| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 860, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 2.9457364341085272e-05, | |
| "loss": 5.0754, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.558139534883721e-05, | |
| "loss": 3.432, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 2.170542635658915e-05, | |
| "loss": 2.6262, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 1.7829457364341087e-05, | |
| "loss": 1.9731, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 1.3953488372093024e-05, | |
| "loss": 1.537, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 1.0077519379844961e-05, | |
| "loss": 1.2258, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 6.201550387596899e-06, | |
| "loss": 0.9553, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 2.325581395348837e-06, | |
| "loss": 0.8734, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 860, | |
| "total_flos": 1570430252839776.0, | |
| "train_loss": 2.1122635065123094, | |
| "train_runtime": 432.5219, | |
| "train_samples_per_second": 23.675, | |
| "train_steps_per_second": 1.988 | |
| } | |
| ], | |
| "max_steps": 860, | |
| "num_train_epochs": 10, | |
| "total_flos": 1570430252839776.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |