| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.943181818181818, | |
| "global_step": 7000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.6448863636363636e-05, | |
| "loss": 1.4544, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.289772727272727e-05, | |
| "loss": 1.0914, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 3.934659090909091e-05, | |
| "loss": 0.9652, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.579545454545455e-05, | |
| "loss": 0.8756, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.2244318181818185e-05, | |
| "loss": 0.7816, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 2.869318181818182e-05, | |
| "loss": 0.7508, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 2.5142045454545453e-05, | |
| "loss": 0.6598, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 2.1590909090909093e-05, | |
| "loss": 0.6049, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 1.8039772727272727e-05, | |
| "loss": 0.5639, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 1.4488636363636366e-05, | |
| "loss": 0.5258, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.09375e-05, | |
| "loss": 0.4905, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 7.386363636363637e-06, | |
| "loss": 0.4758, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 3.835227272727273e-06, | |
| "loss": 0.4578, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 2.840909090909091e-07, | |
| "loss": 0.4302, | |
| "step": 7000 | |
| } | |
| ], | |
| "max_steps": 7040, | |
| "num_train_epochs": 10, | |
| "total_flos": 2593764359423028.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |