| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.23873470605789315, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.994026284348866e-05, | |
| "loss": 0.7347, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.98805256869773e-05, | |
| "loss": 0.6862, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.982078853046596e-05, | |
| "loss": 0.6505, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.97610513739546e-05, | |
| "loss": 0.6177, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.970131421744326e-05, | |
| "loss": 0.5992, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.96415770609319e-05, | |
| "loss": 0.5671, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.958183990442056e-05, | |
| "loss": 0.5407, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.952210274790921e-05, | |
| "loss": 0.5186, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.946236559139786e-05, | |
| "loss": 0.5068, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.940262843488651e-05, | |
| "loss": 0.5162, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.934289127837514e-05, | |
| "loss": 0.527, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.928315412186381e-05, | |
| "loss": 0.4849, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.922341696535246e-05, | |
| "loss": 0.477, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.916367980884111e-05, | |
| "loss": 0.4702, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.910394265232975e-05, | |
| "loss": 0.4504, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.90442054958184e-05, | |
| "loss": 0.4422, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.898446833930706e-05, | |
| "loss": 0.4383, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.892473118279571e-05, | |
| "loss": 0.4356, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.886499402628435e-05, | |
| "loss": 0.4446, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.8805256869773e-05, | |
| "loss": 0.4251, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.874551971326166e-05, | |
| "loss": 0.4201, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.868578255675031e-05, | |
| "loss": 0.4389, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.862604540023895e-05, | |
| "loss": 0.4319, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.85663082437276e-05, | |
| "loss": 0.455, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.850657108721625e-05, | |
| "loss": 0.4297, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.844683393070491e-05, | |
| "loss": 0.4383, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.838709677419355e-05, | |
| "loss": 0.4658, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.83273596176822e-05, | |
| "loss": 0.4499, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.826762246117085e-05, | |
| "loss": 0.41, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.820788530465951e-05, | |
| "loss": 0.3866, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.814814814814815e-05, | |
| "loss": 0.3897, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.80884109916368e-05, | |
| "loss": 0.4053, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.802867383512545e-05, | |
| "loss": 0.4022, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.79689366786141e-05, | |
| "loss": 0.4074, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.790919952210275e-05, | |
| "loss": 0.4, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.78494623655914e-05, | |
| "loss": 0.4282, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.778972520908005e-05, | |
| "loss": 0.4028, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.77299880525687e-05, | |
| "loss": 0.3937, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.767025089605735e-05, | |
| "loss": 0.3665, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.7610513739546e-05, | |
| "loss": 0.3743, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_accuracy": 0.7354377926759424, | |
| "eval_accuracy_sklearn": 0.7354377926759424, | |
| "eval_f1": 0.7483577402505849, | |
| "eval_loss": 0.5406955480575562, | |
| "eval_precision": 0.8060886373049841, | |
| "eval_recall": 0.6983434005143233, | |
| "eval_runtime": 162.5908, | |
| "eval_samples_per_second": 182.563, | |
| "eval_steps_per_second": 11.415, | |
| "step": 200 | |
| } | |
| ], | |
| "max_steps": 8370, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.6942572068864e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |