| { | |
| "best_metric": 4.557596215181799, | |
| "best_model_checkpoint": "./checkpoint-6000", | |
| "epoch": 2.205, | |
| "global_step": 6000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.898977360288234e-06, | |
| "loss": 0.3855, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.160712527409633e-06, | |
| "loss": 0.16, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.85912902234906e-06, | |
| "loss": 0.1358, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.344547104469332e-06, | |
| "loss": 0.157, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.716963756434345e-06, | |
| "loss": 0.1538, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.019180844200955e-06, | |
| "loss": 0.1557, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.27351214279797e-06, | |
| "loss": 0.2211, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.49307723936858e-06, | |
| "loss": 0.1959, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.686247975778677e-06, | |
| "loss": 0.1804, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.858694625217149e-06, | |
| "loss": 0.1335, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.014436199608479e-06, | |
| "loss": 0.1477, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.156425255148058e-06, | |
| "loss": 0.1371, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.28689473531776e-06, | |
| "loss": 0.1334, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.407574351377137e-06, | |
| "loss": 0.1238, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.519831289296397e-06, | |
| "loss": 0.1206, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.624764935335318e-06, | |
| "loss": 0.1167, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.723272550712454e-06, | |
| "loss": 0.1192, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.812484046603779e-06, | |
| "loss": 0.1158, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.900435550016748e-06, | |
| "loss": 0.1409, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.98382788472848e-06, | |
| "loss": 0.1188, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.965454545454546e-06, | |
| "loss": 0.1257, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.1328, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.874545454545455e-06, | |
| "loss": 0.1327, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.82909090909091e-06, | |
| "loss": 0.1511, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.783636363636365e-06, | |
| "loss": 0.1401, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.738181818181818e-06, | |
| "loss": 0.1285, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.692727272727274e-06, | |
| "loss": 0.1363, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.64727272727273e-06, | |
| "loss": 0.131, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.601818181818183e-06, | |
| "loss": 0.112, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.556363636363637e-06, | |
| "loss": 0.1205, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.510909090909092e-06, | |
| "loss": 0.1215, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.465454545454546e-06, | |
| "loss": 0.1366, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.1123, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.374545454545455e-06, | |
| "loss": 0.1121, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.32909090909091e-06, | |
| "loss": 0.1142, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.283636363636364e-06, | |
| "loss": 0.1252, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.23818181818182e-06, | |
| "loss": 0.122, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.192727272727273e-06, | |
| "loss": 0.1449, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.147272727272727e-06, | |
| "loss": 0.1786, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.101818181818183e-06, | |
| "loss": 0.1684, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 0.1619873046875, | |
| "eval_runtime": 4324.26, | |
| "eval_samples_per_second": 3.469, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 6.461965766532001, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.056363636363638e-06, | |
| "loss": 0.135, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.010909090909092e-06, | |
| "loss": 0.1217, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.965454545454546e-06, | |
| "loss": 0.11, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.1428, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.874545454545456e-06, | |
| "loss": 0.1382, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.82909090909091e-06, | |
| "loss": 0.1444, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.783636363636364e-06, | |
| "loss": 0.1476, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.73818181818182e-06, | |
| "loss": 0.1249, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.692727272727273e-06, | |
| "loss": 0.1512, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.647272727272727e-06, | |
| "loss": 0.1272, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.601818181818182e-06, | |
| "loss": 0.1376, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.556363636363638e-06, | |
| "loss": 0.1316, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.510909090909092e-06, | |
| "loss": 0.1233, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.465454545454547e-06, | |
| "loss": 0.1411, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.1325, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.374545454545454e-06, | |
| "loss": 0.1305, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.32909090909091e-06, | |
| "loss": 0.1339, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.283636363636365e-06, | |
| "loss": 0.1362, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.238181818181819e-06, | |
| "loss": 0.1251, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.192727272727273e-06, | |
| "loss": 0.1444, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.147272727272728e-06, | |
| "loss": 0.1391, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.101818181818184e-06, | |
| "loss": 0.1318, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.056363636363636e-06, | |
| "loss": 0.1303, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.010909090909091e-06, | |
| "loss": 0.1293, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.965454545454547e-06, | |
| "loss": 0.1238, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.1288, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.874545454545454e-06, | |
| "loss": 0.1144, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.82909090909091e-06, | |
| "loss": 0.1304, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.783636363636365e-06, | |
| "loss": 0.1139, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.738181818181819e-06, | |
| "loss": 0.1238, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.692727272727273e-06, | |
| "loss": 0.1416, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.647272727272728e-06, | |
| "loss": 0.1587, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.6018181818181826e-06, | |
| "loss": 0.1372, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.556363636363637e-06, | |
| "loss": 0.1307, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.510909090909092e-06, | |
| "loss": 0.1288, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.4654545454545455e-06, | |
| "loss": 0.1367, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.1222, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.3745454545454556e-06, | |
| "loss": 0.1054, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.32909090909091e-06, | |
| "loss": 0.1028, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.283636363636364e-06, | |
| "loss": 0.1174, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_loss": 0.141845703125, | |
| "eval_runtime": 4306.3621, | |
| "eval_samples_per_second": 3.484, | |
| "eval_steps_per_second": 0.218, | |
| "eval_wer": 5.566260897299595, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.2381818181818185e-06, | |
| "loss": 0.1285, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.192727272727273e-06, | |
| "loss": 0.1318, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.1472727272727285e-06, | |
| "loss": 0.1123, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.101818181818182e-06, | |
| "loss": 0.1207, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.056363636363637e-06, | |
| "loss": 0.1016, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.0109090909090915e-06, | |
| "loss": 0.1027, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.965454545454546e-06, | |
| "loss": 0.1232, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.1082, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.874545454545455e-06, | |
| "loss": 0.0897, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.82909090909091e-06, | |
| "loss": 0.131, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.7836363636363644e-06, | |
| "loss": 0.1078, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.738181818181819e-06, | |
| "loss": 0.1435, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.692727272727273e-06, | |
| "loss": 0.127, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.647272727272727e-06, | |
| "loss": 0.1172, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.601818181818183e-06, | |
| "loss": 0.1097, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.5563636363636374e-06, | |
| "loss": 0.0885, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.510909090909091e-06, | |
| "loss": 0.0665, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.465454545454546e-06, | |
| "loss": 0.0648, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.0727, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.374545454545456e-06, | |
| "loss": 0.0684, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.3290909090909096e-06, | |
| "loss": 0.0646, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.283636363636364e-06, | |
| "loss": 0.0683, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.238181818181819e-06, | |
| "loss": 0.0664, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.192727272727273e-06, | |
| "loss": 0.0509, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.147272727272727e-06, | |
| "loss": 0.0534, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.101818181818182e-06, | |
| "loss": 0.0579, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.056363636363637e-06, | |
| "loss": 0.064, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.010909090909092e-06, | |
| "loss": 0.0574, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.965454545454546e-06, | |
| "loss": 0.0517, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.0569, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.874545454545455e-06, | |
| "loss": 0.06, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 5.82909090909091e-06, | |
| "loss": 0.0583, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 5.783636363636365e-06, | |
| "loss": 0.0695, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.7381818181818185e-06, | |
| "loss": 0.087, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.692727272727273e-06, | |
| "loss": 0.0846, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.647272727272728e-06, | |
| "loss": 0.0706, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.601818181818183e-06, | |
| "loss": 0.0616, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.556363636363636e-06, | |
| "loss": 0.0552, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.5109090909090914e-06, | |
| "loss": 0.0743, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.465454545454546e-06, | |
| "loss": 0.069, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_loss": 0.1400146484375, | |
| "eval_runtime": 4316.3895, | |
| "eval_samples_per_second": 3.476, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 5.28651924303636, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.075, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.374545454545454e-06, | |
| "loss": 0.083, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.329090909090909e-06, | |
| "loss": 0.0801, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.283636363636364e-06, | |
| "loss": 0.0741, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.238181818181819e-06, | |
| "loss": 0.0639, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.192727272727273e-06, | |
| "loss": 0.0657, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.147272727272727e-06, | |
| "loss": 0.0681, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.101818181818182e-06, | |
| "loss": 0.0614, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.056363636363637e-06, | |
| "loss": 0.0505, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.010909090909092e-06, | |
| "loss": 0.0647, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.965454545454546e-06, | |
| "loss": 0.0693, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.0682, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.8763636363636366e-06, | |
| "loss": 0.0623, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.830909090909091e-06, | |
| "loss": 0.0647, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.785454545454546e-06, | |
| "loss": 0.0514, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.74e-06, | |
| "loss": 0.0563, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.694545454545455e-06, | |
| "loss": 0.0652, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.6490909090909095e-06, | |
| "loss": 0.0544, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.603636363636364e-06, | |
| "loss": 0.0494, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.558181818181819e-06, | |
| "loss": 0.0669, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.512727272727273e-06, | |
| "loss": 0.058, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.467272727272728e-06, | |
| "loss": 0.0721, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.4218181818181825e-06, | |
| "loss": 0.0708, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.376363636363637e-06, | |
| "loss": 0.0681, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.330909090909091e-06, | |
| "loss": 0.0606, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.285454545454546e-06, | |
| "loss": 0.0685, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.0748, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.194545454545455e-06, | |
| "loss": 0.0626, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.149090909090909e-06, | |
| "loss": 0.0745, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.103636363636364e-06, | |
| "loss": 0.0632, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.0581818181818184e-06, | |
| "loss": 0.0678, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.012727272727273e-06, | |
| "loss": 0.0652, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.967272727272728e-06, | |
| "loss": 0.0585, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.921818181818182e-06, | |
| "loss": 0.0689, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.876363636363637e-06, | |
| "loss": 0.0655, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.830909090909091e-06, | |
| "loss": 0.0673, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.7854545454545456e-06, | |
| "loss": 0.063, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.74e-06, | |
| "loss": 0.0649, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.694545454545455e-06, | |
| "loss": 0.0628, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.6490909090909094e-06, | |
| "loss": 0.0649, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "eval_loss": 0.1314697265625, | |
| "eval_runtime": 4287.8068, | |
| "eval_samples_per_second": 3.499, | |
| "eval_steps_per_second": 0.219, | |
| "eval_wer": 4.89315330640017, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.603636363636364e-06, | |
| "loss": 0.0687, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.5581818181818186e-06, | |
| "loss": 0.0693, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.512727272727273e-06, | |
| "loss": 0.063, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.4672727272727273e-06, | |
| "loss": 0.0594, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.4218181818181824e-06, | |
| "loss": 0.0619, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.3763636363636365e-06, | |
| "loss": 0.0607, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.330909090909091e-06, | |
| "loss": 0.0579, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.2872727272727274e-06, | |
| "loss": 0.0629, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.241818181818182e-06, | |
| "loss": 0.0571, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.196363636363637e-06, | |
| "loss": 0.0595, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.150909090909091e-06, | |
| "loss": 0.0854, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.1054545454545458e-06, | |
| "loss": 0.0728, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.0612, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.014545454545455e-06, | |
| "loss": 0.0631, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.969090909090909e-06, | |
| "loss": 0.0757, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.9254545454545458e-06, | |
| "loss": 0.0728, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.88e-06, | |
| "loss": 0.102, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.834545454545455e-06, | |
| "loss": 0.0933, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.789090909090909e-06, | |
| "loss": 0.1039, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7436363636363637e-06, | |
| "loss": 0.0669, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.6981818181818188e-06, | |
| "loss": 0.0697, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.652727272727273e-06, | |
| "loss": 0.0587, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.6072727272727275e-06, | |
| "loss": 0.061, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.561818181818182e-06, | |
| "loss": 0.0591, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.5163636363636367e-06, | |
| "loss": 0.0572, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.470909090909091e-06, | |
| "loss": 0.0584, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.4254545454545455e-06, | |
| "loss": 0.0548, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.38e-06, | |
| "loss": 0.0561, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.3345454545454547e-06, | |
| "loss": 0.0608, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.2890909090909093e-06, | |
| "loss": 0.0644, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.243636363636364e-06, | |
| "loss": 0.0525, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.1981818181818185e-06, | |
| "loss": 0.0494, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.152727272727273e-06, | |
| "loss": 0.0504, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.1072727272727277e-06, | |
| "loss": 0.0444, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.0618181818181823e-06, | |
| "loss": 0.0381, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.0163636363636364e-06, | |
| "loss": 0.0343, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.970909090909091e-06, | |
| "loss": 0.0397, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.9254545454545456e-06, | |
| "loss": 0.0392, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 0.0314, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.8345454545454546e-06, | |
| "loss": 0.0334, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "eval_loss": 0.1368408203125, | |
| "eval_runtime": 4380.8288, | |
| "eval_samples_per_second": 3.425, | |
| "eval_steps_per_second": 0.214, | |
| "eval_wer": 4.684509887305975, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.7890909090909092e-06, | |
| "loss": 0.0366, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.7436363636363638e-06, | |
| "loss": 0.0393, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.6981818181818182e-06, | |
| "loss": 0.0384, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.6527272727272728e-06, | |
| "loss": 0.0374, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.6072727272727274e-06, | |
| "loss": 0.0337, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5618181818181817e-06, | |
| "loss": 0.0307, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5163636363636363e-06, | |
| "loss": 0.0328, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.470909090909091e-06, | |
| "loss": 0.0389, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.4254545454545457e-06, | |
| "loss": 0.0289, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 0.0362, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.3345454545454547e-06, | |
| "loss": 0.035, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.2890909090909093e-06, | |
| "loss": 0.0381, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.2436363636363637e-06, | |
| "loss": 0.0383, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.1981818181818183e-06, | |
| "loss": 0.0501, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.152727272727273e-06, | |
| "loss": 0.0354, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.1072727272727275e-06, | |
| "loss": 0.0373, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.0618181818181819e-06, | |
| "loss": 0.0354, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.0163636363636365e-06, | |
| "loss": 0.0344, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.70909090909091e-07, | |
| "loss": 0.034, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.254545454545454e-07, | |
| "loss": 0.0364, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.8e-07, | |
| "loss": 0.0399, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.345454545454545e-07, | |
| "loss": 0.0355, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 7.890909090909092e-07, | |
| "loss": 0.0372, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 7.436363636363637e-07, | |
| "loss": 0.0284, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.981818181818182e-07, | |
| "loss": 0.0312, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 6.527272727272728e-07, | |
| "loss": 0.0296, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 6.072727272727273e-07, | |
| "loss": 0.0349, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.618181818181819e-07, | |
| "loss": 0.0316, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.163636363636364e-07, | |
| "loss": 0.0305, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.70909090909091e-07, | |
| "loss": 0.0289, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.2545454545454547e-07, | |
| "loss": 0.0296, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.8e-07, | |
| "loss": 0.0319, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.3454545454545456e-07, | |
| "loss": 0.0325, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.890909090909091e-07, | |
| "loss": 0.0442, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.436363636363637e-07, | |
| "loss": 0.0448, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.9818181818181819e-07, | |
| "loss": 0.0407, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.5272727272727275e-07, | |
| "loss": 0.0386, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.0727272727272728e-07, | |
| "loss": 0.0327, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.181818181818183e-08, | |
| "loss": 0.0378, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.6363636363636366e-08, | |
| "loss": 0.037, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "eval_loss": 0.1331787109375, | |
| "eval_runtime": 4331.8871, | |
| "eval_samples_per_second": 3.463, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 4.557596215181799, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "step": 6000, | |
| "total_flos": 8.151660428090115e+20, | |
| "train_loss": 0.08706036122639974, | |
| "train_runtime": 112219.1034, | |
| "train_samples_per_second": 3.422, | |
| "train_steps_per_second": 0.053 | |
| } | |
| ], | |
| "max_steps": 6000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 8.151660428090115e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |