| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9983987189751802, | |
| "global_step": 624, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.996794871794872e-05, | |
| "loss": 2.5842, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.967948717948718e-05, | |
| "loss": 2.511, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.935897435897436e-05, | |
| "loss": 2.4961, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.903846153846154e-05, | |
| "loss": 2.4203, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.8717948717948718e-05, | |
| "loss": 2.3722, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8397435897435897e-05, | |
| "loss": 2.4072, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.807692307692308e-05, | |
| "loss": 2.4484, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.775641025641026e-05, | |
| "loss": 2.3817, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.7435897435897438e-05, | |
| "loss": 2.4138, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.7115384615384617e-05, | |
| "loss": 2.3693, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.6794871794871796e-05, | |
| "loss": 2.3947, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.6474358974358975e-05, | |
| "loss": 2.3659, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.6153846153846154e-05, | |
| "loss": 2.3695, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.5833333333333333e-05, | |
| "loss": 2.343, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.5512820512820516e-05, | |
| "loss": 2.3434, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5192307692307693e-05, | |
| "loss": 2.3407, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4871794871794874e-05, | |
| "loss": 2.3595, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.4551282051282051e-05, | |
| "loss": 2.3485, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.4230769230769232e-05, | |
| "loss": 2.3488, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.3910256410256411e-05, | |
| "loss": 2.4007, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.3589743589743592e-05, | |
| "loss": 2.3197, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.3269230769230769e-05, | |
| "loss": 2.324, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.294871794871795e-05, | |
| "loss": 2.3546, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.2628205128205129e-05, | |
| "loss": 2.3547, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.230769230769231e-05, | |
| "loss": 2.3702, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.1987179487179487e-05, | |
| "loss": 2.3071, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.1666666666666668e-05, | |
| "loss": 2.3233, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1346153846153847e-05, | |
| "loss": 2.3361, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1025641025641028e-05, | |
| "loss": 2.3152, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.0705128205128205e-05, | |
| "loss": 2.3653, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.0384615384615386e-05, | |
| "loss": 2.3031, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0064102564102565e-05, | |
| "loss": 2.3035, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.743589743589744e-06, | |
| "loss": 1.8449, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.423076923076923e-06, | |
| "loss": 1.7192, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.102564102564104e-06, | |
| "loss": 1.7361, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.782051282051283e-06, | |
| "loss": 1.7584, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.461538461538462e-06, | |
| "loss": 1.7269, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 8.141025641025641e-06, | |
| "loss": 1.7208, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.820512820512822e-06, | |
| "loss": 1.7485, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 1.7506, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.17948717948718e-06, | |
| "loss": 1.7193, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.858974358974359e-06, | |
| "loss": 1.6919, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.538461538461539e-06, | |
| "loss": 1.7516, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.217948717948718e-06, | |
| "loss": 1.7129, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.897435897435898e-06, | |
| "loss": 1.7253, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.576923076923077e-06, | |
| "loss": 1.7259, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.256410256410257e-06, | |
| "loss": 1.7065, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.935897435897436e-06, | |
| "loss": 1.709, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 1.6774, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.294871794871795e-06, | |
| "loss": 1.7154, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.974358974358974e-06, | |
| "loss": 1.6767, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.653846153846154e-06, | |
| "loss": 1.7006, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.6998, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 3.012820512820513e-06, | |
| "loss": 1.7088, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.6923076923076923e-06, | |
| "loss": 1.6739, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.371794871794872e-06, | |
| "loss": 1.6906, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0512820512820513e-06, | |
| "loss": 1.7699, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.7307692307692308e-06, | |
| "loss": 1.6989, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.4102564102564104e-06, | |
| "loss": 1.6954, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.0897435897435899e-06, | |
| "loss": 1.7052, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.692307692307694e-07, | |
| "loss": 1.7458, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.4871794871794876e-07, | |
| "loss": 1.7088, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.282051282051282e-07, | |
| "loss": 1.7122, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 624, | |
| "total_flos": 1.810492168297513e+17, | |
| "train_loss": 2.042313535626118, | |
| "train_runtime": 4896.3607, | |
| "train_samples_per_second": 1.02, | |
| "train_steps_per_second": 0.127 | |
| } | |
| ], | |
| "max_steps": 624, | |
| "num_train_epochs": 2, | |
| "total_flos": 1.810492168297513e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |