| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 5160, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.8e-05, | |
| "loss": 3.0809, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.5633, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.4e-05, | |
| "loss": 1.2472, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.0916, | |
| "step": 2064 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 1e-05, | |
| "loss": 0.999, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.9357, | |
| "step": 3096 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 6e-06, | |
| "loss": 0.8904, | |
| "step": 3612 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.8592, | |
| "step": 4128 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.8373, | |
| "step": 4644 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.8272, | |
| "step": 5160 | |
| } | |
| ], | |
| "max_steps": 5160, | |
| "num_train_epochs": 10, | |
| "total_flos": 322635103469568.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |