| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.727626459143968, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.048638132295719845, | |
| "grad_norm": 7.16464900970459, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.0297, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09727626459143969, | |
| "grad_norm": 4.811949729919434, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.786, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14591439688715954, | |
| "grad_norm": 4.57177734375, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.5046, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.19455252918287938, | |
| "grad_norm": 4.517965793609619, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.4671, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24319066147859922, | |
| "grad_norm": 4.065127849578857, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.4381, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2918287937743191, | |
| "grad_norm": 3.5869154930114746, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4062, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3404669260700389, | |
| "grad_norm": 4.262431621551514, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.3951, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.38910505836575876, | |
| "grad_norm": 4.2176713943481445, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.3862, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4377431906614786, | |
| "grad_norm": 3.8358681201934814, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.364, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.48638132295719844, | |
| "grad_norm": 3.8311147689819336, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3425, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5350194552529183, | |
| "grad_norm": 3.7137532234191895, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.339, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5836575875486382, | |
| "grad_norm": 4.2145280838012695, | |
| "learning_rate": 6e-06, | |
| "loss": 0.3393, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.632295719844358, | |
| "grad_norm": 3.540754556655884, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.3154, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.6809338521400778, | |
| "grad_norm": 3.7712759971618652, | |
| "learning_rate": 7e-06, | |
| "loss": 0.319, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7295719844357976, | |
| "grad_norm": 3.6147518157958984, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3127, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.7782101167315175, | |
| "grad_norm": 3.6948869228363037, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2965, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8268482490272373, | |
| "grad_norm": 3.350135087966919, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3001, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.8754863813229572, | |
| "grad_norm": 3.7692646980285645, | |
| "learning_rate": 9e-06, | |
| "loss": 0.281, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9241245136186771, | |
| "grad_norm": 3.2747561931610107, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2707, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.9727626459143969, | |
| "grad_norm": 3.260094165802002, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2708, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0214007782101167, | |
| "grad_norm": 2.917370080947876, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.2415, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.0700389105058365, | |
| "grad_norm": 2.564760208129883, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.1714, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.1186770428015564, | |
| "grad_norm": 2.844514846801758, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.1842, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.1673151750972763, | |
| "grad_norm": 3.1123480796813965, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.171, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.2159533073929962, | |
| "grad_norm": 3.253563642501831, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.1773, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.264591439688716, | |
| "grad_norm": 2.3145253658294678, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.1779, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.3132295719844358, | |
| "grad_norm": 2.493607521057129, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.1717, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.3618677042801557, | |
| "grad_norm": 2.081571578979492, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.1747, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.4105058365758754, | |
| "grad_norm": 2.9359965324401855, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.1591, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.4591439688715953, | |
| "grad_norm": 2.9549202919006348, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.1556, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.5077821011673151, | |
| "grad_norm": 3.0995261669158936, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.1553, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.556420233463035, | |
| "grad_norm": 2.3519606590270996, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.1502, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.605058365758755, | |
| "grad_norm": 2.3519246578216553, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.1398, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.6536964980544746, | |
| "grad_norm": 2.292020797729492, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.1518, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.7023346303501945, | |
| "grad_norm": 2.7264564037323, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.1393, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.7509727626459144, | |
| "grad_norm": 2.202409267425537, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.1432, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.7996108949416343, | |
| "grad_norm": 2.6967380046844482, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.1492, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.8482490272373542, | |
| "grad_norm": 2.8609344959259033, | |
| "learning_rate": 9e-06, | |
| "loss": 0.1351, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.8968871595330739, | |
| "grad_norm": 1.9660629034042358, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.1378, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.9455252918287937, | |
| "grad_norm": 2.9408161640167236, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.1429, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.9455252918287937, | |
| "eval_loss": 0.2754097580909729, | |
| "eval_runtime": 2154.7003, | |
| "eval_samples_per_second": 2.497, | |
| "eval_steps_per_second": 0.156, | |
| "eval_wer": 0.22082388961787744, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.9941634241245136, | |
| "grad_norm": 2.337907075881958, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.137, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.0428015564202333, | |
| "grad_norm": 1.9045900106430054, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.0728, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.0914396887159534, | |
| "grad_norm": 2.0690503120422363, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.0615, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.140077821011673, | |
| "grad_norm": 2.0722174644470215, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.0582, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.188715953307393, | |
| "grad_norm": 1.7977361679077148, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.0589, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.237354085603113, | |
| "grad_norm": 1.5375025272369385, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.0657, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.2859922178988326, | |
| "grad_norm": 1.9439505338668823, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.0648, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.3346303501945527, | |
| "grad_norm": 1.7909560203552246, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.0594, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.3832684824902723, | |
| "grad_norm": 1.7498406171798706, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.0627, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.4319066147859925, | |
| "grad_norm": 1.557906985282898, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0585, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.480544747081712, | |
| "grad_norm": 1.3372323513031006, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.0614, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.529182879377432, | |
| "grad_norm": 1.8402326107025146, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.0601, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.5778210116731515, | |
| "grad_norm": 2.003889560699463, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.0575, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.6264591439688716, | |
| "grad_norm": 1.5208065509796143, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.0563, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.6750972762645917, | |
| "grad_norm": 2.1009633541107178, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.0633, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.7237354085603114, | |
| "grad_norm": 1.367943286895752, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.0572, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.772373540856031, | |
| "grad_norm": 1.914768934249878, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.0541, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.8210116731517507, | |
| "grad_norm": 2.1714184284210205, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.059, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.869649805447471, | |
| "grad_norm": 1.951029658317566, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.0571, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.9182879377431905, | |
| "grad_norm": 1.6378835439682007, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0563, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.9669260700389106, | |
| "grad_norm": 1.8460659980773926, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.0556, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.0155642023346303, | |
| "grad_norm": 1.0332473516464233, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.045, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.06420233463035, | |
| "grad_norm": 1.5045939683914185, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.0227, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.11284046692607, | |
| "grad_norm": 2.1660637855529785, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.0275, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.1614785992217898, | |
| "grad_norm": 1.159175992012024, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.024, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.21011673151751, | |
| "grad_norm": 0.9240212440490723, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.0271, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.2587548638132295, | |
| "grad_norm": 0.9973633885383606, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.0266, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.307392996108949, | |
| "grad_norm": 1.5561943054199219, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.0267, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.3560311284046693, | |
| "grad_norm": 1.2981663942337036, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.0284, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.404669260700389, | |
| "grad_norm": 1.1085954904556274, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0239, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.453307392996109, | |
| "grad_norm": 1.219834566116333, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.0244, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.501945525291829, | |
| "grad_norm": 1.1478172540664673, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0273, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.5505836575875485, | |
| "grad_norm": 1.5546625852584839, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.0242, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.5992217898832686, | |
| "grad_norm": 1.5303540229797363, | |
| "learning_rate": 7e-06, | |
| "loss": 0.0248, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.6478599221789882, | |
| "grad_norm": 1.2220957279205322, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.0268, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.6964980544747084, | |
| "grad_norm": 1.076720952987671, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.0246, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.745136186770428, | |
| "grad_norm": 0.6398268342018127, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.0251, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.7937743190661477, | |
| "grad_norm": 1.0957626104354858, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.024, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.842412451361868, | |
| "grad_norm": 1.3786637783050537, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.0221, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.8910505836575875, | |
| "grad_norm": 1.0472965240478516, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.0232, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.8910505836575875, | |
| "eval_loss": 0.2916410565376282, | |
| "eval_runtime": 2132.1371, | |
| "eval_samples_per_second": 2.524, | |
| "eval_steps_per_second": 0.158, | |
| "eval_wer": 0.19905364030378941, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.9396887159533076, | |
| "grad_norm": 1.0177409648895264, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.0238, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.9883268482490273, | |
| "grad_norm": 1.0486767292022705, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.0245, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.036964980544747, | |
| "grad_norm": 0.7433611750602722, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.0138, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.085603112840467, | |
| "grad_norm": 0.4771675765514374, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.0109, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.134241245136187, | |
| "grad_norm": 0.6232401132583618, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.0109, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.182879377431907, | |
| "grad_norm": 0.6728478670120239, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.0103, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.2315175097276265, | |
| "grad_norm": 1.6818970441818237, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.0114, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.280155642023346, | |
| "grad_norm": 0.36247119307518005, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.011, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.328793774319066, | |
| "grad_norm": 1.3438676595687866, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.0126, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.377431906614786, | |
| "grad_norm": 0.8739299774169922, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0104, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.426070038910506, | |
| "grad_norm": 0.6817359328269958, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.0118, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.474708171206226, | |
| "grad_norm": 0.5845357775688171, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0125, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.523346303501945, | |
| "grad_norm": 1.067460298538208, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.0099, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.571984435797665, | |
| "grad_norm": 0.7725332379341125, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.0123, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.620622568093385, | |
| "grad_norm": 0.8168444633483887, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.01, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.669260700389105, | |
| "grad_norm": 0.7538411617279053, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.0098, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.717898832684825, | |
| "grad_norm": 0.3141120970249176, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.0107, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.766536964980545, | |
| "grad_norm": 0.887184202671051, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.0095, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.815175097276264, | |
| "grad_norm": 0.5846179127693176, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.0099, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.863813229571985, | |
| "grad_norm": 0.3938518464565277, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0091, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.912451361867705, | |
| "grad_norm": 0.9777388572692871, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.0093, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.961089494163424, | |
| "grad_norm": 0.7862864136695862, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.0111, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 5.009727626459144, | |
| "grad_norm": 0.7504919171333313, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.0094, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.058365758754864, | |
| "grad_norm": 0.2351156324148178, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.0049, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.107003891050583, | |
| "grad_norm": 0.13562794029712677, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.0043, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.155642023346304, | |
| "grad_norm": 0.33899495005607605, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.0048, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.2042801556420235, | |
| "grad_norm": 0.9408302307128906, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.0051, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.252918287937743, | |
| "grad_norm": 0.35573306679725647, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.0053, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.301556420233463, | |
| "grad_norm": 0.7676318883895874, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.0042, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.3501945525291825, | |
| "grad_norm": 0.6776612401008606, | |
| "learning_rate": 5e-06, | |
| "loss": 0.004, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.398832684824903, | |
| "grad_norm": 1.0338044166564941, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.0047, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.447470817120623, | |
| "grad_norm": 0.2744717001914978, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0068, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.496108949416342, | |
| "grad_norm": 0.7075155973434448, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.0061, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.544747081712062, | |
| "grad_norm": 0.1820390671491623, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.0052, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.593385214007782, | |
| "grad_norm": 0.33624935150146484, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.0038, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.642023346303502, | |
| "grad_norm": 0.1461501270532608, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.0046, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.690661478599222, | |
| "grad_norm": 0.33367106318473816, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.0042, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.739299610894942, | |
| "grad_norm": 0.32830071449279785, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.0043, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.787937743190661, | |
| "grad_norm": 1.2332854270935059, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.0039, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 5.836575875486381, | |
| "grad_norm": 0.6161640286445618, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0046, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.836575875486381, | |
| "eval_loss": 0.3218872845172882, | |
| "eval_runtime": 2114.549, | |
| "eval_samples_per_second": 2.545, | |
| "eval_steps_per_second": 0.159, | |
| "eval_wer": 0.18784047079406735, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.885214007782102, | |
| "grad_norm": 2.608947992324829, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.0061, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 5.933852140077821, | |
| "grad_norm": 0.25549063086509705, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.0044, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.982490272373541, | |
| "grad_norm": 1.4720717668533325, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.0051, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.031128404669261, | |
| "grad_norm": 0.7722542881965637, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.0037, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.07976653696498, | |
| "grad_norm": 0.35967299342155457, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.0029, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.1284046692607, | |
| "grad_norm": 0.7472084164619446, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.0029, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.1770428015564205, | |
| "grad_norm": 0.08338561654090881, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.0023, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.22568093385214, | |
| "grad_norm": 0.49297797679901123, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0022, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.27431906614786, | |
| "grad_norm": 0.08119112253189087, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.0019, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.3229571984435795, | |
| "grad_norm": 0.13106492161750793, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0027, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.3715953307393, | |
| "grad_norm": 0.18573686480522156, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.003, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.42023346303502, | |
| "grad_norm": 0.09228133410215378, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.004, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.468871595330739, | |
| "grad_norm": 0.8169698715209961, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.0029, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.517509727626459, | |
| "grad_norm": 0.23130229115486145, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.0026, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.566147859922179, | |
| "grad_norm": 0.10798731446266174, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0036, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.614785992217898, | |
| "grad_norm": 0.09145894646644592, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0024, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.663424124513619, | |
| "grad_norm": 0.3099443018436432, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0018, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 6.712062256809339, | |
| "grad_norm": 0.13762035965919495, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.0021, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.760700389105058, | |
| "grad_norm": 0.11253529787063599, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0012, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 6.809338521400778, | |
| "grad_norm": 0.054800793528556824, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.0019, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.857976653696498, | |
| "grad_norm": 0.06803842633962631, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0024, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 6.906614785992218, | |
| "grad_norm": 0.08114957064390182, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0019, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 6.955252918287938, | |
| "grad_norm": 0.06155654788017273, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0017, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 7.003891050583658, | |
| "grad_norm": 0.09375619143247604, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0015, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.052529182879377, | |
| "grad_norm": 0.04578198865056038, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0011, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.101167315175097, | |
| "grad_norm": 0.036920215934515, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0012, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.1498054474708175, | |
| "grad_norm": 0.07471055537462234, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0009, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.198443579766537, | |
| "grad_norm": 0.05370425805449486, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0009, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.247081712062257, | |
| "grad_norm": 0.04795600846409798, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.2957198443579765, | |
| "grad_norm": 0.030901705846190453, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0007, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.344357976653696, | |
| "grad_norm": 0.05389130115509033, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0012, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.392996108949417, | |
| "grad_norm": 0.05646834522485733, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0009, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.441634241245136, | |
| "grad_norm": 0.04085018113255501, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.490272373540856, | |
| "grad_norm": 0.024638062343001366, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0009, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.538910505836576, | |
| "grad_norm": 0.03571788966655731, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0011, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.587548638132295, | |
| "grad_norm": 0.05909154936671257, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0007, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.636186770428015, | |
| "grad_norm": 0.028353404253721237, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 7.684824902723736, | |
| "grad_norm": 0.16029401123523712, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0009, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.733463035019455, | |
| "grad_norm": 0.04438905417919159, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0008, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 7.782101167315175, | |
| "grad_norm": 0.035433579236269, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0009, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.782101167315175, | |
| "eval_loss": 0.3454054296016693, | |
| "eval_runtime": 2148.9325, | |
| "eval_samples_per_second": 2.504, | |
| "eval_steps_per_second": 0.157, | |
| "eval_wer": 0.1831881983379061, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.830739299610895, | |
| "grad_norm": 0.028996312990784645, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0008, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 7.879377431906615, | |
| "grad_norm": 0.038502488285303116, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0014, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 7.928015564202335, | |
| "grad_norm": 0.06861083209514618, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0009, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 7.976653696498055, | |
| "grad_norm": 0.07425787299871445, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0011, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 8.025291828793774, | |
| "grad_norm": 0.025021173059940338, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0006, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.073929961089494, | |
| "grad_norm": 0.031610701233148575, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.122568093385214, | |
| "grad_norm": 0.02210886962711811, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0008, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.171206225680933, | |
| "grad_norm": 0.03183664008975029, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0007, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.219844357976653, | |
| "grad_norm": 0.024653365835547447, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0005, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.268482490272374, | |
| "grad_norm": 0.02229795977473259, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.317120622568094, | |
| "grad_norm": 0.025678085163235664, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0005, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.365758754863814, | |
| "grad_norm": 0.025166384875774384, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0005, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.414396887159533, | |
| "grad_norm": 0.017260603606700897, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0005, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 8.463035019455253, | |
| "grad_norm": 0.016240952536463737, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0005, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 8.511673151750973, | |
| "grad_norm": 0.021515797823667526, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0006, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 8.560311284046692, | |
| "grad_norm": 0.023731861263513565, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0006, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.608949416342412, | |
| "grad_norm": 0.027586709707975388, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0006, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 8.657587548638132, | |
| "grad_norm": 0.04710078611969948, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0005, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 8.706225680933851, | |
| "grad_norm": 0.030895188450813293, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0005, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 8.754863813229573, | |
| "grad_norm": 0.04257024824619293, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0006, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.803501945525293, | |
| "grad_norm": 0.022679895162582397, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0006, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 8.852140077821012, | |
| "grad_norm": 0.028052836656570435, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0006, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 8.900778210116732, | |
| "grad_norm": 0.022373031824827194, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0005, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 8.949416342412452, | |
| "grad_norm": 0.02304161712527275, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0005, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.998054474708171, | |
| "grad_norm": 0.020944086834788322, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0005, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.04669260700389, | |
| "grad_norm": 0.018772481009364128, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0005, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.09533073929961, | |
| "grad_norm": 0.019685110077261925, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0005, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.14396887159533, | |
| "grad_norm": 0.017540233209729195, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0004, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.19260700389105, | |
| "grad_norm": 0.022134965285658836, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0004, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.24124513618677, | |
| "grad_norm": 0.01725279912352562, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0005, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.289883268482491, | |
| "grad_norm": 0.01975845918059349, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0004, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.33852140077821, | |
| "grad_norm": 0.021399999037384987, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0005, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 9.38715953307393, | |
| "grad_norm": 0.020681940019130707, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0004, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 9.43579766536965, | |
| "grad_norm": 0.020026126876473427, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0005, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 9.48443579766537, | |
| "grad_norm": 0.030826667323708534, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0004, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 9.53307392996109, | |
| "grad_norm": 0.017975399270653725, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0004, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.581712062256809, | |
| "grad_norm": 0.014251296408474445, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0005, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 9.630350194552529, | |
| "grad_norm": 0.018128497526049614, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0005, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.678988326848248, | |
| "grad_norm": 0.018901441246271133, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0005, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 9.727626459143968, | |
| "grad_norm": 0.01760680228471756, | |
| "learning_rate": 0.0, | |
| "loss": 0.0004, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.727626459143968, | |
| "eval_loss": 0.36376532912254333, | |
| "eval_runtime": 2157.1617, | |
| "eval_samples_per_second": 2.494, | |
| "eval_steps_per_second": 0.156, | |
| "eval_wer": 0.18173684838363355, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.727626459143968, | |
| "step": 5000, | |
| "total_flos": 5.4332453168676864e+20, | |
| "train_loss": 0.06844911024216563, | |
| "train_runtime": 58020.7767, | |
| "train_samples_per_second": 2.758, | |
| "train_steps_per_second": 0.086 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "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": 5.4332453168676864e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |