| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 696, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00013142857142857143, | |
| "loss": 1.5363, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00019297124600638977, | |
| "loss": 1.1994, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0001782747603833866, | |
| "loss": 1.0903, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0001635782747603834, | |
| "loss": 0.9799, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0001488817891373802, | |
| "loss": 0.9738, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.000134185303514377, | |
| "loss": 0.8681, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0001194888178913738, | |
| "loss": 0.8991, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00010479233226837062, | |
| "loss": 0.9033, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.009584664536742e-05, | |
| "loss": 0.9134, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 7.539936102236423e-05, | |
| "loss": 0.8864, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 6.070287539936103e-05, | |
| "loss": 0.7979, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.600638977635783e-05, | |
| "loss": 0.8006, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.130990415335464e-05, | |
| "loss": 0.7841, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.6613418530351437e-05, | |
| "loss": 0.8221, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.9169329073482426e-06, | |
| "loss": 0.803, | |
| "step": 690 | |
| } | |
| ], | |
| "logging_steps": 46, | |
| "max_steps": 696, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 5.645876465133158e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |