|
{ |
|
"best_metric": 5.357219480798112, |
|
"best_model_checkpoint": "./output/small/yt-special-batch88/checkpoint-5000", |
|
"epoch": 7.9239302694136295, |
|
"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": 1.4731, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.400000000000001e-07, |
|
"loss": 1.1628, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.44e-06, |
|
"loss": 0.8201, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.94e-06, |
|
"loss": 1.416, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 2.4400000000000004e-06, |
|
"loss": 2.4172, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.9400000000000002e-06, |
|
"loss": 3.5424, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 3.4200000000000007e-06, |
|
"loss": 6.2909, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 3.920000000000001e-06, |
|
"loss": 8.9606, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.42e-06, |
|
"loss": 11.2381, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.86e-06, |
|
"loss": 14.5607, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 5.36e-06, |
|
"loss": 15.6953, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 5.86e-06, |
|
"loss": 18.715, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 6.360000000000001e-06, |
|
"loss": 24.623, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 6.860000000000001e-06, |
|
"loss": 30.027, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 7.360000000000001e-06, |
|
"loss": 35.2724, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 7.860000000000001e-06, |
|
"loss": 36.2271, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 8.36e-06, |
|
"loss": 39.7986, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 8.860000000000002e-06, |
|
"loss": 41.6492, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 9.360000000000002e-06, |
|
"loss": 43.7759, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 9.86e-06, |
|
"loss": 40.5691, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 9.960000000000001e-06, |
|
"loss": 45.5387, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 9.904444444444445e-06, |
|
"loss": 47.2133, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 9.84888888888889e-06, |
|
"loss": 46.206, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 9.793333333333333e-06, |
|
"loss": 47.8666, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 9.737777777777779e-06, |
|
"loss": 46.5186, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 9.682222222222223e-06, |
|
"loss": 42.6377, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 9.626666666666667e-06, |
|
"loss": 44.6983, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 9.571111111111113e-06, |
|
"loss": 43.3703, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 9.515555555555557e-06, |
|
"loss": 45.3232, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 9.460000000000001e-06, |
|
"loss": 41.2427, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 9.404444444444445e-06, |
|
"loss": 39.9395, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 9.348888888888889e-06, |
|
"loss": 40.3131, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 9.293333333333335e-06, |
|
"loss": 39.7779, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 9.237777777777779e-06, |
|
"loss": 41.2274, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 9.182222222222223e-06, |
|
"loss": 38.6106, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 9.126666666666667e-06, |
|
"loss": 36.0439, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 9.07111111111111e-06, |
|
"loss": 38.9444, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 9.015555555555557e-06, |
|
"loss": 40.5574, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 8.96e-06, |
|
"loss": 37.7151, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 8.904444444444446e-06, |
|
"loss": 37.1656, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"eval_loss": 31.41523551940918, |
|
"eval_runtime": 9519.214, |
|
"eval_samples_per_second": 0.53, |
|
"eval_steps_per_second": 0.066, |
|
"eval_wer": 569.1439605234928, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 8.848888888888889e-06, |
|
"loss": 36.8762, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 8.793333333333334e-06, |
|
"loss": 33.3106, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 8.737777777777778e-06, |
|
"loss": 34.9416, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 8.682222222222222e-06, |
|
"loss": 33.5493, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 8.626666666666668e-06, |
|
"loss": 35.4618, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 8.571111111111112e-06, |
|
"loss": 34.874, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 8.515555555555556e-06, |
|
"loss": 32.8104, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 8.46e-06, |
|
"loss": 31.1796, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 8.404444444444444e-06, |
|
"loss": 35.0476, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 8.34888888888889e-06, |
|
"loss": 31.6942, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 8.293333333333334e-06, |
|
"loss": 30.7686, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 8.237777777777778e-06, |
|
"loss": 24.2201, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 8.182222222222222e-06, |
|
"loss": 28.1463, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 8.126666666666668e-06, |
|
"loss": 24.295, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 8.071111111111112e-06, |
|
"loss": 23.8333, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 8.015555555555556e-06, |
|
"loss": 26.2859, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 7.960000000000002e-06, |
|
"loss": 25.793, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 7.904444444444444e-06, |
|
"loss": 26.0106, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 7.84888888888889e-06, |
|
"loss": 22.3316, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 7.793333333333334e-06, |
|
"loss": 25.0446, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 2.42, |
|
"learning_rate": 7.737777777777778e-06, |
|
"loss": 23.6882, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 7.682222222222224e-06, |
|
"loss": 25.7088, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"learning_rate": 7.626666666666668e-06, |
|
"loss": 24.0767, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 2.54, |
|
"learning_rate": 7.571111111111112e-06, |
|
"loss": 22.8335, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 2.58, |
|
"learning_rate": 7.515555555555556e-06, |
|
"loss": 21.7638, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 7.4600000000000006e-06, |
|
"loss": 20.7384, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 2.65, |
|
"learning_rate": 7.4044444444444455e-06, |
|
"loss": 23.4662, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 7.3488888888888895e-06, |
|
"loss": 23.6754, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 2.73, |
|
"learning_rate": 7.2933333333333335e-06, |
|
"loss": 20.6096, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 7.237777777777778e-06, |
|
"loss": 21.258, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 7.1822222222222224e-06, |
|
"loss": 23.6693, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 2.85, |
|
"learning_rate": 7.126666666666667e-06, |
|
"loss": 24.9978, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 7.071111111111112e-06, |
|
"loss": 22.2995, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 7.015555555555556e-06, |
|
"loss": 22.3644, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 2.97, |
|
"learning_rate": 6.96e-06, |
|
"loss": 21.1688, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"learning_rate": 6.904444444444444e-06, |
|
"loss": 19.0633, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 3.05, |
|
"learning_rate": 6.848888888888889e-06, |
|
"loss": 15.0111, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 6.793333333333334e-06, |
|
"loss": 15.5402, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 3.13, |
|
"learning_rate": 6.737777777777779e-06, |
|
"loss": 14.2953, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 3.17, |
|
"learning_rate": 6.682222222222223e-06, |
|
"loss": 15.0344, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 3.17, |
|
"eval_loss": 13.20719051361084, |
|
"eval_runtime": 6258.6543, |
|
"eval_samples_per_second": 0.806, |
|
"eval_steps_per_second": 0.101, |
|
"eval_wer": 144.3488521776443, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 3.21, |
|
"learning_rate": 6.626666666666667e-06, |
|
"loss": 16.1598, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 3.25, |
|
"learning_rate": 6.571111111111111e-06, |
|
"loss": 17.0358, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 3.29, |
|
"learning_rate": 6.515555555555556e-06, |
|
"loss": 16.2851, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 3.33, |
|
"learning_rate": 6.460000000000001e-06, |
|
"loss": 14.0717, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 3.37, |
|
"learning_rate": 6.404444444444446e-06, |
|
"loss": 15.4156, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 3.41, |
|
"learning_rate": 6.348888888888889e-06, |
|
"loss": 14.0956, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 3.45, |
|
"learning_rate": 6.293333333333334e-06, |
|
"loss": 14.6356, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 3.49, |
|
"learning_rate": 6.237777777777778e-06, |
|
"loss": 14.1948, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 3.53, |
|
"learning_rate": 6.182222222222223e-06, |
|
"loss": 14.6206, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 3.57, |
|
"learning_rate": 6.126666666666668e-06, |
|
"loss": 15.4542, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 3.61, |
|
"learning_rate": 6.0711111111111125e-06, |
|
"loss": 14.2712, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 3.65, |
|
"learning_rate": 6.015555555555556e-06, |
|
"loss": 15.7296, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 3.68, |
|
"learning_rate": 5.9600000000000005e-06, |
|
"loss": 13.1578, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 3.72, |
|
"learning_rate": 5.9044444444444446e-06, |
|
"loss": 13.4703, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 3.76, |
|
"learning_rate": 5.8488888888888895e-06, |
|
"loss": 12.8878, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 3.8, |
|
"learning_rate": 5.793333333333334e-06, |
|
"loss": 14.3464, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 3.84, |
|
"learning_rate": 5.737777777777778e-06, |
|
"loss": 13.4684, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 3.88, |
|
"learning_rate": 5.682222222222222e-06, |
|
"loss": 12.8048, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 3.92, |
|
"learning_rate": 5.626666666666667e-06, |
|
"loss": 13.5923, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 3.96, |
|
"learning_rate": 5.571111111111111e-06, |
|
"loss": 13.2108, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 5.515555555555556e-06, |
|
"loss": 12.2199, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 4.04, |
|
"learning_rate": 5.460000000000001e-06, |
|
"loss": 8.721, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 5.404444444444444e-06, |
|
"loss": 8.5954, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 4.12, |
|
"learning_rate": 5.348888888888889e-06, |
|
"loss": 8.8759, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 5.293333333333334e-06, |
|
"loss": 7.916, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 4.2, |
|
"learning_rate": 5.237777777777778e-06, |
|
"loss": 8.5696, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 4.24, |
|
"learning_rate": 5.182222222222223e-06, |
|
"loss": 8.1185, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 4.28, |
|
"learning_rate": 5.128888888888889e-06, |
|
"loss": 8.5324, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 4.32, |
|
"learning_rate": 5.073333333333334e-06, |
|
"loss": 8.1725, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 4.36, |
|
"learning_rate": 5.0177777777777785e-06, |
|
"loss": 8.3494, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 4.4, |
|
"learning_rate": 4.9622222222222225e-06, |
|
"loss": 8.0217, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 4.44, |
|
"learning_rate": 4.9066666666666666e-06, |
|
"loss": 8.8349, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 4.48, |
|
"learning_rate": 4.8511111111111114e-06, |
|
"loss": 7.6893, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 4.52, |
|
"learning_rate": 4.795555555555556e-06, |
|
"loss": 7.5403, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 4.56, |
|
"learning_rate": 4.74e-06, |
|
"loss": 8.2893, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 4.6, |
|
"learning_rate": 4.684444444444444e-06, |
|
"loss": 7.4306, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 4.64, |
|
"learning_rate": 4.628888888888889e-06, |
|
"loss": 8.2434, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 4.68, |
|
"learning_rate": 4.573333333333333e-06, |
|
"loss": 8.3564, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 4.71, |
|
"learning_rate": 4.517777777777778e-06, |
|
"loss": 8.1566, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 4.75, |
|
"learning_rate": 4.462222222222223e-06, |
|
"loss": 7.6075, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 4.75, |
|
"eval_loss": 5.894593238830566, |
|
"eval_runtime": 4130.5161, |
|
"eval_samples_per_second": 1.221, |
|
"eval_steps_per_second": 0.153, |
|
"eval_wer": 42.383608667667886, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 4.79, |
|
"learning_rate": 4.406666666666667e-06, |
|
"loss": 8.4462, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 4.83, |
|
"learning_rate": 4.351111111111111e-06, |
|
"loss": 7.9077, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 4.87, |
|
"learning_rate": 4.295555555555556e-06, |
|
"loss": 7.0686, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 4.91, |
|
"learning_rate": 4.24e-06, |
|
"loss": 6.9951, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 4.95, |
|
"learning_rate": 4.184444444444445e-06, |
|
"loss": 8.6228, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 4.99, |
|
"learning_rate": 4.12888888888889e-06, |
|
"loss": 7.4706, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 5.03, |
|
"learning_rate": 4.073333333333334e-06, |
|
"loss": 5.4224, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 5.07, |
|
"learning_rate": 4.017777777777778e-06, |
|
"loss": 4.8876, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 5.11, |
|
"learning_rate": 3.962222222222222e-06, |
|
"loss": 4.8291, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 5.15, |
|
"learning_rate": 3.906666666666667e-06, |
|
"loss": 5.6413, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 5.19, |
|
"learning_rate": 3.851111111111112e-06, |
|
"loss": 5.3154, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 5.23, |
|
"learning_rate": 3.7955555555555557e-06, |
|
"loss": 5.1792, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 5.27, |
|
"learning_rate": 3.74e-06, |
|
"loss": 4.6619, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 5.31, |
|
"learning_rate": 3.684444444444445e-06, |
|
"loss": 4.7989, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 5.35, |
|
"learning_rate": 3.628888888888889e-06, |
|
"loss": 4.6165, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 5.39, |
|
"learning_rate": 3.5733333333333336e-06, |
|
"loss": 4.812, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 5.43, |
|
"learning_rate": 3.5177777777777784e-06, |
|
"loss": 4.5789, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 5.47, |
|
"learning_rate": 3.4622222222222225e-06, |
|
"loss": 5.4889, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 5.51, |
|
"learning_rate": 3.406666666666667e-06, |
|
"loss": 4.7111, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 5.55, |
|
"learning_rate": 3.351111111111112e-06, |
|
"loss": 3.8969, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 5.59, |
|
"learning_rate": 3.295555555555556e-06, |
|
"loss": 4.9748, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 5.63, |
|
"learning_rate": 3.2400000000000003e-06, |
|
"loss": 4.6032, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 5.67, |
|
"learning_rate": 3.1844444444444444e-06, |
|
"loss": 4.4056, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 5.71, |
|
"learning_rate": 3.1288888888888892e-06, |
|
"loss": 4.3673, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 5.74, |
|
"learning_rate": 3.0733333333333337e-06, |
|
"loss": 4.433, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 5.78, |
|
"learning_rate": 3.0177777777777777e-06, |
|
"loss": 4.3417, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 5.82, |
|
"learning_rate": 2.9622222222222226e-06, |
|
"loss": 4.7957, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 5.86, |
|
"learning_rate": 2.906666666666667e-06, |
|
"loss": 4.6015, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 5.9, |
|
"learning_rate": 2.851111111111111e-06, |
|
"loss": 4.1873, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 5.94, |
|
"learning_rate": 2.795555555555556e-06, |
|
"loss": 3.2233, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 5.98, |
|
"learning_rate": 2.7400000000000004e-06, |
|
"loss": 3.7264, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 6.02, |
|
"learning_rate": 2.6844444444444445e-06, |
|
"loss": 2.7572, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 6.06, |
|
"learning_rate": 2.6288888888888894e-06, |
|
"loss": 2.911, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 6.1, |
|
"learning_rate": 2.573333333333334e-06, |
|
"loss": 2.3855, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 6.14, |
|
"learning_rate": 2.517777777777778e-06, |
|
"loss": 2.274, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 6.18, |
|
"learning_rate": 2.4622222222222223e-06, |
|
"loss": 2.4091, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 6.22, |
|
"learning_rate": 2.4066666666666668e-06, |
|
"loss": 2.8948, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 6.26, |
|
"learning_rate": 2.3511111111111112e-06, |
|
"loss": 2.8496, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 6.3, |
|
"learning_rate": 2.2955555555555557e-06, |
|
"loss": 2.7595, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 6.34, |
|
"learning_rate": 2.24e-06, |
|
"loss": 2.5225, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 6.34, |
|
"eval_loss": 2.0157761573791504, |
|
"eval_runtime": 4033.8091, |
|
"eval_samples_per_second": 1.25, |
|
"eval_steps_per_second": 0.156, |
|
"eval_wer": 19.54301652006007, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 6.38, |
|
"learning_rate": 2.1844444444444446e-06, |
|
"loss": 2.421, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 6.42, |
|
"learning_rate": 2.128888888888889e-06, |
|
"loss": 2.5867, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 6.46, |
|
"learning_rate": 2.0733333333333335e-06, |
|
"loss": 2.3535, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 6.5, |
|
"learning_rate": 2.017777777777778e-06, |
|
"loss": 2.1776, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 6.54, |
|
"learning_rate": 1.9622222222222224e-06, |
|
"loss": 2.1593, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 6.58, |
|
"learning_rate": 1.906666666666667e-06, |
|
"loss": 1.9538, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 6.62, |
|
"learning_rate": 1.8511111111111114e-06, |
|
"loss": 2.1601, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 6.66, |
|
"learning_rate": 1.7955555555555556e-06, |
|
"loss": 1.9078, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 6.7, |
|
"learning_rate": 1.74e-06, |
|
"loss": 1.9699, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 6.74, |
|
"learning_rate": 1.6844444444444447e-06, |
|
"loss": 2.0632, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 6.77, |
|
"learning_rate": 1.628888888888889e-06, |
|
"loss": 2.1845, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 6.81, |
|
"learning_rate": 1.5733333333333334e-06, |
|
"loss": 1.8937, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 6.85, |
|
"learning_rate": 1.5177777777777781e-06, |
|
"loss": 1.8585, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 6.89, |
|
"learning_rate": 1.4622222222222224e-06, |
|
"loss": 2.1634, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 6.93, |
|
"learning_rate": 1.4066666666666668e-06, |
|
"loss": 1.9549, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 6.97, |
|
"learning_rate": 1.351111111111111e-06, |
|
"loss": 1.9286, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 7.01, |
|
"learning_rate": 1.2955555555555557e-06, |
|
"loss": 1.3151, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 7.05, |
|
"learning_rate": 1.2400000000000002e-06, |
|
"loss": 1.0216, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 7.09, |
|
"learning_rate": 1.1844444444444447e-06, |
|
"loss": 0.9815, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 7.13, |
|
"learning_rate": 1.128888888888889e-06, |
|
"loss": 0.9841, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 7.17, |
|
"learning_rate": 1.0733333333333334e-06, |
|
"loss": 1.1402, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 7.21, |
|
"learning_rate": 1.0177777777777778e-06, |
|
"loss": 0.9043, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 7.25, |
|
"learning_rate": 9.622222222222223e-07, |
|
"loss": 0.9213, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 7.29, |
|
"learning_rate": 9.066666666666668e-07, |
|
"loss": 1.0344, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 7.33, |
|
"learning_rate": 8.511111111111112e-07, |
|
"loss": 1.0114, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 7.37, |
|
"learning_rate": 7.955555555555557e-07, |
|
"loss": 0.857, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 7.41, |
|
"learning_rate": 7.4e-07, |
|
"loss": 0.7789, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 7.45, |
|
"learning_rate": 6.844444444444446e-07, |
|
"loss": 0.787, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 7.49, |
|
"learning_rate": 6.288888888888889e-07, |
|
"loss": 0.9653, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 7.53, |
|
"learning_rate": 5.733333333333334e-07, |
|
"loss": 0.6772, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 7.57, |
|
"learning_rate": 5.177777777777778e-07, |
|
"loss": 0.6037, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 7.61, |
|
"learning_rate": 4.6222222222222225e-07, |
|
"loss": 0.8272, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 7.65, |
|
"learning_rate": 4.0666666666666666e-07, |
|
"loss": 0.7372, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 7.69, |
|
"learning_rate": 3.5111111111111117e-07, |
|
"loss": 0.8339, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 7.73, |
|
"learning_rate": 2.9555555555555557e-07, |
|
"loss": 0.6063, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 7.77, |
|
"learning_rate": 2.4000000000000003e-07, |
|
"loss": 0.6655, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 7.81, |
|
"learning_rate": 1.8444444444444446e-07, |
|
"loss": 0.5042, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 7.84, |
|
"learning_rate": 1.288888888888889e-07, |
|
"loss": 0.7717, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 7.88, |
|
"learning_rate": 7.333333333333334e-08, |
|
"loss": 0.4491, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 7.92, |
|
"learning_rate": 1.777777777777778e-08, |
|
"loss": 0.5364, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 7.92, |
|
"eval_loss": 0.2601960599422455, |
|
"eval_runtime": 3887.6238, |
|
"eval_samples_per_second": 1.297, |
|
"eval_steps_per_second": 0.162, |
|
"eval_wer": 5.357219480798112, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 7.92, |
|
"step": 5000, |
|
"total_flos": 1.15333155237888e+19, |
|
"train_loss": 14.672283467102051, |
|
"train_runtime": 56962.9565, |
|
"train_samples_per_second": 0.702, |
|
"train_steps_per_second": 0.088 |
|
} |
|
], |
|
"max_steps": 5000, |
|
"num_train_epochs": 8, |
|
"total_flos": 1.15333155237888e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|