| { | |
| "best_metric": 12.431561996779388, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch4-small/checkpoint-4000", | |
| "epoch": 2.420135527589545, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.3401, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.1668, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.0854, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.849, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.5879, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.5093, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.4755, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.4808, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.4138, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.4688, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.4251, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.393, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.4098, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.4471, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.4498, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.4121, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.3592, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.352, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.3844, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.3635, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.3939, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.4215, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.4197, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.3824, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.3912, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.3353, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.3565, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.4071, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.426, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.4129, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.383, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.3947, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.373, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.365, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3594, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.4095, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.3394, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.2869, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.3483, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.3372, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_loss": 0.28929781913757324, | |
| "eval_runtime": 3047.9194, | |
| "eval_samples_per_second": 1.188, | |
| "eval_steps_per_second": 0.594, | |
| "eval_wer": 16.112261329652632, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.4129, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.3579, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.2991, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.3637, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.3843, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.3125, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.3964, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.3582, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.3196, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.3412, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.3271, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.3105, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.3362, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.3495, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.3975, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.3277, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.3868, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.3865, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.3027, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.363, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.3396, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.3708, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.3011, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.3037, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.3304, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.2976, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.313, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.3158, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.3008, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.2925, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.3376, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.302, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.3393, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.3359, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.2902, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.3257, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.3115, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.3705, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.318, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.2785, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 0.2590125799179077, | |
| "eval_runtime": 3122.5731, | |
| "eval_samples_per_second": 1.16, | |
| "eval_steps_per_second": 0.58, | |
| "eval_wer": 14.603174603174605, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.3832, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.35, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.2637, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.153, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.1344, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.102, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.1003, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.1636, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.1431, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.1382, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.1282, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.1492, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.1227, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.1327, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.1096, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.1512, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.1499, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.1132, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.1207, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.1107, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.1102, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.112, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.1672, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.1669, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.1338, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.1239, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.1317, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.1605, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.1302, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.1157, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.1581, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.1144, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.1515, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.1444, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.1072, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.1451, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.1129, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.1068, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.1161, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.1318, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_loss": 0.253537118434906, | |
| "eval_runtime": 3054.2788, | |
| "eval_samples_per_second": 1.186, | |
| "eval_steps_per_second": 0.593, | |
| "eval_wer": 13.85323211410168, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.1086, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.1147, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.1052, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.1155, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.1474, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.1887, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.1448, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.1153, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.1047, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.1426, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.1337, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.1225, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.1289, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.1191, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.0918, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.1275, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.1221, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.1171, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.1118, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.1068, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.1391, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.1484, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.1057, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.1257, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.1112, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.1119, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.1006, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.1004, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.1205, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.1031, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.1386, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.1371, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.1293, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.1037, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.1407, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.1186, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.1002, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.1335, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.1343, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.1384, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 0.23332776129245758, | |
| "eval_runtime": 5315.3349, | |
| "eval_samples_per_second": 0.681, | |
| "eval_steps_per_second": 0.341, | |
| "eval_wer": 12.431561996779388, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.1094, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.1046, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.1465, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.1261, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.126, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.0729, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.0477, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.0607, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.0482, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.0437, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.0519, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.042, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.0358, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.0483, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0537, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.0431, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.0527, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.0684, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.0367, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.0367, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.0472, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.0559, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.04, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.0378, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.0663, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.0312, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.0563, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.0451, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.0525, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.0866, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.0397, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.0452, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.0431, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.0384, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.0379, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.0458, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.0374, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.0553, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.0568, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.0541, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_loss": 0.2426508367061615, | |
| "eval_runtime": 6981.3896, | |
| "eval_samples_per_second": 0.519, | |
| "eval_steps_per_second": 0.259, | |
| "eval_wer": 12.564987347596043, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "step": 5000, | |
| "total_flos": 5.76997649399808e+18, | |
| "train_loss": 0.2271066749215126, | |
| "train_runtime": 40080.8286, | |
| "train_samples_per_second": 0.499, | |
| "train_steps_per_second": 0.125 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 3, | |
| "total_flos": 5.76997649399808e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |