| { |
| "best_metric": 0.41804637619742235, |
| "best_model_checkpoint": "andyWuTw/homework001-finetuned-cola/run-4/checkpoint-536", |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 804, |
| "is_hyper_param_search": true, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 1.0, |
| "eval_loss": 0.5137542486190796, |
| "eval_matthews_correlation": 0.409237984741568, |
| "eval_runtime": 0.7271, |
| "eval_samples_per_second": 1434.443, |
| "eval_steps_per_second": 90.77, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 2.6489687533496514e-06, |
| "loss": 0.5131, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_loss": 0.517956554889679, |
| "eval_matthews_correlation": 0.41804637619742235, |
| "eval_runtime": 0.8254, |
| "eval_samples_per_second": 1263.604, |
| "eval_steps_per_second": 79.96, |
| "step": 536 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_loss": 0.5104890465736389, |
| "eval_matthews_correlation": 0.4177399098153259, |
| "eval_runtime": 0.7699, |
| "eval_samples_per_second": 1354.644, |
| "eval_steps_per_second": 85.721, |
| "step": 804 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 804, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "total_flos": 95368247819904.0, |
| "train_batch_size": 32, |
| "trial_name": null, |
| "trial_params": { |
| "learning_rate": 7.005825255569473e-06, |
| "num_train_epochs": 3, |
| "per_device_train_batch_size": 32, |
| "seed": 40 |
| } |
| } |
|
|
