| { | |
| "best_metric": 5.408751772230669, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 20.018, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "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": 1.0204, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.746, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.3574, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.2282, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.2058, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.1918, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.187, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.1812, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.1882, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.2561, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.2527, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.2408, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.2106, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.2208, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.2327, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.2039, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.1939, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.1409, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.1443, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.1619, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.975789473684211e-06, | |
| "loss": 0.163, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.949473684210526e-06, | |
| "loss": 0.2023, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.923157894736844e-06, | |
| "loss": 0.1394, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.89684210526316e-06, | |
| "loss": 0.1366, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.870526315789474e-06, | |
| "loss": 0.1357, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.84421052631579e-06, | |
| "loss": 0.1182, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.817894736842106e-06, | |
| "loss": 0.1172, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.791578947368422e-06, | |
| "loss": 0.1117, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.765263157894737e-06, | |
| "loss": 0.149, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.738947368421054e-06, | |
| "loss": 0.1412, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.712631578947369e-06, | |
| "loss": 0.1389, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.686315789473684e-06, | |
| "loss": 0.1166, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.1132, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.633684210526316e-06, | |
| "loss": 0.1156, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.607368421052632e-06, | |
| "loss": 0.112, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.581052631578947e-06, | |
| "loss": 0.0988, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.554736842105264e-06, | |
| "loss": 0.0742, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.52842105263158e-06, | |
| "loss": 0.0642, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.502105263157896e-06, | |
| "loss": 0.0683, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.475789473684212e-06, | |
| "loss": 0.0917, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.19435766339302063, | |
| "eval_runtime": 3575.9982, | |
| "eval_samples_per_second": 4.393, | |
| "eval_steps_per_second": 0.137, | |
| "eval_wer": 6.85604675481309, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.449473684210527e-06, | |
| "loss": 0.1175, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.423157894736842e-06, | |
| "loss": 0.2018, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.396842105263159e-06, | |
| "loss": 0.1576, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.370526315789474e-06, | |
| "loss": 0.1525, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.34421052631579e-06, | |
| "loss": 0.1179, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.317894736842105e-06, | |
| "loss": 0.136, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.291578947368422e-06, | |
| "loss": 0.1421, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.265263157894737e-06, | |
| "loss": 0.1398, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.238947368421052e-06, | |
| "loss": 0.0908, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.21263157894737e-06, | |
| "loss": 0.0944, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.186315789473685e-06, | |
| "loss": 0.1004, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.16e-06, | |
| "loss": 0.1261, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.133684210526317e-06, | |
| "loss": 0.1641, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.107368421052632e-06, | |
| "loss": 0.1871, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.081052631578949e-06, | |
| "loss": 0.1261, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.054736842105264e-06, | |
| "loss": 0.1048, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.02842105263158e-06, | |
| "loss": 0.1084, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.002105263157895e-06, | |
| "loss": 0.0935, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.97578947368421e-06, | |
| "loss": 0.09, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.949473684210527e-06, | |
| "loss": 0.1406, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 8.923157894736842e-06, | |
| "loss": 0.1142, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 8.896842105263159e-06, | |
| "loss": 0.1147, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 8.870526315789474e-06, | |
| "loss": 0.1113, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 8.84421052631579e-06, | |
| "loss": 0.1017, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 8.817894736842107e-06, | |
| "loss": 0.0943, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 8.791578947368422e-06, | |
| "loss": 0.1025, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 8.765263157894739e-06, | |
| "loss": 0.1106, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 8.738947368421053e-06, | |
| "loss": 0.0973, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 8.712631578947368e-06, | |
| "loss": 0.1006, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 8.686315789473685e-06, | |
| "loss": 0.1063, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.1021, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 8.633684210526317e-06, | |
| "loss": 0.0952, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.607368421052632e-06, | |
| "loss": 0.1036, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.581052631578948e-06, | |
| "loss": 0.1087, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.554736842105263e-06, | |
| "loss": 0.1038, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 8.528421052631578e-06, | |
| "loss": 0.1063, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 8.502105263157897e-06, | |
| "loss": 0.0947, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 8.475789473684212e-06, | |
| "loss": 0.0935, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 8.449473684210527e-06, | |
| "loss": 0.088, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 8.423157894736843e-06, | |
| "loss": 0.0927, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.18174529075622559, | |
| "eval_runtime": 3417.7584, | |
| "eval_samples_per_second": 4.596, | |
| "eval_steps_per_second": 0.144, | |
| "eval_wer": 6.143946829872271, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.396842105263158e-06, | |
| "loss": 0.1055, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.370526315789475e-06, | |
| "loss": 0.0981, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.34421052631579e-06, | |
| "loss": 0.0963, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.317894736842107e-06, | |
| "loss": 0.105, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.291578947368422e-06, | |
| "loss": 0.0814, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.265263157894737e-06, | |
| "loss": 0.0898, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.238947368421053e-06, | |
| "loss": 0.0953, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 8.212631578947368e-06, | |
| "loss": 0.1041, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.186315789473685e-06, | |
| "loss": 0.0988, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.16e-06, | |
| "loss": 0.094, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.133684210526316e-06, | |
| "loss": 0.1029, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 8.107368421052633e-06, | |
| "loss": 0.1002, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.081052631578948e-06, | |
| "loss": 0.0932, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.054736842105265e-06, | |
| "loss": 0.105, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.02842105263158e-06, | |
| "loss": 0.0969, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.002105263157895e-06, | |
| "loss": 0.0907, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 7.975789473684211e-06, | |
| "loss": 0.0844, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 7.949473684210526e-06, | |
| "loss": 0.0839, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 7.923157894736843e-06, | |
| "loss": 0.0884, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 7.896842105263158e-06, | |
| "loss": 0.0883, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 7.870526315789475e-06, | |
| "loss": 0.126, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 7.84421052631579e-06, | |
| "loss": 0.1088, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 7.817894736842105e-06, | |
| "loss": 0.0882, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 7.791578947368423e-06, | |
| "loss": 0.0832, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 7.765263157894738e-06, | |
| "loss": 0.0745, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 7.738947368421053e-06, | |
| "loss": 0.1139, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 7.71263157894737e-06, | |
| "loss": 0.0985, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 7.686315789473685e-06, | |
| "loss": 0.0797, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.0937, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 7.633684210526316e-06, | |
| "loss": 0.0865, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 7.607368421052632e-06, | |
| "loss": 0.0705, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 7.581052631578948e-06, | |
| "loss": 0.1005, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 7.554736842105264e-06, | |
| "loss": 0.2297, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 7.5284210526315794e-06, | |
| "loss": 0.1151, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 7.502105263157895e-06, | |
| "loss": 0.1592, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 7.475789473684211e-06, | |
| "loss": 0.1107, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 7.449473684210526e-06, | |
| "loss": 0.113, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 7.4231578947368436e-06, | |
| "loss": 0.1513, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 7.3968421052631585e-06, | |
| "loss": 0.0466, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.370526315789474e-06, | |
| "loss": 0.0456, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "eval_loss": 0.18054412305355072, | |
| "eval_runtime": 3441.4578, | |
| "eval_samples_per_second": 4.564, | |
| "eval_steps_per_second": 0.143, | |
| "eval_wer": 6.26263015069574, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.34421052631579e-06, | |
| "loss": 0.0445, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.317894736842106e-06, | |
| "loss": 0.0387, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 7.291578947368422e-06, | |
| "loss": 0.0377, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.265263157894738e-06, | |
| "loss": 0.0372, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.2389473684210534e-06, | |
| "loss": 0.041, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.212631578947369e-06, | |
| "loss": 0.0363, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.186315789473684e-06, | |
| "loss": 0.0364, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.16e-06, | |
| "loss": 0.0397, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.133684210526316e-06, | |
| "loss": 0.0391, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.107368421052632e-06, | |
| "loss": 0.0353, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 7.0810526315789475e-06, | |
| "loss": 0.0336, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.054736842105264e-06, | |
| "loss": 0.0408, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.02842105263158e-06, | |
| "loss": 0.0356, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 7.002105263157896e-06, | |
| "loss": 0.0427, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 6.975789473684212e-06, | |
| "loss": 0.0376, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 6.9494736842105275e-06, | |
| "loss": 0.0333, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 6.9231578947368424e-06, | |
| "loss": 0.0397, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 6.896842105263158e-06, | |
| "loss": 0.0495, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 6.870526315789474e-06, | |
| "loss": 0.0343, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.84421052631579e-06, | |
| "loss": 0.0778, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.817894736842106e-06, | |
| "loss": 0.0664, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.7915789473684215e-06, | |
| "loss": 0.0503, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.765263157894737e-06, | |
| "loss": 0.0496, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.738947368421052e-06, | |
| "loss": 0.0477, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.71263157894737e-06, | |
| "loss": 0.0538, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.686315789473685e-06, | |
| "loss": 0.0528, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.0314, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 6.6336842105263164e-06, | |
| "loss": 0.0305, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 6.607368421052632e-06, | |
| "loss": 0.0302, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 6.581052631578948e-06, | |
| "loss": 0.0467, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 6.554736842105264e-06, | |
| "loss": 0.0615, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 6.52842105263158e-06, | |
| "loss": 0.0846, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 6.5021052631578955e-06, | |
| "loss": 0.0502, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 6.4757894736842105e-06, | |
| "loss": 0.0338, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 6.449473684210526e-06, | |
| "loss": 0.0397, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 6.423157894736842e-06, | |
| "loss": 0.032, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 6.396842105263158e-06, | |
| "loss": 0.0297, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 6.370526315789474e-06, | |
| "loss": 0.0136, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 6.3442105263157904e-06, | |
| "loss": 0.014, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.317894736842106e-06, | |
| "loss": 0.0343, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "eval_loss": 0.20974387228488922, | |
| "eval_runtime": 3414.1266, | |
| "eval_samples_per_second": 4.601, | |
| "eval_steps_per_second": 0.144, | |
| "eval_wer": 6.177306465995624, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.291578947368422e-06, | |
| "loss": 0.0198, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.265263157894738e-06, | |
| "loss": 0.0169, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 6.238947368421054e-06, | |
| "loss": 0.0147, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.212631578947369e-06, | |
| "loss": 0.0172, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.1863157894736845e-06, | |
| "loss": 0.0182, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.16e-06, | |
| "loss": 0.017, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.133684210526316e-06, | |
| "loss": 0.0104, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.107368421052632e-06, | |
| "loss": 0.0088, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.081052631578948e-06, | |
| "loss": 0.0103, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.054736842105264e-06, | |
| "loss": 0.0189, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 6.0284210526315786e-06, | |
| "loss": 0.0218, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 6.002105263157896e-06, | |
| "loss": 0.0323, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 5.975789473684212e-06, | |
| "loss": 0.0129, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 5.949473684210527e-06, | |
| "loss": 0.0099, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 5.923157894736843e-06, | |
| "loss": 0.0134, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 5.8968421052631585e-06, | |
| "loss": 0.0093, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5.870526315789474e-06, | |
| "loss": 0.0115, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5.84421052631579e-06, | |
| "loss": 0.0174, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 5.817894736842106e-06, | |
| "loss": 0.0147, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 5.791578947368422e-06, | |
| "loss": 0.0156, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 5.765263157894737e-06, | |
| "loss": 0.0156, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 5.7389473684210526e-06, | |
| "loss": 0.0115, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.712631578947368e-06, | |
| "loss": 0.0118, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.686315789473684e-06, | |
| "loss": 0.0145, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.66e-06, | |
| "loss": 0.0102, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.633684210526317e-06, | |
| "loss": 0.0133, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 5.6073684210526325e-06, | |
| "loss": 0.0141, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 5.581052631578948e-06, | |
| "loss": 0.014, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 5.554736842105264e-06, | |
| "loss": 0.011, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 5.52842105263158e-06, | |
| "loss": 0.012, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 5.502105263157895e-06, | |
| "loss": 0.0124, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 5.475789473684211e-06, | |
| "loss": 0.0128, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 5.4494736842105266e-06, | |
| "loss": 0.0132, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 5.423157894736842e-06, | |
| "loss": 0.0128, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 5.396842105263158e-06, | |
| "loss": 0.0107, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 5.370526315789474e-06, | |
| "loss": 0.0148, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.34421052631579e-06, | |
| "loss": 0.0131, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.317894736842105e-06, | |
| "loss": 0.006, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.291578947368422e-06, | |
| "loss": 0.0049, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.265263157894738e-06, | |
| "loss": 0.0046, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "eval_loss": 0.22919650375843048, | |
| "eval_runtime": 3664.5783, | |
| "eval_samples_per_second": 4.286, | |
| "eval_steps_per_second": 0.134, | |
| "eval_wer": 5.937373698493042, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.238947368421053e-06, | |
| "loss": 0.0047, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 10.01, | |
| "learning_rate": 5.212631578947369e-06, | |
| "loss": 0.0036, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.186315789473685e-06, | |
| "loss": 0.0046, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.1600000000000006e-06, | |
| "loss": 0.005, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.133684210526316e-06, | |
| "loss": 0.0042, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 5.107368421052632e-06, | |
| "loss": 0.005, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.081052631578948e-06, | |
| "loss": 0.0043, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.054736842105263e-06, | |
| "loss": 0.0037, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.028421052631579e-06, | |
| "loss": 0.003, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.002105263157895e-06, | |
| "loss": 0.0037, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 4.975789473684211e-06, | |
| "loss": 0.0047, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 4.949473684210527e-06, | |
| "loss": 0.0045, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 4.923157894736842e-06, | |
| "loss": 0.0038, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 4.896842105263158e-06, | |
| "loss": 0.0041, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 4.870526315789474e-06, | |
| "loss": 0.004, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 4.84421052631579e-06, | |
| "loss": 0.0049, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 4.817894736842106e-06, | |
| "loss": 0.0042, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 4.791578947368421e-06, | |
| "loss": 0.0075, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 4.765263157894737e-06, | |
| "loss": 0.0129, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 4.738947368421053e-06, | |
| "loss": 0.0067, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 4.712631578947369e-06, | |
| "loss": 0.0071, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 4.6863157894736845e-06, | |
| "loss": 0.0051, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 4.66e-06, | |
| "loss": 0.007, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 4.633684210526316e-06, | |
| "loss": 0.007, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 4.607368421052632e-06, | |
| "loss": 0.0062, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 4.581052631578948e-06, | |
| "loss": 0.0045, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 4.5547368421052636e-06, | |
| "loss": 0.0041, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 4.528421052631579e-06, | |
| "loss": 0.0048, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 4.502105263157895e-06, | |
| "loss": 0.007, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 4.475789473684211e-06, | |
| "loss": 0.0092, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.449473684210527e-06, | |
| "loss": 0.0121, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.423157894736843e-06, | |
| "loss": 0.0049, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.3968421052631585e-06, | |
| "loss": 0.0052, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 4.370526315789474e-06, | |
| "loss": 0.0053, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 4.344210526315789e-06, | |
| "loss": 0.0045, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.317894736842105e-06, | |
| "loss": 0.0364, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.291578947368422e-06, | |
| "loss": 0.1029, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.2652631578947376e-06, | |
| "loss": 0.0873, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.238947368421053e-06, | |
| "loss": 0.0854, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.212631578947368e-06, | |
| "loss": 0.0829, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "eval_loss": 0.18142107129096985, | |
| "eval_runtime": 3453.737, | |
| "eval_samples_per_second": 4.548, | |
| "eval_steps_per_second": 0.142, | |
| "eval_wer": 6.0643969283473504, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 4.186315789473684e-06, | |
| "loss": 0.0766, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.16e-06, | |
| "loss": 0.0843, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.133684210526316e-06, | |
| "loss": 0.0897, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.1073684210526325e-06, | |
| "loss": 0.0749, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.0810526315789474e-06, | |
| "loss": 0.0845, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.054736842105263e-06, | |
| "loss": 0.0723, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.028421052631579e-06, | |
| "loss": 0.0691, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.002105263157895e-06, | |
| "loss": 0.074, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 3.975789473684211e-06, | |
| "loss": 0.0826, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 3.9494736842105265e-06, | |
| "loss": 0.0829, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 3.923157894736842e-06, | |
| "loss": 0.0716, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 3.896842105263158e-06, | |
| "loss": 0.0872, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 3.870526315789474e-06, | |
| "loss": 0.096, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 12.05, | |
| "learning_rate": 3.84421052631579e-06, | |
| "loss": 0.1299, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 12.05, | |
| "learning_rate": 3.817894736842106e-06, | |
| "loss": 0.079, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 3.791578947368421e-06, | |
| "loss": 0.0314, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 3.7652631578947373e-06, | |
| "loss": 0.0057, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 3.738947368421053e-06, | |
| "loss": 0.0086, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 3.712631578947369e-06, | |
| "loss": 0.0054, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 3.6863157894736847e-06, | |
| "loss": 0.0043, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 3.66e-06, | |
| "loss": 0.0035, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 3.633684210526316e-06, | |
| "loss": 0.004, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 3.6073684210526317e-06, | |
| "loss": 0.0038, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 3.581052631578948e-06, | |
| "loss": 0.0038, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 13.02, | |
| "learning_rate": 3.554736842105264e-06, | |
| "loss": 0.0022, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.528421052631579e-06, | |
| "loss": 0.0023, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.502105263157895e-06, | |
| "loss": 0.0024, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.475789473684211e-06, | |
| "loss": 0.0034, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.4494736842105262e-06, | |
| "loss": 0.0036, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.423157894736842e-06, | |
| "loss": 0.0071, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.3968421052631583e-06, | |
| "loss": 0.0038, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.370526315789474e-06, | |
| "loss": 0.0022, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 3.34421052631579e-06, | |
| "loss": 0.0026, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 3.3178947368421053e-06, | |
| "loss": 0.0021, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 3.291578947368421e-06, | |
| "loss": 0.002, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 3.265263157894737e-06, | |
| "loss": 0.0024, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.238947368421053e-06, | |
| "loss": 0.0024, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.212631578947369e-06, | |
| "loss": 0.0025, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.1863157894736844e-06, | |
| "loss": 0.0027, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.1600000000000002e-06, | |
| "loss": 0.0021, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "eval_loss": 0.2318367063999176, | |
| "eval_runtime": 3428.375, | |
| "eval_samples_per_second": 4.582, | |
| "eval_steps_per_second": 0.143, | |
| "eval_wer": 5.709630028804763, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.133684210526316e-06, | |
| "loss": 0.0024, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.107368421052632e-06, | |
| "loss": 0.0028, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.0810526315789473e-06, | |
| "loss": 0.0023, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 3.0547368421052635e-06, | |
| "loss": 0.0025, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 3.0284210526315793e-06, | |
| "loss": 0.0024, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 3.002105263157895e-06, | |
| "loss": 0.002, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 2.975789473684211e-06, | |
| "loss": 0.002, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 14.03, | |
| "learning_rate": 2.9494736842105264e-06, | |
| "loss": 0.0025, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 2.923157894736842e-06, | |
| "loss": 0.0027, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 2.896842105263158e-06, | |
| "loss": 0.0023, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 2.8705263157894742e-06, | |
| "loss": 0.0023, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 2.84421052631579e-06, | |
| "loss": 0.0027, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 14.05, | |
| "learning_rate": 2.8178947368421055e-06, | |
| "loss": 0.0025, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 14.05, | |
| "learning_rate": 2.7915789473684213e-06, | |
| "loss": 0.0019, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 2.765263157894737e-06, | |
| "loss": 0.0217, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 2.7389473684210525e-06, | |
| "loss": 0.0268, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 2.7126315789473683e-06, | |
| "loss": 0.0533, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 2.6863157894736845e-06, | |
| "loss": 0.0405, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 2.6600000000000004e-06, | |
| "loss": 0.0381, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 2.633684210526316e-06, | |
| "loss": 0.029, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 2.6073684210526316e-06, | |
| "loss": 0.0233, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 2.5810526315789474e-06, | |
| "loss": 0.0381, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 2.5547368421052632e-06, | |
| "loss": 0.0414, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "learning_rate": 2.528421052631579e-06, | |
| "loss": 0.0293, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.5021052631578953e-06, | |
| "loss": 0.0321, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.4757894736842107e-06, | |
| "loss": 0.0335, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.4494736842105265e-06, | |
| "loss": 0.0284, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 15.03, | |
| "learning_rate": 2.4231578947368423e-06, | |
| "loss": 0.0356, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.396842105263158e-06, | |
| "loss": 0.0719, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.371578947368421e-06, | |
| "loss": 0.0801, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.3452631578947373e-06, | |
| "loss": 0.0499, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.3189473684210527e-06, | |
| "loss": 0.0688, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 2.2926315789473685e-06, | |
| "loss": 0.0289, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "learning_rate": 2.2663157894736843e-06, | |
| "loss": 0.0849, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.24e-06, | |
| "loss": 0.0463, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 2.213684210526316e-06, | |
| "loss": 0.0362, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.187368421052632e-06, | |
| "loss": 0.0367, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.1610526315789476e-06, | |
| "loss": 0.0314, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.1347368421052634e-06, | |
| "loss": 0.0343, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.1084210526315792e-06, | |
| "loss": 0.0288, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "eval_loss": 0.18713879585266113, | |
| "eval_runtime": 3412.9803, | |
| "eval_samples_per_second": 4.602, | |
| "eval_steps_per_second": 0.144, | |
| "eval_wer": 5.575549952847438, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.082105263157895e-06, | |
| "loss": 0.0299, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.055789473684211e-06, | |
| "loss": 0.0339, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.0294736842105263e-06, | |
| "loss": 0.036, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 16.02, | |
| "learning_rate": 2.003157894736842e-06, | |
| "loss": 0.0327, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 1.976842105263158e-06, | |
| "loss": 0.0374, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 1.9505263157894737e-06, | |
| "loss": 0.0346, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 1.9242105263157896e-06, | |
| "loss": 0.0359, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 16.03, | |
| "learning_rate": 1.8978947368421056e-06, | |
| "loss": 0.0314, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 1.8715789473684212e-06, | |
| "loss": 0.035, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 1.845263157894737e-06, | |
| "loss": 0.0348, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 1.8189473684210528e-06, | |
| "loss": 0.0334, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 1.7926315789473686e-06, | |
| "loss": 0.0297, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 16.05, | |
| "learning_rate": 1.7663157894736843e-06, | |
| "loss": 0.0269, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 1.74e-06, | |
| "loss": 0.0436, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 1.713684210526316e-06, | |
| "loss": 0.0037, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 1.6873684210526317e-06, | |
| "loss": 0.0025, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 1.6610526315789473e-06, | |
| "loss": 0.0024, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 1.6347368421052633e-06, | |
| "loss": 0.0022, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 1.6084210526315792e-06, | |
| "loss": 0.0019, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 1.5821052631578948e-06, | |
| "loss": 0.0018, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 1.5557894736842106e-06, | |
| "loss": 0.0022, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 1.5294736842105264e-06, | |
| "loss": 0.0019, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 1.5031578947368422e-06, | |
| "loss": 0.0019, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.4768421052631578e-06, | |
| "loss": 0.0018, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.4505263157894739e-06, | |
| "loss": 0.0018, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.4242105263157897e-06, | |
| "loss": 0.0016, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 17.03, | |
| "learning_rate": 1.3978947368421053e-06, | |
| "loss": 0.0015, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.3715789473684213e-06, | |
| "loss": 0.0017, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.345263157894737e-06, | |
| "loss": 0.0016, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.3189473684210527e-06, | |
| "loss": 0.0016, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 1.2926315789473683e-06, | |
| "loss": 0.0016, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 1.2663157894736844e-06, | |
| "loss": 0.0016, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 1.2400000000000002e-06, | |
| "loss": 0.0016, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 1.213684210526316e-06, | |
| "loss": 0.0699, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 1.1873684210526316e-06, | |
| "loss": 0.1066, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.1610526315789474e-06, | |
| "loss": 0.0957, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.1347368421052633e-06, | |
| "loss": 0.1084, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.108421052631579e-06, | |
| "loss": 0.0939, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 18.01, | |
| "learning_rate": 1.082105263157895e-06, | |
| "loss": 0.1022, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.0557894736842105e-06, | |
| "loss": 0.1297, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "eval_loss": 0.18305575847625732, | |
| "eval_runtime": 3456.4879, | |
| "eval_samples_per_second": 4.544, | |
| "eval_steps_per_second": 0.142, | |
| "eval_wer": 5.6884594904957115, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.0294736842105265e-06, | |
| "loss": 0.1294, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.0031578947368421e-06, | |
| "loss": 0.0878, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 9.76842105263158e-07, | |
| "loss": 0.0956, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 9.505263157894738e-07, | |
| "loss": 0.162, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 9.242105263157895e-07, | |
| "loss": 0.147, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 8.978947368421054e-07, | |
| "loss": 0.1374, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 18.03, | |
| "learning_rate": 8.715789473684211e-07, | |
| "loss": 0.1413, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 8.452631578947369e-07, | |
| "loss": 0.1494, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 8.189473684210527e-07, | |
| "loss": 0.1295, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 7.926315789473685e-07, | |
| "loss": 0.1396, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 7.663157894736843e-07, | |
| "loss": 0.1625, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 7.4e-07, | |
| "loss": 0.173, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 7.136842105263158e-07, | |
| "loss": 0.1071, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 6.873684210526315e-07, | |
| "loss": 0.0207, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 6.610526315789475e-07, | |
| "loss": 0.0031, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 6.347368421052633e-07, | |
| "loss": 0.0029, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 6.08421052631579e-07, | |
| "loss": 0.0024, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 5.821052631578948e-07, | |
| "loss": 0.0022, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 5.557894736842106e-07, | |
| "loss": 0.002, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 5.294736842105263e-07, | |
| "loss": 0.0022, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 5.031578947368422e-07, | |
| "loss": 0.0021, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.768421052631579e-07, | |
| "loss": 0.0019, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 4.505263157894737e-07, | |
| "loss": 0.0018, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 4.242105263157895e-07, | |
| "loss": 0.0018, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 3.9789473684210533e-07, | |
| "loss": 0.0018, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 3.715789473684211e-07, | |
| "loss": 0.0016, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 19.03, | |
| "learning_rate": 3.4526315789473686e-07, | |
| "loss": 0.0017, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 3.1894736842105263e-07, | |
| "loss": 0.0018, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 2.9263157894736845e-07, | |
| "loss": 0.0017, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 2.663157894736842e-07, | |
| "loss": 0.0018, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 0.0018, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 19.05, | |
| "learning_rate": 2.136842105263158e-07, | |
| "loss": 0.0016, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 1.8736842105263162e-07, | |
| "loss": 0.0142, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 1.6105263157894738e-07, | |
| "loss": 0.0614, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 1.3473684210526317e-07, | |
| "loss": 0.0736, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 1.0842105263157895e-07, | |
| "loss": 0.0581, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 8.210526315789474e-08, | |
| "loss": 0.0402, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 20.01, | |
| "learning_rate": 5.578947368421053e-08, | |
| "loss": 0.0473, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 2.9473684210526318e-08, | |
| "loss": 0.0376, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 3.1578947368421054e-09, | |
| "loss": 0.0377, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "eval_loss": 0.19153279066085815, | |
| "eval_runtime": 3452.2902, | |
| "eval_samples_per_second": 4.55, | |
| "eval_steps_per_second": 0.142, | |
| "eval_wer": 5.408751772230669, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "step": 10000, | |
| "total_flos": 6.530238440275968e+20, | |
| "train_loss": 0.0632013293929398, | |
| "train_runtime": 159334.9333, | |
| "train_samples_per_second": 4.017, | |
| "train_steps_per_second": 0.063 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 6.530238440275968e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |