| { | |
| "best_metric": 23.77271681619508, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch12-base/checkpoint-4000", | |
| "epoch": 7.256894049346879, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.6178, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.5622, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.3573, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.1809, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.9557, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.807, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.7347, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.7561, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.6812, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.711, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.6641, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.6543, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.6148, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.6127, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.5817, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.613, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.5465, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.5562, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.6016, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.5573, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.5493, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.522, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.5193, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.5282, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.5183, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.5304, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.527, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.4992, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.3507, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.344, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.392, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.3685, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.367, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.3448, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3755, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.3775, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.3387, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.3721, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.3831, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.3536, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_loss": 0.4082595407962799, | |
| "eval_runtime": 401.1017, | |
| "eval_samples_per_second": 9.03, | |
| "eval_steps_per_second": 1.506, | |
| "eval_wer": 26.188175753393146, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.3454, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.3647, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.3482, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.3475, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.3471, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.3415, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.3612, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.3525, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.3536, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.3449, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.3343, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.3271, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.35, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.3739, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.3376, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.2202, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.2069, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.2024, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.2183, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.1858, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.2051, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.2023, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.1891, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.2067, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.1905, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.2083, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.1987, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.2146, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.2211, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.2161, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.1922, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.1897, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.1973, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.187, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.2022, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.2181, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.2331, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.1941, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.196, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.2171, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_loss": 0.3793865740299225, | |
| "eval_runtime": 291.3301, | |
| "eval_samples_per_second": 12.433, | |
| "eval_steps_per_second": 2.073, | |
| "eval_wer": 24.481251437773178, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.2101, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.1878, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.1685, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.121, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.1144, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.1229, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.1122, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.1087, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.1141, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.1254, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.1167, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.1208, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.1117, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.1142, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.1289, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.1238, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.1247, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.119, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.1128, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.1174, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.128, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.1169, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.1164, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.1266, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.1192, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.1237, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.1284, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.1382, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.1158, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.1315, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.0887, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.069, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.0662, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.0736, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.073, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.0757, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.0669, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.0705, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.0748, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.0604, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "eval_loss": 0.3954152464866638, | |
| "eval_runtime": 272.7073, | |
| "eval_samples_per_second": 13.282, | |
| "eval_steps_per_second": 2.215, | |
| "eval_wer": 24.559466298596732, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.0752, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.0678, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.0744, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.0814, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.0749, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.0736, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.0718, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.0724, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.0758, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.0802, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.0733, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.067, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.0786, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.0688, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.078, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.0734, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.0691, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.0752, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.0422, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.0439, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.0398, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.0448, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.0507, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.0421, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.0453, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.0405, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.0578, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.0386, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.0423, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.0445, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.0406, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.0487, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.0446, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.0471, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.0487, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.0505, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.049, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.0466, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.0477, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.0531, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "eval_loss": 0.40786442160606384, | |
| "eval_runtime": 251.3352, | |
| "eval_samples_per_second": 14.411, | |
| "eval_steps_per_second": 2.403, | |
| "eval_wer": 23.77271681619508, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.0489, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.0476, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.0424, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.0427, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.0406, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.0345, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.0292, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.0324, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.0329, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.0293, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.0307, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.0294, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.0322, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.0322, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.0361, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.0285, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.0342, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.0307, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.0285, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.0296, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.0295, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.0267, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.0344, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.0317, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.0317, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.0334, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.0355, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.0282, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.033, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.0297, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.0273, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.0325, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.0268, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.0241, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.0252, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.0251, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.0229, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.0223, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.0236, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.0245, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "eval_loss": 0.42402514815330505, | |
| "eval_runtime": 233.6904, | |
| "eval_samples_per_second": 15.499, | |
| "eval_steps_per_second": 2.585, | |
| "eval_wer": 23.92914653784219, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "step": 5000, | |
| "total_flos": 3.88842162978816e+18, | |
| "train_loss": 0.2170446047782898, | |
| "train_runtime": 4566.507, | |
| "train_samples_per_second": 13.139, | |
| "train_steps_per_second": 1.095 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 8, | |
| "total_flos": 3.88842162978816e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |