| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.9680624556423, | |
| "global_step": 7000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.64513839602555e-05, | |
| "loss": 3.8375, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.2902767920511e-05, | |
| "loss": 3.6126, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.9354151880766505e-05, | |
| "loss": 3.5033, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.5805535841022005e-05, | |
| "loss": 3.3288, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.22569198012775e-05, | |
| "loss": 3.2926, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.8708303761533e-05, | |
| "loss": 3.2287, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.5159687721788504e-05, | |
| "loss": 3.1489, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.1611071682044005e-05, | |
| "loss": 3.1289, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.8062455642299502e-05, | |
| "loss": 3.0657, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.4513839602555005e-05, | |
| "loss": 3.0318, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 1.0965223562810504e-05, | |
| "loss": 3.0376, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 7.416607523066005e-06, | |
| "loss": 2.9734, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 3.867991483321505e-06, | |
| "loss": 2.9613, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.19375443577005e-07, | |
| "loss": 2.9708, | |
| "step": 7000 | |
| } | |
| ], | |
| "max_steps": 7045, | |
| "num_train_epochs": 5, | |
| "total_flos": 3656520695808000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |