| { | |
| "best_metric": 3.8273540533062804, | |
| "best_model_checkpoint": "./whisper-medium-id/checkpoint-3000", | |
| "epoch": 3.3101621979476996, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8e-08, | |
| "loss": 0.086, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.8e-08, | |
| "loss": 0.0456, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.4799999999999998e-07, | |
| "loss": 0.061, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.98e-07, | |
| "loss": 0.068, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.48e-07, | |
| "loss": 0.0314, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.98e-07, | |
| "loss": 0.0716, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.4799999999999994e-07, | |
| "loss": 0.0464, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.98e-07, | |
| "loss": 0.0438, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.48e-07, | |
| "loss": 0.0551, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.979999999999999e-07, | |
| "loss": 0.0408, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 5.48e-07, | |
| "loss": 0.0489, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.979999999999999e-07, | |
| "loss": 0.0696, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 6.48e-07, | |
| "loss": 0.0325, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 6.979999999999999e-07, | |
| "loss": 0.0518, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.48e-07, | |
| "loss": 0.06, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 7.98e-07, | |
| "loss": 0.0415, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 8.48e-07, | |
| "loss": 0.0543, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.98e-07, | |
| "loss": 0.0489, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.479999999999999e-07, | |
| "loss": 0.0522, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.98e-07, | |
| "loss": 0.0526, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.974736842105262e-07, | |
| "loss": 0.0443, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.948421052631578e-07, | |
| "loss": 0.0476, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.922105263157894e-07, | |
| "loss": 0.0442, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.89578947368421e-07, | |
| "loss": 0.0664, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.869473684210526e-07, | |
| "loss": 0.0468, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.843157894736842e-07, | |
| "loss": 0.0565, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.816842105263157e-07, | |
| "loss": 0.0414, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.790526315789473e-07, | |
| "loss": 0.0625, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.76421052631579e-07, | |
| "loss": 0.0367, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.737894736842105e-07, | |
| "loss": 0.0636, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.71157894736842e-07, | |
| "loss": 0.036, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.685263157894736e-07, | |
| "loss": 0.0498, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.658947368421052e-07, | |
| "loss": 0.0513, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.632631578947368e-07, | |
| "loss": 0.0303, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.606315789473684e-07, | |
| "loss": 0.0397, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.58e-07, | |
| "loss": 0.0264, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.553684210526316e-07, | |
| "loss": 0.0559, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.52736842105263e-07, | |
| "loss": 0.0382, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.501052631578947e-07, | |
| "loss": 0.0526, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.474736842105263e-07, | |
| "loss": 0.0427, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_loss": 0.06637874245643616, | |
| "eval_runtime": 2541.3578, | |
| "eval_samples_per_second": 1.424, | |
| "eval_steps_per_second": 0.089, | |
| "eval_wer": 4.380706446555381, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.448421052631579e-07, | |
| "loss": 0.0374, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.422105263157895e-07, | |
| "loss": 0.0472, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.39578947368421e-07, | |
| "loss": 0.0392, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.369473684210526e-07, | |
| "loss": 0.0389, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.343157894736842e-07, | |
| "loss": 0.0344, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.316842105263157e-07, | |
| "loss": 0.0391, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.290526315789473e-07, | |
| "loss": 0.0342, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.264210526315789e-07, | |
| "loss": 0.0474, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.237894736842104e-07, | |
| "loss": 0.0485, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.21157894736842e-07, | |
| "loss": 0.0301, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.185263157894736e-07, | |
| "loss": 0.0347, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.158947368421053e-07, | |
| "loss": 0.0395, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.132631578947369e-07, | |
| "loss": 0.0243, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.106315789473684e-07, | |
| "loss": 0.0508, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.08e-07, | |
| "loss": 0.0416, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.053684210526315e-07, | |
| "loss": 0.0246, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.027368421052631e-07, | |
| "loss": 0.0255, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.001052631578947e-07, | |
| "loss": 0.0406, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.974736842105263e-07, | |
| "loss": 0.035, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.948421052631578e-07, | |
| "loss": 0.0492, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.922105263157894e-07, | |
| "loss": 0.026, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.89578947368421e-07, | |
| "loss": 0.024, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.869473684210526e-07, | |
| "loss": 0.0425, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.843157894736842e-07, | |
| "loss": 0.0552, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.816842105263157e-07, | |
| "loss": 0.0514, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.790526315789474e-07, | |
| "loss": 0.0377, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.764210526315789e-07, | |
| "loss": 0.0454, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.737894736842105e-07, | |
| "loss": 0.0351, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.711578947368421e-07, | |
| "loss": 0.044, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.685263157894737e-07, | |
| "loss": 0.0474, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.658947368421052e-07, | |
| "loss": 0.0263, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.632631578947368e-07, | |
| "loss": 0.0399, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.606315789473684e-07, | |
| "loss": 0.0306, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.58e-07, | |
| "loss": 0.0474, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.553684210526316e-07, | |
| "loss": 0.0312, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.52736842105263e-07, | |
| "loss": 0.0371, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.501052631578946e-07, | |
| "loss": 0.0306, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.474736842105262e-07, | |
| "loss": 0.05, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.448421052631579e-07, | |
| "loss": 0.0478, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 8.422105263157895e-07, | |
| "loss": 0.042, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_loss": 0.06577838957309723, | |
| "eval_runtime": 2452.9212, | |
| "eval_samples_per_second": 1.475, | |
| "eval_steps_per_second": 0.093, | |
| "eval_wer": 3.942635801899843, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.396842105263158e-07, | |
| "loss": 0.0423, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.370526315789473e-07, | |
| "loss": 0.0422, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 8.344210526315789e-07, | |
| "loss": 0.0357, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.317894736842105e-07, | |
| "loss": 0.0464, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.291578947368421e-07, | |
| "loss": 0.0522, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.265263157894737e-07, | |
| "loss": 0.0419, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.238947368421051e-07, | |
| "loss": 0.0484, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.212631578947367e-07, | |
| "loss": 0.0275, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.186315789473683e-07, | |
| "loss": 0.0419, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.159999999999999e-07, | |
| "loss": 0.0308, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.133684210526316e-07, | |
| "loss": 0.0612, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.107368421052632e-07, | |
| "loss": 0.0444, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.081052631578947e-07, | |
| "loss": 0.0379, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 8.054736842105263e-07, | |
| "loss": 0.045, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.028421052631579e-07, | |
| "loss": 0.0362, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.002105263157895e-07, | |
| "loss": 0.0436, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.97578947368421e-07, | |
| "loss": 0.0187, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.949473684210526e-07, | |
| "loss": 0.0283, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.923157894736841e-07, | |
| "loss": 0.0419, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.896842105263157e-07, | |
| "loss": 0.031, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.870526315789473e-07, | |
| "loss": 0.0306, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.844210526315789e-07, | |
| "loss": 0.0379, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.817894736842105e-07, | |
| "loss": 0.0344, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.791578947368421e-07, | |
| "loss": 0.0318, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.765263157894737e-07, | |
| "loss": 0.0466, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.738947368421053e-07, | |
| "loss": 0.0546, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.712631578947368e-07, | |
| "loss": 0.042, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.686315789473684e-07, | |
| "loss": 0.0505, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.66e-07, | |
| "loss": 0.0315, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 7.633684210526315e-07, | |
| "loss": 0.0272, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 7.607368421052631e-07, | |
| "loss": 0.0288, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.581052631578947e-07, | |
| "loss": 0.0352, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.554736842105263e-07, | |
| "loss": 0.0589, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.528421052631579e-07, | |
| "loss": 0.0456, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.502105263157894e-07, | |
| "loss": 0.0338, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 7.47578947368421e-07, | |
| "loss": 0.0385, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 7.449473684210525e-07, | |
| "loss": 0.0344, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.423157894736842e-07, | |
| "loss": 0.0361, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.396842105263158e-07, | |
| "loss": 0.0539, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.370526315789474e-07, | |
| "loss": 0.0265, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_loss": 0.06573929637670517, | |
| "eval_runtime": 2464.359, | |
| "eval_samples_per_second": 1.468, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.8273540533062804, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.344210526315789e-07, | |
| "loss": 0.0242, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.317894736842105e-07, | |
| "loss": 0.025, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.291578947368421e-07, | |
| "loss": 0.0211, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.265263157894737e-07, | |
| "loss": 0.0178, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.238947368421053e-07, | |
| "loss": 0.0216, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 7.212631578947368e-07, | |
| "loss": 0.0262, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.186315789473683e-07, | |
| "loss": 0.0177, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.159999999999999e-07, | |
| "loss": 0.0281, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 7.133684210526315e-07, | |
| "loss": 0.0173, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.107368421052631e-07, | |
| "loss": 0.0179, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.081052631578946e-07, | |
| "loss": 0.0357, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.054736842105263e-07, | |
| "loss": 0.0169, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7.028421052631579e-07, | |
| "loss": 0.0147, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.002105263157895e-07, | |
| "loss": 0.0229, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.975789473684211e-07, | |
| "loss": 0.0192, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.949473684210527e-07, | |
| "loss": 0.0199, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.923157894736841e-07, | |
| "loss": 0.043, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.896842105263157e-07, | |
| "loss": 0.0325, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.870526315789473e-07, | |
| "loss": 0.015, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.844210526315789e-07, | |
| "loss": 0.0206, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.817894736842105e-07, | |
| "loss": 0.0221, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.79157894736842e-07, | |
| "loss": 0.0131, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.765263157894736e-07, | |
| "loss": 0.0202, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.738947368421052e-07, | |
| "loss": 0.0254, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.712631578947369e-07, | |
| "loss": 0.0231, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.686315789473685e-07, | |
| "loss": 0.015, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.66e-07, | |
| "loss": 0.0181, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.633684210526315e-07, | |
| "loss": 0.0175, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 6.607368421052631e-07, | |
| "loss": 0.0225, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 6.581052631578947e-07, | |
| "loss": 0.0222, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 6.554736842105263e-07, | |
| "loss": 0.0177, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 6.528421052631579e-07, | |
| "loss": 0.0148, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.502105263157894e-07, | |
| "loss": 0.0147, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 6.47578947368421e-07, | |
| "loss": 0.0217, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.449473684210526e-07, | |
| "loss": 0.0229, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 6.423157894736842e-07, | |
| "loss": 0.0247, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.396842105263157e-07, | |
| "loss": 0.0151, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.370526315789472e-07, | |
| "loss": 0.0344, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.344210526315789e-07, | |
| "loss": 0.0156, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.317894736842105e-07, | |
| "loss": 0.0211, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_loss": 0.0678883045911789, | |
| "eval_runtime": 2456.4416, | |
| "eval_samples_per_second": 1.473, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.8365765931937656, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.291578947368421e-07, | |
| "loss": 0.0259, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 6.266315789473684e-07, | |
| "loss": 0.0251, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.24e-07, | |
| "loss": 0.0272, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 6.213684210526316e-07, | |
| "loss": 0.0217, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.187368421052631e-07, | |
| "loss": 0.0252, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.161052631578947e-07, | |
| "loss": 0.0171, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.134736842105263e-07, | |
| "loss": 0.0205, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.108421052631578e-07, | |
| "loss": 0.0279, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.082105263157894e-07, | |
| "loss": 0.0235, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 6.05578947368421e-07, | |
| "loss": 0.0317, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.029473684210526e-07, | |
| "loss": 0.0166, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.003157894736842e-07, | |
| "loss": 0.0247, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.976842105263158e-07, | |
| "loss": 0.0368, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.950526315789474e-07, | |
| "loss": 0.0269, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.92421052631579e-07, | |
| "loss": 0.0228, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.897894736842105e-07, | |
| "loss": 0.018, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.871578947368421e-07, | |
| "loss": 0.0124, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.845263157894736e-07, | |
| "loss": 0.0138, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 5.818947368421052e-07, | |
| "loss": 0.0257, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 5.792631578947368e-07, | |
| "loss": 0.0249, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 5.766315789473683e-07, | |
| "loss": 0.0296, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.739999999999999e-07, | |
| "loss": 0.0132, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 5.713684210526315e-07, | |
| "loss": 0.0251, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5.687368421052632e-07, | |
| "loss": 0.0291, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 5.661052631578948e-07, | |
| "loss": 0.0288, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 5.634736842105264e-07, | |
| "loss": 0.0246, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 5.608421052631579e-07, | |
| "loss": 0.0401, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 5.582105263157894e-07, | |
| "loss": 0.0256, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 5.55578947368421e-07, | |
| "loss": 0.0194, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 5.529473684210526e-07, | |
| "loss": 0.0233, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 5.503157894736842e-07, | |
| "loss": 0.0228, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 5.476842105263157e-07, | |
| "loss": 0.0124, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 5.450526315789473e-07, | |
| "loss": 0.0276, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.424210526315789e-07, | |
| "loss": 0.0197, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.397894736842105e-07, | |
| "loss": 0.024, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 5.371578947368421e-07, | |
| "loss": 0.0164, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 5.345263157894738e-07, | |
| "loss": 0.0239, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 5.318947368421052e-07, | |
| "loss": 0.0203, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 5.292631578947368e-07, | |
| "loss": 0.0153, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 5.266315789473684e-07, | |
| "loss": 0.0212, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "eval_loss": 0.06819388270378113, | |
| "eval_runtime": 2455.7945, | |
| "eval_samples_per_second": 1.473, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.8411878631375083, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 5.24e-07, | |
| "loss": 0.0194, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 5.213684210526316e-07, | |
| "loss": 0.0234, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 5.187368421052631e-07, | |
| "loss": 0.0218, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 5.161052631578947e-07, | |
| "loss": 0.0208, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.134736842105263e-07, | |
| "loss": 0.0284, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 5.108421052631579e-07, | |
| "loss": 0.0227, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 5.082105263157895e-07, | |
| "loss": 0.0204, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 5.05578947368421e-07, | |
| "loss": 0.0274, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 5.029473684210525e-07, | |
| "loss": 0.0259, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 5.003157894736841e-07, | |
| "loss": 0.0153, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.976842105263158e-07, | |
| "loss": 0.0199, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.950526315789474e-07, | |
| "loss": 0.0176, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.924210526315789e-07, | |
| "loss": 0.0213, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.897894736842105e-07, | |
| "loss": 0.0253, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.871578947368421e-07, | |
| "loss": 0.0248, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.845263157894737e-07, | |
| "loss": 0.0277, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.818947368421053e-07, | |
| "loss": 0.0185, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.792631578947368e-07, | |
| "loss": 0.0151, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.766315789473684e-07, | |
| "loss": 0.0185, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.7399999999999993e-07, | |
| "loss": 0.0149, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.7136842105263156e-07, | |
| "loss": 0.0267, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.6873684210526315e-07, | |
| "loss": 0.0136, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.6610526315789473e-07, | |
| "loss": 0.0249, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.634736842105263e-07, | |
| "loss": 0.0177, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.6084210526315784e-07, | |
| "loss": 0.022, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.582105263157894e-07, | |
| "loss": 0.0232, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.55578947368421e-07, | |
| "loss": 0.0232, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.5294736842105264e-07, | |
| "loss": 0.0166, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.503157894736842e-07, | |
| "loss": 0.0287, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.476842105263158e-07, | |
| "loss": 0.0263, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.4505263157894733e-07, | |
| "loss": 0.0138, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.424210526315789e-07, | |
| "loss": 0.029, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.397894736842105e-07, | |
| "loss": 0.0218, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.371578947368421e-07, | |
| "loss": 0.0154, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.345263157894737e-07, | |
| "loss": 0.0259, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.3189473684210524e-07, | |
| "loss": 0.029, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.292631578947368e-07, | |
| "loss": 0.0122, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.266315789473684e-07, | |
| "loss": 0.0175, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.24e-07, | |
| "loss": 0.0126, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.213684210526315e-07, | |
| "loss": 0.0206, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_loss": 0.06833098828792572, | |
| "eval_runtime": 2456.4618, | |
| "eval_samples_per_second": 1.473, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.8688554827999635, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.1873684210526315e-07, | |
| "loss": 0.0229, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.1610526315789473e-07, | |
| "loss": 0.0361, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.134736842105263e-07, | |
| "loss": 0.0197, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 4.108421052631579e-07, | |
| "loss": 0.0151, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 4.082105263157894e-07, | |
| "loss": 0.0095, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.05578947368421e-07, | |
| "loss": 0.0135, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 4.029473684210526e-07, | |
| "loss": 0.0136, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 4.003157894736842e-07, | |
| "loss": 0.0164, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.976842105263158e-07, | |
| "loss": 0.012, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.950526315789474e-07, | |
| "loss": 0.0128, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.924210526315789e-07, | |
| "loss": 0.0138, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.898947368421052e-07, | |
| "loss": 0.0108, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.8726315789473684e-07, | |
| "loss": 0.0212, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.846315789473684e-07, | |
| "loss": 0.0093, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.82e-07, | |
| "loss": 0.0148, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.7936842105263153e-07, | |
| "loss": 0.018, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 3.767368421052631e-07, | |
| "loss": 0.0157, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.741052631578947e-07, | |
| "loss": 0.0242, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.714736842105263e-07, | |
| "loss": 0.0094, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.688421052631579e-07, | |
| "loss": 0.0157, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.6621052631578944e-07, | |
| "loss": 0.0094, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.63578947368421e-07, | |
| "loss": 0.0125, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.609473684210526e-07, | |
| "loss": 0.0087, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.583157894736842e-07, | |
| "loss": 0.0084, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.5568421052631577e-07, | |
| "loss": 0.0112, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.5305263157894735e-07, | |
| "loss": 0.0154, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.5042105263157894e-07, | |
| "loss": 0.0124, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.477894736842105e-07, | |
| "loss": 0.0086, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.451578947368421e-07, | |
| "loss": 0.009, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.425263157894737e-07, | |
| "loss": 0.0139, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.398947368421052e-07, | |
| "loss": 0.0074, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.372631578947368e-07, | |
| "loss": 0.0097, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.346315789473684e-07, | |
| "loss": 0.0175, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 3.32e-07, | |
| "loss": 0.0168, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.293684210526316e-07, | |
| "loss": 0.0143, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.267368421052631e-07, | |
| "loss": 0.0143, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.241052631578947e-07, | |
| "loss": 0.014, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.214736842105263e-07, | |
| "loss": 0.0113, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.1884210526315786e-07, | |
| "loss": 0.016, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.162105263157895e-07, | |
| "loss": 0.0166, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "eval_loss": 0.07106726616621017, | |
| "eval_runtime": 2458.2732, | |
| "eval_samples_per_second": 1.472, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.9656921516185557, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.1357894736842103e-07, | |
| "loss": 0.0185, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.109473684210526e-07, | |
| "loss": 0.011, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.083157894736842e-07, | |
| "loss": 0.0123, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.056842105263158e-07, | |
| "loss": 0.0112, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 3.0305263157894736e-07, | |
| "loss": 0.0087, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.004210526315789e-07, | |
| "loss": 0.0129, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.977894736842105e-07, | |
| "loss": 0.0149, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.951578947368421e-07, | |
| "loss": 0.0138, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.925263157894737e-07, | |
| "loss": 0.017, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.8989473684210526e-07, | |
| "loss": 0.0078, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 2.872631578947368e-07, | |
| "loss": 0.0138, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.846315789473684e-07, | |
| "loss": 0.0188, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.8199999999999996e-07, | |
| "loss": 0.0139, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.793684210526316e-07, | |
| "loss": 0.02, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 2.767368421052632e-07, | |
| "loss": 0.0093, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.741052631578947e-07, | |
| "loss": 0.0132, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 2.714736842105263e-07, | |
| "loss": 0.0081, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.6884210526315787e-07, | |
| "loss": 0.0179, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.6621052631578945e-07, | |
| "loss": 0.0172, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.635789473684211e-07, | |
| "loss": 0.021, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.609473684210526e-07, | |
| "loss": 0.0188, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.583157894736842e-07, | |
| "loss": 0.0157, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 2.556842105263158e-07, | |
| "loss": 0.0125, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.5305263157894736e-07, | |
| "loss": 0.0061, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.5042105263157894e-07, | |
| "loss": 0.0124, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 2.477894736842105e-07, | |
| "loss": 0.0094, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 2.451578947368421e-07, | |
| "loss": 0.0172, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.425263157894737e-07, | |
| "loss": 0.0153, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.3989473684210527e-07, | |
| "loss": 0.0174, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.3726315789473682e-07, | |
| "loss": 0.0152, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.346315789473684e-07, | |
| "loss": 0.0094, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.32e-07, | |
| "loss": 0.008, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.2936842105263157e-07, | |
| "loss": 0.0119, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.2673684210526315e-07, | |
| "loss": 0.014, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.241052631578947e-07, | |
| "loss": 0.0103, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.2147368421052631e-07, | |
| "loss": 0.0111, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.188421052631579e-07, | |
| "loss": 0.0095, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.1621052631578945e-07, | |
| "loss": 0.0181, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.1357894736842106e-07, | |
| "loss": 0.0138, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.1094736842105261e-07, | |
| "loss": 0.0095, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "eval_loss": 0.07166043668985367, | |
| "eval_runtime": 2459.2401, | |
| "eval_samples_per_second": 1.471, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.9979710412247536, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.083157894736842e-07, | |
| "loss": 0.0153, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.056842105263158e-07, | |
| "loss": 0.0126, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.0305263157894736e-07, | |
| "loss": 0.0127, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.0042105263157894e-07, | |
| "loss": 0.0095, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.977894736842105e-07, | |
| "loss": 0.0134, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1.951578947368421e-07, | |
| "loss": 0.0123, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.925263157894737e-07, | |
| "loss": 0.014, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.8989473684210524e-07, | |
| "loss": 0.0149, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.8726315789473685e-07, | |
| "loss": 0.0087, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.846315789473684e-07, | |
| "loss": 0.0164, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.82e-07, | |
| "loss": 0.0162, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.7936842105263157e-07, | |
| "loss": 0.0128, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.7673684210526315e-07, | |
| "loss": 0.0156, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.7410526315789473e-07, | |
| "loss": 0.0094, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.714736842105263e-07, | |
| "loss": 0.0109, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.688421052631579e-07, | |
| "loss": 0.0087, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.6621052631578948e-07, | |
| "loss": 0.0096, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.6368421052631577e-07, | |
| "loss": 0.0112, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.6105263157894735e-07, | |
| "loss": 0.0105, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.5842105263157894e-07, | |
| "loss": 0.014, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.5578947368421052e-07, | |
| "loss": 0.0128, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.531578947368421e-07, | |
| "loss": 0.022, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.5052631578947366e-07, | |
| "loss": 0.0203, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.4789473684210526e-07, | |
| "loss": 0.017, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.4526315789473685e-07, | |
| "loss": 0.0098, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.426315789473684e-07, | |
| "loss": 0.0137, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 1.4e-07, | |
| "loss": 0.0155, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.3736842105263156e-07, | |
| "loss": 0.0131, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.3473684210526315e-07, | |
| "loss": 0.0116, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.3210526315789473e-07, | |
| "loss": 0.0072, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.294736842105263e-07, | |
| "loss": 0.0172, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.268421052631579e-07, | |
| "loss": 0.0096, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.2421052631578947e-07, | |
| "loss": 0.0076, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.2157894736842106e-07, | |
| "loss": 0.0151, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.1894736842105262e-07, | |
| "loss": 0.0067, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.163157894736842e-07, | |
| "loss": 0.0118, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.1368421052631579e-07, | |
| "loss": 0.0117, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.1105263157894737e-07, | |
| "loss": 0.0332, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.0842105263157894e-07, | |
| "loss": 0.0187, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.0578947368421052e-07, | |
| "loss": 0.0122, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "eval_loss": 0.07136213779449463, | |
| "eval_runtime": 2461.9446, | |
| "eval_samples_per_second": 1.47, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.9795259614497835, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.031578947368421e-07, | |
| "loss": 0.0075, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.0052631578947368e-07, | |
| "loss": 0.0071, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 9.789473684210525e-08, | |
| "loss": 0.0103, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.526315789473683e-08, | |
| "loss": 0.0099, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 9.263157894736842e-08, | |
| "loss": 0.004, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 9e-08, | |
| "loss": 0.0144, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.736842105263158e-08, | |
| "loss": 0.009, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 8.473684210526315e-08, | |
| "loss": 0.0111, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 8.210526315789473e-08, | |
| "loss": 0.0104, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.947368421052631e-08, | |
| "loss": 0.0068, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.68421052631579e-08, | |
| "loss": 0.0067, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 7.421052631578948e-08, | |
| "loss": 0.0086, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 7.157894736842105e-08, | |
| "loss": 0.0156, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.894736842105263e-08, | |
| "loss": 0.0114, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 6.631578947368421e-08, | |
| "loss": 0.0115, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 6.368421052631579e-08, | |
| "loss": 0.0157, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.105263157894736e-08, | |
| "loss": 0.0097, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.842105263157895e-08, | |
| "loss": 0.0131, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.578947368421052e-08, | |
| "loss": 0.0166, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.3157894736842105e-08, | |
| "loss": 0.007, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 5.052631578947368e-08, | |
| "loss": 0.0086, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 4.789473684210526e-08, | |
| "loss": 0.0171, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.5263157894736844e-08, | |
| "loss": 0.0088, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 4.263157894736842e-08, | |
| "loss": 0.0094, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 4e-08, | |
| "loss": 0.0112, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.7368421052631576e-08, | |
| "loss": 0.0156, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.473684210526316e-08, | |
| "loss": 0.0164, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.210526315789473e-08, | |
| "loss": 0.0045, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 2.9473684210526315e-08, | |
| "loss": 0.0099, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 2.6842105263157894e-08, | |
| "loss": 0.0079, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 2.4210526315789472e-08, | |
| "loss": 0.0055, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 2.157894736842105e-08, | |
| "loss": 0.0112, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.8947368421052632e-08, | |
| "loss": 0.0101, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.631578947368421e-08, | |
| "loss": 0.0072, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.368421052631579e-08, | |
| "loss": 0.0047, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.1052631578947366e-08, | |
| "loss": 0.0059, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 8.421052631578947e-09, | |
| "loss": 0.0121, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 5.789473684210526e-09, | |
| "loss": 0.0039, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.157894736842105e-09, | |
| "loss": 0.0086, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 5.263157894736842e-10, | |
| "loss": 0.0049, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "eval_loss": 0.07196281105279922, | |
| "eval_runtime": 2460.623, | |
| "eval_samples_per_second": 1.47, | |
| "eval_steps_per_second": 0.092, | |
| "eval_wer": 3.9887485013372683, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "step": 10000, | |
| "total_flos": 1.632967852032e+20, | |
| "train_loss": 0.024496208691596986, | |
| "train_runtime": 75684.0421, | |
| "train_samples_per_second": 2.114, | |
| "train_steps_per_second": 0.132 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 4, | |
| "total_flos": 1.632967852032e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |