| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.724473257698541, | |
| "global_step": 6000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.594813614262561e-05, | |
| "loss": 0.2452, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.189627228525122e-05, | |
| "loss": 0.1664, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.7844408427876825e-05, | |
| "loss": 0.1397, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.379254457050243e-05, | |
| "loss": 0.1179, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 2.974068071312804e-05, | |
| "loss": 0.0962, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 2.568881685575365e-05, | |
| "loss": 0.0743, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 2.1636952998379258e-05, | |
| "loss": 0.0589, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 1.7585089141004862e-05, | |
| "loss": 0.0486, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 1.3533225283630471e-05, | |
| "loss": 0.0398, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 9.481361426256078e-06, | |
| "loss": 0.0327, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 5.429497568881686e-06, | |
| "loss": 0.0268, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 1.3776337115072935e-06, | |
| "loss": 0.0236, | |
| "step": 6000 | |
| } | |
| ], | |
| "max_steps": 6170, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.3626404244078336e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |