| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9991417946028416, | |
| "eval_steps": 500, | |
| "global_step": 7863, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0003, | |
| "loss": 0.4456, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00028467799601788834, | |
| "loss": 0.3755, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00024162486124017595, | |
| "loss": 0.3375, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00017970536279917184, | |
| "loss": 0.3148, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00011166926925795655, | |
| "loss": 0.294, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 5.152580983231701e-05, | |
| "loss": 0.2481, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1.1659051053259278e-05, | |
| "loss": 0.2294, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 7863, | |
| "total_flos": 5.78415163140735e+17, | |
| "train_loss": 0.31008846705520515, | |
| "train_runtime": 47529.999, | |
| "train_samples_per_second": 0.662, | |
| "train_steps_per_second": 0.165 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 7863, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 5.78415163140735e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |