| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 8.856682769726248, | |
| "global_step": 5500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.597423510466989e-05, | |
| "loss": 0.3372, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.194847020933977e-05, | |
| "loss": 0.2233, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.7922705314009665e-05, | |
| "loss": 0.1856, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.389694041867955e-05, | |
| "loss": 0.1587, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 2.9871175523349438e-05, | |
| "loss": 0.1332, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.5845410628019323e-05, | |
| "loss": 0.1059, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 2.1819645732689212e-05, | |
| "loss": 0.0869, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 1.77938808373591e-05, | |
| "loss": 0.0743, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 1.3768115942028985e-05, | |
| "loss": 0.0633, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 9.742351046698874e-06, | |
| "loss": 0.0537, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 5.71658615136876e-06, | |
| "loss": 0.0457, | |
| "step": 5500 | |
| } | |
| ], | |
| "max_steps": 6210, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.4222892248394192e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |