| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.35475234270415, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.999665327978581e-05, | |
| "loss": 1.1175, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.9993306559571626e-05, | |
| "loss": 0.7692, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.998995983935743e-05, | |
| "loss": 0.6592, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.9986613119143243e-05, | |
| "loss": 0.6046, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 4.998326639892905e-05, | |
| "loss": 0.5642, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.997991967871486e-05, | |
| "loss": 0.5396, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 4.9976572958500676e-05, | |
| "loss": 0.5121, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 4.997322623828648e-05, | |
| "loss": 0.4977, | |
| "step": 4000 | |
| } | |
| ], | |
| "max_steps": 7470000, | |
| "num_train_epochs": 10000, | |
| "total_flos": 1857418627645440.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |