| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.992, | |
| "eval_steps": 500, | |
| "global_step": 62, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.920192797165511e-05, | |
| "loss": 0.2628, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.685866540361456e-05, | |
| "loss": 0.2224, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.3119819680728e-05, | |
| "loss": 0.3017, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.822410025817406e-05, | |
| "loss": 0.2817, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.2484078074333954e-05, | |
| "loss": 0.2966, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.6266229220967818e-05, | |
| "loss": 0.261, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9967536997783494e-05, | |
| "loss": 0.2825, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.399014621105914e-05, | |
| "loss": 0.2833, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.715687931944449e-06, | |
| "loss": 0.2848, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.480913969818098e-06, | |
| "loss": 0.3022, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.5561966963229924e-06, | |
| "loss": 0.2568, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.2826691520262114e-07, | |
| "loss": 0.2653, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "step": 62, | |
| "total_flos": 2762117699272704.0, | |
| "train_loss": 0.27888556161234457, | |
| "train_runtime": 1689.5979, | |
| "train_samples_per_second": 0.296, | |
| "train_steps_per_second": 0.037 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 62, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 2762117699272704.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |