| { | |
| "best_metric": 0.8705702821418305, | |
| "best_model_checkpoint": "./phobert_results_v2/checkpoint-3470", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 3470, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02881844380403458, | |
| "grad_norm": 4.723288059234619, | |
| "learning_rate": 1.994236311239193e-05, | |
| "loss": 1.9047, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05763688760806916, | |
| "grad_norm": 32.87104034423828, | |
| "learning_rate": 1.9884726224783863e-05, | |
| "loss": 1.7269, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08645533141210375, | |
| "grad_norm": 5.206538677215576, | |
| "learning_rate": 1.9827089337175795e-05, | |
| "loss": 1.6865, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11527377521613832, | |
| "grad_norm": 52.5133056640625, | |
| "learning_rate": 1.9769452449567724e-05, | |
| "loss": 1.633, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1440922190201729, | |
| "grad_norm": 12.493169784545898, | |
| "learning_rate": 1.9711815561959656e-05, | |
| "loss": 1.553, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1729106628242075, | |
| "grad_norm": 7.675703048706055, | |
| "learning_rate": 1.9654178674351588e-05, | |
| "loss": 1.4131, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2017291066282421, | |
| "grad_norm": 5.984454154968262, | |
| "learning_rate": 1.9596541786743517e-05, | |
| "loss": 1.3606, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.23054755043227665, | |
| "grad_norm": 9.403727531433105, | |
| "learning_rate": 1.953890489913545e-05, | |
| "loss": 1.2202, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.25936599423631124, | |
| "grad_norm": 7.890481472015381, | |
| "learning_rate": 1.9481268011527378e-05, | |
| "loss": 1.186, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2881844380403458, | |
| "grad_norm": 12.102771759033203, | |
| "learning_rate": 1.942363112391931e-05, | |
| "loss": 1.0865, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3170028818443804, | |
| "grad_norm": 17.669885635375977, | |
| "learning_rate": 1.936599423631124e-05, | |
| "loss": 1.124, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.345821325648415, | |
| "grad_norm": 10.14146614074707, | |
| "learning_rate": 1.930835734870317e-05, | |
| "loss": 1.1307, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3746397694524496, | |
| "grad_norm": 9.363780975341797, | |
| "learning_rate": 1.9250720461095104e-05, | |
| "loss": 1.0582, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4034582132564842, | |
| "grad_norm": 22.122907638549805, | |
| "learning_rate": 1.9193083573487033e-05, | |
| "loss": 1.007, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4322766570605187, | |
| "grad_norm": 22.921249389648438, | |
| "learning_rate": 1.9135446685878965e-05, | |
| "loss": 1.0046, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.4610951008645533, | |
| "grad_norm": 13.880660057067871, | |
| "learning_rate": 1.9077809798270894e-05, | |
| "loss": 0.9786, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4899135446685879, | |
| "grad_norm": 13.75207805633545, | |
| "learning_rate": 1.9020172910662826e-05, | |
| "loss": 0.9166, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5187319884726225, | |
| "grad_norm": 15.701948165893555, | |
| "learning_rate": 1.8962536023054755e-05, | |
| "loss": 0.8942, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.547550432276657, | |
| "grad_norm": 13.106974601745605, | |
| "learning_rate": 1.8904899135446687e-05, | |
| "loss": 0.8905, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5763688760806917, | |
| "grad_norm": 13.854475021362305, | |
| "learning_rate": 1.884726224783862e-05, | |
| "loss": 0.9089, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6051873198847262, | |
| "grad_norm": 14.22427749633789, | |
| "learning_rate": 1.878962536023055e-05, | |
| "loss": 0.8688, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6340057636887608, | |
| "grad_norm": 14.088136672973633, | |
| "learning_rate": 1.873198847262248e-05, | |
| "loss": 0.859, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6628242074927954, | |
| "grad_norm": 15.746428489685059, | |
| "learning_rate": 1.867435158501441e-05, | |
| "loss": 0.8356, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.69164265129683, | |
| "grad_norm": 10.937832832336426, | |
| "learning_rate": 1.861671469740634e-05, | |
| "loss": 0.8699, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7204610951008645, | |
| "grad_norm": 14.396600723266602, | |
| "learning_rate": 1.855907780979827e-05, | |
| "loss": 0.7789, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7492795389048992, | |
| "grad_norm": 11.801300048828125, | |
| "learning_rate": 1.8501440922190203e-05, | |
| "loss": 0.8713, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7780979827089337, | |
| "grad_norm": 10.435481071472168, | |
| "learning_rate": 1.8443804034582135e-05, | |
| "loss": 0.7961, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8069164265129684, | |
| "grad_norm": 11.2293062210083, | |
| "learning_rate": 1.8386167146974067e-05, | |
| "loss": 0.8397, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.8357348703170029, | |
| "grad_norm": 12.809613227844238, | |
| "learning_rate": 1.8328530259365996e-05, | |
| "loss": 0.8297, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.8645533141210374, | |
| "grad_norm": 14.791847229003906, | |
| "learning_rate": 1.8270893371757928e-05, | |
| "loss": 0.7709, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8933717579250721, | |
| "grad_norm": 14.587478637695312, | |
| "learning_rate": 1.8213256484149857e-05, | |
| "loss": 0.7552, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.9221902017291066, | |
| "grad_norm": 16.79636001586914, | |
| "learning_rate": 1.815561959654179e-05, | |
| "loss": 0.7937, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.9510086455331412, | |
| "grad_norm": 8.510680198669434, | |
| "learning_rate": 1.8097982708933718e-05, | |
| "loss": 0.7181, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.9798270893371758, | |
| "grad_norm": 20.466522216796875, | |
| "learning_rate": 1.804034582132565e-05, | |
| "loss": 0.712, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7955036748811067, | |
| "eval_f1_macro": 0.795530535937033, | |
| "eval_f1_weighted": 0.795536244762841, | |
| "eval_loss": 0.6264312267303467, | |
| "eval_precision_macro": 0.7978076489526655, | |
| "eval_precision_weighted": 0.797819457255313, | |
| "eval_recall_macro": 0.7955041514180622, | |
| "eval_recall_weighted": 0.7955036748811067, | |
| "eval_runtime": 29.7702, | |
| "eval_samples_per_second": 233.086, | |
| "eval_steps_per_second": 14.578, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.0086455331412103, | |
| "grad_norm": 17.28911018371582, | |
| "learning_rate": 1.7982708933717582e-05, | |
| "loss": 0.6512, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.037463976945245, | |
| "grad_norm": 28.79306411743164, | |
| "learning_rate": 1.792507204610951e-05, | |
| "loss": 0.6036, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.0662824207492796, | |
| "grad_norm": 13.444112777709961, | |
| "learning_rate": 1.7867435158501444e-05, | |
| "loss": 0.552, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.0951008645533142, | |
| "grad_norm": 22.068620681762695, | |
| "learning_rate": 1.7809798270893372e-05, | |
| "loss": 0.577, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.1239193083573487, | |
| "grad_norm": 13.942420959472656, | |
| "learning_rate": 1.7752161383285305e-05, | |
| "loss": 0.5573, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.1527377521613833, | |
| "grad_norm": 13.652227401733398, | |
| "learning_rate": 1.7694524495677234e-05, | |
| "loss": 0.5401, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.181556195965418, | |
| "grad_norm": 22.252243041992188, | |
| "learning_rate": 1.7636887608069166e-05, | |
| "loss": 0.558, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.2103746397694524, | |
| "grad_norm": 17.13372802734375, | |
| "learning_rate": 1.7579250720461095e-05, | |
| "loss": 0.514, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.239193083573487, | |
| "grad_norm": 17.164276123046875, | |
| "learning_rate": 1.7521613832853027e-05, | |
| "loss": 0.5137, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.2680115273775217, | |
| "grad_norm": 14.449616432189941, | |
| "learning_rate": 1.746397694524496e-05, | |
| "loss": 0.5, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.2968299711815563, | |
| "grad_norm": 11.61601734161377, | |
| "learning_rate": 1.7406340057636888e-05, | |
| "loss": 0.5071, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.3256484149855907, | |
| "grad_norm": 20.6407527923584, | |
| "learning_rate": 1.734870317002882e-05, | |
| "loss": 0.4803, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.3544668587896254, | |
| "grad_norm": 18.60149383544922, | |
| "learning_rate": 1.729106628242075e-05, | |
| "loss": 0.4967, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.38328530259366, | |
| "grad_norm": 22.12944793701172, | |
| "learning_rate": 1.723342939481268e-05, | |
| "loss": 0.4652, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.4121037463976944, | |
| "grad_norm": 18.460689544677734, | |
| "learning_rate": 1.717579250720461e-05, | |
| "loss": 0.4417, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.440922190201729, | |
| "grad_norm": 19.452363967895508, | |
| "learning_rate": 1.7118155619596542e-05, | |
| "loss": 0.5082, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4697406340057637, | |
| "grad_norm": 41.46109390258789, | |
| "learning_rate": 1.7060518731988475e-05, | |
| "loss": 0.463, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.4985590778097984, | |
| "grad_norm": 25.584379196166992, | |
| "learning_rate": 1.7002881844380407e-05, | |
| "loss": 0.4788, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.527377521613833, | |
| "grad_norm": 16.92909049987793, | |
| "learning_rate": 1.6945244956772336e-05, | |
| "loss": 0.4752, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.5561959654178674, | |
| "grad_norm": 10.342531204223633, | |
| "learning_rate": 1.6887608069164268e-05, | |
| "loss": 0.4848, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.585014409221902, | |
| "grad_norm": 15.596243858337402, | |
| "learning_rate": 1.6829971181556197e-05, | |
| "loss": 0.5158, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.6138328530259365, | |
| "grad_norm": 17.031354904174805, | |
| "learning_rate": 1.6772334293948126e-05, | |
| "loss": 0.4577, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.6426512968299711, | |
| "grad_norm": 13.550045013427734, | |
| "learning_rate": 1.6714697406340058e-05, | |
| "loss": 0.4947, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.6714697406340058, | |
| "grad_norm": 10.102880477905273, | |
| "learning_rate": 1.665706051873199e-05, | |
| "loss": 0.4681, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.7002881844380404, | |
| "grad_norm": 10.968811988830566, | |
| "learning_rate": 1.6599423631123922e-05, | |
| "loss": 0.452, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.729106628242075, | |
| "grad_norm": 4.670314311981201, | |
| "learning_rate": 1.654178674351585e-05, | |
| "loss": 0.4346, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7579250720461095, | |
| "grad_norm": 28.008899688720703, | |
| "learning_rate": 1.6484149855907783e-05, | |
| "loss": 0.4404, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.7867435158501441, | |
| "grad_norm": 27.77347183227539, | |
| "learning_rate": 1.6426512968299712e-05, | |
| "loss": 0.4547, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.8155619596541785, | |
| "grad_norm": 21.949289321899414, | |
| "learning_rate": 1.6368876080691644e-05, | |
| "loss": 0.4924, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.8443804034582132, | |
| "grad_norm": 2.8766281604766846, | |
| "learning_rate": 1.6311239193083573e-05, | |
| "loss": 0.3776, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.8731988472622478, | |
| "grad_norm": 38.812625885009766, | |
| "learning_rate": 1.6253602305475506e-05, | |
| "loss": 0.4176, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.9020172910662825, | |
| "grad_norm": 24.132482528686523, | |
| "learning_rate": 1.6195965417867438e-05, | |
| "loss": 0.4407, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.9308357348703171, | |
| "grad_norm": 19.72426414489746, | |
| "learning_rate": 1.613832853025937e-05, | |
| "loss": 0.4503, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.9596541786743515, | |
| "grad_norm": 16.077899932861328, | |
| "learning_rate": 1.60806916426513e-05, | |
| "loss": 0.4244, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.9884726224783862, | |
| "grad_norm": 21.258529663085938, | |
| "learning_rate": 1.6023054755043228e-05, | |
| "loss": 0.4161, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.87101887880098, | |
| "eval_f1_macro": 0.8705600204306078, | |
| "eval_f1_weighted": 0.8705702821418305, | |
| "eval_loss": 0.43536442518234253, | |
| "eval_precision_macro": 0.8749786652431168, | |
| "eval_precision_weighted": 0.8749977247467932, | |
| "eval_recall_macro": 0.8710178763247105, | |
| "eval_recall_weighted": 0.87101887880098, | |
| "eval_runtime": 29.9836, | |
| "eval_samples_per_second": 231.426, | |
| "eval_steps_per_second": 14.475, | |
| "step": 3470 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 17350, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 7302448933401600.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |