{ "best_metric": 27.36142051090262, "best_model_checkpoint": "./whisper-small-fa-aug/checkpoint-13000", "epoch": 8.0248, "global_step": 15000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.2000000000000006e-07, "loss": 3.8, "step": 25 }, { "epoch": 0.0, "learning_rate": 9.200000000000001e-07, "loss": 3.0497, "step": 50 }, { "epoch": 0.0, "learning_rate": 1.42e-06, "loss": 2.0102, "step": 75 }, { "epoch": 0.0, "learning_rate": 1.9200000000000003e-06, "loss": 1.5948, "step": 100 }, { "epoch": 0.0, "learning_rate": 2.42e-06, "loss": 1.4094, "step": 125 }, { "epoch": 0.0, "learning_rate": 2.92e-06, "loss": 1.2611, "step": 150 }, { "epoch": 0.0, "learning_rate": 3.4200000000000007e-06, "loss": 1.1366, "step": 175 }, { "epoch": 0.01, "learning_rate": 3.920000000000001e-06, "loss": 1.0817, "step": 200 }, { "epoch": 0.01, "learning_rate": 4.42e-06, "loss": 1.0324, "step": 225 }, { "epoch": 0.01, "learning_rate": 4.92e-06, "loss": 0.9176, "step": 250 }, { "epoch": 0.01, "learning_rate": 5.420000000000001e-06, "loss": 0.7975, "step": 275 }, { "epoch": 0.01, "learning_rate": 5.92e-06, "loss": 0.7079, "step": 300 }, { "epoch": 0.01, "learning_rate": 6.42e-06, "loss": 0.5832, "step": 325 }, { "epoch": 0.01, "learning_rate": 6.92e-06, "loss": 0.4469, "step": 350 }, { "epoch": 0.01, "learning_rate": 7.420000000000001e-06, "loss": 0.3913, "step": 375 }, { "epoch": 0.01, "learning_rate": 7.92e-06, "loss": 0.4189, "step": 400 }, { "epoch": 0.01, "learning_rate": 8.42e-06, "loss": 0.3756, "step": 425 }, { "epoch": 0.01, "learning_rate": 8.920000000000001e-06, "loss": 0.3876, "step": 450 }, { "epoch": 0.01, "learning_rate": 9.42e-06, "loss": 0.3802, "step": 475 }, { "epoch": 0.01, "learning_rate": 9.920000000000002e-06, "loss": 0.4061, "step": 500 }, { "epoch": 0.01, "learning_rate": 9.994683544303798e-06, "loss": 0.3885, "step": 525 }, { "epoch": 0.01, "learning_rate": 9.988354430379748e-06, "loss": 0.3449, "step": 550 }, { "epoch": 0.01, "learning_rate": 9.982025316455697e-06, "loss": 0.3292, "step": 575 }, { "epoch": 0.01, "learning_rate": 9.975696202531647e-06, "loss": 0.3439, "step": 600 }, { "epoch": 0.02, "learning_rate": 9.969367088607596e-06, "loss": 0.3031, "step": 625 }, { "epoch": 0.02, "learning_rate": 9.963037974683545e-06, "loss": 0.3159, "step": 650 }, { "epoch": 0.02, "learning_rate": 9.956708860759495e-06, "loss": 0.3304, "step": 675 }, { "epoch": 0.02, "learning_rate": 9.950379746835444e-06, "loss": 0.2918, "step": 700 }, { "epoch": 0.02, "learning_rate": 9.944050632911392e-06, "loss": 0.2785, "step": 725 }, { "epoch": 0.02, "learning_rate": 9.937721518987341e-06, "loss": 0.2618, "step": 750 }, { "epoch": 0.02, "learning_rate": 9.931392405063291e-06, "loss": 0.2862, "step": 775 }, { "epoch": 0.02, "learning_rate": 9.92506329113924e-06, "loss": 0.2962, "step": 800 }, { "epoch": 0.02, "learning_rate": 9.91873417721519e-06, "loss": 0.2935, "step": 825 }, { "epoch": 0.02, "learning_rate": 9.912405063291141e-06, "loss": 0.2827, "step": 850 }, { "epoch": 0.02, "learning_rate": 9.90607594936709e-06, "loss": 0.2702, "step": 875 }, { "epoch": 0.02, "learning_rate": 9.89974683544304e-06, "loss": 0.2987, "step": 900 }, { "epoch": 0.02, "learning_rate": 9.89341772151899e-06, "loss": 0.3067, "step": 925 }, { "epoch": 0.02, "learning_rate": 9.887088607594939e-06, "loss": 0.2962, "step": 950 }, { "epoch": 0.02, "learning_rate": 9.880759493670888e-06, "loss": 0.2807, "step": 975 }, { "epoch": 0.03, "learning_rate": 9.874430379746836e-06, "loss": 0.2533, "step": 1000 }, { "epoch": 0.03, "eval_loss": 0.3274388611316681, "eval_runtime": 1481.9106, "eval_samples_per_second": 7.025, "eval_steps_per_second": 0.439, "eval_wer": 40.657571667832, "step": 1000 }, { "epoch": 0.03, "learning_rate": 9.868101265822785e-06, "loss": 0.21, "step": 1025 }, { "epoch": 0.03, "learning_rate": 9.861772151898735e-06, "loss": 0.2463, "step": 1050 }, { "epoch": 0.03, "learning_rate": 9.855443037974684e-06, "loss": 0.2688, "step": 1075 }, { "epoch": 0.03, "learning_rate": 9.849113924050633e-06, "loss": 0.2293, "step": 1100 }, { "epoch": 0.03, "learning_rate": 9.842784810126583e-06, "loss": 0.2283, "step": 1125 }, { "epoch": 0.03, "learning_rate": 9.836455696202532e-06, "loss": 0.228, "step": 1150 }, { "epoch": 0.03, "learning_rate": 9.830126582278482e-06, "loss": 0.1919, "step": 1175 }, { "epoch": 0.03, "learning_rate": 9.823797468354431e-06, "loss": 0.2173, "step": 1200 }, { "epoch": 0.03, "learning_rate": 9.81746835443038e-06, "loss": 0.2066, "step": 1225 }, { "epoch": 0.03, "learning_rate": 9.81113924050633e-06, "loss": 0.1826, "step": 1250 }, { "epoch": 0.03, "learning_rate": 9.80481012658228e-06, "loss": 0.1855, "step": 1275 }, { "epoch": 0.03, "learning_rate": 9.798481012658229e-06, "loss": 0.1702, "step": 1300 }, { "epoch": 0.03, "learning_rate": 9.792151898734178e-06, "loss": 0.1714, "step": 1325 }, { "epoch": 0.03, "learning_rate": 9.785822784810128e-06, "loss": 0.1686, "step": 1350 }, { "epoch": 0.03, "learning_rate": 9.779493670886077e-06, "loss": 0.1762, "step": 1375 }, { "epoch": 0.04, "learning_rate": 9.773164556962027e-06, "loss": 0.1722, "step": 1400 }, { "epoch": 0.04, "learning_rate": 9.766835443037976e-06, "loss": 0.2051, "step": 1425 }, { "epoch": 0.04, "learning_rate": 9.760506329113924e-06, "loss": 0.2071, "step": 1450 }, { "epoch": 0.04, "learning_rate": 9.754177215189873e-06, "loss": 0.1664, "step": 1475 }, { "epoch": 0.04, "learning_rate": 9.747848101265823e-06, "loss": 0.1871, "step": 1500 }, { "epoch": 0.04, "learning_rate": 9.741518987341772e-06, "loss": 0.1893, "step": 1525 }, { "epoch": 0.04, "learning_rate": 9.735189873417721e-06, "loss": 0.1756, "step": 1550 }, { "epoch": 0.04, "learning_rate": 9.728860759493671e-06, "loss": 0.1687, "step": 1575 }, { "epoch": 0.04, "learning_rate": 9.72253164556962e-06, "loss": 0.1685, "step": 1600 }, { "epoch": 0.04, "learning_rate": 9.71620253164557e-06, "loss": 0.1805, "step": 1625 }, { "epoch": 0.04, "learning_rate": 9.70987341772152e-06, "loss": 0.1868, "step": 1650 }, { "epoch": 0.04, "learning_rate": 9.70354430379747e-06, "loss": 0.212, "step": 1675 }, { "epoch": 0.04, "learning_rate": 9.69721518987342e-06, "loss": 0.2019, "step": 1700 }, { "epoch": 0.04, "learning_rate": 9.69088607594937e-06, "loss": 0.2076, "step": 1725 }, { "epoch": 0.04, "learning_rate": 9.684556962025317e-06, "loss": 0.1971, "step": 1750 }, { "epoch": 1.0, "learning_rate": 9.678227848101266e-06, "loss": 0.2424, "step": 1775 }, { "epoch": 1.0, "learning_rate": 9.671898734177216e-06, "loss": 0.2346, "step": 1800 }, { "epoch": 1.0, "learning_rate": 9.665569620253165e-06, "loss": 0.2436, "step": 1825 }, { "epoch": 1.0, "learning_rate": 9.659240506329115e-06, "loss": 0.2613, "step": 1850 }, { "epoch": 1.0, "learning_rate": 9.652911392405064e-06, "loss": 0.2781, "step": 1875 }, { "epoch": 1.0, "learning_rate": 9.646582278481013e-06, "loss": 0.2465, "step": 1900 }, { "epoch": 1.0, "learning_rate": 9.640253164556963e-06, "loss": 0.2386, "step": 1925 }, { "epoch": 1.0, "learning_rate": 9.633924050632912e-06, "loss": 0.2298, "step": 1950 }, { "epoch": 1.01, "learning_rate": 9.627594936708862e-06, "loss": 0.264, "step": 1975 }, { "epoch": 1.01, "learning_rate": 9.621265822784811e-06, "loss": 0.2077, "step": 2000 }, { "epoch": 1.01, "eval_loss": 0.25397753715515137, "eval_runtime": 1477.9319, "eval_samples_per_second": 7.044, "eval_steps_per_second": 0.44, "eval_wer": 35.12252231743606, "step": 2000 }, { "epoch": 1.01, "learning_rate": 9.61493670886076e-06, "loss": 0.192, "step": 2025 }, { "epoch": 1.01, "learning_rate": 9.60860759493671e-06, "loss": 0.1903, "step": 2050 }, { "epoch": 1.01, "learning_rate": 9.60227848101266e-06, "loss": 0.2135, "step": 2075 }, { "epoch": 1.01, "learning_rate": 9.595949367088609e-06, "loss": 0.1917, "step": 2100 }, { "epoch": 1.01, "learning_rate": 9.589620253164558e-06, "loss": 0.1579, "step": 2125 }, { "epoch": 1.01, "learning_rate": 9.583291139240508e-06, "loss": 0.1779, "step": 2150 }, { "epoch": 1.01, "learning_rate": 9.576962025316457e-06, "loss": 0.1504, "step": 2175 }, { "epoch": 1.01, "learning_rate": 9.570632911392405e-06, "loss": 0.1574, "step": 2200 }, { "epoch": 1.01, "learning_rate": 9.564303797468354e-06, "loss": 0.1757, "step": 2225 }, { "epoch": 1.01, "learning_rate": 9.557974683544304e-06, "loss": 0.1996, "step": 2250 }, { "epoch": 1.01, "learning_rate": 9.551645569620253e-06, "loss": 0.1716, "step": 2275 }, { "epoch": 1.01, "learning_rate": 9.545316455696203e-06, "loss": 0.1509, "step": 2300 }, { "epoch": 1.01, "learning_rate": 9.538987341772152e-06, "loss": 0.1402, "step": 2325 }, { "epoch": 1.01, "learning_rate": 9.532658227848101e-06, "loss": 0.1612, "step": 2350 }, { "epoch": 1.02, "learning_rate": 9.526329113924051e-06, "loss": 0.1357, "step": 2375 }, { "epoch": 1.02, "learning_rate": 9.52e-06, "loss": 0.1426, "step": 2400 }, { "epoch": 1.02, "learning_rate": 9.51367088607595e-06, "loss": 0.1644, "step": 2425 }, { "epoch": 1.02, "learning_rate": 9.5073417721519e-06, "loss": 0.1196, "step": 2450 }, { "epoch": 1.02, "learning_rate": 9.501012658227849e-06, "loss": 0.1332, "step": 2475 }, { "epoch": 1.02, "learning_rate": 9.494683544303798e-06, "loss": 0.1096, "step": 2500 }, { "epoch": 1.02, "learning_rate": 9.488354430379747e-06, "loss": 0.1263, "step": 2525 }, { "epoch": 1.02, "learning_rate": 9.482025316455697e-06, "loss": 0.1467, "step": 2550 }, { "epoch": 1.02, "learning_rate": 9.475696202531646e-06, "loss": 0.1372, "step": 2575 }, { "epoch": 1.02, "learning_rate": 9.469367088607596e-06, "loss": 0.1256, "step": 2600 }, { "epoch": 1.02, "learning_rate": 9.463037974683545e-06, "loss": 0.1319, "step": 2625 }, { "epoch": 1.02, "learning_rate": 9.456708860759495e-06, "loss": 0.1385, "step": 2650 }, { "epoch": 1.02, "learning_rate": 9.450379746835444e-06, "loss": 0.147, "step": 2675 }, { "epoch": 1.02, "learning_rate": 9.444050632911393e-06, "loss": 0.1527, "step": 2700 }, { "epoch": 1.02, "learning_rate": 9.437721518987343e-06, "loss": 0.1357, "step": 2725 }, { "epoch": 1.02, "learning_rate": 9.431392405063292e-06, "loss": 0.122, "step": 2750 }, { "epoch": 1.03, "learning_rate": 9.425063291139242e-06, "loss": 0.0918, "step": 2775 }, { "epoch": 1.03, "learning_rate": 9.418734177215191e-06, "loss": 0.1218, "step": 2800 }, { "epoch": 1.03, "learning_rate": 9.41240506329114e-06, "loss": 0.139, "step": 2825 }, { "epoch": 1.03, "learning_rate": 9.40607594936709e-06, "loss": 0.1205, "step": 2850 }, { "epoch": 1.03, "learning_rate": 9.39974683544304e-06, "loss": 0.1158, "step": 2875 }, { "epoch": 1.03, "learning_rate": 9.393417721518989e-06, "loss": 0.1156, "step": 2900 }, { "epoch": 1.03, "learning_rate": 9.387088607594937e-06, "loss": 0.1044, "step": 2925 }, { "epoch": 1.03, "learning_rate": 9.380759493670886e-06, "loss": 0.112, "step": 2950 }, { "epoch": 1.03, "learning_rate": 9.374430379746835e-06, "loss": 0.1023, "step": 2975 }, { "epoch": 1.03, "learning_rate": 9.368101265822785e-06, "loss": 0.0966, "step": 3000 }, { "epoch": 1.03, "eval_loss": 0.2638459801673889, "eval_runtime": 1500.936, "eval_samples_per_second": 6.936, "eval_steps_per_second": 0.434, "eval_wer": 37.93069806013106, "step": 3000 }, { "epoch": 1.03, "learning_rate": 9.361772151898734e-06, "loss": 0.0942, "step": 3025 }, { "epoch": 1.03, "learning_rate": 9.355443037974684e-06, "loss": 0.0826, "step": 3050 }, { "epoch": 1.03, "learning_rate": 9.349113924050633e-06, "loss": 0.086, "step": 3075 }, { "epoch": 1.03, "learning_rate": 9.342784810126583e-06, "loss": 0.0956, "step": 3100 }, { "epoch": 1.03, "learning_rate": 9.336455696202532e-06, "loss": 0.0935, "step": 3125 }, { "epoch": 1.03, "learning_rate": 9.330126582278481e-06, "loss": 0.0923, "step": 3150 }, { "epoch": 1.04, "learning_rate": 9.323797468354431e-06, "loss": 0.1163, "step": 3175 }, { "epoch": 1.04, "learning_rate": 9.31746835443038e-06, "loss": 0.1099, "step": 3200 }, { "epoch": 1.04, "learning_rate": 9.31113924050633e-06, "loss": 0.0879, "step": 3225 }, { "epoch": 1.04, "learning_rate": 9.30481012658228e-06, "loss": 0.0994, "step": 3250 }, { "epoch": 1.04, "learning_rate": 9.298481012658229e-06, "loss": 0.1069, "step": 3275 }, { "epoch": 1.04, "learning_rate": 9.292151898734178e-06, "loss": 0.0957, "step": 3300 }, { "epoch": 1.04, "learning_rate": 9.285822784810127e-06, "loss": 0.0866, "step": 3325 }, { "epoch": 1.04, "learning_rate": 9.279493670886077e-06, "loss": 0.0965, "step": 3350 }, { "epoch": 1.04, "learning_rate": 9.273164556962026e-06, "loss": 0.0952, "step": 3375 }, { "epoch": 1.04, "learning_rate": 9.266835443037976e-06, "loss": 0.1178, "step": 3400 }, { "epoch": 1.04, "learning_rate": 9.260506329113925e-06, "loss": 0.1154, "step": 3425 }, { "epoch": 1.04, "learning_rate": 9.254177215189875e-06, "loss": 0.1075, "step": 3450 }, { "epoch": 1.04, "learning_rate": 9.247848101265824e-06, "loss": 0.1161, "step": 3475 }, { "epoch": 1.04, "learning_rate": 9.241518987341773e-06, "loss": 0.1153, "step": 3500 }, { "epoch": 2.0, "learning_rate": 9.235189873417723e-06, "loss": 0.1343, "step": 3525 }, { "epoch": 2.0, "learning_rate": 9.228860759493672e-06, "loss": 0.1374, "step": 3550 }, { "epoch": 2.0, "learning_rate": 9.222531645569622e-06, "loss": 0.1436, "step": 3575 }, { "epoch": 2.0, "learning_rate": 9.216202531645571e-06, "loss": 0.1561, "step": 3600 }, { "epoch": 2.0, "learning_rate": 9.20987341772152e-06, "loss": 0.1709, "step": 3625 }, { "epoch": 2.0, "learning_rate": 9.20354430379747e-06, "loss": 0.142, "step": 3650 }, { "epoch": 2.0, "learning_rate": 9.197215189873418e-06, "loss": 0.1389, "step": 3675 }, { "epoch": 2.0, "learning_rate": 9.190886075949367e-06, "loss": 0.1398, "step": 3700 }, { "epoch": 2.01, "learning_rate": 9.184556962025317e-06, "loss": 0.1526, "step": 3725 }, { "epoch": 2.01, "learning_rate": 9.178227848101266e-06, "loss": 0.1273, "step": 3750 }, { "epoch": 2.01, "learning_rate": 9.171898734177215e-06, "loss": 0.1119, "step": 3775 }, { "epoch": 2.01, "learning_rate": 9.165569620253165e-06, "loss": 0.1088, "step": 3800 }, { "epoch": 2.01, "learning_rate": 9.159493670886076e-06, "loss": 0.1362, "step": 3825 }, { "epoch": 2.01, "learning_rate": 9.153164556962026e-06, "loss": 0.1247, "step": 3850 }, { "epoch": 2.01, "learning_rate": 9.146835443037975e-06, "loss": 0.0974, "step": 3875 }, { "epoch": 2.01, "learning_rate": 9.140506329113925e-06, "loss": 0.1218, "step": 3900 }, { "epoch": 2.01, "learning_rate": 9.134177215189874e-06, "loss": 0.0846, "step": 3925 }, { "epoch": 2.01, "learning_rate": 9.127848101265823e-06, "loss": 0.0964, "step": 3950 }, { "epoch": 2.01, "learning_rate": 9.121518987341773e-06, "loss": 0.1166, "step": 3975 }, { "epoch": 2.01, "learning_rate": 9.115189873417722e-06, "loss": 0.1212, "step": 4000 }, { "epoch": 2.01, "eval_loss": 0.22517943382263184, "eval_runtime": 1537.556, "eval_samples_per_second": 6.771, "eval_steps_per_second": 0.423, "eval_wer": 69.04990325045935, "step": 4000 }, { "epoch": 2.01, "learning_rate": 9.108860759493672e-06, "loss": 0.1076, "step": 4025 }, { "epoch": 2.01, "learning_rate": 9.102531645569621e-06, "loss": 0.086, "step": 4050 }, { "epoch": 2.01, "learning_rate": 9.09620253164557e-06, "loss": 0.0773, "step": 4075 }, { "epoch": 2.01, "learning_rate": 9.08987341772152e-06, "loss": 0.098, "step": 4100 }, { "epoch": 2.02, "learning_rate": 9.08354430379747e-06, "loss": 0.0747, "step": 4125 }, { "epoch": 2.02, "learning_rate": 9.077215189873419e-06, "loss": 0.0895, "step": 4150 }, { "epoch": 2.02, "learning_rate": 9.070886075949368e-06, "loss": 0.0957, "step": 4175 }, { "epoch": 2.02, "learning_rate": 9.064556962025318e-06, "loss": 0.0789, "step": 4200 }, { "epoch": 2.02, "learning_rate": 9.058227848101267e-06, "loss": 0.0718, "step": 4225 }, { "epoch": 2.02, "learning_rate": 9.051898734177215e-06, "loss": 0.0737, "step": 4250 }, { "epoch": 2.02, "learning_rate": 9.045569620253164e-06, "loss": 0.0731, "step": 4275 }, { "epoch": 2.02, "learning_rate": 9.039240506329114e-06, "loss": 0.087, "step": 4300 }, { "epoch": 2.02, "learning_rate": 9.032911392405063e-06, "loss": 0.0875, "step": 4325 }, { "epoch": 2.02, "learning_rate": 9.026582278481013e-06, "loss": 0.0768, "step": 4350 }, { "epoch": 2.02, "learning_rate": 9.020253164556962e-06, "loss": 0.0842, "step": 4375 }, { "epoch": 2.02, "learning_rate": 9.013924050632911e-06, "loss": 0.092, "step": 4400 }, { "epoch": 2.02, "learning_rate": 9.007594936708861e-06, "loss": 0.0949, "step": 4425 }, { "epoch": 2.02, "learning_rate": 9.00126582278481e-06, "loss": 0.0989, "step": 4450 }, { "epoch": 2.02, "learning_rate": 8.99493670886076e-06, "loss": 0.0908, "step": 4475 }, { "epoch": 2.02, "learning_rate": 8.98860759493671e-06, "loss": 0.0783, "step": 4500 }, { "epoch": 2.03, "learning_rate": 8.982278481012659e-06, "loss": 0.0651, "step": 4525 }, { "epoch": 2.03, "learning_rate": 8.975949367088608e-06, "loss": 0.0791, "step": 4550 }, { "epoch": 2.03, "learning_rate": 8.969620253164557e-06, "loss": 0.0829, "step": 4575 }, { "epoch": 2.03, "learning_rate": 8.963291139240507e-06, "loss": 0.0797, "step": 4600 }, { "epoch": 2.03, "learning_rate": 8.956962025316456e-06, "loss": 0.0743, "step": 4625 }, { "epoch": 2.03, "learning_rate": 8.950632911392406e-06, "loss": 0.0814, "step": 4650 }, { "epoch": 2.03, "learning_rate": 8.944303797468355e-06, "loss": 0.0594, "step": 4675 }, { "epoch": 2.03, "learning_rate": 8.937974683544305e-06, "loss": 0.0709, "step": 4700 }, { "epoch": 2.03, "learning_rate": 8.931645569620254e-06, "loss": 0.0621, "step": 4725 }, { "epoch": 2.03, "learning_rate": 8.925316455696203e-06, "loss": 0.0599, "step": 4750 }, { "epoch": 2.03, "learning_rate": 8.918987341772153e-06, "loss": 0.0625, "step": 4775 }, { "epoch": 2.03, "learning_rate": 8.912658227848102e-06, "loss": 0.0541, "step": 4800 }, { "epoch": 2.03, "learning_rate": 8.906329113924052e-06, "loss": 0.059, "step": 4825 }, { "epoch": 2.03, "learning_rate": 8.900000000000001e-06, "loss": 0.0576, "step": 4850 }, { "epoch": 2.03, "learning_rate": 8.89367088607595e-06, "loss": 0.0623, "step": 4875 }, { "epoch": 2.03, "learning_rate": 8.8873417721519e-06, "loss": 0.0592, "step": 4900 }, { "epoch": 2.04, "learning_rate": 8.88101265822785e-06, "loss": 0.0692, "step": 4925 }, { "epoch": 2.04, "learning_rate": 8.874683544303799e-06, "loss": 0.077, "step": 4950 }, { "epoch": 2.04, "learning_rate": 8.868354430379748e-06, "loss": 0.0662, "step": 4975 }, { "epoch": 2.04, "learning_rate": 8.862025316455696e-06, "loss": 0.0605, "step": 5000 }, { "epoch": 2.04, "eval_loss": 0.2468421906232834, "eval_runtime": 1442.9116, "eval_samples_per_second": 7.215, "eval_steps_per_second": 0.451, "eval_wer": 33.46883689165677, "step": 5000 }, { "epoch": 2.04, "learning_rate": 8.855696202531645e-06, "loss": 0.0769, "step": 5025 }, { "epoch": 2.04, "learning_rate": 8.849367088607595e-06, "loss": 0.0601, "step": 5050 }, { "epoch": 2.04, "learning_rate": 8.843037974683544e-06, "loss": 0.0698, "step": 5075 }, { "epoch": 2.04, "learning_rate": 8.836708860759494e-06, "loss": 0.0593, "step": 5100 }, { "epoch": 2.04, "learning_rate": 8.830379746835443e-06, "loss": 0.0521, "step": 5125 }, { "epoch": 2.04, "learning_rate": 8.824050632911393e-06, "loss": 0.0687, "step": 5150 }, { "epoch": 2.04, "learning_rate": 8.817721518987342e-06, "loss": 0.0818, "step": 5175 }, { "epoch": 2.04, "learning_rate": 8.811392405063291e-06, "loss": 0.0706, "step": 5200 }, { "epoch": 2.04, "learning_rate": 8.805063291139241e-06, "loss": 0.0734, "step": 5225 }, { "epoch": 2.04, "learning_rate": 8.79873417721519e-06, "loss": 0.0795, "step": 5250 }, { "epoch": 3.0, "learning_rate": 8.79240506329114e-06, "loss": 0.0905, "step": 5275 }, { "epoch": 3.0, "learning_rate": 8.78607594936709e-06, "loss": 0.086, "step": 5300 }, { "epoch": 3.0, "learning_rate": 8.779746835443039e-06, "loss": 0.105, "step": 5325 }, { "epoch": 3.0, "learning_rate": 8.773417721518988e-06, "loss": 0.0981, "step": 5350 }, { "epoch": 3.0, "learning_rate": 8.767088607594937e-06, "loss": 0.1131, "step": 5375 }, { "epoch": 3.0, "learning_rate": 8.760759493670887e-06, "loss": 0.0916, "step": 5400 }, { "epoch": 3.0, "learning_rate": 8.754430379746836e-06, "loss": 0.0974, "step": 5425 }, { "epoch": 3.0, "learning_rate": 8.748101265822786e-06, "loss": 0.0899, "step": 5450 }, { "epoch": 3.01, "learning_rate": 8.741772151898735e-06, "loss": 0.1104, "step": 5475 }, { "epoch": 3.01, "learning_rate": 8.735443037974685e-06, "loss": 0.0818, "step": 5500 }, { "epoch": 3.01, "learning_rate": 8.729113924050634e-06, "loss": 0.0764, "step": 5525 }, { "epoch": 3.01, "learning_rate": 8.722784810126583e-06, "loss": 0.0709, "step": 5550 }, { "epoch": 3.01, "learning_rate": 8.716455696202533e-06, "loss": 0.0891, "step": 5575 }, { "epoch": 3.01, "learning_rate": 8.710126582278482e-06, "loss": 0.0869, "step": 5600 }, { "epoch": 3.01, "learning_rate": 8.703797468354432e-06, "loss": 0.0631, "step": 5625 }, { "epoch": 3.01, "learning_rate": 8.697468354430381e-06, "loss": 0.0795, "step": 5650 }, { "epoch": 3.01, "learning_rate": 8.69113924050633e-06, "loss": 0.0603, "step": 5675 }, { "epoch": 3.01, "learning_rate": 8.68481012658228e-06, "loss": 0.064, "step": 5700 }, { "epoch": 3.01, "learning_rate": 8.67848101265823e-06, "loss": 0.0738, "step": 5725 }, { "epoch": 3.01, "learning_rate": 8.672151898734177e-06, "loss": 0.0796, "step": 5750 }, { "epoch": 3.01, "learning_rate": 8.665822784810127e-06, "loss": 0.0723, "step": 5775 }, { "epoch": 3.01, "learning_rate": 8.659493670886076e-06, "loss": 0.0592, "step": 5800 }, { "epoch": 3.01, "learning_rate": 8.653164556962025e-06, "loss": 0.051, "step": 5825 }, { "epoch": 3.01, "learning_rate": 8.646835443037975e-06, "loss": 0.0716, "step": 5850 }, { "epoch": 3.02, "learning_rate": 8.640506329113924e-06, "loss": 0.0598, "step": 5875 }, { "epoch": 3.02, "learning_rate": 8.634177215189874e-06, "loss": 0.0524, "step": 5900 }, { "epoch": 3.02, "learning_rate": 8.627848101265823e-06, "loss": 0.0623, "step": 5925 }, { "epoch": 3.02, "learning_rate": 8.621518987341773e-06, "loss": 0.054, "step": 5950 }, { "epoch": 3.02, "learning_rate": 8.615189873417722e-06, "loss": 0.057, "step": 5975 }, { "epoch": 3.02, "learning_rate": 8.608860759493671e-06, "loss": 0.0479, "step": 6000 }, { "epoch": 3.02, "eval_loss": 0.24428367614746094, "eval_runtime": 1457.5568, "eval_samples_per_second": 7.143, "eval_steps_per_second": 0.447, "eval_wer": 36.51929299663409, "step": 6000 }, { "epoch": 3.02, "learning_rate": 8.602531645569621e-06, "loss": 0.0501, "step": 6025 }, { "epoch": 3.02, "learning_rate": 8.59620253164557e-06, "loss": 0.0527, "step": 6050 }, { "epoch": 3.02, "learning_rate": 8.58987341772152e-06, "loss": 0.0558, "step": 6075 }, { "epoch": 3.02, "learning_rate": 8.58354430379747e-06, "loss": 0.0548, "step": 6100 }, { "epoch": 3.02, "learning_rate": 8.577215189873419e-06, "loss": 0.0503, "step": 6125 }, { "epoch": 3.02, "learning_rate": 8.570886075949368e-06, "loss": 0.0641, "step": 6150 }, { "epoch": 3.02, "learning_rate": 8.564556962025316e-06, "loss": 0.0748, "step": 6175 }, { "epoch": 3.02, "learning_rate": 8.558227848101267e-06, "loss": 0.0775, "step": 6200 }, { "epoch": 3.02, "learning_rate": 8.551898734177216e-06, "loss": 0.0662, "step": 6225 }, { "epoch": 3.02, "learning_rate": 8.545569620253166e-06, "loss": 0.0591, "step": 6250 }, { "epoch": 3.03, "learning_rate": 8.539240506329115e-06, "loss": 0.0421, "step": 6275 }, { "epoch": 3.03, "learning_rate": 8.532911392405065e-06, "loss": 0.0565, "step": 6300 }, { "epoch": 3.03, "learning_rate": 8.526582278481014e-06, "loss": 0.0574, "step": 6325 }, { "epoch": 3.03, "learning_rate": 8.520253164556963e-06, "loss": 0.0466, "step": 6350 }, { "epoch": 3.03, "learning_rate": 8.513924050632913e-06, "loss": 0.0486, "step": 6375 }, { "epoch": 3.03, "learning_rate": 8.507594936708862e-06, "loss": 0.0502, "step": 6400 }, { "epoch": 3.03, "learning_rate": 8.501265822784812e-06, "loss": 0.0415, "step": 6425 }, { "epoch": 3.03, "learning_rate": 8.494936708860761e-06, "loss": 0.0449, "step": 6450 }, { "epoch": 3.03, "learning_rate": 8.488607594936709e-06, "loss": 0.0525, "step": 6475 }, { "epoch": 3.03, "learning_rate": 8.482278481012658e-06, "loss": 0.0401, "step": 6500 }, { "epoch": 3.03, "learning_rate": 8.475949367088608e-06, "loss": 0.0502, "step": 6525 }, { "epoch": 3.03, "learning_rate": 8.469620253164557e-06, "loss": 0.0453, "step": 6550 }, { "epoch": 3.03, "learning_rate": 8.463291139240507e-06, "loss": 0.0442, "step": 6575 }, { "epoch": 3.03, "learning_rate": 8.456962025316456e-06, "loss": 0.0423, "step": 6600 }, { "epoch": 3.03, "learning_rate": 8.450632911392405e-06, "loss": 0.0518, "step": 6625 }, { "epoch": 3.03, "learning_rate": 8.444303797468355e-06, "loss": 0.0479, "step": 6650 }, { "epoch": 3.04, "learning_rate": 8.437974683544304e-06, "loss": 0.0521, "step": 6675 }, { "epoch": 3.04, "learning_rate": 8.431645569620254e-06, "loss": 0.0436, "step": 6700 }, { "epoch": 3.04, "learning_rate": 8.425316455696203e-06, "loss": 0.0539, "step": 6725 }, { "epoch": 3.04, "learning_rate": 8.418987341772153e-06, "loss": 0.0446, "step": 6750 }, { "epoch": 3.04, "learning_rate": 8.412658227848102e-06, "loss": 0.0521, "step": 6775 }, { "epoch": 3.04, "learning_rate": 8.406329113924051e-06, "loss": 0.0391, "step": 6800 }, { "epoch": 3.04, "learning_rate": 8.400000000000001e-06, "loss": 0.0513, "step": 6825 }, { "epoch": 3.04, "learning_rate": 8.39367088607595e-06, "loss": 0.0449, "step": 6850 }, { "epoch": 3.04, "learning_rate": 8.3873417721519e-06, "loss": 0.0463, "step": 6875 }, { "epoch": 3.04, "learning_rate": 8.38101265822785e-06, "loss": 0.0495, "step": 6900 }, { "epoch": 3.04, "learning_rate": 8.374683544303797e-06, "loss": 0.0628, "step": 6925 }, { "epoch": 3.04, "learning_rate": 8.368354430379746e-06, "loss": 0.0594, "step": 6950 }, { "epoch": 3.04, "learning_rate": 8.362025316455696e-06, "loss": 0.0573, "step": 6975 }, { "epoch": 3.04, "learning_rate": 8.355696202531645e-06, "loss": 0.057, "step": 7000 }, { "epoch": 3.04, "eval_loss": 0.25129273533821106, "eval_runtime": 1479.2043, "eval_samples_per_second": 7.038, "eval_steps_per_second": 0.44, "eval_wer": 57.010683100538216, "step": 7000 }, { "epoch": 4.0, "learning_rate": 8.349367088607596e-06, "loss": 0.0534, "step": 7025 }, { "epoch": 4.0, "learning_rate": 8.343037974683546e-06, "loss": 0.0663, "step": 7050 }, { "epoch": 4.0, "learning_rate": 8.336708860759495e-06, "loss": 0.0698, "step": 7075 }, { "epoch": 4.0, "learning_rate": 8.330379746835445e-06, "loss": 0.0633, "step": 7100 }, { "epoch": 4.0, "learning_rate": 8.324050632911394e-06, "loss": 0.0808, "step": 7125 }, { "epoch": 4.0, "learning_rate": 8.317721518987343e-06, "loss": 0.0699, "step": 7150 }, { "epoch": 4.0, "learning_rate": 8.311392405063293e-06, "loss": 0.0736, "step": 7175 }, { "epoch": 4.0, "learning_rate": 8.305063291139242e-06, "loss": 0.069, "step": 7200 }, { "epoch": 4.01, "learning_rate": 8.29873417721519e-06, "loss": 0.0718, "step": 7225 }, { "epoch": 4.01, "learning_rate": 8.29240506329114e-06, "loss": 0.0674, "step": 7250 }, { "epoch": 4.01, "learning_rate": 8.286075949367089e-06, "loss": 0.0568, "step": 7275 }, { "epoch": 4.01, "learning_rate": 8.279746835443038e-06, "loss": 0.0497, "step": 7300 }, { "epoch": 4.01, "learning_rate": 8.273417721518988e-06, "loss": 0.0563, "step": 7325 }, { "epoch": 4.01, "learning_rate": 8.267088607594937e-06, "loss": 0.0649, "step": 7350 }, { "epoch": 4.01, "learning_rate": 8.260759493670887e-06, "loss": 0.042, "step": 7375 }, { "epoch": 4.01, "learning_rate": 8.254430379746836e-06, "loss": 0.0559, "step": 7400 }, { "epoch": 4.01, "learning_rate": 8.248101265822785e-06, "loss": 0.0476, "step": 7425 }, { "epoch": 4.01, "learning_rate": 8.241772151898735e-06, "loss": 0.0427, "step": 7450 }, { "epoch": 4.01, "learning_rate": 8.235443037974684e-06, "loss": 0.0533, "step": 7475 }, { "epoch": 4.01, "learning_rate": 8.229113924050634e-06, "loss": 0.0633, "step": 7500 }, { "epoch": 4.01, "learning_rate": 8.222784810126583e-06, "loss": 0.055, "step": 7525 }, { "epoch": 4.01, "learning_rate": 8.216455696202533e-06, "loss": 0.043, "step": 7550 }, { "epoch": 4.01, "learning_rate": 8.210126582278482e-06, "loss": 0.0421, "step": 7575 }, { "epoch": 4.01, "learning_rate": 8.203797468354431e-06, "loss": 0.0519, "step": 7600 }, { "epoch": 4.02, "learning_rate": 8.197468354430381e-06, "loss": 0.0398, "step": 7625 }, { "epoch": 4.02, "learning_rate": 8.19113924050633e-06, "loss": 0.041, "step": 7650 }, { "epoch": 4.02, "learning_rate": 8.184810126582278e-06, "loss": 0.0477, "step": 7675 }, { "epoch": 4.02, "learning_rate": 8.178481012658227e-06, "loss": 0.046, "step": 7700 }, { "epoch": 4.02, "learning_rate": 8.172151898734177e-06, "loss": 0.0465, "step": 7725 }, { "epoch": 4.02, "learning_rate": 8.165822784810126e-06, "loss": 0.0359, "step": 7750 }, { "epoch": 4.02, "learning_rate": 8.159493670886076e-06, "loss": 0.0362, "step": 7775 }, { "epoch": 4.02, "learning_rate": 8.153164556962025e-06, "loss": 0.04, "step": 7800 }, { "epoch": 4.02, "learning_rate": 8.146835443037975e-06, "loss": 0.0433, "step": 7825 }, { "epoch": 4.02, "learning_rate": 8.140506329113926e-06, "loss": 0.0399, "step": 7850 }, { "epoch": 4.02, "learning_rate": 8.134177215189875e-06, "loss": 0.0396, "step": 7875 }, { "epoch": 4.02, "learning_rate": 8.127848101265825e-06, "loss": 0.0472, "step": 7900 }, { "epoch": 4.02, "learning_rate": 8.121518987341774e-06, "loss": 0.0533, "step": 7925 }, { "epoch": 4.02, "learning_rate": 8.115189873417722e-06, "loss": 0.0564, "step": 7950 }, { "epoch": 4.02, "learning_rate": 8.108860759493671e-06, "loss": 0.0508, "step": 7975 }, { "epoch": 4.02, "learning_rate": 8.10253164556962e-06, "loss": 0.0375, "step": 8000 }, { "epoch": 4.02, "eval_loss": 0.252152681350708, "eval_runtime": 1456.2607, "eval_samples_per_second": 7.149, "eval_steps_per_second": 0.447, "eval_wer": 44.38120945056017, "step": 8000 }, { "epoch": 4.03, "learning_rate": 8.09620253164557e-06, "loss": 0.0296, "step": 8025 }, { "epoch": 4.03, "learning_rate": 8.08987341772152e-06, "loss": 0.0395, "step": 8050 }, { "epoch": 4.03, "learning_rate": 8.083544303797469e-06, "loss": 0.0409, "step": 8075 }, { "epoch": 4.03, "learning_rate": 8.077215189873418e-06, "loss": 0.0343, "step": 8100 }, { "epoch": 4.03, "learning_rate": 8.070886075949368e-06, "loss": 0.0321, "step": 8125 }, { "epoch": 4.03, "learning_rate": 8.064556962025317e-06, "loss": 0.0333, "step": 8150 }, { "epoch": 4.03, "learning_rate": 8.058227848101267e-06, "loss": 0.0327, "step": 8175 }, { "epoch": 4.03, "learning_rate": 8.051898734177216e-06, "loss": 0.0318, "step": 8200 }, { "epoch": 4.03, "learning_rate": 8.045569620253165e-06, "loss": 0.0405, "step": 8225 }, { "epoch": 4.03, "learning_rate": 8.039240506329115e-06, "loss": 0.0382, "step": 8250 }, { "epoch": 4.03, "learning_rate": 8.032911392405064e-06, "loss": 0.0317, "step": 8275 }, { "epoch": 4.03, "learning_rate": 8.026582278481014e-06, "loss": 0.0285, "step": 8300 }, { "epoch": 4.03, "learning_rate": 8.020253164556963e-06, "loss": 0.0285, "step": 8325 }, { "epoch": 4.03, "learning_rate": 8.013924050632913e-06, "loss": 0.0269, "step": 8350 }, { "epoch": 4.03, "learning_rate": 8.007594936708862e-06, "loss": 0.0334, "step": 8375 }, { "epoch": 4.03, "learning_rate": 8.00126582278481e-06, "loss": 0.0406, "step": 8400 }, { "epoch": 4.04, "learning_rate": 7.99493670886076e-06, "loss": 0.038, "step": 8425 }, { "epoch": 4.04, "learning_rate": 7.988607594936709e-06, "loss": 0.0315, "step": 8450 }, { "epoch": 4.04, "learning_rate": 7.982278481012658e-06, "loss": 0.0329, "step": 8475 }, { "epoch": 4.04, "learning_rate": 7.975949367088607e-06, "loss": 0.0373, "step": 8500 }, { "epoch": 4.04, "learning_rate": 7.969620253164557e-06, "loss": 0.0514, "step": 8525 }, { "epoch": 4.04, "learning_rate": 7.963291139240506e-06, "loss": 0.0349, "step": 8550 }, { "epoch": 4.04, "learning_rate": 7.956962025316456e-06, "loss": 0.0385, "step": 8575 }, { "epoch": 4.04, "learning_rate": 7.950632911392405e-06, "loss": 0.0367, "step": 8600 }, { "epoch": 4.04, "learning_rate": 7.944303797468355e-06, "loss": 0.0317, "step": 8625 }, { "epoch": 4.04, "learning_rate": 7.937974683544304e-06, "loss": 0.0368, "step": 8650 }, { "epoch": 4.04, "learning_rate": 7.931645569620255e-06, "loss": 0.045, "step": 8675 }, { "epoch": 4.04, "learning_rate": 7.925316455696203e-06, "loss": 0.0456, "step": 8700 }, { "epoch": 4.04, "learning_rate": 7.918987341772152e-06, "loss": 0.0421, "step": 8725 }, { "epoch": 4.04, "learning_rate": 7.912658227848102e-06, "loss": 0.043, "step": 8750 }, { "epoch": 5.0, "learning_rate": 7.906329113924051e-06, "loss": 0.0428, "step": 8775 }, { "epoch": 5.0, "learning_rate": 7.9e-06, "loss": 0.0459, "step": 8800 }, { "epoch": 5.0, "learning_rate": 7.89367088607595e-06, "loss": 0.0467, "step": 8825 }, { "epoch": 5.0, "learning_rate": 7.8873417721519e-06, "loss": 0.0462, "step": 8850 }, { "epoch": 5.0, "learning_rate": 7.881012658227849e-06, "loss": 0.0578, "step": 8875 }, { "epoch": 5.0, "learning_rate": 7.874683544303798e-06, "loss": 0.0422, "step": 8900 }, { "epoch": 5.0, "learning_rate": 7.868354430379748e-06, "loss": 0.0546, "step": 8925 }, { "epoch": 5.0, "learning_rate": 7.862025316455697e-06, "loss": 0.0515, "step": 8950 }, { "epoch": 5.01, "learning_rate": 7.855696202531647e-06, "loss": 0.0541, "step": 8975 }, { "epoch": 5.01, "learning_rate": 7.849367088607596e-06, "loss": 0.0479, "step": 9000 }, { "epoch": 5.01, "eval_loss": 0.2375001162290573, "eval_runtime": 1456.1088, "eval_samples_per_second": 7.15, "eval_steps_per_second": 0.447, "eval_wer": 31.58262735979447, "step": 9000 }, { "epoch": 5.01, "learning_rate": 7.843037974683545e-06, "loss": 0.0434, "step": 9025 }, { "epoch": 5.01, "learning_rate": 7.836708860759495e-06, "loss": 0.033, "step": 9050 }, { "epoch": 5.01, "learning_rate": 7.830379746835444e-06, "loss": 0.0494, "step": 9075 }, { "epoch": 5.01, "learning_rate": 7.824050632911394e-06, "loss": 0.0522, "step": 9100 }, { "epoch": 5.01, "learning_rate": 7.817721518987343e-06, "loss": 0.0338, "step": 9125 }, { "epoch": 5.01, "learning_rate": 7.811392405063291e-06, "loss": 0.0395, "step": 9150 }, { "epoch": 5.01, "learning_rate": 7.80506329113924e-06, "loss": 0.0332, "step": 9175 }, { "epoch": 5.01, "learning_rate": 7.79873417721519e-06, "loss": 0.0381, "step": 9200 }, { "epoch": 5.01, "learning_rate": 7.79240506329114e-06, "loss": 0.0409, "step": 9225 }, { "epoch": 5.01, "learning_rate": 7.786075949367089e-06, "loss": 0.0529, "step": 9250 }, { "epoch": 5.01, "learning_rate": 7.779746835443038e-06, "loss": 0.0498, "step": 9275 }, { "epoch": 5.01, "learning_rate": 7.773417721518987e-06, "loss": 0.0379, "step": 9300 }, { "epoch": 5.01, "learning_rate": 7.767088607594937e-06, "loss": 0.0316, "step": 9325 }, { "epoch": 5.01, "learning_rate": 7.760759493670886e-06, "loss": 0.0398, "step": 9350 }, { "epoch": 5.02, "learning_rate": 7.754430379746836e-06, "loss": 0.0427, "step": 9375 }, { "epoch": 5.02, "learning_rate": 7.748101265822785e-06, "loss": 0.0363, "step": 9400 }, { "epoch": 5.02, "learning_rate": 7.741772151898735e-06, "loss": 0.0393, "step": 9425 }, { "epoch": 5.02, "learning_rate": 7.735443037974684e-06, "loss": 0.0377, "step": 9450 }, { "epoch": 5.02, "learning_rate": 7.729113924050633e-06, "loss": 0.0313, "step": 9475 }, { "epoch": 5.02, "learning_rate": 7.722784810126583e-06, "loss": 0.0359, "step": 9500 }, { "epoch": 5.02, "learning_rate": 7.716455696202532e-06, "loss": 0.0289, "step": 9525 }, { "epoch": 5.02, "learning_rate": 7.710126582278482e-06, "loss": 0.0354, "step": 9550 }, { "epoch": 5.02, "learning_rate": 7.703797468354431e-06, "loss": 0.0309, "step": 9575 }, { "epoch": 5.02, "learning_rate": 7.69746835443038e-06, "loss": 0.0321, "step": 9600 }, { "epoch": 5.02, "learning_rate": 7.69113924050633e-06, "loss": 0.0392, "step": 9625 }, { "epoch": 5.02, "learning_rate": 7.68481012658228e-06, "loss": 0.0362, "step": 9650 }, { "epoch": 5.02, "learning_rate": 7.678481012658229e-06, "loss": 0.0358, "step": 9675 }, { "epoch": 5.02, "learning_rate": 7.672151898734178e-06, "loss": 0.045, "step": 9700 }, { "epoch": 5.02, "learning_rate": 7.665822784810128e-06, "loss": 0.0401, "step": 9725 }, { "epoch": 5.02, "learning_rate": 7.659493670886077e-06, "loss": 0.0357, "step": 9750 }, { "epoch": 5.03, "learning_rate": 7.653164556962027e-06, "loss": 0.0256, "step": 9775 }, { "epoch": 5.03, "learning_rate": 7.646835443037976e-06, "loss": 0.0358, "step": 9800 }, { "epoch": 5.03, "learning_rate": 7.640506329113925e-06, "loss": 0.0361, "step": 9825 }, { "epoch": 5.03, "learning_rate": 7.634177215189875e-06, "loss": 0.0365, "step": 9850 }, { "epoch": 5.03, "learning_rate": 7.6278481012658234e-06, "loss": 0.0344, "step": 9875 }, { "epoch": 5.03, "learning_rate": 7.621518987341773e-06, "loss": 0.0349, "step": 9900 }, { "epoch": 5.03, "learning_rate": 7.615189873417722e-06, "loss": 0.0289, "step": 9925 }, { "epoch": 5.03, "learning_rate": 7.608860759493672e-06, "loss": 0.0259, "step": 9950 }, { "epoch": 5.03, "learning_rate": 7.602531645569621e-06, "loss": 0.0289, "step": 9975 }, { "epoch": 5.03, "learning_rate": 7.59620253164557e-06, "loss": 0.0269, "step": 10000 }, { "epoch": 5.03, "eval_loss": 0.2615405321121216, "eval_runtime": 1420.7197, "eval_samples_per_second": 7.328, "eval_steps_per_second": 0.458, "eval_wer": 36.23310948145498, "step": 10000 }, { "epoch": 5.03, "learning_rate": 7.589873417721519e-06, "loss": 0.0261, "step": 10025 }, { "epoch": 5.03, "learning_rate": 7.583544303797469e-06, "loss": 0.0278, "step": 10050 }, { "epoch": 5.03, "learning_rate": 7.577215189873418e-06, "loss": 0.0248, "step": 10075 }, { "epoch": 5.03, "learning_rate": 7.5708860759493674e-06, "loss": 0.0263, "step": 10100 }, { "epoch": 5.03, "learning_rate": 7.564556962025317e-06, "loss": 0.0253, "step": 10125 }, { "epoch": 5.03, "learning_rate": 7.558227848101266e-06, "loss": 0.0248, "step": 10150 }, { "epoch": 5.04, "learning_rate": 7.551898734177216e-06, "loss": 0.0281, "step": 10175 }, { "epoch": 5.04, "learning_rate": 7.545569620253165e-06, "loss": 0.0276, "step": 10200 }, { "epoch": 5.04, "learning_rate": 7.539240506329114e-06, "loss": 0.0285, "step": 10225 }, { "epoch": 5.04, "learning_rate": 7.532911392405063e-06, "loss": 0.0265, "step": 10250 }, { "epoch": 5.04, "learning_rate": 7.526582278481013e-06, "loss": 0.0288, "step": 10275 }, { "epoch": 5.04, "learning_rate": 7.520253164556963e-06, "loss": 0.0247, "step": 10300 }, { "epoch": 5.04, "learning_rate": 7.513924050632912e-06, "loss": 0.0281, "step": 10325 }, { "epoch": 5.04, "learning_rate": 7.507594936708862e-06, "loss": 0.0262, "step": 10350 }, { "epoch": 5.04, "learning_rate": 7.501265822784811e-06, "loss": 0.0243, "step": 10375 }, { "epoch": 5.04, "learning_rate": 7.494936708860761e-06, "loss": 0.0331, "step": 10400 }, { "epoch": 5.04, "learning_rate": 7.48860759493671e-06, "loss": 0.0346, "step": 10425 }, { "epoch": 5.04, "learning_rate": 7.4822784810126594e-06, "loss": 0.0348, "step": 10450 }, { "epoch": 5.04, "learning_rate": 7.475949367088609e-06, "loss": 0.0272, "step": 10475 }, { "epoch": 5.04, "learning_rate": 7.4696202531645574e-06, "loss": 0.0326, "step": 10500 }, { "epoch": 6.0, "learning_rate": 7.463291139240507e-06, "loss": 0.033, "step": 10525 }, { "epoch": 6.0, "learning_rate": 7.456962025316456e-06, "loss": 0.0333, "step": 10550 }, { "epoch": 6.0, "learning_rate": 7.450632911392406e-06, "loss": 0.0371, "step": 10575 }, { "epoch": 6.0, "learning_rate": 7.444303797468355e-06, "loss": 0.0359, "step": 10600 }, { "epoch": 6.0, "learning_rate": 7.437974683544305e-06, "loss": 0.0438, "step": 10625 }, { "epoch": 6.0, "learning_rate": 7.431645569620254e-06, "loss": 0.0382, "step": 10650 }, { "epoch": 6.0, "learning_rate": 7.4253164556962034e-06, "loss": 0.0337, "step": 10675 }, { "epoch": 6.0, "learning_rate": 7.418987341772153e-06, "loss": 0.0422, "step": 10700 }, { "epoch": 6.01, "learning_rate": 7.4126582278481014e-06, "loss": 0.0467, "step": 10725 }, { "epoch": 6.01, "learning_rate": 7.406329113924051e-06, "loss": 0.0437, "step": 10750 }, { "epoch": 6.01, "learning_rate": 7.4e-06, "loss": 0.0328, "step": 10775 }, { "epoch": 6.01, "learning_rate": 7.39367088607595e-06, "loss": 0.0295, "step": 10800 }, { "epoch": 6.01, "learning_rate": 7.387341772151899e-06, "loss": 0.0431, "step": 10825 }, { "epoch": 6.01, "learning_rate": 7.381012658227849e-06, "loss": 0.0456, "step": 10850 }, { "epoch": 6.01, "learning_rate": 7.374683544303798e-06, "loss": 0.0267, "step": 10875 }, { "epoch": 6.01, "learning_rate": 7.3683544303797474e-06, "loss": 0.0334, "step": 10900 }, { "epoch": 6.01, "learning_rate": 7.362025316455697e-06, "loss": 0.0296, "step": 10925 }, { "epoch": 6.01, "learning_rate": 7.3556962025316454e-06, "loss": 0.0261, "step": 10950 }, { "epoch": 6.01, "learning_rate": 7.349367088607595e-06, "loss": 0.0299, "step": 10975 }, { "epoch": 6.01, "learning_rate": 7.343037974683544e-06, "loss": 0.0379, "step": 11000 }, { "epoch": 6.01, "eval_loss": 0.23971551656723022, "eval_runtime": 1435.6772, "eval_samples_per_second": 7.252, "eval_steps_per_second": 0.453, "eval_wer": 31.7712483129807, "step": 11000 }, { "epoch": 6.01, "learning_rate": 7.336708860759494e-06, "loss": 0.0457, "step": 11025 }, { "epoch": 6.01, "learning_rate": 7.330379746835443e-06, "loss": 0.0275, "step": 11050 }, { "epoch": 6.01, "learning_rate": 7.324050632911393e-06, "loss": 0.0254, "step": 11075 }, { "epoch": 6.01, "learning_rate": 7.317721518987342e-06, "loss": 0.035, "step": 11100 }, { "epoch": 6.02, "learning_rate": 7.311392405063292e-06, "loss": 0.0273, "step": 11125 }, { "epoch": 6.02, "learning_rate": 7.305063291139242e-06, "loss": 0.0227, "step": 11150 }, { "epoch": 6.02, "learning_rate": 7.298734177215191e-06, "loss": 0.0263, "step": 11175 }, { "epoch": 6.02, "learning_rate": 7.2924050632911406e-06, "loss": 0.0294, "step": 11200 }, { "epoch": 6.02, "learning_rate": 7.28607594936709e-06, "loss": 0.026, "step": 11225 }, { "epoch": 6.02, "learning_rate": 7.279746835443039e-06, "loss": 0.0251, "step": 11250 }, { "epoch": 6.02, "learning_rate": 7.273417721518988e-06, "loss": 0.024, "step": 11275 }, { "epoch": 6.02, "learning_rate": 7.2670886075949374e-06, "loss": 0.0344, "step": 11300 }, { "epoch": 6.02, "learning_rate": 7.260759493670887e-06, "loss": 0.0338, "step": 11325 }, { "epoch": 6.02, "learning_rate": 7.254430379746836e-06, "loss": 0.0266, "step": 11350 }, { "epoch": 6.02, "learning_rate": 7.248101265822786e-06, "loss": 0.0271, "step": 11375 }, { "epoch": 6.02, "learning_rate": 7.241772151898735e-06, "loss": 0.0338, "step": 11400 }, { "epoch": 6.02, "learning_rate": 7.2354430379746846e-06, "loss": 0.0321, "step": 11425 }, { "epoch": 6.02, "learning_rate": 7.229113924050634e-06, "loss": 0.0385, "step": 11450 }, { "epoch": 6.02, "learning_rate": 7.222784810126583e-06, "loss": 0.0284, "step": 11475 }, { "epoch": 6.02, "learning_rate": 7.216455696202532e-06, "loss": 0.0268, "step": 11500 }, { "epoch": 6.03, "learning_rate": 7.2101265822784814e-06, "loss": 0.0191, "step": 11525 }, { "epoch": 6.03, "learning_rate": 7.203797468354431e-06, "loss": 0.0243, "step": 11550 }, { "epoch": 6.03, "learning_rate": 7.19746835443038e-06, "loss": 0.0228, "step": 11575 }, { "epoch": 6.03, "learning_rate": 7.19113924050633e-06, "loss": 0.0187, "step": 11600 }, { "epoch": 6.03, "learning_rate": 7.184810126582279e-06, "loss": 0.0255, "step": 11625 }, { "epoch": 6.03, "learning_rate": 7.1784810126582286e-06, "loss": 0.0218, "step": 11650 }, { "epoch": 6.03, "learning_rate": 7.172151898734178e-06, "loss": 0.0249, "step": 11675 }, { "epoch": 6.03, "learning_rate": 7.165822784810127e-06, "loss": 0.0254, "step": 11700 }, { "epoch": 6.03, "learning_rate": 7.159493670886076e-06, "loss": 0.0323, "step": 11725 }, { "epoch": 6.03, "learning_rate": 7.1531645569620254e-06, "loss": 0.021, "step": 11750 }, { "epoch": 6.03, "learning_rate": 7.146835443037975e-06, "loss": 0.0242, "step": 11775 }, { "epoch": 6.03, "learning_rate": 7.140506329113924e-06, "loss": 0.0233, "step": 11800 }, { "epoch": 6.03, "learning_rate": 7.134177215189874e-06, "loss": 0.0237, "step": 11825 }, { "epoch": 6.03, "learning_rate": 7.127848101265823e-06, "loss": 0.0257, "step": 11850 }, { "epoch": 6.03, "learning_rate": 7.121518987341773e-06, "loss": 0.0256, "step": 11875 }, { "epoch": 6.03, "learning_rate": 7.115189873417722e-06, "loss": 0.0264, "step": 11900 }, { "epoch": 6.04, "learning_rate": 7.108860759493671e-06, "loss": 0.0254, "step": 11925 }, { "epoch": 6.04, "learning_rate": 7.102531645569622e-06, "loss": 0.025, "step": 11950 }, { "epoch": 6.04, "learning_rate": 7.096202531645571e-06, "loss": 0.0219, "step": 11975 }, { "epoch": 6.04, "learning_rate": 7.08987341772152e-06, "loss": 0.0241, "step": 12000 }, { "epoch": 6.04, "eval_loss": 0.26222535967826843, "eval_runtime": 1460.2321, "eval_samples_per_second": 7.13, "eval_steps_per_second": 0.446, "eval_wer": 32.87370526350022, "step": 12000 }, { "epoch": 6.04, "learning_rate": 7.083544303797469e-06, "loss": 0.0337, "step": 12025 }, { "epoch": 6.04, "learning_rate": 7.0772151898734186e-06, "loss": 0.0179, "step": 12050 }, { "epoch": 6.04, "learning_rate": 7.070886075949368e-06, "loss": 0.0174, "step": 12075 }, { "epoch": 6.04, "learning_rate": 7.0645569620253174e-06, "loss": 0.0211, "step": 12100 }, { "epoch": 6.04, "learning_rate": 7.058227848101267e-06, "loss": 0.0224, "step": 12125 }, { "epoch": 6.04, "learning_rate": 7.051898734177216e-06, "loss": 0.0226, "step": 12150 }, { "epoch": 6.04, "learning_rate": 7.045569620253166e-06, "loss": 0.0271, "step": 12175 }, { "epoch": 6.04, "learning_rate": 7.039240506329114e-06, "loss": 0.0264, "step": 12200 }, { "epoch": 6.04, "learning_rate": 7.032911392405064e-06, "loss": 0.031, "step": 12225 }, { "epoch": 6.04, "learning_rate": 7.026582278481013e-06, "loss": 0.022, "step": 12250 }, { "epoch": 7.0, "learning_rate": 7.0202531645569626e-06, "loss": 0.0266, "step": 12275 }, { "epoch": 7.0, "learning_rate": 7.013924050632912e-06, "loss": 0.0269, "step": 12300 }, { "epoch": 7.0, "learning_rate": 7.0075949367088614e-06, "loss": 0.028, "step": 12325 }, { "epoch": 7.0, "learning_rate": 7.001265822784811e-06, "loss": 0.0361, "step": 12350 }, { "epoch": 7.0, "learning_rate": 6.99493670886076e-06, "loss": 0.0383, "step": 12375 }, { "epoch": 7.0, "learning_rate": 6.98860759493671e-06, "loss": 0.0312, "step": 12400 }, { "epoch": 7.0, "learning_rate": 6.982278481012658e-06, "loss": 0.0275, "step": 12425 }, { "epoch": 7.0, "learning_rate": 6.975949367088608e-06, "loss": 0.0407, "step": 12450 }, { "epoch": 7.01, "learning_rate": 6.969620253164557e-06, "loss": 0.0436, "step": 12475 }, { "epoch": 7.01, "learning_rate": 6.9632911392405066e-06, "loss": 0.0359, "step": 12500 }, { "epoch": 7.01, "learning_rate": 6.956962025316456e-06, "loss": 0.0281, "step": 12525 }, { "epoch": 7.01, "learning_rate": 6.9506329113924054e-06, "loss": 0.0283, "step": 12550 }, { "epoch": 7.01, "learning_rate": 6.944303797468355e-06, "loss": 0.0248, "step": 12575 }, { "epoch": 7.01, "learning_rate": 6.937974683544304e-06, "loss": 0.0295, "step": 12600 }, { "epoch": 7.01, "learning_rate": 6.931645569620254e-06, "loss": 0.0287, "step": 12625 }, { "epoch": 7.01, "learning_rate": 6.925316455696202e-06, "loss": 0.0261, "step": 12650 }, { "epoch": 7.01, "learning_rate": 6.918987341772152e-06, "loss": 0.0259, "step": 12675 }, { "epoch": 7.01, "learning_rate": 6.912658227848101e-06, "loss": 0.0265, "step": 12700 }, { "epoch": 7.01, "learning_rate": 6.906329113924051e-06, "loss": 0.0239, "step": 12725 }, { "epoch": 7.01, "learning_rate": 6.9e-06, "loss": 0.0342, "step": 12750 }, { "epoch": 7.01, "learning_rate": 6.89367088607595e-06, "loss": 0.0253, "step": 12775 }, { "epoch": 7.01, "learning_rate": 6.8873417721519e-06, "loss": 0.0317, "step": 12800 }, { "epoch": 7.01, "learning_rate": 6.881012658227849e-06, "loss": 0.0233, "step": 12825 }, { "epoch": 7.01, "learning_rate": 6.8746835443037986e-06, "loss": 0.0198, "step": 12850 }, { "epoch": 7.02, "learning_rate": 6.868354430379748e-06, "loss": 0.018, "step": 12875 }, { "epoch": 7.02, "learning_rate": 6.862025316455697e-06, "loss": 0.0182, "step": 12900 }, { "epoch": 7.02, "learning_rate": 6.855949367088608e-06, "loss": 0.0279, "step": 12925 }, { "epoch": 7.02, "learning_rate": 6.8496202531645574e-06, "loss": 0.0201, "step": 12950 }, { "epoch": 7.02, "learning_rate": 6.843291139240507e-06, "loss": 0.0222, "step": 12975 }, { "epoch": 7.02, "learning_rate": 6.836962025316456e-06, "loss": 0.0236, "step": 13000 }, { "epoch": 7.02, "eval_loss": 0.2616036534309387, "eval_runtime": 1403.0214, "eval_samples_per_second": 7.42, "eval_steps_per_second": 0.464, "eval_wer": 27.36142051090262, "step": 13000 }, { "epoch": 7.02, "learning_rate": 6.830632911392405e-06, "loss": 0.0159, "step": 13025 }, { "epoch": 7.02, "learning_rate": 6.824303797468354e-06, "loss": 0.0186, "step": 13050 }, { "epoch": 7.02, "learning_rate": 6.817974683544304e-06, "loss": 0.0246, "step": 13075 }, { "epoch": 7.02, "learning_rate": 6.811645569620253e-06, "loss": 0.0198, "step": 13100 }, { "epoch": 7.02, "learning_rate": 6.805316455696203e-06, "loss": 0.0287, "step": 13125 }, { "epoch": 7.02, "learning_rate": 6.798987341772152e-06, "loss": 0.0255, "step": 13150 }, { "epoch": 7.02, "learning_rate": 6.792658227848102e-06, "loss": 0.035, "step": 13175 }, { "epoch": 7.02, "learning_rate": 6.786329113924052e-06, "loss": 0.0358, "step": 13200 }, { "epoch": 7.02, "learning_rate": 6.780000000000001e-06, "loss": 0.0283, "step": 13225 }, { "epoch": 7.02, "learning_rate": 6.773670886075951e-06, "loss": 0.0203, "step": 13250 }, { "epoch": 7.03, "learning_rate": 6.7673417721519e-06, "loss": 0.0193, "step": 13275 }, { "epoch": 7.03, "learning_rate": 6.7610126582278494e-06, "loss": 0.0174, "step": 13300 }, { "epoch": 7.03, "learning_rate": 6.754683544303798e-06, "loss": 0.0279, "step": 13325 }, { "epoch": 7.03, "learning_rate": 6.7483544303797474e-06, "loss": 0.023, "step": 13350 }, { "epoch": 7.03, "learning_rate": 6.742025316455697e-06, "loss": 0.0266, "step": 13375 }, { "epoch": 7.03, "learning_rate": 6.735696202531646e-06, "loss": 0.024, "step": 13400 }, { "epoch": 7.03, "learning_rate": 6.729367088607596e-06, "loss": 0.0164, "step": 13425 }, { "epoch": 7.03, "learning_rate": 6.723037974683545e-06, "loss": 0.019, "step": 13450 }, { "epoch": 7.03, "learning_rate": 6.716708860759495e-06, "loss": 0.0209, "step": 13475 }, { "epoch": 7.03, "learning_rate": 6.710379746835444e-06, "loss": 0.0241, "step": 13500 }, { "epoch": 7.03, "learning_rate": 6.7040506329113934e-06, "loss": 0.0233, "step": 13525 }, { "epoch": 7.03, "learning_rate": 6.697721518987342e-06, "loss": 0.0204, "step": 13550 }, { "epoch": 7.03, "learning_rate": 6.6913924050632914e-06, "loss": 0.0131, "step": 13575 }, { "epoch": 7.03, "learning_rate": 6.685063291139241e-06, "loss": 0.0268, "step": 13600 }, { "epoch": 7.03, "learning_rate": 6.67873417721519e-06, "loss": 0.0156, "step": 13625 }, { "epoch": 7.03, "learning_rate": 6.67240506329114e-06, "loss": 0.0205, "step": 13650 }, { "epoch": 7.04, "learning_rate": 6.666075949367089e-06, "loss": 0.0217, "step": 13675 }, { "epoch": 7.04, "learning_rate": 6.659746835443039e-06, "loss": 0.0258, "step": 13700 }, { "epoch": 7.04, "learning_rate": 6.653417721518988e-06, "loss": 0.0211, "step": 13725 }, { "epoch": 7.04, "learning_rate": 6.647088607594937e-06, "loss": 0.0188, "step": 13750 }, { "epoch": 7.04, "learning_rate": 6.640759493670886e-06, "loss": 0.0234, "step": 13775 }, { "epoch": 7.04, "learning_rate": 6.6344303797468355e-06, "loss": 0.0202, "step": 13800 }, { "epoch": 7.04, "learning_rate": 6.628101265822785e-06, "loss": 0.0251, "step": 13825 }, { "epoch": 7.04, "learning_rate": 6.621772151898734e-06, "loss": 0.017, "step": 13850 }, { "epoch": 7.04, "learning_rate": 6.615443037974684e-06, "loss": 0.0254, "step": 13875 }, { "epoch": 7.04, "learning_rate": 6.609113924050633e-06, "loss": 0.0289, "step": 13900 }, { "epoch": 7.04, "learning_rate": 6.602784810126583e-06, "loss": 0.0238, "step": 13925 }, { "epoch": 7.04, "learning_rate": 6.596455696202532e-06, "loss": 0.0218, "step": 13950 }, { "epoch": 7.04, "learning_rate": 6.590126582278481e-06, "loss": 0.0289, "step": 13975 }, { "epoch": 7.04, "learning_rate": 6.583797468354432e-06, "loss": 0.0271, "step": 14000 }, { "epoch": 7.04, "eval_loss": 0.26684999465942383, "eval_runtime": 1448.0228, "eval_samples_per_second": 7.19, "eval_steps_per_second": 0.45, "eval_wer": 42.464105107400115, "step": 14000 }, { "epoch": 8.0, "learning_rate": 6.577468354430381e-06, "loss": 0.0292, "step": 14025 }, { "epoch": 8.0, "learning_rate": 6.57113924050633e-06, "loss": 0.0192, "step": 14050 }, { "epoch": 8.0, "learning_rate": 6.564810126582279e-06, "loss": 0.027, "step": 14075 }, { "epoch": 8.0, "learning_rate": 6.558481012658229e-06, "loss": 0.03, "step": 14100 }, { "epoch": 8.0, "learning_rate": 6.552151898734178e-06, "loss": 0.0375, "step": 14125 }, { "epoch": 8.0, "learning_rate": 6.5458227848101274e-06, "loss": 0.0338, "step": 14150 }, { "epoch": 8.0, "learning_rate": 6.539493670886077e-06, "loss": 0.0305, "step": 14175 }, { "epoch": 8.0, "learning_rate": 6.533164556962026e-06, "loss": 0.0266, "step": 14200 }, { "epoch": 8.01, "learning_rate": 6.526835443037976e-06, "loss": 0.0378, "step": 14225 }, { "epoch": 8.01, "learning_rate": 6.520506329113925e-06, "loss": 0.0256, "step": 14250 }, { "epoch": 8.01, "learning_rate": 6.514177215189874e-06, "loss": 0.0203, "step": 14275 }, { "epoch": 8.01, "learning_rate": 6.507848101265823e-06, "loss": 0.0212, "step": 14300 }, { "epoch": 8.01, "learning_rate": 6.501518987341773e-06, "loss": 0.0204, "step": 14325 }, { "epoch": 8.01, "learning_rate": 6.495189873417722e-06, "loss": 0.0259, "step": 14350 }, { "epoch": 8.01, "learning_rate": 6.4888607594936714e-06, "loss": 0.0214, "step": 14375 }, { "epoch": 8.01, "learning_rate": 6.482531645569621e-06, "loss": 0.021, "step": 14400 }, { "epoch": 8.01, "learning_rate": 6.47620253164557e-06, "loss": 0.0241, "step": 14425 }, { "epoch": 8.01, "learning_rate": 6.46987341772152e-06, "loss": 0.0216, "step": 14450 }, { "epoch": 8.01, "learning_rate": 6.463544303797469e-06, "loss": 0.0209, "step": 14475 }, { "epoch": 8.01, "learning_rate": 6.457215189873418e-06, "loss": 0.0292, "step": 14500 }, { "epoch": 8.01, "learning_rate": 6.450886075949367e-06, "loss": 0.0286, "step": 14525 }, { "epoch": 8.01, "learning_rate": 6.444556962025317e-06, "loss": 0.0218, "step": 14550 }, { "epoch": 8.01, "learning_rate": 6.438227848101266e-06, "loss": 0.0209, "step": 14575 }, { "epoch": 8.01, "learning_rate": 6.4318987341772154e-06, "loss": 0.0244, "step": 14600 }, { "epoch": 8.02, "learning_rate": 6.425569620253165e-06, "loss": 0.0212, "step": 14625 }, { "epoch": 8.02, "learning_rate": 6.419240506329114e-06, "loss": 0.0155, "step": 14650 }, { "epoch": 8.02, "learning_rate": 6.412911392405064e-06, "loss": 0.0209, "step": 14675 }, { "epoch": 8.02, "learning_rate": 6.406582278481013e-06, "loss": 0.0203, "step": 14700 }, { "epoch": 8.02, "learning_rate": 6.400253164556962e-06, "loss": 0.0192, "step": 14725 }, { "epoch": 8.02, "learning_rate": 6.393924050632911e-06, "loss": 0.0216, "step": 14750 }, { "epoch": 8.02, "learning_rate": 6.387594936708861e-06, "loss": 0.012, "step": 14775 }, { "epoch": 8.02, "learning_rate": 6.38126582278481e-06, "loss": 0.0241, "step": 14800 }, { "epoch": 8.02, "learning_rate": 6.37493670886076e-06, "loss": 0.0192, "step": 14825 }, { "epoch": 8.02, "learning_rate": 6.36860759493671e-06, "loss": 0.0202, "step": 14850 }, { "epoch": 8.02, "learning_rate": 6.362278481012659e-06, "loss": 0.0257, "step": 14875 }, { "epoch": 8.02, "learning_rate": 6.3559493670886086e-06, "loss": 0.0218, "step": 14900 }, { "epoch": 8.02, "learning_rate": 6.349620253164558e-06, "loss": 0.0232, "step": 14925 }, { "epoch": 8.02, "learning_rate": 6.3432911392405074e-06, "loss": 0.03, "step": 14950 }, { "epoch": 8.02, "learning_rate": 6.337215189873418e-06, "loss": 0.0252, "step": 14975 }, { "epoch": 8.02, "learning_rate": 6.3308860759493675e-06, "loss": 0.0176, "step": 15000 }, { "epoch": 8.02, "eval_loss": 0.2596448063850403, "eval_runtime": 1417.6481, "eval_samples_per_second": 7.344, "eval_steps_per_second": 0.459, "eval_wer": 28.52241499861786, "step": 15000 } ], "max_steps": 40000, "num_train_epochs": 9223372036854775807, "total_flos": 6.92374092447744e+19, "trial_name": null, "trial_params": null }