| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.0, | |
| "global_step": 2475, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3e-05, | |
| "loss": 1.3813, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 52.7923516381019, | |
| "eval_f1": 63.221978497760105, | |
| "eval_runtime": 8.4582, | |
| "eval_samples_per_second": 1014.045, | |
| "eval_steps_per_second": 3.192, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.7629, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 54.54121487699662, | |
| "eval_f1": 64.07410986090608, | |
| "eval_runtime": 7.5065, | |
| "eval_samples_per_second": 1142.604, | |
| "eval_steps_per_second": 3.597, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.57, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_exact_match": 53.806692316660836, | |
| "eval_f1": 63.7117853356075, | |
| "eval_runtime": 7.5, | |
| "eval_samples_per_second": 1143.598, | |
| "eval_steps_per_second": 3.6, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4292, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_exact_match": 52.8040107263612, | |
| "eval_f1": 62.67781767783121, | |
| "eval_runtime": 7.5005, | |
| "eval_samples_per_second": 1143.53, | |
| "eval_steps_per_second": 3.6, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3211, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_exact_match": 55.08919202518363, | |
| "eval_f1": 64.1535471677265, | |
| "eval_runtime": 7.4798, | |
| "eval_samples_per_second": 1146.69, | |
| "eval_steps_per_second": 3.61, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.2406, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_exact_match": 55.065873848665035, | |
| "eval_f1": 63.05978576562645, | |
| "eval_runtime": 7.4813, | |
| "eval_samples_per_second": 1146.454, | |
| "eval_steps_per_second": 3.609, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1761, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_exact_match": 53.14212428588085, | |
| "eval_f1": 62.77110644240899, | |
| "eval_runtime": 7.5138, | |
| "eval_samples_per_second": 1141.504, | |
| "eval_steps_per_second": 3.593, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1357, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_exact_match": 55.858691850297305, | |
| "eval_f1": 63.800808981047425, | |
| "eval_runtime": 7.4789, | |
| "eval_samples_per_second": 1146.834, | |
| "eval_steps_per_second": 3.61, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1075, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_exact_match": 57.910691383933774, | |
| "eval_f1": 65.09187195276797, | |
| "eval_runtime": 7.5025, | |
| "eval_samples_per_second": 1143.219, | |
| "eval_steps_per_second": 3.599, | |
| "step": 2475 | |
| } | |
| ], | |
| "max_steps": 2750, | |
| "num_train_epochs": 10, | |
| "total_flos": 784827225735168.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |