| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 45460, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.600087989441267e-05, | |
| "loss": 1.1857, | |
| "step": 9092 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.1187185049057007, | |
| "eval_runtime": 193.7266, | |
| "eval_samples_per_second": 41.522, | |
| "eval_steps_per_second": 10.381, | |
| "step": 9092 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.2003079630444347e-05, | |
| "loss": 1.0728, | |
| "step": 18184 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.0675935745239258, | |
| "eval_runtime": 193.8852, | |
| "eval_samples_per_second": 41.488, | |
| "eval_steps_per_second": 10.372, | |
| "step": 18184 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.003519577650682e-06, | |
| "loss": 1.0286, | |
| "step": 27276 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.045861005783081, | |
| "eval_runtime": 193.844, | |
| "eval_samples_per_second": 41.497, | |
| "eval_steps_per_second": 10.374, | |
| "step": 27276 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.004839419269689e-06, | |
| "loss": 1.0046, | |
| "step": 36368 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.0346417427062988, | |
| "eval_runtime": 191.4491, | |
| "eval_samples_per_second": 42.016, | |
| "eval_steps_per_second": 10.504, | |
| "step": 36368 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 7.039155301363837e-09, | |
| "loss": 0.9914, | |
| "step": 45460 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 1.0310282707214355, | |
| "eval_runtime": 191.6107, | |
| "eval_samples_per_second": 41.981, | |
| "eval_steps_per_second": 10.495, | |
| "step": 45460 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 45460, | |
| "total_flos": 1.8998637018218496e+17, | |
| "train_loss": 1.0566342449062363, | |
| "train_runtime": 22320.8822, | |
| "train_samples_per_second": 16.293, | |
| "train_steps_per_second": 2.037 | |
| } | |
| ], | |
| "max_steps": 45460, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.8998637018218496e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |