| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 6783, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.631431519976412e-05, | |
| "loss": 4.4714, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.262863039952824e-05, | |
| "loss": 4.0522, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.894294559929235e-05, | |
| "loss": 3.9579, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.525726079905647e-05, | |
| "loss": 3.8858, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.157157599882059e-05, | |
| "loss": 3.8457, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7885891198584695e-05, | |
| "loss": 3.8249, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.4200206398348816e-05, | |
| "loss": 3.7971, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0514521598112928e-05, | |
| "loss": 3.7473, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6828836797877046e-05, | |
| "loss": 3.7236, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3143151997641162e-05, | |
| "loss": 3.7065, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.457467197405279e-06, | |
| "loss": 3.6722, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.771782397169395e-06, | |
| "loss": 3.6964, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.0860975969335103e-06, | |
| "loss": 3.6654, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 6783, | |
| "total_flos": 1.1525089201938432e+16, | |
| "train_loss": 3.843147566900938, | |
| "train_runtime": 1064.9117, | |
| "train_samples_per_second": 38.212, | |
| "train_steps_per_second": 6.37 | |
| } | |
| ], | |
| "max_steps": 6783, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.1525089201938432e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |