| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 5310, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5291902071563094e-05, | |
| "loss": 1.6091, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.058380414312618e-05, | |
| "loss": 1.3109, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.587570621468927e-05, | |
| "loss": 1.2431, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.116760828625235e-05, | |
| "loss": 1.1485, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6459510357815442e-05, | |
| "loss": 1.0738, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.175141242937853e-05, | |
| "loss": 1.0668, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.704331450094162e-05, | |
| "loss": 1.0542, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2335216572504708e-05, | |
| "loss": 0.9543, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.627118644067798e-06, | |
| "loss": 0.9464, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.919020715630885e-06, | |
| "loss": 0.9416, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 5310, | |
| "total_flos": 5.05049054183424e+16, | |
| "train_loss": 1.1233460162319033, | |
| "train_runtime": 7598.915, | |
| "train_samples_per_second": 1.398, | |
| "train_steps_per_second": 0.699 | |
| } | |
| ], | |
| "max_steps": 5310, | |
| "num_train_epochs": 3, | |
| "total_flos": 5.05049054183424e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |