| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.999955993017558, | |
| "global_step": 85210, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.054069295242593e-05, | |
| "loss": 1.6188, | |
| "step": 8521 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.049172706882305e-05, | |
| "loss": 1.5369, | |
| "step": 17042 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.044276118522018e-05, | |
| "loss": 1.5126, | |
| "step": 25563 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.039379530161728e-05, | |
| "loss": 1.496, | |
| "step": 34084 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.0344829418014404e-05, | |
| "loss": 1.4796, | |
| "step": 42605 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.029586353441152e-05, | |
| "loss": 1.4672, | |
| "step": 51126 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.0246897650808646e-05, | |
| "loss": 1.4581, | |
| "step": 59647 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.0197931767205764e-05, | |
| "loss": 1.451, | |
| "step": 68168 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1.0148965883602883e-05, | |
| "loss": 1.4451, | |
| "step": 76689 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 9.999999999999998e-08, | |
| "loss": 1.4408, | |
| "step": 85210 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 85210, | |
| "total_flos": 5.230860312759798e+18, | |
| "train_loss": 1.4906291372968254, | |
| "train_runtime": 109697.9633, | |
| "train_samples_per_second": 24.858, | |
| "train_steps_per_second": 0.777 | |
| } | |
| ], | |
| "max_steps": 85210, | |
| "num_train_epochs": 10, | |
| "total_flos": 5.230860312759798e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |