| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 5763, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3e-05, | |
| "loss": 3.21, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7912026726057907e-05, | |
| "loss": 1.4143, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.5824053452115813e-05, | |
| "loss": 1.2072, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "exact_match": 72.9641693811075, | |
| "f1": 84.24962401921071, | |
| "step": 1921 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.3736080178173723e-05, | |
| "loss": 1.1298, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.1648106904231625e-05, | |
| "loss": 0.7843, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.956013363028953e-05, | |
| "loss": 0.8011, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.7472160356347438e-05, | |
| "loss": 0.7699, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "exact_match": 73.0727470141151, | |
| "f1": 85.57152971417061, | |
| "step": 3842 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5384187082405344e-05, | |
| "loss": 0.6582, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.3296213808463252e-05, | |
| "loss": 0.4987, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 1.120824053452116e-05, | |
| "loss": 0.4904, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.120267260579064e-06, | |
| "loss": 0.4922, | |
| "step": 5500 | |
| } | |
| ], | |
| "max_steps": 7684, | |
| "num_train_epochs": 4, | |
| "total_flos": 1.429443287350272e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |