{ "batch_size": 16, "accum_iter": 1, "epochs": 1, "warmup_epochs": 0.01, "lr": 4e-05, "min_lr": 2e-06, "wd": 0.01, "clip_grad": 4.0, "init_from": null, "data_config": "configs/data/448exp.yaml", "cache_ann_on_disk": true, "length_clustering": true, "num_workers": 8, "pin_mem": true, "seed": 0, "output_dir": "output/7B_all_lr4e-5_min_lr2e-6", "save_interval": 1, "save_iteration_interval": 2000, "only_save_trainable": false, "ckpt_max_keep": 2, "auto_resume": true, "resume_path": null, "model_parallel_size": 1, "data_parallel": "fsdp", "precision": "bf16", "grad_precision": "fp32", "checkpointing": true, "max_seq_len": 2048, "mask_image_logits": false, "dropout": 0.01, "z_loss_weight": 1e-05, "model_size": "7B", "world_size": 32, "rank": 0, "gpu": 0, "local_rank": 0, "dist_url": "env://", "distributed": true, "dist_backend": "nccl" }