|
{ |
|
"best_metric": 18.417108833893636, |
|
"best_model_checkpoint": "./checkpoint-5000", |
|
"epoch": 6.0112, |
|
"eval_steps": 1000, |
|
"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": 1.669, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.400000000000001e-07, |
|
"loss": 1.5129, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.44e-06, |
|
"loss": 1.2674, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.94e-06, |
|
"loss": 1.0082, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.4400000000000004e-06, |
|
"loss": 0.8399, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.9400000000000002e-06, |
|
"loss": 0.7935, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.44e-06, |
|
"loss": 0.7408, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.94e-06, |
|
"loss": 0.6816, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.440000000000001e-06, |
|
"loss": 0.6786, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.94e-06, |
|
"loss": 0.6969, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.4400000000000004e-06, |
|
"loss": 0.6485, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.94e-06, |
|
"loss": 0.5832, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.440000000000001e-06, |
|
"loss": 0.5365, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.9400000000000005e-06, |
|
"loss": 0.4981, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 7.440000000000001e-06, |
|
"loss": 0.5089, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 7.94e-06, |
|
"loss": 0.4927, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 8.44e-06, |
|
"loss": 0.5053, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 8.94e-06, |
|
"loss": 0.491, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.440000000000001e-06, |
|
"loss": 0.4335, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.940000000000001e-06, |
|
"loss": 0.3663, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.951111111111111e-06, |
|
"loss": 0.3869, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.895555555555557e-06, |
|
"loss": 0.3972, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.84e-06, |
|
"loss": 0.3781, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.784444444444445e-06, |
|
"loss": 0.359, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.72888888888889e-06, |
|
"loss": 0.3472, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.673333333333334e-06, |
|
"loss": 0.3711, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.617777777777778e-06, |
|
"loss": 0.3371, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.562222222222223e-06, |
|
"loss": 0.3427, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.506666666666667e-06, |
|
"loss": 0.3115, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.451111111111112e-06, |
|
"loss": 0.3236, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.395555555555556e-06, |
|
"loss": 0.3023, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.340000000000002e-06, |
|
"loss": 0.3073, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 9.284444444444444e-06, |
|
"loss": 0.336, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 9.22888888888889e-06, |
|
"loss": 0.3505, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 9.173333333333334e-06, |
|
"loss": 0.3154, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 9.117777777777778e-06, |
|
"loss": 0.3212, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 9.062222222222224e-06, |
|
"loss": 0.2841, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 9.006666666666666e-06, |
|
"loss": 0.2533, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.951111111111112e-06, |
|
"loss": 0.271, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.895555555555556e-06, |
|
"loss": 0.2826, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"eval_loss": 0.3471951186656952, |
|
"eval_runtime": 1237.3622, |
|
"eval_samples_per_second": 5.327, |
|
"eval_steps_per_second": 0.666, |
|
"eval_wer": 24.934181214305966, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 8.84e-06, |
|
"loss": 0.2507, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.784444444444446e-06, |
|
"loss": 0.2705, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 8.72888888888889e-06, |
|
"loss": 0.2895, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.673333333333334e-06, |
|
"loss": 0.2562, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 8.617777777777778e-06, |
|
"loss": 0.2131, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 8.562222222222224e-06, |
|
"loss": 0.2321, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 8.506666666666668e-06, |
|
"loss": 0.2034, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 8.451111111111112e-06, |
|
"loss": 0.2057, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 8.395555555555557e-06, |
|
"loss": 0.1869, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 8.34e-06, |
|
"loss": 0.2028, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 8.284444444444446e-06, |
|
"loss": 0.1924, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 8.22888888888889e-06, |
|
"loss": 0.1676, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 8.173333333333334e-06, |
|
"loss": 0.1521, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 8.11777777777778e-06, |
|
"loss": 0.1512, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 8.062222222222222e-06, |
|
"loss": 0.1613, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 8.006666666666667e-06, |
|
"loss": 0.1463, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 7.951111111111111e-06, |
|
"loss": 0.1434, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.895555555555557e-06, |
|
"loss": 0.1657, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 7.840000000000001e-06, |
|
"loss": 0.1337, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 7.784444444444445e-06, |
|
"loss": 0.1457, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 7.72888888888889e-06, |
|
"loss": 0.1262, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 7.673333333333333e-06, |
|
"loss": 0.1225, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 7.617777777777778e-06, |
|
"loss": 0.1327, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 7.562222222222223e-06, |
|
"loss": 0.1256, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 7.506666666666668e-06, |
|
"loss": 0.1258, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 7.451111111111111e-06, |
|
"loss": 0.1436, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 7.395555555555556e-06, |
|
"loss": 0.1599, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 7.340000000000001e-06, |
|
"loss": 0.1238, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 7.284444444444445e-06, |
|
"loss": 0.1355, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 7.22888888888889e-06, |
|
"loss": 0.1269, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 7.173333333333335e-06, |
|
"loss": 0.1139, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 7.117777777777778e-06, |
|
"loss": 0.1299, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 7.062222222222223e-06, |
|
"loss": 0.1183, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 7.006666666666667e-06, |
|
"loss": 0.1129, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 6.951111111111112e-06, |
|
"loss": 0.1178, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 6.8955555555555565e-06, |
|
"loss": 0.1212, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 6.8400000000000014e-06, |
|
"loss": 0.1163, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 6.784444444444445e-06, |
|
"loss": 0.1063, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 6.7288888888888895e-06, |
|
"loss": 0.0954, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 6.6733333333333335e-06, |
|
"loss": 0.0872, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"eval_loss": 0.30121901631355286, |
|
"eval_runtime": 1165.0993, |
|
"eval_samples_per_second": 5.657, |
|
"eval_steps_per_second": 0.707, |
|
"eval_wer": 20.266110413544496, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 6.617777777777778e-06, |
|
"loss": 0.0807, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 6.562222222222223e-06, |
|
"loss": 0.0877, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 6.5066666666666665e-06, |
|
"loss": 0.0924, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 6.451111111111111e-06, |
|
"loss": 0.084, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 6.395555555555556e-06, |
|
"loss": 0.0668, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 6.34e-06, |
|
"loss": 0.0734, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 6.284444444444445e-06, |
|
"loss": 0.0604, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 6.22888888888889e-06, |
|
"loss": 0.0679, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 6.173333333333333e-06, |
|
"loss": 0.0683, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 6.117777777777778e-06, |
|
"loss": 0.0634, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 6.062222222222223e-06, |
|
"loss": 0.0571, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 6.006666666666667e-06, |
|
"loss": 0.0642, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 5.951111111111112e-06, |
|
"loss": 0.0602, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 5.895555555555557e-06, |
|
"loss": 0.0469, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 5.84e-06, |
|
"loss": 0.0546, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 5.784444444444445e-06, |
|
"loss": 0.0578, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 5.72888888888889e-06, |
|
"loss": 0.0534, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 5.673333333333334e-06, |
|
"loss": 0.0568, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 5.617777777777779e-06, |
|
"loss": 0.0553, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"learning_rate": 5.562222222222222e-06, |
|
"loss": 0.0671, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"learning_rate": 5.506666666666667e-06, |
|
"loss": 0.061, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 3.02, |
|
"learning_rate": 5.451111111111112e-06, |
|
"loss": 0.0612, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 3.02, |
|
"learning_rate": 5.3955555555555565e-06, |
|
"loss": 0.049, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 5.3400000000000005e-06, |
|
"loss": 0.0451, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 5.2844444444444454e-06, |
|
"loss": 0.0558, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 3.04, |
|
"learning_rate": 5.228888888888889e-06, |
|
"loss": 0.0496, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 3.04, |
|
"learning_rate": 5.1733333333333335e-06, |
|
"loss": 0.0477, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 3.05, |
|
"learning_rate": 5.117777777777778e-06, |
|
"loss": 0.0523, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 3.05, |
|
"learning_rate": 5.062222222222222e-06, |
|
"loss": 0.0498, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 3.06, |
|
"learning_rate": 5.006666666666667e-06, |
|
"loss": 0.0488, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 3.06, |
|
"learning_rate": 4.951111111111111e-06, |
|
"loss": 0.0452, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 3.07, |
|
"learning_rate": 4.895555555555556e-06, |
|
"loss": 0.043, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 3.07, |
|
"learning_rate": 4.84e-06, |
|
"loss": 0.0345, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 3.08, |
|
"learning_rate": 4.784444444444445e-06, |
|
"loss": 0.0376, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 3.08, |
|
"learning_rate": 4.728888888888889e-06, |
|
"loss": 0.0384, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 4.673333333333333e-06, |
|
"loss": 0.0367, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 4.617777777777778e-06, |
|
"loss": 0.0344, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 3.1, |
|
"learning_rate": 4.562222222222222e-06, |
|
"loss": 0.0306, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 3.1, |
|
"learning_rate": 4.506666666666667e-06, |
|
"loss": 0.024, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 3.11, |
|
"learning_rate": 4.451111111111112e-06, |
|
"loss": 0.0275, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 3.11, |
|
"eval_loss": 0.3085091710090637, |
|
"eval_runtime": 1167.7383, |
|
"eval_samples_per_second": 5.644, |
|
"eval_steps_per_second": 0.706, |
|
"eval_wer": 19.302118352302642, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 3.11, |
|
"learning_rate": 4.395555555555556e-06, |
|
"loss": 0.0252, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"learning_rate": 4.34e-06, |
|
"loss": 0.0258, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"learning_rate": 4.284444444444445e-06, |
|
"loss": 0.0271, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 3.13, |
|
"learning_rate": 4.228888888888889e-06, |
|
"loss": 0.0256, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 3.13, |
|
"learning_rate": 4.173333333333334e-06, |
|
"loss": 0.0245, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"learning_rate": 4.117777777777779e-06, |
|
"loss": 0.0233, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"learning_rate": 4.062222222222223e-06, |
|
"loss": 0.0231, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 3.15, |
|
"learning_rate": 4.006666666666667e-06, |
|
"loss": 0.0192, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 3.15, |
|
"learning_rate": 3.951111111111112e-06, |
|
"loss": 0.0226, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 3.16, |
|
"learning_rate": 3.895555555555556e-06, |
|
"loss": 0.0191, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 3.16, |
|
"learning_rate": 3.8400000000000005e-06, |
|
"loss": 0.0205, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 3.784444444444445e-06, |
|
"loss": 0.0187, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 4.01, |
|
"learning_rate": 3.728888888888889e-06, |
|
"loss": 0.0294, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 4.01, |
|
"learning_rate": 3.673333333333334e-06, |
|
"loss": 0.0232, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 3.617777777777778e-06, |
|
"loss": 0.0241, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 3.5622222222222224e-06, |
|
"loss": 0.019, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 4.03, |
|
"learning_rate": 3.5066666666666673e-06, |
|
"loss": 0.0214, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 4.03, |
|
"learning_rate": 3.4511111111111113e-06, |
|
"loss": 0.0221, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 4.04, |
|
"learning_rate": 3.3955555555555558e-06, |
|
"loss": 0.0196, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 4.04, |
|
"learning_rate": 3.3400000000000006e-06, |
|
"loss": 0.02, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 4.05, |
|
"learning_rate": 3.2844444444444447e-06, |
|
"loss": 0.0207, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 4.05, |
|
"learning_rate": 3.228888888888889e-06, |
|
"loss": 0.0198, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 4.06, |
|
"learning_rate": 3.173333333333334e-06, |
|
"loss": 0.0189, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 4.06, |
|
"learning_rate": 3.117777777777778e-06, |
|
"loss": 0.0163, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 4.07, |
|
"learning_rate": 3.0622222222222225e-06, |
|
"loss": 0.0138, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 4.07, |
|
"learning_rate": 3.0066666666666674e-06, |
|
"loss": 0.0165, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 2.9511111111111114e-06, |
|
"loss": 0.0164, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 2.895555555555556e-06, |
|
"loss": 0.0163, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 4.09, |
|
"learning_rate": 2.84e-06, |
|
"loss": 0.014, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 4.09, |
|
"learning_rate": 2.784444444444445e-06, |
|
"loss": 0.0138, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 4.1, |
|
"learning_rate": 2.7288888888888893e-06, |
|
"loss": 0.0113, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 4.1, |
|
"learning_rate": 2.6733333333333333e-06, |
|
"loss": 0.0098, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 4.11, |
|
"learning_rate": 2.617777777777778e-06, |
|
"loss": 0.01, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 4.11, |
|
"learning_rate": 2.5622222222222226e-06, |
|
"loss": 0.0101, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 4.12, |
|
"learning_rate": 2.5066666666666667e-06, |
|
"loss": 0.0112, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 4.12, |
|
"learning_rate": 2.451111111111111e-06, |
|
"loss": 0.0088, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 4.13, |
|
"learning_rate": 2.3955555555555556e-06, |
|
"loss": 0.0119, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 4.13, |
|
"learning_rate": 2.3400000000000005e-06, |
|
"loss": 0.0112, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 4.14, |
|
"learning_rate": 2.2844444444444445e-06, |
|
"loss": 0.0086, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 4.14, |
|
"learning_rate": 2.228888888888889e-06, |
|
"loss": 0.0086, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 4.14, |
|
"eval_loss": 0.32967904210090637, |
|
"eval_runtime": 1171.2337, |
|
"eval_samples_per_second": 5.627, |
|
"eval_steps_per_second": 0.704, |
|
"eval_wer": 18.751265745878733, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 4.15, |
|
"learning_rate": 2.1733333333333334e-06, |
|
"loss": 0.0101, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 4.15, |
|
"learning_rate": 2.117777777777778e-06, |
|
"loss": 0.0072, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 2.0622222222222223e-06, |
|
"loss": 0.0078, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 2.006666666666667e-06, |
|
"loss": 0.0089, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 1.9511111111111113e-06, |
|
"loss": 0.009, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 1.8955555555555557e-06, |
|
"loss": 0.0111, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 1.8400000000000002e-06, |
|
"loss": 0.0101, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 5.02, |
|
"learning_rate": 1.7844444444444444e-06, |
|
"loss": 0.0117, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 5.02, |
|
"learning_rate": 1.728888888888889e-06, |
|
"loss": 0.0086, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 5.03, |
|
"learning_rate": 1.6733333333333335e-06, |
|
"loss": 0.0077, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 5.03, |
|
"learning_rate": 1.6177777777777778e-06, |
|
"loss": 0.0091, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 5.04, |
|
"learning_rate": 1.5622222222222225e-06, |
|
"loss": 0.0077, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 5.04, |
|
"learning_rate": 1.506666666666667e-06, |
|
"loss": 0.0079, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 5.05, |
|
"learning_rate": 1.4511111111111112e-06, |
|
"loss": 0.0099, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 5.05, |
|
"learning_rate": 1.3955555555555556e-06, |
|
"loss": 0.0078, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 5.06, |
|
"learning_rate": 1.34e-06, |
|
"loss": 0.0072, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 5.06, |
|
"learning_rate": 1.2844444444444445e-06, |
|
"loss": 0.0068, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 5.07, |
|
"learning_rate": 1.228888888888889e-06, |
|
"loss": 0.0068, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 5.07, |
|
"learning_rate": 1.1733333333333335e-06, |
|
"loss": 0.0059, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 5.08, |
|
"learning_rate": 1.117777777777778e-06, |
|
"loss": 0.0058, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 5.08, |
|
"learning_rate": 1.0622222222222222e-06, |
|
"loss": 0.008, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 5.09, |
|
"learning_rate": 1.0066666666666668e-06, |
|
"loss": 0.005, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 5.09, |
|
"learning_rate": 9.511111111111111e-07, |
|
"loss": 0.0059, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 5.1, |
|
"learning_rate": 8.955555555555557e-07, |
|
"loss": 0.0053, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 5.1, |
|
"learning_rate": 8.400000000000001e-07, |
|
"loss": 0.0048, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 5.11, |
|
"learning_rate": 7.844444444444445e-07, |
|
"loss": 0.0048, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 5.11, |
|
"learning_rate": 7.28888888888889e-07, |
|
"loss": 0.0056, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 5.12, |
|
"learning_rate": 6.733333333333334e-07, |
|
"loss": 0.0051, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 5.12, |
|
"learning_rate": 6.177777777777778e-07, |
|
"loss": 0.0049, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 5.13, |
|
"learning_rate": 5.622222222222223e-07, |
|
"loss": 0.0051, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 5.13, |
|
"learning_rate": 5.066666666666667e-07, |
|
"loss": 0.0052, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 5.14, |
|
"learning_rate": 4.511111111111111e-07, |
|
"loss": 0.0047, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 5.14, |
|
"learning_rate": 3.9555555555555557e-07, |
|
"loss": 0.0045, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 5.15, |
|
"learning_rate": 3.4000000000000003e-07, |
|
"loss": 0.0046, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 5.15, |
|
"learning_rate": 2.844444444444445e-07, |
|
"loss": 0.0042, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 5.16, |
|
"learning_rate": 2.2888888888888892e-07, |
|
"loss": 0.0047, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 5.16, |
|
"learning_rate": 1.7333333333333335e-07, |
|
"loss": 0.0042, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"learning_rate": 1.1777777777777778e-07, |
|
"loss": 0.0053, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"learning_rate": 6.222222222222223e-08, |
|
"loss": 0.0062, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"learning_rate": 6.666666666666667e-09, |
|
"loss": 0.0051, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"eval_loss": 0.3390321731567383, |
|
"eval_runtime": 1157.2511, |
|
"eval_samples_per_second": 5.695, |
|
"eval_steps_per_second": 0.712, |
|
"eval_wer": 18.417108833893636, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 6.01, |
|
"step": 5000, |
|
"total_flos": 2.308336900079616e+19, |
|
"train_loss": 0.15512495822161437, |
|
"train_runtime": 16533.0342, |
|
"train_samples_per_second": 4.839, |
|
"train_steps_per_second": 0.302 |
|
} |
|
], |
|
"logging_steps": 25, |
|
"max_steps": 5000, |
|
"num_train_epochs": 9223372036854775807, |
|
"save_steps": 1000, |
|
"total_flos": 2.308336900079616e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|