| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 7448370, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.5e-05, | |
| "loss": 5.1426, | |
| "step": 744837 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4e-05, | |
| "loss": 4.3897, | |
| "step": 1489674 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.5e-05, | |
| "loss": 4.271, | |
| "step": 2234511 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3e-05, | |
| "loss": 4.2161, | |
| "step": 2979348 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.5e-05, | |
| "loss": 4.174, | |
| "step": 3724185 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2e-05, | |
| "loss": 4.1243, | |
| "step": 4469022 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 1.5e-05, | |
| "loss": 4.0682, | |
| "step": 5213859 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 1e-05, | |
| "loss": 4.0117, | |
| "step": 5958696 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5e-06, | |
| "loss": 3.9571, | |
| "step": 6703533 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.0, | |
| "loss": 3.9111, | |
| "step": 7448370 | |
| } | |
| ], | |
| "max_steps": 7448370, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.4787097128956529e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |