| { | |
| "best_metric": 11.36701337295691, | |
| "best_model_checkpoint": "./data/finetuningRuns/whisper-md-el-intlv-xs/checkpoint-10000", | |
| "epoch": 24.875621890547265, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.84e-07, | |
| "loss": 0.5528, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 7.84e-07, | |
| "loss": 0.4833, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.1839999999999998e-06, | |
| "loss": 0.3464, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.584e-06, | |
| "loss": 0.2632, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.984e-06, | |
| "loss": 0.2358, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.384e-06, | |
| "loss": 0.2058, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.7839999999999995e-06, | |
| "loss": 0.1928, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.184e-06, | |
| "loss": 0.1642, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.584e-06, | |
| "loss": 0.1726, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9839999999999995e-06, | |
| "loss": 0.1561, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.384e-06, | |
| "loss": 0.1498, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.7839999999999994e-06, | |
| "loss": 0.1418, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.184e-06, | |
| "loss": 0.14, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.583999999999999e-06, | |
| "loss": 0.1299, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.984e-06, | |
| "loss": 0.1209, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.384e-06, | |
| "loss": 0.1251, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.784e-06, | |
| "loss": 0.0814, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.184e-06, | |
| "loss": 0.0831, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.583999999999999e-06, | |
| "loss": 0.0725, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.984e-06, | |
| "loss": 0.0773, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.97978947368421e-06, | |
| "loss": 0.0706, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.958736842105263e-06, | |
| "loss": 0.0723, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.937684210526315e-06, | |
| "loss": 0.0731, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.916631578947368e-06, | |
| "loss": 0.0781, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.89557894736842e-06, | |
| "loss": 0.0744, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 7.874526315789473e-06, | |
| "loss": 0.0652, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 7.853473684210526e-06, | |
| "loss": 0.0652, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.832421052631579e-06, | |
| "loss": 0.0714, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.811368421052631e-06, | |
| "loss": 0.0576, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 7.790315789473684e-06, | |
| "loss": 0.0605, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.769263157894737e-06, | |
| "loss": 0.0669, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 7.74821052631579e-06, | |
| "loss": 0.0591, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.727157894736842e-06, | |
| "loss": 0.0327, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.706105263157894e-06, | |
| "loss": 0.0254, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.685052631578947e-06, | |
| "loss": 0.0241, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 7.664e-06, | |
| "loss": 0.026, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.642947368421052e-06, | |
| "loss": 0.0259, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.621894736842104e-06, | |
| "loss": 0.0269, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.600842105263158e-06, | |
| "loss": 0.0228, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.5797894736842104e-06, | |
| "loss": 0.0251, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "eval_loss": 0.22164975106716156, | |
| "eval_runtime": 1081.9546, | |
| "eval_samples_per_second": 1.568, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 12.583580980683507, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.558736842105263e-06, | |
| "loss": 0.0242, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.537684210526316e-06, | |
| "loss": 0.0281, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.516631578947368e-06, | |
| "loss": 0.0245, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 7.495578947368421e-06, | |
| "loss": 0.0253, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 7.474526315789474e-06, | |
| "loss": 0.0241, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 7.4534736842105255e-06, | |
| "loss": 0.0232, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.432421052631578e-06, | |
| "loss": 0.0208, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 7.411368421052631e-06, | |
| "loss": 0.0203, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.3903157894736835e-06, | |
| "loss": 0.0123, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 7.369263157894736e-06, | |
| "loss": 0.0106, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 7.348210526315789e-06, | |
| "loss": 0.0092, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 7.327157894736842e-06, | |
| "loss": 0.0089, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 7.306105263157895e-06, | |
| "loss": 0.0117, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 7.285052631578948e-06, | |
| "loss": 0.0117, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 7.264e-06, | |
| "loss": 0.0084, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 7.242947368421052e-06, | |
| "loss": 0.01, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 7.221894736842105e-06, | |
| "loss": 0.0119, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 7.200842105263157e-06, | |
| "loss": 0.0105, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 7.17978947368421e-06, | |
| "loss": 0.0098, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 7.158736842105263e-06, | |
| "loss": 0.01, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 7.137684210526315e-06, | |
| "loss": 0.0098, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 7.116631578947368e-06, | |
| "loss": 0.0084, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 7.095578947368421e-06, | |
| "loss": 0.0105, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 7.074526315789473e-06, | |
| "loss": 0.0113, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 7.053473684210526e-06, | |
| "loss": 0.006, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 7.0324210526315795e-06, | |
| "loss": 0.0044, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 7.011368421052631e-06, | |
| "loss": 0.0042, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 6.990315789473684e-06, | |
| "loss": 0.0057, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 6.9692631578947366e-06, | |
| "loss": 0.0047, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 6.948210526315789e-06, | |
| "loss": 0.0055, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 6.927157894736842e-06, | |
| "loss": 0.005, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 6.9061052631578945e-06, | |
| "loss": 0.0055, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 6.885052631578947e-06, | |
| "loss": 0.0049, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 6.864e-06, | |
| "loss": 0.0041, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 6.8429473684210525e-06, | |
| "loss": 0.0052, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 6.821894736842104e-06, | |
| "loss": 0.0044, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 6.800842105263157e-06, | |
| "loss": 0.004, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 6.77978947368421e-06, | |
| "loss": 0.0046, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 6.758736842105263e-06, | |
| "loss": 0.0061, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 6.737684210526316e-06, | |
| "loss": 0.0051, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "eval_loss": 0.2874409854412079, | |
| "eval_runtime": 1079.0071, | |
| "eval_samples_per_second": 1.572, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 12.295690936106983, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 6.7166315789473684e-06, | |
| "loss": 0.004, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 6.695578947368421e-06, | |
| "loss": 0.0032, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 6.674526315789474e-06, | |
| "loss": 0.0032, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 6.653473684210526e-06, | |
| "loss": 0.0028, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 6.632421052631579e-06, | |
| "loss": 0.0028, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 6.611368421052631e-06, | |
| "loss": 0.0028, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 6.5903157894736835e-06, | |
| "loss": 0.0027, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 6.569263157894736e-06, | |
| "loss": 0.0028, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 6.548210526315789e-06, | |
| "loss": 0.0031, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 6.5271578947368415e-06, | |
| "loss": 0.0027, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 6.506105263157894e-06, | |
| "loss": 0.0027, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 6.485052631578947e-06, | |
| "loss": 0.0027, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 6.464e-06, | |
| "loss": 0.0028, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 6.442947368421053e-06, | |
| "loss": 0.0022, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 6.421894736842106e-06, | |
| "loss": 0.0027, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 6.400842105263157e-06, | |
| "loss": 0.0032, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 6.37978947368421e-06, | |
| "loss": 0.003, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 6.358736842105263e-06, | |
| "loss": 0.0018, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 6.337684210526315e-06, | |
| "loss": 0.0019, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 6.316631578947368e-06, | |
| "loss": 0.0019, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 6.295578947368421e-06, | |
| "loss": 0.0021, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 6.274526315789473e-06, | |
| "loss": 0.0028, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 6.253473684210526e-06, | |
| "loss": 0.0027, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 6.232421052631579e-06, | |
| "loss": 0.0021, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 6.211368421052631e-06, | |
| "loss": 0.0015, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 6.190315789473683e-06, | |
| "loss": 0.0018, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 6.169263157894737e-06, | |
| "loss": 0.0019, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 6.148210526315789e-06, | |
| "loss": 0.0016, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 6.127157894736842e-06, | |
| "loss": 0.0015, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 6.1061052631578946e-06, | |
| "loss": 0.002, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 6.085052631578947e-06, | |
| "loss": 0.0028, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 6.064e-06, | |
| "loss": 0.0022, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 6.0429473684210525e-06, | |
| "loss": 0.002, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 6.021894736842105e-06, | |
| "loss": 0.0014, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 6.000842105263158e-06, | |
| "loss": 0.0019, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 5.97978947368421e-06, | |
| "loss": 0.0013, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 5.958736842105262e-06, | |
| "loss": 0.0016, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 5.937684210526315e-06, | |
| "loss": 0.0012, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 5.916631578947368e-06, | |
| "loss": 0.0019, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 5.895578947368421e-06, | |
| "loss": 0.0015, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "eval_loss": 0.32812297344207764, | |
| "eval_runtime": 1086.8672, | |
| "eval_samples_per_second": 1.56, | |
| "eval_steps_per_second": 0.195, | |
| "eval_wer": 11.905646359583951, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 5.874526315789474e-06, | |
| "loss": 0.0016, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "learning_rate": 5.8534736842105264e-06, | |
| "loss": 0.0021, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 5.832421052631579e-06, | |
| "loss": 0.001, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 5.811368421052632e-06, | |
| "loss": 0.001, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 5.790315789473684e-06, | |
| "loss": 0.0011, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 5.769263157894736e-06, | |
| "loss": 0.0028, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 5.748210526315789e-06, | |
| "loss": 0.0018, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 5.7271578947368415e-06, | |
| "loss": 0.0017, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 5.706105263157894e-06, | |
| "loss": 0.0021, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 5.685052631578947e-06, | |
| "loss": 0.0019, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 5.6639999999999995e-06, | |
| "loss": 0.0019, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 5.642947368421052e-06, | |
| "loss": 0.0026, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 5.621894736842105e-06, | |
| "loss": 0.0016, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 5.600842105263158e-06, | |
| "loss": 0.0014, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 5.579789473684211e-06, | |
| "loss": 0.0019, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 5.558736842105263e-06, | |
| "loss": 0.0018, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 5.537684210526315e-06, | |
| "loss": 0.0014, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 5.516631578947368e-06, | |
| "loss": 0.0021, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 5.495578947368421e-06, | |
| "loss": 0.0015, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 5.474526315789473e-06, | |
| "loss": 0.0021, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 5.453473684210526e-06, | |
| "loss": 0.0024, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 5.432421052631579e-06, | |
| "loss": 0.0017, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 5.411368421052631e-06, | |
| "loss": 0.0021, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.390315789473684e-06, | |
| "loss": 0.0013, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.369263157894737e-06, | |
| "loss": 0.0015, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 5.3482105263157884e-06, | |
| "loss": 0.0016, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 5.327157894736842e-06, | |
| "loss": 0.0011, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 5.306105263157895e-06, | |
| "loss": 0.0013, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 5.285052631578947e-06, | |
| "loss": 0.0013, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 5.264e-06, | |
| "loss": 0.0014, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "learning_rate": 5.2429473684210526e-06, | |
| "loss": 0.0012, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 5.221894736842105e-06, | |
| "loss": 0.0017, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 5.200842105263158e-06, | |
| "loss": 0.0013, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 9.58, | |
| "learning_rate": 5.1797894736842105e-06, | |
| "loss": 0.0014, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 5.158736842105263e-06, | |
| "loss": 0.0017, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 5.137684210526315e-06, | |
| "loss": 0.0025, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 5.116631578947368e-06, | |
| "loss": 0.0018, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 5.09557894736842e-06, | |
| "loss": 0.0017, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 5.074526315789473e-06, | |
| "loss": 0.0018, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 5.053473684210526e-06, | |
| "loss": 0.0017, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "eval_loss": 0.31783589720726013, | |
| "eval_runtime": 1082.091, | |
| "eval_samples_per_second": 1.567, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 12.592867756315007, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.032421052631579e-06, | |
| "loss": 0.0015, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "learning_rate": 5.011368421052632e-06, | |
| "loss": 0.0007, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 4.9903157894736844e-06, | |
| "loss": 0.0009, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "learning_rate": 4.969263157894737e-06, | |
| "loss": 0.0012, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "learning_rate": 4.94821052631579e-06, | |
| "loss": 0.0011, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 4.9271578947368415e-06, | |
| "loss": 0.0008, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 4.906105263157894e-06, | |
| "loss": 0.0015, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "learning_rate": 4.885052631578947e-06, | |
| "loss": 0.0012, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "learning_rate": 4.8639999999999995e-06, | |
| "loss": 0.0014, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 4.842947368421052e-06, | |
| "loss": 0.0009, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "learning_rate": 4.821894736842105e-06, | |
| "loss": 0.0011, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 10.7, | |
| "learning_rate": 4.8008421052631575e-06, | |
| "loss": 0.0008, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 10.76, | |
| "learning_rate": 4.77978947368421e-06, | |
| "loss": 0.0015, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "learning_rate": 4.758736842105264e-06, | |
| "loss": 0.0012, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 10.88, | |
| "learning_rate": 4.737684210526316e-06, | |
| "loss": 0.0016, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "learning_rate": 4.716631578947369e-06, | |
| "loss": 0.002, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 4.695578947368421e-06, | |
| "loss": 0.0016, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 11.07, | |
| "learning_rate": 4.674526315789473e-06, | |
| "loss": 0.0005, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 11.13, | |
| "learning_rate": 4.653473684210526e-06, | |
| "loss": 0.0007, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 4.632421052631579e-06, | |
| "loss": 0.0006, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 4.611368421052631e-06, | |
| "loss": 0.0008, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 4.590315789473684e-06, | |
| "loss": 0.0005, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 4.569263157894737e-06, | |
| "loss": 0.0005, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 4.548210526315789e-06, | |
| "loss": 0.0006, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 11.5, | |
| "learning_rate": 4.527157894736842e-06, | |
| "loss": 0.0004, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 11.57, | |
| "learning_rate": 4.506105263157894e-06, | |
| "loss": 0.0005, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 4.4850526315789464e-06, | |
| "loss": 0.0006, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "learning_rate": 4.464e-06, | |
| "loss": 0.0007, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "learning_rate": 4.442947368421053e-06, | |
| "loss": 0.0012, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 11.82, | |
| "learning_rate": 4.421894736842105e-06, | |
| "loss": 0.0007, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 4.400842105263158e-06, | |
| "loss": 0.0006, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 11.94, | |
| "learning_rate": 4.3797894736842106e-06, | |
| "loss": 0.0009, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.358736842105263e-06, | |
| "loss": 0.0005, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 4.337684210526316e-06, | |
| "loss": 0.0005, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 4.3166315789473685e-06, | |
| "loss": 0.0004, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 4.29557894736842e-06, | |
| "loss": 0.0006, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 12.25, | |
| "learning_rate": 4.274526315789473e-06, | |
| "loss": 0.0006, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 4.253473684210526e-06, | |
| "loss": 0.0005, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "learning_rate": 4.232421052631578e-06, | |
| "loss": 0.0005, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 4.211368421052631e-06, | |
| "loss": 0.0008, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "eval_loss": 0.3448905348777771, | |
| "eval_runtime": 1083.7087, | |
| "eval_samples_per_second": 1.565, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 11.979940564635958, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 4.190315789473684e-06, | |
| "loss": 0.0004, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 4.169263157894737e-06, | |
| "loss": 0.0013, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 12.62, | |
| "learning_rate": 4.14821052631579e-06, | |
| "loss": 0.0009, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 12.69, | |
| "learning_rate": 4.1271578947368424e-06, | |
| "loss": 0.0007, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 4.106105263157895e-06, | |
| "loss": 0.0007, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 4.085052631578947e-06, | |
| "loss": 0.0004, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "learning_rate": 4.0639999999999995e-06, | |
| "loss": 0.0005, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 4.042947368421052e-06, | |
| "loss": 0.0005, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 4.021894736842105e-06, | |
| "loss": 0.0006, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 13.06, | |
| "learning_rate": 4.0008421052631575e-06, | |
| "loss": 0.0009, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 13.12, | |
| "learning_rate": 3.97978947368421e-06, | |
| "loss": 0.0003, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 13.18, | |
| "learning_rate": 3.958736842105263e-06, | |
| "loss": 0.0007, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 13.25, | |
| "learning_rate": 3.9376842105263155e-06, | |
| "loss": 0.0002, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 13.31, | |
| "learning_rate": 3.916631578947368e-06, | |
| "loss": 0.0003, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "learning_rate": 3.895578947368421e-06, | |
| "loss": 0.0002, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 13.43, | |
| "learning_rate": 3.8745263157894734e-06, | |
| "loss": 0.0005, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 3.853473684210526e-06, | |
| "loss": 0.0002, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 3.832421052631579e-06, | |
| "loss": 0.0004, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 3.8113684210526314e-06, | |
| "loss": 0.0003, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 3.790315789473684e-06, | |
| "loss": 0.0002, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 13.74, | |
| "learning_rate": 3.7692631578947367e-06, | |
| "loss": 0.0003, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 13.81, | |
| "learning_rate": 3.7482105263157894e-06, | |
| "loss": 0.0002, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 13.87, | |
| "learning_rate": 3.727157894736842e-06, | |
| "loss": 0.0002, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 3.7061052631578943e-06, | |
| "loss": 0.0004, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "learning_rate": 3.6850526315789473e-06, | |
| "loss": 0.0003, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 14.05, | |
| "learning_rate": 3.664e-06, | |
| "loss": 0.0004, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 14.12, | |
| "learning_rate": 3.6429473684210526e-06, | |
| "loss": 0.0004, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 14.18, | |
| "learning_rate": 3.6218947368421053e-06, | |
| "loss": 0.0002, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 14.24, | |
| "learning_rate": 3.6008421052631575e-06, | |
| "loss": 0.0001, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "learning_rate": 3.57978947368421e-06, | |
| "loss": 0.0004, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "learning_rate": 3.558736842105263e-06, | |
| "loss": 0.0003, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 14.43, | |
| "learning_rate": 3.537684210526316e-06, | |
| "loss": 0.0001, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "learning_rate": 3.5166315789473686e-06, | |
| "loss": 0.0001, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 14.55, | |
| "learning_rate": 3.495578947368421e-06, | |
| "loss": 0.0007, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 14.61, | |
| "learning_rate": 3.4745263157894735e-06, | |
| "loss": 0.0003, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "learning_rate": 3.453473684210526e-06, | |
| "loss": 0.0002, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 14.74, | |
| "learning_rate": 3.4324210526315788e-06, | |
| "loss": 0.0002, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 14.8, | |
| "learning_rate": 3.4113684210526314e-06, | |
| "loss": 0.0001, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "learning_rate": 3.390315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 14.93, | |
| "learning_rate": 3.3692631578947367e-06, | |
| "loss": 0.0001, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 14.93, | |
| "eval_loss": 0.3638215661048889, | |
| "eval_runtime": 1086.1753, | |
| "eval_samples_per_second": 1.561, | |
| "eval_steps_per_second": 0.195, | |
| "eval_wer": 11.710624071322437, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 14.99, | |
| "learning_rate": 3.3482105263157894e-06, | |
| "loss": 0.0001, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 3.327157894736842e-06, | |
| "loss": 0.0001, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 3.3061052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 3.285052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 15.24, | |
| "learning_rate": 3.2639999999999996e-06, | |
| "loss": 0.0001, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 15.3, | |
| "learning_rate": 3.2429473684210527e-06, | |
| "loss": 0.0001, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 15.36, | |
| "learning_rate": 3.2218947368421053e-06, | |
| "loss": 0.0001, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 15.42, | |
| "learning_rate": 3.200842105263158e-06, | |
| "loss": 0.0001, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 15.49, | |
| "learning_rate": 3.17978947368421e-06, | |
| "loss": 0.0001, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 15.55, | |
| "learning_rate": 3.158736842105263e-06, | |
| "loss": 0.0001, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 15.61, | |
| "learning_rate": 3.1376842105263155e-06, | |
| "loss": 0.0001, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 3.1166315789473686e-06, | |
| "loss": 0.0001, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 15.73, | |
| "learning_rate": 3.0955789473684212e-06, | |
| "loss": 0.0001, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 15.8, | |
| "learning_rate": 3.0745263157894735e-06, | |
| "loss": 0.0001, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 15.86, | |
| "learning_rate": 3.053473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 15.92, | |
| "learning_rate": 3.0324210526315788e-06, | |
| "loss": 0.0001, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "learning_rate": 3.0113684210526314e-06, | |
| "loss": 0.0001, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.990315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 16.11, | |
| "learning_rate": 2.9692631578947367e-06, | |
| "loss": 0.0001, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "learning_rate": 2.9482105263157894e-06, | |
| "loss": 0.0001, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 16.23, | |
| "learning_rate": 2.927157894736842e-06, | |
| "loss": 0.0001, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 16.29, | |
| "learning_rate": 2.9061052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 2.8850526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 2.8639999999999996e-06, | |
| "loss": 0.0001, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 16.48, | |
| "learning_rate": 2.8429473684210523e-06, | |
| "loss": 0.0001, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 2.8218947368421053e-06, | |
| "loss": 0.0001, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 16.6, | |
| "learning_rate": 2.800842105263158e-06, | |
| "loss": 0.0001, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 2.7797894736842106e-06, | |
| "loss": 0.0001, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "learning_rate": 2.758736842105263e-06, | |
| "loss": 0.0001, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 16.79, | |
| "learning_rate": 2.7376842105263155e-06, | |
| "loss": 0.0001, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 16.85, | |
| "learning_rate": 2.716631578947368e-06, | |
| "loss": 0.0001, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 16.92, | |
| "learning_rate": 2.695578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 16.98, | |
| "learning_rate": 2.674526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 2.653473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 17.1, | |
| "learning_rate": 2.632421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 17.16, | |
| "learning_rate": 2.6113684210526315e-06, | |
| "loss": 0.0001, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 2.590315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 17.29, | |
| "learning_rate": 2.5692631578947368e-06, | |
| "loss": 0.0001, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 17.35, | |
| "learning_rate": 2.548210526315789e-06, | |
| "loss": 0.0001, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "learning_rate": 2.527157894736842e-06, | |
| "loss": 0.0001, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "eval_loss": 0.39095476269721985, | |
| "eval_runtime": 1082.2874, | |
| "eval_samples_per_second": 1.567, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 11.49702823179792, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 2.5061052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 17.54, | |
| "learning_rate": 2.4850526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 17.6, | |
| "learning_rate": 2.464e-06, | |
| "loss": 0.0001, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 17.66, | |
| "learning_rate": 2.4429473684210523e-06, | |
| "loss": 0.0001, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 17.72, | |
| "learning_rate": 2.421894736842105e-06, | |
| "loss": 0.0001, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 17.79, | |
| "learning_rate": 2.400842105263158e-06, | |
| "loss": 0.0001, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 2.3797894736842107e-06, | |
| "loss": 0.0001, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 17.91, | |
| "learning_rate": 2.3587368421052633e-06, | |
| "loss": 0.0001, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 17.97, | |
| "learning_rate": 2.3376842105263155e-06, | |
| "loss": 0.0001, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 2.316631578947368e-06, | |
| "loss": 0.0001, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 18.1, | |
| "learning_rate": 2.295578947368421e-06, | |
| "loss": 0.0001, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 18.16, | |
| "learning_rate": 2.2745263157894735e-06, | |
| "loss": 0.0001, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 2.2534736842105266e-06, | |
| "loss": 0.0001, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 18.28, | |
| "learning_rate": 2.232421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 18.35, | |
| "learning_rate": 2.2113684210526315e-06, | |
| "loss": 0.0001, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 18.41, | |
| "learning_rate": 2.190315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 18.47, | |
| "learning_rate": 2.1692631578947368e-06, | |
| "loss": 0.0001, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 18.53, | |
| "learning_rate": 2.1482105263157894e-06, | |
| "loss": 0.0001, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 18.59, | |
| "learning_rate": 2.1271578947368417e-06, | |
| "loss": 0.0001, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 18.66, | |
| "learning_rate": 2.1061052631578947e-06, | |
| "loss": 0.0001, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 18.72, | |
| "learning_rate": 2.0850526315789474e-06, | |
| "loss": 0.0, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 18.78, | |
| "learning_rate": 2.064e-06, | |
| "loss": 0.0001, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 18.84, | |
| "learning_rate": 2.0429473684210527e-06, | |
| "loss": 0.0001, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 18.91, | |
| "learning_rate": 2.021894736842105e-06, | |
| "loss": 0.0001, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 2.0008421052631576e-06, | |
| "loss": 0.0, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 1.9797894736842103e-06, | |
| "loss": 0.0001, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 19.09, | |
| "learning_rate": 1.958736842105263e-06, | |
| "loss": 0.0, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 1.937684210526316e-06, | |
| "loss": 0.0, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 19.22, | |
| "learning_rate": 1.9166315789473682e-06, | |
| "loss": 0.0, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 19.28, | |
| "learning_rate": 1.8955789473684209e-06, | |
| "loss": 0.0, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 19.34, | |
| "learning_rate": 1.8745263157894735e-06, | |
| "loss": 0.0, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 19.4, | |
| "learning_rate": 1.8534736842105264e-06, | |
| "loss": 0.0001, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 19.47, | |
| "learning_rate": 1.8324210526315788e-06, | |
| "loss": 0.0, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 19.53, | |
| "learning_rate": 1.8113684210526315e-06, | |
| "loss": 0.0, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 1.7903157894736841e-06, | |
| "loss": 0.0, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 19.65, | |
| "learning_rate": 1.7692631578947368e-06, | |
| "loss": 0.0, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 19.71, | |
| "learning_rate": 1.7482105263157895e-06, | |
| "loss": 0.0, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 19.78, | |
| "learning_rate": 1.727157894736842e-06, | |
| "loss": 0.0, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 19.84, | |
| "learning_rate": 1.7061052631578948e-06, | |
| "loss": 0.0, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 19.9, | |
| "learning_rate": 1.6850526315789472e-06, | |
| "loss": 0.0, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 19.9, | |
| "eval_loss": 0.40417808294296265, | |
| "eval_runtime": 1080.6756, | |
| "eval_samples_per_second": 1.569, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 11.394873699851411, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 19.96, | |
| "learning_rate": 1.6639999999999999e-06, | |
| "loss": 0.0, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 1.6429473684210527e-06, | |
| "loss": 0.0, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 20.09, | |
| "learning_rate": 1.6218947368421052e-06, | |
| "loss": 0.0, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 20.15, | |
| "learning_rate": 1.6008421052631578e-06, | |
| "loss": 0.0, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 20.21, | |
| "learning_rate": 1.5797894736842105e-06, | |
| "loss": 0.0, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 20.27, | |
| "learning_rate": 1.5587368421052631e-06, | |
| "loss": 0.0, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 20.34, | |
| "learning_rate": 1.5376842105263158e-06, | |
| "loss": 0.0, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 20.4, | |
| "learning_rate": 1.5166315789473682e-06, | |
| "loss": 0.0, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 20.46, | |
| "learning_rate": 1.495578947368421e-06, | |
| "loss": 0.0, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 20.52, | |
| "learning_rate": 1.4745263157894735e-06, | |
| "loss": 0.0, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 20.58, | |
| "learning_rate": 1.4534736842105262e-06, | |
| "loss": 0.0, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 20.65, | |
| "learning_rate": 1.432421052631579e-06, | |
| "loss": 0.0, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 20.71, | |
| "learning_rate": 1.4113684210526315e-06, | |
| "loss": 0.0, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 20.77, | |
| "learning_rate": 1.3903157894736842e-06, | |
| "loss": 0.0, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 20.83, | |
| "learning_rate": 1.3692631578947366e-06, | |
| "loss": 0.0, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 20.9, | |
| "learning_rate": 1.3482105263157895e-06, | |
| "loss": 0.0, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 20.96, | |
| "learning_rate": 1.3271578947368421e-06, | |
| "loss": 0.0, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 21.02, | |
| "learning_rate": 1.3061052631578946e-06, | |
| "loss": 0.0, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 21.08, | |
| "learning_rate": 1.2850526315789474e-06, | |
| "loss": 0.0, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 21.14, | |
| "learning_rate": 1.2639999999999999e-06, | |
| "loss": 0.0, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 21.21, | |
| "learning_rate": 1.2429473684210525e-06, | |
| "loss": 0.0, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 21.27, | |
| "learning_rate": 1.2218947368421054e-06, | |
| "loss": 0.0, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 21.33, | |
| "learning_rate": 1.2008421052631578e-06, | |
| "loss": 0.0, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 21.39, | |
| "learning_rate": 1.1797894736842105e-06, | |
| "loss": 0.0, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 21.46, | |
| "learning_rate": 1.158736842105263e-06, | |
| "loss": 0.0, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 21.52, | |
| "learning_rate": 1.1376842105263158e-06, | |
| "loss": 0.0, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 21.58, | |
| "learning_rate": 1.1166315789473685e-06, | |
| "loss": 0.0, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 21.64, | |
| "learning_rate": 1.095578947368421e-06, | |
| "loss": 0.0, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 21.7, | |
| "learning_rate": 1.0745263157894738e-06, | |
| "loss": 0.0, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 21.77, | |
| "learning_rate": 1.0534736842105262e-06, | |
| "loss": 0.0, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 21.83, | |
| "learning_rate": 1.0324210526315789e-06, | |
| "loss": 0.0, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 21.89, | |
| "learning_rate": 1.0113684210526315e-06, | |
| "loss": 0.0, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 21.95, | |
| "learning_rate": 9.903157894736842e-07, | |
| "loss": 0.0, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 22.01, | |
| "learning_rate": 9.692631578947368e-07, | |
| "loss": 0.0, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 22.08, | |
| "learning_rate": 9.482105263157895e-07, | |
| "loss": 0.0, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 22.14, | |
| "learning_rate": 9.27157894736842e-07, | |
| "loss": 0.0, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 22.2, | |
| "learning_rate": 9.061052631578947e-07, | |
| "loss": 0.0, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 22.26, | |
| "learning_rate": 8.850526315789472e-07, | |
| "loss": 0.0, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 22.33, | |
| "learning_rate": 8.639999999999999e-07, | |
| "loss": 0.0, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 22.39, | |
| "learning_rate": 8.429473684210527e-07, | |
| "loss": 0.0, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 22.39, | |
| "eval_loss": 0.41294780373573303, | |
| "eval_runtime": 1082.768, | |
| "eval_samples_per_second": 1.566, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 11.413447251114412, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 22.45, | |
| "learning_rate": 8.218947368421052e-07, | |
| "loss": 0.0, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 22.51, | |
| "learning_rate": 8.008421052631579e-07, | |
| "loss": 0.0, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 22.57, | |
| "learning_rate": 7.797894736842104e-07, | |
| "loss": 0.0, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 22.64, | |
| "learning_rate": 7.587368421052631e-07, | |
| "loss": 0.0, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 22.7, | |
| "learning_rate": 7.376842105263158e-07, | |
| "loss": 0.0, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 22.76, | |
| "learning_rate": 7.166315789473684e-07, | |
| "loss": 0.0, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 22.82, | |
| "learning_rate": 6.95578947368421e-07, | |
| "loss": 0.0, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 22.89, | |
| "learning_rate": 6.745263157894736e-07, | |
| "loss": 0.0, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 22.95, | |
| "learning_rate": 6.534736842105262e-07, | |
| "loss": 0.0, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 23.01, | |
| "learning_rate": 6.32421052631579e-07, | |
| "loss": 0.0, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 23.07, | |
| "learning_rate": 6.113684210526315e-07, | |
| "loss": 0.0, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 23.13, | |
| "learning_rate": 5.903157894736842e-07, | |
| "loss": 0.0, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 23.2, | |
| "learning_rate": 5.692631578947368e-07, | |
| "loss": 0.0, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 23.26, | |
| "learning_rate": 5.482105263157894e-07, | |
| "loss": 0.0, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 23.32, | |
| "learning_rate": 5.271578947368422e-07, | |
| "loss": 0.0, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 23.38, | |
| "learning_rate": 5.061052631578947e-07, | |
| "loss": 0.0, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 23.45, | |
| "learning_rate": 4.850526315789474e-07, | |
| "loss": 0.0, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 23.51, | |
| "learning_rate": 4.64e-07, | |
| "loss": 0.0, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 23.57, | |
| "learning_rate": 4.4294736842105263e-07, | |
| "loss": 0.0, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 23.63, | |
| "learning_rate": 4.2189473684210523e-07, | |
| "loss": 0.0, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 23.69, | |
| "learning_rate": 4.008421052631579e-07, | |
| "loss": 0.0, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 23.76, | |
| "learning_rate": 3.797894736842105e-07, | |
| "loss": 0.0, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 23.82, | |
| "learning_rate": 3.5873684210526314e-07, | |
| "loss": 0.0, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 23.88, | |
| "learning_rate": 3.376842105263158e-07, | |
| "loss": 0.0, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 23.94, | |
| "learning_rate": 3.166315789473684e-07, | |
| "loss": 0.0, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 2.9557894736842105e-07, | |
| "loss": 0.0, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 24.07, | |
| "learning_rate": 2.7452631578947365e-07, | |
| "loss": 0.0, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 24.13, | |
| "learning_rate": 2.534736842105263e-07, | |
| "loss": 0.0, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 24.19, | |
| "learning_rate": 2.3242105263157894e-07, | |
| "loss": 0.0, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 24.25, | |
| "learning_rate": 2.1136842105263157e-07, | |
| "loss": 0.0, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 24.32, | |
| "learning_rate": 1.903157894736842e-07, | |
| "loss": 0.0, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 24.38, | |
| "learning_rate": 1.6926315789473682e-07, | |
| "loss": 0.0, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 24.44, | |
| "learning_rate": 1.4821052631578948e-07, | |
| "loss": 0.0, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 24.5, | |
| "learning_rate": 1.271578947368421e-07, | |
| "loss": 0.0, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 24.56, | |
| "learning_rate": 1.0610526315789472e-07, | |
| "loss": 0.0, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 24.63, | |
| "learning_rate": 8.505263157894736e-08, | |
| "loss": 0.0, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 24.69, | |
| "learning_rate": 6.4e-08, | |
| "loss": 0.0, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 24.75, | |
| "learning_rate": 4.2947368421052626e-08, | |
| "loss": 0.0, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 24.81, | |
| "learning_rate": 2.189473684210526e-08, | |
| "loss": 0.0, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 24.88, | |
| "learning_rate": 8.421052631578947e-10, | |
| "loss": 0.0, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 24.88, | |
| "eval_loss": 0.4167824387550354, | |
| "eval_runtime": 1084.1284, | |
| "eval_samples_per_second": 1.564, | |
| "eval_steps_per_second": 0.196, | |
| "eval_wer": 11.36701337295691, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 24.88, | |
| "step": 10000, | |
| "total_flos": 3.264466032997171e+20, | |
| "train_loss": 0.013932611434056889, | |
| "train_runtime": 53727.7715, | |
| "train_samples_per_second": 5.956, | |
| "train_steps_per_second": 0.186 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 25, | |
| "total_flos": 3.264466032997171e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |