{ "best_metric": 5.895082837397793, "best_model_checkpoint": "./output1/checkpoint-12000", "epoch": 2.1053333333333333, "global_step": 12000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.6000000000000004e-07, "loss": 0.7515, "step": 25 }, { "epoch": 0.0, "learning_rate": 9.400000000000001e-07, "loss": 0.5055, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.44e-06, "loss": 0.207, "step": 75 }, { "epoch": 0.01, "learning_rate": 1.94e-06, "loss": 0.144, "step": 100 }, { "epoch": 0.01, "learning_rate": 2.4400000000000004e-06, "loss": 0.1265, "step": 125 }, { "epoch": 0.01, "learning_rate": 2.9400000000000002e-06, "loss": 0.1558, "step": 150 }, { "epoch": 0.01, "learning_rate": 3.44e-06, "loss": 0.085, "step": 175 }, { "epoch": 0.02, "learning_rate": 3.94e-06, "loss": 0.1084, "step": 200 }, { "epoch": 0.02, "learning_rate": 4.440000000000001e-06, "loss": 0.1233, "step": 225 }, { "epoch": 0.02, "learning_rate": 4.94e-06, "loss": 0.1032, "step": 250 }, { "epoch": 0.02, "learning_rate": 5.4400000000000004e-06, "loss": 0.109, "step": 275 }, { "epoch": 0.03, "learning_rate": 5.94e-06, "loss": 0.1042, "step": 300 }, { "epoch": 0.03, "learning_rate": 6.42e-06, "loss": 0.1276, "step": 325 }, { "epoch": 0.03, "learning_rate": 6.92e-06, "loss": 0.1224, "step": 350 }, { "epoch": 0.03, "learning_rate": 7.420000000000001e-06, "loss": 0.1291, "step": 375 }, { "epoch": 0.03, "learning_rate": 7.9e-06, "loss": 0.143, "step": 400 }, { "epoch": 0.04, "learning_rate": 8.400000000000001e-06, "loss": 0.1331, "step": 425 }, { "epoch": 0.04, "learning_rate": 8.880000000000001e-06, "loss": 0.1605, "step": 450 }, { "epoch": 0.04, "learning_rate": 9.38e-06, "loss": 0.1887, "step": 475 }, { "epoch": 0.04, "learning_rate": 9.88e-06, "loss": 0.183, "step": 500 }, { "epoch": 0.04, "learning_rate": 9.983478260869566e-06, "loss": 0.1646, "step": 525 }, { "epoch": 0.05, "learning_rate": 9.961739130434783e-06, "loss": 0.166, "step": 550 }, { "epoch": 0.05, "learning_rate": 9.940000000000001e-06, "loss": 0.1888, "step": 575 }, { "epoch": 0.05, "learning_rate": 9.918260869565218e-06, "loss": 0.1969, "step": 600 }, { "epoch": 0.05, "learning_rate": 9.896521739130437e-06, "loss": 0.2212, "step": 625 }, { "epoch": 0.05, "learning_rate": 9.874782608695653e-06, "loss": 0.1807, "step": 650 }, { "epoch": 0.06, "learning_rate": 9.853043478260872e-06, "loss": 0.1986, "step": 675 }, { "epoch": 0.06, "learning_rate": 9.831304347826087e-06, "loss": 0.2535, "step": 700 }, { "epoch": 0.06, "learning_rate": 9.809565217391305e-06, "loss": 0.188, "step": 725 }, { "epoch": 0.06, "learning_rate": 9.787826086956522e-06, "loss": 0.2109, "step": 750 }, { "epoch": 0.06, "learning_rate": 9.76608695652174e-06, "loss": 0.2326, "step": 775 }, { "epoch": 0.07, "learning_rate": 9.744347826086957e-06, "loss": 0.2208, "step": 800 }, { "epoch": 0.07, "learning_rate": 9.722608695652175e-06, "loss": 0.17, "step": 825 }, { "epoch": 0.07, "learning_rate": 9.700869565217392e-06, "loss": 0.1559, "step": 850 }, { "epoch": 0.07, "learning_rate": 9.67913043478261e-06, "loss": 0.1538, "step": 875 }, { "epoch": 0.07, "learning_rate": 9.657391304347827e-06, "loss": 0.1684, "step": 900 }, { "epoch": 0.08, "learning_rate": 9.635652173913044e-06, "loss": 0.1854, "step": 925 }, { "epoch": 0.08, "learning_rate": 9.61391304347826e-06, "loss": 0.1413, "step": 950 }, { "epoch": 0.08, "learning_rate": 9.592173913043479e-06, "loss": 0.1716, "step": 975 }, { "epoch": 0.08, "learning_rate": 9.570434782608696e-06, "loss": 0.138, "step": 1000 }, { "epoch": 0.08, "eval_loss": 0.21006987988948822, "eval_runtime": 15602.3465, "eval_samples_per_second": 0.689, "eval_steps_per_second": 0.172, "eval_wer": 11.52884252217674, "step": 1000 }, { "epoch": 0.09, "learning_rate": 9.548695652173914e-06, "loss": 0.1526, "step": 1025 }, { "epoch": 0.09, "learning_rate": 9.52695652173913e-06, "loss": 0.1578, "step": 1050 }, { "epoch": 0.09, "learning_rate": 9.50521739130435e-06, "loss": 0.1465, "step": 1075 }, { "epoch": 0.09, "learning_rate": 9.483478260869566e-06, "loss": 0.1575, "step": 1100 }, { "epoch": 0.09, "learning_rate": 9.461739130434784e-06, "loss": 0.1656, "step": 1125 }, { "epoch": 0.1, "learning_rate": 9.440000000000001e-06, "loss": 0.1623, "step": 1150 }, { "epoch": 0.1, "learning_rate": 9.418260869565218e-06, "loss": 0.1428, "step": 1175 }, { "epoch": 0.1, "learning_rate": 9.396521739130435e-06, "loss": 0.1652, "step": 1200 }, { "epoch": 0.1, "learning_rate": 9.374782608695653e-06, "loss": 0.1585, "step": 1225 }, { "epoch": 0.1, "learning_rate": 9.35304347826087e-06, "loss": 0.1826, "step": 1250 }, { "epoch": 0.11, "learning_rate": 9.331304347826088e-06, "loss": 0.2007, "step": 1275 }, { "epoch": 0.11, "learning_rate": 9.309565217391305e-06, "loss": 0.1528, "step": 1300 }, { "epoch": 0.11, "learning_rate": 9.287826086956523e-06, "loss": 0.1651, "step": 1325 }, { "epoch": 0.11, "learning_rate": 9.26608695652174e-06, "loss": 0.1637, "step": 1350 }, { "epoch": 0.11, "learning_rate": 9.244347826086957e-06, "loss": 0.1564, "step": 1375 }, { "epoch": 0.12, "learning_rate": 9.222608695652175e-06, "loss": 0.2099, "step": 1400 }, { "epoch": 0.12, "learning_rate": 9.200869565217392e-06, "loss": 0.1688, "step": 1425 }, { "epoch": 0.12, "learning_rate": 9.17913043478261e-06, "loss": 0.1772, "step": 1450 }, { "epoch": 0.12, "learning_rate": 9.157391304347827e-06, "loss": 0.1279, "step": 1475 }, { "epoch": 0.12, "learning_rate": 9.135652173913044e-06, "loss": 0.1674, "step": 1500 }, { "epoch": 0.13, "learning_rate": 9.113913043478262e-06, "loss": 0.1627, "step": 1525 }, { "epoch": 0.13, "learning_rate": 9.092173913043479e-06, "loss": 0.136, "step": 1550 }, { "epoch": 0.13, "learning_rate": 9.070434782608697e-06, "loss": 0.1263, "step": 1575 }, { "epoch": 0.13, "learning_rate": 9.048695652173914e-06, "loss": 0.171, "step": 1600 }, { "epoch": 0.14, "learning_rate": 9.02695652173913e-06, "loss": 0.1328, "step": 1625 }, { "epoch": 0.14, "learning_rate": 9.005217391304349e-06, "loss": 0.1416, "step": 1650 }, { "epoch": 0.14, "learning_rate": 8.983478260869566e-06, "loss": 0.1135, "step": 1675 }, { "epoch": 0.14, "learning_rate": 8.961739130434784e-06, "loss": 0.1153, "step": 1700 }, { "epoch": 0.14, "learning_rate": 8.94e-06, "loss": 0.1127, "step": 1725 }, { "epoch": 0.15, "learning_rate": 8.91826086956522e-06, "loss": 0.1757, "step": 1750 }, { "epoch": 0.15, "learning_rate": 8.896521739130436e-06, "loss": 0.123, "step": 1775 }, { "epoch": 0.15, "learning_rate": 8.874782608695653e-06, "loss": 0.1382, "step": 1800 }, { "epoch": 0.15, "learning_rate": 8.85304347826087e-06, "loss": 0.1585, "step": 1825 }, { "epoch": 0.15, "learning_rate": 8.831304347826088e-06, "loss": 0.1523, "step": 1850 }, { "epoch": 0.16, "learning_rate": 8.809565217391304e-06, "loss": 0.1452, "step": 1875 }, { "epoch": 0.16, "learning_rate": 8.787826086956523e-06, "loss": 0.1314, "step": 1900 }, { "epoch": 0.16, "learning_rate": 8.76608695652174e-06, "loss": 0.1321, "step": 1925 }, { "epoch": 0.16, "learning_rate": 8.744347826086958e-06, "loss": 0.1958, "step": 1950 }, { "epoch": 0.16, "learning_rate": 8.722608695652175e-06, "loss": 0.1185, "step": 1975 }, { "epoch": 0.17, "learning_rate": 8.700869565217393e-06, "loss": 0.121, "step": 2000 }, { "epoch": 0.17, "eval_loss": 0.19874396920204163, "eval_runtime": 15276.9212, "eval_samples_per_second": 0.703, "eval_steps_per_second": 0.176, "eval_wer": 10.445780549799414, "step": 2000 }, { "epoch": 0.17, "learning_rate": 8.679130434782608e-06, "loss": 0.1296, "step": 2025 }, { "epoch": 0.17, "learning_rate": 8.657391304347827e-06, "loss": 0.1132, "step": 2050 }, { "epoch": 0.17, "learning_rate": 8.635652173913043e-06, "loss": 0.1355, "step": 2075 }, { "epoch": 0.17, "learning_rate": 8.613913043478262e-06, "loss": 0.1465, "step": 2100 }, { "epoch": 0.18, "learning_rate": 8.592173913043478e-06, "loss": 0.1085, "step": 2125 }, { "epoch": 0.18, "learning_rate": 8.570434782608697e-06, "loss": 0.107, "step": 2150 }, { "epoch": 0.18, "learning_rate": 8.548695652173914e-06, "loss": 0.1419, "step": 2175 }, { "epoch": 0.18, "learning_rate": 8.526956521739132e-06, "loss": 0.1084, "step": 2200 }, { "epoch": 0.19, "learning_rate": 8.505217391304349e-06, "loss": 0.1579, "step": 2225 }, { "epoch": 0.19, "learning_rate": 8.483478260869565e-06, "loss": 0.1201, "step": 2250 }, { "epoch": 0.19, "learning_rate": 8.461739130434782e-06, "loss": 0.1467, "step": 2275 }, { "epoch": 0.19, "learning_rate": 8.44e-06, "loss": 0.1499, "step": 2300 }, { "epoch": 0.19, "learning_rate": 8.418260869565217e-06, "loss": 0.1215, "step": 2325 }, { "epoch": 0.2, "learning_rate": 8.396521739130436e-06, "loss": 0.1166, "step": 2350 }, { "epoch": 0.2, "learning_rate": 8.374782608695652e-06, "loss": 0.1864, "step": 2375 }, { "epoch": 0.2, "learning_rate": 8.35304347826087e-06, "loss": 0.116, "step": 2400 }, { "epoch": 0.2, "learning_rate": 8.331304347826087e-06, "loss": 0.1805, "step": 2425 }, { "epoch": 0.2, "learning_rate": 8.309565217391306e-06, "loss": 0.1684, "step": 2450 }, { "epoch": 0.21, "learning_rate": 8.287826086956523e-06, "loss": 0.1299, "step": 2475 }, { "epoch": 0.21, "learning_rate": 8.26608695652174e-06, "loss": 0.1236, "step": 2500 }, { "epoch": 0.21, "learning_rate": 8.244347826086958e-06, "loss": 0.1397, "step": 2525 }, { "epoch": 0.21, "learning_rate": 8.222608695652174e-06, "loss": 0.12, "step": 2550 }, { "epoch": 0.21, "learning_rate": 8.200869565217391e-06, "loss": 0.1409, "step": 2575 }, { "epoch": 0.22, "learning_rate": 8.17913043478261e-06, "loss": 0.1524, "step": 2600 }, { "epoch": 0.22, "learning_rate": 8.157391304347826e-06, "loss": 0.1116, "step": 2625 }, { "epoch": 0.22, "learning_rate": 8.135652173913045e-06, "loss": 0.1701, "step": 2650 }, { "epoch": 0.22, "learning_rate": 8.113913043478261e-06, "loss": 0.1405, "step": 2675 }, { "epoch": 0.23, "learning_rate": 8.092173913043478e-06, "loss": 0.1487, "step": 2700 }, { "epoch": 0.23, "learning_rate": 8.070434782608697e-06, "loss": 0.154, "step": 2725 }, { "epoch": 0.23, "learning_rate": 8.048695652173913e-06, "loss": 0.1094, "step": 2750 }, { "epoch": 0.23, "learning_rate": 8.026956521739132e-06, "loss": 0.1155, "step": 2775 }, { "epoch": 0.23, "learning_rate": 8.005217391304348e-06, "loss": 0.1061, "step": 2800 }, { "epoch": 0.24, "learning_rate": 7.983478260869567e-06, "loss": 0.1016, "step": 2825 }, { "epoch": 0.24, "learning_rate": 7.961739130434783e-06, "loss": 0.1161, "step": 2850 }, { "epoch": 0.24, "learning_rate": 7.94e-06, "loss": 0.162, "step": 2875 }, { "epoch": 0.24, "learning_rate": 7.918260869565219e-06, "loss": 0.1337, "step": 2900 }, { "epoch": 0.24, "learning_rate": 7.896521739130435e-06, "loss": 0.1315, "step": 2925 }, { "epoch": 0.25, "learning_rate": 7.874782608695652e-06, "loss": 0.1255, "step": 2950 }, { "epoch": 0.25, "learning_rate": 7.85304347826087e-06, "loss": 0.1197, "step": 2975 }, { "epoch": 0.25, "learning_rate": 7.831304347826087e-06, "loss": 0.1413, "step": 3000 }, { "epoch": 0.25, "eval_loss": 0.19560454785823822, "eval_runtime": 15398.3583, "eval_samples_per_second": 0.698, "eval_steps_per_second": 0.174, "eval_wer": 10.467217214662679, "step": 3000 }, { "epoch": 0.25, "learning_rate": 7.809565217391306e-06, "loss": 0.1344, "step": 3025 }, { "epoch": 0.25, "learning_rate": 7.787826086956522e-06, "loss": 0.1171, "step": 3050 }, { "epoch": 0.26, "learning_rate": 7.76608695652174e-06, "loss": 0.1117, "step": 3075 }, { "epoch": 0.26, "learning_rate": 7.744347826086957e-06, "loss": 0.1098, "step": 3100 }, { "epoch": 0.26, "learning_rate": 7.722608695652176e-06, "loss": 0.1347, "step": 3125 }, { "epoch": 0.26, "learning_rate": 7.700869565217391e-06, "loss": 0.1394, "step": 3150 }, { "epoch": 0.26, "learning_rate": 7.67913043478261e-06, "loss": 0.1074, "step": 3175 }, { "epoch": 0.27, "learning_rate": 7.657391304347826e-06, "loss": 0.1457, "step": 3200 }, { "epoch": 0.27, "learning_rate": 7.635652173913044e-06, "loss": 0.0925, "step": 3225 }, { "epoch": 0.27, "learning_rate": 7.613913043478262e-06, "loss": 0.1306, "step": 3250 }, { "epoch": 0.27, "learning_rate": 7.592173913043479e-06, "loss": 0.1184, "step": 3275 }, { "epoch": 0.28, "learning_rate": 7.570434782608696e-06, "loss": 0.1422, "step": 3300 }, { "epoch": 0.28, "learning_rate": 7.548695652173914e-06, "loss": 0.1566, "step": 3325 }, { "epoch": 0.28, "learning_rate": 7.526956521739131e-06, "loss": 0.1272, "step": 3350 }, { "epoch": 0.28, "learning_rate": 7.505217391304348e-06, "loss": 0.1034, "step": 3375 }, { "epoch": 0.28, "learning_rate": 7.483478260869566e-06, "loss": 0.1078, "step": 3400 }, { "epoch": 0.29, "learning_rate": 7.461739130434783e-06, "loss": 0.1605, "step": 3425 }, { "epoch": 0.29, "learning_rate": 7.440000000000001e-06, "loss": 0.1183, "step": 3450 }, { "epoch": 0.29, "learning_rate": 7.418260869565218e-06, "loss": 0.1643, "step": 3475 }, { "epoch": 0.29, "learning_rate": 7.396521739130436e-06, "loss": 0.1368, "step": 3500 }, { "epoch": 0.29, "learning_rate": 7.3747826086956535e-06, "loss": 0.153, "step": 3525 }, { "epoch": 0.3, "learning_rate": 7.353043478260871e-06, "loss": 0.1649, "step": 3550 }, { "epoch": 0.3, "learning_rate": 7.331304347826087e-06, "loss": 0.1097, "step": 3575 }, { "epoch": 0.3, "learning_rate": 7.3095652173913044e-06, "loss": 0.181, "step": 3600 }, { "epoch": 0.3, "learning_rate": 7.287826086956522e-06, "loss": 0.1397, "step": 3625 }, { "epoch": 0.3, "learning_rate": 7.2660869565217396e-06, "loss": 0.1492, "step": 3650 }, { "epoch": 0.31, "learning_rate": 7.244347826086957e-06, "loss": 0.155, "step": 3675 }, { "epoch": 0.31, "learning_rate": 7.222608695652175e-06, "loss": 0.1197, "step": 3700 }, { "epoch": 0.31, "learning_rate": 7.200869565217392e-06, "loss": 0.1274, "step": 3725 }, { "epoch": 0.31, "learning_rate": 7.17913043478261e-06, "loss": 0.1389, "step": 3750 }, { "epoch": 0.31, "learning_rate": 7.157391304347827e-06, "loss": 0.1216, "step": 3775 }, { "epoch": 0.32, "learning_rate": 7.135652173913043e-06, "loss": 0.113, "step": 3800 }, { "epoch": 0.32, "learning_rate": 7.113913043478261e-06, "loss": 0.1224, "step": 3825 }, { "epoch": 0.32, "learning_rate": 7.092173913043478e-06, "loss": 0.1227, "step": 3850 }, { "epoch": 0.32, "learning_rate": 7.070434782608696e-06, "loss": 0.1076, "step": 3875 }, { "epoch": 0.33, "learning_rate": 7.0486956521739135e-06, "loss": 0.1453, "step": 3900 }, { "epoch": 0.33, "learning_rate": 7.026956521739131e-06, "loss": 0.1581, "step": 3925 }, { "epoch": 0.33, "learning_rate": 7.005217391304349e-06, "loss": 0.1636, "step": 3950 }, { "epoch": 0.33, "learning_rate": 6.983478260869566e-06, "loss": 0.1145, "step": 3975 }, { "epoch": 0.33, "learning_rate": 6.961739130434784e-06, "loss": 0.1158, "step": 4000 }, { "epoch": 0.33, "eval_loss": 0.17780181765556335, "eval_runtime": 15586.0168, "eval_samples_per_second": 0.689, "eval_steps_per_second": 0.172, "eval_wer": 9.372926513071262, "step": 4000 }, { "epoch": 0.34, "learning_rate": 6.9400000000000005e-06, "loss": 0.1216, "step": 4025 }, { "epoch": 0.34, "learning_rate": 6.918260869565217e-06, "loss": 0.1074, "step": 4050 }, { "epoch": 0.34, "learning_rate": 6.896521739130435e-06, "loss": 0.0898, "step": 4075 }, { "epoch": 0.34, "learning_rate": 6.874782608695652e-06, "loss": 0.0964, "step": 4100 }, { "epoch": 0.34, "learning_rate": 6.85304347826087e-06, "loss": 0.0995, "step": 4125 }, { "epoch": 0.35, "learning_rate": 6.8313043478260874e-06, "loss": 0.0868, "step": 4150 }, { "epoch": 0.35, "learning_rate": 6.809565217391305e-06, "loss": 0.091, "step": 4175 }, { "epoch": 0.35, "learning_rate": 6.7878260869565226e-06, "loss": 0.0614, "step": 4200 }, { "epoch": 0.35, "learning_rate": 6.76608695652174e-06, "loss": 0.0776, "step": 4225 }, { "epoch": 0.35, "learning_rate": 6.744347826086957e-06, "loss": 0.0648, "step": 4250 }, { "epoch": 0.36, "learning_rate": 6.722608695652174e-06, "loss": 0.1021, "step": 4275 }, { "epoch": 0.36, "learning_rate": 6.700869565217392e-06, "loss": 0.0831, "step": 4300 }, { "epoch": 0.36, "learning_rate": 6.6791304347826095e-06, "loss": 0.1161, "step": 4325 }, { "epoch": 0.36, "learning_rate": 6.657391304347826e-06, "loss": 0.101, "step": 4350 }, { "epoch": 0.36, "learning_rate": 6.635652173913044e-06, "loss": 0.0892, "step": 4375 }, { "epoch": 0.37, "learning_rate": 6.613913043478261e-06, "loss": 0.0943, "step": 4400 }, { "epoch": 0.37, "learning_rate": 6.592173913043479e-06, "loss": 0.074, "step": 4425 }, { "epoch": 0.37, "learning_rate": 6.5704347826086965e-06, "loss": 0.1021, "step": 4450 }, { "epoch": 0.37, "learning_rate": 6.548695652173913e-06, "loss": 0.1037, "step": 4475 }, { "epoch": 0.38, "learning_rate": 6.526956521739131e-06, "loss": 0.0684, "step": 4500 }, { "epoch": 0.38, "learning_rate": 6.505217391304348e-06, "loss": 0.0572, "step": 4525 }, { "epoch": 0.38, "learning_rate": 6.483478260869566e-06, "loss": 0.0867, "step": 4550 }, { "epoch": 0.38, "learning_rate": 6.4617391304347835e-06, "loss": 0.0896, "step": 4575 }, { "epoch": 0.38, "learning_rate": 6.440000000000001e-06, "loss": 0.0818, "step": 4600 }, { "epoch": 0.39, "learning_rate": 6.418260869565219e-06, "loss": 0.0866, "step": 4625 }, { "epoch": 0.39, "learning_rate": 6.397391304347826e-06, "loss": 0.1059, "step": 4650 }, { "epoch": 0.39, "learning_rate": 6.375652173913044e-06, "loss": 0.0829, "step": 4675 }, { "epoch": 0.39, "learning_rate": 6.353913043478261e-06, "loss": 0.0926, "step": 4700 }, { "epoch": 0.39, "learning_rate": 6.332173913043479e-06, "loss": 0.124, "step": 4725 }, { "epoch": 0.4, "learning_rate": 6.3104347826086964e-06, "loss": 0.102, "step": 4750 }, { "epoch": 0.4, "learning_rate": 6.288695652173914e-06, "loss": 0.1063, "step": 4775 }, { "epoch": 0.4, "learning_rate": 6.2669565217391316e-06, "loss": 0.0903, "step": 4800 }, { "epoch": 0.4, "learning_rate": 6.245217391304348e-06, "loss": 0.1055, "step": 4825 }, { "epoch": 0.4, "learning_rate": 6.223478260869565e-06, "loss": 0.095, "step": 4850 }, { "epoch": 0.41, "learning_rate": 6.2017391304347826e-06, "loss": 0.1251, "step": 4875 }, { "epoch": 0.41, "learning_rate": 6.18e-06, "loss": 0.1235, "step": 4900 }, { "epoch": 0.41, "learning_rate": 6.158260869565218e-06, "loss": 0.1182, "step": 4925 }, { "epoch": 0.41, "learning_rate": 6.136521739130435e-06, "loss": 0.078, "step": 4950 }, { "epoch": 0.41, "learning_rate": 6.114782608695653e-06, "loss": 0.1615, "step": 4975 }, { "epoch": 0.42, "learning_rate": 6.09304347826087e-06, "loss": 0.1056, "step": 5000 }, { "epoch": 0.42, "eval_loss": 0.17951148748397827, "eval_runtime": 15573.2195, "eval_samples_per_second": 0.69, "eval_steps_per_second": 0.172, "eval_wer": 9.779202351908374, "step": 5000 }, { "epoch": 0.42, "learning_rate": 6.071304347826087e-06, "loss": 0.0967, "step": 5025 }, { "epoch": 0.42, "learning_rate": 6.049565217391305e-06, "loss": 0.1342, "step": 5050 }, { "epoch": 0.42, "learning_rate": 6.027826086956522e-06, "loss": 0.1332, "step": 5075 }, { "epoch": 0.42, "learning_rate": 6.00608695652174e-06, "loss": 0.1087, "step": 5100 }, { "epoch": 0.43, "learning_rate": 5.984347826086957e-06, "loss": 0.1395, "step": 5125 }, { "epoch": 0.43, "learning_rate": 5.962608695652175e-06, "loss": 0.1379, "step": 5150 }, { "epoch": 0.43, "learning_rate": 5.940869565217392e-06, "loss": 0.1202, "step": 5175 }, { "epoch": 0.43, "learning_rate": 5.919130434782609e-06, "loss": 0.1312, "step": 5200 }, { "epoch": 0.44, "learning_rate": 5.897391304347827e-06, "loss": 0.0865, "step": 5225 }, { "epoch": 0.44, "learning_rate": 5.8756521739130435e-06, "loss": 0.0991, "step": 5250 }, { "epoch": 0.44, "learning_rate": 5.853913043478261e-06, "loss": 0.128, "step": 5275 }, { "epoch": 0.44, "learning_rate": 5.832173913043479e-06, "loss": 0.1183, "step": 5300 }, { "epoch": 0.44, "learning_rate": 5.810434782608696e-06, "loss": 0.119, "step": 5325 }, { "epoch": 0.45, "learning_rate": 5.788695652173914e-06, "loss": 0.1268, "step": 5350 }, { "epoch": 1.0, "learning_rate": 5.766956521739131e-06, "loss": 0.1383, "step": 5375 }, { "epoch": 1.0, "learning_rate": 5.745217391304349e-06, "loss": 0.0993, "step": 5400 }, { "epoch": 1.0, "learning_rate": 5.723478260869566e-06, "loss": 0.0893, "step": 5425 }, { "epoch": 1.01, "learning_rate": 5.701739130434784e-06, "loss": 0.0548, "step": 5450 }, { "epoch": 1.01, "learning_rate": 5.68e-06, "loss": 0.0671, "step": 5475 }, { "epoch": 1.01, "learning_rate": 5.658260869565217e-06, "loss": 0.0435, "step": 5500 }, { "epoch": 1.01, "learning_rate": 5.636521739130435e-06, "loss": 0.0544, "step": 5525 }, { "epoch": 1.02, "learning_rate": 5.6147826086956525e-06, "loss": 0.0625, "step": 5550 }, { "epoch": 1.02, "learning_rate": 5.59304347826087e-06, "loss": 0.071, "step": 5575 }, { "epoch": 1.02, "learning_rate": 5.571304347826088e-06, "loss": 0.077, "step": 5600 }, { "epoch": 1.02, "learning_rate": 5.549565217391305e-06, "loss": 0.0567, "step": 5625 }, { "epoch": 1.02, "learning_rate": 5.527826086956523e-06, "loss": 0.0631, "step": 5650 }, { "epoch": 1.03, "learning_rate": 5.50608695652174e-06, "loss": 0.0718, "step": 5675 }, { "epoch": 1.03, "learning_rate": 5.484347826086956e-06, "loss": 0.0538, "step": 5700 }, { "epoch": 1.03, "learning_rate": 5.462608695652174e-06, "loss": 0.0381, "step": 5725 }, { "epoch": 1.03, "learning_rate": 5.440869565217391e-06, "loss": 0.0745, "step": 5750 }, { "epoch": 1.03, "learning_rate": 5.419130434782609e-06, "loss": 0.0686, "step": 5775 }, { "epoch": 1.04, "learning_rate": 5.3973913043478264e-06, "loss": 0.0492, "step": 5800 }, { "epoch": 1.04, "learning_rate": 5.375652173913044e-06, "loss": 0.07, "step": 5825 }, { "epoch": 1.04, "learning_rate": 5.3539130434782616e-06, "loss": 0.0663, "step": 5850 }, { "epoch": 1.04, "learning_rate": 5.332173913043479e-06, "loss": 0.0527, "step": 5875 }, { "epoch": 1.04, "learning_rate": 5.310434782608697e-06, "loss": 0.0503, "step": 5900 }, { "epoch": 1.05, "learning_rate": 5.288695652173913e-06, "loss": 0.0514, "step": 5925 }, { "epoch": 1.05, "learning_rate": 5.26695652173913e-06, "loss": 0.0703, "step": 5950 }, { "epoch": 1.05, "learning_rate": 5.245217391304348e-06, "loss": 0.0631, "step": 5975 }, { "epoch": 1.05, "learning_rate": 5.223478260869565e-06, "loss": 0.056, "step": 6000 }, { "epoch": 1.05, "eval_loss": 0.1559927612543106, "eval_runtime": 15677.864, "eval_samples_per_second": 0.685, "eval_steps_per_second": 0.171, "eval_wer": 7.692700305217276, "step": 6000 }, { "epoch": 1.05, "learning_rate": 5.201739130434783e-06, "loss": 0.0762, "step": 6025 }, { "epoch": 1.06, "learning_rate": 5.18e-06, "loss": 0.0747, "step": 6050 }, { "epoch": 1.06, "learning_rate": 5.158260869565218e-06, "loss": 0.0696, "step": 6075 }, { "epoch": 1.06, "learning_rate": 5.1365217391304355e-06, "loss": 0.0732, "step": 6100 }, { "epoch": 1.06, "learning_rate": 5.114782608695653e-06, "loss": 0.0465, "step": 6125 }, { "epoch": 1.07, "learning_rate": 5.09304347826087e-06, "loss": 0.0589, "step": 6150 }, { "epoch": 1.07, "learning_rate": 5.071304347826087e-06, "loss": 0.0379, "step": 6175 }, { "epoch": 1.07, "learning_rate": 5.049565217391305e-06, "loss": 0.0527, "step": 6200 }, { "epoch": 1.07, "learning_rate": 5.0278260869565225e-06, "loss": 0.0426, "step": 6225 }, { "epoch": 1.07, "learning_rate": 5.006086956521739e-06, "loss": 0.0435, "step": 6250 }, { "epoch": 1.08, "learning_rate": 4.984347826086957e-06, "loss": 0.0498, "step": 6275 }, { "epoch": 1.08, "learning_rate": 4.962608695652174e-06, "loss": 0.0561, "step": 6300 }, { "epoch": 1.08, "learning_rate": 4.940869565217392e-06, "loss": 0.0555, "step": 6325 }, { "epoch": 1.08, "learning_rate": 4.9191304347826094e-06, "loss": 0.0301, "step": 6350 }, { "epoch": 1.08, "learning_rate": 4.897391304347827e-06, "loss": 0.0508, "step": 6375 }, { "epoch": 1.09, "learning_rate": 4.875652173913044e-06, "loss": 0.0539, "step": 6400 }, { "epoch": 1.09, "learning_rate": 4.853913043478261e-06, "loss": 0.0505, "step": 6425 }, { "epoch": 1.09, "learning_rate": 4.832173913043479e-06, "loss": 0.0442, "step": 6450 }, { "epoch": 1.09, "learning_rate": 4.810434782608696e-06, "loss": 0.0352, "step": 6475 }, { "epoch": 1.09, "learning_rate": 4.788695652173914e-06, "loss": 0.0475, "step": 6500 }, { "epoch": 1.1, "learning_rate": 4.766956521739131e-06, "loss": 0.0377, "step": 6525 }, { "epoch": 1.1, "learning_rate": 4.745217391304348e-06, "loss": 0.0367, "step": 6550 }, { "epoch": 1.1, "learning_rate": 4.723478260869566e-06, "loss": 0.0545, "step": 6575 }, { "epoch": 1.1, "learning_rate": 4.701739130434783e-06, "loss": 0.0416, "step": 6600 }, { "epoch": 1.1, "learning_rate": 4.680000000000001e-06, "loss": 0.0544, "step": 6625 }, { "epoch": 1.11, "learning_rate": 4.658260869565218e-06, "loss": 0.0699, "step": 6650 }, { "epoch": 1.11, "learning_rate": 4.636521739130435e-06, "loss": 0.0502, "step": 6675 }, { "epoch": 1.11, "learning_rate": 4.614782608695653e-06, "loss": 0.0378, "step": 6700 }, { "epoch": 1.11, "learning_rate": 4.59304347826087e-06, "loss": 0.0616, "step": 6725 }, { "epoch": 1.12, "learning_rate": 4.572173913043479e-06, "loss": 0.062, "step": 6750 }, { "epoch": 1.12, "learning_rate": 4.5504347826086955e-06, "loss": 0.0466, "step": 6775 }, { "epoch": 1.12, "learning_rate": 4.528695652173913e-06, "loss": 0.0534, "step": 6800 }, { "epoch": 1.12, "learning_rate": 4.506956521739131e-06, "loss": 0.0429, "step": 6825 }, { "epoch": 1.12, "learning_rate": 4.485217391304348e-06, "loss": 0.0427, "step": 6850 }, { "epoch": 1.13, "learning_rate": 4.463478260869566e-06, "loss": 0.0486, "step": 6875 }, { "epoch": 1.13, "learning_rate": 4.4417391304347825e-06, "loss": 0.0494, "step": 6900 }, { "epoch": 1.13, "learning_rate": 4.42e-06, "loss": 0.0322, "step": 6925 }, { "epoch": 1.13, "learning_rate": 4.398260869565218e-06, "loss": 0.0463, "step": 6950 }, { "epoch": 1.13, "learning_rate": 4.376521739130435e-06, "loss": 0.0419, "step": 6975 }, { "epoch": 1.14, "learning_rate": 4.354782608695653e-06, "loss": 0.0323, "step": 7000 }, { "epoch": 1.14, "eval_loss": 0.14599578082561493, "eval_runtime": 15650.561, "eval_samples_per_second": 0.686, "eval_steps_per_second": 0.172, "eval_wer": 7.14453416085665, "step": 7000 }, { "epoch": 1.14, "learning_rate": 4.33304347826087e-06, "loss": 0.0308, "step": 7025 }, { "epoch": 1.14, "learning_rate": 4.311304347826087e-06, "loss": 0.0281, "step": 7050 }, { "epoch": 1.14, "learning_rate": 4.2895652173913046e-06, "loss": 0.0412, "step": 7075 }, { "epoch": 1.14, "learning_rate": 4.267826086956522e-06, "loss": 0.0434, "step": 7100 }, { "epoch": 1.15, "learning_rate": 4.24608695652174e-06, "loss": 0.0412, "step": 7125 }, { "epoch": 1.15, "learning_rate": 4.224347826086957e-06, "loss": 0.0419, "step": 7150 }, { "epoch": 1.15, "learning_rate": 4.202608695652175e-06, "loss": 0.0423, "step": 7175 }, { "epoch": 1.15, "learning_rate": 4.1808695652173915e-06, "loss": 0.0353, "step": 7200 }, { "epoch": 1.15, "learning_rate": 4.159130434782609e-06, "loss": 0.0444, "step": 7225 }, { "epoch": 1.16, "learning_rate": 4.137391304347827e-06, "loss": 0.0309, "step": 7250 }, { "epoch": 1.16, "learning_rate": 4.115652173913044e-06, "loss": 0.0379, "step": 7275 }, { "epoch": 1.16, "learning_rate": 4.093913043478262e-06, "loss": 0.0308, "step": 7300 }, { "epoch": 1.16, "learning_rate": 4.0721739130434785e-06, "loss": 0.0411, "step": 7325 }, { "epoch": 1.17, "learning_rate": 4.050434782608696e-06, "loss": 0.0402, "step": 7350 }, { "epoch": 1.17, "learning_rate": 4.028695652173914e-06, "loss": 0.0309, "step": 7375 }, { "epoch": 1.17, "learning_rate": 4.006956521739131e-06, "loss": 0.0191, "step": 7400 }, { "epoch": 1.17, "learning_rate": 3.985217391304348e-06, "loss": 0.0289, "step": 7425 }, { "epoch": 1.17, "learning_rate": 3.9634782608695655e-06, "loss": 0.0349, "step": 7450 }, { "epoch": 1.18, "learning_rate": 3.941739130434783e-06, "loss": 0.0527, "step": 7475 }, { "epoch": 1.18, "learning_rate": 3.920000000000001e-06, "loss": 0.031, "step": 7500 }, { "epoch": 1.18, "learning_rate": 3.898260869565218e-06, "loss": 0.0334, "step": 7525 }, { "epoch": 1.18, "learning_rate": 3.876521739130435e-06, "loss": 0.0367, "step": 7550 }, { "epoch": 1.18, "learning_rate": 3.8547826086956524e-06, "loss": 0.0295, "step": 7575 }, { "epoch": 1.19, "learning_rate": 3.83304347826087e-06, "loss": 0.0329, "step": 7600 }, { "epoch": 1.19, "learning_rate": 3.8113043478260876e-06, "loss": 0.0245, "step": 7625 }, { "epoch": 1.19, "learning_rate": 3.7895652173913043e-06, "loss": 0.036, "step": 7650 }, { "epoch": 1.19, "learning_rate": 3.767826086956522e-06, "loss": 0.0353, "step": 7675 }, { "epoch": 1.19, "learning_rate": 3.7460869565217394e-06, "loss": 0.0436, "step": 7700 }, { "epoch": 1.2, "learning_rate": 3.724347826086957e-06, "loss": 0.0382, "step": 7725 }, { "epoch": 1.2, "learning_rate": 3.702608695652174e-06, "loss": 0.0489, "step": 7750 }, { "epoch": 1.2, "learning_rate": 3.6808695652173917e-06, "loss": 0.0412, "step": 7775 }, { "epoch": 1.2, "learning_rate": 3.66e-06, "loss": 0.0395, "step": 7800 }, { "epoch": 1.2, "learning_rate": 3.6382608695652177e-06, "loss": 0.0495, "step": 7825 }, { "epoch": 1.21, "learning_rate": 3.6165217391304352e-06, "loss": 0.031, "step": 7850 }, { "epoch": 1.21, "learning_rate": 3.5947826086956524e-06, "loss": 0.0339, "step": 7875 }, { "epoch": 1.21, "learning_rate": 3.5730434782608695e-06, "loss": 0.0577, "step": 7900 }, { "epoch": 1.21, "learning_rate": 3.551304347826087e-06, "loss": 0.0458, "step": 7925 }, { "epoch": 1.22, "learning_rate": 3.5295652173913046e-06, "loss": 0.0327, "step": 7950 }, { "epoch": 1.22, "learning_rate": 3.507826086956522e-06, "loss": 0.0309, "step": 7975 }, { "epoch": 1.22, "learning_rate": 3.4860869565217393e-06, "loss": 0.0213, "step": 8000 }, { "epoch": 1.22, "eval_loss": 0.1490851193666458, "eval_runtime": 15525.496, "eval_samples_per_second": 0.692, "eval_steps_per_second": 0.173, "eval_wer": 7.28438287925033, "step": 8000 }, { "epoch": 1.22, "learning_rate": 3.464347826086957e-06, "loss": 0.047, "step": 8025 }, { "epoch": 1.22, "learning_rate": 3.4426086956521745e-06, "loss": 0.028, "step": 8050 }, { "epoch": 1.23, "learning_rate": 3.4208695652173916e-06, "loss": 0.0419, "step": 8075 }, { "epoch": 1.23, "learning_rate": 3.3991304347826087e-06, "loss": 0.0449, "step": 8100 }, { "epoch": 1.23, "learning_rate": 3.3773913043478263e-06, "loss": 0.0447, "step": 8125 }, { "epoch": 1.23, "learning_rate": 3.355652173913044e-06, "loss": 0.0405, "step": 8150 }, { "epoch": 1.23, "learning_rate": 3.3339130434782614e-06, "loss": 0.0429, "step": 8175 }, { "epoch": 1.24, "learning_rate": 3.312173913043479e-06, "loss": 0.0329, "step": 8200 }, { "epoch": 1.24, "learning_rate": 3.2904347826086957e-06, "loss": 0.0215, "step": 8225 }, { "epoch": 1.24, "learning_rate": 3.2686956521739133e-06, "loss": 0.044, "step": 8250 }, { "epoch": 1.24, "learning_rate": 3.246956521739131e-06, "loss": 0.043, "step": 8275 }, { "epoch": 1.24, "learning_rate": 3.2252173913043484e-06, "loss": 0.0333, "step": 8300 }, { "epoch": 1.25, "learning_rate": 3.203478260869565e-06, "loss": 0.031, "step": 8325 }, { "epoch": 1.25, "learning_rate": 3.1817391304347827e-06, "loss": 0.031, "step": 8350 }, { "epoch": 1.25, "learning_rate": 3.1600000000000002e-06, "loss": 0.0252, "step": 8375 }, { "epoch": 1.25, "learning_rate": 3.138260869565218e-06, "loss": 0.0407, "step": 8400 }, { "epoch": 1.25, "learning_rate": 3.116521739130435e-06, "loss": 0.0233, "step": 8425 }, { "epoch": 1.26, "learning_rate": 3.094782608695652e-06, "loss": 0.0346, "step": 8450 }, { "epoch": 1.26, "learning_rate": 3.0730434782608696e-06, "loss": 0.0424, "step": 8475 }, { "epoch": 1.26, "learning_rate": 3.051304347826087e-06, "loss": 0.0516, "step": 8500 }, { "epoch": 1.26, "learning_rate": 3.0295652173913048e-06, "loss": 0.0343, "step": 8525 }, { "epoch": 1.27, "learning_rate": 3.007826086956522e-06, "loss": 0.0224, "step": 8550 }, { "epoch": 1.27, "learning_rate": 2.9860869565217395e-06, "loss": 0.0337, "step": 8575 }, { "epoch": 1.27, "learning_rate": 2.9643478260869566e-06, "loss": 0.0337, "step": 8600 }, { "epoch": 1.27, "learning_rate": 2.942608695652174e-06, "loss": 0.0373, "step": 8625 }, { "epoch": 1.27, "learning_rate": 2.9208695652173913e-06, "loss": 0.0319, "step": 8650 }, { "epoch": 1.28, "learning_rate": 2.899130434782609e-06, "loss": 0.0308, "step": 8675 }, { "epoch": 1.28, "learning_rate": 2.8773913043478264e-06, "loss": 0.0318, "step": 8700 }, { "epoch": 1.28, "learning_rate": 2.855652173913044e-06, "loss": 0.0432, "step": 8725 }, { "epoch": 1.28, "learning_rate": 2.8339130434782616e-06, "loss": 0.03, "step": 8750 }, { "epoch": 1.28, "learning_rate": 2.8121739130434783e-06, "loss": 0.0432, "step": 8775 }, { "epoch": 1.29, "learning_rate": 2.790434782608696e-06, "loss": 0.0332, "step": 8800 }, { "epoch": 1.29, "learning_rate": 2.7686956521739134e-06, "loss": 0.035, "step": 8825 }, { "epoch": 1.29, "learning_rate": 2.746956521739131e-06, "loss": 0.0418, "step": 8850 }, { "epoch": 1.29, "learning_rate": 2.7252173913043477e-06, "loss": 0.0305, "step": 8875 }, { "epoch": 1.29, "learning_rate": 2.7034782608695652e-06, "loss": 0.0277, "step": 8900 }, { "epoch": 1.3, "learning_rate": 2.681739130434783e-06, "loss": 0.0363, "step": 8925 }, { "epoch": 1.3, "learning_rate": 2.6600000000000004e-06, "loss": 0.0387, "step": 8950 }, { "epoch": 1.3, "learning_rate": 2.638260869565218e-06, "loss": 0.0474, "step": 8975 }, { "epoch": 1.3, "learning_rate": 2.6165217391304346e-06, "loss": 0.051, "step": 9000 }, { "epoch": 1.3, "eval_loss": 0.14573915302753448, "eval_runtime": 15496.944, "eval_samples_per_second": 0.693, "eval_steps_per_second": 0.173, "eval_wer": 6.958749732041689, "step": 9000 }, { "epoch": 1.3, "learning_rate": 2.594782608695652e-06, "loss": 0.0288, "step": 9025 }, { "epoch": 1.31, "learning_rate": 2.5730434782608698e-06, "loss": 0.0292, "step": 9050 }, { "epoch": 1.31, "learning_rate": 2.5513043478260873e-06, "loss": 0.0423, "step": 9075 }, { "epoch": 1.31, "learning_rate": 2.5295652173913045e-06, "loss": 0.0383, "step": 9100 }, { "epoch": 1.31, "learning_rate": 2.507826086956522e-06, "loss": 0.0267, "step": 9125 }, { "epoch": 1.32, "learning_rate": 2.486086956521739e-06, "loss": 0.0253, "step": 9150 }, { "epoch": 1.32, "learning_rate": 2.4643478260869567e-06, "loss": 0.0297, "step": 9175 }, { "epoch": 1.32, "learning_rate": 2.4426086956521743e-06, "loss": 0.027, "step": 9200 }, { "epoch": 1.32, "learning_rate": 2.4208695652173914e-06, "loss": 0.0331, "step": 9225 }, { "epoch": 1.32, "learning_rate": 2.399130434782609e-06, "loss": 0.0347, "step": 9250 }, { "epoch": 1.33, "learning_rate": 2.3773913043478266e-06, "loss": 0.0476, "step": 9275 }, { "epoch": 1.33, "learning_rate": 2.3556521739130437e-06, "loss": 0.0271, "step": 9300 }, { "epoch": 1.33, "learning_rate": 2.3339130434782613e-06, "loss": 0.0381, "step": 9325 }, { "epoch": 1.33, "learning_rate": 2.3121739130434784e-06, "loss": 0.032, "step": 9350 }, { "epoch": 1.33, "learning_rate": 2.290434782608696e-06, "loss": 0.0387, "step": 9375 }, { "epoch": 1.34, "learning_rate": 2.268695652173913e-06, "loss": 0.0248, "step": 9400 }, { "epoch": 1.34, "learning_rate": 2.2469565217391307e-06, "loss": 0.025, "step": 9425 }, { "epoch": 1.34, "learning_rate": 2.225217391304348e-06, "loss": 0.0143, "step": 9450 }, { "epoch": 1.34, "learning_rate": 2.2034782608695654e-06, "loss": 0.0317, "step": 9475 }, { "epoch": 1.34, "learning_rate": 2.181739130434783e-06, "loss": 0.0327, "step": 9500 }, { "epoch": 1.35, "learning_rate": 2.16e-06, "loss": 0.0145, "step": 9525 }, { "epoch": 1.35, "learning_rate": 2.1382608695652176e-06, "loss": 0.0378, "step": 9550 }, { "epoch": 1.35, "learning_rate": 2.1165217391304348e-06, "loss": 0.0154, "step": 9575 }, { "epoch": 1.35, "learning_rate": 2.0947826086956523e-06, "loss": 0.018, "step": 9600 }, { "epoch": 1.35, "learning_rate": 2.0730434782608695e-06, "loss": 0.0266, "step": 9625 }, { "epoch": 1.36, "learning_rate": 2.051304347826087e-06, "loss": 0.0201, "step": 9650 }, { "epoch": 1.36, "learning_rate": 2.0295652173913046e-06, "loss": 0.0245, "step": 9675 }, { "epoch": 1.36, "learning_rate": 2.0078260869565217e-06, "loss": 0.0259, "step": 9700 }, { "epoch": 1.36, "learning_rate": 1.9860869565217393e-06, "loss": 0.0186, "step": 9725 }, { "epoch": 1.37, "learning_rate": 1.964347826086957e-06, "loss": 0.0185, "step": 9750 }, { "epoch": 1.37, "learning_rate": 1.942608695652174e-06, "loss": 0.0183, "step": 9775 }, { "epoch": 1.37, "learning_rate": 1.9208695652173916e-06, "loss": 0.0204, "step": 9800 }, { "epoch": 1.37, "learning_rate": 1.899130434782609e-06, "loss": 0.0143, "step": 9825 }, { "epoch": 1.37, "learning_rate": 1.8773913043478263e-06, "loss": 0.0252, "step": 9850 }, { "epoch": 1.38, "learning_rate": 1.8556521739130436e-06, "loss": 0.0219, "step": 9875 }, { "epoch": 1.38, "learning_rate": 1.833913043478261e-06, "loss": 0.016, "step": 9900 }, { "epoch": 1.38, "learning_rate": 1.8121739130434785e-06, "loss": 0.0254, "step": 9925 }, { "epoch": 1.38, "learning_rate": 1.7904347826086957e-06, "loss": 0.0297, "step": 9950 }, { "epoch": 1.38, "learning_rate": 1.7686956521739132e-06, "loss": 0.0172, "step": 9975 }, { "epoch": 1.39, "learning_rate": 1.7469565217391304e-06, "loss": 0.0196, "step": 10000 }, { "epoch": 1.39, "eval_loss": 0.14200527966022491, "eval_runtime": 15478.2147, "eval_samples_per_second": 0.694, "eval_steps_per_second": 0.174, "eval_wer": 6.608617539275033, "step": 10000 }, { "epoch": 1.39, "learning_rate": 1.725217391304348e-06, "loss": 0.0269, "step": 10025 }, { "epoch": 1.39, "learning_rate": 1.7034782608695655e-06, "loss": 0.0351, "step": 10050 }, { "epoch": 1.39, "learning_rate": 1.6817391304347826e-06, "loss": 0.0201, "step": 10075 }, { "epoch": 1.39, "learning_rate": 1.6600000000000002e-06, "loss": 0.0255, "step": 10100 }, { "epoch": 1.4, "learning_rate": 1.6382608695652176e-06, "loss": 0.0242, "step": 10125 }, { "epoch": 1.4, "learning_rate": 1.616521739130435e-06, "loss": 0.0374, "step": 10150 }, { "epoch": 1.4, "learning_rate": 1.5947826086956523e-06, "loss": 0.0318, "step": 10175 }, { "epoch": 1.4, "learning_rate": 1.5730434782608698e-06, "loss": 0.0325, "step": 10200 }, { "epoch": 1.4, "learning_rate": 1.551304347826087e-06, "loss": 0.0334, "step": 10225 }, { "epoch": 1.41, "learning_rate": 1.5295652173913045e-06, "loss": 0.015, "step": 10250 }, { "epoch": 1.41, "learning_rate": 1.507826086956522e-06, "loss": 0.0224, "step": 10275 }, { "epoch": 1.41, "learning_rate": 1.4860869565217392e-06, "loss": 0.0314, "step": 10300 }, { "epoch": 1.41, "learning_rate": 1.4652173913043479e-06, "loss": 0.0402, "step": 10325 }, { "epoch": 1.42, "learning_rate": 1.4434782608695652e-06, "loss": 0.0403, "step": 10350 }, { "epoch": 1.42, "learning_rate": 1.4217391304347828e-06, "loss": 0.0324, "step": 10375 }, { "epoch": 1.42, "learning_rate": 1.4000000000000001e-06, "loss": 0.0257, "step": 10400 }, { "epoch": 1.42, "learning_rate": 1.3782608695652175e-06, "loss": 0.0314, "step": 10425 }, { "epoch": 1.42, "learning_rate": 1.356521739130435e-06, "loss": 0.0314, "step": 10450 }, { "epoch": 1.43, "learning_rate": 1.3347826086956522e-06, "loss": 0.0302, "step": 10475 }, { "epoch": 1.43, "learning_rate": 1.3130434782608698e-06, "loss": 0.0282, "step": 10500 }, { "epoch": 1.43, "learning_rate": 1.291304347826087e-06, "loss": 0.0281, "step": 10525 }, { "epoch": 1.43, "learning_rate": 1.2695652173913045e-06, "loss": 0.0403, "step": 10550 }, { "epoch": 1.43, "learning_rate": 1.2478260869565218e-06, "loss": 0.0359, "step": 10575 }, { "epoch": 1.44, "learning_rate": 1.2260869565217392e-06, "loss": 0.0332, "step": 10600 }, { "epoch": 1.44, "learning_rate": 1.2043478260869567e-06, "loss": 0.0261, "step": 10625 }, { "epoch": 1.44, "learning_rate": 1.182608695652174e-06, "loss": 0.0418, "step": 10650 }, { "epoch": 1.44, "learning_rate": 1.1608695652173914e-06, "loss": 0.0343, "step": 10675 }, { "epoch": 1.44, "learning_rate": 1.1391304347826088e-06, "loss": 0.0181, "step": 10700 }, { "epoch": 1.45, "learning_rate": 1.1173913043478261e-06, "loss": 0.0346, "step": 10725 }, { "epoch": 2.0, "learning_rate": 1.0956521739130435e-06, "loss": 0.0273, "step": 10750 }, { "epoch": 2.0, "learning_rate": 1.0739130434782608e-06, "loss": 0.0151, "step": 10775 }, { "epoch": 2.01, "learning_rate": 1.0521739130434784e-06, "loss": 0.0224, "step": 10800 }, { "epoch": 2.01, "learning_rate": 1.0304347826086957e-06, "loss": 0.013, "step": 10825 }, { "epoch": 2.01, "learning_rate": 1.008695652173913e-06, "loss": 0.0273, "step": 10850 }, { "epoch": 2.01, "learning_rate": 9.869565217391304e-07, "loss": 0.0115, "step": 10875 }, { "epoch": 2.01, "learning_rate": 9.65217391304348e-07, "loss": 0.015, "step": 10900 }, { "epoch": 2.02, "learning_rate": 9.434782608695653e-07, "loss": 0.0083, "step": 10925 }, { "epoch": 2.02, "learning_rate": 9.217391304347827e-07, "loss": 0.0126, "step": 10950 }, { "epoch": 2.02, "learning_rate": 9.000000000000001e-07, "loss": 0.0207, "step": 10975 }, { "epoch": 2.02, "learning_rate": 8.782608695652174e-07, "loss": 0.019, "step": 11000 }, { "epoch": 2.02, "eval_loss": 0.13033150136470795, "eval_runtime": 15439.3203, "eval_samples_per_second": 0.696, "eval_steps_per_second": 0.174, "eval_wer": 6.05534742708982, "step": 11000 }, { "epoch": 2.02, "learning_rate": 8.565217391304348e-07, "loss": 0.0107, "step": 11025 }, { "epoch": 2.03, "learning_rate": 8.347826086956523e-07, "loss": 0.014, "step": 11050 }, { "epoch": 2.03, "learning_rate": 8.130434782608697e-07, "loss": 0.0162, "step": 11075 }, { "epoch": 2.03, "learning_rate": 7.91304347826087e-07, "loss": 0.028, "step": 11100 }, { "epoch": 2.03, "learning_rate": 7.695652173913045e-07, "loss": 0.0058, "step": 11125 }, { "epoch": 2.03, "learning_rate": 7.478260869565218e-07, "loss": 0.0145, "step": 11150 }, { "epoch": 2.04, "learning_rate": 7.260869565217392e-07, "loss": 0.0122, "step": 11175 }, { "epoch": 2.04, "learning_rate": 7.043478260869565e-07, "loss": 0.0125, "step": 11200 }, { "epoch": 2.04, "learning_rate": 6.826086956521739e-07, "loss": 0.0177, "step": 11225 }, { "epoch": 2.04, "learning_rate": 6.608695652173913e-07, "loss": 0.0119, "step": 11250 }, { "epoch": 2.04, "learning_rate": 6.391304347826087e-07, "loss": 0.0147, "step": 11275 }, { "epoch": 2.05, "learning_rate": 6.173913043478262e-07, "loss": 0.0137, "step": 11300 }, { "epoch": 2.05, "learning_rate": 5.956521739130435e-07, "loss": 0.0077, "step": 11325 }, { "epoch": 2.05, "learning_rate": 5.739130434782609e-07, "loss": 0.0253, "step": 11350 }, { "epoch": 2.05, "learning_rate": 5.521739130434783e-07, "loss": 0.0197, "step": 11375 }, { "epoch": 2.06, "learning_rate": 5.304347826086957e-07, "loss": 0.0182, "step": 11400 }, { "epoch": 2.06, "learning_rate": 5.086956521739131e-07, "loss": 0.022, "step": 11425 }, { "epoch": 2.06, "learning_rate": 4.869565217391305e-07, "loss": 0.0084, "step": 11450 }, { "epoch": 2.06, "learning_rate": 4.652173913043478e-07, "loss": 0.013, "step": 11475 }, { "epoch": 2.06, "learning_rate": 4.4347826086956523e-07, "loss": 0.0059, "step": 11500 }, { "epoch": 2.07, "learning_rate": 4.217391304347827e-07, "loss": 0.0173, "step": 11525 }, { "epoch": 2.07, "learning_rate": 4.0000000000000003e-07, "loss": 0.0118, "step": 11550 }, { "epoch": 2.07, "learning_rate": 3.7826086956521744e-07, "loss": 0.0171, "step": 11575 }, { "epoch": 2.07, "learning_rate": 3.565217391304348e-07, "loss": 0.0144, "step": 11600 }, { "epoch": 2.07, "learning_rate": 3.347826086956522e-07, "loss": 0.0097, "step": 11625 }, { "epoch": 2.08, "learning_rate": 3.130434782608696e-07, "loss": 0.0185, "step": 11650 }, { "epoch": 2.08, "learning_rate": 2.91304347826087e-07, "loss": 0.01, "step": 11675 }, { "epoch": 2.08, "learning_rate": 2.6956521739130435e-07, "loss": 0.0213, "step": 11700 }, { "epoch": 2.08, "learning_rate": 2.4782608695652176e-07, "loss": 0.0232, "step": 11725 }, { "epoch": 2.08, "learning_rate": 2.2608695652173916e-07, "loss": 0.0201, "step": 11750 }, { "epoch": 2.09, "learning_rate": 2.0434782608695654e-07, "loss": 0.0088, "step": 11775 }, { "epoch": 2.09, "learning_rate": 1.8260869565217394e-07, "loss": 0.0094, "step": 11800 }, { "epoch": 2.09, "learning_rate": 1.6086956521739132e-07, "loss": 0.0047, "step": 11825 }, { "epoch": 2.09, "learning_rate": 1.3913043478260872e-07, "loss": 0.005, "step": 11850 }, { "epoch": 2.09, "learning_rate": 1.1739130434782609e-07, "loss": 0.0143, "step": 11875 }, { "epoch": 2.1, "learning_rate": 9.565217391304348e-08, "loss": 0.0058, "step": 11900 }, { "epoch": 2.1, "learning_rate": 7.391304347826088e-08, "loss": 0.0158, "step": 11925 }, { "epoch": 2.1, "learning_rate": 5.217391304347827e-08, "loss": 0.0112, "step": 11950 }, { "epoch": 2.1, "learning_rate": 3.0434782608695655e-08, "loss": 0.0217, "step": 11975 }, { "epoch": 2.11, "learning_rate": 8.695652173913045e-09, "loss": 0.0124, "step": 12000 }, { "epoch": 2.11, "eval_loss": 0.13104619085788727, "eval_runtime": 15542.0789, "eval_samples_per_second": 0.691, "eval_steps_per_second": 0.173, "eval_wer": 5.895082837397793, "step": 12000 }, { "epoch": 2.11, "step": 12000, "total_flos": 1.019122089984e+20, "train_loss": 0.08123690730022888, "train_runtime": 223165.8508, "train_samples_per_second": 0.215, "train_steps_per_second": 0.054 } ], "max_steps": 12000, "num_train_epochs": 9223372036854775807, "total_flos": 1.019122089984e+20, "trial_name": null, "trial_params": null }