| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 39, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00011961025641025642, | |
| "loss": 5.2002, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00010202051282051282, | |
| "loss": 4.6709, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.443076923076924e-05, | |
| "loss": 4.2521, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.684102564102564e-05, | |
| "loss": 4.258, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.925128205128205e-05, | |
| "loss": 4.0036, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.166153846153846e-05, | |
| "loss": 3.865, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4071794871794871e-05, | |
| "loss": 3.769, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 39, | |
| "total_flos": 20380778496000.0, | |
| "train_loss": 4.229723147856883, | |
| "train_runtime": 16.2558, | |
| "train_samples_per_second": 2.399, | |
| "train_steps_per_second": 2.399 | |
| } | |
| ], | |
| "max_steps": 39, | |
| "num_train_epochs": 1, | |
| "total_flos": 20380778496000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |