| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.151345291479821, | |
| "eval_steps": 500, | |
| "global_step": 405, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999993108758315e-05, | |
| "loss": 1.5531, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9982776840027333e-05, | |
| "loss": 1.177, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.993116668755721e-05, | |
| "loss": 1.1029, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_loss": 1.0746039152145386, | |
| "eval_runtime": 3.9638, | |
| "eval_samples_per_second": 126.14, | |
| "eval_steps_per_second": 21.192, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.984534732057208e-05, | |
| "loss": 1.036, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.9725614355209207e-05, | |
| "loss": 0.8394, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.957238022747188e-05, | |
| "loss": 0.8271, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "eval_loss": 1.1265954971313477, | |
| "eval_runtime": 3.6594, | |
| "eval_samples_per_second": 136.636, | |
| "eval_steps_per_second": 22.955, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.9386172772539162e-05, | |
| "loss": 0.667, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.916763340656793e-05, | |
| "loss": 0.5326, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.8917514917250276e-05, | |
| "loss": 0.5131, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "eval_loss": 1.3422094583511353, | |
| "eval_runtime": 4.9335, | |
| "eval_samples_per_second": 101.349, | |
| "eval_steps_per_second": 17.027, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "step": 405, | |
| "total_flos": 63559811727360.0, | |
| "train_loss": 0.8341983215308484, | |
| "train_runtime": 1136.3762, | |
| "train_samples_per_second": 56.514, | |
| "train_steps_per_second": 2.355 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 2676, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 63559811727360.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |