| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.82905982905983, | |
| "global_step": 4600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.95260663507109e-05, | |
| "loss": 0.2546, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 4.3601895734597157e-05, | |
| "loss": 0.0295, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.767772511848342e-05, | |
| "loss": 0.017, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.175355450236967e-05, | |
| "loss": 0.0107, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 2.5829383886255927e-05, | |
| "loss": 0.007, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 1.990521327014218e-05, | |
| "loss": 0.005, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 1.3981042654028437e-05, | |
| "loss": 0.0033, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 8.056872037914693e-06, | |
| "loss": 0.0026, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 2.1327014218009483e-06, | |
| "loss": 0.0018, | |
| "step": 4500 | |
| } | |
| ], | |
| "max_steps": 4680, | |
| "num_train_epochs": 10, | |
| "total_flos": 6017085844864512.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |