| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "global_step": 68, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.000109940736055617, | |
| "loss": 4.4243, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.982671888105512e-05, | |
| "loss": 4.0618, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.6324050628611986e-06, | |
| "loss": 4.0138, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.0275104488948473e-05, | |
| "loss": 3.7661, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.227039054081748e-05, | |
| "loss": 3.7248, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011929601172133719, | |
| "loss": 3.9344, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.00013603195463831566, | |
| "loss": 3.8787, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.917765120627052e-05, | |
| "loss": 3.5252, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.8022348793729525e-05, | |
| "loss": 3.5674, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.1680453616843376e-06, | |
| "loss": 3.4848, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.7903988278662747e-05, | |
| "loss": 3.2603, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.492960945918241e-05, | |
| "loss": 3.4291, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 0.0001269248955110515, | |
| "loss": 3.4645, | |
| "step": 65 | |
| } | |
| ], | |
| "max_steps": 68, | |
| "num_train_epochs": 4, | |
| "total_flos": 67413344256000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |