| { | |
| "best_metric": 0.4587719298245614, | |
| "best_model_checkpoint": "model/mert_cmp_single/checkpoint-835", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 835, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.0428173542022705, | |
| "learning_rate": 0.004, | |
| "loss": 1.9863, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.34210526315789475, | |
| "eval_f1_macro": 0.11514252156075641, | |
| "eval_f1_micro": 0.34210526315789475, | |
| "eval_f1_weighted": 0.23796259505732095, | |
| "eval_loss": 1.8524572849273682, | |
| "eval_precision_macro": 0.11245959365658408, | |
| "eval_precision_micro": 0.34210526315789475, | |
| "eval_precision_weighted": 0.20273893408852728, | |
| "eval_recall_macro": 0.1590998496911623, | |
| "eval_recall_micro": 0.34210526315789475, | |
| "eval_recall_weighted": 0.34210526315789475, | |
| "eval_runtime": 384.7963, | |
| "eval_samples_per_second": 2.963, | |
| "eval_steps_per_second": 0.187, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 3.109973430633545, | |
| "learning_rate": 0.003, | |
| "loss": 1.825, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.3394736842105263, | |
| "eval_f1_macro": 0.13922160871783656, | |
| "eval_f1_micro": 0.3394736842105263, | |
| "eval_f1_weighted": 0.26941182225189036, | |
| "eval_loss": 1.8022291660308838, | |
| "eval_precision_macro": 0.11904513486359071, | |
| "eval_precision_micro": 0.3394736842105263, | |
| "eval_precision_weighted": 0.24822549940815403, | |
| "eval_recall_macro": 0.2242063492063492, | |
| "eval_recall_micro": 0.3394736842105263, | |
| "eval_recall_weighted": 0.3394736842105263, | |
| "eval_runtime": 388.714, | |
| "eval_samples_per_second": 2.933, | |
| "eval_steps_per_second": 0.185, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 3.6947247982025146, | |
| "learning_rate": 0.002, | |
| "loss": 1.7777, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.387719298245614, | |
| "eval_f1_macro": 0.18116391167343948, | |
| "eval_f1_micro": 0.387719298245614, | |
| "eval_f1_weighted": 0.32465915019441605, | |
| "eval_loss": 1.6795759201049805, | |
| "eval_precision_macro": 0.1841779545892386, | |
| "eval_precision_micro": 0.387719298245614, | |
| "eval_precision_weighted": 0.312286629527785, | |
| "eval_recall_macro": 0.2353197126803938, | |
| "eval_recall_micro": 0.387719298245614, | |
| "eval_recall_weighted": 0.387719298245614, | |
| "eval_runtime": 388.0064, | |
| "eval_samples_per_second": 2.938, | |
| "eval_steps_per_second": 0.186, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 12.209613800048828, | |
| "learning_rate": 0.001, | |
| "loss": 1.7034, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.4192982456140351, | |
| "eval_f1_macro": 0.2541449877226969, | |
| "eval_f1_micro": 0.4192982456140351, | |
| "eval_f1_weighted": 0.3766885793183303, | |
| "eval_loss": 1.5931727886199951, | |
| "eval_precision_macro": 0.2826136618574851, | |
| "eval_precision_micro": 0.4192982456140351, | |
| "eval_precision_weighted": 0.3787058942522796, | |
| "eval_recall_macro": 0.2971802170384961, | |
| "eval_recall_micro": 0.4192982456140351, | |
| "eval_recall_weighted": 0.4192982456140351, | |
| "eval_runtime": 387.3989, | |
| "eval_samples_per_second": 2.943, | |
| "eval_steps_per_second": 0.186, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 3.8179590702056885, | |
| "learning_rate": 0.0, | |
| "loss": 1.5985, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.4587719298245614, | |
| "eval_f1_macro": 0.2877411293172874, | |
| "eval_f1_micro": 0.4587719298245614, | |
| "eval_f1_weighted": 0.4111312997459781, | |
| "eval_loss": 1.5373002290725708, | |
| "eval_precision_macro": 0.29145524831736175, | |
| "eval_precision_micro": 0.4587719298245614, | |
| "eval_precision_weighted": 0.38923064340369684, | |
| "eval_recall_macro": 0.3265703379867467, | |
| "eval_recall_micro": 0.4587719298245614, | |
| "eval_recall_weighted": 0.4587719298245614, | |
| "eval_runtime": 387.2757, | |
| "eval_samples_per_second": 2.944, | |
| "eval_steps_per_second": 0.186, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 835, | |
| "total_flos": 0.0, | |
| "train_loss": 1.7781848290723241, | |
| "train_runtime": 8256.8443, | |
| "train_samples_per_second": 1.611, | |
| "train_steps_per_second": 0.101 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 835, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |