| { | |
| "best_metric": 14.849506681653555, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 1.0, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0025, | |
| "grad_norm": 32.17699432373047, | |
| "learning_rate": 7.875e-07, | |
| "loss": 3.5416, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 17.338768005371094, | |
| "learning_rate": 1.7249999999999998e-06, | |
| "loss": 2.8498, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0075, | |
| "grad_norm": 13.783746719360352, | |
| "learning_rate": 2.6624999999999995e-06, | |
| "loss": 1.9843, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 10.80038070678711, | |
| "learning_rate": 3.6e-06, | |
| "loss": 1.6113, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0125, | |
| "grad_norm": 9.404985427856445, | |
| "learning_rate": 4.537499999999999e-06, | |
| "loss": 1.3704, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 8.393142700195312, | |
| "learning_rate": 5.474999999999999e-06, | |
| "loss": 1.1666, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0175, | |
| "grad_norm": 9.127829551696777, | |
| "learning_rate": 6.4125e-06, | |
| "loss": 1.0771, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 9.412626266479492, | |
| "learning_rate": 7.35e-06, | |
| "loss": 1.0148, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0225, | |
| "grad_norm": 7.45149564743042, | |
| "learning_rate": 8.2875e-06, | |
| "loss": 0.8875, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 10.142014503479004, | |
| "learning_rate": 9.224999999999999e-06, | |
| "loss": 0.8914, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.0275, | |
| "grad_norm": 8.410262107849121, | |
| "learning_rate": 1.01625e-05, | |
| "loss": 0.8811, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 9.67717170715332, | |
| "learning_rate": 1.1099999999999999e-05, | |
| "loss": 0.8374, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.0325, | |
| "grad_norm": 7.850519180297852, | |
| "learning_rate": 1.20375e-05, | |
| "loss": 0.7848, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 7.787046909332275, | |
| "learning_rate": 1.2974999999999999e-05, | |
| "loss": 0.7056, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.0375, | |
| "grad_norm": 8.000933647155762, | |
| "learning_rate": 1.39125e-05, | |
| "loss": 0.6118, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.588298320770264, | |
| "learning_rate": 1.485e-05, | |
| "loss": 0.5839, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.0425, | |
| "grad_norm": 7.084440231323242, | |
| "learning_rate": 1.5787499999999997e-05, | |
| "loss": 0.5352, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 6.439165115356445, | |
| "learning_rate": 1.6725e-05, | |
| "loss": 0.5099, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.0475, | |
| "grad_norm": 6.848193168640137, | |
| "learning_rate": 1.76625e-05, | |
| "loss": 0.4845, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.957454204559326, | |
| "learning_rate": 1.8599999999999998e-05, | |
| "loss": 0.4591, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.0525, | |
| "grad_norm": 7.173631191253662, | |
| "learning_rate": 1.95375e-05, | |
| "loss": 0.4265, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 5.627034664154053, | |
| "learning_rate": 2.0475e-05, | |
| "loss": 0.4271, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.0575, | |
| "grad_norm": 5.964080333709717, | |
| "learning_rate": 2.1412499999999995e-05, | |
| "loss": 0.4051, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 8.408676147460938, | |
| "learning_rate": 2.2349999999999998e-05, | |
| "loss": 0.3964, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.0625, | |
| "grad_norm": 6.321172714233398, | |
| "learning_rate": 2.3287499999999997e-05, | |
| "loss": 0.3565, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 5.115601062774658, | |
| "learning_rate": 2.4225e-05, | |
| "loss": 0.3746, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.0675, | |
| "grad_norm": 5.302433967590332, | |
| "learning_rate": 2.51625e-05, | |
| "loss": 0.3636, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.260375499725342, | |
| "learning_rate": 2.6099999999999997e-05, | |
| "loss": 0.3597, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.0725, | |
| "grad_norm": 8.463918685913086, | |
| "learning_rate": 2.7037499999999997e-05, | |
| "loss": 0.399, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 6.9618659019470215, | |
| "learning_rate": 2.7975e-05, | |
| "loss": 0.5304, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.0775, | |
| "grad_norm": 6.6647162437438965, | |
| "learning_rate": 2.8912499999999998e-05, | |
| "loss": 0.5135, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 7.29068660736084, | |
| "learning_rate": 2.985e-05, | |
| "loss": 0.5595, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.0825, | |
| "grad_norm": 5.9147186279296875, | |
| "learning_rate": 3.0787499999999996e-05, | |
| "loss": 0.3989, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 4.840363502502441, | |
| "learning_rate": 3.1725e-05, | |
| "loss": 0.3295, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.0875, | |
| "grad_norm": 4.7903571128845215, | |
| "learning_rate": 3.2662499999999994e-05, | |
| "loss": 0.3158, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 6.355573654174805, | |
| "learning_rate": 3.36e-05, | |
| "loss": 0.4046, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.0925, | |
| "grad_norm": 7.575016975402832, | |
| "learning_rate": 3.45375e-05, | |
| "loss": 0.4779, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 6.783919334411621, | |
| "learning_rate": 3.5474999999999995e-05, | |
| "loss": 0.4578, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.0975, | |
| "grad_norm": 7.364860534667969, | |
| "learning_rate": 3.64125e-05, | |
| "loss": 0.6591, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 7.520545959472656, | |
| "learning_rate": 3.735e-05, | |
| "loss": 0.586, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.6248713731765747, | |
| "eval_runtime": 39.9662, | |
| "eval_samples_per_second": 42.311, | |
| "eval_steps_per_second": 2.652, | |
| "eval_wer": 34.16385662545272, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1025, | |
| "grad_norm": 6.746083736419678, | |
| "learning_rate": 3.74125e-05, | |
| "loss": 0.5033, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 5.35268497467041, | |
| "learning_rate": 3.730833333333333e-05, | |
| "loss": 0.3954, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.1075, | |
| "grad_norm": 4.840243816375732, | |
| "learning_rate": 3.7204166666666665e-05, | |
| "loss": 0.3024, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 4.398126602172852, | |
| "learning_rate": 3.7099999999999994e-05, | |
| "loss": 0.2543, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1125, | |
| "grad_norm": 4.504413604736328, | |
| "learning_rate": 3.699583333333333e-05, | |
| "loss": 0.2439, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 4.229017734527588, | |
| "learning_rate": 3.6891666666666664e-05, | |
| "loss": 0.2602, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.1175, | |
| "grad_norm": 4.006415367126465, | |
| "learning_rate": 3.678749999999999e-05, | |
| "loss": 0.2588, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 4.367706298828125, | |
| "learning_rate": 3.668333333333333e-05, | |
| "loss": 0.2419, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.1225, | |
| "grad_norm": 7.554067611694336, | |
| "learning_rate": 3.6579166666666664e-05, | |
| "loss": 0.3699, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 6.349033355712891, | |
| "learning_rate": 3.6475e-05, | |
| "loss": 0.3811, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.1275, | |
| "grad_norm": 6.685649394989014, | |
| "learning_rate": 3.6370833333333334e-05, | |
| "loss": 0.4109, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 8.145482063293457, | |
| "learning_rate": 3.626666666666666e-05, | |
| "loss": 0.4295, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.1325, | |
| "grad_norm": 6.097599983215332, | |
| "learning_rate": 3.61625e-05, | |
| "loss": 0.3691, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 6.148138999938965, | |
| "learning_rate": 3.6058333333333333e-05, | |
| "loss": 0.3754, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.1375, | |
| "grad_norm": 6.437056541442871, | |
| "learning_rate": 3.595416666666666e-05, | |
| "loss": 0.3569, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 6.0700201988220215, | |
| "learning_rate": 3.585e-05, | |
| "loss": 0.3443, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.1425, | |
| "grad_norm": 6.010077476501465, | |
| "learning_rate": 3.5745833333333326e-05, | |
| "loss": 0.3516, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 4.950198173522949, | |
| "learning_rate": 3.564166666666666e-05, | |
| "loss": 0.3309, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.1475, | |
| "grad_norm": 4.597211837768555, | |
| "learning_rate": 3.5537499999999996e-05, | |
| "loss": 0.2661, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 5.0821356773376465, | |
| "learning_rate": 3.543333333333333e-05, | |
| "loss": 0.2346, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.1525, | |
| "grad_norm": 3.8110575675964355, | |
| "learning_rate": 3.532916666666667e-05, | |
| "loss": 0.2383, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 5.058354377746582, | |
| "learning_rate": 3.5224999999999996e-05, | |
| "loss": 0.2311, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.1575, | |
| "grad_norm": 4.210774898529053, | |
| "learning_rate": 3.512083333333333e-05, | |
| "loss": 0.205, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 4.318747043609619, | |
| "learning_rate": 3.5016666666666666e-05, | |
| "loss": 0.215, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.1625, | |
| "grad_norm": 5.172638893127441, | |
| "learning_rate": 3.49125e-05, | |
| "loss": 0.3064, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 5.396555423736572, | |
| "learning_rate": 3.480833333333333e-05, | |
| "loss": 0.3197, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.1675, | |
| "grad_norm": 7.734130382537842, | |
| "learning_rate": 3.4704166666666665e-05, | |
| "loss": 0.3303, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 4.788872241973877, | |
| "learning_rate": 3.4599999999999994e-05, | |
| "loss": 0.3104, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.1725, | |
| "grad_norm": 4.634150505065918, | |
| "learning_rate": 3.449583333333333e-05, | |
| "loss": 0.26, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 4.166738510131836, | |
| "learning_rate": 3.4391666666666665e-05, | |
| "loss": 0.2411, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.1775, | |
| "grad_norm": 4.349499702453613, | |
| "learning_rate": 3.42875e-05, | |
| "loss": 0.2139, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 4.308148384094238, | |
| "learning_rate": 3.418333333333333e-05, | |
| "loss": 0.218, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1825, | |
| "grad_norm": 3.210516929626465, | |
| "learning_rate": 3.4079166666666664e-05, | |
| "loss": 0.2069, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 3.348567247390747, | |
| "learning_rate": 3.3975e-05, | |
| "loss": 0.193, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.1875, | |
| "grad_norm": 4.138334274291992, | |
| "learning_rate": 3.3870833333333334e-05, | |
| "loss": 0.1939, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 4.503148555755615, | |
| "learning_rate": 3.376666666666666e-05, | |
| "loss": 0.1921, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.1925, | |
| "grad_norm": 3.379345417022705, | |
| "learning_rate": 3.36625e-05, | |
| "loss": 0.1829, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 4.82003927230835, | |
| "learning_rate": 3.355833333333333e-05, | |
| "loss": 0.2564, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.1975, | |
| "grad_norm": 5.72125768661499, | |
| "learning_rate": 3.345416666666666e-05, | |
| "loss": 0.3326, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 5.574431896209717, | |
| "learning_rate": 3.335e-05, | |
| "loss": 0.3145, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.5048365592956543, | |
| "eval_runtime": 37.7406, | |
| "eval_samples_per_second": 44.806, | |
| "eval_steps_per_second": 2.809, | |
| "eval_wer": 25.259148245285374, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2025, | |
| "grad_norm": 4.822042942047119, | |
| "learning_rate": 3.324583333333333e-05, | |
| "loss": 0.2662, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 4.082330226898193, | |
| "learning_rate": 3.314166666666666e-05, | |
| "loss": 0.1935, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.2075, | |
| "grad_norm": 5.841582775115967, | |
| "learning_rate": 3.30375e-05, | |
| "loss": 0.186, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 3.377840042114258, | |
| "learning_rate": 3.293333333333333e-05, | |
| "loss": 0.1834, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.2125, | |
| "grad_norm": 3.8289568424224854, | |
| "learning_rate": 3.282916666666667e-05, | |
| "loss": 0.1594, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 3.3037023544311523, | |
| "learning_rate": 3.2724999999999996e-05, | |
| "loss": 0.1623, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.2175, | |
| "grad_norm": 4.1908860206604, | |
| "learning_rate": 3.262083333333333e-05, | |
| "loss": 0.1674, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 6.681819438934326, | |
| "learning_rate": 3.2516666666666666e-05, | |
| "loss": 0.2856, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.2225, | |
| "grad_norm": 5.509127616882324, | |
| "learning_rate": 3.2412499999999995e-05, | |
| "loss": 0.2918, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 4.549745559692383, | |
| "learning_rate": 3.230833333333333e-05, | |
| "loss": 0.3008, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2275, | |
| "grad_norm": 4.073692798614502, | |
| "learning_rate": 3.2204166666666666e-05, | |
| "loss": 0.2116, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 3.217360734939575, | |
| "learning_rate": 3.2099999999999994e-05, | |
| "loss": 0.1622, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.2325, | |
| "grad_norm": 4.068457126617432, | |
| "learning_rate": 3.199583333333333e-05, | |
| "loss": 0.1538, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 6.21290922164917, | |
| "learning_rate": 3.1891666666666665e-05, | |
| "loss": 0.1543, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2375, | |
| "grad_norm": 2.9277803897857666, | |
| "learning_rate": 3.17875e-05, | |
| "loss": 0.158, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 3.4852840900421143, | |
| "learning_rate": 3.168333333333333e-05, | |
| "loss": 0.1626, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2425, | |
| "grad_norm": 4.382246971130371, | |
| "learning_rate": 3.1579166666666664e-05, | |
| "loss": 0.1642, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 2.6627678871154785, | |
| "learning_rate": 3.1475e-05, | |
| "loss": 0.1526, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.2475, | |
| "grad_norm": 3.2728075981140137, | |
| "learning_rate": 3.1370833333333335e-05, | |
| "loss": 0.1624, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 3.8267433643341064, | |
| "learning_rate": 3.126666666666666e-05, | |
| "loss": 0.1857, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.2525, | |
| "grad_norm": 9.670909881591797, | |
| "learning_rate": 3.11625e-05, | |
| "loss": 0.2492, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 4.275064945220947, | |
| "learning_rate": 3.105833333333333e-05, | |
| "loss": 0.238, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.2575, | |
| "grad_norm": 4.751267910003662, | |
| "learning_rate": 3.095416666666666e-05, | |
| "loss": 0.2788, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 3.571152925491333, | |
| "learning_rate": 3.085e-05, | |
| "loss": 0.2056, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.2625, | |
| "grad_norm": 4.068671226501465, | |
| "learning_rate": 3.074583333333333e-05, | |
| "loss": 0.1741, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 3.274651527404785, | |
| "learning_rate": 3.064166666666666e-05, | |
| "loss": 0.1571, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.2675, | |
| "grad_norm": 3.0895748138427734, | |
| "learning_rate": 3.05375e-05, | |
| "loss": 0.149, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 2.9620044231414795, | |
| "learning_rate": 3.0433333333333332e-05, | |
| "loss": 0.1584, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.2725, | |
| "grad_norm": 3.2053639888763428, | |
| "learning_rate": 3.0329166666666664e-05, | |
| "loss": 0.1676, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 4.407464981079102, | |
| "learning_rate": 3.0225e-05, | |
| "loss": 0.1799, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.2775, | |
| "grad_norm": 5.430395603179932, | |
| "learning_rate": 3.0120833333333328e-05, | |
| "loss": 0.3165, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 4.443249702453613, | |
| "learning_rate": 3.0016666666666663e-05, | |
| "loss": 0.2934, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.2825, | |
| "grad_norm": 4.94956111907959, | |
| "learning_rate": 2.9912499999999995e-05, | |
| "loss": 0.2587, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 4.102962493896484, | |
| "learning_rate": 2.980833333333333e-05, | |
| "loss": 0.2637, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.2875, | |
| "grad_norm": 3.9422686100006104, | |
| "learning_rate": 2.9704166666666662e-05, | |
| "loss": 0.2561, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 4.891707897186279, | |
| "learning_rate": 2.9599999999999998e-05, | |
| "loss": 0.2476, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.2925, | |
| "grad_norm": 3.3687989711761475, | |
| "learning_rate": 2.949583333333333e-05, | |
| "loss": 0.2015, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 3.280766010284424, | |
| "learning_rate": 2.9391666666666665e-05, | |
| "loss": 0.1618, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.2975, | |
| "grad_norm": 3.6520583629608154, | |
| "learning_rate": 2.9287499999999997e-05, | |
| "loss": 0.1685, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 4.03117561340332, | |
| "learning_rate": 2.9183333333333332e-05, | |
| "loss": 0.225, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.4838997423648834, | |
| "eval_runtime": 38.3108, | |
| "eval_samples_per_second": 44.139, | |
| "eval_steps_per_second": 2.767, | |
| "eval_wer": 22.05570126139628, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3025, | |
| "grad_norm": 5.466115951538086, | |
| "learning_rate": 2.9079166666666664e-05, | |
| "loss": 0.2752, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 6.833763599395752, | |
| "learning_rate": 2.8974999999999996e-05, | |
| "loss": 0.3203, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3075, | |
| "grad_norm": 3.3256447315216064, | |
| "learning_rate": 2.8870833333333328e-05, | |
| "loss": 0.2206, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 3.841505289077759, | |
| "learning_rate": 2.8766666666666663e-05, | |
| "loss": 0.1708, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 3.2223970890045166, | |
| "learning_rate": 2.8662499999999995e-05, | |
| "loss": 0.1507, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 4.438643932342529, | |
| "learning_rate": 2.855833333333333e-05, | |
| "loss": 0.1827, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.3175, | |
| "grad_norm": 4.88558292388916, | |
| "learning_rate": 2.8454166666666663e-05, | |
| "loss": 0.2668, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 5.326033115386963, | |
| "learning_rate": 2.8349999999999998e-05, | |
| "loss": 0.247, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.3225, | |
| "grad_norm": 4.686404228210449, | |
| "learning_rate": 2.824583333333333e-05, | |
| "loss": 0.2666, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 5.8094401359558105, | |
| "learning_rate": 2.8141666666666665e-05, | |
| "loss": 0.2178, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.3275, | |
| "grad_norm": 5.154630661010742, | |
| "learning_rate": 2.80375e-05, | |
| "loss": 0.2241, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 3.6322267055511475, | |
| "learning_rate": 2.7933333333333332e-05, | |
| "loss": 0.2393, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.3325, | |
| "grad_norm": 3.901859998703003, | |
| "learning_rate": 2.7829166666666668e-05, | |
| "loss": 0.188, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 3.8515241146087646, | |
| "learning_rate": 2.7724999999999996e-05, | |
| "loss": 0.1522, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.3375, | |
| "grad_norm": 3.150157928466797, | |
| "learning_rate": 2.7620833333333328e-05, | |
| "loss": 0.1356, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 3.471564769744873, | |
| "learning_rate": 2.7516666666666664e-05, | |
| "loss": 0.1435, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.3425, | |
| "grad_norm": 3.938312292098999, | |
| "learning_rate": 2.7412499999999995e-05, | |
| "loss": 0.1445, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 2.5613808631896973, | |
| "learning_rate": 2.730833333333333e-05, | |
| "loss": 0.1363, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3475, | |
| "grad_norm": 3.3892834186553955, | |
| "learning_rate": 2.7204166666666663e-05, | |
| "loss": 0.1322, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 4.404175281524658, | |
| "learning_rate": 2.7099999999999998e-05, | |
| "loss": 0.1602, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.3525, | |
| "grad_norm": 3.336073398590088, | |
| "learning_rate": 2.6995833333333333e-05, | |
| "loss": 0.1488, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 5.919867992401123, | |
| "learning_rate": 2.6891666666666665e-05, | |
| "loss": 0.1476, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.3575, | |
| "grad_norm": 5.114974021911621, | |
| "learning_rate": 2.67875e-05, | |
| "loss": 0.2274, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 4.443377494812012, | |
| "learning_rate": 2.6683333333333333e-05, | |
| "loss": 0.2627, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.3625, | |
| "grad_norm": 4.095592021942139, | |
| "learning_rate": 2.657916666666666e-05, | |
| "loss": 0.2451, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 2.603665828704834, | |
| "learning_rate": 2.6474999999999996e-05, | |
| "loss": 0.1531, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3675, | |
| "grad_norm": 3.089383125305176, | |
| "learning_rate": 2.637083333333333e-05, | |
| "loss": 0.1334, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 3.058241605758667, | |
| "learning_rate": 2.6266666666666664e-05, | |
| "loss": 0.1322, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.3725, | |
| "grad_norm": 3.3943984508514404, | |
| "learning_rate": 2.61625e-05, | |
| "loss": 0.159, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 4.98250150680542, | |
| "learning_rate": 2.605833333333333e-05, | |
| "loss": 0.2044, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.3775, | |
| "grad_norm": 3.9143283367156982, | |
| "learning_rate": 2.5954166666666666e-05, | |
| "loss": 0.2495, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 5.307981014251709, | |
| "learning_rate": 2.5849999999999998e-05, | |
| "loss": 0.2502, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.3825, | |
| "grad_norm": 4.974552631378174, | |
| "learning_rate": 2.5745833333333333e-05, | |
| "loss": 0.2502, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 5.2868547439575195, | |
| "learning_rate": 2.5641666666666665e-05, | |
| "loss": 0.2662, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.3875, | |
| "grad_norm": 4.530261039733887, | |
| "learning_rate": 2.55375e-05, | |
| "loss": 0.2359, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 3.932204484939575, | |
| "learning_rate": 2.543333333333333e-05, | |
| "loss": 0.1757, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.3925, | |
| "grad_norm": 3.0662739276885986, | |
| "learning_rate": 2.5329166666666665e-05, | |
| "loss": 0.148, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 3.4663736820220947, | |
| "learning_rate": 2.5224999999999997e-05, | |
| "loss": 0.1411, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.3975, | |
| "grad_norm": 6.448814868927002, | |
| "learning_rate": 2.5120833333333332e-05, | |
| "loss": 0.2409, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 4.826459884643555, | |
| "learning_rate": 2.5016666666666664e-05, | |
| "loss": 0.3003, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.45400944352149963, | |
| "eval_runtime": 38.2487, | |
| "eval_samples_per_second": 44.211, | |
| "eval_steps_per_second": 2.771, | |
| "eval_wer": 20.307231172723867, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4025, | |
| "grad_norm": 3.8398826122283936, | |
| "learning_rate": 2.49125e-05, | |
| "loss": 0.2244, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 5.592238426208496, | |
| "learning_rate": 2.480833333333333e-05, | |
| "loss": 0.3029, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.4075, | |
| "grad_norm": 4.635577201843262, | |
| "learning_rate": 2.4704166666666666e-05, | |
| "loss": 0.2194, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 3.2737910747528076, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 0.2102, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.4125, | |
| "grad_norm": 2.5521626472473145, | |
| "learning_rate": 2.4495833333333334e-05, | |
| "loss": 0.1517, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 3.2408368587493896, | |
| "learning_rate": 2.4391666666666666e-05, | |
| "loss": 0.1331, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.4175, | |
| "grad_norm": 3.243713855743408, | |
| "learning_rate": 2.4287499999999997e-05, | |
| "loss": 0.1385, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 2.6619813442230225, | |
| "learning_rate": 2.418333333333333e-05, | |
| "loss": 0.1307, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.4225, | |
| "grad_norm": 2.726177930831909, | |
| "learning_rate": 2.4079166666666665e-05, | |
| "loss": 0.1399, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 3.1518099308013916, | |
| "learning_rate": 2.3974999999999997e-05, | |
| "loss": 0.1344, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.4275, | |
| "grad_norm": 3.95133376121521, | |
| "learning_rate": 2.3870833333333332e-05, | |
| "loss": 0.1512, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 4.805422306060791, | |
| "learning_rate": 2.3766666666666664e-05, | |
| "loss": 0.2212, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.4325, | |
| "grad_norm": 4.867978572845459, | |
| "learning_rate": 2.36625e-05, | |
| "loss": 0.2005, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 3.448626756668091, | |
| "learning_rate": 2.355833333333333e-05, | |
| "loss": 0.2085, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.4375, | |
| "grad_norm": 2.675185441970825, | |
| "learning_rate": 2.3454166666666666e-05, | |
| "loss": 0.1263, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 3.467017412185669, | |
| "learning_rate": 2.335e-05, | |
| "loss": 0.1122, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.4425, | |
| "grad_norm": 3.4933815002441406, | |
| "learning_rate": 2.3245833333333334e-05, | |
| "loss": 0.1266, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 3.4990930557250977, | |
| "learning_rate": 2.3141666666666666e-05, | |
| "loss": 0.1466, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.4475, | |
| "grad_norm": 3.077288866043091, | |
| "learning_rate": 2.3037499999999998e-05, | |
| "loss": 0.1486, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 3.375373363494873, | |
| "learning_rate": 2.293333333333333e-05, | |
| "loss": 0.1391, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4525, | |
| "grad_norm": 4.026995658874512, | |
| "learning_rate": 2.2829166666666665e-05, | |
| "loss": 0.1807, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 4.551440238952637, | |
| "learning_rate": 2.2724999999999997e-05, | |
| "loss": 0.2373, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.4575, | |
| "grad_norm": 5.197234153747559, | |
| "learning_rate": 2.2620833333333332e-05, | |
| "loss": 0.2225, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 4.593672752380371, | |
| "learning_rate": 2.2516666666666664e-05, | |
| "loss": 0.2394, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.4625, | |
| "grad_norm": 4.616493225097656, | |
| "learning_rate": 2.24125e-05, | |
| "loss": 0.2431, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 5.233091831207275, | |
| "learning_rate": 2.230833333333333e-05, | |
| "loss": 0.2357, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.4675, | |
| "grad_norm": 4.357009410858154, | |
| "learning_rate": 2.2204166666666667e-05, | |
| "loss": 0.2041, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 3.481781482696533, | |
| "learning_rate": 2.21e-05, | |
| "loss": 0.1817, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.4725, | |
| "grad_norm": 4.179861545562744, | |
| "learning_rate": 2.1995833333333334e-05, | |
| "loss": 0.1953, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 3.9264073371887207, | |
| "learning_rate": 2.1891666666666662e-05, | |
| "loss": 0.1927, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.4775, | |
| "grad_norm": 3.4452853202819824, | |
| "learning_rate": 2.1787499999999998e-05, | |
| "loss": 0.1443, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 2.9945247173309326, | |
| "learning_rate": 2.168333333333333e-05, | |
| "loss": 0.1587, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.4825, | |
| "grad_norm": 3.666703701019287, | |
| "learning_rate": 2.1579166666666665e-05, | |
| "loss": 0.1314, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 3.7032406330108643, | |
| "learning_rate": 2.1474999999999997e-05, | |
| "loss": 0.1524, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.4875, | |
| "grad_norm": 3.9737789630889893, | |
| "learning_rate": 2.1370833333333332e-05, | |
| "loss": 0.1605, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 4.30121374130249, | |
| "learning_rate": 2.1266666666666664e-05, | |
| "loss": 0.1834, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.4925, | |
| "grad_norm": 3.051647424697876, | |
| "learning_rate": 2.11625e-05, | |
| "loss": 0.1651, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 3.1889472007751465, | |
| "learning_rate": 2.105833333333333e-05, | |
| "loss": 0.1392, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.4975, | |
| "grad_norm": 3.226409912109375, | |
| "learning_rate": 2.0954166666666667e-05, | |
| "loss": 0.1286, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 3.729316234588623, | |
| "learning_rate": 2.085e-05, | |
| "loss": 0.132, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_loss": 0.45741865038871765, | |
| "eval_runtime": 38.6756, | |
| "eval_samples_per_second": 43.723, | |
| "eval_steps_per_second": 2.741, | |
| "eval_wer": 19.014612214312475, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5025, | |
| "grad_norm": 3.4599833488464355, | |
| "learning_rate": 2.074583333333333e-05, | |
| "loss": 0.1868, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 3.7839012145996094, | |
| "learning_rate": 2.0641666666666662e-05, | |
| "loss": 0.1635, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.5075, | |
| "grad_norm": 4.264291286468506, | |
| "learning_rate": 2.0537499999999998e-05, | |
| "loss": 0.2522, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 4.526363849639893, | |
| "learning_rate": 2.043333333333333e-05, | |
| "loss": 0.2296, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.5125, | |
| "grad_norm": 4.191585063934326, | |
| "learning_rate": 2.0329166666666665e-05, | |
| "loss": 0.1969, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 5.047796249389648, | |
| "learning_rate": 2.0224999999999997e-05, | |
| "loss": 0.204, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.5175, | |
| "grad_norm": 2.665344476699829, | |
| "learning_rate": 2.0120833333333332e-05, | |
| "loss": 0.1704, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 2.922651529312134, | |
| "learning_rate": 2.0016666666666664e-05, | |
| "loss": 0.117, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.5225, | |
| "grad_norm": 2.9956483840942383, | |
| "learning_rate": 1.99125e-05, | |
| "loss": 0.123, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 2.745727777481079, | |
| "learning_rate": 1.980833333333333e-05, | |
| "loss": 0.1223, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.5275, | |
| "grad_norm": 2.3079681396484375, | |
| "learning_rate": 1.9704166666666667e-05, | |
| "loss": 0.1229, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 3.4931304454803467, | |
| "learning_rate": 1.9599999999999995e-05, | |
| "loss": 0.1287, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.5325, | |
| "grad_norm": 3.3155760765075684, | |
| "learning_rate": 1.949583333333333e-05, | |
| "loss": 0.1284, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 4.483437538146973, | |
| "learning_rate": 1.9391666666666663e-05, | |
| "loss": 0.1695, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.5375, | |
| "grad_norm": 4.182435512542725, | |
| "learning_rate": 1.9287499999999998e-05, | |
| "loss": 0.2029, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 3.4684391021728516, | |
| "learning_rate": 1.918333333333333e-05, | |
| "loss": 0.1907, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.5425, | |
| "grad_norm": 4.1064019203186035, | |
| "learning_rate": 1.9079166666666665e-05, | |
| "loss": 0.1847, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 3.6676344871520996, | |
| "learning_rate": 1.8974999999999997e-05, | |
| "loss": 0.1644, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.5475, | |
| "grad_norm": 3.6105685234069824, | |
| "learning_rate": 1.8870833333333332e-05, | |
| "loss": 0.1368, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 3.2299535274505615, | |
| "learning_rate": 1.8766666666666664e-05, | |
| "loss": 0.138, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.5525, | |
| "grad_norm": 5.327032566070557, | |
| "learning_rate": 1.8662499999999996e-05, | |
| "loss": 0.1786, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 4.6841583251953125, | |
| "learning_rate": 1.855833333333333e-05, | |
| "loss": 0.2034, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.5575, | |
| "grad_norm": 4.341950416564941, | |
| "learning_rate": 1.8454166666666663e-05, | |
| "loss": 0.1611, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 2.254387378692627, | |
| "learning_rate": 1.835e-05, | |
| "loss": 0.1087, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.5625, | |
| "grad_norm": 3.0610556602478027, | |
| "learning_rate": 1.8245833333333334e-05, | |
| "loss": 0.0877, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 2.6597042083740234, | |
| "learning_rate": 1.8141666666666663e-05, | |
| "loss": 0.0893, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.5675, | |
| "grad_norm": 2.638615369796753, | |
| "learning_rate": 1.8037499999999998e-05, | |
| "loss": 0.1155, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 3.690338373184204, | |
| "learning_rate": 1.793333333333333e-05, | |
| "loss": 0.1252, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.5725, | |
| "grad_norm": 2.847730875015259, | |
| "learning_rate": 1.7829166666666665e-05, | |
| "loss": 0.1228, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 2.7520906925201416, | |
| "learning_rate": 1.7725e-05, | |
| "loss": 0.1162, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.5775, | |
| "grad_norm": 2.643385648727417, | |
| "learning_rate": 1.7620833333333332e-05, | |
| "loss": 0.1168, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 2.70414400100708, | |
| "learning_rate": 1.7516666666666664e-05, | |
| "loss": 0.1092, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.5825, | |
| "grad_norm": 2.194267988204956, | |
| "learning_rate": 1.7412499999999996e-05, | |
| "loss": 0.0949, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 3.784864902496338, | |
| "learning_rate": 1.730833333333333e-05, | |
| "loss": 0.1041, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.5875, | |
| "grad_norm": 3.313042402267456, | |
| "learning_rate": 1.7204166666666667e-05, | |
| "loss": 0.1105, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 3.7411179542541504, | |
| "learning_rate": 1.71e-05, | |
| "loss": 0.1396, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.5925, | |
| "grad_norm": 5.693575859069824, | |
| "learning_rate": 1.6995833333333334e-05, | |
| "loss": 0.2024, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 3.940126419067383, | |
| "learning_rate": 1.6891666666666663e-05, | |
| "loss": 0.1936, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.5975, | |
| "grad_norm": 4.423816680908203, | |
| "learning_rate": 1.6787499999999998e-05, | |
| "loss": 0.1702, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 3.7899911403656006, | |
| "learning_rate": 1.6683333333333333e-05, | |
| "loss": 0.1588, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.4380021393299103, | |
| "eval_runtime": 37.918, | |
| "eval_samples_per_second": 44.596, | |
| "eval_steps_per_second": 2.796, | |
| "eval_wer": 17.821905832396652, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6025, | |
| "grad_norm": 3.4838547706604004, | |
| "learning_rate": 1.6579166666666665e-05, | |
| "loss": 0.1486, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 3.3172895908355713, | |
| "learning_rate": 1.6475e-05, | |
| "loss": 0.1349, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.6075, | |
| "grad_norm": 2.6951234340667725, | |
| "learning_rate": 1.6370833333333333e-05, | |
| "loss": 0.1201, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.4754366874694824, | |
| "learning_rate": 1.6266666666666665e-05, | |
| "loss": 0.1109, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.6125, | |
| "grad_norm": 3.2808432579040527, | |
| "learning_rate": 1.61625e-05, | |
| "loss": 0.115, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 3.401995897293091, | |
| "learning_rate": 1.6058333333333332e-05, | |
| "loss": 0.1293, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.6175, | |
| "grad_norm": 3.313483476638794, | |
| "learning_rate": 1.5954166666666667e-05, | |
| "loss": 0.1285, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 4.194374084472656, | |
| "learning_rate": 1.585e-05, | |
| "loss": 0.129, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.6225, | |
| "grad_norm": 3.408337354660034, | |
| "learning_rate": 1.574583333333333e-05, | |
| "loss": 0.1212, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 3.5416505336761475, | |
| "learning_rate": 1.5641666666666666e-05, | |
| "loss": 0.1146, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.6275, | |
| "grad_norm": 3.736720561981201, | |
| "learning_rate": 1.5537499999999998e-05, | |
| "loss": 0.133, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 4.025073051452637, | |
| "learning_rate": 1.5433333333333334e-05, | |
| "loss": 0.1445, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.6325, | |
| "grad_norm": 3.1371307373046875, | |
| "learning_rate": 1.5329166666666665e-05, | |
| "loss": 0.1943, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 3.793010711669922, | |
| "learning_rate": 1.5224999999999999e-05, | |
| "loss": 0.176, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.6375, | |
| "grad_norm": 3.112964630126953, | |
| "learning_rate": 1.5120833333333331e-05, | |
| "loss": 0.1647, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 2.877613067626953, | |
| "learning_rate": 1.5016666666666665e-05, | |
| "loss": 0.1252, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.6425, | |
| "grad_norm": 2.556334972381592, | |
| "learning_rate": 1.4912499999999998e-05, | |
| "loss": 0.1137, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 5.248802661895752, | |
| "learning_rate": 1.4808333333333332e-05, | |
| "loss": 0.1087, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.6475, | |
| "grad_norm": 3.568530321121216, | |
| "learning_rate": 1.4704166666666666e-05, | |
| "loss": 0.1548, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 3.2436602115631104, | |
| "learning_rate": 1.4599999999999997e-05, | |
| "loss": 0.1494, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.6525, | |
| "grad_norm": 5.127964019775391, | |
| "learning_rate": 1.4495833333333331e-05, | |
| "loss": 0.1935, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 3.848689079284668, | |
| "learning_rate": 1.4391666666666665e-05, | |
| "loss": 0.1781, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.6575, | |
| "grad_norm": 3.3769173622131348, | |
| "learning_rate": 1.4287499999999998e-05, | |
| "loss": 0.178, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 3.279507875442505, | |
| "learning_rate": 1.4183333333333332e-05, | |
| "loss": 0.1559, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.6625, | |
| "grad_norm": 4.212521553039551, | |
| "learning_rate": 1.4079166666666666e-05, | |
| "loss": 0.1667, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.665, | |
| "grad_norm": 3.3866727352142334, | |
| "learning_rate": 1.3974999999999997e-05, | |
| "loss": 0.1966, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.6675, | |
| "grad_norm": 3.52656888961792, | |
| "learning_rate": 1.3870833333333331e-05, | |
| "loss": 0.1675, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 2.9528002738952637, | |
| "learning_rate": 1.3766666666666665e-05, | |
| "loss": 0.1612, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.6725, | |
| "grad_norm": 3.4069855213165283, | |
| "learning_rate": 1.3662499999999998e-05, | |
| "loss": 0.1162, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "grad_norm": 3.048247814178467, | |
| "learning_rate": 1.3558333333333334e-05, | |
| "loss": 0.105, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.6775, | |
| "grad_norm": 3.930851459503174, | |
| "learning_rate": 1.3454166666666664e-05, | |
| "loss": 0.1272, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 4.210860729217529, | |
| "learning_rate": 1.3349999999999998e-05, | |
| "loss": 0.2174, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.6825, | |
| "grad_norm": 6.0455498695373535, | |
| "learning_rate": 1.3245833333333331e-05, | |
| "loss": 0.194, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.685, | |
| "grad_norm": 5.270689487457275, | |
| "learning_rate": 1.3141666666666665e-05, | |
| "loss": 0.2003, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.6875, | |
| "grad_norm": 4.297138690948486, | |
| "learning_rate": 1.30375e-05, | |
| "loss": 0.1923, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 4.304157257080078, | |
| "learning_rate": 1.2933333333333334e-05, | |
| "loss": 0.1884, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.6925, | |
| "grad_norm": 3.3254096508026123, | |
| "learning_rate": 1.2829166666666664e-05, | |
| "loss": 0.1713, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.695, | |
| "grad_norm": 4.185650825500488, | |
| "learning_rate": 1.2724999999999998e-05, | |
| "loss": 0.1756, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.6975, | |
| "grad_norm": 4.220348358154297, | |
| "learning_rate": 1.2620833333333333e-05, | |
| "loss": 0.1958, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 4.203312873840332, | |
| "learning_rate": 1.2516666666666667e-05, | |
| "loss": 0.1841, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 0.4394695460796356, | |
| "eval_runtime": 37.5658, | |
| "eval_samples_per_second": 45.014, | |
| "eval_steps_per_second": 2.822, | |
| "eval_wer": 16.666666666666664, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7025, | |
| "grad_norm": 6.67822265625, | |
| "learning_rate": 1.2416666666666666e-05, | |
| "loss": 0.3408, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.705, | |
| "grad_norm": 8.843494415283203, | |
| "learning_rate": 1.2312499999999998e-05, | |
| "loss": 0.6013, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.7075, | |
| "grad_norm": 5.624876499176025, | |
| "learning_rate": 1.2208333333333331e-05, | |
| "loss": 0.5257, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 4.507925987243652, | |
| "learning_rate": 1.2104166666666665e-05, | |
| "loss": 0.4356, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.7125, | |
| "grad_norm": 4.30970573425293, | |
| "learning_rate": 1.1999999999999999e-05, | |
| "loss": 0.2381, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.715, | |
| "grad_norm": 5.139924049377441, | |
| "learning_rate": 1.1895833333333332e-05, | |
| "loss": 0.185, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.7175, | |
| "grad_norm": 2.9631428718566895, | |
| "learning_rate": 1.1791666666666666e-05, | |
| "loss": 0.17, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 3.6727359294891357, | |
| "learning_rate": 1.1687499999999998e-05, | |
| "loss": 0.1323, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.7225, | |
| "grad_norm": 2.7629499435424805, | |
| "learning_rate": 1.1583333333333331e-05, | |
| "loss": 0.128, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "grad_norm": 3.040910482406616, | |
| "learning_rate": 1.1479166666666665e-05, | |
| "loss": 0.1371, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.7275, | |
| "grad_norm": 2.7423293590545654, | |
| "learning_rate": 1.1374999999999999e-05, | |
| "loss": 0.1128, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 2.2156248092651367, | |
| "learning_rate": 1.1270833333333332e-05, | |
| "loss": 0.0964, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.7325, | |
| "grad_norm": 2.783275604248047, | |
| "learning_rate": 1.1166666666666664e-05, | |
| "loss": 0.101, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.735, | |
| "grad_norm": 4.380770206451416, | |
| "learning_rate": 1.1062499999999998e-05, | |
| "loss": 0.132, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.7375, | |
| "grad_norm": 3.904127597808838, | |
| "learning_rate": 1.0958333333333331e-05, | |
| "loss": 0.1807, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 5.420929431915283, | |
| "learning_rate": 1.0854166666666665e-05, | |
| "loss": 0.1436, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.7425, | |
| "grad_norm": 2.929569721221924, | |
| "learning_rate": 1.075e-05, | |
| "loss": 0.1648, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.745, | |
| "grad_norm": 3.0535011291503906, | |
| "learning_rate": 1.0645833333333334e-05, | |
| "loss": 0.1166, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.7475, | |
| "grad_norm": 2.630368232727051, | |
| "learning_rate": 1.0541666666666664e-05, | |
| "loss": 0.1081, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 4.1855621337890625, | |
| "learning_rate": 1.0437499999999998e-05, | |
| "loss": 0.1033, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.7525, | |
| "grad_norm": 4.28782844543457, | |
| "learning_rate": 1.0333333333333332e-05, | |
| "loss": 0.2037, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.755, | |
| "grad_norm": 5.351161956787109, | |
| "learning_rate": 1.0229166666666667e-05, | |
| "loss": 0.2651, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.7575, | |
| "grad_norm": 5.073176383972168, | |
| "learning_rate": 1.0125e-05, | |
| "loss": 0.2933, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 2.562958002090454, | |
| "learning_rate": 1.0020833333333334e-05, | |
| "loss": 0.1829, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.7625, | |
| "grad_norm": 4.164181709289551, | |
| "learning_rate": 9.916666666666664e-06, | |
| "loss": 0.1309, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.765, | |
| "grad_norm": 2.616945743560791, | |
| "learning_rate": 9.8125e-06, | |
| "loss": 0.1077, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.7675, | |
| "grad_norm": 4.083223342895508, | |
| "learning_rate": 9.708333333333333e-06, | |
| "loss": 0.1442, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 4.046257972717285, | |
| "learning_rate": 9.604166666666667e-06, | |
| "loss": 0.1725, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.7725, | |
| "grad_norm": 4.065029621124268, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.1833, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "grad_norm": 3.321071147918701, | |
| "learning_rate": 9.39583333333333e-06, | |
| "loss": 0.1671, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.7775, | |
| "grad_norm": 3.388200044631958, | |
| "learning_rate": 9.291666666666666e-06, | |
| "loss": 0.1196, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 4.393247604370117, | |
| "learning_rate": 9.1875e-06, | |
| "loss": 0.1399, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.7825, | |
| "grad_norm": 3.463327646255493, | |
| "learning_rate": 9.083333333333333e-06, | |
| "loss": 0.3125, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.785, | |
| "grad_norm": 2.6306939125061035, | |
| "learning_rate": 8.979166666666665e-06, | |
| "loss": 0.153, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.7875, | |
| "grad_norm": 2.3278794288635254, | |
| "learning_rate": 8.874999999999999e-06, | |
| "loss": 0.1141, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 2.850830554962158, | |
| "learning_rate": 8.770833333333333e-06, | |
| "loss": 0.1096, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.7925, | |
| "grad_norm": 2.4107863903045654, | |
| "learning_rate": 8.666666666666666e-06, | |
| "loss": 0.1117, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.795, | |
| "grad_norm": 3.420828104019165, | |
| "learning_rate": 8.5625e-06, | |
| "loss": 0.1071, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.7975, | |
| "grad_norm": 3.005918025970459, | |
| "learning_rate": 8.458333333333333e-06, | |
| "loss": 0.1146, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 3.3296029567718506, | |
| "learning_rate": 8.354166666666665e-06, | |
| "loss": 0.143, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.3718983232975006, | |
| "eval_runtime": 38.8288, | |
| "eval_samples_per_second": 43.55, | |
| "eval_steps_per_second": 2.73, | |
| "eval_wer": 15.448982140626951, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8025, | |
| "grad_norm": 4.029442310333252, | |
| "learning_rate": 8.249999999999999e-06, | |
| "loss": 0.1663, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.805, | |
| "grad_norm": 3.7387983798980713, | |
| "learning_rate": 8.145833333333333e-06, | |
| "loss": 0.1808, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.8075, | |
| "grad_norm": 3.0396807193756104, | |
| "learning_rate": 8.041666666666666e-06, | |
| "loss": 0.1351, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 2.9825220108032227, | |
| "learning_rate": 7.9375e-06, | |
| "loss": 0.1215, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.8125, | |
| "grad_norm": 2.495398998260498, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.1054, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.815, | |
| "grad_norm": 3.750054121017456, | |
| "learning_rate": 7.729166666666665e-06, | |
| "loss": 0.1049, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.8175, | |
| "grad_norm": 2.1268110275268555, | |
| "learning_rate": 7.625e-06, | |
| "loss": 0.0932, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 3.2984964847564697, | |
| "learning_rate": 7.520833333333332e-06, | |
| "loss": 0.0759, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.8225, | |
| "grad_norm": 2.470698356628418, | |
| "learning_rate": 7.416666666666666e-06, | |
| "loss": 0.0983, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "grad_norm": 3.2774899005889893, | |
| "learning_rate": 7.3125e-06, | |
| "loss": 0.085, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.8275, | |
| "grad_norm": 2.46321177482605, | |
| "learning_rate": 7.208333333333333e-06, | |
| "loss": 0.0916, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 2.0468456745147705, | |
| "learning_rate": 7.104166666666666e-06, | |
| "loss": 0.0951, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.8325, | |
| "grad_norm": 3.481823444366455, | |
| "learning_rate": 7e-06, | |
| "loss": 0.1368, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.835, | |
| "grad_norm": 4.122819423675537, | |
| "learning_rate": 6.895833333333333e-06, | |
| "loss": 0.1719, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.8375, | |
| "grad_norm": 4.750814914703369, | |
| "learning_rate": 6.791666666666666e-06, | |
| "loss": 0.2073, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.771611452102661, | |
| "learning_rate": 6.6875e-06, | |
| "loss": 0.1592, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.8425, | |
| "grad_norm": 4.536616325378418, | |
| "learning_rate": 6.583333333333333e-06, | |
| "loss": 0.1153, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.845, | |
| "grad_norm": 2.6289072036743164, | |
| "learning_rate": 6.479166666666666e-06, | |
| "loss": 0.1098, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.8475, | |
| "grad_norm": 3.1746957302093506, | |
| "learning_rate": 6.375e-06, | |
| "loss": 0.1216, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 3.029151439666748, | |
| "learning_rate": 6.270833333333333e-06, | |
| "loss": 0.1374, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.8525, | |
| "grad_norm": 3.8162496089935303, | |
| "learning_rate": 6.166666666666666e-06, | |
| "loss": 0.1408, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.855, | |
| "grad_norm": 5.568809509277344, | |
| "learning_rate": 6.062499999999999e-06, | |
| "loss": 0.1948, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.8575, | |
| "grad_norm": 5.649611473083496, | |
| "learning_rate": 5.958333333333333e-06, | |
| "loss": 0.3217, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 7.035798072814941, | |
| "learning_rate": 5.854166666666666e-06, | |
| "loss": 0.4378, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.8625, | |
| "grad_norm": 7.135374546051025, | |
| "learning_rate": 5.749999999999999e-06, | |
| "loss": 0.5134, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.865, | |
| "grad_norm": 3.085472345352173, | |
| "learning_rate": 5.645833333333333e-06, | |
| "loss": 0.3075, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.8675, | |
| "grad_norm": 4.60648250579834, | |
| "learning_rate": 5.541666666666666e-06, | |
| "loss": 0.2117, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 5.39269495010376, | |
| "learning_rate": 5.437499999999999e-06, | |
| "loss": 0.2447, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.8725, | |
| "grad_norm": 3.4270975589752197, | |
| "learning_rate": 5.333333333333333e-06, | |
| "loss": 0.1657, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 2.6738994121551514, | |
| "learning_rate": 5.2291666666666664e-06, | |
| "loss": 0.1087, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.8775, | |
| "grad_norm": 2.7085695266723633, | |
| "learning_rate": 5.124999999999999e-06, | |
| "loss": 0.1076, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 3.1228623390197754, | |
| "learning_rate": 5.020833333333333e-06, | |
| "loss": 0.1216, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.8825, | |
| "grad_norm": 5.188724994659424, | |
| "learning_rate": 4.9166666666666665e-06, | |
| "loss": 0.154, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.885, | |
| "grad_norm": 3.700319528579712, | |
| "learning_rate": 4.812499999999999e-06, | |
| "loss": 0.1491, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.8875, | |
| "grad_norm": 3.185534954071045, | |
| "learning_rate": 4.708333333333333e-06, | |
| "loss": 0.1517, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 3.0981645584106445, | |
| "learning_rate": 4.6041666666666665e-06, | |
| "loss": 0.1092, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.8925, | |
| "grad_norm": 3.850358247756958, | |
| "learning_rate": 4.499999999999999e-06, | |
| "loss": 0.107, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.895, | |
| "grad_norm": 3.169626235961914, | |
| "learning_rate": 4.395833333333333e-06, | |
| "loss": 0.115, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.8975, | |
| "grad_norm": 2.9183895587921143, | |
| "learning_rate": 4.2916666666666665e-06, | |
| "loss": 0.0914, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 2.877211570739746, | |
| "learning_rate": 4.187499999999999e-06, | |
| "loss": 0.0967, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 0.3684903383255005, | |
| "eval_runtime": 39.0288, | |
| "eval_samples_per_second": 43.327, | |
| "eval_steps_per_second": 2.716, | |
| "eval_wer": 15.136755339078306, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9025, | |
| "grad_norm": 3.196434497833252, | |
| "learning_rate": 4.083333333333333e-06, | |
| "loss": 0.1113, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.905, | |
| "grad_norm": 3.862992525100708, | |
| "learning_rate": 3.9791666666666665e-06, | |
| "loss": 0.1354, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.9075, | |
| "grad_norm": 4.684475898742676, | |
| "learning_rate": 3.874999999999999e-06, | |
| "loss": 0.1739, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 3.8043711185455322, | |
| "learning_rate": 3.770833333333333e-06, | |
| "loss": 0.1725, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.9125, | |
| "grad_norm": 3.2041592597961426, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.1385, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.915, | |
| "grad_norm": 2.448774576187134, | |
| "learning_rate": 3.5624999999999998e-06, | |
| "loss": 0.1294, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.9175, | |
| "grad_norm": 3.0288240909576416, | |
| "learning_rate": 3.458333333333333e-06, | |
| "loss": 0.1252, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 3.7271015644073486, | |
| "learning_rate": 3.354166666666666e-06, | |
| "loss": 0.1297, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.9225, | |
| "grad_norm": 3.9993860721588135, | |
| "learning_rate": 3.25e-06, | |
| "loss": 0.1503, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "grad_norm": 4.390246391296387, | |
| "learning_rate": 3.145833333333333e-06, | |
| "loss": 0.1744, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.9275, | |
| "grad_norm": 3.318673849105835, | |
| "learning_rate": 3.041666666666666e-06, | |
| "loss": 0.1926, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 2.6013784408569336, | |
| "learning_rate": 2.9375e-06, | |
| "loss": 0.1183, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.9325, | |
| "grad_norm": 2.507906675338745, | |
| "learning_rate": 2.833333333333333e-06, | |
| "loss": 0.1198, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.935, | |
| "grad_norm": 2.4156315326690674, | |
| "learning_rate": 2.7291666666666662e-06, | |
| "loss": 0.0964, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 3.535658359527588, | |
| "learning_rate": 2.625e-06, | |
| "loss": 0.125, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 3.782447099685669, | |
| "learning_rate": 2.520833333333333e-06, | |
| "loss": 0.15, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.9425, | |
| "grad_norm": 3.4775750637054443, | |
| "learning_rate": 2.4166666666666663e-06, | |
| "loss": 0.169, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.945, | |
| "grad_norm": 2.6878416538238525, | |
| "learning_rate": 2.3125e-06, | |
| "loss": 0.2931, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.9475, | |
| "grad_norm": 3.1754825115203857, | |
| "learning_rate": 2.208333333333333e-06, | |
| "loss": 0.1597, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 4.0340256690979, | |
| "learning_rate": 2.1041666666666667e-06, | |
| "loss": 0.1347, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.9525, | |
| "grad_norm": 4.122541904449463, | |
| "learning_rate": 2e-06, | |
| "loss": 0.1548, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.955, | |
| "grad_norm": 3.8141324520111084, | |
| "learning_rate": 1.8958333333333331e-06, | |
| "loss": 0.1713, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.9575, | |
| "grad_norm": 4.313726902008057, | |
| "learning_rate": 1.7916666666666665e-06, | |
| "loss": 0.1928, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 3.0914995670318604, | |
| "learning_rate": 1.6874999999999997e-06, | |
| "loss": 0.1501, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.9625, | |
| "grad_norm": 2.9334723949432373, | |
| "learning_rate": 1.5833333333333331e-06, | |
| "loss": 0.1221, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.965, | |
| "grad_norm": 2.4183266162872314, | |
| "learning_rate": 1.4791666666666663e-06, | |
| "loss": 0.0928, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.9675, | |
| "grad_norm": 2.296832323074341, | |
| "learning_rate": 1.375e-06, | |
| "loss": 0.0959, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 2.63551926612854, | |
| "learning_rate": 1.2708333333333334e-06, | |
| "loss": 0.0966, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.9725, | |
| "grad_norm": 1.874773621559143, | |
| "learning_rate": 1.1666666666666666e-06, | |
| "loss": 0.0958, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "grad_norm": 2.397674322128296, | |
| "learning_rate": 1.0624999999999998e-06, | |
| "loss": 0.1009, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.9775, | |
| "grad_norm": 3.000075101852417, | |
| "learning_rate": 9.583333333333334e-07, | |
| "loss": 0.1095, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.8197531700134277, | |
| "learning_rate": 8.541666666666666e-07, | |
| "loss": 0.1535, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.9825, | |
| "grad_norm": 3.8931257724761963, | |
| "learning_rate": 7.499999999999999e-07, | |
| "loss": 0.1639, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.985, | |
| "grad_norm": 3.463376045227051, | |
| "learning_rate": 6.458333333333332e-07, | |
| "loss": 0.1494, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.9875, | |
| "grad_norm": 2.9888062477111816, | |
| "learning_rate": 5.416666666666666e-07, | |
| "loss": 0.1483, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 7.088611602783203, | |
| "learning_rate": 4.375e-07, | |
| "loss": 0.2288, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.9925, | |
| "grad_norm": 2.751157283782959, | |
| "learning_rate": 3.333333333333333e-07, | |
| "loss": 0.1954, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.995, | |
| "grad_norm": 3.0588839054107666, | |
| "learning_rate": 2.2916666666666666e-07, | |
| "loss": 0.1068, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.9975, | |
| "grad_norm": 2.712277412414551, | |
| "learning_rate": 1.25e-07, | |
| "loss": 0.0997, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.309202194213867, | |
| "learning_rate": 2.083333333333333e-08, | |
| "loss": 0.1059, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.37193799018859863, | |
| "eval_runtime": 39.6288, | |
| "eval_samples_per_second": 42.671, | |
| "eval_steps_per_second": 2.675, | |
| "eval_wer": 14.849506681653555, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 10000, | |
| "total_flos": 7.8780432384e+18, | |
| "train_loss": 0.2454334835767746, | |
| "train_runtime": 3974.1996, | |
| "train_samples_per_second": 80.519, | |
| "train_steps_per_second": 2.516 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7.8780432384e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |