| { | |
| "best_metric": 0.166651669293941, | |
| "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-0/checkpoint-3207", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 3207, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.447053301406701e-07, | |
| "loss": 0.6327, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.702122880016472e-07, | |
| "loss": 0.5907, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.5891121625900269, | |
| "eval_matthews_correlation": 0.0, | |
| "eval_runtime": 0.6963, | |
| "eval_samples_per_second": 1497.889, | |
| "eval_steps_per_second": 94.785, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.957192458626242e-07, | |
| "loss": 0.5573, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.2122620372360136e-07, | |
| "loss": 0.5487, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.55595862865448, | |
| "eval_matthews_correlation": 0.12300105566407585, | |
| "eval_runtime": 0.7049, | |
| "eval_samples_per_second": 1479.688, | |
| "eval_steps_per_second": 93.633, | |
| "step": 2138 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.4673316158457843e-07, | |
| "loss": 0.5211, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.22401194455555e-08, | |
| "loss": 0.5205, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.55258709192276, | |
| "eval_matthews_correlation": 0.166651669293941, | |
| "eval_runtime": 1.1749, | |
| "eval_samples_per_second": 887.771, | |
| "eval_steps_per_second": 56.177, | |
| "step": 3207 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 3207, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 113236081840572.0, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 1.119198372279693e-06, | |
| "num_train_epochs": 3, | |
| "per_device_train_batch_size": 8, | |
| "seed": 19 | |
| } | |
| } | |