| { | |
| "best_metric": 32.89771261927907, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 4.784688995215311, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4e-07, | |
| "loss": 2.3155, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 2.1634, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.9426, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.7462, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.6197, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 1.4952, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.44e-06, | |
| "loss": 1.4331, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.94e-06, | |
| "loss": 1.371, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 1.3486, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.94e-06, | |
| "loss": 1.3221, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 1.2895, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.94e-06, | |
| "loss": 1.2749, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 1.2541, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 1.2288, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 1.1923, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.94e-06, | |
| "loss": 1.1419, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.44e-06, | |
| "loss": 1.0368, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.9198, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.7921, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.6873, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.6003, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.5659, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.5085, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.4888, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.4533, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.4459, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.4343, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.4404, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.4121, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.3909, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.4023, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.3668, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.3935, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.3603, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3692, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.3575, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.3669, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.3523, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.3386, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.3362, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "eval_loss": 0.35355815291404724, | |
| "eval_runtime": 949.8137, | |
| "eval_samples_per_second": 8.794, | |
| "eval_steps_per_second": 1.1, | |
| "eval_wer": 45.08599996032297, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.3308, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.3359, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.3072, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.3012, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.2906, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.3154, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.2893, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.2936, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.2903, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.2935, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.2876, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.2741, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.2853, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.2831, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.294, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.2734, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.2787, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.279, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.2719, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.2849, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.275, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.2614, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.2723, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.2634, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.276, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.2654, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.2609, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.265, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.2749, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.2605, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.2596, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.2596, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.2608, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.2568, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.2578, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.2522, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.2633, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.2447, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.25, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.2395, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_loss": 0.2744958996772766, | |
| "eval_runtime": 931.5383, | |
| "eval_samples_per_second": 8.967, | |
| "eval_steps_per_second": 1.122, | |
| "eval_wer": 37.171424603725676, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.2532, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.564444444444446e-06, | |
| "loss": 0.2471, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.508888888888889e-06, | |
| "loss": 0.2449, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.453333333333334e-06, | |
| "loss": 0.2472, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.397777777777778e-06, | |
| "loss": 0.22, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.342222222222223e-06, | |
| "loss": 0.2187, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.286666666666668e-06, | |
| "loss": 0.2188, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.231111111111111e-06, | |
| "loss": 0.2214, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.175555555555556e-06, | |
| "loss": 0.2182, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 6.120000000000001e-06, | |
| "loss": 0.2178, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 6.064444444444445e-06, | |
| "loss": 0.217, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.00888888888889e-06, | |
| "loss": 0.2222, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.9533333333333345e-06, | |
| "loss": 0.2263, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.897777777777778e-06, | |
| "loss": 0.2221, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.8422222222222226e-06, | |
| "loss": 0.208, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.7866666666666674e-06, | |
| "loss": 0.2225, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5.7311111111111115e-06, | |
| "loss": 0.2034, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5.675555555555556e-06, | |
| "loss": 0.2089, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.2154, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.2087, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.2109, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.2184, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.2112, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.2139, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.206, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.2118, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.2143, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.2041, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.2007, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.2151, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.2073, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.2161, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.2189, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.2023, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.2038, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.2053, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.2069, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.2017, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.2052, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.205, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "eval_loss": 0.24848520755767822, | |
| "eval_runtime": 930.6976, | |
| "eval_samples_per_second": 8.975, | |
| "eval_steps_per_second": 1.123, | |
| "eval_wer": 34.73525502410379, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.2092, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.2108, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.204, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.2009, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.21, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.199, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.1826, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.1816, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.1781, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.1895, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.1844, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.1859, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.1903, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.1804, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.1807, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.1873, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.1868, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.1804, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.1788, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.172, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.1791, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.1796, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.1741, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.1922, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.1901, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.1825, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.175, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.1781, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.1747, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.1771, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.1809, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.1836, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.1753, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.174, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.1809, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.1792, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.1788, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.1814, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.189, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.1795, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "eval_loss": 0.23519772291183472, | |
| "eval_runtime": 929.4402, | |
| "eval_samples_per_second": 8.987, | |
| "eval_steps_per_second": 1.124, | |
| "eval_wer": 33.24687047433888, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.1722, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.1846, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.1692, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.1884, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.1737, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.1821, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.1789, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.1666, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.1689, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.1652, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.1662, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.1564, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.168, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.1586, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.1628, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.1553, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.1617, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.1652, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.1612, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.1657, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.1685, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.1591, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.1569, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.1722, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.1668, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.1631, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.1522, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.1684, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.1607, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.18, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.169, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.1683, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.1615, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.1698, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.1699, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.1599, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.1659, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.1704, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.1618, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.1578, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "eval_loss": 0.23136980831623077, | |
| "eval_runtime": 935.756, | |
| "eval_samples_per_second": 8.926, | |
| "eval_steps_per_second": 1.117, | |
| "eval_wer": 32.89771261927907, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "step": 5000, | |
| "total_flos": 1.96813215203328e+18, | |
| "train_loss": 0.3487478602409363, | |
| "train_runtime": 6610.9497, | |
| "train_samples_per_second": 12.101, | |
| "train_steps_per_second": 0.756 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.96813215203328e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |