|
{ |
|
"best_metric": 62.6897432259895, |
|
"best_model_checkpoint": "./checkpoint-5000", |
|
"epoch": 1.103, |
|
"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": 0.3482, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 9.400000000000001e-07, |
|
"loss": 0.309, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.44e-06, |
|
"loss": 0.2591, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 1.94e-06, |
|
"loss": 0.2917, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.4400000000000004e-06, |
|
"loss": 0.2494, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 2.9400000000000002e-06, |
|
"loss": 0.2889, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.44e-06, |
|
"loss": 0.2748, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 3.94e-06, |
|
"loss": 0.2483, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.440000000000001e-06, |
|
"loss": 0.2066, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 4.94e-06, |
|
"loss": 0.2021, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.4400000000000004e-06, |
|
"loss": 0.278, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 5.94e-06, |
|
"loss": 0.1931, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.440000000000001e-06, |
|
"loss": 0.2872, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 6.9400000000000005e-06, |
|
"loss": 0.2465, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 7.420000000000001e-06, |
|
"loss": 0.2475, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 7.92e-06, |
|
"loss": 0.2316, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 8.42e-06, |
|
"loss": 0.2259, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 8.920000000000001e-06, |
|
"loss": 0.2359, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.42e-06, |
|
"loss": 0.2914, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.920000000000002e-06, |
|
"loss": 0.2437, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.953333333333333e-06, |
|
"loss": 0.3181, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.89777777777778e-06, |
|
"loss": 0.3076, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.842222222222223e-06, |
|
"loss": 0.3039, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.786666666666667e-06, |
|
"loss": 0.2872, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.731111111111113e-06, |
|
"loss": 0.3232, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.675555555555555e-06, |
|
"loss": 0.2911, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.620000000000001e-06, |
|
"loss": 0.2918, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.564444444444445e-06, |
|
"loss": 0.2764, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.508888888888889e-06, |
|
"loss": 0.2656, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.453333333333335e-06, |
|
"loss": 0.3079, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.397777777777779e-06, |
|
"loss": 0.2563, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.342222222222223e-06, |
|
"loss": 0.1979, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.286666666666667e-06, |
|
"loss": 0.2031, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.231111111111111e-06, |
|
"loss": 0.2258, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.175555555555557e-06, |
|
"loss": 0.2336, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.12e-06, |
|
"loss": 0.2432, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.064444444444447e-06, |
|
"loss": 0.2779, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.008888888888889e-06, |
|
"loss": 0.2275, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 8.953333333333335e-06, |
|
"loss": 0.2701, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 8.897777777777779e-06, |
|
"loss": 0.2264, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"eval_loss": 0.31018269062042236, |
|
"eval_runtime": 6168.4947, |
|
"eval_samples_per_second": 0.746, |
|
"eval_steps_per_second": 0.746, |
|
"eval_wer": 79.35877429422614, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 8.842222222222223e-06, |
|
"loss": 0.215, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 8.786666666666668e-06, |
|
"loss": 0.2394, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 8.73111111111111e-06, |
|
"loss": 0.2623, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 8.675555555555556e-06, |
|
"loss": 0.2657, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 8.62e-06, |
|
"loss": 0.3247, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 8.564444444444445e-06, |
|
"loss": 0.2637, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 8.50888888888889e-06, |
|
"loss": 0.2622, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 8.453333333333334e-06, |
|
"loss": 0.2092, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 8.397777777777778e-06, |
|
"loss": 0.2378, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 8.342222222222222e-06, |
|
"loss": 0.2824, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.286666666666668e-06, |
|
"loss": 0.2592, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.231111111111112e-06, |
|
"loss": 0.2408, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 8.175555555555556e-06, |
|
"loss": 0.2474, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 8.120000000000002e-06, |
|
"loss": 0.2583, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 8.064444444444444e-06, |
|
"loss": 0.2186, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 8.00888888888889e-06, |
|
"loss": 0.2107, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 7.953333333333334e-06, |
|
"loss": 0.2439, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 7.897777777777778e-06, |
|
"loss": 0.2146, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 7.842222222222224e-06, |
|
"loss": 0.2741, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 7.786666666666666e-06, |
|
"loss": 0.1977, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 7.731111111111112e-06, |
|
"loss": 0.2165, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 7.675555555555556e-06, |
|
"loss": 0.2317, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 7.620000000000001e-06, |
|
"loss": 0.2177, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 7.564444444444446e-06, |
|
"loss": 0.2827, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 7.50888888888889e-06, |
|
"loss": 0.2084, |
|
"step": 1625 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 7.453333333333334e-06, |
|
"loss": 0.2599, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 7.3977777777777786e-06, |
|
"loss": 0.2752, |
|
"step": 1675 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 7.342222222222223e-06, |
|
"loss": 0.2562, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 7.2866666666666675e-06, |
|
"loss": 0.2396, |
|
"step": 1725 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 7.231111111111112e-06, |
|
"loss": 0.271, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 7.1755555555555556e-06, |
|
"loss": 0.265, |
|
"step": 1775 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 7.1200000000000004e-06, |
|
"loss": 0.2844, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 7.0644444444444445e-06, |
|
"loss": 0.3365, |
|
"step": 1825 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 7.008888888888889e-06, |
|
"loss": 0.3543, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 6.953333333333334e-06, |
|
"loss": 0.3603, |
|
"step": 1875 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 6.897777777777779e-06, |
|
"loss": 0.3471, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 6.842222222222222e-06, |
|
"loss": 0.3541, |
|
"step": 1925 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 6.786666666666667e-06, |
|
"loss": 0.2702, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 6.731111111111111e-06, |
|
"loss": 0.2737, |
|
"step": 1975 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 6.675555555555556e-06, |
|
"loss": 0.3195, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"eval_loss": 0.28298914432525635, |
|
"eval_runtime": 6173.4536, |
|
"eval_samples_per_second": 0.746, |
|
"eval_steps_per_second": 0.746, |
|
"eval_wer": 78.19548872180451, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 6.620000000000001e-06, |
|
"loss": 0.3136, |
|
"step": 2025 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 6.564444444444446e-06, |
|
"loss": 0.3176, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 6.508888888888889e-06, |
|
"loss": 0.2692, |
|
"step": 2075 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 6.453333333333334e-06, |
|
"loss": 0.2629, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 6.397777777777778e-06, |
|
"loss": 0.3176, |
|
"step": 2125 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 6.342222222222223e-06, |
|
"loss": 0.2751, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 6.286666666666668e-06, |
|
"loss": 0.3283, |
|
"step": 2175 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 6.231111111111111e-06, |
|
"loss": 0.3355, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 6.175555555555556e-06, |
|
"loss": 0.3352, |
|
"step": 2225 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 6.120000000000001e-06, |
|
"loss": 0.3002, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 6.064444444444445e-06, |
|
"loss": 0.4066, |
|
"step": 2275 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 6.00888888888889e-06, |
|
"loss": 0.3046, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 5.9533333333333345e-06, |
|
"loss": 0.4013, |
|
"step": 2325 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 5.897777777777778e-06, |
|
"loss": 0.296, |
|
"step": 2350 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 5.8422222222222226e-06, |
|
"loss": 0.312, |
|
"step": 2375 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 5.7866666666666674e-06, |
|
"loss": 0.2965, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 5.7311111111111115e-06, |
|
"loss": 0.406, |
|
"step": 2425 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 5.675555555555556e-06, |
|
"loss": 0.376, |
|
"step": 2450 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 5.620000000000001e-06, |
|
"loss": 0.3161, |
|
"step": 2475 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 5.5644444444444444e-06, |
|
"loss": 0.3798, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 5.508888888888889e-06, |
|
"loss": 0.2596, |
|
"step": 2525 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 5.453333333333334e-06, |
|
"loss": 0.3665, |
|
"step": 2550 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 5.397777777777778e-06, |
|
"loss": 0.4132, |
|
"step": 2575 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 5.342222222222223e-06, |
|
"loss": 0.2845, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 5.286666666666666e-06, |
|
"loss": 0.4019, |
|
"step": 2625 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 5.231111111111111e-06, |
|
"loss": 0.3766, |
|
"step": 2650 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 5.175555555555556e-06, |
|
"loss": 0.3659, |
|
"step": 2675 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 5.12e-06, |
|
"loss": 0.4297, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 5.064444444444445e-06, |
|
"loss": 0.3215, |
|
"step": 2725 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 5.00888888888889e-06, |
|
"loss": 0.457, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.953333333333334e-06, |
|
"loss": 0.3057, |
|
"step": 2775 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.897777777777778e-06, |
|
"loss": 0.3736, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.842222222222223e-06, |
|
"loss": 0.3088, |
|
"step": 2825 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.786666666666667e-06, |
|
"loss": 0.3513, |
|
"step": 2850 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.731111111111112e-06, |
|
"loss": 0.3594, |
|
"step": 2875 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 4.675555555555556e-06, |
|
"loss": 0.3347, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 4.620000000000001e-06, |
|
"loss": 0.3025, |
|
"step": 2925 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 4.564444444444445e-06, |
|
"loss": 0.2782, |
|
"step": 2950 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 4.50888888888889e-06, |
|
"loss": 0.3664, |
|
"step": 2975 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 4.453333333333334e-06, |
|
"loss": 0.3905, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"eval_loss": 0.25082916021347046, |
|
"eval_runtime": 6168.713, |
|
"eval_samples_per_second": 0.746, |
|
"eval_steps_per_second": 0.746, |
|
"eval_wer": 72.9181444176479, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 4.397777777777778e-06, |
|
"loss": 0.382, |
|
"step": 3025 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 4.3422222222222225e-06, |
|
"loss": 0.3521, |
|
"step": 3050 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 4.2866666666666666e-06, |
|
"loss": 0.2286, |
|
"step": 3075 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 4.2311111111111114e-06, |
|
"loss": 0.3386, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 4.175555555555556e-06, |
|
"loss": 0.2871, |
|
"step": 3125 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 4.12e-06, |
|
"loss": 0.3036, |
|
"step": 3150 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 4.064444444444444e-06, |
|
"loss": 0.3341, |
|
"step": 3175 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 4.008888888888889e-06, |
|
"loss": 0.2758, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 3.953333333333333e-06, |
|
"loss": 0.2324, |
|
"step": 3225 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 3.897777777777778e-06, |
|
"loss": 0.3621, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 3.844444444444445e-06, |
|
"loss": 0.3616, |
|
"step": 3275 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 3.7888888888888893e-06, |
|
"loss": 0.2718, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 3.7333333333333337e-06, |
|
"loss": 0.3232, |
|
"step": 3325 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 3.6777777777777778e-06, |
|
"loss": 0.3179, |
|
"step": 3350 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 3.6222222222222226e-06, |
|
"loss": 0.2973, |
|
"step": 3375 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 3.566666666666667e-06, |
|
"loss": 0.2818, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 3.511111111111111e-06, |
|
"loss": 0.2213, |
|
"step": 3425 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 3.455555555555556e-06, |
|
"loss": 0.2455, |
|
"step": 3450 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 3.4000000000000005e-06, |
|
"loss": 0.2238, |
|
"step": 3475 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 3.3444444444444445e-06, |
|
"loss": 0.2123, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 3.2888888888888894e-06, |
|
"loss": 0.3331, |
|
"step": 3525 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 3.2333333333333334e-06, |
|
"loss": 0.2183, |
|
"step": 3550 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 3.177777777777778e-06, |
|
"loss": 0.2723, |
|
"step": 3575 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.1222222222222228e-06, |
|
"loss": 0.2617, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 3.066666666666667e-06, |
|
"loss": 0.3038, |
|
"step": 3625 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.0111111111111113e-06, |
|
"loss": 0.2658, |
|
"step": 3650 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 2.955555555555556e-06, |
|
"loss": 0.2076, |
|
"step": 3675 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 2.9e-06, |
|
"loss": 0.2348, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 2.8444444444444446e-06, |
|
"loss": 0.2532, |
|
"step": 3725 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 2.788888888888889e-06, |
|
"loss": 0.2042, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.7333333333333336e-06, |
|
"loss": 0.2718, |
|
"step": 3775 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.677777777777778e-06, |
|
"loss": 0.252, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.6222222222222225e-06, |
|
"loss": 0.2803, |
|
"step": 3825 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.566666666666667e-06, |
|
"loss": 0.2946, |
|
"step": 3850 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.5111111111111114e-06, |
|
"loss": 0.2075, |
|
"step": 3875 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.455555555555556e-06, |
|
"loss": 0.2604, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.4000000000000003e-06, |
|
"loss": 0.2709, |
|
"step": 3925 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.3444444444444448e-06, |
|
"loss": 0.2368, |
|
"step": 3950 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.2888888888888892e-06, |
|
"loss": 0.1744, |
|
"step": 3975 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.2333333333333333e-06, |
|
"loss": 0.2478, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"eval_loss": 0.2407277226448059, |
|
"eval_runtime": 6088.2767, |
|
"eval_samples_per_second": 0.756, |
|
"eval_steps_per_second": 0.756, |
|
"eval_wer": 68.84664491417223, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.1777777777777777e-06, |
|
"loss": 0.2375, |
|
"step": 4025 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.1222222222222226e-06, |
|
"loss": 0.1942, |
|
"step": 4050 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.0666666666666666e-06, |
|
"loss": 0.2232, |
|
"step": 4075 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.011111111111111e-06, |
|
"loss": 0.2589, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.955555555555556e-06, |
|
"loss": 0.2511, |
|
"step": 4125 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.9000000000000002e-06, |
|
"loss": 0.1862, |
|
"step": 4150 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 1.8444444444444445e-06, |
|
"loss": 0.2908, |
|
"step": 4175 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.788888888888889e-06, |
|
"loss": 0.2375, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.7333333333333336e-06, |
|
"loss": 0.1627, |
|
"step": 4225 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.6777777777777779e-06, |
|
"loss": 0.1947, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.6222222222222223e-06, |
|
"loss": 0.2152, |
|
"step": 4275 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.566666666666667e-06, |
|
"loss": 0.192, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.5111111111111112e-06, |
|
"loss": 0.2595, |
|
"step": 4325 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.4555555555555557e-06, |
|
"loss": 0.2527, |
|
"step": 4350 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.4000000000000001e-06, |
|
"loss": 0.2079, |
|
"step": 4375 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.3444444444444446e-06, |
|
"loss": 0.2328, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.288888888888889e-06, |
|
"loss": 0.1896, |
|
"step": 4425 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.2333333333333335e-06, |
|
"loss": 0.2328, |
|
"step": 4450 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 1.1777777777777778e-06, |
|
"loss": 0.224, |
|
"step": 4475 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 1.1222222222222222e-06, |
|
"loss": 0.18, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 1.066666666666667e-06, |
|
"loss": 0.1317, |
|
"step": 4525 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 1.0111111111111111e-06, |
|
"loss": 0.103, |
|
"step": 4550 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 9.555555555555556e-07, |
|
"loss": 0.1106, |
|
"step": 4575 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 9.000000000000001e-07, |
|
"loss": 0.114, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 8.444444444444445e-07, |
|
"loss": 0.1121, |
|
"step": 4625 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 7.888888888888889e-07, |
|
"loss": 0.1263, |
|
"step": 4650 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 7.333333333333334e-07, |
|
"loss": 0.1342, |
|
"step": 4675 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 6.777777777777779e-07, |
|
"loss": 0.0707, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 6.222222222222223e-07, |
|
"loss": 0.0887, |
|
"step": 4725 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 5.666666666666667e-07, |
|
"loss": 0.1056, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 5.111111111111112e-07, |
|
"loss": 0.0806, |
|
"step": 4775 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 4.5555555555555563e-07, |
|
"loss": 0.0775, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 4.0000000000000003e-07, |
|
"loss": 0.0878, |
|
"step": 4825 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 3.444444444444445e-07, |
|
"loss": 0.1029, |
|
"step": 4850 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 2.888888888888889e-07, |
|
"loss": 0.076, |
|
"step": 4875 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 2.3333333333333336e-07, |
|
"loss": 0.0924, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 1.777777777777778e-07, |
|
"loss": 0.0959, |
|
"step": 4925 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 1.2222222222222225e-07, |
|
"loss": 0.0648, |
|
"step": 4950 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 6.666666666666668e-08, |
|
"loss": 0.0603, |
|
"step": 4975 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 1.1111111111111112e-08, |
|
"loss": 0.0922, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"eval_loss": 0.2164831906557083, |
|
"eval_runtime": 6141.6159, |
|
"eval_samples_per_second": 0.75, |
|
"eval_steps_per_second": 0.75, |
|
"eval_wer": 62.6897432259895, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"step": 5000, |
|
"total_flos": 1.02060490752e+19, |
|
"train_loss": 0.2594885242700577, |
|
"train_runtime": 35915.1077, |
|
"train_samples_per_second": 0.278, |
|
"train_steps_per_second": 0.139 |
|
} |
|
], |
|
"max_steps": 5000, |
|
"num_train_epochs": 9223372036854775807, |
|
"total_flos": 1.02060490752e+19, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|