| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.962962962962963, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.506172839506173e-05, | |
| "loss": 1.0616, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.5579138398170471, | |
| "eval_runtime": 1778.2352, | |
| "eval_samples_per_second": 3.14, | |
| "eval_steps_per_second": 0.393, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0123456790123458e-05, | |
| "loss": 0.6432, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.4459874629974365, | |
| "eval_runtime": 1778.2941, | |
| "eval_samples_per_second": 3.14, | |
| "eval_steps_per_second": 0.393, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.185185185185185e-06, | |
| "loss": 0.5407, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.469135802469136e-07, | |
| "loss": 0.499, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 2025, | |
| "num_train_epochs": 3, | |
| "total_flos": 2071956701034000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |