frames. ], tot_loss[loss=0.1963, simple_loss=0.2838, pruned_loss=0.05442, over 3051838.38 frames. ], batch size: 191, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:03:58,890 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.627e+02 2.585e+02 3.000e+02 3.629e+02 8.129e+02, threshold=6.000e+02, percent-clipped=2.0 2023-05-02 19:04:00,875 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8989, 2.3424, 1.8722, 2.0695, 2.6467, 2.2759, 2.5072, 2.8008], device='cuda:7'), covar=tensor([0.0291, 0.0483, 0.0703, 0.0604, 0.0360, 0.0487, 0.0293, 0.0325], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0244, 0.0234, 0.0235, 0.0245, 0.0242, 0.0240, 0.0243], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:04:16,128 INFO [train.py:904] (7/8) Epoch 29, batch 7300, loss[loss=0.2672, simple_loss=0.32, pruned_loss=0.1071, over 11684.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2838, pruned_loss=0.05476, over 3052543.80 frames. ], batch size: 248, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:05:04,771 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-05-02 19:05:32,034 INFO [train.py:904] (7/8) Epoch 29, batch 7350, loss[loss=0.2022, simple_loss=0.287, pruned_loss=0.05874, over 15340.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2851, pruned_loss=0.05583, over 3037552.28 frames. ], batch size: 190, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:06:32,138 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.793e+02 2.867e+02 3.369e+02 3.998e+02 9.840e+02, threshold=6.738e+02, percent-clipped=8.0 2023-05-02 19:06:45,507 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-05-02 19:06:49,565 INFO [train.py:904] (7/8) Epoch 29, batch 7400, loss[loss=0.2024, simple_loss=0.2867, pruned_loss=0.05908, over 17062.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2857, pruned_loss=0.05632, over 3040044.22 frames. ], batch size: 53, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:08:07,288 INFO [train.py:904] (7/8) Epoch 29, batch 7450, loss[loss=0.2015, simple_loss=0.2873, pruned_loss=0.05784, over 16735.00 frames. ], tot_loss[loss=0.1998, simple_loss=0.2864, pruned_loss=0.05664, over 3051480.31 frames. ], batch size: 124, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:08:10,494 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1006, 2.4505, 2.5550, 2.0049, 2.6990, 2.7963, 2.4220, 2.3691], device='cuda:7'), covar=tensor([0.0753, 0.0299, 0.0280, 0.0927, 0.0146, 0.0294, 0.0472, 0.0488], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0111, 0.0104, 0.0139, 0.0088, 0.0132, 0.0130, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:08:15,988 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-05-02 19:09:10,902 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.991e+02 2.789e+02 3.286e+02 3.823e+02 6.627e+02, threshold=6.571e+02, percent-clipped=0.0 2023-05-02 19:09:28,118 INFO [train.py:904] (7/8) Epoch 29, batch 7500, loss[loss=0.1773, simple_loss=0.2692, pruned_loss=0.04266, over 16812.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2867, pruned_loss=0.05598, over 3064365.99 frames. ], batch size: 102, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:10:01,653 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=291725.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:10:45,771 INFO [train.py:904] (7/8) Epoch 29, batch 7550, loss[loss=0.1882, simple_loss=0.2722, pruned_loss=0.05217, over 16653.00 frames. ], tot_loss[loss=0.199, simple_loss=0.2857, pruned_loss=0.05614, over 3054953.72 frames. ], batch size: 134, lr: 2.31e-03, grad_scale: 4.0 2023-05-02 19:11:15,451 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=291773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:11:19,702 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=291776.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:11:44,714 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.882e+02 2.637e+02 3.162e+02 3.774e+02 1.302e+03, threshold=6.323e+02, percent-clipped=2.0 2023-05-02 19:11:58,241 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9911, 3.0346, 1.9856, 3.1992, 2.3711, 3.2984, 2.2162, 2.5353], device='cuda:7'), covar=tensor([0.0367, 0.0453, 0.1651, 0.0273, 0.0844, 0.0687, 0.1446, 0.0815], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0181, 0.0196, 0.0174, 0.0180, 0.0221, 0.0205, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:12:01,417 INFO [train.py:904] (7/8) Epoch 29, batch 7600, loss[loss=0.1977, simple_loss=0.2861, pruned_loss=0.05462, over 16301.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2857, pruned_loss=0.05681, over 3033502.81 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:12:18,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1402, 2.3158, 2.3990, 3.8454, 2.1876, 2.6591, 2.3803, 2.4820], device='cuda:7'), covar=tensor([0.1538, 0.3523, 0.3052, 0.0623, 0.4274, 0.2483, 0.3743, 0.3378], device='cuda:7'), in_proj_covar=tensor([0.0423, 0.0476, 0.0387, 0.0337, 0.0447, 0.0547, 0.0448, 0.0558], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:12:52,618 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=291837.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:12:54,530 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3692, 3.5052, 3.7717, 2.0933, 3.2317, 2.4616, 3.6846, 3.8039], device='cuda:7'), covar=tensor([0.0252, 0.0865, 0.0569, 0.2245, 0.0841, 0.1008, 0.0614, 0.0986], device='cuda:7'), in_proj_covar=tensor([0.0161, 0.0172, 0.0171, 0.0158, 0.0149, 0.0134, 0.0146, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 19:12:56,161 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9283, 4.1950, 4.0129, 4.0708, 3.7386, 3.8236, 3.8330, 4.1680], device='cuda:7'), covar=tensor([0.1106, 0.0884, 0.1031, 0.0842, 0.0813, 0.1521, 0.0984, 0.1029], device='cuda:7'), in_proj_covar=tensor([0.0721, 0.0866, 0.0714, 0.0674, 0.0552, 0.0555, 0.0726, 0.0681], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:13:18,972 INFO [train.py:904] (7/8) Epoch 29, batch 7650, loss[loss=0.1771, simple_loss=0.2746, pruned_loss=0.03983, over 16736.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2853, pruned_loss=0.05665, over 3049479.73 frames. ], batch size: 83, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:13:36,443 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1747, 2.0582, 2.6779, 3.0881, 2.9315, 3.4631, 2.1408, 3.5312], device='cuda:7'), covar=tensor([0.0247, 0.0639, 0.0387, 0.0350, 0.0360, 0.0201, 0.0704, 0.0159], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0197, 0.0186, 0.0191, 0.0208, 0.0166, 0.0203, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:14:20,910 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.976e+02 2.915e+02 3.313e+02 4.275e+02 8.487e+02, threshold=6.627e+02, percent-clipped=5.0 2023-05-02 19:14:34,133 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4793, 1.7156, 2.1563, 2.3883, 2.4448, 2.6791, 1.8527, 2.6443], device='cuda:7'), covar=tensor([0.0251, 0.0625, 0.0350, 0.0369, 0.0390, 0.0262, 0.0661, 0.0219], device='cuda:7'), in_proj_covar=tensor([0.0196, 0.0197, 0.0185, 0.0191, 0.0208, 0.0166, 0.0203, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:14:36,022 INFO [train.py:904] (7/8) Epoch 29, batch 7700, loss[loss=0.195, simple_loss=0.2809, pruned_loss=0.05451, over 16490.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2844, pruned_loss=0.05623, over 3071078.57 frames. ], batch size: 68, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:15:53,500 INFO [train.py:904] (7/8) Epoch 29, batch 7750, loss[loss=0.1892, simple_loss=0.2819, pruned_loss=0.0482, over 17197.00 frames. ], tot_loss[loss=0.1983, simple_loss=0.2847, pruned_loss=0.05594, over 3063754.32 frames. ], batch size: 44, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:16:55,759 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.177e+02 2.797e+02 3.314e+02 3.963e+02 7.468e+02, threshold=6.627e+02, percent-clipped=1.0 2023-05-02 19:17:14,571 INFO [train.py:904] (7/8) Epoch 29, batch 7800, loss[loss=0.1848, simple_loss=0.277, pruned_loss=0.04633, over 16507.00 frames. ], tot_loss[loss=0.1995, simple_loss=0.2855, pruned_loss=0.05676, over 3064451.33 frames. ], batch size: 75, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:18:30,017 INFO [train.py:904] (7/8) Epoch 29, batch 7850, loss[loss=0.1977, simple_loss=0.2988, pruned_loss=0.04826, over 16811.00 frames. ], tot_loss[loss=0.2001, simple_loss=0.2867, pruned_loss=0.05674, over 3056347.92 frames. ], batch size: 39, lr: 2.30e-03, grad_scale: 2.0 2023-05-02 19:18:31,039 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=292054.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:19:30,100 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.705e+02 3.148e+02 3.782e+02 7.710e+02, threshold=6.297e+02, percent-clipped=1.0 2023-05-02 19:19:43,449 INFO [train.py:904] (7/8) Epoch 29, batch 7900, loss[loss=0.1942, simple_loss=0.2874, pruned_loss=0.05052, over 16629.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2858, pruned_loss=0.05619, over 3065500.83 frames. ], batch size: 134, lr: 2.30e-03, grad_scale: 2.0 2023-05-02 19:19:51,504 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3435, 2.9195, 2.6672, 2.3030, 2.2717, 2.3354, 2.8958, 2.8189], device='cuda:7'), covar=tensor([0.2805, 0.0758, 0.1765, 0.2798, 0.2634, 0.2381, 0.0587, 0.1498], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0275, 0.0315, 0.0329, 0.0308, 0.0279, 0.0306, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 19:20:00,041 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=292115.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:20:26,274 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=292132.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:20:41,448 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0383, 4.1098, 3.9546, 3.6956, 3.7288, 4.0511, 3.7104, 3.8488], device='cuda:7'), covar=tensor([0.0605, 0.0724, 0.0339, 0.0312, 0.0710, 0.0483, 0.1126, 0.0630], device='cuda:7'), in_proj_covar=tensor([0.0310, 0.0470, 0.0363, 0.0363, 0.0360, 0.0418, 0.0251, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:20:47,181 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-02 19:21:01,246 INFO [train.py:904] (7/8) Epoch 29, batch 7950, loss[loss=0.1864, simple_loss=0.2758, pruned_loss=0.04849, over 16663.00 frames. ], tot_loss[loss=0.2002, simple_loss=0.2865, pruned_loss=0.05691, over 3059954.96 frames. ], batch size: 89, lr: 2.30e-03, grad_scale: 2.0 2023-05-02 19:22:03,660 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.577e+02 3.154e+02 3.832e+02 7.252e+02, threshold=6.308e+02, percent-clipped=1.0 2023-05-02 19:22:18,476 INFO [train.py:904] (7/8) Epoch 29, batch 8000, loss[loss=0.2115, simple_loss=0.3055, pruned_loss=0.05869, over 16843.00 frames. ], tot_loss[loss=0.2009, simple_loss=0.2869, pruned_loss=0.05746, over 3050152.14 frames. ], batch size: 102, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:23:20,985 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8373, 5.1425, 4.9235, 4.9360, 4.6754, 4.6144, 4.5178, 5.2212], device='cuda:7'), covar=tensor([0.1223, 0.0820, 0.1054, 0.0918, 0.0826, 0.1179, 0.1225, 0.0862], device='cuda:7'), in_proj_covar=tensor([0.0720, 0.0864, 0.0714, 0.0673, 0.0552, 0.0554, 0.0725, 0.0680], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:23:21,154 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.6587, 2.6519, 2.3238, 3.8606, 2.6356, 3.8572, 1.5539, 2.8716], device='cuda:7'), covar=tensor([0.1429, 0.0822, 0.1382, 0.0211, 0.0233, 0.0400, 0.1812, 0.0842], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0183, 0.0203, 0.0206, 0.0209, 0.0220, 0.0212, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:23:31,120 INFO [train.py:904] (7/8) Epoch 29, batch 8050, loss[loss=0.1919, simple_loss=0.2793, pruned_loss=0.05223, over 16609.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2868, pruned_loss=0.05689, over 3062447.32 frames. ], batch size: 57, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:23:37,645 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 19:24:12,822 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4142, 3.5240, 2.0799, 3.9312, 2.6593, 3.9118, 2.2108, 2.7736], device='cuda:7'), covar=tensor([0.0376, 0.0444, 0.1894, 0.0296, 0.0931, 0.0682, 0.1747, 0.0912], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0180, 0.0196, 0.0173, 0.0180, 0.0220, 0.0204, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:24:32,482 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 2.613e+02 3.145e+02 3.848e+02 6.977e+02, threshold=6.289e+02, percent-clipped=2.0 2023-05-02 19:24:37,631 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9754, 2.8102, 2.6785, 4.6997, 3.4469, 4.1238, 1.7291, 3.1963], device='cuda:7'), covar=tensor([0.1323, 0.0818, 0.1244, 0.0154, 0.0309, 0.0436, 0.1696, 0.0820], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0183, 0.0203, 0.0206, 0.0210, 0.0220, 0.0213, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:24:46,385 INFO [train.py:904] (7/8) Epoch 29, batch 8100, loss[loss=0.1936, simple_loss=0.2797, pruned_loss=0.05373, over 16223.00 frames. ], tot_loss[loss=0.1986, simple_loss=0.2859, pruned_loss=0.05566, over 3078255.53 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:25:36,799 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 19:25:49,653 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-05-02 19:26:01,431 INFO [train.py:904] (7/8) Epoch 29, batch 8150, loss[loss=0.1853, simple_loss=0.2637, pruned_loss=0.05345, over 16697.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2832, pruned_loss=0.05454, over 3085721.66 frames. ], batch size: 57, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:26:32,605 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2798, 4.3928, 4.6851, 4.6263, 4.6627, 4.3856, 4.3789, 4.3186], device='cuda:7'), covar=tensor([0.0370, 0.0623, 0.0357, 0.0439, 0.0441, 0.0437, 0.0900, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0440, 0.0498, 0.0479, 0.0443, 0.0526, 0.0506, 0.0584, 0.0406], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 19:27:01,243 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.736e+02 3.221e+02 3.909e+02 8.294e+02, threshold=6.443e+02, percent-clipped=4.0 2023-05-02 19:27:13,367 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-05-02 19:27:15,047 INFO [train.py:904] (7/8) Epoch 29, batch 8200, loss[loss=0.1898, simple_loss=0.2803, pruned_loss=0.0497, over 16743.00 frames. ], tot_loss[loss=0.1939, simple_loss=0.2807, pruned_loss=0.05359, over 3116496.20 frames. ], batch size: 134, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:27:25,896 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=292410.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:28:00,927 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=292432.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:28:34,808 INFO [train.py:904] (7/8) Epoch 29, batch 8250, loss[loss=0.19, simple_loss=0.2866, pruned_loss=0.04672, over 16169.00 frames. ], tot_loss[loss=0.1912, simple_loss=0.2798, pruned_loss=0.05129, over 3112695.73 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:29:15,008 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2703, 3.4305, 3.8387, 2.1649, 3.3390, 2.4944, 3.6639, 3.6915], device='cuda:7'), covar=tensor([0.0226, 0.0882, 0.0505, 0.2122, 0.0734, 0.0999, 0.0549, 0.0910], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0170, 0.0170, 0.0156, 0.0147, 0.0133, 0.0145, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 19:29:17,991 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=292480.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:29:41,444 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.184e+02 2.675e+02 3.477e+02 6.200e+02, threshold=5.351e+02, percent-clipped=0.0 2023-05-02 19:29:55,813 INFO [train.py:904] (7/8) Epoch 29, batch 8300, loss[loss=0.1703, simple_loss=0.2752, pruned_loss=0.03264, over 16651.00 frames. ], tot_loss[loss=0.1866, simple_loss=0.2768, pruned_loss=0.04822, over 3099028.59 frames. ], batch size: 83, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:30:29,469 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6288, 1.8053, 2.2974, 2.6578, 2.5530, 2.9189, 2.0148, 2.9958], device='cuda:7'), covar=tensor([0.0269, 0.0646, 0.0446, 0.0364, 0.0426, 0.0246, 0.0673, 0.0233], device='cuda:7'), in_proj_covar=tensor([0.0195, 0.0196, 0.0185, 0.0190, 0.0207, 0.0165, 0.0203, 0.0166], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:31:15,571 INFO [train.py:904] (7/8) Epoch 29, batch 8350, loss[loss=0.1794, simple_loss=0.2739, pruned_loss=0.04243, over 15286.00 frames. ], tot_loss[loss=0.1852, simple_loss=0.2769, pruned_loss=0.04673, over 3101180.66 frames. ], batch size: 190, lr: 2.30e-03, grad_scale: 4.0 2023-05-02 19:32:20,508 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.032e+02 2.659e+02 3.191e+02 5.260e+02, threshold=5.318e+02, percent-clipped=0.0 2023-05-02 19:32:25,651 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6846, 4.4796, 4.7221, 4.8592, 5.0336, 4.5376, 5.0273, 5.0327], device='cuda:7'), covar=tensor([0.2027, 0.1472, 0.1809, 0.0835, 0.0619, 0.1065, 0.0640, 0.0866], device='cuda:7'), in_proj_covar=tensor([0.0672, 0.0820, 0.0944, 0.0835, 0.0638, 0.0656, 0.0698, 0.0814], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:32:36,090 INFO [train.py:904] (7/8) Epoch 29, batch 8400, loss[loss=0.1652, simple_loss=0.2681, pruned_loss=0.03113, over 16784.00 frames. ], tot_loss[loss=0.182, simple_loss=0.2744, pruned_loss=0.04477, over 3099423.60 frames. ], batch size: 102, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:33:56,344 INFO [train.py:904] (7/8) Epoch 29, batch 8450, loss[loss=0.1624, simple_loss=0.2604, pruned_loss=0.03217, over 17260.00 frames. ], tot_loss[loss=0.1793, simple_loss=0.2726, pruned_loss=0.04297, over 3101131.45 frames. ], batch size: 52, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:34:14,476 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0706, 3.1193, 2.0242, 3.2842, 2.3693, 3.3282, 2.2726, 2.6481], device='cuda:7'), covar=tensor([0.0312, 0.0378, 0.1468, 0.0296, 0.0783, 0.0513, 0.1419, 0.0667], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0177, 0.0193, 0.0170, 0.0177, 0.0216, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:34:26,903 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-05-02 19:35:03,484 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.196e+02 2.552e+02 2.926e+02 6.173e+02, threshold=5.103e+02, percent-clipped=1.0 2023-05-02 19:35:10,515 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7535, 3.9269, 2.4836, 4.3595, 3.0770, 4.3159, 2.7486, 3.2356], device='cuda:7'), covar=tensor([0.0296, 0.0340, 0.1524, 0.0292, 0.0728, 0.0488, 0.1316, 0.0650], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0177, 0.0192, 0.0169, 0.0176, 0.0215, 0.0200, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:35:19,270 INFO [train.py:904] (7/8) Epoch 29, batch 8500, loss[loss=0.1519, simple_loss=0.2338, pruned_loss=0.03501, over 11732.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2691, pruned_loss=0.04097, over 3099606.55 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:35:28,951 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=292710.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:36:25,494 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-05-02 19:36:40,877 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5693, 2.8935, 3.3403, 2.0852, 2.8536, 2.1326, 3.1377, 3.0973], device='cuda:7'), covar=tensor([0.0286, 0.0990, 0.0505, 0.2125, 0.0833, 0.1085, 0.0649, 0.0988], device='cuda:7'), in_proj_covar=tensor([0.0158, 0.0168, 0.0168, 0.0155, 0.0146, 0.0132, 0.0143, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 19:36:42,996 INFO [train.py:904] (7/8) Epoch 29, batch 8550, loss[loss=0.1831, simple_loss=0.2634, pruned_loss=0.05136, over 11833.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.2669, pruned_loss=0.04017, over 3072433.12 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:36:52,496 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=292758.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:38:04,316 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.654e+02 2.162e+02 2.570e+02 3.257e+02 5.024e+02, threshold=5.139e+02, percent-clipped=0.0 2023-05-02 19:38:21,506 INFO [train.py:904] (7/8) Epoch 29, batch 8600, loss[loss=0.1564, simple_loss=0.2479, pruned_loss=0.03242, over 12322.00 frames. ], tot_loss[loss=0.1726, simple_loss=0.2669, pruned_loss=0.0392, over 3061547.18 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:40:01,877 INFO [train.py:904] (7/8) Epoch 29, batch 8650, loss[loss=0.1547, simple_loss=0.2541, pruned_loss=0.02762, over 15237.00 frames. ], tot_loss[loss=0.1706, simple_loss=0.2653, pruned_loss=0.03796, over 3056215.77 frames. ], batch size: 190, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:41:31,356 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.024e+02 2.393e+02 2.870e+02 4.192e+02, threshold=4.785e+02, percent-clipped=0.0 2023-05-02 19:41:48,482 INFO [train.py:904] (7/8) Epoch 29, batch 8700, loss[loss=0.1599, simple_loss=0.2481, pruned_loss=0.03584, over 11986.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2625, pruned_loss=0.03679, over 3047890.96 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:43:26,849 INFO [train.py:904] (7/8) Epoch 29, batch 8750, loss[loss=0.2008, simple_loss=0.2966, pruned_loss=0.05252, over 16717.00 frames. ], tot_loss[loss=0.1683, simple_loss=0.2631, pruned_loss=0.03679, over 3049858.03 frames. ], batch size: 134, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:44:01,935 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=292967.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:44:57,308 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-05-02 19:45:00,945 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 2.165e+02 2.592e+02 3.174e+02 6.684e+02, threshold=5.184e+02, percent-clipped=4.0 2023-05-02 19:45:20,133 INFO [train.py:904] (7/8) Epoch 29, batch 8800, loss[loss=0.164, simple_loss=0.2665, pruned_loss=0.03076, over 16868.00 frames. ], tot_loss[loss=0.1663, simple_loss=0.2612, pruned_loss=0.03566, over 3054665.87 frames. ], batch size: 96, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:46:12,301 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293028.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:47:05,849 INFO [train.py:904] (7/8) Epoch 29, batch 8850, loss[loss=0.1746, simple_loss=0.2801, pruned_loss=0.03455, over 16228.00 frames. ], tot_loss[loss=0.167, simple_loss=0.264, pruned_loss=0.03505, over 3056770.51 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:47:24,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 19:48:05,927 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293081.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:48:35,607 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.279e+02 2.601e+02 3.166e+02 4.771e+02, threshold=5.201e+02, percent-clipped=0.0 2023-05-02 19:48:55,730 INFO [train.py:904] (7/8) Epoch 29, batch 8900, loss[loss=0.1733, simple_loss=0.2659, pruned_loss=0.04038, over 16870.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2646, pruned_loss=0.03443, over 3066965.66 frames. ], batch size: 116, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:49:09,414 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1529, 3.4589, 3.4655, 2.3624, 3.1791, 3.4983, 3.3674, 1.8958], device='cuda:7'), covar=tensor([0.0590, 0.0075, 0.0071, 0.0448, 0.0127, 0.0120, 0.0096, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0136, 0.0088, 0.0089, 0.0132, 0.0101, 0.0113, 0.0097, 0.0128], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-02 19:49:24,544 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293117.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:49:26,256 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8265, 1.4279, 1.6456, 1.7595, 1.8657, 1.9011, 1.6774, 1.8488], device='cuda:7'), covar=tensor([0.0287, 0.0487, 0.0257, 0.0351, 0.0374, 0.0236, 0.0496, 0.0184], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0194, 0.0184, 0.0188, 0.0206, 0.0163, 0.0201, 0.0165], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:50:03,688 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8829, 3.9279, 4.0125, 3.8053, 3.9347, 4.3496, 3.9847, 3.7121], device='cuda:7'), covar=tensor([0.2109, 0.2197, 0.2438, 0.2392, 0.2720, 0.1582, 0.1583, 0.2605], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0620, 0.0689, 0.0503, 0.0673, 0.0709, 0.0536, 0.0671], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 19:50:34,662 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293142.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:50:42,583 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 19:51:01,972 INFO [train.py:904] (7/8) Epoch 29, batch 8950, loss[loss=0.1491, simple_loss=0.241, pruned_loss=0.02857, over 12454.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.2637, pruned_loss=0.03486, over 3079696.37 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:51:53,110 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293178.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:52:32,342 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.460e+02 2.109e+02 2.425e+02 3.053e+02 5.475e+02, threshold=4.851e+02, percent-clipped=2.0 2023-05-02 19:52:53,153 INFO [train.py:904] (7/8) Epoch 29, batch 9000, loss[loss=0.1507, simple_loss=0.2398, pruned_loss=0.03078, over 16533.00 frames. ], tot_loss[loss=0.1641, simple_loss=0.2605, pruned_loss=0.03388, over 3072482.25 frames. ], batch size: 75, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:52:53,154 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 19:53:02,751 INFO [train.py:938] (7/8) Epoch 29, validation: loss=0.1431, simple_loss=0.2465, pruned_loss=0.01987, over 944034.00 frames. 2023-05-02 19:53:02,752 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 19:53:29,319 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0704, 4.0199, 3.9876, 3.1519, 3.9231, 1.8322, 3.7613, 3.5062], device='cuda:7'), covar=tensor([0.0144, 0.0135, 0.0181, 0.0291, 0.0122, 0.3146, 0.0161, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0173, 0.0211, 0.0182, 0.0188, 0.0216, 0.0199, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:54:47,969 INFO [train.py:904] (7/8) Epoch 29, batch 9050, loss[loss=0.1714, simple_loss=0.2585, pruned_loss=0.04212, over 12629.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2617, pruned_loss=0.03454, over 3082124.17 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:55:15,583 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0675, 2.0972, 2.1929, 3.4782, 2.0501, 2.4018, 2.2351, 2.2432], device='cuda:7'), covar=tensor([0.1394, 0.3950, 0.3396, 0.0665, 0.4748, 0.2856, 0.4132, 0.3631], device='cuda:7'), in_proj_covar=tensor([0.0415, 0.0468, 0.0381, 0.0329, 0.0439, 0.0535, 0.0441, 0.0547], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 19:56:15,026 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.113e+02 2.508e+02 2.955e+02 4.534e+02, threshold=5.015e+02, percent-clipped=0.0 2023-05-02 19:56:34,938 INFO [train.py:904] (7/8) Epoch 29, batch 9100, loss[loss=0.1584, simple_loss=0.2482, pruned_loss=0.03436, over 12259.00 frames. ], tot_loss[loss=0.1658, simple_loss=0.2613, pruned_loss=0.0351, over 3077546.82 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:56:49,145 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1501, 2.5358, 2.6144, 1.9436, 2.7964, 2.8036, 2.5686, 2.5058], device='cuda:7'), covar=tensor([0.0654, 0.0280, 0.0264, 0.1025, 0.0132, 0.0257, 0.0449, 0.0451], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0108, 0.0099, 0.0136, 0.0085, 0.0128, 0.0127, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 19:57:17,602 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293323.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:58:34,120 INFO [train.py:904] (7/8) Epoch 29, batch 9150, loss[loss=0.1496, simple_loss=0.2414, pruned_loss=0.0289, over 12022.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2619, pruned_loss=0.035, over 3051331.84 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 19:58:51,922 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293361.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 19:58:57,359 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293364.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:00:04,475 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 2.108e+02 2.614e+02 3.370e+02 9.014e+02, threshold=5.228e+02, percent-clipped=3.0 2023-05-02 20:00:20,601 INFO [train.py:904] (7/8) Epoch 29, batch 9200, loss[loss=0.1756, simple_loss=0.2733, pruned_loss=0.03891, over 16225.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2578, pruned_loss=0.03415, over 3051288.37 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:00:55,093 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293422.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:01:00,715 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=293425.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:01:09,723 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 20:01:23,548 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293437.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:01:54,173 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-02 20:01:58,354 INFO [train.py:904] (7/8) Epoch 29, batch 9250, loss[loss=0.1664, simple_loss=0.2524, pruned_loss=0.04021, over 12501.00 frames. ], tot_loss[loss=0.1628, simple_loss=0.2573, pruned_loss=0.03411, over 3041423.79 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:02:39,159 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293473.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:03:23,292 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-05-02 20:03:28,490 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.175e+02 2.514e+02 3.115e+02 5.451e+02, threshold=5.029e+02, percent-clipped=2.0 2023-05-02 20:03:49,268 INFO [train.py:904] (7/8) Epoch 29, batch 9300, loss[loss=0.1342, simple_loss=0.2303, pruned_loss=0.01902, over 16777.00 frames. ], tot_loss[loss=0.1616, simple_loss=0.2559, pruned_loss=0.03363, over 3041295.05 frames. ], batch size: 76, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:05:33,324 INFO [train.py:904] (7/8) Epoch 29, batch 9350, loss[loss=0.1718, simple_loss=0.26, pruned_loss=0.04182, over 15361.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2553, pruned_loss=0.03346, over 3038091.49 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:06:29,234 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3133, 5.2995, 5.0264, 4.4422, 5.1327, 2.0370, 4.8444, 4.8816], device='cuda:7'), covar=tensor([0.0094, 0.0099, 0.0221, 0.0406, 0.0116, 0.2623, 0.0174, 0.0211], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0171, 0.0208, 0.0180, 0.0186, 0.0215, 0.0197, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:06:52,958 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8752, 1.4504, 1.7429, 1.7493, 1.9439, 1.9118, 1.7285, 1.8746], device='cuda:7'), covar=tensor([0.0341, 0.0520, 0.0286, 0.0384, 0.0403, 0.0271, 0.0538, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0192, 0.0192, 0.0182, 0.0186, 0.0205, 0.0162, 0.0200, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:06:56,973 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 1.973e+02 2.358e+02 2.873e+02 5.255e+02, threshold=4.716e+02, percent-clipped=1.0 2023-05-02 20:07:15,895 INFO [train.py:904] (7/8) Epoch 29, batch 9400, loss[loss=0.1591, simple_loss=0.2496, pruned_loss=0.03426, over 12109.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2559, pruned_loss=0.03331, over 3033788.62 frames. ], batch size: 246, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:07:39,441 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3367, 4.4370, 4.2438, 3.9594, 3.9491, 4.3461, 4.0794, 4.1088], device='cuda:7'), covar=tensor([0.0684, 0.0643, 0.0409, 0.0343, 0.0880, 0.0599, 0.0758, 0.0618], device='cuda:7'), in_proj_covar=tensor([0.0303, 0.0457, 0.0354, 0.0355, 0.0350, 0.0407, 0.0244, 0.0422], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:07:53,827 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=293623.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:08:55,066 INFO [train.py:904] (7/8) Epoch 29, batch 9450, loss[loss=0.1514, simple_loss=0.2501, pruned_loss=0.02632, over 15289.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2575, pruned_loss=0.03345, over 3029894.01 frames. ], batch size: 190, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:09:29,882 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=293671.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:10:18,285 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.101e+02 2.359e+02 2.998e+02 5.430e+02, threshold=4.718e+02, percent-clipped=2.0 2023-05-02 20:10:34,120 INFO [train.py:904] (7/8) Epoch 29, batch 9500, loss[loss=0.1509, simple_loss=0.2347, pruned_loss=0.03357, over 12904.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2564, pruned_loss=0.03317, over 3043805.38 frames. ], batch size: 247, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:11:03,204 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293717.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:11:08,956 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=293720.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:11:42,016 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=293737.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:12:19,496 INFO [train.py:904] (7/8) Epoch 29, batch 9550, loss[loss=0.1617, simple_loss=0.2586, pruned_loss=0.03239, over 16703.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2561, pruned_loss=0.03312, over 3051790.17 frames. ], batch size: 76, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:12:30,352 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2286, 3.3701, 3.7967, 2.1401, 3.2275, 2.4685, 3.5703, 3.5434], device='cuda:7'), covar=tensor([0.0259, 0.1029, 0.0493, 0.2244, 0.0764, 0.1028, 0.0616, 0.1131], device='cuda:7'), in_proj_covar=tensor([0.0156, 0.0164, 0.0166, 0.0153, 0.0144, 0.0129, 0.0141, 0.0177], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:12:42,594 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8846, 3.8489, 3.9869, 3.7431, 3.9097, 4.3378, 3.9402, 3.6373], device='cuda:7'), covar=tensor([0.2038, 0.2326, 0.2407, 0.2386, 0.2711, 0.1544, 0.1772, 0.2780], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0618, 0.0688, 0.0501, 0.0669, 0.0709, 0.0533, 0.0669], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 20:12:59,118 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=293773.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:13:25,493 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=293785.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:13:43,640 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.273e+02 2.766e+02 3.522e+02 6.419e+02, threshold=5.532e+02, percent-clipped=5.0 2023-05-02 20:13:59,143 INFO [train.py:904] (7/8) Epoch 29, batch 9600, loss[loss=0.1896, simple_loss=0.2865, pruned_loss=0.04637, over 16733.00 frames. ], tot_loss[loss=0.1631, simple_loss=0.2582, pruned_loss=0.03405, over 3049311.75 frames. ], batch size: 83, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:14:19,028 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2540, 3.0298, 3.2464, 1.7477, 3.4001, 3.5172, 2.7786, 2.7674], device='cuda:7'), covar=tensor([0.0857, 0.0359, 0.0286, 0.1311, 0.0130, 0.0197, 0.0569, 0.0521], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0109, 0.0099, 0.0136, 0.0085, 0.0127, 0.0127, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 20:14:32,486 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=293821.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:14:32,668 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0743, 2.5277, 2.5886, 1.8702, 2.7467, 2.8102, 2.4842, 2.4684], device='cuda:7'), covar=tensor([0.0689, 0.0292, 0.0254, 0.1020, 0.0133, 0.0260, 0.0493, 0.0438], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0109, 0.0099, 0.0136, 0.0085, 0.0127, 0.0127, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-02 20:15:45,758 INFO [train.py:904] (7/8) Epoch 29, batch 9650, loss[loss=0.1788, simple_loss=0.2724, pruned_loss=0.0426, over 15347.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2602, pruned_loss=0.03453, over 3046053.38 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:16:47,400 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6158, 2.9373, 3.3318, 1.9928, 2.8526, 2.1823, 3.1589, 3.1125], device='cuda:7'), covar=tensor([0.0296, 0.0924, 0.0526, 0.2222, 0.0839, 0.1082, 0.0697, 0.1060], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0165, 0.0167, 0.0154, 0.0144, 0.0130, 0.0142, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:17:16,480 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 2.271e+02 2.644e+02 3.266e+02 5.877e+02, threshold=5.288e+02, percent-clipped=1.0 2023-05-02 20:17:19,745 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6089, 3.7528, 2.3019, 4.2232, 2.8420, 4.1221, 2.3810, 2.9640], device='cuda:7'), covar=tensor([0.0328, 0.0393, 0.1694, 0.0244, 0.0926, 0.0544, 0.1686, 0.0873], device='cuda:7'), in_proj_covar=tensor([0.0171, 0.0174, 0.0190, 0.0166, 0.0175, 0.0212, 0.0200, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:17:35,789 INFO [train.py:904] (7/8) Epoch 29, batch 9700, loss[loss=0.1694, simple_loss=0.2519, pruned_loss=0.04349, over 12487.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2598, pruned_loss=0.03455, over 3051472.41 frames. ], batch size: 248, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:18:10,754 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 20:19:17,224 INFO [train.py:904] (7/8) Epoch 29, batch 9750, loss[loss=0.1605, simple_loss=0.257, pruned_loss=0.03202, over 15349.00 frames. ], tot_loss[loss=0.164, simple_loss=0.2587, pruned_loss=0.03467, over 3067300.33 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:20:25,364 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=293987.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:20:38,044 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 2.282e+02 2.658e+02 3.259e+02 1.207e+03, threshold=5.315e+02, percent-clipped=3.0 2023-05-02 20:20:56,837 INFO [train.py:904] (7/8) Epoch 29, batch 9800, loss[loss=0.1473, simple_loss=0.2529, pruned_loss=0.02088, over 16434.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.259, pruned_loss=0.0338, over 3080870.67 frames. ], batch size: 68, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:21:10,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0577, 3.0668, 1.9043, 3.3084, 2.3175, 3.3125, 2.1815, 2.5880], device='cuda:7'), covar=tensor([0.0368, 0.0441, 0.1708, 0.0262, 0.0941, 0.0668, 0.1596, 0.0869], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0175, 0.0191, 0.0167, 0.0176, 0.0213, 0.0201, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:21:22,977 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294017.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:21:24,700 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0793, 4.1563, 3.9983, 3.7306, 3.7984, 4.0801, 3.7593, 3.8711], device='cuda:7'), covar=tensor([0.0551, 0.0604, 0.0313, 0.0266, 0.0608, 0.0439, 0.0974, 0.0542], device='cuda:7'), in_proj_covar=tensor([0.0301, 0.0455, 0.0353, 0.0352, 0.0347, 0.0404, 0.0243, 0.0420], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:21:27,518 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294020.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:21:33,322 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-05-02 20:22:29,388 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294048.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:22:39,776 INFO [train.py:904] (7/8) Epoch 29, batch 9850, loss[loss=0.162, simple_loss=0.2589, pruned_loss=0.03253, over 15259.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2601, pruned_loss=0.03358, over 3091563.63 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 16.0 2023-05-02 20:23:02,391 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294065.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:23:09,658 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294068.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:24:14,911 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 2.054e+02 2.475e+02 2.961e+02 4.915e+02, threshold=4.950e+02, percent-clipped=0.0 2023-05-02 20:24:32,433 INFO [train.py:904] (7/8) Epoch 29, batch 9900, loss[loss=0.1631, simple_loss=0.2546, pruned_loss=0.03582, over 12827.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.26, pruned_loss=0.03335, over 3082007.09 frames. ], batch size: 250, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:25:24,255 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 20:25:45,646 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.67 vs. limit=2.0 2023-05-02 20:26:02,235 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8524, 3.7501, 3.9915, 3.7502, 3.9502, 4.3192, 4.0007, 3.7122], device='cuda:7'), covar=tensor([0.2002, 0.2796, 0.2268, 0.2453, 0.2632, 0.1736, 0.1750, 0.2507], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0616, 0.0686, 0.0501, 0.0667, 0.0707, 0.0532, 0.0665], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 20:26:27,782 INFO [train.py:904] (7/8) Epoch 29, batch 9950, loss[loss=0.1417, simple_loss=0.2491, pruned_loss=0.01716, over 16886.00 frames. ], tot_loss[loss=0.1652, simple_loss=0.2623, pruned_loss=0.03401, over 3074970.02 frames. ], batch size: 102, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:26:32,285 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3978, 2.4364, 2.4617, 4.0897, 2.3093, 2.7453, 2.4198, 2.5353], device='cuda:7'), covar=tensor([0.1263, 0.3607, 0.3074, 0.0499, 0.3825, 0.2470, 0.4056, 0.2983], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0465, 0.0380, 0.0328, 0.0437, 0.0531, 0.0439, 0.0544], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:27:07,209 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294170.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:28:08,142 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.053e+02 2.347e+02 2.830e+02 4.908e+02, threshold=4.695e+02, percent-clipped=0.0 2023-05-02 20:28:27,682 INFO [train.py:904] (7/8) Epoch 29, batch 10000, loss[loss=0.164, simple_loss=0.2679, pruned_loss=0.03007, over 15241.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2613, pruned_loss=0.03358, over 3093163.21 frames. ], batch size: 191, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:28:41,427 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5638, 3.2811, 3.6049, 2.0281, 3.7576, 3.8597, 3.0602, 3.0919], device='cuda:7'), covar=tensor([0.0710, 0.0277, 0.0220, 0.1118, 0.0092, 0.0143, 0.0390, 0.0405], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0108, 0.0098, 0.0135, 0.0084, 0.0126, 0.0126, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 20:28:49,998 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 20:29:21,866 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294231.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:29:48,574 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0327, 2.2178, 2.2033, 3.6533, 2.1066, 2.5306, 2.3502, 2.3455], device='cuda:7'), covar=tensor([0.1440, 0.3737, 0.3332, 0.0616, 0.4423, 0.2713, 0.3825, 0.3570], device='cuda:7'), in_proj_covar=tensor([0.0411, 0.0463, 0.0379, 0.0326, 0.0436, 0.0529, 0.0437, 0.0542], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:30:08,535 INFO [train.py:904] (7/8) Epoch 29, batch 10050, loss[loss=0.1586, simple_loss=0.2601, pruned_loss=0.0285, over 16808.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2617, pruned_loss=0.03345, over 3113123.05 frames. ], batch size: 83, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:31:20,499 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3753, 2.4924, 2.1426, 2.3378, 2.8000, 2.4747, 2.7321, 3.0132], device='cuda:7'), covar=tensor([0.0168, 0.0537, 0.0637, 0.0561, 0.0357, 0.0492, 0.0248, 0.0321], device='cuda:7'), in_proj_covar=tensor([0.0219, 0.0237, 0.0226, 0.0226, 0.0235, 0.0234, 0.0229, 0.0233], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:31:27,888 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.082e+02 2.498e+02 3.055e+02 5.920e+02, threshold=4.997e+02, percent-clipped=2.0 2023-05-02 20:31:41,189 INFO [train.py:904] (7/8) Epoch 29, batch 10100, loss[loss=0.1573, simple_loss=0.2492, pruned_loss=0.03269, over 16229.00 frames. ], tot_loss[loss=0.1634, simple_loss=0.2609, pruned_loss=0.033, over 3119986.67 frames. ], batch size: 165, lr: 2.30e-03, grad_scale: 8.0 2023-05-02 20:32:42,471 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294339.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:32:46,858 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294343.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:33:20,750 INFO [train.py:904] (7/8) Epoch 30, batch 0, loss[loss=0.2129, simple_loss=0.2922, pruned_loss=0.06685, over 16381.00 frames. ], tot_loss[loss=0.2129, simple_loss=0.2922, pruned_loss=0.06685, over 16381.00 frames. ], batch size: 165, lr: 2.26e-03, grad_scale: 8.0 2023-05-02 20:33:20,750 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 20:33:28,214 INFO [train.py:938] (7/8) Epoch 30, validation: loss=0.1426, simple_loss=0.2458, pruned_loss=0.01974, over 944034.00 frames. 2023-05-02 20:33:28,215 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 20:33:53,550 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-02 20:34:10,707 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9415, 3.2570, 2.8593, 5.2590, 4.2647, 4.4755, 1.8268, 3.4858], device='cuda:7'), covar=tensor([0.1378, 0.0725, 0.1239, 0.0182, 0.0285, 0.0458, 0.1699, 0.0743], device='cuda:7'), in_proj_covar=tensor([0.0172, 0.0178, 0.0198, 0.0199, 0.0200, 0.0214, 0.0209, 0.0196], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:34:28,827 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.544e+02 2.426e+02 2.887e+02 3.382e+02 7.415e+02, threshold=5.774e+02, percent-clipped=4.0 2023-05-02 20:34:30,113 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294400.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:34:36,211 INFO [train.py:904] (7/8) Epoch 30, batch 50, loss[loss=0.172, simple_loss=0.2716, pruned_loss=0.03618, over 17141.00 frames. ], tot_loss[loss=0.1755, simple_loss=0.2641, pruned_loss=0.04347, over 747943.15 frames. ], batch size: 48, lr: 2.26e-03, grad_scale: 1.0 2023-05-02 20:35:02,182 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294423.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:35:22,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5054, 1.7281, 2.1506, 2.2829, 2.4867, 2.4315, 1.8781, 2.5756], device='cuda:7'), covar=tensor([0.0220, 0.0587, 0.0370, 0.0384, 0.0366, 0.0375, 0.0627, 0.0243], device='cuda:7'), in_proj_covar=tensor([0.0193, 0.0194, 0.0183, 0.0187, 0.0206, 0.0162, 0.0200, 0.0163], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:35:43,886 INFO [train.py:904] (7/8) Epoch 30, batch 100, loss[loss=0.1866, simple_loss=0.2644, pruned_loss=0.05442, over 16857.00 frames. ], tot_loss[loss=0.174, simple_loss=0.2626, pruned_loss=0.04275, over 1320738.83 frames. ], batch size: 90, lr: 2.26e-03, grad_scale: 1.0 2023-05-02 20:35:48,585 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.3189, 5.3296, 5.1410, 4.7423, 5.1162, 2.1338, 4.9889, 5.0193], device='cuda:7'), covar=tensor([0.0104, 0.0102, 0.0243, 0.0361, 0.0118, 0.2547, 0.0137, 0.0240], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0170, 0.0208, 0.0178, 0.0186, 0.0215, 0.0196, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:36:24,575 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294484.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:36:32,244 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294488.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:36:46,207 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.392e+02 2.778e+02 3.445e+02 1.255e+03, threshold=5.557e+02, percent-clipped=5.0 2023-05-02 20:36:51,466 INFO [train.py:904] (7/8) Epoch 30, batch 150, loss[loss=0.1854, simple_loss=0.2767, pruned_loss=0.04708, over 15507.00 frames. ], tot_loss[loss=0.1709, simple_loss=0.2601, pruned_loss=0.04084, over 1768909.87 frames. ], batch size: 190, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:37:20,812 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294526.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:37:32,842 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-02 20:37:50,439 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5553, 3.5597, 2.1816, 3.7641, 2.8575, 3.7085, 2.3198, 2.9390], device='cuda:7'), covar=tensor([0.0308, 0.0470, 0.1671, 0.0445, 0.0787, 0.0907, 0.1511, 0.0750], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0179, 0.0194, 0.0170, 0.0178, 0.0217, 0.0204, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:37:53,411 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294549.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:37:59,964 INFO [train.py:904] (7/8) Epoch 30, batch 200, loss[loss=0.1957, simple_loss=0.2687, pruned_loss=0.0613, over 16900.00 frames. ], tot_loss[loss=0.1713, simple_loss=0.2601, pruned_loss=0.04128, over 2116060.19 frames. ], batch size: 116, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:38:23,526 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294571.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:38:41,590 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.9640, 4.1079, 2.9691, 4.7064, 3.3818, 4.6441, 2.9697, 3.6431], device='cuda:7'), covar=tensor([0.0391, 0.0460, 0.1447, 0.0354, 0.0849, 0.0573, 0.1444, 0.0709], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0179, 0.0194, 0.0171, 0.0179, 0.0217, 0.0204, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:39:01,687 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.189e+02 2.574e+02 3.148e+02 2.009e+03, threshold=5.148e+02, percent-clipped=2.0 2023-05-02 20:39:06,664 INFO [train.py:904] (7/8) Epoch 30, batch 250, loss[loss=0.1768, simple_loss=0.2723, pruned_loss=0.04063, over 16665.00 frames. ], tot_loss[loss=0.1701, simple_loss=0.2577, pruned_loss=0.04127, over 2384753.77 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:39:45,476 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=294632.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:40:01,782 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294643.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:40:02,963 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1048, 5.6937, 5.8374, 5.5307, 5.6140, 6.1673, 5.6435, 5.3403], device='cuda:7'), covar=tensor([0.1011, 0.1886, 0.2334, 0.2139, 0.2583, 0.0926, 0.1564, 0.2338], device='cuda:7'), in_proj_covar=tensor([0.0421, 0.0634, 0.0707, 0.0513, 0.0686, 0.0726, 0.0544, 0.0681], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 20:40:12,517 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9083, 2.8318, 2.7619, 5.0345, 4.0719, 4.4012, 1.6743, 3.2283], device='cuda:7'), covar=tensor([0.1343, 0.0836, 0.1207, 0.0202, 0.0196, 0.0390, 0.1718, 0.0799], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0181, 0.0201, 0.0204, 0.0204, 0.0217, 0.0212, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:40:15,321 INFO [train.py:904] (7/8) Epoch 30, batch 300, loss[loss=0.1502, simple_loss=0.2327, pruned_loss=0.03383, over 16808.00 frames. ], tot_loss[loss=0.1668, simple_loss=0.2542, pruned_loss=0.03968, over 2583780.95 frames. ], batch size: 102, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:41:05,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2905, 5.2185, 5.1175, 4.5863, 4.7747, 5.1475, 5.2077, 4.7701], device='cuda:7'), covar=tensor([0.0643, 0.0640, 0.0422, 0.0439, 0.1288, 0.0542, 0.0299, 0.0933], device='cuda:7'), in_proj_covar=tensor([0.0309, 0.0466, 0.0361, 0.0362, 0.0356, 0.0415, 0.0248, 0.0431], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:41:06,779 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294691.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:41:12,841 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294695.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:41:19,996 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.370e+02 1.963e+02 2.186e+02 2.662e+02 5.332e+02, threshold=4.373e+02, percent-clipped=1.0 2023-05-02 20:41:25,725 INFO [train.py:904] (7/8) Epoch 30, batch 350, loss[loss=0.1697, simple_loss=0.2695, pruned_loss=0.03495, over 17068.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2528, pruned_loss=0.03892, over 2742512.47 frames. ], batch size: 55, lr: 2.25e-03, grad_scale: 1.0 2023-05-02 20:42:33,030 INFO [train.py:904] (7/8) Epoch 30, batch 400, loss[loss=0.1717, simple_loss=0.2444, pruned_loss=0.04954, over 16845.00 frames. ], tot_loss[loss=0.1642, simple_loss=0.2512, pruned_loss=0.0386, over 2872090.29 frames. ], batch size: 96, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:42:55,457 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7309, 3.6435, 4.3691, 2.4494, 3.5480, 2.9360, 4.0668, 3.9372], device='cuda:7'), covar=tensor([0.0216, 0.1089, 0.0444, 0.2104, 0.0752, 0.0912, 0.0536, 0.1192], device='cuda:7'), in_proj_covar=tensor([0.0160, 0.0168, 0.0169, 0.0157, 0.0147, 0.0132, 0.0144, 0.0181], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:43:06,973 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294779.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:43:34,533 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.286e+02 2.158e+02 2.521e+02 2.985e+02 1.602e+03, threshold=5.041e+02, percent-clipped=2.0 2023-05-02 20:43:41,863 INFO [train.py:904] (7/8) Epoch 30, batch 450, loss[loss=0.1688, simple_loss=0.2639, pruned_loss=0.03685, over 16728.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2496, pruned_loss=0.03777, over 2967381.70 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:43:50,082 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8988, 2.3599, 2.3656, 3.6479, 2.8595, 3.8067, 1.6549, 2.8599], device='cuda:7'), covar=tensor([0.1456, 0.0919, 0.1460, 0.0262, 0.0149, 0.0433, 0.1809, 0.0954], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0181, 0.0200, 0.0203, 0.0203, 0.0217, 0.0211, 0.0199], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:44:12,107 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294826.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:44:16,244 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2940, 4.1256, 4.3505, 4.4765, 4.5535, 4.1259, 4.3644, 4.5407], device='cuda:7'), covar=tensor([0.1625, 0.1217, 0.1299, 0.0697, 0.0615, 0.1238, 0.3555, 0.0821], device='cuda:7'), in_proj_covar=tensor([0.0679, 0.0828, 0.0954, 0.0843, 0.0641, 0.0664, 0.0704, 0.0817], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:44:36,120 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294844.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:44:49,619 INFO [train.py:904] (7/8) Epoch 30, batch 500, loss[loss=0.1496, simple_loss=0.2315, pruned_loss=0.0338, over 11934.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2487, pruned_loss=0.03732, over 3043103.32 frames. ], batch size: 246, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:45:17,752 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=294874.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:45:52,586 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.033e+02 2.399e+02 2.919e+02 5.551e+02, threshold=4.798e+02, percent-clipped=2.0 2023-05-02 20:45:57,309 INFO [train.py:904] (7/8) Epoch 30, batch 550, loss[loss=0.1674, simple_loss=0.2658, pruned_loss=0.03451, over 16785.00 frames. ], tot_loss[loss=0.1605, simple_loss=0.2481, pruned_loss=0.03646, over 3101675.86 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:46:29,056 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=294927.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:47:00,860 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 20:47:05,467 INFO [train.py:904] (7/8) Epoch 30, batch 600, loss[loss=0.1744, simple_loss=0.2703, pruned_loss=0.03924, over 16730.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2482, pruned_loss=0.0376, over 3149169.58 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:47:18,802 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8000, 1.8560, 2.3545, 2.6399, 2.6305, 2.5823, 1.8954, 2.7844], device='cuda:7'), covar=tensor([0.0209, 0.0599, 0.0403, 0.0384, 0.0383, 0.0430, 0.0691, 0.0251], device='cuda:7'), in_proj_covar=tensor([0.0199, 0.0200, 0.0188, 0.0193, 0.0212, 0.0167, 0.0206, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 20:47:53,694 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=294989.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:48:01,974 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=294995.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:48:07,359 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.157e+02 2.572e+02 3.014e+02 4.443e+02, threshold=5.143e+02, percent-clipped=0.0 2023-05-02 20:48:13,122 INFO [train.py:904] (7/8) Epoch 30, batch 650, loss[loss=0.1713, simple_loss=0.2642, pruned_loss=0.03922, over 17098.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2476, pruned_loss=0.03717, over 3197943.38 frames. ], batch size: 47, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:48:42,025 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.7301, 6.1113, 5.8578, 5.9299, 5.4508, 5.5567, 5.4053, 6.2552], device='cuda:7'), covar=tensor([0.1422, 0.0966, 0.1096, 0.0974, 0.0988, 0.0643, 0.1480, 0.0954], device='cuda:7'), in_proj_covar=tensor([0.0723, 0.0868, 0.0716, 0.0677, 0.0555, 0.0553, 0.0732, 0.0684], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 20:49:04,985 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295043.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:49:15,943 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295050.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 20:49:20,779 INFO [train.py:904] (7/8) Epoch 30, batch 700, loss[loss=0.1549, simple_loss=0.2438, pruned_loss=0.03303, over 16701.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2474, pruned_loss=0.03708, over 3222541.87 frames. ], batch size: 89, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:49:39,362 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295068.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:49:54,489 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295079.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:50:03,930 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 20:50:23,317 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.027e+02 2.322e+02 2.790e+02 6.383e+02, threshold=4.644e+02, percent-clipped=2.0 2023-05-02 20:50:27,893 INFO [train.py:904] (7/8) Epoch 30, batch 750, loss[loss=0.1584, simple_loss=0.2395, pruned_loss=0.03867, over 12183.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2474, pruned_loss=0.03698, over 3236924.31 frames. ], batch size: 248, lr: 2.25e-03, grad_scale: 2.0 2023-05-02 20:50:56,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4252, 3.3735, 3.6106, 2.5873, 3.3221, 3.6989, 3.4557, 2.1251], device='cuda:7'), covar=tensor([0.0542, 0.0259, 0.0075, 0.0432, 0.0150, 0.0129, 0.0119, 0.0565], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0092, 0.0093, 0.0137, 0.0104, 0.0117, 0.0100, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 20:51:00,101 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295127.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:51:03,390 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295129.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:51:24,565 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295144.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:51:38,210 INFO [train.py:904] (7/8) Epoch 30, batch 800, loss[loss=0.1549, simple_loss=0.2364, pruned_loss=0.03669, over 16872.00 frames. ], tot_loss[loss=0.16, simple_loss=0.2467, pruned_loss=0.03667, over 3252081.84 frames. ], batch size: 90, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:52:29,870 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295192.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:52:42,512 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.068e+02 2.393e+02 2.794e+02 7.464e+02, threshold=4.786e+02, percent-clipped=2.0 2023-05-02 20:52:43,780 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-05-02 20:52:48,748 INFO [train.py:904] (7/8) Epoch 30, batch 850, loss[loss=0.1658, simple_loss=0.2525, pruned_loss=0.03956, over 16807.00 frames. ], tot_loss[loss=0.1597, simple_loss=0.2463, pruned_loss=0.03658, over 3267326.13 frames. ], batch size: 102, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:53:18,848 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295225.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:53:21,293 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295227.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:53:56,733 INFO [train.py:904] (7/8) Epoch 30, batch 900, loss[loss=0.1536, simple_loss=0.2361, pruned_loss=0.03554, over 16798.00 frames. ], tot_loss[loss=0.1589, simple_loss=0.2459, pruned_loss=0.036, over 3288314.16 frames. ], batch size: 102, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:54:27,487 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295275.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:54:34,262 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295279.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:54:43,633 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295286.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:54:46,322 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.41 vs. limit=5.0 2023-05-02 20:54:47,575 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-02 20:55:02,821 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.063e+02 2.465e+02 2.923e+02 4.814e+02, threshold=4.929e+02, percent-clipped=1.0 2023-05-02 20:55:08,087 INFO [train.py:904] (7/8) Epoch 30, batch 950, loss[loss=0.1603, simple_loss=0.248, pruned_loss=0.03628, over 16711.00 frames. ], tot_loss[loss=0.1598, simple_loss=0.2464, pruned_loss=0.03661, over 3303164.54 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:55:49,196 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295333.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:55:59,239 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295340.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:56:06,349 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295345.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 20:56:17,482 INFO [train.py:904] (7/8) Epoch 30, batch 1000, loss[loss=0.1414, simple_loss=0.232, pruned_loss=0.02542, over 17172.00 frames. ], tot_loss[loss=0.1591, simple_loss=0.2455, pruned_loss=0.03641, over 3299404.28 frames. ], batch size: 46, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:56:24,373 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295359.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:57:11,788 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2197, 3.9107, 4.3432, 2.2314, 4.5068, 4.7259, 3.5086, 3.5990], device='cuda:7'), covar=tensor([0.0727, 0.0301, 0.0276, 0.1241, 0.0107, 0.0183, 0.0466, 0.0456], device='cuda:7'), in_proj_covar=tensor([0.0148, 0.0111, 0.0102, 0.0139, 0.0087, 0.0132, 0.0130, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 20:57:12,978 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295394.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:57:20,047 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.691e+02 2.023e+02 2.384e+02 2.929e+02 5.067e+02, threshold=4.769e+02, percent-clipped=1.0 2023-05-02 20:57:26,448 INFO [train.py:904] (7/8) Epoch 30, batch 1050, loss[loss=0.1383, simple_loss=0.2384, pruned_loss=0.01916, over 17127.00 frames. ], tot_loss[loss=0.1583, simple_loss=0.2448, pruned_loss=0.03585, over 3304970.92 frames. ], batch size: 47, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:57:48,321 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295420.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:57:54,262 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295424.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 20:58:36,221 INFO [train.py:904] (7/8) Epoch 30, batch 1100, loss[loss=0.1503, simple_loss=0.2398, pruned_loss=0.03044, over 17199.00 frames. ], tot_loss[loss=0.158, simple_loss=0.2443, pruned_loss=0.03583, over 3319893.06 frames. ], batch size: 44, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 20:59:38,547 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.061e+02 2.413e+02 2.897e+02 1.352e+03, threshold=4.827e+02, percent-clipped=9.0 2023-05-02 20:59:43,313 INFO [train.py:904] (7/8) Epoch 30, batch 1150, loss[loss=0.1496, simple_loss=0.2275, pruned_loss=0.03588, over 16881.00 frames. ], tot_loss[loss=0.1579, simple_loss=0.2447, pruned_loss=0.03559, over 3320055.71 frames. ], batch size: 96, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:00:45,283 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.7213, 1.8349, 1.6770, 1.5378, 1.9399, 1.6089, 1.5734, 1.9126], device='cuda:7'), covar=tensor([0.0275, 0.0394, 0.0503, 0.0482, 0.0275, 0.0366, 0.0235, 0.0299], device='cuda:7'), in_proj_covar=tensor([0.0239, 0.0252, 0.0239, 0.0241, 0.0250, 0.0249, 0.0247, 0.0250], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:00:51,769 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0307, 4.3914, 3.0880, 2.5033, 2.7743, 2.6991, 4.8094, 3.6087], device='cuda:7'), covar=tensor([0.2801, 0.0592, 0.1930, 0.3052, 0.2995, 0.2182, 0.0376, 0.1580], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0278, 0.0316, 0.0331, 0.0308, 0.0282, 0.0307, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:00:52,306 INFO [train.py:904] (7/8) Epoch 30, batch 1200, loss[loss=0.1605, simple_loss=0.2454, pruned_loss=0.03783, over 16834.00 frames. ], tot_loss[loss=0.1572, simple_loss=0.2432, pruned_loss=0.03563, over 3311252.57 frames. ], batch size: 42, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:00:55,681 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295556.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:00:58,015 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295558.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:01:02,504 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-05-02 21:01:29,942 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295581.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:01:55,321 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.025e+02 2.431e+02 2.981e+02 5.369e+02, threshold=4.863e+02, percent-clipped=2.0 2023-05-02 21:02:00,811 INFO [train.py:904] (7/8) Epoch 30, batch 1250, loss[loss=0.1717, simple_loss=0.2359, pruned_loss=0.05381, over 16780.00 frames. ], tot_loss[loss=0.1581, simple_loss=0.2431, pruned_loss=0.03654, over 3314799.57 frames. ], batch size: 83, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:02:08,755 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0662, 3.0540, 3.3557, 2.1470, 2.9333, 2.2888, 3.5388, 3.4329], device='cuda:7'), covar=tensor([0.0251, 0.1007, 0.0604, 0.2063, 0.0928, 0.1107, 0.0582, 0.0984], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0171, 0.0171, 0.0158, 0.0149, 0.0133, 0.0146, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 21:02:19,327 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295617.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:02:21,425 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295619.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:02:34,283 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295629.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:02:43,102 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295635.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:02:58,111 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295645.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:03:08,465 INFO [train.py:904] (7/8) Epoch 30, batch 1300, loss[loss=0.1577, simple_loss=0.2306, pruned_loss=0.04239, over 16780.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2425, pruned_loss=0.03647, over 3315453.69 frames. ], batch size: 83, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:03:52,669 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295685.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:03:57,773 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295689.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:03:59,052 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295690.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:04:02,885 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295693.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:04:11,930 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.150e+02 2.443e+02 2.917e+02 6.484e+02, threshold=4.887e+02, percent-clipped=2.0 2023-05-02 21:04:18,067 INFO [train.py:904] (7/8) Epoch 30, batch 1350, loss[loss=0.1319, simple_loss=0.2138, pruned_loss=0.02497, over 15840.00 frames. ], tot_loss[loss=0.1569, simple_loss=0.2422, pruned_loss=0.03577, over 3319557.70 frames. ], batch size: 35, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:04:33,169 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295715.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:04:46,300 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295724.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:05:17,231 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295746.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:05:26,459 INFO [train.py:904] (7/8) Epoch 30, batch 1400, loss[loss=0.1562, simple_loss=0.2389, pruned_loss=0.03672, over 17222.00 frames. ], tot_loss[loss=0.1568, simple_loss=0.2422, pruned_loss=0.03576, over 3314866.08 frames. ], batch size: 44, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:05:52,150 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295772.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:06:21,360 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.09 vs. limit=2.0 2023-05-02 21:06:30,030 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 2.078e+02 2.399e+02 2.795e+02 5.908e+02, threshold=4.799e+02, percent-clipped=2.0 2023-05-02 21:06:35,134 INFO [train.py:904] (7/8) Epoch 30, batch 1450, loss[loss=0.1459, simple_loss=0.2417, pruned_loss=0.02508, over 17102.00 frames. ], tot_loss[loss=0.1564, simple_loss=0.2418, pruned_loss=0.03546, over 3321193.56 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:06:57,646 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295820.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:07:44,322 INFO [train.py:904] (7/8) Epoch 30, batch 1500, loss[loss=0.1665, simple_loss=0.2434, pruned_loss=0.0448, over 16913.00 frames. ], tot_loss[loss=0.1563, simple_loss=0.2417, pruned_loss=0.03551, over 3313730.65 frames. ], batch size: 109, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:08:11,970 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=295873.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:08:19,146 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3791, 2.3946, 2.4374, 4.1708, 2.3261, 2.8014, 2.4691, 2.5406], device='cuda:7'), covar=tensor([0.1543, 0.3917, 0.3427, 0.0637, 0.4468, 0.2761, 0.3897, 0.4106], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0483, 0.0394, 0.0344, 0.0451, 0.0553, 0.0456, 0.0567], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:08:22,737 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:08:22,770 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295881.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:08:47,916 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.156e+02 2.462e+02 3.031e+02 5.873e+02, threshold=4.924e+02, percent-clipped=1.0 2023-05-02 21:08:54,107 INFO [train.py:904] (7/8) Epoch 30, batch 1550, loss[loss=0.1771, simple_loss=0.2632, pruned_loss=0.0455, over 15526.00 frames. ], tot_loss[loss=0.1577, simple_loss=0.2429, pruned_loss=0.03628, over 3322421.61 frames. ], batch size: 191, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:09:06,256 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295912.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:09:08,351 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295914.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:09:29,568 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295929.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:09:36,383 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=295934.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:09:37,395 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295935.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:10:02,950 INFO [train.py:904] (7/8) Epoch 30, batch 1600, loss[loss=0.1566, simple_loss=0.2494, pruned_loss=0.0319, over 16836.00 frames. ], tot_loss[loss=0.1596, simple_loss=0.2451, pruned_loss=0.03711, over 3331184.06 frames. ], batch size: 42, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:10:36,693 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8009, 2.7226, 2.6684, 4.9328, 3.9186, 4.2935, 1.7592, 3.1818], device='cuda:7'), covar=tensor([0.1381, 0.0880, 0.1266, 0.0254, 0.0231, 0.0405, 0.1617, 0.0790], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0184, 0.0203, 0.0209, 0.0208, 0.0221, 0.0214, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:10:42,973 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=295983.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:10:45,981 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=295985.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:10:51,154 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=295989.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:11:05,448 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.204e+02 2.497e+02 2.999e+02 5.396e+02, threshold=4.994e+02, percent-clipped=1.0 2023-05-02 21:11:14,075 INFO [train.py:904] (7/8) Epoch 30, batch 1650, loss[loss=0.1815, simple_loss=0.2706, pruned_loss=0.0462, over 16629.00 frames. ], tot_loss[loss=0.1606, simple_loss=0.2463, pruned_loss=0.03745, over 3324184.53 frames. ], batch size: 62, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:11:30,769 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296015.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:00,962 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296037.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:06,277 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296041.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:16,943 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296049.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:12:24,451 INFO [train.py:904] (7/8) Epoch 30, batch 1700, loss[loss=0.2151, simple_loss=0.2987, pruned_loss=0.06577, over 12231.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2485, pruned_loss=0.03789, over 3318479.14 frames. ], batch size: 247, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:12:36,862 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296063.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:13:26,988 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.214e+02 2.660e+02 3.287e+02 5.248e+02, threshold=5.320e+02, percent-clipped=2.0 2023-05-02 21:13:32,347 INFO [train.py:904] (7/8) Epoch 30, batch 1750, loss[loss=0.1364, simple_loss=0.2234, pruned_loss=0.02475, over 16813.00 frames. ], tot_loss[loss=0.1625, simple_loss=0.2497, pruned_loss=0.03769, over 3314827.12 frames. ], batch size: 39, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:13:40,818 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296110.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:13:42,381 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.64 vs. limit=5.0 2023-05-02 21:14:24,292 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8488, 2.5707, 2.5566, 4.0641, 3.2323, 3.9954, 1.6214, 2.9592], device='cuda:7'), covar=tensor([0.1415, 0.0779, 0.1194, 0.0175, 0.0131, 0.0396, 0.1650, 0.0837], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0184, 0.0203, 0.0209, 0.0208, 0.0221, 0.0214, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:14:32,729 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8312, 2.6529, 2.6652, 4.0648, 3.2865, 4.0266, 1.7047, 2.9359], device='cuda:7'), covar=tensor([0.1449, 0.0746, 0.1140, 0.0184, 0.0142, 0.0388, 0.1648, 0.0877], device='cuda:7'), in_proj_covar=tensor([0.0176, 0.0185, 0.0203, 0.0209, 0.0208, 0.0221, 0.0214, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:14:41,103 INFO [train.py:904] (7/8) Epoch 30, batch 1800, loss[loss=0.1413, simple_loss=0.2331, pruned_loss=0.02478, over 17184.00 frames. ], tot_loss[loss=0.1626, simple_loss=0.2501, pruned_loss=0.03753, over 3312547.83 frames. ], batch size: 46, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:15:13,719 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296176.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:15:18,315 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 21:15:47,289 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 2.165e+02 2.480e+02 2.953e+02 5.408e+02, threshold=4.959e+02, percent-clipped=1.0 2023-05-02 21:15:51,606 INFO [train.py:904] (7/8) Epoch 30, batch 1850, loss[loss=0.1675, simple_loss=0.2467, pruned_loss=0.04411, over 16688.00 frames. ], tot_loss[loss=0.163, simple_loss=0.2509, pruned_loss=0.03761, over 3311912.90 frames. ], batch size: 134, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:16:03,843 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296212.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:16:06,028 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296214.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:16:28,000 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296229.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 21:16:57,128 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1105, 5.4398, 5.2159, 5.2378, 4.9873, 4.9506, 4.8009, 5.5644], device='cuda:7'), covar=tensor([0.1406, 0.0874, 0.1077, 0.0922, 0.0844, 0.0957, 0.1357, 0.0853], device='cuda:7'), in_proj_covar=tensor([0.0740, 0.0891, 0.0734, 0.0695, 0.0570, 0.0564, 0.0750, 0.0702], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:17:02,031 INFO [train.py:904] (7/8) Epoch 30, batch 1900, loss[loss=0.1722, simple_loss=0.2711, pruned_loss=0.03658, over 16637.00 frames. ], tot_loss[loss=0.162, simple_loss=0.25, pruned_loss=0.03696, over 3313656.61 frames. ], batch size: 62, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:17:05,283 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-05-02 21:17:10,075 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296260.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:17:13,524 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296262.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:17:42,741 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-05-02 21:17:46,280 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296285.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:17:56,045 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-05-02 21:18:06,423 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.062e+02 2.316e+02 2.814e+02 5.874e+02, threshold=4.633e+02, percent-clipped=2.0 2023-05-02 21:18:10,614 INFO [train.py:904] (7/8) Epoch 30, batch 1950, loss[loss=0.1565, simple_loss=0.2429, pruned_loss=0.03503, over 15801.00 frames. ], tot_loss[loss=0.1623, simple_loss=0.2504, pruned_loss=0.03715, over 3304323.48 frames. ], batch size: 35, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:18:27,073 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296315.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:18:52,567 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296333.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:19:02,356 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296341.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:19:11,725 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7709, 1.9229, 2.3808, 2.6301, 2.7049, 2.6757, 2.0310, 2.9036], device='cuda:7'), covar=tensor([0.0244, 0.0624, 0.0413, 0.0367, 0.0389, 0.0433, 0.0637, 0.0228], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0202, 0.0191, 0.0196, 0.0214, 0.0171, 0.0207, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 21:19:21,061 INFO [train.py:904] (7/8) Epoch 30, batch 2000, loss[loss=0.1414, simple_loss=0.2229, pruned_loss=0.02998, over 16793.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.2502, pruned_loss=0.03714, over 3306231.76 frames. ], batch size: 102, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:19:37,585 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.46 vs. limit=2.0 2023-05-02 21:19:51,161 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296376.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:20:01,465 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6115, 3.6944, 2.5246, 4.4038, 3.0725, 4.3294, 2.6053, 3.2043], device='cuda:7'), covar=tensor([0.0384, 0.0501, 0.1539, 0.0325, 0.0797, 0.0527, 0.1492, 0.0751], device='cuda:7'), in_proj_covar=tensor([0.0180, 0.0186, 0.0198, 0.0179, 0.0183, 0.0226, 0.0209, 0.0186], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:20:09,547 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296389.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:20:23,096 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7131, 4.6660, 4.6168, 4.0597, 4.6600, 2.0621, 4.4538, 4.2469], device='cuda:7'), covar=tensor([0.0184, 0.0141, 0.0206, 0.0333, 0.0131, 0.2830, 0.0153, 0.0272], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0179, 0.0217, 0.0188, 0.0196, 0.0223, 0.0207, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:20:25,434 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.133e+02 2.447e+02 2.850e+02 4.858e+02, threshold=4.894e+02, percent-clipped=1.0 2023-05-02 21:20:29,600 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9287, 2.1522, 2.6794, 2.9257, 2.7673, 3.4308, 2.5390, 3.4554], device='cuda:7'), covar=tensor([0.0342, 0.0632, 0.0414, 0.0434, 0.0465, 0.0243, 0.0558, 0.0200], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0202, 0.0191, 0.0197, 0.0214, 0.0171, 0.0207, 0.0171], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 21:20:30,272 INFO [train.py:904] (7/8) Epoch 30, batch 2050, loss[loss=0.154, simple_loss=0.2498, pruned_loss=0.0291, over 17188.00 frames. ], tot_loss[loss=0.1621, simple_loss=0.25, pruned_loss=0.03706, over 3312698.34 frames. ], batch size: 46, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:20:32,333 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296405.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:21:12,419 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296434.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:21:39,722 INFO [train.py:904] (7/8) Epoch 30, batch 2100, loss[loss=0.1939, simple_loss=0.2786, pruned_loss=0.05465, over 11813.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2506, pruned_loss=0.03763, over 3306532.84 frames. ], batch size: 248, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:21:43,795 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5930, 4.5365, 4.4595, 3.8977, 4.5064, 1.8625, 4.2935, 4.0192], device='cuda:7'), covar=tensor([0.0175, 0.0157, 0.0217, 0.0332, 0.0135, 0.3026, 0.0166, 0.0273], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0179, 0.0217, 0.0187, 0.0195, 0.0222, 0.0207, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:22:09,617 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296476.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:33,179 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296492.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:36,852 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296495.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:39,127 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296497.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:22:46,676 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.431e+02 2.088e+02 2.487e+02 2.838e+02 5.578e+02, threshold=4.974e+02, percent-clipped=1.0 2023-05-02 21:22:48,904 INFO [train.py:904] (7/8) Epoch 30, batch 2150, loss[loss=0.1672, simple_loss=0.2497, pruned_loss=0.04236, over 16807.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2512, pruned_loss=0.03831, over 3307533.07 frames. ], batch size: 102, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:23:03,763 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296514.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:23:16,240 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296524.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:23:23,552 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296529.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:23:35,337 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0935, 5.1691, 5.5333, 5.5168, 5.5622, 5.2373, 5.1732, 4.9477], device='cuda:7'), covar=tensor([0.0374, 0.0569, 0.0425, 0.0433, 0.0506, 0.0364, 0.0939, 0.0467], device='cuda:7'), in_proj_covar=tensor([0.0448, 0.0510, 0.0487, 0.0451, 0.0534, 0.0517, 0.0593, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 21:23:43,109 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7566, 4.8158, 5.0066, 4.8094, 4.8090, 5.4606, 4.9633, 4.6590], device='cuda:7'), covar=tensor([0.1495, 0.2221, 0.2305, 0.2292, 0.2866, 0.1099, 0.1776, 0.2506], device='cuda:7'), in_proj_covar=tensor([0.0438, 0.0654, 0.0731, 0.0532, 0.0710, 0.0749, 0.0562, 0.0706], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:23:56,944 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296553.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:23:57,540 INFO [train.py:904] (7/8) Epoch 30, batch 2200, loss[loss=0.164, simple_loss=0.2455, pruned_loss=0.04121, over 15842.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2514, pruned_loss=0.03855, over 3320791.70 frames. ], batch size: 35, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:24:04,159 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296558.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:24:13,080 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9429, 2.1659, 2.2231, 3.4812, 2.1361, 2.4211, 2.2649, 2.2996], device='cuda:7'), covar=tensor([0.1711, 0.3744, 0.3438, 0.0842, 0.4276, 0.2837, 0.3917, 0.3516], device='cuda:7'), in_proj_covar=tensor([0.0431, 0.0484, 0.0395, 0.0346, 0.0451, 0.0555, 0.0457, 0.0568], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:24:27,052 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296575.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:24:30,754 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296577.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:25:04,060 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.512e+02 2.151e+02 2.512e+02 3.134e+02 5.584e+02, threshold=5.024e+02, percent-clipped=2.0 2023-05-02 21:25:06,332 INFO [train.py:904] (7/8) Epoch 30, batch 2250, loss[loss=0.2299, simple_loss=0.3071, pruned_loss=0.07632, over 11818.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2522, pruned_loss=0.03879, over 3322706.40 frames. ], batch size: 246, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:25:20,050 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:25:31,248 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.21 vs. limit=5.0 2023-05-02 21:26:17,736 INFO [train.py:904] (7/8) Epoch 30, batch 2300, loss[loss=0.1689, simple_loss=0.2442, pruned_loss=0.04683, over 16756.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2529, pruned_loss=0.03893, over 3319733.10 frames. ], batch size: 124, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:26:41,411 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296671.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:26:45,195 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296674.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:27:05,648 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296688.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:27:24,077 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.179e+02 2.509e+02 3.030e+02 4.955e+02, threshold=5.019e+02, percent-clipped=0.0 2023-05-02 21:27:27,335 INFO [train.py:904] (7/8) Epoch 30, batch 2350, loss[loss=0.1767, simple_loss=0.2613, pruned_loss=0.04601, over 16689.00 frames. ], tot_loss[loss=0.1655, simple_loss=0.2535, pruned_loss=0.03877, over 3322033.87 frames. ], batch size: 134, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:27:28,836 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296705.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:28:30,545 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0305, 4.6622, 4.6297, 3.1569, 3.7920, 4.5923, 3.9911, 2.9119], device='cuda:7'), covar=tensor([0.0512, 0.0065, 0.0047, 0.0433, 0.0166, 0.0093, 0.0109, 0.0457], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0136, 0.0105, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 21:28:30,588 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296749.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:28:36,423 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=296753.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:28:37,280 INFO [train.py:904] (7/8) Epoch 30, batch 2400, loss[loss=0.1527, simple_loss=0.2433, pruned_loss=0.03103, over 16811.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.2536, pruned_loss=0.03863, over 3327441.57 frames. ], batch size: 39, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:28:54,536 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 21:29:27,153 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296790.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:29:43,968 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.201e+02 2.707e+02 3.099e+02 6.549e+02, threshold=5.413e+02, percent-clipped=3.0 2023-05-02 21:29:46,350 INFO [train.py:904] (7/8) Epoch 30, batch 2450, loss[loss=0.1727, simple_loss=0.271, pruned_loss=0.03719, over 17039.00 frames. ], tot_loss[loss=0.165, simple_loss=0.2538, pruned_loss=0.03806, over 3324315.62 frames. ], batch size: 53, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:30:10,949 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2440, 4.1821, 4.1629, 3.8846, 3.9676, 4.2207, 3.9091, 4.0345], device='cuda:7'), covar=tensor([0.0627, 0.0790, 0.0333, 0.0302, 0.0674, 0.0466, 0.0805, 0.0642], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0499, 0.0385, 0.0388, 0.0382, 0.0445, 0.0265, 0.0461], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:30:48,334 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296848.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:30:51,430 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4531, 3.2412, 2.6522, 2.1655, 2.2616, 2.2920, 3.3640, 2.9789], device='cuda:7'), covar=tensor([0.3085, 0.0801, 0.2015, 0.3146, 0.2838, 0.2325, 0.0699, 0.1728], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0278, 0.0317, 0.0330, 0.0309, 0.0282, 0.0307, 0.0357], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:30:55,453 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296853.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:30:56,392 INFO [train.py:904] (7/8) Epoch 30, batch 2500, loss[loss=0.1474, simple_loss=0.2348, pruned_loss=0.03, over 15957.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2539, pruned_loss=0.0377, over 3329827.14 frames. ], batch size: 35, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:31:19,668 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296870.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:31:24,748 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296873.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:31:59,209 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7417, 3.4267, 3.8211, 1.9828, 3.8727, 3.9254, 3.1904, 3.0125], device='cuda:7'), covar=tensor([0.0804, 0.0274, 0.0194, 0.1274, 0.0131, 0.0222, 0.0441, 0.0437], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0140, 0.0088, 0.0134, 0.0132, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:32:04,916 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.483e+02 2.203e+02 2.647e+02 3.147e+02 6.132e+02, threshold=5.294e+02, percent-clipped=2.0 2023-05-02 21:32:07,804 INFO [train.py:904] (7/8) Epoch 30, batch 2550, loss[loss=0.1808, simple_loss=0.2657, pruned_loss=0.04793, over 16268.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2539, pruned_loss=0.03791, over 3329295.65 frames. ], batch size: 165, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:32:49,131 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296934.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 21:32:49,247 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8892, 4.4658, 3.1677, 2.4478, 2.6662, 2.7482, 4.8127, 3.6092], device='cuda:7'), covar=tensor([0.2951, 0.0490, 0.1845, 0.3118, 0.3186, 0.2161, 0.0354, 0.1558], device='cuda:7'), in_proj_covar=tensor([0.0340, 0.0278, 0.0317, 0.0330, 0.0310, 0.0283, 0.0307, 0.0358], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:32:54,150 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296938.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:33:15,503 INFO [train.py:904] (7/8) Epoch 30, batch 2600, loss[loss=0.1593, simple_loss=0.2445, pruned_loss=0.03701, over 16677.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.2536, pruned_loss=0.03756, over 3335203.59 frames. ], batch size: 89, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:33:36,745 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=296969.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:33:39,727 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=296971.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:34:18,414 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=296999.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:34:22,262 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.341e+02 2.007e+02 2.417e+02 2.925e+02 5.072e+02, threshold=4.833e+02, percent-clipped=0.0 2023-05-02 21:34:24,428 INFO [train.py:904] (7/8) Epoch 30, batch 2650, loss[loss=0.1691, simple_loss=0.266, pruned_loss=0.03605, over 17219.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2543, pruned_loss=0.03738, over 3341915.18 frames. ], batch size: 44, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:34:46,004 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297019.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:34:57,935 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-05-02 21:35:13,099 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.2329, 3.4465, 3.6440, 2.4617, 3.2956, 3.7285, 3.4373, 2.1184], device='cuda:7'), covar=tensor([0.0602, 0.0153, 0.0067, 0.0455, 0.0145, 0.0100, 0.0108, 0.0557], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0092, 0.0093, 0.0137, 0.0105, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 21:35:20,548 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297044.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:35:34,417 INFO [train.py:904] (7/8) Epoch 30, batch 2700, loss[loss=0.1566, simple_loss=0.2539, pruned_loss=0.02967, over 17109.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2546, pruned_loss=0.03714, over 3345932.63 frames. ], batch size: 49, lr: 2.25e-03, grad_scale: 8.0 2023-05-02 21:36:20,064 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7805, 3.9678, 2.5598, 4.5348, 3.1587, 4.4634, 2.6125, 3.3216], device='cuda:7'), covar=tensor([0.0410, 0.0431, 0.1652, 0.0383, 0.0826, 0.0560, 0.1655, 0.0774], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0186, 0.0200, 0.0180, 0.0185, 0.0227, 0.0210, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:36:24,008 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297090.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:36:42,011 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.120e+02 2.454e+02 2.893e+02 5.534e+02, threshold=4.909e+02, percent-clipped=1.0 2023-05-02 21:36:43,154 INFO [train.py:904] (7/8) Epoch 30, batch 2750, loss[loss=0.177, simple_loss=0.2682, pruned_loss=0.04284, over 16700.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2549, pruned_loss=0.03641, over 3346744.41 frames. ], batch size: 57, lr: 2.25e-03, grad_scale: 4.0 2023-05-02 21:37:30,859 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297138.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:37:44,379 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297148.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:37:50,759 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297153.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:37:51,598 INFO [train.py:904] (7/8) Epoch 30, batch 2800, loss[loss=0.1487, simple_loss=0.2445, pruned_loss=0.02643, over 17176.00 frames. ], tot_loss[loss=0.1643, simple_loss=0.255, pruned_loss=0.03681, over 3333421.99 frames. ], batch size: 46, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:38:14,254 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297170.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:38:26,496 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 21:38:49,782 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297196.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:38:57,320 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297201.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:39:00,132 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.070e+02 2.512e+02 3.097e+02 5.652e+02, threshold=5.024e+02, percent-clipped=1.0 2023-05-02 21:39:01,292 INFO [train.py:904] (7/8) Epoch 30, batch 2850, loss[loss=0.1839, simple_loss=0.2566, pruned_loss=0.05563, over 16920.00 frames. ], tot_loss[loss=0.1645, simple_loss=0.2547, pruned_loss=0.03712, over 3330480.18 frames. ], batch size: 109, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:39:15,440 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9501, 4.3913, 4.3718, 3.1671, 3.5798, 4.3512, 3.8445, 2.6812], device='cuda:7'), covar=tensor([0.0451, 0.0076, 0.0052, 0.0382, 0.0181, 0.0096, 0.0105, 0.0472], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0091, 0.0092, 0.0135, 0.0104, 0.0116, 0.0099, 0.0131], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 21:39:21,722 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297218.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:39:26,938 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=297222.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:39:37,169 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297229.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 21:40:04,184 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7838, 4.7415, 5.1331, 5.1018, 5.2029, 4.8428, 4.7869, 4.7009], device='cuda:7'), covar=tensor([0.0392, 0.0727, 0.0521, 0.0530, 0.0558, 0.0521, 0.1113, 0.0589], device='cuda:7'), in_proj_covar=tensor([0.0450, 0.0512, 0.0489, 0.0452, 0.0533, 0.0516, 0.0595, 0.0415], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 21:40:10,543 INFO [train.py:904] (7/8) Epoch 30, batch 2900, loss[loss=0.1378, simple_loss=0.2255, pruned_loss=0.02505, over 17028.00 frames. ], tot_loss[loss=0.1648, simple_loss=0.2541, pruned_loss=0.03768, over 3317623.51 frames. ], batch size: 41, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:40:33,194 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297269.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:40:51,741 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=297283.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:41:08,403 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297294.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:41:22,344 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.045e+02 2.480e+02 2.886e+02 5.342e+02, threshold=4.960e+02, percent-clipped=3.0 2023-05-02 21:41:22,359 INFO [train.py:904] (7/8) Epoch 30, batch 2950, loss[loss=0.1761, simple_loss=0.2616, pruned_loss=0.04528, over 16620.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2534, pruned_loss=0.03823, over 3307455.49 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:41:41,561 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297317.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:42:15,172 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6484, 2.7042, 2.3796, 2.5125, 2.9690, 2.7055, 3.1451, 3.1467], device='cuda:7'), covar=tensor([0.0204, 0.0508, 0.0588, 0.0506, 0.0333, 0.0461, 0.0304, 0.0344], device='cuda:7'), in_proj_covar=tensor([0.0242, 0.0251, 0.0239, 0.0241, 0.0251, 0.0249, 0.0249, 0.0252], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:42:18,970 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297344.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:42:33,298 INFO [train.py:904] (7/8) Epoch 30, batch 3000, loss[loss=0.1621, simple_loss=0.2397, pruned_loss=0.04231, over 16875.00 frames. ], tot_loss[loss=0.1649, simple_loss=0.2531, pruned_loss=0.03834, over 3309646.00 frames. ], batch size: 109, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:42:33,298 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 21:42:42,092 INFO [train.py:938] (7/8) Epoch 30, validation: loss=0.1331, simple_loss=0.238, pruned_loss=0.01415, over 944034.00 frames. 2023-05-02 21:42:42,093 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 21:43:35,256 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297392.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:43:45,167 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9351, 3.1862, 2.8958, 5.1918, 4.1218, 4.4129, 1.7996, 3.3051], device='cuda:7'), covar=tensor([0.1385, 0.0738, 0.1202, 0.0189, 0.0281, 0.0455, 0.1715, 0.0789], device='cuda:7'), in_proj_covar=tensor([0.0175, 0.0185, 0.0204, 0.0210, 0.0209, 0.0222, 0.0214, 0.0202], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:43:51,744 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.205e+02 2.625e+02 3.080e+02 5.870e+02, threshold=5.251e+02, percent-clipped=1.0 2023-05-02 21:43:51,762 INFO [train.py:904] (7/8) Epoch 30, batch 3050, loss[loss=0.1888, simple_loss=0.2704, pruned_loss=0.05359, over 12502.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2526, pruned_loss=0.03841, over 3303095.72 frames. ], batch size: 246, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:45:01,432 INFO [train.py:904] (7/8) Epoch 30, batch 3100, loss[loss=0.1611, simple_loss=0.2515, pruned_loss=0.03541, over 17119.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2522, pruned_loss=0.03845, over 3310443.22 frames. ], batch size: 48, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:45:55,847 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2570, 5.8532, 5.9658, 5.6365, 5.7771, 6.3288, 5.8235, 5.4756], device='cuda:7'), covar=tensor([0.0944, 0.2033, 0.2568, 0.2040, 0.2579, 0.1019, 0.1609, 0.2440], device='cuda:7'), in_proj_covar=tensor([0.0442, 0.0659, 0.0734, 0.0533, 0.0714, 0.0751, 0.0563, 0.0712], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:46:07,377 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.135e+02 2.482e+02 3.007e+02 9.801e+02, threshold=4.963e+02, percent-clipped=2.0 2023-05-02 21:46:07,399 INFO [train.py:904] (7/8) Epoch 30, batch 3150, loss[loss=0.1761, simple_loss=0.2726, pruned_loss=0.0398, over 17049.00 frames. ], tot_loss[loss=0.164, simple_loss=0.252, pruned_loss=0.038, over 3319407.32 frames. ], batch size: 55, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:46:42,750 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297529.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 21:47:17,210 INFO [train.py:904] (7/8) Epoch 30, batch 3200, loss[loss=0.1669, simple_loss=0.257, pruned_loss=0.03842, over 16552.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.2511, pruned_loss=0.03733, over 3326170.75 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:47:37,634 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2759, 5.6177, 5.3770, 5.3932, 5.0889, 5.0749, 4.9924, 5.7464], device='cuda:7'), covar=tensor([0.1393, 0.1001, 0.1134, 0.1029, 0.0953, 0.0894, 0.1236, 0.0967], device='cuda:7'), in_proj_covar=tensor([0.0751, 0.0903, 0.0745, 0.0704, 0.0581, 0.0571, 0.0759, 0.0710], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:47:49,870 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297577.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:47:51,724 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=297578.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:48:13,582 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297594.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:48:27,241 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.037e+02 2.359e+02 2.785e+02 6.437e+02, threshold=4.718e+02, percent-clipped=1.0 2023-05-02 21:48:27,262 INFO [train.py:904] (7/8) Epoch 30, batch 3250, loss[loss=0.168, simple_loss=0.2588, pruned_loss=0.03864, over 17065.00 frames. ], tot_loss[loss=0.1629, simple_loss=0.251, pruned_loss=0.03738, over 3326496.14 frames. ], batch size: 55, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:49:20,241 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297642.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:49:35,385 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7161, 4.1421, 2.9381, 2.3747, 2.7104, 2.6243, 4.5189, 3.4941], device='cuda:7'), covar=tensor([0.3062, 0.0592, 0.1917, 0.2996, 0.2824, 0.2174, 0.0389, 0.1404], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0280, 0.0319, 0.0334, 0.0312, 0.0285, 0.0310, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:49:36,042 INFO [train.py:904] (7/8) Epoch 30, batch 3300, loss[loss=0.1757, simple_loss=0.268, pruned_loss=0.04173, over 16688.00 frames. ], tot_loss[loss=0.1638, simple_loss=0.2523, pruned_loss=0.03772, over 3321173.56 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:49:37,545 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-05-02 21:50:32,866 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7627, 3.5943, 4.0162, 2.2636, 4.1187, 4.1403, 3.2166, 3.2081], device='cuda:7'), covar=tensor([0.0833, 0.0270, 0.0246, 0.1121, 0.0114, 0.0235, 0.0463, 0.0443], device='cuda:7'), in_proj_covar=tensor([0.0150, 0.0112, 0.0104, 0.0139, 0.0089, 0.0135, 0.0132, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:50:46,301 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.242e+02 2.585e+02 3.281e+02 4.783e+02, threshold=5.170e+02, percent-clipped=1.0 2023-05-02 21:50:46,321 INFO [train.py:904] (7/8) Epoch 30, batch 3350, loss[loss=0.185, simple_loss=0.2767, pruned_loss=0.04663, over 16656.00 frames. ], tot_loss[loss=0.166, simple_loss=0.2545, pruned_loss=0.03878, over 3314473.52 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:51:37,764 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3052, 2.1390, 1.7970, 1.8290, 2.3817, 2.1094, 2.1513, 2.4258], device='cuda:7'), covar=tensor([0.0306, 0.0411, 0.0548, 0.0493, 0.0249, 0.0371, 0.0259, 0.0316], device='cuda:7'), in_proj_covar=tensor([0.0243, 0.0253, 0.0241, 0.0243, 0.0253, 0.0251, 0.0251, 0.0254], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:51:56,071 INFO [train.py:904] (7/8) Epoch 30, batch 3400, loss[loss=0.212, simple_loss=0.297, pruned_loss=0.06347, over 12253.00 frames. ], tot_loss[loss=0.1657, simple_loss=0.254, pruned_loss=0.03873, over 3310889.70 frames. ], batch size: 246, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:52:20,910 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1745, 5.6200, 5.7413, 5.3599, 5.4830, 6.1286, 5.6087, 5.2961], device='cuda:7'), covar=tensor([0.1074, 0.2221, 0.2638, 0.2237, 0.2854, 0.0971, 0.1634, 0.2435], device='cuda:7'), in_proj_covar=tensor([0.0440, 0.0656, 0.0730, 0.0531, 0.0710, 0.0749, 0.0560, 0.0709], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 21:53:07,134 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.981e+02 2.351e+02 2.739e+02 4.469e+02, threshold=4.702e+02, percent-clipped=0.0 2023-05-02 21:53:07,155 INFO [train.py:904] (7/8) Epoch 30, batch 3450, loss[loss=0.1752, simple_loss=0.2536, pruned_loss=0.04838, over 11731.00 frames. ], tot_loss[loss=0.1639, simple_loss=0.2525, pruned_loss=0.03764, over 3307713.01 frames. ], batch size: 246, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:53:59,531 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.54 vs. limit=2.0 2023-05-02 21:54:17,205 INFO [train.py:904] (7/8) Epoch 30, batch 3500, loss[loss=0.163, simple_loss=0.2628, pruned_loss=0.03161, over 17024.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.251, pruned_loss=0.03765, over 3296509.15 frames. ], batch size: 50, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:54:51,995 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=297878.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:55:28,095 INFO [train.py:904] (7/8) Epoch 30, batch 3550, loss[loss=0.1592, simple_loss=0.2558, pruned_loss=0.03123, over 17092.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2494, pruned_loss=0.03707, over 3294756.82 frames. ], batch size: 55, lr: 2.24e-03, grad_scale: 4.0 2023-05-02 21:55:29,305 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.419e+02 1.998e+02 2.319e+02 2.785e+02 6.034e+02, threshold=4.637e+02, percent-clipped=1.0 2023-05-02 21:55:48,599 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9051, 2.1864, 2.5845, 2.9365, 2.7672, 3.4452, 2.5765, 3.4906], device='cuda:7'), covar=tensor([0.0354, 0.0629, 0.0463, 0.0418, 0.0450, 0.0251, 0.0542, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0206, 0.0204, 0.0194, 0.0199, 0.0218, 0.0175, 0.0210, 0.0174], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 21:55:59,121 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=297926.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:56:37,502 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5631, 5.9762, 5.7350, 5.7869, 5.3917, 5.4680, 5.2625, 6.1156], device='cuda:7'), covar=tensor([0.1491, 0.0937, 0.1067, 0.0931, 0.0914, 0.0677, 0.1536, 0.0934], device='cuda:7'), in_proj_covar=tensor([0.0750, 0.0904, 0.0743, 0.0703, 0.0580, 0.0570, 0.0760, 0.0711], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 21:56:38,333 INFO [train.py:904] (7/8) Epoch 30, batch 3600, loss[loss=0.1421, simple_loss=0.2356, pruned_loss=0.02426, over 17143.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2481, pruned_loss=0.03672, over 3294776.62 frames. ], batch size: 46, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:56:42,492 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=297957.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:57:18,361 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-05-02 21:57:35,815 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5866, 3.6165, 2.3412, 3.8627, 2.9493, 3.8062, 2.3541, 2.9602], device='cuda:7'), covar=tensor([0.0296, 0.0490, 0.1511, 0.0353, 0.0774, 0.0802, 0.1527, 0.0769], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0187, 0.0199, 0.0181, 0.0184, 0.0227, 0.0209, 0.0187], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 21:57:53,340 INFO [train.py:904] (7/8) Epoch 30, batch 3650, loss[loss=0.1712, simple_loss=0.2444, pruned_loss=0.04903, over 16457.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2476, pruned_loss=0.03701, over 3304807.52 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 21:57:55,115 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.298e+02 2.680e+02 3.365e+02 6.432e+02, threshold=5.360e+02, percent-clipped=4.0 2023-05-02 21:58:14,276 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298018.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 21:59:06,235 INFO [train.py:904] (7/8) Epoch 30, batch 3700, loss[loss=0.1744, simple_loss=0.2546, pruned_loss=0.04712, over 16488.00 frames. ], tot_loss[loss=0.1612, simple_loss=0.2461, pruned_loss=0.03818, over 3289122.95 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:00:17,697 INFO [train.py:904] (7/8) Epoch 30, batch 3750, loss[loss=0.1611, simple_loss=0.2343, pruned_loss=0.04392, over 16405.00 frames. ], tot_loss[loss=0.1624, simple_loss=0.2461, pruned_loss=0.03932, over 3284114.92 frames. ], batch size: 75, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:00:19,707 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.710e+02 2.048e+02 2.384e+02 2.845e+02 4.773e+02, threshold=4.768e+02, percent-clipped=0.0 2023-05-02 22:00:43,461 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7063, 2.4763, 2.0665, 2.2136, 2.8006, 2.5308, 2.6856, 2.8652], device='cuda:7'), covar=tensor([0.0299, 0.0487, 0.0596, 0.0638, 0.0294, 0.0449, 0.0238, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0242, 0.0252, 0.0240, 0.0242, 0.0252, 0.0250, 0.0251, 0.0253], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:01:30,455 INFO [train.py:904] (7/8) Epoch 30, batch 3800, loss[loss=0.1617, simple_loss=0.2525, pruned_loss=0.03551, over 16846.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2466, pruned_loss=0.04015, over 3275640.36 frames. ], batch size: 42, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:02:43,885 INFO [train.py:904] (7/8) Epoch 30, batch 3850, loss[loss=0.1676, simple_loss=0.2449, pruned_loss=0.0451, over 16416.00 frames. ], tot_loss[loss=0.1644, simple_loss=0.2471, pruned_loss=0.04088, over 3278519.57 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:02:44,949 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.556e+02 2.341e+02 2.538e+02 2.759e+02 9.437e+02, threshold=5.075e+02, percent-clipped=1.0 2023-05-02 22:03:53,276 INFO [train.py:904] (7/8) Epoch 30, batch 3900, loss[loss=0.1935, simple_loss=0.2655, pruned_loss=0.06076, over 16892.00 frames. ], tot_loss[loss=0.1647, simple_loss=0.2466, pruned_loss=0.0414, over 3280436.91 frames. ], batch size: 109, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:04:49,564 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8598, 4.7461, 4.7570, 4.4516, 4.5007, 4.8247, 4.6253, 4.6051], device='cuda:7'), covar=tensor([0.0708, 0.0953, 0.0380, 0.0354, 0.0871, 0.0714, 0.0465, 0.0645], device='cuda:7'), in_proj_covar=tensor([0.0330, 0.0500, 0.0387, 0.0388, 0.0383, 0.0447, 0.0264, 0.0461], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:05:04,298 INFO [train.py:904] (7/8) Epoch 30, batch 3950, loss[loss=0.175, simple_loss=0.2476, pruned_loss=0.05119, over 16942.00 frames. ], tot_loss[loss=0.1653, simple_loss=0.2467, pruned_loss=0.04201, over 3290363.63 frames. ], batch size: 109, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:05:05,536 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.189e+02 2.590e+02 3.423e+02 7.666e+02, threshold=5.180e+02, percent-clipped=4.0 2023-05-02 22:05:17,570 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298313.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:05:17,777 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7247, 3.8318, 2.9139, 2.2933, 2.4580, 2.4913, 3.9045, 3.3164], device='cuda:7'), covar=tensor([0.2881, 0.0609, 0.1896, 0.3385, 0.2968, 0.2183, 0.0581, 0.1532], device='cuda:7'), in_proj_covar=tensor([0.0341, 0.0280, 0.0319, 0.0334, 0.0313, 0.0284, 0.0309, 0.0361], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:05:55,135 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1741, 3.1379, 3.5269, 2.3030, 3.0273, 2.4308, 3.6333, 3.5833], device='cuda:7'), covar=tensor([0.0240, 0.0972, 0.0656, 0.1979, 0.0878, 0.1058, 0.0515, 0.0877], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0174, 0.0172, 0.0159, 0.0150, 0.0134, 0.0147, 0.0188], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 22:06:15,422 INFO [train.py:904] (7/8) Epoch 30, batch 4000, loss[loss=0.1639, simple_loss=0.2438, pruned_loss=0.04201, over 17046.00 frames. ], tot_loss[loss=0.1659, simple_loss=0.2467, pruned_loss=0.0425, over 3297455.71 frames. ], batch size: 53, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:07:04,857 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 22:07:18,898 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 22:07:25,665 INFO [train.py:904] (7/8) Epoch 30, batch 4050, loss[loss=0.1709, simple_loss=0.2643, pruned_loss=0.03873, over 15511.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2481, pruned_loss=0.04209, over 3297005.90 frames. ], batch size: 190, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:07:27,599 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 1.991e+02 2.231e+02 2.617e+02 4.473e+02, threshold=4.462e+02, percent-clipped=0.0 2023-05-02 22:08:18,020 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0348, 2.5799, 2.4338, 4.3824, 2.9746, 3.9804, 1.6769, 2.8813], device='cuda:7'), covar=tensor([0.1244, 0.0930, 0.1420, 0.0170, 0.0308, 0.0382, 0.1721, 0.0944], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0183, 0.0202, 0.0209, 0.0208, 0.0220, 0.0212, 0.0201], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 22:08:37,711 INFO [train.py:904] (7/8) Epoch 30, batch 4100, loss[loss=0.1842, simple_loss=0.2662, pruned_loss=0.05104, over 12127.00 frames. ], tot_loss[loss=0.1667, simple_loss=0.25, pruned_loss=0.04176, over 3288967.68 frames. ], batch size: 246, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:09:20,237 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298482.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:09:53,970 INFO [train.py:904] (7/8) Epoch 30, batch 4150, loss[loss=0.2049, simple_loss=0.2947, pruned_loss=0.05756, over 16894.00 frames. ], tot_loss[loss=0.1719, simple_loss=0.2564, pruned_loss=0.04371, over 3244769.29 frames. ], batch size: 116, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:09:56,059 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.028e+02 2.326e+02 2.808e+02 4.631e+02, threshold=4.653e+02, percent-clipped=1.0 2023-05-02 22:10:01,834 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1214, 3.7141, 3.6493, 2.3372, 3.2964, 3.6792, 3.3607, 2.0678], device='cuda:7'), covar=tensor([0.0642, 0.0057, 0.0073, 0.0504, 0.0139, 0.0137, 0.0130, 0.0530], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0092, 0.0095, 0.0138, 0.0106, 0.0119, 0.0102, 0.0134], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 22:10:35,512 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298530.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:10:56,388 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298543.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:11:12,261 INFO [train.py:904] (7/8) Epoch 30, batch 4200, loss[loss=0.1934, simple_loss=0.2863, pruned_loss=0.05024, over 16474.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2633, pruned_loss=0.04525, over 3220194.73 frames. ], batch size: 75, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:11:22,731 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5823, 3.5989, 3.9986, 2.3043, 3.2339, 2.6122, 3.9230, 3.9727], device='cuda:7'), covar=tensor([0.0229, 0.0897, 0.0512, 0.2107, 0.0868, 0.0928, 0.0579, 0.0990], device='cuda:7'), in_proj_covar=tensor([0.0164, 0.0175, 0.0173, 0.0160, 0.0150, 0.0134, 0.0148, 0.0188], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 22:12:08,283 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298591.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:12:26,990 INFO [train.py:904] (7/8) Epoch 30, batch 4250, loss[loss=0.164, simple_loss=0.2622, pruned_loss=0.03289, over 16760.00 frames. ], tot_loss[loss=0.1791, simple_loss=0.2671, pruned_loss=0.0456, over 3203870.15 frames. ], batch size: 124, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:12:28,281 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.166e+02 2.498e+02 2.846e+02 5.761e+02, threshold=4.997e+02, percent-clipped=3.0 2023-05-02 22:12:40,753 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=298613.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:12:41,392 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-05-02 22:12:59,072 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-05-02 22:13:04,658 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298629.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 22:13:25,997 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9918, 3.2558, 3.1753, 2.0981, 2.9999, 3.2487, 3.0391, 1.9367], device='cuda:7'), covar=tensor([0.0621, 0.0073, 0.0096, 0.0524, 0.0135, 0.0141, 0.0130, 0.0523], device='cuda:7'), in_proj_covar=tensor([0.0142, 0.0092, 0.0094, 0.0138, 0.0106, 0.0118, 0.0101, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 22:13:41,914 INFO [train.py:904] (7/8) Epoch 30, batch 4300, loss[loss=0.2115, simple_loss=0.295, pruned_loss=0.06399, over 11783.00 frames. ], tot_loss[loss=0.1792, simple_loss=0.2684, pruned_loss=0.04501, over 3193771.88 frames. ], batch size: 247, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:13:53,099 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=298661.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:14:35,110 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=298689.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:14:36,450 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298690.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:14:56,163 INFO [train.py:904] (7/8) Epoch 30, batch 4350, loss[loss=0.1977, simple_loss=0.2872, pruned_loss=0.05408, over 16526.00 frames. ], tot_loss[loss=0.1817, simple_loss=0.2718, pruned_loss=0.0458, over 3194590.66 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:14:57,382 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.491e+02 2.186e+02 2.538e+02 2.866e+02 4.288e+02, threshold=5.076e+02, percent-clipped=0.0 2023-05-02 22:16:04,762 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=298750.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:16:09,621 INFO [train.py:904] (7/8) Epoch 30, batch 4400, loss[loss=0.1824, simple_loss=0.2804, pruned_loss=0.04217, over 16660.00 frames. ], tot_loss[loss=0.1838, simple_loss=0.2738, pruned_loss=0.04684, over 3184367.97 frames. ], batch size: 76, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:16:37,377 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6924, 1.8808, 2.4478, 2.6617, 2.6658, 3.0447, 2.0426, 2.9913], device='cuda:7'), covar=tensor([0.0267, 0.0593, 0.0349, 0.0345, 0.0340, 0.0223, 0.0608, 0.0177], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0202, 0.0191, 0.0197, 0.0215, 0.0171, 0.0207, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 22:16:49,279 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6557, 4.7809, 4.9917, 4.6462, 4.8008, 5.3555, 4.8129, 4.4470], device='cuda:7'), covar=tensor([0.1221, 0.1674, 0.1979, 0.2060, 0.2304, 0.0854, 0.1550, 0.2564], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0643, 0.0715, 0.0523, 0.0695, 0.0735, 0.0552, 0.0695], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:17:21,883 INFO [train.py:904] (7/8) Epoch 30, batch 4450, loss[loss=0.1914, simple_loss=0.2823, pruned_loss=0.05025, over 16386.00 frames. ], tot_loss[loss=0.1876, simple_loss=0.2782, pruned_loss=0.04845, over 3192007.79 frames. ], batch size: 68, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:17:23,560 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.049e+02 2.382e+02 3.028e+02 5.771e+02, threshold=4.764e+02, percent-clipped=2.0 2023-05-02 22:17:37,636 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9546, 2.3661, 2.2599, 3.6546, 2.1955, 2.6334, 2.3864, 2.4275], device='cuda:7'), covar=tensor([0.1564, 0.3262, 0.3115, 0.0636, 0.4127, 0.2426, 0.3322, 0.3356], device='cuda:7'), in_proj_covar=tensor([0.0429, 0.0482, 0.0391, 0.0344, 0.0449, 0.0555, 0.0454, 0.0566], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:17:59,416 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5956, 4.4745, 4.6352, 4.7958, 4.9147, 4.4899, 4.9272, 4.9451], device='cuda:7'), covar=tensor([0.1590, 0.1088, 0.1387, 0.0653, 0.0457, 0.0963, 0.0557, 0.0518], device='cuda:7'), in_proj_covar=tensor([0.0700, 0.0850, 0.0981, 0.0869, 0.0659, 0.0686, 0.0720, 0.0838], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:18:10,732 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298838.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 22:18:12,005 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5418, 3.5291, 2.2239, 4.2482, 2.8621, 4.1597, 2.5237, 2.9957], device='cuda:7'), covar=tensor([0.0337, 0.0438, 0.1839, 0.0142, 0.0868, 0.0458, 0.1552, 0.0846], device='cuda:7'), in_proj_covar=tensor([0.0179, 0.0184, 0.0197, 0.0176, 0.0182, 0.0223, 0.0205, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 22:18:23,458 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5524, 4.6562, 4.4412, 4.1162, 4.1753, 4.5502, 4.2283, 4.2583], device='cuda:7'), covar=tensor([0.0522, 0.0321, 0.0248, 0.0245, 0.0644, 0.0317, 0.0602, 0.0509], device='cuda:7'), in_proj_covar=tensor([0.0322, 0.0486, 0.0379, 0.0380, 0.0374, 0.0437, 0.0259, 0.0450], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:18:32,517 INFO [train.py:904] (7/8) Epoch 30, batch 4500, loss[loss=0.196, simple_loss=0.2935, pruned_loss=0.04925, over 16737.00 frames. ], tot_loss[loss=0.1884, simple_loss=0.2786, pruned_loss=0.04911, over 3189454.78 frames. ], batch size: 124, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:19:19,861 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298886.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:19:44,881 INFO [train.py:904] (7/8) Epoch 30, batch 4550, loss[loss=0.1887, simple_loss=0.2747, pruned_loss=0.05135, over 17032.00 frames. ], tot_loss[loss=0.1898, simple_loss=0.2793, pruned_loss=0.05014, over 3201465.92 frames. ], batch size: 55, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:19:46,103 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 1.782e+02 2.097e+02 2.367e+02 4.731e+02, threshold=4.194e+02, percent-clipped=0.0 2023-05-02 22:20:57,288 INFO [train.py:904] (7/8) Epoch 30, batch 4600, loss[loss=0.1927, simple_loss=0.266, pruned_loss=0.05977, over 11712.00 frames. ], tot_loss[loss=0.1903, simple_loss=0.2802, pruned_loss=0.05016, over 3221668.11 frames. ], batch size: 246, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:21:14,630 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.9430, 1.9792, 2.6917, 2.8738, 2.8326, 3.3155, 2.2733, 3.3144], device='cuda:7'), covar=tensor([0.0239, 0.0585, 0.0322, 0.0325, 0.0347, 0.0191, 0.0601, 0.0165], device='cuda:7'), in_proj_covar=tensor([0.0203, 0.0202, 0.0191, 0.0198, 0.0216, 0.0172, 0.0208, 0.0172], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 22:21:29,656 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5628, 4.6235, 4.8383, 4.8011, 4.8498, 4.5598, 4.5495, 4.3519], device='cuda:7'), covar=tensor([0.0320, 0.0460, 0.0357, 0.0370, 0.0464, 0.0390, 0.0909, 0.0557], device='cuda:7'), in_proj_covar=tensor([0.0442, 0.0505, 0.0482, 0.0446, 0.0528, 0.0510, 0.0587, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 22:21:43,598 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=298985.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:22:04,323 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.2768, 5.3221, 5.5716, 5.5318, 5.6563, 5.2910, 5.1886, 4.8690], device='cuda:7'), covar=tensor([0.0302, 0.0410, 0.0313, 0.0391, 0.0520, 0.0302, 0.1070, 0.0558], device='cuda:7'), in_proj_covar=tensor([0.0442, 0.0504, 0.0481, 0.0445, 0.0528, 0.0510, 0.0587, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 22:22:09,592 INFO [train.py:904] (7/8) Epoch 30, batch 4650, loss[loss=0.1674, simple_loss=0.2517, pruned_loss=0.04154, over 16759.00 frames. ], tot_loss[loss=0.19, simple_loss=0.2792, pruned_loss=0.05036, over 3214346.00 frames. ], batch size: 83, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:22:10,879 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.373e+02 1.831e+02 2.100e+02 2.427e+02 4.549e+02, threshold=4.199e+02, percent-clipped=1.0 2023-05-02 22:22:28,144 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.6151, 5.9066, 5.6607, 5.7759, 5.4119, 5.3195, 5.3272, 6.0682], device='cuda:7'), covar=tensor([0.1179, 0.0817, 0.0964, 0.0794, 0.0788, 0.0672, 0.1098, 0.0699], device='cuda:7'), in_proj_covar=tensor([0.0730, 0.0880, 0.0722, 0.0685, 0.0565, 0.0559, 0.0738, 0.0692], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:22:30,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8973, 4.0087, 4.1621, 4.1237, 4.1507, 3.9606, 3.9613, 3.9010], device='cuda:7'), covar=tensor([0.0340, 0.0492, 0.0356, 0.0381, 0.0430, 0.0411, 0.0711, 0.0547], device='cuda:7'), in_proj_covar=tensor([0.0441, 0.0503, 0.0480, 0.0444, 0.0527, 0.0509, 0.0586, 0.0412], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 22:22:51,786 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9625, 5.1822, 5.0153, 5.0269, 4.7483, 4.6858, 4.6203, 5.3070], device='cuda:7'), covar=tensor([0.1158, 0.0836, 0.0858, 0.0804, 0.0753, 0.0940, 0.1112, 0.0725], device='cuda:7'), in_proj_covar=tensor([0.0730, 0.0880, 0.0722, 0.0685, 0.0565, 0.0559, 0.0738, 0.0692], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:23:03,552 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3562, 4.4560, 4.2508, 3.9176, 3.9347, 4.3570, 4.0340, 4.0848], device='cuda:7'), covar=tensor([0.0553, 0.0475, 0.0279, 0.0270, 0.0705, 0.0400, 0.0791, 0.0475], device='cuda:7'), in_proj_covar=tensor([0.0320, 0.0484, 0.0377, 0.0379, 0.0372, 0.0434, 0.0258, 0.0447], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:23:09,778 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299045.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:23:12,066 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1581, 4.0198, 4.2131, 4.3512, 4.4513, 4.1144, 4.4081, 4.5192], device='cuda:7'), covar=tensor([0.1784, 0.1217, 0.1480, 0.0732, 0.0545, 0.1218, 0.0853, 0.0611], device='cuda:7'), in_proj_covar=tensor([0.0694, 0.0841, 0.0975, 0.0860, 0.0652, 0.0679, 0.0712, 0.0830], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:23:22,504 INFO [train.py:904] (7/8) Epoch 30, batch 4700, loss[loss=0.1693, simple_loss=0.2601, pruned_loss=0.03924, over 16479.00 frames. ], tot_loss[loss=0.1879, simple_loss=0.2767, pruned_loss=0.04948, over 3220604.58 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:24:36,521 INFO [train.py:904] (7/8) Epoch 30, batch 4750, loss[loss=0.1684, simple_loss=0.2585, pruned_loss=0.03918, over 16280.00 frames. ], tot_loss[loss=0.1839, simple_loss=0.2727, pruned_loss=0.04752, over 3209419.64 frames. ], batch size: 165, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:24:37,724 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 1.904e+02 2.113e+02 2.472e+02 4.285e+02, threshold=4.226e+02, percent-clipped=1.0 2023-05-02 22:24:58,558 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-05-02 22:25:08,366 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299126.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:25:25,961 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299138.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:25:49,336 INFO [train.py:904] (7/8) Epoch 30, batch 4800, loss[loss=0.1744, simple_loss=0.2728, pruned_loss=0.03804, over 16273.00 frames. ], tot_loss[loss=0.1795, simple_loss=0.2688, pruned_loss=0.04506, over 3216483.78 frames. ], batch size: 165, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:26:37,339 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299186.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:26:37,393 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299186.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:26:39,181 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299187.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:26:46,061 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299191.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:27:04,371 INFO [train.py:904] (7/8) Epoch 30, batch 4850, loss[loss=0.1874, simple_loss=0.2781, pruned_loss=0.04836, over 16560.00 frames. ], tot_loss[loss=0.1801, simple_loss=0.2701, pruned_loss=0.04509, over 3194487.48 frames. ], batch size: 57, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:27:06,464 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.292e+02 1.894e+02 2.171e+02 2.574e+02 8.129e+02, threshold=4.343e+02, percent-clipped=1.0 2023-05-02 22:27:22,513 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.3707, 4.4460, 4.6827, 4.6512, 4.6782, 4.4219, 4.4020, 4.3081], device='cuda:7'), covar=tensor([0.0298, 0.0483, 0.0360, 0.0366, 0.0392, 0.0346, 0.0856, 0.0503], device='cuda:7'), in_proj_covar=tensor([0.0441, 0.0503, 0.0481, 0.0446, 0.0526, 0.0509, 0.0587, 0.0412], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 22:27:49,443 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299234.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:28:02,958 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299243.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:28:03,270 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-02 22:28:08,028 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.4641, 3.5703, 3.6852, 3.6605, 3.6767, 3.5341, 3.5573, 3.5644], device='cuda:7'), covar=tensor([0.0364, 0.0689, 0.0449, 0.0423, 0.0475, 0.0488, 0.0703, 0.0482], device='cuda:7'), in_proj_covar=tensor([0.0441, 0.0503, 0.0481, 0.0445, 0.0526, 0.0509, 0.0586, 0.0411], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 22:28:16,518 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299252.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:28:18,994 INFO [train.py:904] (7/8) Epoch 30, batch 4900, loss[loss=0.1571, simple_loss=0.25, pruned_loss=0.03211, over 17221.00 frames. ], tot_loss[loss=0.1786, simple_loss=0.2696, pruned_loss=0.04384, over 3177549.58 frames. ], batch size: 45, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:28:43,839 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7403, 4.8639, 5.0354, 4.7289, 4.8495, 5.4225, 4.8892, 4.5795], device='cuda:7'), covar=tensor([0.1104, 0.1884, 0.2028, 0.2008, 0.2611, 0.1120, 0.1588, 0.2566], device='cuda:7'), in_proj_covar=tensor([0.0433, 0.0637, 0.0708, 0.0520, 0.0692, 0.0730, 0.0548, 0.0691], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:29:06,103 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299285.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 22:29:33,780 INFO [train.py:904] (7/8) Epoch 30, batch 4950, loss[loss=0.1715, simple_loss=0.2635, pruned_loss=0.03972, over 17144.00 frames. ], tot_loss[loss=0.1769, simple_loss=0.2683, pruned_loss=0.04277, over 3192561.46 frames. ], batch size: 48, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:29:34,268 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299304.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:29:34,856 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 1.940e+02 2.221e+02 2.664e+02 5.987e+02, threshold=4.443e+02, percent-clipped=2.0 2023-05-02 22:29:38,514 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-05-02 22:30:02,811 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299323.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:30:17,899 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299333.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 22:30:36,125 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299345.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:30:43,377 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.90 vs. limit=5.0 2023-05-02 22:30:47,903 INFO [train.py:904] (7/8) Epoch 30, batch 5000, loss[loss=0.2001, simple_loss=0.2951, pruned_loss=0.05261, over 15416.00 frames. ], tot_loss[loss=0.178, simple_loss=0.2703, pruned_loss=0.04287, over 3201275.28 frames. ], batch size: 190, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:31:33,691 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299384.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:31:46,454 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299393.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:31:47,909 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5460, 3.7974, 2.7307, 2.1921, 2.6695, 2.4625, 4.0762, 3.2654], device='cuda:7'), covar=tensor([0.3421, 0.0767, 0.2277, 0.3017, 0.2809, 0.2339, 0.0594, 0.1545], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0277, 0.0316, 0.0331, 0.0310, 0.0282, 0.0307, 0.0356], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:31:49,100 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.1885, 2.4186, 2.3918, 3.9029, 2.3143, 2.7350, 2.4336, 2.5231], device='cuda:7'), covar=tensor([0.1642, 0.3625, 0.3315, 0.0655, 0.4194, 0.2691, 0.3885, 0.3345], device='cuda:7'), in_proj_covar=tensor([0.0429, 0.0483, 0.0392, 0.0344, 0.0449, 0.0555, 0.0456, 0.0567], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:32:01,708 INFO [train.py:904] (7/8) Epoch 30, batch 5050, loss[loss=0.174, simple_loss=0.262, pruned_loss=0.04298, over 16412.00 frames. ], tot_loss[loss=0.1787, simple_loss=0.2712, pruned_loss=0.0431, over 3196897.36 frames. ], batch size: 35, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:32:02,878 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.023e+02 2.400e+02 2.818e+02 3.992e+02, threshold=4.801e+02, percent-clipped=0.0 2023-05-02 22:32:06,450 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0617, 2.4093, 2.0025, 2.2277, 2.7520, 2.4375, 2.5417, 2.9319], device='cuda:7'), covar=tensor([0.0183, 0.0541, 0.0659, 0.0521, 0.0285, 0.0450, 0.0221, 0.0315], device='cuda:7'), in_proj_covar=tensor([0.0236, 0.0246, 0.0236, 0.0237, 0.0247, 0.0245, 0.0245, 0.0248], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:33:13,776 INFO [train.py:904] (7/8) Epoch 30, batch 5100, loss[loss=0.1776, simple_loss=0.2655, pruned_loss=0.04481, over 16868.00 frames. ], tot_loss[loss=0.1774, simple_loss=0.2694, pruned_loss=0.04267, over 3219448.05 frames. ], batch size: 109, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:33:42,840 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7859, 4.8546, 4.6791, 4.2794, 4.2726, 4.7443, 4.6450, 4.4780], device='cuda:7'), covar=tensor([0.0685, 0.0487, 0.0366, 0.0379, 0.1225, 0.0580, 0.0382, 0.0694], device='cuda:7'), in_proj_covar=tensor([0.0318, 0.0482, 0.0374, 0.0376, 0.0371, 0.0433, 0.0255, 0.0444], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:33:57,983 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299482.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:34:14,242 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2754, 2.3714, 1.7625, 1.9974, 2.6423, 2.3693, 2.8041, 3.0048], device='cuda:7'), covar=tensor([0.0170, 0.0683, 0.0919, 0.0753, 0.0401, 0.0619, 0.0276, 0.0387], device='cuda:7'), in_proj_covar=tensor([0.0235, 0.0246, 0.0236, 0.0236, 0.0247, 0.0244, 0.0244, 0.0247], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:34:30,233 INFO [train.py:904] (7/8) Epoch 30, batch 5150, loss[loss=0.1745, simple_loss=0.2763, pruned_loss=0.03631, over 15331.00 frames. ], tot_loss[loss=0.1764, simple_loss=0.2691, pruned_loss=0.04182, over 3201762.28 frames. ], batch size: 191, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:34:30,749 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.6507, 3.7214, 1.9346, 4.1796, 2.7020, 4.0763, 2.1242, 2.9139], device='cuda:7'), covar=tensor([0.0297, 0.0388, 0.2103, 0.0273, 0.0953, 0.0529, 0.1948, 0.0939], device='cuda:7'), in_proj_covar=tensor([0.0177, 0.0182, 0.0195, 0.0173, 0.0180, 0.0220, 0.0203, 0.0183], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 22:34:31,341 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 1.896e+02 2.234e+02 2.667e+02 3.487e+02, threshold=4.469e+02, percent-clipped=0.0 2023-05-02 22:35:33,158 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299547.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:35:43,074 INFO [train.py:904] (7/8) Epoch 30, batch 5200, loss[loss=0.1749, simple_loss=0.2689, pruned_loss=0.04048, over 16412.00 frames. ], tot_loss[loss=0.1747, simple_loss=0.2672, pruned_loss=0.04107, over 3210866.10 frames. ], batch size: 146, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:36:42,242 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-05-02 22:36:48,438 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299599.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:36:55,729 INFO [train.py:904] (7/8) Epoch 30, batch 5250, loss[loss=0.166, simple_loss=0.263, pruned_loss=0.03446, over 16890.00 frames. ], tot_loss[loss=0.1727, simple_loss=0.2645, pruned_loss=0.04044, over 3217514.66 frames. ], batch size: 116, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:36:56,956 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.002e+02 2.220e+02 2.747e+02 4.551e+02, threshold=4.439e+02, percent-clipped=1.0 2023-05-02 22:38:06,903 INFO [train.py:904] (7/8) Epoch 30, batch 5300, loss[loss=0.1667, simple_loss=0.2573, pruned_loss=0.03808, over 15394.00 frames. ], tot_loss[loss=0.1698, simple_loss=0.2611, pruned_loss=0.03923, over 3218362.36 frames. ], batch size: 190, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:38:42,833 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=299679.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:39:18,177 INFO [train.py:904] (7/8) Epoch 30, batch 5350, loss[loss=0.1799, simple_loss=0.2693, pruned_loss=0.0452, over 16662.00 frames. ], tot_loss[loss=0.1688, simple_loss=0.2597, pruned_loss=0.03893, over 3220440.37 frames. ], batch size: 62, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:39:19,353 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.376e+02 1.953e+02 2.222e+02 2.592e+02 5.137e+02, threshold=4.444e+02, percent-clipped=1.0 2023-05-02 22:39:30,682 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.84 vs. limit=5.0 2023-05-02 22:40:29,714 INFO [train.py:904] (7/8) Epoch 30, batch 5400, loss[loss=0.1756, simple_loss=0.2699, pruned_loss=0.04061, over 16531.00 frames. ], tot_loss[loss=0.1705, simple_loss=0.2619, pruned_loss=0.03955, over 3229028.43 frames. ], batch size: 75, lr: 2.24e-03, grad_scale: 8.0 2023-05-02 22:40:46,305 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-05-02 22:40:51,621 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299769.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:40:54,894 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.66 vs. limit=2.0 2023-05-02 22:41:10,090 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299782.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:41:35,391 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=299798.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 22:41:46,664 INFO [train.py:904] (7/8) Epoch 30, batch 5450, loss[loss=0.1936, simple_loss=0.276, pruned_loss=0.05558, over 12239.00 frames. ], tot_loss[loss=0.1736, simple_loss=0.265, pruned_loss=0.04111, over 3207602.49 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:41:47,800 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 2.110e+02 2.402e+02 2.864e+02 4.092e+02, threshold=4.805e+02, percent-clipped=0.0 2023-05-02 22:42:27,055 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299830.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:42:27,310 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299830.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:42:51,990 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299847.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:43:02,909 INFO [train.py:904] (7/8) Epoch 30, batch 5500, loss[loss=0.2058, simple_loss=0.2942, pruned_loss=0.05869, over 16812.00 frames. ], tot_loss[loss=0.1803, simple_loss=0.2714, pruned_loss=0.04464, over 3181285.71 frames. ], batch size: 39, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:43:11,724 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=299859.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 22:44:09,163 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299895.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:44:14,820 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299899.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:44:21,847 INFO [train.py:904] (7/8) Epoch 30, batch 5550, loss[loss=0.2128, simple_loss=0.3015, pruned_loss=0.06199, over 16591.00 frames. ], tot_loss[loss=0.189, simple_loss=0.2785, pruned_loss=0.04969, over 3130477.43 frames. ], batch size: 75, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:44:23,798 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 2.904e+02 3.261e+02 4.114e+02 7.155e+02, threshold=6.523e+02, percent-clipped=13.0 2023-05-02 22:45:09,859 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8464, 4.8949, 4.7333, 4.3684, 4.3755, 4.8176, 4.6887, 4.5372], device='cuda:7'), covar=tensor([0.0689, 0.0615, 0.0350, 0.0391, 0.1119, 0.0506, 0.0451, 0.0727], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0485, 0.0377, 0.0377, 0.0372, 0.0436, 0.0256, 0.0446], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:45:14,759 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-02 22:45:32,581 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=299947.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:45:43,013 INFO [train.py:904] (7/8) Epoch 30, batch 5600, loss[loss=0.2224, simple_loss=0.3118, pruned_loss=0.06655, over 15281.00 frames. ], tot_loss[loss=0.1954, simple_loss=0.2831, pruned_loss=0.05385, over 3077110.10 frames. ], batch size: 191, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:46:27,215 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=299979.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:47:11,378 INFO [train.py:904] (7/8) Epoch 30, batch 5650, loss[loss=0.216, simple_loss=0.2969, pruned_loss=0.0675, over 16588.00 frames. ], tot_loss[loss=0.2027, simple_loss=0.2885, pruned_loss=0.05843, over 3039935.40 frames. ], batch size: 134, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:47:13,276 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.022e+02 3.189e+02 3.868e+02 4.538e+02 6.994e+02, threshold=7.735e+02, percent-clipped=1.0 2023-05-02 22:47:16,643 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.65 vs. limit=2.0 2023-05-02 22:47:37,900 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-05-02 22:47:39,126 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6564, 4.8484, 4.9974, 4.7749, 4.8538, 5.3821, 4.8399, 4.6050], device='cuda:7'), covar=tensor([0.1243, 0.1918, 0.2609, 0.2010, 0.2445, 0.1099, 0.1900, 0.2632], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0636, 0.0708, 0.0516, 0.0687, 0.0726, 0.0546, 0.0687], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 22:47:48,226 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300027.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:48:27,603 INFO [train.py:904] (7/8) Epoch 30, batch 5700, loss[loss=0.2751, simple_loss=0.3376, pruned_loss=0.1063, over 11341.00 frames. ], tot_loss[loss=0.2047, simple_loss=0.2899, pruned_loss=0.05976, over 3049637.87 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 16.0 2023-05-02 22:48:46,517 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5295, 4.5964, 4.4161, 4.0708, 4.1088, 4.5142, 4.2795, 4.2557], device='cuda:7'), covar=tensor([0.0669, 0.0535, 0.0318, 0.0353, 0.0879, 0.0471, 0.0596, 0.0644], device='cuda:7'), in_proj_covar=tensor([0.0319, 0.0484, 0.0375, 0.0376, 0.0370, 0.0434, 0.0256, 0.0445], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 22:48:57,372 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=300073.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:49:45,360 INFO [train.py:904] (7/8) Epoch 30, batch 5750, loss[loss=0.2113, simple_loss=0.2827, pruned_loss=0.06993, over 10979.00 frames. ], tot_loss[loss=0.2068, simple_loss=0.2922, pruned_loss=0.06067, over 3036390.69 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:49:49,227 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.972e+02 3.119e+02 3.894e+02 5.149e+02 1.112e+03, threshold=7.788e+02, percent-clipped=3.0 2023-05-02 22:50:19,342 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300125.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:50:35,636 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=300134.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:51:06,746 INFO [train.py:904] (7/8) Epoch 30, batch 5800, loss[loss=0.2456, simple_loss=0.3079, pruned_loss=0.09161, over 11959.00 frames. ], tot_loss[loss=0.2048, simple_loss=0.2912, pruned_loss=0.05923, over 3046467.71 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:51:07,775 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300154.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 22:52:25,621 INFO [train.py:904] (7/8) Epoch 30, batch 5850, loss[loss=0.2069, simple_loss=0.2926, pruned_loss=0.06056, over 17079.00 frames. ], tot_loss[loss=0.2022, simple_loss=0.2892, pruned_loss=0.05761, over 3053238.98 frames. ], batch size: 53, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:52:28,952 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.843e+02 2.638e+02 2.992e+02 3.617e+02 7.830e+02, threshold=5.985e+02, percent-clipped=1.0 2023-05-02 22:52:35,170 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8934, 2.7891, 2.4852, 4.7941, 3.3617, 4.0732, 1.8368, 2.9665], device='cuda:7'), covar=tensor([0.1291, 0.0830, 0.1380, 0.0180, 0.0353, 0.0520, 0.1565, 0.0938], device='cuda:7'), in_proj_covar=tensor([0.0173, 0.0182, 0.0201, 0.0207, 0.0207, 0.0218, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 22:53:48,230 INFO [train.py:904] (7/8) Epoch 30, batch 5900, loss[loss=0.2184, simple_loss=0.2845, pruned_loss=0.0761, over 11312.00 frames. ], tot_loss[loss=0.2011, simple_loss=0.2883, pruned_loss=0.05688, over 3088733.77 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:54:17,856 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-05-02 22:55:09,239 INFO [train.py:904] (7/8) Epoch 30, batch 5950, loss[loss=0.1866, simple_loss=0.2908, pruned_loss=0.04117, over 16813.00 frames. ], tot_loss[loss=0.2006, simple_loss=0.2896, pruned_loss=0.05576, over 3106796.69 frames. ], batch size: 102, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:55:12,900 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.875e+02 2.782e+02 3.249e+02 4.004e+02 6.648e+02, threshold=6.499e+02, percent-clipped=2.0 2023-05-02 22:56:29,066 INFO [train.py:904] (7/8) Epoch 30, batch 6000, loss[loss=0.1845, simple_loss=0.2747, pruned_loss=0.04717, over 16910.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2889, pruned_loss=0.05561, over 3104095.16 frames. ], batch size: 109, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:56:29,066 INFO [train.py:929] (7/8) Computing validation loss 2023-05-02 22:56:39,792 INFO [train.py:938] (7/8) Epoch 30, validation: loss=0.1471, simple_loss=0.2591, pruned_loss=0.01755, over 944034.00 frames. 2023-05-02 22:56:39,792 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-02 22:57:57,811 INFO [train.py:904] (7/8) Epoch 30, batch 6050, loss[loss=0.2232, simple_loss=0.2927, pruned_loss=0.07688, over 11171.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2877, pruned_loss=0.05544, over 3087890.32 frames. ], batch size: 249, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:58:01,296 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 2.541e+02 2.904e+02 3.584e+02 6.768e+02, threshold=5.808e+02, percent-clipped=1.0 2023-05-02 22:58:29,463 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=300425.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:58:35,973 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300429.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 22:59:15,671 INFO [train.py:904] (7/8) Epoch 30, batch 6100, loss[loss=0.2318, simple_loss=0.3008, pruned_loss=0.08142, over 11280.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2871, pruned_loss=0.05449, over 3091462.24 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 22:59:17,390 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=300454.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 22:59:19,719 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-05-02 22:59:44,081 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7453, 4.1770, 4.1663, 2.8401, 3.7716, 4.2433, 3.7681, 2.4914], device='cuda:7'), covar=tensor([0.0533, 0.0054, 0.0060, 0.0413, 0.0110, 0.0117, 0.0101, 0.0430], device='cuda:7'), in_proj_covar=tensor([0.0141, 0.0093, 0.0094, 0.0137, 0.0105, 0.0118, 0.0101, 0.0133], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 22:59:47,470 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300473.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:00:32,834 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300502.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 23:00:34,969 INFO [train.py:904] (7/8) Epoch 30, batch 6150, loss[loss=0.1742, simple_loss=0.2594, pruned_loss=0.04453, over 17221.00 frames. ], tot_loss[loss=0.1961, simple_loss=0.2848, pruned_loss=0.05374, over 3100060.25 frames. ], batch size: 52, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:00:38,229 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.879e+02 2.817e+02 3.397e+02 3.962e+02 7.100e+02, threshold=6.795e+02, percent-clipped=3.0 2023-05-02 23:01:53,689 INFO [train.py:904] (7/8) Epoch 30, batch 6200, loss[loss=0.1912, simple_loss=0.2794, pruned_loss=0.05153, over 16873.00 frames. ], tot_loss[loss=0.1945, simple_loss=0.2827, pruned_loss=0.05312, over 3110964.00 frames. ], batch size: 116, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:03:13,166 INFO [train.py:904] (7/8) Epoch 30, batch 6250, loss[loss=0.1901, simple_loss=0.2883, pruned_loss=0.04592, over 16590.00 frames. ], tot_loss[loss=0.1944, simple_loss=0.2826, pruned_loss=0.05315, over 3107178.54 frames. ], batch size: 76, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:03:16,360 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 2.845e+02 3.157e+02 4.016e+02 6.897e+02, threshold=6.314e+02, percent-clipped=1.0 2023-05-02 23:03:34,788 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=300617.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:04:30,742 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.2691, 3.4303, 3.6988, 2.1960, 3.1693, 2.5264, 3.7546, 3.8767], device='cuda:7'), covar=tensor([0.0226, 0.0850, 0.0624, 0.2119, 0.0834, 0.0983, 0.0511, 0.0867], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0172, 0.0172, 0.0158, 0.0149, 0.0134, 0.0146, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 23:04:31,372 INFO [train.py:904] (7/8) Epoch 30, batch 6300, loss[loss=0.2006, simple_loss=0.3028, pruned_loss=0.04919, over 16738.00 frames. ], tot_loss[loss=0.1941, simple_loss=0.2827, pruned_loss=0.05275, over 3115736.60 frames. ], batch size: 62, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:05:09,697 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8302, 1.4928, 1.7440, 1.7114, 1.8797, 1.8920, 1.6371, 1.8171], device='cuda:7'), covar=tensor([0.0270, 0.0446, 0.0241, 0.0339, 0.0313, 0.0201, 0.0488, 0.0161], device='cuda:7'), in_proj_covar=tensor([0.0200, 0.0200, 0.0188, 0.0195, 0.0212, 0.0169, 0.0206, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:7') 2023-05-02 23:05:11,421 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=300678.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:05:50,453 INFO [train.py:904] (7/8) Epoch 30, batch 6350, loss[loss=0.2071, simple_loss=0.286, pruned_loss=0.06411, over 16816.00 frames. ], tot_loss[loss=0.1955, simple_loss=0.2835, pruned_loss=0.05379, over 3106360.99 frames. ], batch size: 116, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:05:53,985 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.820e+02 3.153e+02 4.098e+02 7.645e+02, threshold=6.307e+02, percent-clipped=4.0 2023-05-02 23:06:02,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.2124, 3.4432, 3.5525, 3.5249, 3.5232, 3.4165, 3.2360, 3.4604], device='cuda:7'), covar=tensor([0.0820, 0.1184, 0.0919, 0.0847, 0.0980, 0.1122, 0.1387, 0.0864], device='cuda:7'), in_proj_covar=tensor([0.0442, 0.0504, 0.0482, 0.0447, 0.0528, 0.0511, 0.0586, 0.0412], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 23:06:28,000 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8305, 2.8976, 2.4697, 2.7379, 3.2833, 2.9211, 3.3409, 3.4891], device='cuda:7'), covar=tensor([0.0113, 0.0425, 0.0550, 0.0454, 0.0274, 0.0415, 0.0265, 0.0288], device='cuda:7'), in_proj_covar=tensor([0.0231, 0.0241, 0.0231, 0.0232, 0.0242, 0.0239, 0.0239, 0.0241], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:06:29,165 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=300729.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:07:07,469 INFO [train.py:904] (7/8) Epoch 30, batch 6400, loss[loss=0.2008, simple_loss=0.2875, pruned_loss=0.05707, over 15340.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2847, pruned_loss=0.05561, over 3074565.19 frames. ], batch size: 191, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:07:42,793 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=300777.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:08:19,410 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=300802.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:08:21,359 INFO [train.py:904] (7/8) Epoch 30, batch 6450, loss[loss=0.181, simple_loss=0.2558, pruned_loss=0.05316, over 11313.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2843, pruned_loss=0.0546, over 3085350.32 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:08:24,285 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.836e+02 3.339e+02 4.246e+02 8.060e+02, threshold=6.678e+02, percent-clipped=2.0 2023-05-02 23:09:38,131 INFO [train.py:904] (7/8) Epoch 30, batch 6500, loss[loss=0.2064, simple_loss=0.2954, pruned_loss=0.05869, over 16195.00 frames. ], tot_loss[loss=0.1947, simple_loss=0.282, pruned_loss=0.0537, over 3093323.11 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:09:52,574 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=300863.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 23:10:37,244 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0143, 4.0085, 3.9704, 3.0182, 3.8588, 1.8387, 3.5917, 3.3639], device='cuda:7'), covar=tensor([0.0251, 0.0205, 0.0235, 0.0402, 0.0190, 0.3208, 0.0213, 0.0333], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0178, 0.0217, 0.0189, 0.0195, 0.0221, 0.0206, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:10:58,881 INFO [train.py:904] (7/8) Epoch 30, batch 6550, loss[loss=0.1978, simple_loss=0.2971, pruned_loss=0.04928, over 17192.00 frames. ], tot_loss[loss=0.1962, simple_loss=0.2845, pruned_loss=0.05397, over 3102638.72 frames. ], batch size: 44, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:11:01,653 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.987e+02 2.626e+02 3.208e+02 3.708e+02 1.015e+03, threshold=6.415e+02, percent-clipped=2.0 2023-05-02 23:12:19,435 INFO [train.py:904] (7/8) Epoch 30, batch 6600, loss[loss=0.1915, simple_loss=0.2778, pruned_loss=0.05266, over 16654.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.286, pruned_loss=0.05416, over 3093032.18 frames. ], batch size: 62, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:12:48,895 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300973.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:13:39,000 INFO [train.py:904] (7/8) Epoch 30, batch 6650, loss[loss=0.1889, simple_loss=0.2734, pruned_loss=0.05217, over 16554.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2863, pruned_loss=0.05521, over 3085515.21 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:13:43,934 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 2.674e+02 3.483e+02 3.945e+02 7.489e+02, threshold=6.965e+02, percent-clipped=2.0 2023-05-02 23:14:55,126 INFO [train.py:904] (7/8) Epoch 30, batch 6700, loss[loss=0.1961, simple_loss=0.2863, pruned_loss=0.05294, over 15367.00 frames. ], tot_loss[loss=0.1985, simple_loss=0.2856, pruned_loss=0.05569, over 3074388.11 frames. ], batch size: 191, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:15:49,394 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6194, 4.5861, 4.4362, 3.5101, 4.5127, 1.6003, 4.2139, 4.0593], device='cuda:7'), covar=tensor([0.0108, 0.0101, 0.0230, 0.0488, 0.0121, 0.3302, 0.0168, 0.0336], device='cuda:7'), in_proj_covar=tensor([0.0187, 0.0177, 0.0217, 0.0188, 0.0194, 0.0221, 0.0205, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:16:11,414 INFO [train.py:904] (7/8) Epoch 30, batch 6750, loss[loss=0.1858, simple_loss=0.2788, pruned_loss=0.04642, over 16731.00 frames. ], tot_loss[loss=0.1976, simple_loss=0.2842, pruned_loss=0.05543, over 3083785.35 frames. ], batch size: 124, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:16:15,732 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 2.866e+02 3.351e+02 4.005e+02 5.751e+02, threshold=6.702e+02, percent-clipped=0.0 2023-05-02 23:17:07,637 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301140.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:17:29,025 INFO [train.py:904] (7/8) Epoch 30, batch 6800, loss[loss=0.2065, simple_loss=0.2945, pruned_loss=0.0592, over 16806.00 frames. ], tot_loss[loss=0.1982, simple_loss=0.2849, pruned_loss=0.05579, over 3071935.39 frames. ], batch size: 124, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:17:35,479 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=301158.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 23:18:28,716 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.7783, 3.8705, 3.9505, 3.7159, 3.9110, 4.2618, 3.9205, 3.6646], device='cuda:7'), covar=tensor([0.2198, 0.2107, 0.2735, 0.2329, 0.2367, 0.1639, 0.1690, 0.2444], device='cuda:7'), in_proj_covar=tensor([0.0435, 0.0645, 0.0719, 0.0526, 0.0697, 0.0735, 0.0555, 0.0701], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:18:43,945 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301201.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:18:47,471 INFO [train.py:904] (7/8) Epoch 30, batch 6850, loss[loss=0.1843, simple_loss=0.2921, pruned_loss=0.03823, over 16323.00 frames. ], tot_loss[loss=0.1989, simple_loss=0.2857, pruned_loss=0.05609, over 3071096.50 frames. ], batch size: 35, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:18:51,708 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.931e+02 2.806e+02 3.426e+02 4.107e+02 8.215e+02, threshold=6.851e+02, percent-clipped=6.0 2023-05-02 23:20:04,220 INFO [train.py:904] (7/8) Epoch 30, batch 6900, loss[loss=0.2257, simple_loss=0.2991, pruned_loss=0.07611, over 11195.00 frames. ], tot_loss[loss=0.1991, simple_loss=0.2876, pruned_loss=0.05527, over 3080277.65 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:20:35,499 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=301273.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:21:19,485 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1521, 3.2629, 3.5988, 2.0949, 3.0690, 2.3345, 3.5697, 3.6440], device='cuda:7'), covar=tensor([0.0257, 0.0930, 0.0620, 0.2269, 0.0905, 0.1048, 0.0594, 0.0977], device='cuda:7'), in_proj_covar=tensor([0.0162, 0.0173, 0.0172, 0.0158, 0.0149, 0.0134, 0.0147, 0.0185], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005], device='cuda:7') 2023-05-02 23:21:22,707 INFO [train.py:904] (7/8) Epoch 30, batch 6950, loss[loss=0.2387, simple_loss=0.3105, pruned_loss=0.08343, over 11398.00 frames. ], tot_loss[loss=0.2003, simple_loss=0.2886, pruned_loss=0.05607, over 3092722.38 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:21:26,936 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.877e+02 3.278e+02 4.034e+02 6.348e+02, threshold=6.555e+02, percent-clipped=0.0 2023-05-02 23:21:50,651 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=301321.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:22:07,237 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.8563, 3.7462, 3.9050, 4.0068, 4.0930, 3.6807, 4.0501, 4.1252], device='cuda:7'), covar=tensor([0.1668, 0.1181, 0.1319, 0.0712, 0.0653, 0.1886, 0.0954, 0.0791], device='cuda:7'), in_proj_covar=tensor([0.0677, 0.0823, 0.0951, 0.0841, 0.0639, 0.0662, 0.0703, 0.0813], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:22:15,647 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2023-05-02 23:22:38,942 INFO [train.py:904] (7/8) Epoch 30, batch 7000, loss[loss=0.2151, simple_loss=0.2886, pruned_loss=0.07081, over 11309.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2884, pruned_loss=0.05508, over 3095474.61 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:22:54,508 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3757, 3.4961, 2.0568, 3.8387, 2.5641, 3.7957, 2.1302, 2.7528], device='cuda:7'), covar=tensor([0.0363, 0.0436, 0.1904, 0.0296, 0.0974, 0.0734, 0.1704, 0.0881], device='cuda:7'), in_proj_covar=tensor([0.0178, 0.0183, 0.0197, 0.0174, 0.0180, 0.0222, 0.0205, 0.0184], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-02 23:23:55,579 INFO [train.py:904] (7/8) Epoch 30, batch 7050, loss[loss=0.1963, simple_loss=0.2885, pruned_loss=0.05207, over 16372.00 frames. ], tot_loss[loss=0.2, simple_loss=0.2895, pruned_loss=0.05522, over 3110871.41 frames. ], batch size: 146, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:24:00,751 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.808e+02 2.564e+02 3.175e+02 3.880e+02 7.852e+02, threshold=6.351e+02, percent-clipped=2.0 2023-05-02 23:24:38,959 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301430.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:25:14,081 INFO [train.py:904] (7/8) Epoch 30, batch 7100, loss[loss=0.1843, simple_loss=0.2769, pruned_loss=0.04581, over 16377.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.2881, pruned_loss=0.05518, over 3088694.88 frames. ], batch size: 35, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:25:21,060 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=301458.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:25:36,719 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4177, 2.9551, 2.7140, 2.3026, 2.2938, 2.3553, 2.9313, 2.8536], device='cuda:7'), covar=tensor([0.2489, 0.0731, 0.1534, 0.2471, 0.2315, 0.2179, 0.0492, 0.1380], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0277, 0.0316, 0.0330, 0.0307, 0.0281, 0.0305, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:26:08,415 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6424, 3.8130, 2.9007, 2.3494, 2.5357, 2.5140, 4.1599, 3.3558], device='cuda:7'), covar=tensor([0.3114, 0.0732, 0.1901, 0.2750, 0.2776, 0.2247, 0.0422, 0.1462], device='cuda:7'), in_proj_covar=tensor([0.0339, 0.0277, 0.0316, 0.0330, 0.0308, 0.0282, 0.0305, 0.0354], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:26:14,344 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301491.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:26:21,592 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=301496.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:26:21,664 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3794, 3.4944, 3.6144, 3.5962, 3.6135, 3.4485, 3.4798, 3.4919], device='cuda:7'), covar=tensor([0.0448, 0.0768, 0.0560, 0.0529, 0.0601, 0.0651, 0.0862, 0.0638], device='cuda:7'), in_proj_covar=tensor([0.0444, 0.0509, 0.0485, 0.0449, 0.0530, 0.0515, 0.0591, 0.0413], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 23:26:33,764 INFO [train.py:904] (7/8) Epoch 30, batch 7150, loss[loss=0.1791, simple_loss=0.275, pruned_loss=0.04164, over 16735.00 frames. ], tot_loss[loss=0.1988, simple_loss=0.2865, pruned_loss=0.05555, over 3062044.62 frames. ], batch size: 89, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:26:36,535 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=301506.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:26:37,393 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.858e+02 2.724e+02 3.109e+02 4.036e+02 9.536e+02, threshold=6.218e+02, percent-clipped=1.0 2023-05-02 23:27:49,050 INFO [train.py:904] (7/8) Epoch 30, batch 7200, loss[loss=0.1782, simple_loss=0.2724, pruned_loss=0.04204, over 16147.00 frames. ], tot_loss[loss=0.197, simple_loss=0.2845, pruned_loss=0.05477, over 3034475.61 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:28:48,500 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 23:29:10,980 INFO [train.py:904] (7/8) Epoch 30, batch 7250, loss[loss=0.1927, simple_loss=0.281, pruned_loss=0.05218, over 15420.00 frames. ], tot_loss[loss=0.1949, simple_loss=0.2824, pruned_loss=0.05367, over 3034428.21 frames. ], batch size: 190, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:29:15,149 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.563e+02 2.328e+02 2.684e+02 3.448e+02 5.554e+02, threshold=5.368e+02, percent-clipped=0.0 2023-05-02 23:30:24,980 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0736, 5.1735, 5.5201, 5.4625, 5.5200, 5.1841, 5.1176, 4.8750], device='cuda:7'), covar=tensor([0.0378, 0.0574, 0.0336, 0.0391, 0.0434, 0.0377, 0.1000, 0.0556], device='cuda:7'), in_proj_covar=tensor([0.0444, 0.0507, 0.0484, 0.0448, 0.0528, 0.0513, 0.0589, 0.0412], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-02 23:30:27,550 INFO [train.py:904] (7/8) Epoch 30, batch 7300, loss[loss=0.1942, simple_loss=0.285, pruned_loss=0.05175, over 15343.00 frames. ], tot_loss[loss=0.194, simple_loss=0.2817, pruned_loss=0.05311, over 3057896.06 frames. ], batch size: 191, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:30:55,784 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-05-02 23:31:44,861 INFO [train.py:904] (7/8) Epoch 30, batch 7350, loss[loss=0.2428, simple_loss=0.3061, pruned_loss=0.08978, over 10946.00 frames. ], tot_loss[loss=0.1956, simple_loss=0.283, pruned_loss=0.05407, over 3041958.17 frames. ], batch size: 247, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:31:50,962 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.792e+02 2.739e+02 3.075e+02 3.816e+02 5.602e+02, threshold=6.150e+02, percent-clipped=1.0 2023-05-02 23:32:44,879 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301743.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:33:00,870 INFO [train.py:904] (7/8) Epoch 30, batch 7400, loss[loss=0.1984, simple_loss=0.2937, pruned_loss=0.05154, over 16675.00 frames. ], tot_loss[loss=0.1966, simple_loss=0.2841, pruned_loss=0.05452, over 3048546.32 frames. ], batch size: 134, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:33:54,214 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=301786.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:34:10,808 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=301796.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:34:23,458 INFO [train.py:904] (7/8) Epoch 30, batch 7450, loss[loss=0.1825, simple_loss=0.2724, pruned_loss=0.04628, over 16434.00 frames. ], tot_loss[loss=0.1975, simple_loss=0.2848, pruned_loss=0.05511, over 3053082.07 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:34:24,288 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301804.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:34:29,850 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.990e+02 2.877e+02 3.334e+02 4.200e+02 6.879e+02, threshold=6.668e+02, percent-clipped=3.0 2023-05-02 23:35:30,254 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=301844.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:35:45,211 INFO [train.py:904] (7/8) Epoch 30, batch 7500, loss[loss=0.1918, simple_loss=0.2777, pruned_loss=0.05299, over 16523.00 frames. ], tot_loss[loss=0.1972, simple_loss=0.2853, pruned_loss=0.05455, over 3062246.36 frames. ], batch size: 62, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:36:03,135 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.5639, 3.6756, 2.7819, 2.2588, 2.4694, 2.4547, 3.9111, 3.3147], device='cuda:7'), covar=tensor([0.3117, 0.0668, 0.1961, 0.2926, 0.2839, 0.2282, 0.0447, 0.1312], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0276, 0.0316, 0.0330, 0.0307, 0.0281, 0.0305, 0.0353], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:36:57,925 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301899.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:37:05,279 INFO [train.py:904] (7/8) Epoch 30, batch 7550, loss[loss=0.2093, simple_loss=0.2952, pruned_loss=0.06174, over 16265.00 frames. ], tot_loss[loss=0.1968, simple_loss=0.2845, pruned_loss=0.05455, over 3062379.56 frames. ], batch size: 165, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:37:07,805 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=301905.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:37:11,511 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.823e+02 2.767e+02 3.333e+02 4.100e+02 7.380e+02, threshold=6.666e+02, percent-clipped=1.0 2023-05-02 23:37:12,938 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-05-02 23:37:27,008 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-05-02 23:38:12,619 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.9907, 5.0108, 4.8033, 4.0921, 4.9060, 1.9680, 4.6375, 4.3611], device='cuda:7'), covar=tensor([0.0108, 0.0100, 0.0202, 0.0434, 0.0099, 0.2739, 0.0127, 0.0327], device='cuda:7'), in_proj_covar=tensor([0.0183, 0.0174, 0.0213, 0.0185, 0.0190, 0.0217, 0.0201, 0.0178], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:38:17,535 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=3.97 vs. limit=5.0 2023-05-02 23:38:23,964 INFO [train.py:904] (7/8) Epoch 30, batch 7600, loss[loss=0.1716, simple_loss=0.2587, pruned_loss=0.0423, over 17042.00 frames. ], tot_loss[loss=0.1964, simple_loss=0.2837, pruned_loss=0.05457, over 3054829.34 frames. ], batch size: 53, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:38:34,819 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301960.0, num_to_drop=1, layers_to_drop={2} 2023-05-02 23:38:44,345 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=301966.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:39:47,161 INFO [train.py:904] (7/8) Epoch 30, batch 7650, loss[loss=0.2171, simple_loss=0.3036, pruned_loss=0.06532, over 16514.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2847, pruned_loss=0.0555, over 3074247.90 frames. ], batch size: 68, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:39:53,140 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 2.050e+02 2.820e+02 3.410e+02 4.459e+02 8.175e+02, threshold=6.820e+02, percent-clipped=3.0 2023-05-02 23:41:03,395 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.50 vs. limit=5.0 2023-05-02 23:41:05,120 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-05-02 23:41:06,721 INFO [train.py:904] (7/8) Epoch 30, batch 7700, loss[loss=0.2504, simple_loss=0.3171, pruned_loss=0.09182, over 11375.00 frames. ], tot_loss[loss=0.1977, simple_loss=0.2843, pruned_loss=0.05555, over 3080456.25 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:41:57,506 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302086.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:42:18,159 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302099.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:42:26,386 INFO [train.py:904] (7/8) Epoch 30, batch 7750, loss[loss=0.2321, simple_loss=0.3235, pruned_loss=0.07037, over 16429.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2848, pruned_loss=0.05597, over 3079046.79 frames. ], batch size: 146, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:42:33,837 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.636e+02 3.091e+02 3.789e+02 8.720e+02, threshold=6.183e+02, percent-clipped=2.0 2023-05-02 23:42:48,972 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-05-02 23:43:12,803 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302134.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:43:23,436 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-02 23:43:39,405 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-05-02 23:43:41,543 INFO [train.py:904] (7/8) Epoch 30, batch 7800, loss[loss=0.2622, simple_loss=0.3195, pruned_loss=0.1025, over 11130.00 frames. ], tot_loss[loss=0.1997, simple_loss=0.2858, pruned_loss=0.05679, over 3078622.25 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:44:55,082 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302200.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:45:00,021 INFO [train.py:904] (7/8) Epoch 30, batch 7850, loss[loss=0.2159, simple_loss=0.2917, pruned_loss=0.07007, over 11547.00 frames. ], tot_loss[loss=0.1993, simple_loss=0.2864, pruned_loss=0.05608, over 3091321.73 frames. ], batch size: 249, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:45:06,745 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.895e+02 2.702e+02 3.253e+02 3.935e+02 9.747e+02, threshold=6.506e+02, percent-clipped=3.0 2023-05-02 23:45:31,687 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.60 vs. limit=2.0 2023-05-02 23:45:58,898 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3408, 3.8313, 3.8065, 2.4500, 3.4800, 3.8446, 3.4652, 2.0562], device='cuda:7'), covar=tensor([0.0590, 0.0063, 0.0068, 0.0473, 0.0122, 0.0119, 0.0134, 0.0547], device='cuda:7'), in_proj_covar=tensor([0.0139, 0.0092, 0.0093, 0.0136, 0.0103, 0.0117, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 23:46:15,241 INFO [train.py:904] (7/8) Epoch 30, batch 7900, loss[loss=0.1834, simple_loss=0.2791, pruned_loss=0.04391, over 17148.00 frames. ], tot_loss[loss=0.1979, simple_loss=0.2853, pruned_loss=0.05529, over 3096547.28 frames. ], batch size: 47, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:46:16,931 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302255.0, num_to_drop=1, layers_to_drop={3} 2023-05-02 23:46:17,049 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5010, 4.5679, 4.3984, 4.1051, 4.1109, 4.5048, 4.2230, 4.2200], device='cuda:7'), covar=tensor([0.0645, 0.0580, 0.0310, 0.0319, 0.0841, 0.0456, 0.0622, 0.0648], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0473, 0.0365, 0.0367, 0.0361, 0.0421, 0.0252, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:46:19,162 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8012, 5.1232, 5.3375, 5.0437, 5.1226, 5.6762, 5.1248, 4.9182], device='cuda:7'), covar=tensor([0.1206, 0.1883, 0.2339, 0.1753, 0.2268, 0.0868, 0.1694, 0.2419], device='cuda:7'), in_proj_covar=tensor([0.0433, 0.0642, 0.0718, 0.0524, 0.0694, 0.0736, 0.0555, 0.0699], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:46:25,341 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302261.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:46:25,532 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302261.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:46:28,383 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4076, 4.4980, 4.3038, 4.0000, 4.0050, 4.4245, 4.1280, 4.1494], device='cuda:7'), covar=tensor([0.0688, 0.0600, 0.0334, 0.0351, 0.0843, 0.0516, 0.0647, 0.0671], device='cuda:7'), in_proj_covar=tensor([0.0312, 0.0473, 0.0365, 0.0367, 0.0361, 0.0421, 0.0252, 0.0436], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:46:55,552 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.0200, 2.2621, 2.1997, 3.6086, 2.1648, 2.5314, 2.3039, 2.3447], device='cuda:7'), covar=tensor([0.1612, 0.3533, 0.3371, 0.0732, 0.4325, 0.2487, 0.3764, 0.3445], device='cuda:7'), in_proj_covar=tensor([0.0425, 0.0480, 0.0390, 0.0340, 0.0449, 0.0551, 0.0453, 0.0561], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:47:16,598 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.4367, 4.4261, 4.3107, 3.4982, 4.3882, 1.6025, 4.1475, 3.8517], device='cuda:7'), covar=tensor([0.0144, 0.0116, 0.0215, 0.0371, 0.0109, 0.3270, 0.0161, 0.0346], device='cuda:7'), in_proj_covar=tensor([0.0184, 0.0175, 0.0215, 0.0186, 0.0191, 0.0218, 0.0202, 0.0179], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-02 23:47:22,536 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302296.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:47:32,972 INFO [train.py:904] (7/8) Epoch 30, batch 7950, loss[loss=0.2377, simple_loss=0.298, pruned_loss=0.0887, over 11877.00 frames. ], tot_loss[loss=0.1984, simple_loss=0.2856, pruned_loss=0.05559, over 3110702.09 frames. ], batch size: 246, lr: 2.23e-03, grad_scale: 4.0 2023-05-02 23:47:41,024 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.902e+02 2.712e+02 3.186e+02 3.569e+02 8.017e+02, threshold=6.372e+02, percent-clipped=2.0 2023-05-02 23:47:42,690 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.3455, 3.6730, 3.6953, 2.4970, 3.4019, 3.7286, 3.3917, 2.0887], device='cuda:7'), covar=tensor([0.0558, 0.0080, 0.0072, 0.0438, 0.0120, 0.0130, 0.0129, 0.0528], device='cuda:7'), in_proj_covar=tensor([0.0140, 0.0092, 0.0093, 0.0136, 0.0103, 0.0118, 0.0100, 0.0132], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:7') 2023-05-02 23:48:48,634 INFO [train.py:904] (7/8) Epoch 30, batch 8000, loss[loss=0.1976, simple_loss=0.2918, pruned_loss=0.05168, over 16871.00 frames. ], tot_loss[loss=0.1994, simple_loss=0.2862, pruned_loss=0.05631, over 3096163.83 frames. ], batch size: 116, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:48:50,245 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-05-02 23:48:54,537 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302357.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:49:57,505 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302399.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:50:03,902 INFO [train.py:904] (7/8) Epoch 30, batch 8050, loss[loss=0.1641, simple_loss=0.264, pruned_loss=0.03214, over 16777.00 frames. ], tot_loss[loss=0.1992, simple_loss=0.286, pruned_loss=0.05616, over 3076395.34 frames. ], batch size: 89, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:50:11,662 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.866e+02 2.982e+02 3.316e+02 3.990e+02 1.007e+03, threshold=6.633e+02, percent-clipped=4.0 2023-05-02 23:50:28,318 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.5141, 4.6604, 4.8243, 4.6176, 4.6923, 5.1991, 4.6733, 4.4157], device='cuda:7'), covar=tensor([0.1349, 0.1797, 0.2189, 0.1921, 0.2266, 0.0907, 0.1656, 0.2411], device='cuda:7'), in_proj_covar=tensor([0.0432, 0.0640, 0.0716, 0.0524, 0.0693, 0.0733, 0.0554, 0.0697], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:51:10,641 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302447.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:51:21,030 INFO [train.py:904] (7/8) Epoch 30, batch 8100, loss[loss=0.2196, simple_loss=0.2927, pruned_loss=0.07321, over 11625.00 frames. ], tot_loss[loss=0.198, simple_loss=0.2851, pruned_loss=0.0555, over 3076680.42 frames. ], batch size: 248, lr: 2.23e-03, grad_scale: 8.0 2023-05-02 23:51:56,212 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302477.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:52:36,006 INFO [train.py:904] (7/8) Epoch 30, batch 8150, loss[loss=0.2088, simple_loss=0.2827, pruned_loss=0.06749, over 11761.00 frames. ], tot_loss[loss=0.1967, simple_loss=0.2833, pruned_loss=0.05508, over 3055278.43 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:52:43,488 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.950e+02 2.749e+02 3.387e+02 4.131e+02 6.245e+02, threshold=6.774e+02, percent-clipped=0.0 2023-05-02 23:53:28,207 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302538.0, num_to_drop=1, layers_to_drop={0} 2023-05-02 23:53:52,238 INFO [train.py:904] (7/8) Epoch 30, batch 8200, loss[loss=0.191, simple_loss=0.2832, pruned_loss=0.04939, over 15590.00 frames. ], tot_loss[loss=0.1952, simple_loss=0.2813, pruned_loss=0.0546, over 3061485.33 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:53:53,963 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302555.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:53:55,135 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302556.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:54:03,397 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302561.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:55:13,043 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302603.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:55:13,855 INFO [train.py:904] (7/8) Epoch 30, batch 8250, loss[loss=0.1531, simple_loss=0.2446, pruned_loss=0.03081, over 11914.00 frames. ], tot_loss[loss=0.1919, simple_loss=0.2798, pruned_loss=0.052, over 3055373.71 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:55:22,074 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.554e+02 2.948e+02 3.684e+02 6.863e+02, threshold=5.896e+02, percent-clipped=1.0 2023-05-02 23:55:23,079 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302609.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:56:02,676 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302632.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:56:35,858 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302652.0, num_to_drop=0, layers_to_drop=set() 2023-05-02 23:56:38,714 INFO [train.py:904] (7/8) Epoch 30, batch 8300, loss[loss=0.1854, simple_loss=0.284, pruned_loss=0.04335, over 16387.00 frames. ], tot_loss[loss=0.1875, simple_loss=0.277, pruned_loss=0.04895, over 3050421.51 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:56:48,004 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4626, 3.0377, 2.8003, 2.3105, 2.2194, 2.3509, 2.9895, 2.8224], device='cuda:7'), covar=tensor([0.2789, 0.0710, 0.1551, 0.3128, 0.3122, 0.2494, 0.0504, 0.1622], device='cuda:7'), in_proj_covar=tensor([0.0338, 0.0275, 0.0316, 0.0329, 0.0307, 0.0281, 0.0304, 0.0352], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-02 23:57:43,551 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302693.0, num_to_drop=1, layers_to_drop={1} 2023-05-02 23:58:01,703 INFO [train.py:904] (7/8) Epoch 30, batch 8350, loss[loss=0.1874, simple_loss=0.2831, pruned_loss=0.04587, over 16632.00 frames. ], tot_loss[loss=0.1855, simple_loss=0.2767, pruned_loss=0.04716, over 3049007.99 frames. ], batch size: 134, lr: 2.22e-03, grad_scale: 8.0 2023-05-02 23:58:09,439 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.283e+02 2.623e+02 3.174e+02 4.916e+02, threshold=5.246e+02, percent-clipped=0.0 2023-05-02 23:58:35,910 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-05-02 23:59:22,770 INFO [train.py:904] (7/8) Epoch 30, batch 8400, loss[loss=0.1769, simple_loss=0.2595, pruned_loss=0.04716, over 12016.00 frames. ], tot_loss[loss=0.1827, simple_loss=0.2743, pruned_loss=0.04552, over 3027601.07 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:00:44,440 INFO [train.py:904] (7/8) Epoch 30, batch 8450, loss[loss=0.1713, simple_loss=0.2661, pruned_loss=0.03826, over 15194.00 frames. ], tot_loss[loss=0.1799, simple_loss=0.2723, pruned_loss=0.04378, over 3017722.56 frames. ], batch size: 190, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:00:52,080 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.019e+02 2.324e+02 2.808e+02 4.179e+02, threshold=4.647e+02, percent-clipped=0.0 2023-05-03 00:01:06,769 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-05-03 00:01:20,467 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-05-03 00:01:31,163 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302833.0, num_to_drop=1, layers_to_drop={0} 2023-05-03 00:02:04,575 INFO [train.py:904] (7/8) Epoch 30, batch 8500, loss[loss=0.1581, simple_loss=0.2482, pruned_loss=0.03397, over 15340.00 frames. ], tot_loss[loss=0.176, simple_loss=0.2689, pruned_loss=0.04155, over 3042526.37 frames. ], batch size: 190, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:02:08,506 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302856.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:02:08,685 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1134, 2.3421, 2.4426, 3.0218, 1.9655, 3.2054, 1.9923, 2.8455], device='cuda:7'), covar=tensor([0.1131, 0.0612, 0.0993, 0.0184, 0.0085, 0.0386, 0.1405, 0.0612], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0182, 0.0201, 0.0206, 0.0206, 0.0218, 0.0211, 0.0200], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-03 00:02:48,223 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=302881.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:03:16,214 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-05-03 00:03:22,943 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.8786, 2.6928, 2.9313, 2.1523, 2.7348, 2.1744, 2.7362, 2.8906], device='cuda:7'), covar=tensor([0.0283, 0.1017, 0.0456, 0.1873, 0.0805, 0.0971, 0.0611, 0.0863], device='cuda:7'), in_proj_covar=tensor([0.0159, 0.0169, 0.0169, 0.0156, 0.0147, 0.0132, 0.0144, 0.0182], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0005], device='cuda:7') 2023-05-03 00:03:26,605 INFO [train.py:904] (7/8) Epoch 30, batch 8550, loss[loss=0.1808, simple_loss=0.2768, pruned_loss=0.04244, over 16235.00 frames. ], tot_loss[loss=0.1738, simple_loss=0.2665, pruned_loss=0.04057, over 3026235.38 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:03:28,403 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=302904.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:03:38,838 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.175e+02 2.493e+02 2.958e+02 5.835e+02, threshold=4.986e+02, percent-clipped=2.0 2023-05-03 00:04:45,068 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=302942.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:05:04,938 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=302952.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:05:08,030 INFO [train.py:904] (7/8) Epoch 30, batch 8600, loss[loss=0.1771, simple_loss=0.2755, pruned_loss=0.03931, over 16401.00 frames. ], tot_loss[loss=0.1731, simple_loss=0.2668, pruned_loss=0.0397, over 3033003.87 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:06:16,094 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=302988.0, num_to_drop=1, layers_to_drop={3} 2023-05-03 00:06:39,315 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303000.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:06:46,605 INFO [train.py:904] (7/8) Epoch 30, batch 8650, loss[loss=0.1754, simple_loss=0.2729, pruned_loss=0.039, over 16664.00 frames. ], tot_loss[loss=0.171, simple_loss=0.2651, pruned_loss=0.03844, over 3026638.73 frames. ], batch size: 134, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:06:58,838 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 2.161e+02 2.643e+02 3.389e+02 5.700e+02, threshold=5.286e+02, percent-clipped=3.0 2023-05-03 00:08:31,322 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303052.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:08:34,580 INFO [train.py:904] (7/8) Epoch 30, batch 8700, loss[loss=0.1547, simple_loss=0.2464, pruned_loss=0.0315, over 12433.00 frames. ], tot_loss[loss=0.1682, simple_loss=0.2622, pruned_loss=0.03708, over 3018250.71 frames. ], batch size: 247, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:10:14,436 INFO [train.py:904] (7/8) Epoch 30, batch 8750, loss[loss=0.1638, simple_loss=0.266, pruned_loss=0.03074, over 15343.00 frames. ], tot_loss[loss=0.168, simple_loss=0.2624, pruned_loss=0.03678, over 3030377.65 frames. ], batch size: 190, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:10:25,190 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.109e+02 2.507e+02 3.042e+02 7.094e+02, threshold=5.015e+02, percent-clipped=1.0 2023-05-03 00:10:26,697 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6373, 2.6174, 1.9356, 2.7653, 2.1289, 2.8147, 2.1699, 2.4081], device='cuda:7'), covar=tensor([0.0327, 0.0361, 0.1322, 0.0312, 0.0720, 0.0571, 0.1274, 0.0622], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0178, 0.0193, 0.0170, 0.0177, 0.0216, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-03 00:10:38,690 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303113.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:11:23,664 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303133.0, num_to_drop=1, layers_to_drop={1} 2023-05-03 00:11:31,988 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6959, 2.6938, 1.9601, 2.8664, 2.1829, 2.8951, 2.1696, 2.4387], device='cuda:7'), covar=tensor([0.0334, 0.0338, 0.1424, 0.0299, 0.0795, 0.0408, 0.1378, 0.0689], device='cuda:7'), in_proj_covar=tensor([0.0174, 0.0178, 0.0193, 0.0169, 0.0176, 0.0216, 0.0201, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:7') 2023-05-03 00:11:50,577 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.7407, 2.0093, 2.3520, 2.7533, 2.7088, 3.1946, 2.2460, 3.1850], device='cuda:7'), covar=tensor([0.0279, 0.0586, 0.0430, 0.0343, 0.0396, 0.0202, 0.0568, 0.0186], device='cuda:7'), in_proj_covar=tensor([0.0198, 0.0199, 0.0188, 0.0192, 0.0210, 0.0167, 0.0205, 0.0169], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:12:06,334 INFO [train.py:904] (7/8) Epoch 30, batch 8800, loss[loss=0.1741, simple_loss=0.2593, pruned_loss=0.04446, over 12655.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2608, pruned_loss=0.03567, over 3028799.97 frames. ], batch size: 246, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:13:03,256 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303181.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:13:50,927 INFO [train.py:904] (7/8) Epoch 30, batch 8850, loss[loss=0.1659, simple_loss=0.2711, pruned_loss=0.03032, over 16658.00 frames. ], tot_loss[loss=0.1664, simple_loss=0.2631, pruned_loss=0.03485, over 3021474.27 frames. ], batch size: 134, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:14:00,696 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.249e+02 2.609e+02 3.163e+02 6.486e+02, threshold=5.217e+02, percent-clipped=3.0 2023-05-03 00:14:01,741 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.0285, 3.1924, 3.1925, 2.1660, 2.9088, 3.2766, 3.1252, 1.9755], device='cuda:7'), covar=tensor([0.0586, 0.0078, 0.0080, 0.0480, 0.0160, 0.0101, 0.0117, 0.0525], device='cuda:7'), in_proj_covar=tensor([0.0138, 0.0090, 0.0092, 0.0134, 0.0103, 0.0115, 0.0098, 0.0130], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-03 00:14:32,334 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8180, 1.3657, 1.8087, 1.7365, 1.8797, 1.9468, 1.7567, 1.8388], device='cuda:7'), covar=tensor([0.0310, 0.0494, 0.0266, 0.0356, 0.0378, 0.0227, 0.0506, 0.0178], device='cuda:7'), in_proj_covar=tensor([0.0198, 0.0199, 0.0187, 0.0192, 0.0210, 0.0167, 0.0205, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:15:03,443 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303237.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:15:17,356 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8485, 1.4299, 1.7647, 1.8455, 1.9200, 1.9599, 1.7411, 1.9220], device='cuda:7'), covar=tensor([0.0293, 0.0521, 0.0278, 0.0344, 0.0343, 0.0240, 0.0516, 0.0184], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0199, 0.0187, 0.0192, 0.0210, 0.0167, 0.0205, 0.0168], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:15:20,844 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.5123, 5.5629, 5.3441, 4.9105, 4.9781, 5.4303, 5.3650, 5.1155], device='cuda:7'), covar=tensor([0.0769, 0.0681, 0.0382, 0.0395, 0.1224, 0.0829, 0.0294, 0.0883], device='cuda:7'), in_proj_covar=tensor([0.0308, 0.0466, 0.0360, 0.0361, 0.0354, 0.0415, 0.0248, 0.0428], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:15:38,695 INFO [train.py:904] (7/8) Epoch 30, batch 8900, loss[loss=0.1646, simple_loss=0.2673, pruned_loss=0.03097, over 16675.00 frames. ], tot_loss[loss=0.1661, simple_loss=0.2637, pruned_loss=0.0343, over 3033393.16 frames. ], batch size: 89, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:15:49,617 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303258.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:17:03,923 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303288.0, num_to_drop=1, layers_to_drop={0} 2023-05-03 00:17:15,879 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.0190, 4.0565, 4.3176, 4.2902, 4.3114, 4.0939, 4.0718, 4.0791], device='cuda:7'), covar=tensor([0.0342, 0.0736, 0.0437, 0.0431, 0.0488, 0.0495, 0.0851, 0.0480], device='cuda:7'), in_proj_covar=tensor([0.0430, 0.0489, 0.0469, 0.0435, 0.0514, 0.0496, 0.0568, 0.0400], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:7') 2023-05-03 00:17:43,164 INFO [train.py:904] (7/8) Epoch 30, batch 8950, loss[loss=0.154, simple_loss=0.2447, pruned_loss=0.03162, over 15314.00 frames. ], tot_loss[loss=0.1662, simple_loss=0.2633, pruned_loss=0.03458, over 3058105.64 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:17:53,314 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.462e+02 2.074e+02 2.389e+02 2.805e+02 4.954e+02, threshold=4.779e+02, percent-clipped=0.0 2023-05-03 00:18:16,133 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303319.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:18:22,511 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6343, 2.1528, 1.8462, 1.9364, 2.4328, 2.0957, 1.9381, 2.5348], device='cuda:7'), covar=tensor([0.0215, 0.0490, 0.0644, 0.0594, 0.0357, 0.0494, 0.0256, 0.0349], device='cuda:7'), in_proj_covar=tensor([0.0227, 0.0240, 0.0230, 0.0231, 0.0242, 0.0238, 0.0236, 0.0239], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:18:53,964 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303336.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:19:32,192 INFO [train.py:904] (7/8) Epoch 30, batch 9000, loss[loss=0.1489, simple_loss=0.2466, pruned_loss=0.0256, over 15158.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2603, pruned_loss=0.03349, over 3066994.60 frames. ], batch size: 190, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:19:32,193 INFO [train.py:929] (7/8) Computing validation loss 2023-05-03 00:19:42,081 INFO [train.py:938] (7/8) Epoch 30, validation: loss=0.1431, simple_loss=0.2465, pruned_loss=0.01984, over 944034.00 frames. 2023-05-03 00:19:42,082 INFO [train.py:939] (7/8) Maximum memory allocated so far is 17846MB 2023-05-03 00:20:14,574 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.1346, 5.1105, 4.8803, 4.2829, 4.9262, 2.0324, 4.6628, 4.7222], device='cuda:7'), covar=tensor([0.0119, 0.0097, 0.0234, 0.0373, 0.0141, 0.2721, 0.0155, 0.0244], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0172, 0.0210, 0.0181, 0.0188, 0.0216, 0.0199, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:21:28,488 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303403.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:21:29,175 INFO [train.py:904] (7/8) Epoch 30, batch 9050, loss[loss=0.1692, simple_loss=0.2591, pruned_loss=0.03962, over 16297.00 frames. ], tot_loss[loss=0.1646, simple_loss=0.2614, pruned_loss=0.03393, over 3083750.35 frames. ], batch size: 166, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:21:38,847 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303408.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:21:39,558 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.159e+02 2.466e+02 3.038e+02 8.070e+02, threshold=4.932e+02, percent-clipped=3.0 2023-05-03 00:22:08,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6560, 4.8279, 4.9896, 4.7770, 4.9010, 5.3856, 4.9478, 4.6446], device='cuda:7'), covar=tensor([0.1152, 0.1919, 0.2113, 0.2022, 0.2157, 0.0930, 0.1559, 0.2445], device='cuda:7'), in_proj_covar=tensor([0.0419, 0.0624, 0.0697, 0.0509, 0.0673, 0.0717, 0.0540, 0.0678], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-03 00:23:13,947 INFO [train.py:904] (7/8) Epoch 30, batch 9100, loss[loss=0.1738, simple_loss=0.2734, pruned_loss=0.03714, over 16175.00 frames. ], tot_loss[loss=0.1651, simple_loss=0.2613, pruned_loss=0.0344, over 3109422.78 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:23:33,679 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303464.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:23:59,787 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8860, 2.7579, 2.6443, 1.9452, 2.6081, 2.8042, 2.6795, 1.8356], device='cuda:7'), covar=tensor([0.0521, 0.0105, 0.0096, 0.0436, 0.0164, 0.0130, 0.0128, 0.0547], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0090, 0.0091, 0.0133, 0.0102, 0.0114, 0.0097, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-03 00:24:17,582 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-05-03 00:24:45,753 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.1291, 3.2160, 3.6887, 2.1332, 3.0840, 2.3127, 3.5278, 3.4933], device='cuda:7'), covar=tensor([0.0257, 0.0968, 0.0540, 0.2228, 0.0854, 0.1072, 0.0668, 0.0995], device='cuda:7'), in_proj_covar=tensor([0.0157, 0.0167, 0.0167, 0.0155, 0.0146, 0.0131, 0.0143, 0.0180], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:7') 2023-05-03 00:24:54,436 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-05-03 00:25:05,278 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5706, 3.5290, 3.5403, 2.6723, 3.3977, 1.9744, 3.2567, 2.8414], device='cuda:7'), covar=tensor([0.0149, 0.0142, 0.0199, 0.0218, 0.0126, 0.2739, 0.0140, 0.0324], device='cuda:7'), in_proj_covar=tensor([0.0182, 0.0173, 0.0212, 0.0182, 0.0189, 0.0217, 0.0200, 0.0176], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:25:12,853 INFO [train.py:904] (7/8) Epoch 30, batch 9150, loss[loss=0.1647, simple_loss=0.2606, pruned_loss=0.03443, over 15471.00 frames. ], tot_loss[loss=0.1654, simple_loss=0.262, pruned_loss=0.03443, over 3092356.20 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:25:25,339 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 2.268e+02 2.639e+02 3.152e+02 6.290e+02, threshold=5.277e+02, percent-clipped=4.0 2023-05-03 00:26:18,287 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7877, 4.6481, 4.8395, 4.9530, 5.1235, 4.6755, 5.1476, 5.1392], device='cuda:7'), covar=tensor([0.2017, 0.1194, 0.1577, 0.0808, 0.0618, 0.0987, 0.0760, 0.1111], device='cuda:7'), in_proj_covar=tensor([0.0659, 0.0800, 0.0923, 0.0826, 0.0623, 0.0646, 0.0684, 0.0791], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:26:23,053 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303537.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:26:49,853 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.6166, 1.9648, 2.2361, 2.6152, 2.6417, 2.9560, 2.0704, 2.9103], device='cuda:7'), covar=tensor([0.0304, 0.0627, 0.0435, 0.0417, 0.0416, 0.0252, 0.0640, 0.0187], device='cuda:7'), in_proj_covar=tensor([0.0197, 0.0198, 0.0187, 0.0192, 0.0210, 0.0167, 0.0205, 0.0167], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:26:54,957 INFO [train.py:904] (7/8) Epoch 30, batch 9200, loss[loss=0.1788, simple_loss=0.274, pruned_loss=0.04181, over 16683.00 frames. ], tot_loss[loss=0.1627, simple_loss=0.2582, pruned_loss=0.03358, over 3105092.97 frames. ], batch size: 134, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:27:32,803 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303574.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:27:52,355 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303585.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:28:29,165 INFO [train.py:904] (7/8) Epoch 30, batch 9250, loss[loss=0.1625, simple_loss=0.2559, pruned_loss=0.03455, over 16318.00 frames. ], tot_loss[loss=0.1619, simple_loss=0.2574, pruned_loss=0.03324, over 3094133.92 frames. ], batch size: 146, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:28:41,902 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.600e+02 2.148e+02 2.583e+02 3.115e+02 7.022e+02, threshold=5.165e+02, percent-clipped=1.0 2023-05-03 00:28:49,588 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303614.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:29:39,017 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303635.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:30:19,439 INFO [train.py:904] (7/8) Epoch 30, batch 9300, loss[loss=0.1502, simple_loss=0.2419, pruned_loss=0.02918, over 16912.00 frames. ], tot_loss[loss=0.1608, simple_loss=0.2557, pruned_loss=0.03291, over 3094040.26 frames. ], batch size: 116, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:31:27,497 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.3848, 4.0366, 4.5489, 2.4241, 4.7423, 4.8291, 3.6355, 3.8021], device='cuda:7'), covar=tensor([0.0568, 0.0272, 0.0234, 0.1093, 0.0065, 0.0126, 0.0335, 0.0357], device='cuda:7'), in_proj_covar=tensor([0.0145, 0.0107, 0.0099, 0.0135, 0.0084, 0.0128, 0.0127, 0.0127], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-03 00:32:00,704 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4895, 3.0477, 2.7036, 2.2652, 2.1903, 2.2928, 3.0721, 2.8710], device='cuda:7'), covar=tensor([0.2657, 0.0703, 0.1777, 0.2921, 0.2892, 0.2340, 0.0503, 0.1498], device='cuda:7'), in_proj_covar=tensor([0.0329, 0.0268, 0.0307, 0.0321, 0.0297, 0.0274, 0.0296, 0.0342], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-03 00:32:05,585 INFO [train.py:904] (7/8) Epoch 30, batch 9350, loss[loss=0.1552, simple_loss=0.2553, pruned_loss=0.02753, over 16802.00 frames. ], tot_loss[loss=0.1603, simple_loss=0.2555, pruned_loss=0.03258, over 3110049.10 frames. ], batch size: 102, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:32:13,951 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303708.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:32:16,692 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.024e+02 2.345e+02 2.902e+02 6.854e+02, threshold=4.690e+02, percent-clipped=2.0 2023-05-03 00:33:46,254 INFO [train.py:904] (7/8) Epoch 30, batch 9400, loss[loss=0.16, simple_loss=0.2632, pruned_loss=0.0284, over 16189.00 frames. ], tot_loss[loss=0.1599, simple_loss=0.2554, pruned_loss=0.03225, over 3095406.93 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:33:50,782 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303756.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:33:56,792 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303759.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:35:26,329 INFO [train.py:904] (7/8) Epoch 30, batch 9450, loss[loss=0.1665, simple_loss=0.2625, pruned_loss=0.03526, over 16144.00 frames. ], tot_loss[loss=0.1611, simple_loss=0.2569, pruned_loss=0.03263, over 3082126.90 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:35:36,998 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.278e+02 2.120e+02 2.504e+02 3.073e+02 7.133e+02, threshold=5.009e+02, percent-clipped=4.0 2023-05-03 00:35:38,602 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303810.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:36:01,118 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=2.95 vs. limit=5.0 2023-05-03 00:37:08,354 INFO [train.py:904] (7/8) Epoch 30, batch 9500, loss[loss=0.1549, simple_loss=0.2543, pruned_loss=0.02782, over 15318.00 frames. ], tot_loss[loss=0.1607, simple_loss=0.2563, pruned_loss=0.03255, over 3071532.62 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:37:43,927 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303871.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:38:36,419 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303896.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:38:54,505 INFO [train.py:904] (7/8) Epoch 30, batch 9550, loss[loss=0.1868, simple_loss=0.2832, pruned_loss=0.04516, over 16910.00 frames. ], tot_loss[loss=0.161, simple_loss=0.2562, pruned_loss=0.03292, over 3055291.11 frames. ], batch size: 125, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:39:08,518 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 2.050e+02 2.496e+02 3.159e+02 5.329e+02, threshold=4.992e+02, percent-clipped=3.0 2023-05-03 00:39:18,130 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=303914.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:39:49,806 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=303930.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:40:25,405 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-05-03 00:40:36,090 INFO [train.py:904] (7/8) Epoch 30, batch 9600, loss[loss=0.1773, simple_loss=0.282, pruned_loss=0.0363, over 16257.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2569, pruned_loss=0.03334, over 3046009.26 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:40:43,449 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=303957.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:40:48,325 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-05-03 00:40:53,326 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=303962.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:41:16,204 INFO [zipformer.py:625] (7/8) warmup_begin=2666.7, warmup_end=3333.3, batch_count=303974.0, num_to_drop=1, layers_to_drop={0} 2023-05-03 00:42:31,203 INFO [train.py:904] (7/8) Epoch 30, batch 9650, loss[loss=0.1648, simple_loss=0.2617, pruned_loss=0.034, over 15365.00 frames. ], tot_loss[loss=0.1636, simple_loss=0.2593, pruned_loss=0.0339, over 3051202.28 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:42:48,099 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.176e+02 2.625e+02 3.264e+02 9.247e+02, threshold=5.250e+02, percent-clipped=2.0 2023-05-03 00:43:41,237 INFO [zipformer.py:625] (7/8) warmup_begin=3333.3, warmup_end=4000.0, batch_count=304035.0, num_to_drop=1, layers_to_drop={1} 2023-05-03 00:44:20,837 INFO [train.py:904] (7/8) Epoch 30, batch 9700, loss[loss=0.1867, simple_loss=0.2755, pruned_loss=0.04897, over 12630.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2585, pruned_loss=0.03391, over 3055402.20 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:44:30,781 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=304059.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:46:04,102 INFO [train.py:904] (7/8) Epoch 30, batch 9750, loss[loss=0.158, simple_loss=0.2456, pruned_loss=0.03514, over 12409.00 frames. ], tot_loss[loss=0.1622, simple_loss=0.257, pruned_loss=0.03365, over 3062672.17 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:46:10,429 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=304107.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:46:12,629 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.1330, 2.5473, 2.5956, 1.9471, 2.7391, 2.8072, 2.5438, 2.4890], device='cuda:7'), covar=tensor([0.0632, 0.0270, 0.0239, 0.0991, 0.0137, 0.0250, 0.0429, 0.0451], device='cuda:7'), in_proj_covar=tensor([0.0144, 0.0107, 0.0098, 0.0134, 0.0084, 0.0128, 0.0127, 0.0126], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0003], device='cuda:7') 2023-05-03 00:46:17,049 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.052e+02 2.422e+02 2.861e+02 5.308e+02, threshold=4.844e+02, percent-clipped=2.0 2023-05-03 00:47:41,831 INFO [train.py:904] (7/8) Epoch 30, batch 9800, loss[loss=0.1429, simple_loss=0.2372, pruned_loss=0.02428, over 12369.00 frames. ], tot_loss[loss=0.1613, simple_loss=0.2569, pruned_loss=0.03286, over 3057212.22 frames. ], batch size: 248, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:48:05,279 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=304166.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:48:19,122 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([3.5339, 3.5086, 3.4771, 2.7386, 3.3537, 2.1064, 3.1167, 2.8353], device='cuda:7'), covar=tensor([0.0138, 0.0127, 0.0177, 0.0197, 0.0105, 0.2446, 0.0133, 0.0304], device='cuda:7'), in_proj_covar=tensor([0.0181, 0.0172, 0.0210, 0.0179, 0.0187, 0.0215, 0.0198, 0.0175], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:49:24,991 INFO [train.py:904] (7/8) Epoch 30, batch 9850, loss[loss=0.1721, simple_loss=0.2639, pruned_loss=0.04016, over 16152.00 frames. ], tot_loss[loss=0.1618, simple_loss=0.2581, pruned_loss=0.03277, over 3048632.85 frames. ], batch size: 165, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:49:39,352 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.072e+02 2.330e+02 2.794e+02 5.653e+02, threshold=4.660e+02, percent-clipped=2.0 2023-05-03 00:49:57,753 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.9704, 2.8251, 2.7343, 2.0047, 2.5319, 2.8323, 2.7082, 1.9366], device='cuda:7'), covar=tensor([0.0458, 0.0091, 0.0094, 0.0382, 0.0176, 0.0122, 0.0130, 0.0491], device='cuda:7'), in_proj_covar=tensor([0.0137, 0.0089, 0.0091, 0.0134, 0.0102, 0.0113, 0.0097, 0.0129], device='cuda:7'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:7') 2023-05-03 00:50:19,627 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=304230.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:50:46,920 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([1.8679, 1.4177, 1.7939, 1.7468, 1.9251, 1.9419, 1.7363, 1.9065], device='cuda:7'), covar=tensor([0.0333, 0.0528, 0.0281, 0.0373, 0.0355, 0.0271, 0.0519, 0.0185], device='cuda:7'), in_proj_covar=tensor([0.0194, 0.0196, 0.0184, 0.0189, 0.0207, 0.0164, 0.0201, 0.0164], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:51:13,367 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=304252.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:51:16,024 INFO [train.py:904] (7/8) Epoch 30, batch 9900, loss[loss=0.1567, simple_loss=0.2605, pruned_loss=0.02649, over 15491.00 frames. ], tot_loss[loss=0.1617, simple_loss=0.2583, pruned_loss=0.03258, over 3048168.01 frames. ], batch size: 191, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:51:46,157 INFO [scaling.py:679] (7/8) Whitening: num_groups=1, num_channels=384, metric=4.67 vs. limit=5.0 2023-05-03 00:52:13,322 INFO [zipformer.py:625] (7/8) warmup_begin=666.7, warmup_end=1333.3, batch_count=304278.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:53:14,794 INFO [train.py:904] (7/8) Epoch 30, batch 9950, loss[loss=0.1814, simple_loss=0.269, pruned_loss=0.04693, over 12579.00 frames. ], tot_loss[loss=0.1635, simple_loss=0.2609, pruned_loss=0.03305, over 3052216.34 frames. ], batch size: 250, lr: 2.22e-03, grad_scale: 4.0 2023-05-03 00:53:31,713 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.067e+02 2.374e+02 2.801e+02 5.953e+02, threshold=4.748e+02, percent-clipped=1.0 2023-05-03 00:53:50,832 INFO [scaling.py:679] (7/8) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-05-03 00:54:18,589 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.7996, 5.0263, 5.1611, 4.9830, 5.0536, 5.5681, 5.0375, 4.7656], device='cuda:7'), covar=tensor([0.1058, 0.1942, 0.2640, 0.1824, 0.2110, 0.0915, 0.1597, 0.2232], device='cuda:7'), in_proj_covar=tensor([0.0413, 0.0618, 0.0693, 0.0502, 0.0667, 0.0712, 0.0535, 0.0671], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-03 00:54:21,399 INFO [zipformer.py:625] (7/8) warmup_begin=1333.3, warmup_end=2000.0, batch_count=304330.0, num_to_drop=1, layers_to_drop={1} 2023-05-03 00:55:15,643 INFO [train.py:904] (7/8) Epoch 30, batch 10000, loss[loss=0.1743, simple_loss=0.2833, pruned_loss=0.03262, over 17001.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2604, pruned_loss=0.03297, over 3073643.47 frames. ], batch size: 109, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:55:57,831 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([2.4202, 3.3893, 2.7197, 2.1573, 2.1529, 2.3448, 3.4423, 2.9637], device='cuda:7'), covar=tensor([0.3196, 0.0567, 0.1895, 0.3147, 0.3024, 0.2382, 0.0408, 0.1490], device='cuda:7'), in_proj_covar=tensor([0.0331, 0.0269, 0.0309, 0.0323, 0.0297, 0.0276, 0.0297, 0.0344], device='cuda:7'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:7') 2023-05-03 00:56:04,105 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.8916, 5.1144, 5.2854, 5.0516, 5.1374, 5.6446, 5.1630, 4.8724], device='cuda:7'), covar=tensor([0.0929, 0.1857, 0.2149, 0.1754, 0.2054, 0.0848, 0.1501, 0.2301], device='cuda:7'), in_proj_covar=tensor([0.0412, 0.0616, 0.0691, 0.0501, 0.0666, 0.0710, 0.0534, 0.0668], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:7') 2023-05-03 00:56:04,155 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([5.0795, 5.4057, 5.2034, 5.1697, 4.9636, 4.8979, 4.7667, 5.4840], device='cuda:7'), covar=tensor([0.1168, 0.0855, 0.0869, 0.0808, 0.0727, 0.0862, 0.1273, 0.0792], device='cuda:7'), in_proj_covar=tensor([0.0703, 0.0850, 0.0696, 0.0661, 0.0542, 0.0541, 0.0710, 0.0667], device='cuda:7'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:56:57,648 INFO [train.py:904] (7/8) Epoch 30, batch 10050, loss[loss=0.1558, simple_loss=0.2526, pruned_loss=0.02954, over 16468.00 frames. ], tot_loss[loss=0.1632, simple_loss=0.2607, pruned_loss=0.0329, over 3079686.29 frames. ], batch size: 68, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:57:10,559 INFO [optim.py:368] (7/8) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 2.064e+02 2.395e+02 2.910e+02 6.626e+02, threshold=4.790e+02, percent-clipped=3.0 2023-05-03 00:58:34,295 INFO [train.py:904] (7/8) Epoch 30, batch 10100, loss[loss=0.1542, simple_loss=0.2473, pruned_loss=0.03053, over 16669.00 frames. ], tot_loss[loss=0.1633, simple_loss=0.2606, pruned_loss=0.03301, over 3061744.71 frames. ], batch size: 134, lr: 2.22e-03, grad_scale: 8.0 2023-05-03 00:58:57,196 INFO [zipformer.py:625] (7/8) warmup_begin=2000.0, warmup_end=2666.7, batch_count=304466.0, num_to_drop=0, layers_to_drop=set() 2023-05-03 00:59:12,195 INFO [zipformer.py:1454] (7/8) attn_weights_entropy = tensor([4.6728, 4.4936, 4.7500, 4.8717, 5.0242, 4.5346, 5.0627, 5.0470], device='cuda:7'), covar=tensor([0.1993, 0.1235, 0.1571, 0.0801, 0.0539, 0.0944, 0.0507, 0.0730], device='cuda:7'), in_proj_covar=tensor([0.0651, 0.0787, 0.0904, 0.0815, 0.0613, 0.0635, 0.0674, 0.0781], device='cuda:7'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:7') 2023-05-03 00:59:55,813 INFO [train.py:1169] (7/8) Done!