| { | |
| "best_metric": 0.49867676415016876, | |
| "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-2/checkpoint-1070", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 1605, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 4.918482780456543, | |
| "learning_rate": 5.0273219162249924e-05, | |
| "loss": 0.5455, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.4970824718475342, | |
| "eval_matthews_correlation": 0.3827318387959091, | |
| "eval_runtime": 0.7956, | |
| "eval_samples_per_second": 1310.993, | |
| "eval_steps_per_second": 82.958, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "grad_norm": 8.05403995513916, | |
| "learning_rate": 2.7525156192906067e-05, | |
| "loss": 0.3233, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.5374172329902649, | |
| "eval_matthews_correlation": 0.49867676415016876, | |
| "eval_runtime": 1.1803, | |
| "eval_samples_per_second": 883.698, | |
| "eval_steps_per_second": 55.92, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 12.47011947631836, | |
| "learning_rate": 4.77709322356221e-06, | |
| "loss": 0.1756, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.8386874794960022, | |
| "eval_matthews_correlation": 0.46417453765870365, | |
| "eval_runtime": 0.8078, | |
| "eval_samples_per_second": 1291.106, | |
| "eval_steps_per_second": 81.7, | |
| "step": 1605 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1605, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 128518731784344.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 7.302128213159378e-05, | |
| "num_train_epochs": 3, | |
| "per_device_train_batch_size": 16, | |
| "seed": 39 | |
| } | |
| } | |