{ "best_metric": 8.282966640983934, "best_model_checkpoint": "./checkpoint-18000", "epoch": 1.183, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.4e-07, "loss": 1.1882, "step": 25 }, { "epoch": 0.0, "learning_rate": 9.400000000000001e-07, "loss": 1.0714, "step": 50 }, { "epoch": 0.0, "learning_rate": 1.44e-06, "loss": 0.6108, "step": 75 }, { "epoch": 0.01, "learning_rate": 1.94e-06, "loss": 0.3702, "step": 100 }, { "epoch": 0.01, "learning_rate": 2.42e-06, "loss": 0.3698, "step": 125 }, { "epoch": 0.01, "learning_rate": 2.92e-06, "loss": 0.2707, "step": 150 }, { "epoch": 0.01, "learning_rate": 3.4200000000000007e-06, "loss": 0.2701, "step": 175 }, { "epoch": 0.01, "learning_rate": 3.920000000000001e-06, "loss": 0.3283, "step": 200 }, { "epoch": 0.01, "learning_rate": 4.42e-06, "loss": 0.2796, "step": 225 }, { "epoch": 0.01, "learning_rate": 4.92e-06, "loss": 0.3132, "step": 250 }, { "epoch": 0.01, "learning_rate": 5.420000000000001e-06, "loss": 0.3, "step": 275 }, { "epoch": 0.01, "learning_rate": 5.92e-06, "loss": 0.2772, "step": 300 }, { "epoch": 0.02, "learning_rate": 6.42e-06, "loss": 0.266, "step": 325 }, { "epoch": 0.02, "learning_rate": 6.92e-06, "loss": 0.2774, "step": 350 }, { "epoch": 0.02, "learning_rate": 7.420000000000001e-06, "loss": 0.2049, "step": 375 }, { "epoch": 0.02, "learning_rate": 7.92e-06, "loss": 0.3264, "step": 400 }, { "epoch": 0.02, "learning_rate": 8.42e-06, "loss": 0.2724, "step": 425 }, { "epoch": 0.02, "learning_rate": 8.920000000000001e-06, "loss": 0.2796, "step": 450 }, { "epoch": 0.02, "learning_rate": 9.42e-06, "loss": 0.302, "step": 475 }, { "epoch": 0.03, "learning_rate": 9.920000000000002e-06, "loss": 0.3252, "step": 500 }, { "epoch": 0.03, "learning_rate": 9.98923076923077e-06, "loss": 0.323, "step": 525 }, { "epoch": 0.03, "learning_rate": 9.976410256410257e-06, "loss": 0.3206, "step": 550 }, { "epoch": 0.03, "learning_rate": 9.963589743589744e-06, "loss": 0.3323, "step": 575 }, { "epoch": 0.03, "learning_rate": 9.950769230769232e-06, "loss": 0.3674, "step": 600 }, { "epoch": 0.03, "learning_rate": 9.937948717948719e-06, "loss": 0.3344, "step": 625 }, { "epoch": 0.03, "learning_rate": 9.925128205128206e-06, "loss": 0.3539, "step": 650 }, { "epoch": 0.03, "learning_rate": 9.912307692307693e-06, "loss": 0.3085, "step": 675 }, { "epoch": 0.04, "learning_rate": 9.899487179487181e-06, "loss": 0.299, "step": 700 }, { "epoch": 0.04, "learning_rate": 9.886666666666668e-06, "loss": 0.3678, "step": 725 }, { "epoch": 0.04, "learning_rate": 9.873846153846155e-06, "loss": 0.3156, "step": 750 }, { "epoch": 0.04, "learning_rate": 9.861025641025642e-06, "loss": 0.3126, "step": 775 }, { "epoch": 0.04, "learning_rate": 9.84820512820513e-06, "loss": 0.3105, "step": 800 }, { "epoch": 0.04, "learning_rate": 9.835384615384617e-06, "loss": 0.3582, "step": 825 }, { "epoch": 0.04, "learning_rate": 9.822564102564104e-06, "loss": 0.2785, "step": 850 }, { "epoch": 0.04, "learning_rate": 9.80974358974359e-06, "loss": 0.3243, "step": 875 }, { "epoch": 0.04, "learning_rate": 9.796923076923077e-06, "loss": 0.3486, "step": 900 }, { "epoch": 0.05, "learning_rate": 9.784102564102564e-06, "loss": 0.3677, "step": 925 }, { "epoch": 0.05, "learning_rate": 9.771282051282051e-06, "loss": 0.3743, "step": 950 }, { "epoch": 0.05, "learning_rate": 9.75846153846154e-06, "loss": 0.4701, "step": 975 }, { "epoch": 0.05, "learning_rate": 9.745641025641026e-06, "loss": 0.318, "step": 1000 }, { "epoch": 0.05, "learning_rate": 9.732820512820513e-06, "loss": 0.2627, "step": 1025 }, { "epoch": 0.05, "learning_rate": 9.72e-06, "loss": 0.3449, "step": 1050 }, { "epoch": 0.05, "learning_rate": 9.707179487179487e-06, "loss": 0.3842, "step": 1075 }, { "epoch": 0.06, "learning_rate": 9.694358974358975e-06, "loss": 0.2295, "step": 1100 }, { "epoch": 0.06, "learning_rate": 9.681538461538462e-06, "loss": 0.2823, "step": 1125 }, { "epoch": 0.06, "learning_rate": 9.668717948717949e-06, "loss": 0.421, "step": 1150 }, { "epoch": 0.06, "learning_rate": 9.655897435897436e-06, "loss": 0.3193, "step": 1175 }, { "epoch": 0.06, "learning_rate": 9.643076923076924e-06, "loss": 0.3343, "step": 1200 }, { "epoch": 0.06, "learning_rate": 9.630256410256411e-06, "loss": 0.2728, "step": 1225 }, { "epoch": 0.06, "learning_rate": 9.617435897435898e-06, "loss": 0.2823, "step": 1250 }, { "epoch": 0.06, "learning_rate": 9.604615384615385e-06, "loss": 0.2736, "step": 1275 }, { "epoch": 0.07, "learning_rate": 9.591794871794873e-06, "loss": 0.2643, "step": 1300 }, { "epoch": 0.07, "learning_rate": 9.57897435897436e-06, "loss": 0.2139, "step": 1325 }, { "epoch": 0.07, "learning_rate": 9.566153846153847e-06, "loss": 0.2481, "step": 1350 }, { "epoch": 0.07, "learning_rate": 9.553333333333334e-06, "loss": 0.3043, "step": 1375 }, { "epoch": 0.07, "learning_rate": 9.540512820512822e-06, "loss": 0.2435, "step": 1400 }, { "epoch": 0.07, "learning_rate": 9.52769230769231e-06, "loss": 0.3249, "step": 1425 }, { "epoch": 0.07, "learning_rate": 9.514871794871796e-06, "loss": 0.3418, "step": 1450 }, { "epoch": 0.07, "learning_rate": 9.502051282051283e-06, "loss": 0.357, "step": 1475 }, { "epoch": 0.07, "learning_rate": 9.48923076923077e-06, "loss": 0.287, "step": 1500 }, { "epoch": 0.08, "learning_rate": 9.476410256410257e-06, "loss": 0.3519, "step": 1525 }, { "epoch": 0.08, "learning_rate": 9.463589743589743e-06, "loss": 0.3165, "step": 1550 }, { "epoch": 0.08, "learning_rate": 9.450769230769232e-06, "loss": 0.2987, "step": 1575 }, { "epoch": 0.08, "learning_rate": 9.437948717948719e-06, "loss": 0.3239, "step": 1600 }, { "epoch": 0.08, "learning_rate": 9.425128205128206e-06, "loss": 0.2594, "step": 1625 }, { "epoch": 0.08, "learning_rate": 9.412307692307692e-06, "loss": 0.2979, "step": 1650 }, { "epoch": 0.08, "learning_rate": 9.39948717948718e-06, "loss": 0.2895, "step": 1675 }, { "epoch": 0.09, "learning_rate": 9.386666666666668e-06, "loss": 0.375, "step": 1700 }, { "epoch": 0.09, "learning_rate": 9.373846153846155e-06, "loss": 0.3645, "step": 1725 }, { "epoch": 0.09, "learning_rate": 9.361025641025641e-06, "loss": 0.2496, "step": 1750 }, { "epoch": 0.09, "learning_rate": 9.348717948717949e-06, "loss": 0.3615, "step": 1775 }, { "epoch": 0.09, "learning_rate": 9.335897435897436e-06, "loss": 0.2936, "step": 1800 }, { "epoch": 0.09, "learning_rate": 9.323076923076924e-06, "loss": 0.3259, "step": 1825 }, { "epoch": 0.09, "learning_rate": 9.310256410256411e-06, "loss": 0.3738, "step": 1850 }, { "epoch": 0.09, "learning_rate": 9.297435897435898e-06, "loss": 0.3057, "step": 1875 }, { "epoch": 0.1, "learning_rate": 9.284615384615385e-06, "loss": 0.2409, "step": 1900 }, { "epoch": 0.1, "learning_rate": 9.271794871794873e-06, "loss": 0.323, "step": 1925 }, { "epoch": 0.1, "learning_rate": 9.25897435897436e-06, "loss": 0.3082, "step": 1950 }, { "epoch": 0.1, "learning_rate": 9.246153846153847e-06, "loss": 0.3518, "step": 1975 }, { "epoch": 0.1, "learning_rate": 9.233333333333334e-06, "loss": 0.2652, "step": 2000 }, { "epoch": 0.1, "eval_loss": 0.34685784578323364, "eval_runtime": 19263.2384, "eval_samples_per_second": 0.848, "eval_steps_per_second": 0.848, "eval_wer": 15.353708301229918, "step": 2000 }, { "epoch": 0.1, "learning_rate": 9.220512820512822e-06, "loss": 0.3688, "step": 2025 }, { "epoch": 0.1, "learning_rate": 9.207692307692309e-06, "loss": 0.2243, "step": 2050 }, { "epoch": 0.1, "learning_rate": 9.194871794871796e-06, "loss": 0.2665, "step": 2075 }, { "epoch": 0.1, "learning_rate": 9.182051282051283e-06, "loss": 0.3534, "step": 2100 }, { "epoch": 0.11, "learning_rate": 9.169230769230771e-06, "loss": 0.343, "step": 2125 }, { "epoch": 0.11, "learning_rate": 9.156410256410258e-06, "loss": 0.2907, "step": 2150 }, { "epoch": 0.11, "learning_rate": 9.143589743589745e-06, "loss": 0.3294, "step": 2175 }, { "epoch": 0.11, "learning_rate": 9.130769230769232e-06, "loss": 0.3201, "step": 2200 }, { "epoch": 0.11, "learning_rate": 9.117948717948718e-06, "loss": 0.302, "step": 2225 }, { "epoch": 0.11, "learning_rate": 9.105128205128207e-06, "loss": 0.2924, "step": 2250 }, { "epoch": 0.11, "learning_rate": 9.092307692307692e-06, "loss": 0.292, "step": 2275 }, { "epoch": 0.12, "learning_rate": 9.07948717948718e-06, "loss": 0.3517, "step": 2300 }, { "epoch": 0.12, "learning_rate": 9.066666666666667e-06, "loss": 0.2605, "step": 2325 }, { "epoch": 0.12, "learning_rate": 9.053846153846154e-06, "loss": 0.3457, "step": 2350 }, { "epoch": 0.12, "learning_rate": 9.041025641025641e-06, "loss": 0.3062, "step": 2375 }, { "epoch": 0.12, "learning_rate": 9.028205128205128e-06, "loss": 0.3394, "step": 2400 }, { "epoch": 0.12, "learning_rate": 9.015384615384616e-06, "loss": 0.2356, "step": 2425 }, { "epoch": 0.12, "learning_rate": 9.002564102564103e-06, "loss": 0.2856, "step": 2450 }, { "epoch": 0.12, "learning_rate": 8.98974358974359e-06, "loss": 0.3532, "step": 2475 }, { "epoch": 0.12, "learning_rate": 8.976923076923077e-06, "loss": 0.343, "step": 2500 }, { "epoch": 0.13, "learning_rate": 8.964102564102565e-06, "loss": 0.3008, "step": 2525 }, { "epoch": 0.13, "learning_rate": 8.951282051282052e-06, "loss": 0.3335, "step": 2550 }, { "epoch": 0.13, "learning_rate": 8.938461538461539e-06, "loss": 0.2703, "step": 2575 }, { "epoch": 0.13, "learning_rate": 8.925641025641026e-06, "loss": 0.2936, "step": 2600 }, { "epoch": 0.13, "learning_rate": 8.912820512820514e-06, "loss": 0.3027, "step": 2625 }, { "epoch": 0.13, "learning_rate": 8.900000000000001e-06, "loss": 0.3208, "step": 2650 }, { "epoch": 0.13, "learning_rate": 8.887179487179488e-06, "loss": 0.2921, "step": 2675 }, { "epoch": 0.14, "learning_rate": 8.874358974358975e-06, "loss": 0.2825, "step": 2700 }, { "epoch": 0.14, "learning_rate": 8.861538461538463e-06, "loss": 0.3475, "step": 2725 }, { "epoch": 0.14, "learning_rate": 8.84871794871795e-06, "loss": 0.3619, "step": 2750 }, { "epoch": 0.14, "learning_rate": 8.835897435897437e-06, "loss": 0.2778, "step": 2775 }, { "epoch": 0.14, "learning_rate": 8.823076923076924e-06, "loss": 0.3371, "step": 2800 }, { "epoch": 0.14, "learning_rate": 8.81025641025641e-06, "loss": 0.3207, "step": 2825 }, { "epoch": 0.14, "learning_rate": 8.7974358974359e-06, "loss": 0.326, "step": 2850 }, { "epoch": 0.14, "learning_rate": 8.784615384615386e-06, "loss": 0.3371, "step": 2875 }, { "epoch": 0.14, "learning_rate": 8.771794871794873e-06, "loss": 0.3331, "step": 2900 }, { "epoch": 0.15, "learning_rate": 8.75897435897436e-06, "loss": 0.2987, "step": 2925 }, { "epoch": 0.15, "learning_rate": 8.746153846153846e-06, "loss": 0.2508, "step": 2950 }, { "epoch": 0.15, "learning_rate": 8.733333333333333e-06, "loss": 0.3177, "step": 2975 }, { "epoch": 0.15, "learning_rate": 8.72051282051282e-06, "loss": 0.2891, "step": 3000 }, { "epoch": 0.15, "learning_rate": 8.707692307692309e-06, "loss": 0.2744, "step": 3025 }, { "epoch": 0.15, "learning_rate": 8.694871794871795e-06, "loss": 0.2708, "step": 3050 }, { "epoch": 0.15, "learning_rate": 8.682051282051282e-06, "loss": 0.3248, "step": 3075 }, { "epoch": 0.15, "learning_rate": 8.669230769230769e-06, "loss": 0.352, "step": 3100 }, { "epoch": 0.16, "learning_rate": 8.656410256410258e-06, "loss": 0.258, "step": 3125 }, { "epoch": 0.16, "learning_rate": 8.643589743589744e-06, "loss": 0.2519, "step": 3150 }, { "epoch": 0.16, "learning_rate": 8.630769230769231e-06, "loss": 0.3251, "step": 3175 }, { "epoch": 0.16, "learning_rate": 8.617948717948718e-06, "loss": 0.2979, "step": 3200 }, { "epoch": 0.16, "learning_rate": 8.605128205128207e-06, "loss": 0.2821, "step": 3225 }, { "epoch": 0.16, "learning_rate": 8.592307692307693e-06, "loss": 0.27, "step": 3250 }, { "epoch": 0.16, "learning_rate": 8.57948717948718e-06, "loss": 0.2426, "step": 3275 }, { "epoch": 0.17, "learning_rate": 8.566666666666667e-06, "loss": 0.3178, "step": 3300 }, { "epoch": 0.17, "learning_rate": 8.553846153846156e-06, "loss": 0.3166, "step": 3325 }, { "epoch": 0.17, "learning_rate": 8.541025641025642e-06, "loss": 0.2844, "step": 3350 }, { "epoch": 0.17, "learning_rate": 8.52820512820513e-06, "loss": 0.3083, "step": 3375 }, { "epoch": 0.17, "learning_rate": 8.515384615384616e-06, "loss": 0.3709, "step": 3400 }, { "epoch": 0.17, "learning_rate": 8.502564102564103e-06, "loss": 0.3546, "step": 3425 }, { "epoch": 0.17, "learning_rate": 8.489743589743591e-06, "loss": 0.312, "step": 3450 }, { "epoch": 0.17, "learning_rate": 8.476923076923078e-06, "loss": 0.2522, "step": 3475 }, { "epoch": 0.17, "learning_rate": 8.464102564102565e-06, "loss": 0.2793, "step": 3500 }, { "epoch": 0.18, "learning_rate": 8.451282051282052e-06, "loss": 0.2263, "step": 3525 }, { "epoch": 0.18, "learning_rate": 8.438461538461539e-06, "loss": 0.3577, "step": 3550 }, { "epoch": 0.18, "learning_rate": 8.425641025641026e-06, "loss": 0.2815, "step": 3575 }, { "epoch": 0.18, "learning_rate": 8.412820512820512e-06, "loss": 0.2117, "step": 3600 }, { "epoch": 0.18, "learning_rate": 8.400000000000001e-06, "loss": 0.3198, "step": 3625 }, { "epoch": 0.18, "learning_rate": 8.387179487179488e-06, "loss": 0.3161, "step": 3650 }, { "epoch": 0.18, "learning_rate": 8.374358974358975e-06, "loss": 0.2585, "step": 3675 }, { "epoch": 0.18, "learning_rate": 8.361538461538461e-06, "loss": 0.3117, "step": 3700 }, { "epoch": 0.19, "learning_rate": 8.34871794871795e-06, "loss": 0.2335, "step": 3725 }, { "epoch": 0.19, "learning_rate": 8.335897435897437e-06, "loss": 0.3305, "step": 3750 }, { "epoch": 0.19, "learning_rate": 8.323076923076924e-06, "loss": 0.2932, "step": 3775 }, { "epoch": 0.19, "learning_rate": 8.31025641025641e-06, "loss": 0.2775, "step": 3800 }, { "epoch": 0.19, "learning_rate": 8.297435897435899e-06, "loss": 0.3081, "step": 3825 }, { "epoch": 0.19, "learning_rate": 8.284615384615386e-06, "loss": 0.2745, "step": 3850 }, { "epoch": 0.19, "learning_rate": 8.271794871794873e-06, "loss": 0.3643, "step": 3875 }, { "epoch": 0.2, "learning_rate": 8.25897435897436e-06, "loss": 0.1852, "step": 3900 }, { "epoch": 0.2, "learning_rate": 8.246153846153848e-06, "loss": 0.2124, "step": 3925 }, { "epoch": 0.2, "learning_rate": 8.233333333333335e-06, "loss": 0.3447, "step": 3950 }, { "epoch": 0.2, "learning_rate": 8.220512820512822e-06, "loss": 0.307, "step": 3975 }, { "epoch": 0.2, "learning_rate": 8.207692307692308e-06, "loss": 0.3273, "step": 4000 }, { "epoch": 0.2, "eval_loss": 0.31508541107177734, "eval_runtime": 18903.1169, "eval_samples_per_second": 0.864, "eval_steps_per_second": 0.864, "eval_wer": 14.11408840567303, "step": 4000 }, { "epoch": 0.2, "learning_rate": 8.194871794871795e-06, "loss": 0.3, "step": 4025 }, { "epoch": 0.2, "learning_rate": 8.182051282051284e-06, "loss": 0.3633, "step": 4050 }, { "epoch": 0.2, "learning_rate": 8.16923076923077e-06, "loss": 0.2742, "step": 4075 }, { "epoch": 0.2, "learning_rate": 8.156410256410257e-06, "loss": 0.2882, "step": 4100 }, { "epoch": 0.21, "learning_rate": 8.143589743589744e-06, "loss": 0.3191, "step": 4125 }, { "epoch": 0.21, "learning_rate": 8.130769230769231e-06, "loss": 0.2816, "step": 4150 }, { "epoch": 0.21, "learning_rate": 8.117948717948718e-06, "loss": 0.2698, "step": 4175 }, { "epoch": 0.21, "learning_rate": 8.105128205128205e-06, "loss": 0.3499, "step": 4200 }, { "epoch": 0.21, "learning_rate": 8.092307692307693e-06, "loss": 0.2725, "step": 4225 }, { "epoch": 0.21, "learning_rate": 8.07948717948718e-06, "loss": 0.2902, "step": 4250 }, { "epoch": 0.21, "learning_rate": 8.066666666666667e-06, "loss": 0.2639, "step": 4275 }, { "epoch": 0.21, "learning_rate": 8.053846153846154e-06, "loss": 0.2275, "step": 4300 }, { "epoch": 0.22, "learning_rate": 8.041025641025642e-06, "loss": 0.3251, "step": 4325 }, { "epoch": 0.22, "learning_rate": 8.028205128205129e-06, "loss": 0.291, "step": 4350 }, { "epoch": 0.22, "learning_rate": 8.015384615384616e-06, "loss": 0.3083, "step": 4375 }, { "epoch": 0.22, "learning_rate": 8.002564102564103e-06, "loss": 0.3732, "step": 4400 }, { "epoch": 0.22, "learning_rate": 7.989743589743591e-06, "loss": 0.3353, "step": 4425 }, { "epoch": 0.22, "learning_rate": 7.976923076923078e-06, "loss": 0.2133, "step": 4450 }, { "epoch": 0.22, "learning_rate": 7.964102564102565e-06, "loss": 0.3056, "step": 4475 }, { "epoch": 0.23, "learning_rate": 7.951282051282052e-06, "loss": 0.3075, "step": 4500 }, { "epoch": 0.23, "learning_rate": 7.93846153846154e-06, "loss": 0.275, "step": 4525 }, { "epoch": 0.23, "learning_rate": 7.925641025641027e-06, "loss": 0.2129, "step": 4550 }, { "epoch": 0.23, "learning_rate": 7.912820512820514e-06, "loss": 0.2684, "step": 4575 }, { "epoch": 0.23, "learning_rate": 7.9e-06, "loss": 0.3709, "step": 4600 }, { "epoch": 0.23, "learning_rate": 7.887179487179487e-06, "loss": 0.2442, "step": 4625 }, { "epoch": 0.23, "learning_rate": 7.874358974358976e-06, "loss": 0.2966, "step": 4650 }, { "epoch": 0.23, "learning_rate": 7.861538461538463e-06, "loss": 0.2635, "step": 4675 }, { "epoch": 0.23, "learning_rate": 7.84871794871795e-06, "loss": 0.2965, "step": 4700 }, { "epoch": 0.24, "learning_rate": 7.835897435897436e-06, "loss": 0.2666, "step": 4725 }, { "epoch": 0.24, "learning_rate": 7.823076923076923e-06, "loss": 0.2876, "step": 4750 }, { "epoch": 0.24, "learning_rate": 7.81025641025641e-06, "loss": 0.2022, "step": 4775 }, { "epoch": 0.24, "learning_rate": 7.797435897435897e-06, "loss": 0.2868, "step": 4800 }, { "epoch": 0.24, "learning_rate": 7.784615384615385e-06, "loss": 0.2606, "step": 4825 }, { "epoch": 0.24, "learning_rate": 7.771794871794872e-06, "loss": 0.1949, "step": 4850 }, { "epoch": 0.24, "learning_rate": 7.758974358974359e-06, "loss": 0.2721, "step": 4875 }, { "epoch": 0.24, "learning_rate": 7.746153846153846e-06, "loss": 0.2982, "step": 4900 }, { "epoch": 0.25, "learning_rate": 7.733333333333334e-06, "loss": 0.2637, "step": 4925 }, { "epoch": 0.25, "learning_rate": 7.720512820512821e-06, "loss": 0.263, "step": 4950 }, { "epoch": 0.25, "learning_rate": 7.707692307692308e-06, "loss": 0.3062, "step": 4975 }, { "epoch": 0.25, "learning_rate": 7.694871794871795e-06, "loss": 0.2615, "step": 5000 }, { "epoch": 0.25, "learning_rate": 7.682051282051283e-06, "loss": 0.2551, "step": 5025 }, { "epoch": 0.25, "learning_rate": 7.66923076923077e-06, "loss": 0.2219, "step": 5050 }, { "epoch": 0.25, "learning_rate": 7.656410256410257e-06, "loss": 0.2695, "step": 5075 }, { "epoch": 0.26, "learning_rate": 7.643589743589744e-06, "loss": 0.3119, "step": 5100 }, { "epoch": 0.26, "learning_rate": 7.630769230769232e-06, "loss": 0.2293, "step": 5125 }, { "epoch": 0.26, "learning_rate": 7.617948717948719e-06, "loss": 0.2916, "step": 5150 }, { "epoch": 0.26, "learning_rate": 7.605128205128206e-06, "loss": 0.2858, "step": 5175 }, { "epoch": 0.26, "learning_rate": 7.592307692307693e-06, "loss": 0.242, "step": 5200 }, { "epoch": 0.26, "learning_rate": 7.5794871794871805e-06, "loss": 0.2074, "step": 5225 }, { "epoch": 0.26, "learning_rate": 7.566666666666667e-06, "loss": 0.3358, "step": 5250 }, { "epoch": 0.26, "learning_rate": 7.553846153846155e-06, "loss": 0.2527, "step": 5275 }, { "epoch": 0.27, "learning_rate": 7.541025641025642e-06, "loss": 0.2718, "step": 5300 }, { "epoch": 0.27, "learning_rate": 7.528205128205129e-06, "loss": 0.1725, "step": 5325 }, { "epoch": 0.27, "learning_rate": 7.5153846153846155e-06, "loss": 0.2873, "step": 5350 }, { "epoch": 0.27, "learning_rate": 7.502564102564102e-06, "loss": 0.2617, "step": 5375 }, { "epoch": 0.27, "learning_rate": 7.48974358974359e-06, "loss": 0.299, "step": 5400 }, { "epoch": 0.27, "learning_rate": 7.476923076923077e-06, "loss": 0.2949, "step": 5425 }, { "epoch": 0.27, "learning_rate": 7.4641025641025645e-06, "loss": 0.2067, "step": 5450 }, { "epoch": 0.27, "learning_rate": 7.451282051282051e-06, "loss": 0.258, "step": 5475 }, { "epoch": 0.28, "learning_rate": 7.438461538461539e-06, "loss": 0.2933, "step": 5500 }, { "epoch": 0.28, "learning_rate": 7.425641025641026e-06, "loss": 0.3197, "step": 5525 }, { "epoch": 0.28, "learning_rate": 7.4128205128205135e-06, "loss": 0.2555, "step": 5550 }, { "epoch": 0.28, "learning_rate": 7.4e-06, "loss": 0.287, "step": 5575 }, { "epoch": 0.28, "learning_rate": 7.387179487179488e-06, "loss": 0.1872, "step": 5600 }, { "epoch": 0.28, "learning_rate": 7.374358974358975e-06, "loss": 0.2702, "step": 5625 }, { "epoch": 0.28, "learning_rate": 7.3615384615384625e-06, "loss": 0.2879, "step": 5650 }, { "epoch": 0.28, "learning_rate": 7.348717948717949e-06, "loss": 0.2249, "step": 5675 }, { "epoch": 0.28, "learning_rate": 7.335897435897437e-06, "loss": 0.2461, "step": 5700 }, { "epoch": 0.29, "learning_rate": 7.323076923076924e-06, "loss": 0.2731, "step": 5725 }, { "epoch": 0.29, "learning_rate": 7.3102564102564115e-06, "loss": 0.1719, "step": 5750 }, { "epoch": 0.29, "learning_rate": 7.297435897435898e-06, "loss": 0.2388, "step": 5775 }, { "epoch": 0.29, "learning_rate": 7.284615384615386e-06, "loss": 0.2547, "step": 5800 }, { "epoch": 0.29, "learning_rate": 7.271794871794873e-06, "loss": 0.3051, "step": 5825 }, { "epoch": 0.29, "learning_rate": 7.25897435897436e-06, "loss": 0.255, "step": 5850 }, { "epoch": 0.29, "learning_rate": 7.246153846153847e-06, "loss": 0.3546, "step": 5875 }, { "epoch": 0.29, "learning_rate": 7.233333333333334e-06, "loss": 0.3293, "step": 5900 }, { "epoch": 0.3, "learning_rate": 7.220512820512822e-06, "loss": 0.3334, "step": 5925 }, { "epoch": 0.3, "learning_rate": 7.207692307692308e-06, "loss": 0.2176, "step": 5950 }, { "epoch": 0.3, "learning_rate": 7.1948717948717946e-06, "loss": 0.3174, "step": 5975 }, { "epoch": 0.3, "learning_rate": 7.182051282051282e-06, "loss": 0.2696, "step": 6000 }, { "epoch": 0.3, "eval_loss": 0.29548051953315735, "eval_runtime": 18971.7289, "eval_samples_per_second": 0.861, "eval_steps_per_second": 0.861, "eval_wer": 13.247153497132095, "step": 6000 }, { "epoch": 0.3, "learning_rate": 7.169230769230769e-06, "loss": 0.2577, "step": 6025 }, { "epoch": 0.3, "learning_rate": 7.156410256410257e-06, "loss": 0.3044, "step": 6050 }, { "epoch": 0.3, "learning_rate": 7.1435897435897436e-06, "loss": 0.2737, "step": 6075 }, { "epoch": 0.3, "learning_rate": 7.130769230769231e-06, "loss": 0.2851, "step": 6100 }, { "epoch": 0.31, "learning_rate": 7.117948717948718e-06, "loss": 0.274, "step": 6125 }, { "epoch": 0.31, "learning_rate": 7.105128205128206e-06, "loss": 0.2255, "step": 6150 }, { "epoch": 0.31, "learning_rate": 7.092820512820513e-06, "loss": 0.2786, "step": 6175 }, { "epoch": 0.31, "learning_rate": 7.08e-06, "loss": 0.3505, "step": 6200 }, { "epoch": 0.31, "learning_rate": 7.067179487179488e-06, "loss": 0.1906, "step": 6225 }, { "epoch": 0.31, "learning_rate": 7.0543589743589745e-06, "loss": 0.2226, "step": 6250 }, { "epoch": 0.31, "learning_rate": 7.041538461538462e-06, "loss": 0.3104, "step": 6275 }, { "epoch": 0.32, "learning_rate": 7.028717948717949e-06, "loss": 0.2629, "step": 6300 }, { "epoch": 0.32, "learning_rate": 7.015897435897437e-06, "loss": 0.2858, "step": 6325 }, { "epoch": 0.32, "learning_rate": 7.0030769230769235e-06, "loss": 0.2368, "step": 6350 }, { "epoch": 0.32, "learning_rate": 6.990256410256411e-06, "loss": 0.196, "step": 6375 }, { "epoch": 0.32, "learning_rate": 6.977435897435898e-06, "loss": 0.2733, "step": 6400 }, { "epoch": 0.32, "learning_rate": 6.964615384615386e-06, "loss": 0.2483, "step": 6425 }, { "epoch": 0.32, "learning_rate": 6.9517948717948725e-06, "loss": 0.2746, "step": 6450 }, { "epoch": 0.32, "learning_rate": 6.938974358974359e-06, "loss": 0.2253, "step": 6475 }, { "epoch": 0.33, "learning_rate": 6.926153846153847e-06, "loss": 0.2607, "step": 6500 }, { "epoch": 0.33, "learning_rate": 6.913333333333334e-06, "loss": 0.2735, "step": 6525 }, { "epoch": 0.33, "learning_rate": 6.9005128205128214e-06, "loss": 0.2356, "step": 6550 }, { "epoch": 0.33, "learning_rate": 6.887692307692308e-06, "loss": 0.2034, "step": 6575 }, { "epoch": 0.33, "learning_rate": 6.874871794871796e-06, "loss": 0.2497, "step": 6600 }, { "epoch": 0.33, "learning_rate": 6.862051282051283e-06, "loss": 0.1989, "step": 6625 }, { "epoch": 0.33, "learning_rate": 6.8492307692307704e-06, "loss": 0.2692, "step": 6650 }, { "epoch": 0.33, "learning_rate": 6.836410256410257e-06, "loss": 0.257, "step": 6675 }, { "epoch": 0.34, "learning_rate": 6.823589743589745e-06, "loss": 0.2432, "step": 6700 }, { "epoch": 0.34, "learning_rate": 6.810769230769231e-06, "loss": 0.2754, "step": 6725 }, { "epoch": 0.34, "learning_rate": 6.797948717948718e-06, "loss": 0.2335, "step": 6750 }, { "epoch": 0.34, "learning_rate": 6.785128205128205e-06, "loss": 0.2459, "step": 6775 }, { "epoch": 0.34, "learning_rate": 6.772307692307692e-06, "loss": 0.1958, "step": 6800 }, { "epoch": 0.34, "learning_rate": 6.75948717948718e-06, "loss": 0.2715, "step": 6825 }, { "epoch": 0.34, "learning_rate": 6.746666666666667e-06, "loss": 0.2216, "step": 6850 }, { "epoch": 0.34, "learning_rate": 6.733846153846154e-06, "loss": 0.2473, "step": 6875 }, { "epoch": 0.34, "learning_rate": 6.721025641025641e-06, "loss": 0.2909, "step": 6900 }, { "epoch": 0.35, "learning_rate": 6.708205128205129e-06, "loss": 0.2528, "step": 6925 }, { "epoch": 0.35, "learning_rate": 6.695384615384616e-06, "loss": 0.1718, "step": 6950 }, { "epoch": 0.35, "learning_rate": 6.682564102564103e-06, "loss": 0.3335, "step": 6975 }, { "epoch": 0.35, "learning_rate": 6.66974358974359e-06, "loss": 0.2523, "step": 7000 }, { "epoch": 0.35, "learning_rate": 6.656923076923078e-06, "loss": 0.2021, "step": 7025 }, { "epoch": 0.35, "learning_rate": 6.644102564102565e-06, "loss": 0.2483, "step": 7050 }, { "epoch": 0.35, "learning_rate": 6.6312820512820515e-06, "loss": 0.2333, "step": 7075 }, { "epoch": 0.35, "learning_rate": 6.618461538461539e-06, "loss": 0.1491, "step": 7100 }, { "epoch": 0.36, "learning_rate": 6.605641025641026e-06, "loss": 0.2398, "step": 7125 }, { "epoch": 0.36, "learning_rate": 6.592820512820514e-06, "loss": 0.2222, "step": 7150 }, { "epoch": 0.36, "learning_rate": 6.5800000000000005e-06, "loss": 0.2181, "step": 7175 }, { "epoch": 0.36, "learning_rate": 6.567179487179488e-06, "loss": 0.1906, "step": 7200 }, { "epoch": 0.36, "learning_rate": 6.554358974358975e-06, "loss": 0.2525, "step": 7225 }, { "epoch": 0.36, "learning_rate": 6.541538461538463e-06, "loss": 0.2007, "step": 7250 }, { "epoch": 0.36, "learning_rate": 6.5287179487179495e-06, "loss": 0.242, "step": 7275 }, { "epoch": 0.36, "learning_rate": 6.515897435897437e-06, "loss": 0.2433, "step": 7300 }, { "epoch": 0.37, "learning_rate": 6.503076923076923e-06, "loss": 0.2736, "step": 7325 }, { "epoch": 0.37, "learning_rate": 6.49025641025641e-06, "loss": 0.2124, "step": 7350 }, { "epoch": 0.37, "learning_rate": 6.477435897435898e-06, "loss": 0.1979, "step": 7375 }, { "epoch": 0.37, "learning_rate": 6.4646153846153845e-06, "loss": 0.206, "step": 7400 }, { "epoch": 0.37, "learning_rate": 6.451794871794872e-06, "loss": 0.1941, "step": 7425 }, { "epoch": 0.37, "learning_rate": 6.438974358974359e-06, "loss": 0.2744, "step": 7450 }, { "epoch": 0.37, "learning_rate": 6.426153846153847e-06, "loss": 0.1961, "step": 7475 }, { "epoch": 0.38, "learning_rate": 6.4133333333333335e-06, "loss": 0.2508, "step": 7500 }, { "epoch": 0.38, "learning_rate": 6.400512820512821e-06, "loss": 0.2312, "step": 7525 }, { "epoch": 0.38, "learning_rate": 6.387692307692308e-06, "loss": 0.2264, "step": 7550 }, { "epoch": 0.38, "learning_rate": 6.374871794871796e-06, "loss": 0.2225, "step": 7575 }, { "epoch": 0.38, "learning_rate": 6.3620512820512825e-06, "loss": 0.2373, "step": 7600 }, { "epoch": 0.38, "learning_rate": 6.34923076923077e-06, "loss": 0.2378, "step": 7625 }, { "epoch": 0.38, "learning_rate": 6.336410256410257e-06, "loss": 0.2115, "step": 7650 }, { "epoch": 0.38, "learning_rate": 6.323589743589744e-06, "loss": 0.234, "step": 7675 }, { "epoch": 0.39, "learning_rate": 6.3107692307692315e-06, "loss": 0.2268, "step": 7700 }, { "epoch": 0.39, "learning_rate": 6.297948717948718e-06, "loss": 0.2476, "step": 7725 }, { "epoch": 0.39, "learning_rate": 6.285128205128206e-06, "loss": 0.1547, "step": 7750 }, { "epoch": 0.39, "learning_rate": 6.272307692307693e-06, "loss": 0.2108, "step": 7775 }, { "epoch": 0.39, "learning_rate": 6.2594871794871805e-06, "loss": 0.2428, "step": 7800 }, { "epoch": 0.39, "learning_rate": 6.246666666666667e-06, "loss": 0.2334, "step": 7825 }, { "epoch": 0.39, "learning_rate": 6.233846153846155e-06, "loss": 0.1947, "step": 7850 }, { "epoch": 0.39, "learning_rate": 6.221025641025642e-06, "loss": 0.2222, "step": 7875 }, { "epoch": 0.4, "learning_rate": 6.2082051282051294e-06, "loss": 0.204, "step": 7900 }, { "epoch": 0.4, "learning_rate": 6.195384615384616e-06, "loss": 0.1893, "step": 7925 }, { "epoch": 0.4, "learning_rate": 6.182564102564102e-06, "loss": 0.2714, "step": 7950 }, { "epoch": 0.4, "learning_rate": 6.16974358974359e-06, "loss": 0.2071, "step": 7975 }, { "epoch": 0.4, "learning_rate": 6.156923076923077e-06, "loss": 0.1725, "step": 8000 }, { "epoch": 0.4, "eval_loss": 0.2786870300769806, "eval_runtime": 18917.4176, "eval_samples_per_second": 0.864, "eval_steps_per_second": 0.864, "eval_wer": 11.683360442884457, "step": 8000 }, { "epoch": 0.4, "learning_rate": 6.144102564102564e-06, "loss": 0.1681, "step": 8025 }, { "epoch": 0.4, "learning_rate": 6.131282051282051e-06, "loss": 0.2166, "step": 8050 }, { "epoch": 0.4, "learning_rate": 6.118461538461539e-06, "loss": 0.2152, "step": 8075 }, { "epoch": 0.41, "learning_rate": 6.105641025641026e-06, "loss": 0.15, "step": 8100 }, { "epoch": 0.41, "learning_rate": 6.092820512820513e-06, "loss": 0.1511, "step": 8125 }, { "epoch": 0.41, "learning_rate": 6.08e-06, "loss": 0.1499, "step": 8150 }, { "epoch": 0.41, "learning_rate": 6.067179487179488e-06, "loss": 0.2434, "step": 8175 }, { "epoch": 0.41, "learning_rate": 6.054358974358975e-06, "loss": 0.2041, "step": 8200 }, { "epoch": 0.41, "learning_rate": 6.041538461538462e-06, "loss": 0.186, "step": 8225 }, { "epoch": 0.41, "learning_rate": 6.028717948717949e-06, "loss": 0.1799, "step": 8250 }, { "epoch": 0.41, "learning_rate": 6.015897435897436e-06, "loss": 0.1544, "step": 8275 }, { "epoch": 0.41, "learning_rate": 6.003076923076924e-06, "loss": 0.2298, "step": 8300 }, { "epoch": 0.42, "learning_rate": 5.9902564102564105e-06, "loss": 0.1769, "step": 8325 }, { "epoch": 0.42, "learning_rate": 5.977435897435898e-06, "loss": 0.1666, "step": 8350 }, { "epoch": 0.42, "learning_rate": 5.964615384615385e-06, "loss": 0.1145, "step": 8375 }, { "epoch": 0.42, "learning_rate": 5.951794871794873e-06, "loss": 0.1337, "step": 8400 }, { "epoch": 0.42, "learning_rate": 5.9389743589743595e-06, "loss": 0.202, "step": 8425 }, { "epoch": 0.42, "learning_rate": 5.926153846153847e-06, "loss": 0.2293, "step": 8450 }, { "epoch": 0.42, "learning_rate": 5.913333333333334e-06, "loss": 0.1453, "step": 8475 }, { "epoch": 0.42, "learning_rate": 5.900512820512822e-06, "loss": 0.2571, "step": 8500 }, { "epoch": 0.43, "learning_rate": 5.8876923076923085e-06, "loss": 0.2198, "step": 8525 }, { "epoch": 0.43, "learning_rate": 5.8748717948717945e-06, "loss": 0.1664, "step": 8550 }, { "epoch": 0.43, "learning_rate": 5.862051282051282e-06, "loss": 0.1791, "step": 8575 }, { "epoch": 0.43, "learning_rate": 5.849230769230769e-06, "loss": 0.2011, "step": 8600 }, { "epoch": 0.43, "learning_rate": 5.836410256410257e-06, "loss": 0.1844, "step": 8625 }, { "epoch": 0.43, "learning_rate": 5.8235897435897435e-06, "loss": 0.2135, "step": 8650 }, { "epoch": 0.43, "learning_rate": 5.811282051282053e-06, "loss": 0.1645, "step": 8675 }, { "epoch": 0.43, "learning_rate": 5.7984615384615394e-06, "loss": 0.1625, "step": 8700 }, { "epoch": 0.44, "learning_rate": 5.785641025641025e-06, "loss": 0.2671, "step": 8725 }, { "epoch": 0.44, "learning_rate": 5.772820512820513e-06, "loss": 0.1393, "step": 8750 }, { "epoch": 0.44, "learning_rate": 5.76e-06, "loss": 0.2068, "step": 8775 }, { "epoch": 0.44, "learning_rate": 5.747179487179488e-06, "loss": 0.2262, "step": 8800 }, { "epoch": 0.44, "learning_rate": 5.734358974358974e-06, "loss": 0.2413, "step": 8825 }, { "epoch": 0.44, "learning_rate": 5.721538461538462e-06, "loss": 0.2111, "step": 8850 }, { "epoch": 0.44, "learning_rate": 5.708717948717949e-06, "loss": 0.2413, "step": 8875 }, { "epoch": 0.45, "learning_rate": 5.6958974358974366e-06, "loss": 0.1866, "step": 8900 }, { "epoch": 0.45, "learning_rate": 5.683076923076923e-06, "loss": 0.1819, "step": 8925 }, { "epoch": 0.45, "learning_rate": 5.67025641025641e-06, "loss": 0.2119, "step": 8950 }, { "epoch": 0.45, "learning_rate": 5.657435897435898e-06, "loss": 0.1712, "step": 8975 }, { "epoch": 0.45, "learning_rate": 5.644615384615385e-06, "loss": 0.1718, "step": 9000 }, { "epoch": 0.45, "learning_rate": 5.631794871794872e-06, "loss": 0.259, "step": 9025 }, { "epoch": 0.45, "learning_rate": 5.618974358974359e-06, "loss": 0.1678, "step": 9050 }, { "epoch": 0.45, "learning_rate": 5.606153846153847e-06, "loss": 0.1971, "step": 9075 }, { "epoch": 0.46, "learning_rate": 5.593333333333334e-06, "loss": 0.2426, "step": 9100 }, { "epoch": 0.46, "learning_rate": 5.580512820512821e-06, "loss": 0.1909, "step": 9125 }, { "epoch": 0.46, "learning_rate": 5.567692307692308e-06, "loss": 0.1973, "step": 9150 }, { "epoch": 0.46, "learning_rate": 5.554871794871796e-06, "loss": 0.2049, "step": 9175 }, { "epoch": 0.46, "learning_rate": 5.542051282051283e-06, "loss": 0.263, "step": 9200 }, { "epoch": 0.46, "learning_rate": 5.52923076923077e-06, "loss": 0.1635, "step": 9225 }, { "epoch": 0.46, "learning_rate": 5.516410256410257e-06, "loss": 0.2338, "step": 9250 }, { "epoch": 0.46, "learning_rate": 5.503589743589745e-06, "loss": 0.244, "step": 9275 }, { "epoch": 0.47, "learning_rate": 5.490769230769232e-06, "loss": 0.2346, "step": 9300 }, { "epoch": 0.47, "learning_rate": 5.477948717948718e-06, "loss": 0.193, "step": 9325 }, { "epoch": 0.47, "learning_rate": 5.465128205128205e-06, "loss": 0.1848, "step": 9350 }, { "epoch": 0.47, "learning_rate": 5.452307692307692e-06, "loss": 0.12, "step": 9375 }, { "epoch": 0.47, "learning_rate": 5.43948717948718e-06, "loss": 0.2107, "step": 9400 }, { "epoch": 0.47, "learning_rate": 5.426666666666667e-06, "loss": 0.1724, "step": 9425 }, { "epoch": 0.47, "learning_rate": 5.413846153846154e-06, "loss": 0.1969, "step": 9450 }, { "epoch": 0.47, "learning_rate": 5.401025641025641e-06, "loss": 0.2131, "step": 9475 }, { "epoch": 0.47, "learning_rate": 5.388205128205129e-06, "loss": 0.1624, "step": 9500 }, { "epoch": 0.48, "learning_rate": 5.375384615384616e-06, "loss": 0.2201, "step": 9525 }, { "epoch": 0.48, "learning_rate": 5.3625641025641025e-06, "loss": 0.2565, "step": 9550 }, { "epoch": 0.48, "learning_rate": 5.34974358974359e-06, "loss": 0.1863, "step": 9575 }, { "epoch": 0.48, "learning_rate": 5.336923076923077e-06, "loss": 0.2139, "step": 9600 }, { "epoch": 0.48, "learning_rate": 5.324102564102565e-06, "loss": 0.1954, "step": 9625 }, { "epoch": 0.48, "learning_rate": 5.3112820512820515e-06, "loss": 0.2186, "step": 9650 }, { "epoch": 0.48, "learning_rate": 5.298461538461539e-06, "loss": 0.2129, "step": 9675 }, { "epoch": 0.48, "learning_rate": 5.285641025641026e-06, "loss": 0.1804, "step": 9700 }, { "epoch": 0.49, "learning_rate": 5.272820512820514e-06, "loss": 0.1912, "step": 9725 }, { "epoch": 0.49, "learning_rate": 5.2600000000000005e-06, "loss": 0.1619, "step": 9750 }, { "epoch": 0.49, "learning_rate": 5.247179487179488e-06, "loss": 0.1521, "step": 9775 }, { "epoch": 0.49, "learning_rate": 5.234358974358975e-06, "loss": 0.2549, "step": 9800 }, { "epoch": 0.49, "learning_rate": 5.221538461538463e-06, "loss": 0.1659, "step": 9825 }, { "epoch": 0.49, "learning_rate": 5.2087179487179495e-06, "loss": 0.1936, "step": 9850 }, { "epoch": 0.49, "learning_rate": 5.195897435897437e-06, "loss": 0.1784, "step": 9875 }, { "epoch": 0.49, "learning_rate": 5.183076923076924e-06, "loss": 0.2317, "step": 9900 }, { "epoch": 0.5, "learning_rate": 5.170256410256412e-06, "loss": 0.1954, "step": 9925 }, { "epoch": 0.5, "learning_rate": 5.157435897435898e-06, "loss": 0.2, "step": 9950 }, { "epoch": 0.5, "learning_rate": 5.144615384615384e-06, "loss": 0.1607, "step": 9975 }, { "epoch": 0.5, "learning_rate": 5.131794871794872e-06, "loss": 0.1741, "step": 10000 }, { "epoch": 0.5, "eval_loss": 0.26475706696510315, "eval_runtime": 18928.303, "eval_samples_per_second": 0.863, "eval_steps_per_second": 0.863, "eval_wer": 11.00876066546814, "step": 10000 }, { "epoch": 0.5, "learning_rate": 5.118974358974359e-06, "loss": 0.2249, "step": 10025 }, { "epoch": 0.5, "learning_rate": 5.106153846153847e-06, "loss": 0.1868, "step": 10050 }, { "epoch": 0.5, "learning_rate": 5.093333333333333e-06, "loss": 0.1739, "step": 10075 }, { "epoch": 0.51, "learning_rate": 5.080512820512821e-06, "loss": 0.1457, "step": 10100 }, { "epoch": 0.51, "learning_rate": 5.067692307692308e-06, "loss": 0.1841, "step": 10125 }, { "epoch": 0.51, "learning_rate": 5.054871794871795e-06, "loss": 0.1708, "step": 10150 }, { "epoch": 0.51, "learning_rate": 5.042051282051282e-06, "loss": 0.233, "step": 10175 }, { "epoch": 0.51, "learning_rate": 5.029230769230769e-06, "loss": 0.2166, "step": 10200 }, { "epoch": 0.51, "learning_rate": 5.016410256410257e-06, "loss": 0.1728, "step": 10225 }, { "epoch": 0.51, "learning_rate": 5.003589743589744e-06, "loss": 0.1786, "step": 10250 }, { "epoch": 0.51, "learning_rate": 4.990769230769231e-06, "loss": 0.2963, "step": 10275 }, { "epoch": 0.52, "learning_rate": 4.977948717948718e-06, "loss": 0.1379, "step": 10300 }, { "epoch": 0.52, "learning_rate": 4.965128205128206e-06, "loss": 0.2376, "step": 10325 }, { "epoch": 0.52, "learning_rate": 4.952307692307693e-06, "loss": 0.1692, "step": 10350 }, { "epoch": 0.52, "learning_rate": 4.93948717948718e-06, "loss": 0.1433, "step": 10375 }, { "epoch": 0.52, "learning_rate": 4.926666666666667e-06, "loss": 0.2132, "step": 10400 }, { "epoch": 0.52, "learning_rate": 4.913846153846154e-06, "loss": 0.196, "step": 10425 }, { "epoch": 0.52, "learning_rate": 4.901025641025641e-06, "loss": 0.1758, "step": 10450 }, { "epoch": 0.52, "learning_rate": 4.8882051282051285e-06, "loss": 0.1369, "step": 10475 }, { "epoch": 0.53, "learning_rate": 4.875384615384615e-06, "loss": 0.2515, "step": 10500 }, { "epoch": 0.53, "learning_rate": 4.862564102564103e-06, "loss": 0.1653, "step": 10525 }, { "epoch": 0.53, "learning_rate": 4.84974358974359e-06, "loss": 0.1727, "step": 10550 }, { "epoch": 0.53, "learning_rate": 4.8369230769230775e-06, "loss": 0.2502, "step": 10575 }, { "epoch": 0.53, "learning_rate": 4.824102564102564e-06, "loss": 0.2177, "step": 10600 }, { "epoch": 0.53, "learning_rate": 4.811282051282052e-06, "loss": 0.1941, "step": 10625 }, { "epoch": 0.53, "learning_rate": 4.798461538461539e-06, "loss": 0.2151, "step": 10650 }, { "epoch": 0.53, "learning_rate": 4.7856410256410265e-06, "loss": 0.2528, "step": 10675 }, { "epoch": 0.54, "learning_rate": 4.772820512820513e-06, "loss": 0.2247, "step": 10700 }, { "epoch": 0.54, "learning_rate": 4.76e-06, "loss": 0.2198, "step": 10725 }, { "epoch": 0.54, "learning_rate": 4.747179487179487e-06, "loss": 0.2017, "step": 10750 }, { "epoch": 0.54, "learning_rate": 4.734358974358975e-06, "loss": 0.1903, "step": 10775 }, { "epoch": 0.54, "learning_rate": 4.7215384615384615e-06, "loss": 0.1586, "step": 10800 }, { "epoch": 0.54, "learning_rate": 4.708717948717949e-06, "loss": 0.2279, "step": 10825 }, { "epoch": 0.54, "learning_rate": 4.695897435897436e-06, "loss": 0.2752, "step": 10850 }, { "epoch": 0.54, "learning_rate": 4.683076923076924e-06, "loss": 0.217, "step": 10875 }, { "epoch": 0.55, "learning_rate": 4.6702564102564105e-06, "loss": 0.1995, "step": 10900 }, { "epoch": 0.55, "learning_rate": 4.657435897435898e-06, "loss": 0.1936, "step": 10925 }, { "epoch": 0.55, "learning_rate": 4.644615384615385e-06, "loss": 0.2082, "step": 10950 }, { "epoch": 0.55, "learning_rate": 4.631794871794873e-06, "loss": 0.247, "step": 10975 }, { "epoch": 0.55, "learning_rate": 4.6189743589743595e-06, "loss": 0.1539, "step": 11000 }, { "epoch": 0.55, "learning_rate": 4.606153846153846e-06, "loss": 0.2201, "step": 11025 }, { "epoch": 0.55, "learning_rate": 4.5938461538461546e-06, "loss": 0.2845, "step": 11050 }, { "epoch": 0.55, "learning_rate": 4.581025641025641e-06, "loss": 0.2098, "step": 11075 }, { "epoch": 0.56, "learning_rate": 4.568205128205129e-06, "loss": 0.2216, "step": 11100 }, { "epoch": 0.56, "learning_rate": 4.555384615384616e-06, "loss": 0.1963, "step": 11125 }, { "epoch": 0.56, "learning_rate": 4.5425641025641036e-06, "loss": 0.2575, "step": 11150 }, { "epoch": 0.56, "learning_rate": 4.5297435897435895e-06, "loss": 0.2189, "step": 11175 }, { "epoch": 0.56, "learning_rate": 4.516923076923077e-06, "loss": 0.2377, "step": 11200 }, { "epoch": 0.56, "learning_rate": 4.504102564102564e-06, "loss": 0.2099, "step": 11225 }, { "epoch": 0.56, "learning_rate": 4.491282051282052e-06, "loss": 0.2526, "step": 11250 }, { "epoch": 0.56, "learning_rate": 4.4784615384615385e-06, "loss": 0.248, "step": 11275 }, { "epoch": 0.56, "learning_rate": 4.465641025641026e-06, "loss": 0.225, "step": 11300 }, { "epoch": 0.57, "learning_rate": 4.452820512820513e-06, "loss": 0.2252, "step": 11325 }, { "epoch": 0.57, "learning_rate": 4.440000000000001e-06, "loss": 0.192, "step": 11350 }, { "epoch": 0.57, "learning_rate": 4.4271794871794875e-06, "loss": 0.206, "step": 11375 }, { "epoch": 0.57, "learning_rate": 4.414358974358975e-06, "loss": 0.1675, "step": 11400 }, { "epoch": 0.57, "learning_rate": 4.401538461538462e-06, "loss": 0.1707, "step": 11425 }, { "epoch": 0.57, "learning_rate": 4.38871794871795e-06, "loss": 0.2326, "step": 11450 }, { "epoch": 0.57, "learning_rate": 4.375897435897436e-06, "loss": 0.1907, "step": 11475 }, { "epoch": 0.57, "learning_rate": 4.363076923076923e-06, "loss": 0.2512, "step": 11500 }, { "epoch": 0.58, "learning_rate": 4.35025641025641e-06, "loss": 0.2738, "step": 11525 }, { "epoch": 0.58, "learning_rate": 4.337435897435898e-06, "loss": 0.2091, "step": 11550 }, { "epoch": 0.58, "learning_rate": 4.324615384615385e-06, "loss": 0.1791, "step": 11575 }, { "epoch": 0.58, "learning_rate": 4.311794871794872e-06, "loss": 0.1668, "step": 11600 }, { "epoch": 0.58, "learning_rate": 4.298974358974359e-06, "loss": 0.216, "step": 11625 }, { "epoch": 0.58, "learning_rate": 4.286153846153847e-06, "loss": 0.2948, "step": 11650 }, { "epoch": 0.58, "learning_rate": 4.273333333333334e-06, "loss": 0.1563, "step": 11675 }, { "epoch": 0.58, "learning_rate": 4.260512820512821e-06, "loss": 0.1796, "step": 11700 }, { "epoch": 0.59, "learning_rate": 4.247692307692308e-06, "loss": 0.2025, "step": 11725 }, { "epoch": 0.59, "learning_rate": 4.234871794871796e-06, "loss": 0.1765, "step": 11750 }, { "epoch": 0.59, "learning_rate": 4.222051282051283e-06, "loss": 0.2129, "step": 11775 }, { "epoch": 0.59, "learning_rate": 4.2092307692307695e-06, "loss": 0.2073, "step": 11800 }, { "epoch": 0.59, "learning_rate": 4.196410256410256e-06, "loss": 0.2088, "step": 11825 }, { "epoch": 0.59, "learning_rate": 4.183589743589744e-06, "loss": 0.1748, "step": 11850 }, { "epoch": 0.59, "learning_rate": 4.170769230769231e-06, "loss": 0.2492, "step": 11875 }, { "epoch": 0.59, "learning_rate": 4.1579487179487184e-06, "loss": 0.2043, "step": 11900 }, { "epoch": 0.6, "learning_rate": 4.145128205128205e-06, "loss": 0.1446, "step": 11925 }, { "epoch": 0.6, "learning_rate": 4.132307692307693e-06, "loss": 0.2016, "step": 11950 }, { "epoch": 0.6, "learning_rate": 4.11948717948718e-06, "loss": 0.185, "step": 11975 }, { "epoch": 0.6, "learning_rate": 4.1066666666666674e-06, "loss": 0.2037, "step": 12000 }, { "epoch": 0.6, "eval_loss": 0.24701610207557678, "eval_runtime": 18876.3965, "eval_samples_per_second": 0.866, "eval_steps_per_second": 0.866, "eval_wer": 10.190908312644465, "step": 12000 }, { "epoch": 0.6, "learning_rate": 4.093846153846154e-06, "loss": 0.1759, "step": 12025 }, { "epoch": 0.6, "learning_rate": 4.081025641025642e-06, "loss": 0.1671, "step": 12050 }, { "epoch": 0.6, "learning_rate": 4.068205128205129e-06, "loss": 0.2421, "step": 12075 }, { "epoch": 0.6, "learning_rate": 4.055384615384616e-06, "loss": 0.1862, "step": 12100 }, { "epoch": 0.61, "learning_rate": 4.042564102564102e-06, "loss": 0.2603, "step": 12125 }, { "epoch": 0.61, "learning_rate": 4.02974358974359e-06, "loss": 0.2342, "step": 12150 }, { "epoch": 0.61, "learning_rate": 4.016923076923077e-06, "loss": 0.1512, "step": 12175 }, { "epoch": 0.61, "learning_rate": 4.004102564102565e-06, "loss": 0.2011, "step": 12200 }, { "epoch": 0.61, "learning_rate": 3.991282051282051e-06, "loss": 0.2763, "step": 12225 }, { "epoch": 0.61, "learning_rate": 3.978461538461539e-06, "loss": 0.1574, "step": 12250 }, { "epoch": 0.61, "learning_rate": 3.965641025641026e-06, "loss": 0.2361, "step": 12275 }, { "epoch": 0.61, "learning_rate": 3.9528205128205136e-06, "loss": 0.1887, "step": 12300 }, { "epoch": 0.62, "learning_rate": 3.94e-06, "loss": 0.1513, "step": 12325 }, { "epoch": 0.62, "learning_rate": 3.927179487179488e-06, "loss": 0.2568, "step": 12350 }, { "epoch": 0.62, "learning_rate": 3.914358974358975e-06, "loss": 0.2072, "step": 12375 }, { "epoch": 0.62, "learning_rate": 3.901538461538462e-06, "loss": 0.1641, "step": 12400 }, { "epoch": 0.62, "learning_rate": 3.8887179487179485e-06, "loss": 0.2342, "step": 12425 }, { "epoch": 0.62, "learning_rate": 3.875897435897436e-06, "loss": 0.2017, "step": 12450 }, { "epoch": 0.62, "learning_rate": 3.863076923076923e-06, "loss": 0.2335, "step": 12475 }, { "epoch": 0.62, "learning_rate": 3.850256410256411e-06, "loss": 0.2, "step": 12500 }, { "epoch": 0.63, "learning_rate": 3.8374358974358975e-06, "loss": 0.1501, "step": 12525 }, { "epoch": 0.63, "learning_rate": 3.824615384615385e-06, "loss": 0.1866, "step": 12550 }, { "epoch": 0.63, "learning_rate": 3.811794871794872e-06, "loss": 0.1083, "step": 12575 }, { "epoch": 0.63, "learning_rate": 3.7989743589743593e-06, "loss": 0.1084, "step": 12600 }, { "epoch": 0.63, "learning_rate": 3.7861538461538465e-06, "loss": 0.1642, "step": 12625 }, { "epoch": 0.63, "learning_rate": 3.7733333333333338e-06, "loss": 0.1873, "step": 12650 }, { "epoch": 0.63, "learning_rate": 3.760512820512821e-06, "loss": 0.2075, "step": 12675 }, { "epoch": 0.64, "learning_rate": 3.747692307692308e-06, "loss": 0.1922, "step": 12700 }, { "epoch": 0.64, "learning_rate": 3.734871794871795e-06, "loss": 0.1691, "step": 12725 }, { "epoch": 0.64, "learning_rate": 3.7220512820512823e-06, "loss": 0.1387, "step": 12750 }, { "epoch": 0.64, "learning_rate": 3.7092307692307696e-06, "loss": 0.1885, "step": 12775 }, { "epoch": 0.64, "learning_rate": 3.696410256410257e-06, "loss": 0.1773, "step": 12800 }, { "epoch": 0.64, "learning_rate": 3.6835897435897437e-06, "loss": 0.0919, "step": 12825 }, { "epoch": 0.64, "learning_rate": 3.670769230769231e-06, "loss": 0.1707, "step": 12850 }, { "epoch": 0.64, "learning_rate": 3.657948717948718e-06, "loss": 0.1787, "step": 12875 }, { "epoch": 0.65, "learning_rate": 3.6451282051282054e-06, "loss": 0.2146, "step": 12900 }, { "epoch": 0.65, "learning_rate": 3.6323076923076926e-06, "loss": 0.1601, "step": 12925 }, { "epoch": 0.65, "learning_rate": 3.61948717948718e-06, "loss": 0.128, "step": 12950 }, { "epoch": 0.65, "learning_rate": 3.606666666666667e-06, "loss": 0.185, "step": 12975 }, { "epoch": 0.65, "learning_rate": 3.593846153846154e-06, "loss": 0.1089, "step": 13000 }, { "epoch": 0.65, "learning_rate": 3.5810256410256412e-06, "loss": 0.1424, "step": 13025 }, { "epoch": 0.65, "learning_rate": 3.5682051282051285e-06, "loss": 0.1172, "step": 13050 }, { "epoch": 0.65, "learning_rate": 3.5553846153846157e-06, "loss": 0.1775, "step": 13075 }, { "epoch": 0.66, "learning_rate": 3.542564102564103e-06, "loss": 0.1821, "step": 13100 }, { "epoch": 0.66, "learning_rate": 3.5297435897435898e-06, "loss": 0.1337, "step": 13125 }, { "epoch": 0.66, "learning_rate": 3.516923076923077e-06, "loss": 0.139, "step": 13150 }, { "epoch": 0.66, "learning_rate": 3.5041025641025643e-06, "loss": 0.0985, "step": 13175 }, { "epoch": 0.66, "learning_rate": 3.4912820512820515e-06, "loss": 0.1199, "step": 13200 }, { "epoch": 0.66, "learning_rate": 3.4784615384615388e-06, "loss": 0.1836, "step": 13225 }, { "epoch": 0.66, "learning_rate": 3.465641025641026e-06, "loss": 0.1241, "step": 13250 }, { "epoch": 0.66, "learning_rate": 3.4528205128205133e-06, "loss": 0.1445, "step": 13275 }, { "epoch": 0.67, "learning_rate": 3.44e-06, "loss": 0.1459, "step": 13300 }, { "epoch": 0.67, "learning_rate": 3.4271794871794873e-06, "loss": 0.1358, "step": 13325 }, { "epoch": 0.67, "learning_rate": 3.4143589743589746e-06, "loss": 0.0987, "step": 13350 }, { "epoch": 0.67, "learning_rate": 3.401538461538462e-06, "loss": 0.1298, "step": 13375 }, { "epoch": 0.67, "learning_rate": 3.388717948717949e-06, "loss": 0.1482, "step": 13400 }, { "epoch": 0.67, "learning_rate": 3.375897435897436e-06, "loss": 0.1157, "step": 13425 }, { "epoch": 0.67, "learning_rate": 3.363076923076923e-06, "loss": 0.1875, "step": 13450 }, { "epoch": 0.67, "learning_rate": 3.3502564102564104e-06, "loss": 0.1344, "step": 13475 }, { "epoch": 0.68, "learning_rate": 3.3374358974358977e-06, "loss": 0.0993, "step": 13500 }, { "epoch": 0.68, "learning_rate": 3.324615384615385e-06, "loss": 0.147, "step": 13525 }, { "epoch": 0.68, "learning_rate": 3.311794871794872e-06, "loss": 0.1794, "step": 13550 }, { "epoch": 0.68, "learning_rate": 3.2989743589743594e-06, "loss": 0.1343, "step": 13575 }, { "epoch": 0.68, "learning_rate": 3.2861538461538467e-06, "loss": 0.1366, "step": 13600 }, { "epoch": 0.68, "learning_rate": 3.2733333333333335e-06, "loss": 0.112, "step": 13625 }, { "epoch": 0.68, "learning_rate": 3.2610256410256413e-06, "loss": 0.1506, "step": 13650 }, { "epoch": 0.68, "learning_rate": 3.2482051282051286e-06, "loss": 0.213, "step": 13675 }, { "epoch": 0.69, "learning_rate": 3.235384615384616e-06, "loss": 0.1451, "step": 13700 }, { "epoch": 0.69, "learning_rate": 3.222564102564103e-06, "loss": 0.1778, "step": 13725 }, { "epoch": 0.69, "learning_rate": 3.2097435897435903e-06, "loss": 0.1196, "step": 13750 }, { "epoch": 0.69, "learning_rate": 3.1969230769230776e-06, "loss": 0.0989, "step": 13775 }, { "epoch": 0.69, "learning_rate": 3.184102564102564e-06, "loss": 0.155, "step": 13800 }, { "epoch": 0.69, "learning_rate": 3.171282051282051e-06, "loss": 0.161, "step": 13825 }, { "epoch": 0.69, "learning_rate": 3.1584615384615385e-06, "loss": 0.1065, "step": 13850 }, { "epoch": 0.69, "learning_rate": 3.1456410256410257e-06, "loss": 0.1707, "step": 13875 }, { "epoch": 0.69, "learning_rate": 3.132820512820513e-06, "loss": 0.1299, "step": 13900 }, { "epoch": 0.7, "learning_rate": 3.12e-06, "loss": 0.2309, "step": 13925 }, { "epoch": 0.7, "learning_rate": 3.1071794871794874e-06, "loss": 0.1629, "step": 13950 }, { "epoch": 0.7, "learning_rate": 3.0943589743589747e-06, "loss": 0.1675, "step": 13975 }, { "epoch": 0.7, "learning_rate": 3.081538461538462e-06, "loss": 0.1586, "step": 14000 }, { "epoch": 0.7, "eval_loss": 0.23334650695323944, "eval_runtime": 18772.2505, "eval_samples_per_second": 0.87, "eval_steps_per_second": 0.87, "eval_wer": 9.409582512912708, "step": 14000 }, { "epoch": 0.7, "learning_rate": 3.068717948717949e-06, "loss": 0.1408, "step": 14025 }, { "epoch": 0.7, "learning_rate": 3.0558974358974364e-06, "loss": 0.1751, "step": 14050 }, { "epoch": 0.7, "learning_rate": 3.0430769230769237e-06, "loss": 0.1608, "step": 14075 }, { "epoch": 0.7, "learning_rate": 3.03025641025641e-06, "loss": 0.1322, "step": 14100 }, { "epoch": 0.71, "learning_rate": 3.0174358974358973e-06, "loss": 0.2265, "step": 14125 }, { "epoch": 0.71, "learning_rate": 3.0046153846153846e-06, "loss": 0.1534, "step": 14150 }, { "epoch": 0.71, "learning_rate": 2.991794871794872e-06, "loss": 0.1453, "step": 14175 }, { "epoch": 0.71, "learning_rate": 2.978974358974359e-06, "loss": 0.1697, "step": 14200 }, { "epoch": 0.71, "learning_rate": 2.9661538461538463e-06, "loss": 0.2237, "step": 14225 }, { "epoch": 0.71, "learning_rate": 2.9533333333333336e-06, "loss": 0.1432, "step": 14250 }, { "epoch": 0.71, "learning_rate": 2.940512820512821e-06, "loss": 0.1307, "step": 14275 }, { "epoch": 0.71, "learning_rate": 2.927692307692308e-06, "loss": 0.1965, "step": 14300 }, { "epoch": 0.72, "learning_rate": 2.9148717948717953e-06, "loss": 0.1383, "step": 14325 }, { "epoch": 0.72, "learning_rate": 2.9020512820512826e-06, "loss": 0.1494, "step": 14350 }, { "epoch": 0.72, "learning_rate": 2.88923076923077e-06, "loss": 0.1926, "step": 14375 }, { "epoch": 0.72, "learning_rate": 2.8764102564102562e-06, "loss": 0.1912, "step": 14400 }, { "epoch": 0.72, "learning_rate": 2.8635897435897435e-06, "loss": 0.1918, "step": 14425 }, { "epoch": 0.72, "learning_rate": 2.8507692307692307e-06, "loss": 0.1996, "step": 14450 }, { "epoch": 0.72, "learning_rate": 2.837948717948718e-06, "loss": 0.1997, "step": 14475 }, { "epoch": 0.72, "learning_rate": 2.825128205128205e-06, "loss": 0.201, "step": 14500 }, { "epoch": 0.73, "learning_rate": 2.8123076923076925e-06, "loss": 0.1396, "step": 14525 }, { "epoch": 0.73, "learning_rate": 2.7994871794871797e-06, "loss": 0.2583, "step": 14550 }, { "epoch": 0.73, "learning_rate": 2.786666666666667e-06, "loss": 0.2007, "step": 14575 }, { "epoch": 0.73, "learning_rate": 2.773846153846154e-06, "loss": 0.1886, "step": 14600 }, { "epoch": 0.73, "learning_rate": 2.7610256410256414e-06, "loss": 0.2423, "step": 14625 }, { "epoch": 0.73, "learning_rate": 2.7482051282051287e-06, "loss": 0.2049, "step": 14650 }, { "epoch": 0.73, "learning_rate": 2.735384615384616e-06, "loss": 0.1674, "step": 14675 }, { "epoch": 0.73, "learning_rate": 2.7225641025641023e-06, "loss": 0.1666, "step": 14700 }, { "epoch": 0.74, "learning_rate": 2.7097435897435896e-06, "loss": 0.1195, "step": 14725 }, { "epoch": 0.74, "learning_rate": 2.696923076923077e-06, "loss": 0.1322, "step": 14750 }, { "epoch": 0.74, "learning_rate": 2.684102564102564e-06, "loss": 0.1972, "step": 14775 }, { "epoch": 0.74, "learning_rate": 2.6712820512820513e-06, "loss": 0.1826, "step": 14800 }, { "epoch": 0.74, "learning_rate": 2.6584615384615386e-06, "loss": 0.2226, "step": 14825 }, { "epoch": 0.74, "learning_rate": 2.645641025641026e-06, "loss": 0.1377, "step": 14850 }, { "epoch": 0.74, "learning_rate": 2.632820512820513e-06, "loss": 0.2146, "step": 14875 }, { "epoch": 0.74, "learning_rate": 2.6200000000000003e-06, "loss": 0.113, "step": 14900 }, { "epoch": 0.75, "learning_rate": 2.6071794871794876e-06, "loss": 0.1719, "step": 14925 }, { "epoch": 0.75, "learning_rate": 2.594358974358975e-06, "loss": 0.1296, "step": 14950 }, { "epoch": 0.75, "learning_rate": 2.581538461538462e-06, "loss": 0.2052, "step": 14975 }, { "epoch": 0.75, "learning_rate": 2.5687179487179485e-06, "loss": 0.2401, "step": 15000 }, { "epoch": 0.75, "learning_rate": 2.5558974358974357e-06, "loss": 0.1589, "step": 15025 }, { "epoch": 0.75, "learning_rate": 2.543076923076923e-06, "loss": 0.143, "step": 15050 }, { "epoch": 0.75, "learning_rate": 2.5302564102564102e-06, "loss": 0.1596, "step": 15075 }, { "epoch": 0.76, "learning_rate": 2.5174358974358975e-06, "loss": 0.2011, "step": 15100 }, { "epoch": 0.76, "learning_rate": 2.5046153846153847e-06, "loss": 0.1549, "step": 15125 }, { "epoch": 0.76, "learning_rate": 2.491794871794872e-06, "loss": 0.0953, "step": 15150 }, { "epoch": 0.76, "learning_rate": 2.478974358974359e-06, "loss": 0.1909, "step": 15175 }, { "epoch": 0.76, "learning_rate": 2.4661538461538465e-06, "loss": 0.2364, "step": 15200 }, { "epoch": 0.76, "learning_rate": 2.4533333333333333e-06, "loss": 0.1783, "step": 15225 }, { "epoch": 0.76, "learning_rate": 2.4405128205128205e-06, "loss": 0.1273, "step": 15250 }, { "epoch": 0.76, "learning_rate": 2.4276923076923078e-06, "loss": 0.1827, "step": 15275 }, { "epoch": 0.77, "learning_rate": 2.414871794871795e-06, "loss": 0.1641, "step": 15300 }, { "epoch": 0.77, "learning_rate": 2.4020512820512823e-06, "loss": 0.2446, "step": 15325 }, { "epoch": 0.77, "learning_rate": 2.3892307692307695e-06, "loss": 0.1589, "step": 15350 }, { "epoch": 0.77, "learning_rate": 2.3764102564102563e-06, "loss": 0.1273, "step": 15375 }, { "epoch": 0.77, "learning_rate": 2.3635897435897436e-06, "loss": 0.1311, "step": 15400 }, { "epoch": 0.77, "learning_rate": 2.350769230769231e-06, "loss": 0.1941, "step": 15425 }, { "epoch": 0.77, "learning_rate": 2.337948717948718e-06, "loss": 0.2189, "step": 15450 }, { "epoch": 0.77, "learning_rate": 2.3251282051282053e-06, "loss": 0.2339, "step": 15475 }, { "epoch": 0.78, "learning_rate": 2.3123076923076926e-06, "loss": 0.1909, "step": 15500 }, { "epoch": 0.78, "learning_rate": 2.29948717948718e-06, "loss": 0.1264, "step": 15525 }, { "epoch": 0.78, "learning_rate": 2.2866666666666667e-06, "loss": 0.1251, "step": 15550 }, { "epoch": 0.78, "learning_rate": 2.273846153846154e-06, "loss": 0.1692, "step": 15575 }, { "epoch": 0.78, "learning_rate": 2.261025641025641e-06, "loss": 0.1462, "step": 15600 }, { "epoch": 0.78, "learning_rate": 2.2482051282051284e-06, "loss": 0.1529, "step": 15625 }, { "epoch": 0.78, "learning_rate": 2.2353846153846156e-06, "loss": 0.1641, "step": 15650 }, { "epoch": 0.78, "learning_rate": 2.222564102564103e-06, "loss": 0.1075, "step": 15675 }, { "epoch": 0.79, "learning_rate": 2.2097435897435897e-06, "loss": 0.2809, "step": 15700 }, { "epoch": 0.79, "learning_rate": 2.196923076923077e-06, "loss": 0.199, "step": 15725 }, { "epoch": 0.79, "learning_rate": 2.1841025641025642e-06, "loss": 0.1734, "step": 15750 }, { "epoch": 0.79, "learning_rate": 2.1712820512820515e-06, "loss": 0.1638, "step": 15775 }, { "epoch": 0.79, "learning_rate": 2.1584615384615387e-06, "loss": 0.1426, "step": 15800 }, { "epoch": 0.79, "learning_rate": 2.145641025641026e-06, "loss": 0.1483, "step": 15825 }, { "epoch": 0.79, "learning_rate": 2.1328205128205128e-06, "loss": 0.1769, "step": 15850 }, { "epoch": 0.79, "learning_rate": 2.12e-06, "loss": 0.1557, "step": 15875 }, { "epoch": 0.8, "learning_rate": 2.1071794871794873e-06, "loss": 0.178, "step": 15900 }, { "epoch": 0.8, "learning_rate": 2.0943589743589745e-06, "loss": 0.2075, "step": 15925 }, { "epoch": 0.8, "learning_rate": 2.0815384615384618e-06, "loss": 0.1643, "step": 15950 }, { "epoch": 0.8, "learning_rate": 2.068717948717949e-06, "loss": 0.1982, "step": 15975 }, { "epoch": 0.8, "learning_rate": 2.055897435897436e-06, "loss": 0.1548, "step": 16000 }, { "epoch": 0.8, "eval_loss": 0.21835076808929443, "eval_runtime": 18649.2977, "eval_samples_per_second": 0.876, "eval_steps_per_second": 0.876, "eval_wer": 8.972405330593842, "step": 16000 }, { "epoch": 0.8, "learning_rate": 2.043076923076923e-06, "loss": 0.1384, "step": 16025 }, { "epoch": 0.8, "learning_rate": 2.0302564102564103e-06, "loss": 0.1606, "step": 16050 }, { "epoch": 0.8, "learning_rate": 2.0174358974358976e-06, "loss": 0.1324, "step": 16075 }, { "epoch": 0.81, "learning_rate": 2.004615384615385e-06, "loss": 0.148, "step": 16100 }, { "epoch": 0.81, "learning_rate": 1.991794871794872e-06, "loss": 0.2402, "step": 16125 }, { "epoch": 0.81, "learning_rate": 1.978974358974359e-06, "loss": 0.2398, "step": 16150 }, { "epoch": 0.81, "learning_rate": 1.966153846153846e-06, "loss": 0.1269, "step": 16175 }, { "epoch": 0.81, "learning_rate": 1.9533333333333334e-06, "loss": 0.1933, "step": 16200 }, { "epoch": 0.81, "learning_rate": 1.9405128205128207e-06, "loss": 0.2184, "step": 16225 }, { "epoch": 0.81, "learning_rate": 1.927692307692308e-06, "loss": 0.1813, "step": 16250 }, { "epoch": 0.81, "learning_rate": 1.914871794871795e-06, "loss": 0.1122, "step": 16275 }, { "epoch": 0.81, "learning_rate": 1.9020512820512822e-06, "loss": 0.2157, "step": 16300 }, { "epoch": 0.82, "learning_rate": 1.8892307692307692e-06, "loss": 0.2085, "step": 16325 }, { "epoch": 1.0, "learning_rate": 1.8764102564102565e-06, "loss": 0.1334, "step": 16350 }, { "epoch": 1.0, "learning_rate": 1.8641025641025643e-06, "loss": 0.152, "step": 16375 }, { "epoch": 1.0, "learning_rate": 1.8512820512820516e-06, "loss": 0.1665, "step": 16400 }, { "epoch": 1.0, "learning_rate": 1.8384615384615386e-06, "loss": 0.1826, "step": 16425 }, { "epoch": 1.01, "learning_rate": 1.8256410256410258e-06, "loss": 0.1206, "step": 16450 }, { "epoch": 1.01, "learning_rate": 1.8128205128205129e-06, "loss": 0.1529, "step": 16475 }, { "epoch": 1.01, "learning_rate": 1.8000000000000001e-06, "loss": 0.1462, "step": 16500 }, { "epoch": 1.01, "learning_rate": 1.7871794871794874e-06, "loss": 0.1321, "step": 16525 }, { "epoch": 1.01, "learning_rate": 1.7743589743589746e-06, "loss": 0.1567, "step": 16550 }, { "epoch": 1.01, "learning_rate": 1.7615384615384617e-06, "loss": 0.1313, "step": 16575 }, { "epoch": 1.01, "learning_rate": 1.748717948717949e-06, "loss": 0.1687, "step": 16600 }, { "epoch": 1.01, "learning_rate": 1.735897435897436e-06, "loss": 0.1138, "step": 16625 }, { "epoch": 1.02, "learning_rate": 1.7230769230769232e-06, "loss": 0.1067, "step": 16650 }, { "epoch": 1.02, "learning_rate": 1.7102564102564104e-06, "loss": 0.1678, "step": 16675 }, { "epoch": 1.02, "learning_rate": 1.6974358974358977e-06, "loss": 0.1265, "step": 16700 }, { "epoch": 1.02, "learning_rate": 1.6846153846153847e-06, "loss": 0.1245, "step": 16725 }, { "epoch": 1.02, "learning_rate": 1.671794871794872e-06, "loss": 0.1242, "step": 16750 }, { "epoch": 1.02, "learning_rate": 1.658974358974359e-06, "loss": 0.1266, "step": 16775 }, { "epoch": 1.02, "learning_rate": 1.6461538461538463e-06, "loss": 0.0883, "step": 16800 }, { "epoch": 1.02, "learning_rate": 1.6333333333333335e-06, "loss": 0.1917, "step": 16825 }, { "epoch": 1.03, "learning_rate": 1.6205128205128208e-06, "loss": 0.1437, "step": 16850 }, { "epoch": 1.03, "learning_rate": 1.6076923076923078e-06, "loss": 0.1213, "step": 16875 }, { "epoch": 1.03, "learning_rate": 1.594871794871795e-06, "loss": 0.1518, "step": 16900 }, { "epoch": 1.03, "learning_rate": 1.582051282051282e-06, "loss": 0.1838, "step": 16925 }, { "epoch": 1.03, "learning_rate": 1.5692307692307693e-06, "loss": 0.1318, "step": 16950 }, { "epoch": 1.03, "learning_rate": 1.5564102564102566e-06, "loss": 0.1627, "step": 16975 }, { "epoch": 1.03, "learning_rate": 1.5435897435897438e-06, "loss": 0.1642, "step": 17000 }, { "epoch": 1.03, "learning_rate": 1.5307692307692309e-06, "loss": 0.1483, "step": 17025 }, { "epoch": 1.04, "learning_rate": 1.517948717948718e-06, "loss": 0.1319, "step": 17050 }, { "epoch": 1.04, "learning_rate": 1.5051282051282051e-06, "loss": 0.0869, "step": 17075 }, { "epoch": 1.04, "learning_rate": 1.4923076923076924e-06, "loss": 0.1272, "step": 17100 }, { "epoch": 1.04, "learning_rate": 1.4794871794871796e-06, "loss": 0.1624, "step": 17125 }, { "epoch": 1.04, "learning_rate": 1.4666666666666669e-06, "loss": 0.0825, "step": 17150 }, { "epoch": 1.04, "learning_rate": 1.453846153846154e-06, "loss": 0.1079, "step": 17175 }, { "epoch": 1.04, "learning_rate": 1.4410256410256412e-06, "loss": 0.2062, "step": 17200 }, { "epoch": 1.04, "learning_rate": 1.4282051282051282e-06, "loss": 0.1518, "step": 17225 }, { "epoch": 1.05, "learning_rate": 1.4153846153846155e-06, "loss": 0.1394, "step": 17250 }, { "epoch": 1.05, "learning_rate": 1.4025641025641027e-06, "loss": 0.1006, "step": 17275 }, { "epoch": 1.05, "learning_rate": 1.38974358974359e-06, "loss": 0.1229, "step": 17300 }, { "epoch": 1.05, "learning_rate": 1.376923076923077e-06, "loss": 0.1608, "step": 17325 }, { "epoch": 1.05, "learning_rate": 1.3641025641025642e-06, "loss": 0.093, "step": 17350 }, { "epoch": 1.05, "learning_rate": 1.3512820512820513e-06, "loss": 0.1625, "step": 17375 }, { "epoch": 1.05, "learning_rate": 1.3384615384615385e-06, "loss": 0.1078, "step": 17400 }, { "epoch": 1.05, "learning_rate": 1.3256410256410258e-06, "loss": 0.1066, "step": 17425 }, { "epoch": 1.06, "learning_rate": 1.312820512820513e-06, "loss": 0.1087, "step": 17450 }, { "epoch": 1.06, "learning_rate": 1.3e-06, "loss": 0.1338, "step": 17475 }, { "epoch": 1.06, "learning_rate": 1.2871794871794873e-06, "loss": 0.1197, "step": 17500 }, { "epoch": 1.06, "learning_rate": 1.2743589743589745e-06, "loss": 0.1147, "step": 17525 }, { "epoch": 1.06, "learning_rate": 1.2615384615384616e-06, "loss": 0.0764, "step": 17550 }, { "epoch": 1.06, "learning_rate": 1.2487179487179488e-06, "loss": 0.1418, "step": 17575 }, { "epoch": 1.06, "learning_rate": 1.235897435897436e-06, "loss": 0.18, "step": 17600 }, { "epoch": 1.06, "learning_rate": 1.2230769230769233e-06, "loss": 0.1501, "step": 17625 }, { "epoch": 1.07, "learning_rate": 1.2102564102564104e-06, "loss": 0.1592, "step": 17650 }, { "epoch": 1.07, "learning_rate": 1.1974358974358976e-06, "loss": 0.1357, "step": 17675 }, { "epoch": 1.07, "learning_rate": 1.1846153846153849e-06, "loss": 0.1898, "step": 17700 }, { "epoch": 1.07, "learning_rate": 1.1717948717948719e-06, "loss": 0.1767, "step": 17725 }, { "epoch": 1.07, "learning_rate": 1.1589743589743591e-06, "loss": 0.1572, "step": 17750 }, { "epoch": 1.07, "learning_rate": 1.1461538461538464e-06, "loss": 0.1359, "step": 17775 }, { "epoch": 1.07, "learning_rate": 1.1333333333333334e-06, "loss": 0.1672, "step": 17800 }, { "epoch": 1.07, "learning_rate": 1.1205128205128207e-06, "loss": 0.1239, "step": 17825 }, { "epoch": 1.08, "learning_rate": 1.107692307692308e-06, "loss": 0.1977, "step": 17850 }, { "epoch": 1.08, "learning_rate": 1.094871794871795e-06, "loss": 0.159, "step": 17875 }, { "epoch": 1.08, "learning_rate": 1.0820512820512822e-06, "loss": 0.097, "step": 17900 }, { "epoch": 1.08, "learning_rate": 1.0692307692307695e-06, "loss": 0.0951, "step": 17925 }, { "epoch": 1.08, "learning_rate": 1.0564102564102565e-06, "loss": 0.1054, "step": 17950 }, { "epoch": 1.08, "learning_rate": 1.0435897435897437e-06, "loss": 0.0833, "step": 17975 }, { "epoch": 1.08, "learning_rate": 1.030769230769231e-06, "loss": 0.1799, "step": 18000 }, { "epoch": 1.08, "eval_loss": 0.20637726783752441, "eval_runtime": 18705.3852, "eval_samples_per_second": 0.874, "eval_steps_per_second": 0.874, "eval_wer": 8.282966640983934, "step": 18000 }, { "epoch": 1.08, "learning_rate": 1.017948717948718e-06, "loss": 0.1495, "step": 18025 }, { "epoch": 1.09, "learning_rate": 1.0051282051282053e-06, "loss": 0.1337, "step": 18050 }, { "epoch": 1.09, "learning_rate": 9.923076923076925e-07, "loss": 0.0922, "step": 18075 }, { "epoch": 1.09, "learning_rate": 9.794871794871796e-07, "loss": 0.1244, "step": 18100 }, { "epoch": 1.09, "learning_rate": 9.666666666666668e-07, "loss": 0.1044, "step": 18125 }, { "epoch": 1.09, "learning_rate": 9.53846153846154e-07, "loss": 0.1464, "step": 18150 }, { "epoch": 1.09, "learning_rate": 9.410256410256411e-07, "loss": 0.106, "step": 18175 }, { "epoch": 1.09, "learning_rate": 9.282051282051282e-07, "loss": 0.1591, "step": 18200 }, { "epoch": 1.09, "learning_rate": 9.153846153846155e-07, "loss": 0.1352, "step": 18225 }, { "epoch": 1.1, "learning_rate": 9.025641025641027e-07, "loss": 0.0864, "step": 18250 }, { "epoch": 1.1, "learning_rate": 8.897435897435898e-07, "loss": 0.0852, "step": 18275 }, { "epoch": 1.1, "learning_rate": 8.76923076923077e-07, "loss": 0.0955, "step": 18300 }, { "epoch": 1.1, "learning_rate": 8.641025641025643e-07, "loss": 0.1232, "step": 18325 }, { "epoch": 1.1, "learning_rate": 8.512820512820513e-07, "loss": 0.1048, "step": 18350 }, { "epoch": 1.1, "learning_rate": 8.384615384615385e-07, "loss": 0.1521, "step": 18375 }, { "epoch": 1.1, "learning_rate": 8.256410256410258e-07, "loss": 0.1376, "step": 18400 }, { "epoch": 1.1, "learning_rate": 8.128205128205128e-07, "loss": 0.1381, "step": 18425 }, { "epoch": 1.11, "learning_rate": 8.000000000000001e-07, "loss": 0.0992, "step": 18450 }, { "epoch": 1.11, "learning_rate": 7.871794871794873e-07, "loss": 0.135, "step": 18475 }, { "epoch": 1.11, "learning_rate": 7.743589743589744e-07, "loss": 0.086, "step": 18500 }, { "epoch": 1.11, "learning_rate": 7.615384615384616e-07, "loss": 0.2145, "step": 18525 }, { "epoch": 1.11, "learning_rate": 7.487179487179488e-07, "loss": 0.0918, "step": 18550 }, { "epoch": 1.11, "learning_rate": 7.358974358974359e-07, "loss": 0.1426, "step": 18575 }, { "epoch": 1.11, "learning_rate": 7.230769230769231e-07, "loss": 0.116, "step": 18600 }, { "epoch": 1.11, "learning_rate": 7.102564102564104e-07, "loss": 0.0961, "step": 18625 }, { "epoch": 1.12, "learning_rate": 6.974358974358974e-07, "loss": 0.1061, "step": 18650 }, { "epoch": 1.12, "learning_rate": 6.846153846153847e-07, "loss": 0.1532, "step": 18675 }, { "epoch": 1.12, "learning_rate": 6.717948717948719e-07, "loss": 0.1531, "step": 18700 }, { "epoch": 1.12, "learning_rate": 6.58974358974359e-07, "loss": 0.0914, "step": 18725 }, { "epoch": 1.12, "learning_rate": 6.461538461538462e-07, "loss": 0.102, "step": 18750 }, { "epoch": 1.12, "learning_rate": 6.333333333333334e-07, "loss": 0.1315, "step": 18775 }, { "epoch": 1.12, "learning_rate": 6.205128205128206e-07, "loss": 0.1051, "step": 18800 }, { "epoch": 1.12, "learning_rate": 6.076923076923077e-07, "loss": 0.093, "step": 18825 }, { "epoch": 1.13, "learning_rate": 5.948717948717949e-07, "loss": 0.0999, "step": 18850 }, { "epoch": 1.13, "learning_rate": 5.820512820512821e-07, "loss": 0.0782, "step": 18875 }, { "epoch": 1.13, "learning_rate": 5.692307692307693e-07, "loss": 0.1069, "step": 18900 }, { "epoch": 1.13, "learning_rate": 5.564102564102564e-07, "loss": 0.1121, "step": 18925 }, { "epoch": 1.13, "learning_rate": 5.435897435897437e-07, "loss": 0.0984, "step": 18950 }, { "epoch": 1.13, "learning_rate": 5.307692307692308e-07, "loss": 0.1279, "step": 18975 }, { "epoch": 1.13, "learning_rate": 5.179487179487179e-07, "loss": 0.1472, "step": 19000 }, { "epoch": 1.13, "learning_rate": 5.051282051282052e-07, "loss": 0.1318, "step": 19025 }, { "epoch": 1.14, "learning_rate": 4.923076923076923e-07, "loss": 0.1324, "step": 19050 }, { "epoch": 1.14, "learning_rate": 4.794871794871795e-07, "loss": 0.1236, "step": 19075 }, { "epoch": 1.14, "learning_rate": 4.666666666666667e-07, "loss": 0.0828, "step": 19100 }, { "epoch": 1.14, "learning_rate": 4.5384615384615386e-07, "loss": 0.1349, "step": 19125 }, { "epoch": 1.14, "learning_rate": 4.4102564102564105e-07, "loss": 0.1043, "step": 19150 }, { "epoch": 1.14, "learning_rate": 4.2820512820512825e-07, "loss": 0.1362, "step": 19175 }, { "epoch": 1.14, "learning_rate": 4.153846153846154e-07, "loss": 0.1481, "step": 19200 }, { "epoch": 1.14, "learning_rate": 4.025641025641026e-07, "loss": 0.1191, "step": 19225 }, { "epoch": 1.15, "learning_rate": 3.897435897435898e-07, "loss": 0.1297, "step": 19250 }, { "epoch": 1.15, "learning_rate": 3.769230769230769e-07, "loss": 0.1298, "step": 19275 }, { "epoch": 1.15, "learning_rate": 3.641025641025641e-07, "loss": 0.141, "step": 19300 }, { "epoch": 1.15, "learning_rate": 3.512820512820513e-07, "loss": 0.0957, "step": 19325 }, { "epoch": 1.15, "learning_rate": 3.3846153846153845e-07, "loss": 0.1473, "step": 19350 }, { "epoch": 1.15, "learning_rate": 3.2564102564102565e-07, "loss": 0.1182, "step": 19375 }, { "epoch": 1.15, "learning_rate": 3.1282051282051284e-07, "loss": 0.1619, "step": 19400 }, { "epoch": 1.15, "learning_rate": 3.0000000000000004e-07, "loss": 0.0586, "step": 19425 }, { "epoch": 1.16, "learning_rate": 2.871794871794872e-07, "loss": 0.1108, "step": 19450 }, { "epoch": 1.16, "learning_rate": 2.743589743589744e-07, "loss": 0.0637, "step": 19475 }, { "epoch": 1.16, "learning_rate": 2.6153846153846157e-07, "loss": 0.1143, "step": 19500 }, { "epoch": 1.16, "learning_rate": 2.487179487179487e-07, "loss": 0.1529, "step": 19525 }, { "epoch": 1.16, "learning_rate": 2.3589743589743593e-07, "loss": 0.1034, "step": 19550 }, { "epoch": 1.16, "learning_rate": 2.2307692307692308e-07, "loss": 0.1014, "step": 19575 }, { "epoch": 1.16, "learning_rate": 2.1025641025641027e-07, "loss": 0.0913, "step": 19600 }, { "epoch": 1.16, "learning_rate": 1.9743589743589747e-07, "loss": 0.1101, "step": 19625 }, { "epoch": 1.17, "learning_rate": 1.8461538461538464e-07, "loss": 0.1272, "step": 19650 }, { "epoch": 1.17, "learning_rate": 1.717948717948718e-07, "loss": 0.1547, "step": 19675 }, { "epoch": 1.17, "learning_rate": 1.58974358974359e-07, "loss": 0.1522, "step": 19700 }, { "epoch": 1.17, "learning_rate": 1.4615384615384617e-07, "loss": 0.1595, "step": 19725 }, { "epoch": 1.17, "learning_rate": 1.3333333333333336e-07, "loss": 0.1017, "step": 19750 }, { "epoch": 1.17, "learning_rate": 1.2051282051282053e-07, "loss": 0.1707, "step": 19775 }, { "epoch": 1.17, "learning_rate": 1.076923076923077e-07, "loss": 0.1239, "step": 19800 }, { "epoch": 1.17, "learning_rate": 9.487179487179488e-08, "loss": 0.1607, "step": 19825 }, { "epoch": 1.18, "learning_rate": 8.205128205128206e-08, "loss": 0.0712, "step": 19850 }, { "epoch": 1.18, "learning_rate": 6.923076923076924e-08, "loss": 0.0959, "step": 19875 }, { "epoch": 1.18, "learning_rate": 5.641025641025642e-08, "loss": 0.1177, "step": 19900 }, { "epoch": 1.18, "learning_rate": 4.358974358974359e-08, "loss": 0.1559, "step": 19925 }, { "epoch": 1.18, "learning_rate": 3.076923076923077e-08, "loss": 0.1425, "step": 19950 }, { "epoch": 1.18, "learning_rate": 1.794871794871795e-08, "loss": 0.081, "step": 19975 }, { "epoch": 1.18, "learning_rate": 5.128205128205129e-09, "loss": 0.1165, "step": 20000 }, { "epoch": 1.18, "eval_loss": 0.2029048502445221, "eval_runtime": 18903.4699, "eval_samples_per_second": 0.864, "eval_steps_per_second": 0.864, "eval_wer": 8.323488285820279, "step": 20000 }, { "epoch": 1.18, "step": 20000, "total_flos": 4.08241963008e+19, "train_loss": 0.21553417257070542, "train_runtime": 212423.1023, "train_samples_per_second": 0.188, "train_steps_per_second": 0.094 } ], "max_steps": 20000, "num_train_epochs": 9223372036854775807, "total_flos": 4.08241963008e+19, "trial_name": null, "trial_params": null }