| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.29000429000429, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02145002145002145, | |
| "grad_norm": 9.066595077514648, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.715, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0429000429000429, | |
| "grad_norm": 6.495127201080322, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.3703, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06435006435006435, | |
| "grad_norm": 5.474078178405762, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.0103, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0858000858000858, | |
| "grad_norm": 5.429323196411133, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.9033, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10725010725010725, | |
| "grad_norm": 5.380603790283203, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.7779, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1287001287001287, | |
| "grad_norm": 4.108312606811523, | |
| "learning_rate": 3e-06, | |
| "loss": 0.7064, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.15015015015015015, | |
| "grad_norm": 4.261275291442871, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.6859, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1716001716001716, | |
| "grad_norm": 4.792464256286621, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.6422, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.19305019305019305, | |
| "grad_norm": 4.474643707275391, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.6719, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2145002145002145, | |
| "grad_norm": 5.082165718078613, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6536, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.23595023595023595, | |
| "grad_norm": 5.100755214691162, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.6508, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.2574002574002574, | |
| "grad_norm": 4.228156566619873, | |
| "learning_rate": 6e-06, | |
| "loss": 0.633, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.27885027885027885, | |
| "grad_norm": 4.423853874206543, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.5827, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.3003003003003003, | |
| "grad_norm": 4.693906784057617, | |
| "learning_rate": 7e-06, | |
| "loss": 0.5881, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.32175032175032175, | |
| "grad_norm": 5.522333145141602, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.5953, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.3432003432003432, | |
| "grad_norm": 3.7150588035583496, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.5733, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.36465036465036466, | |
| "grad_norm": 4.65098762512207, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.5741, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3861003861003861, | |
| "grad_norm": 4.152739524841309, | |
| "learning_rate": 9e-06, | |
| "loss": 0.5721, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.40755040755040756, | |
| "grad_norm": 4.5913615226745605, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.5837, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.429000429000429, | |
| "grad_norm": 3.7452423572540283, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5763, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.45045045045045046, | |
| "grad_norm": 4.0456953048706055, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.5743, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.4719004719004719, | |
| "grad_norm": 4.183647155761719, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.5851, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.49335049335049336, | |
| "grad_norm": 3.6905064582824707, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.5423, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.5148005148005148, | |
| "grad_norm": 4.111310005187988, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.5392, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5362505362505362, | |
| "grad_norm": 3.7859528064727783, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.5257, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.5577005577005577, | |
| "grad_norm": 3.810257911682129, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.4958, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5791505791505791, | |
| "grad_norm": 3.279156446456909, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.5103, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.6006006006006006, | |
| "grad_norm": 3.16200852394104, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.5262, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.622050622050622, | |
| "grad_norm": 3.9035494327545166, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.5131, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.6435006435006435, | |
| "grad_norm": 3.071242570877075, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.478, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.6649506649506649, | |
| "grad_norm": 4.14898157119751, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.5035, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.6864006864006864, | |
| "grad_norm": 3.8921945095062256, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.5, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.7078507078507078, | |
| "grad_norm": 3.4418222904205322, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.4676, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.7293007293007293, | |
| "grad_norm": 3.1406946182250977, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.4668, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7507507507507507, | |
| "grad_norm": 3.900789499282837, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.4834, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.7722007722007722, | |
| "grad_norm": 3.2185850143432617, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.4716, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7936507936507936, | |
| "grad_norm": 3.683173179626465, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.4882, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.8151008151008151, | |
| "grad_norm": 3.3974883556365967, | |
| "learning_rate": 9e-06, | |
| "loss": 0.4781, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.8365508365508365, | |
| "grad_norm": 3.78350830078125, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.4872, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.858000858000858, | |
| "grad_norm": 3.633676528930664, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.4429, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.858000858000858, | |
| "eval_loss": 0.46728599071502686, | |
| "eval_runtime": 1674.249, | |
| "eval_samples_per_second": 2.33, | |
| "eval_steps_per_second": 0.146, | |
| "eval_wer": 0.3495363662706462, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.8794508794508794, | |
| "grad_norm": 3.6749162673950195, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.4712, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.9009009009009009, | |
| "grad_norm": 3.0788767337799072, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.4857, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.9223509223509223, | |
| "grad_norm": 3.3249752521514893, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.4406, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.9438009438009438, | |
| "grad_norm": 3.189654588699341, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.4313, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.9652509652509652, | |
| "grad_norm": 3.3606882095336914, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.4517, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.9867009867009867, | |
| "grad_norm": 3.4091646671295166, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.4378, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.008151008151008, | |
| "grad_norm": 2.8111917972564697, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3944, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.0296010296010296, | |
| "grad_norm": 2.7863500118255615, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.3242, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.0510510510510511, | |
| "grad_norm": 2.681203603744507, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.3294, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.0725010725010724, | |
| "grad_norm": 3.21625018119812, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.3232, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.093951093951094, | |
| "grad_norm": 2.1063191890716553, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.3162, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.1154011154011154, | |
| "grad_norm": 2.7239444255828857, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.3372, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.136851136851137, | |
| "grad_norm": 2.6881861686706543, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.3269, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.1583011583011582, | |
| "grad_norm": 3.041693925857544, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.3299, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.1797511797511797, | |
| "grad_norm": 3.6576833724975586, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.3786, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.2012012012012012, | |
| "grad_norm": 3.035702705383301, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.346, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.2226512226512227, | |
| "grad_norm": 2.5944619178771973, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.3309, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.244101244101244, | |
| "grad_norm": 2.5746264457702637, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.3235, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.2655512655512655, | |
| "grad_norm": 2.7788243293762207, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.3212, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.287001287001287, | |
| "grad_norm": 3.6462416648864746, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.332, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.3084513084513085, | |
| "grad_norm": 2.2463419437408447, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.3366, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.3299013299013298, | |
| "grad_norm": 3.5368576049804688, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.3138, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.3513513513513513, | |
| "grad_norm": 2.547539472579956, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.3564, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.3728013728013728, | |
| "grad_norm": 3.0548551082611084, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.3393, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.3942513942513943, | |
| "grad_norm": 2.7717063426971436, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3266, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.4157014157014158, | |
| "grad_norm": 2.6384313106536865, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.3355, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.4371514371514371, | |
| "grad_norm": 2.7044169902801514, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.3375, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.4586014586014586, | |
| "grad_norm": 2.4753165245056152, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.3347, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.4800514800514801, | |
| "grad_norm": 3.0648443698883057, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.3202, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.5015015015015014, | |
| "grad_norm": 3.1359264850616455, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.3303, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.522951522951523, | |
| "grad_norm": 2.995232105255127, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.3197, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.5444015444015444, | |
| "grad_norm": 2.7742152214050293, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.3255, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.565851565851566, | |
| "grad_norm": 2.7029542922973633, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.3147, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.5873015873015874, | |
| "grad_norm": 2.604520082473755, | |
| "learning_rate": 7e-06, | |
| "loss": 0.3235, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.6087516087516087, | |
| "grad_norm": 3.152074098587036, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.319, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.6302016302016302, | |
| "grad_norm": 3.121864080429077, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.2972, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.6516516516516515, | |
| "grad_norm": 2.733600616455078, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.3177, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.673101673101673, | |
| "grad_norm": 2.8984882831573486, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.3142, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.6945516945516945, | |
| "grad_norm": 2.585207939147949, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.3372, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.716001716001716, | |
| "grad_norm": 3.007678985595703, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.3192, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.716001716001716, | |
| "eval_loss": 0.41164642572402954, | |
| "eval_runtime": 1704.3317, | |
| "eval_samples_per_second": 2.289, | |
| "eval_steps_per_second": 0.143, | |
| "eval_wer": 0.2986090988119386, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.7374517374517375, | |
| "grad_norm": 2.6128337383270264, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.3215, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.758901758901759, | |
| "grad_norm": 2.6473169326782227, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.324, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.7803517803517803, | |
| "grad_norm": 2.654451608657837, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.3029, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.8018018018018018, | |
| "grad_norm": 2.335432767868042, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.3197, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.8232518232518231, | |
| "grad_norm": 2.4517362117767334, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.3136, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.8447018447018446, | |
| "grad_norm": 2.7718915939331055, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.3059, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.8661518661518661, | |
| "grad_norm": 2.876389265060425, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.3193, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.8876018876018876, | |
| "grad_norm": 2.892172336578369, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.3097, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.9090519090519091, | |
| "grad_norm": 3.030829429626465, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.3232, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.9305019305019306, | |
| "grad_norm": 2.721433162689209, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.2944, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.951951951951952, | |
| "grad_norm": 3.1279942989349365, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.3063, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.9734019734019734, | |
| "grad_norm": 2.706367254257202, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2998, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.9948519948519947, | |
| "grad_norm": 3.0728838443756104, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.3036, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.016302016302016, | |
| "grad_norm": 2.0992302894592285, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.2477, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.0377520377520377, | |
| "grad_norm": 1.931810975074768, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.2085, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.0592020592020592, | |
| "grad_norm": 2.4714324474334717, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.201, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.0806520806520807, | |
| "grad_norm": 2.1515088081359863, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.1978, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.1021021021021022, | |
| "grad_norm": 2.203587770462036, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.2082, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.1235521235521237, | |
| "grad_norm": 2.3945024013519287, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.2012, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.145002145002145, | |
| "grad_norm": 2.5214550495147705, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.1983, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.1664521664521663, | |
| "grad_norm": 2.8749985694885254, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.202, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.187902187902188, | |
| "grad_norm": 2.302109956741333, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.187, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.2093522093522093, | |
| "grad_norm": 2.303842782974243, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.2007, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.230802230802231, | |
| "grad_norm": 2.1499674320220947, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.2025, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.2522522522522523, | |
| "grad_norm": 2.735565185546875, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.1934, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.273702273702274, | |
| "grad_norm": 2.781230926513672, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.1959, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.2951522951522954, | |
| "grad_norm": 2.934748649597168, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.1923, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.3166023166023164, | |
| "grad_norm": 2.5688586235046387, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.2096, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.338052338052338, | |
| "grad_norm": 2.15342378616333, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.198, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.3595023595023594, | |
| "grad_norm": 2.287332773208618, | |
| "learning_rate": 5e-06, | |
| "loss": 0.2079, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 2.2606818675994873, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.2014, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.4024024024024024, | |
| "grad_norm": 2.0348024368286133, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.2046, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.423852423852424, | |
| "grad_norm": 2.326032876968384, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.1905, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.4453024453024454, | |
| "grad_norm": 2.2641820907592773, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.1959, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.4667524667524665, | |
| "grad_norm": 1.923122763633728, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.199, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.488202488202488, | |
| "grad_norm": 2.273163080215454, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.1983, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.5096525096525095, | |
| "grad_norm": 2.816136598587036, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.1901, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.531102531102531, | |
| "grad_norm": 2.453509569168091, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.1819, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.5525525525525525, | |
| "grad_norm": 1.754265546798706, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.1808, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.574002574002574, | |
| "grad_norm": 2.362611770629883, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.1917, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.574002574002574, | |
| "eval_loss": 0.4086485207080841, | |
| "eval_runtime": 1612.1909, | |
| "eval_samples_per_second": 2.42, | |
| "eval_steps_per_second": 0.151, | |
| "eval_wer": 0.2937071380276248, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.5954525954525955, | |
| "grad_norm": 2.633920669555664, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.188, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.616902616902617, | |
| "grad_norm": 2.290724515914917, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.2027, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.6383526383526386, | |
| "grad_norm": 2.451239585876465, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.1982, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.6598026598026596, | |
| "grad_norm": 2.550907611846924, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.1994, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.681252681252681, | |
| "grad_norm": 2.633718729019165, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.229, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.7027027027027026, | |
| "grad_norm": 2.618665933609009, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.2017, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.724152724152724, | |
| "grad_norm": 2.954728364944458, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.1873, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.7456027456027456, | |
| "grad_norm": 2.27900767326355, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.2034, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.767052767052767, | |
| "grad_norm": 2.202578544616699, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.1955, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.7885027885027887, | |
| "grad_norm": 2.391146659851074, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.1934, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.8099528099528097, | |
| "grad_norm": 2.5382649898529053, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.1907, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.8314028314028317, | |
| "grad_norm": 2.9581174850463867, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.2083, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.8528528528528527, | |
| "grad_norm": 2.907230854034424, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.2074, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.8743028743028742, | |
| "grad_norm": 2.437242031097412, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.193, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.8957528957528957, | |
| "grad_norm": 2.3380300998687744, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.211, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.9172029172029172, | |
| "grad_norm": 3.0222008228302, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.1924, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.9386529386529388, | |
| "grad_norm": 2.763075351715088, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.1989, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.9601029601029603, | |
| "grad_norm": 2.859872817993164, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.1845, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.9815529815529818, | |
| "grad_norm": 2.663336992263794, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.1994, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.003003003003003, | |
| "grad_norm": 1.762622594833374, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.1894, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.0244530244530243, | |
| "grad_norm": 1.7877063751220703, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.1154, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.045903045903046, | |
| "grad_norm": 2.131319761276245, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.1122, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.0673530673530673, | |
| "grad_norm": 1.935207486152649, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.1157, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.088803088803089, | |
| "grad_norm": 2.1667850017547607, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.1165, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.1102531102531104, | |
| "grad_norm": 2.0937955379486084, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.1066, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 3.131703131703132, | |
| "grad_norm": 2.4619686603546143, | |
| "learning_rate": 3e-06, | |
| "loss": 0.1047, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.153153153153153, | |
| "grad_norm": 2.3083388805389404, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.1483, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 3.1746031746031744, | |
| "grad_norm": 2.0462327003479004, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.1183, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.196053196053196, | |
| "grad_norm": 2.2849583625793457, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.1155, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 3.2175032175032174, | |
| "grad_norm": 2.364773750305176, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.1139, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.238953238953239, | |
| "grad_norm": 2.6987013816833496, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.1117, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.2604032604032605, | |
| "grad_norm": 2.8335797786712646, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.1135, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.281853281853282, | |
| "grad_norm": 2.0935230255126953, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.1048, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.3033033033033035, | |
| "grad_norm": 2.1639814376831055, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.1088, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.324753324753325, | |
| "grad_norm": 2.022599458694458, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.1093, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.346203346203346, | |
| "grad_norm": 2.6962931156158447, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.1139, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.3676533676533675, | |
| "grad_norm": 1.9098910093307495, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.1082, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.389103389103389, | |
| "grad_norm": 2.3071112632751465, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.129, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.4105534105534105, | |
| "grad_norm": 1.9279221296310425, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.1153, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.432003432003432, | |
| "grad_norm": 2.1972861289978027, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.1113, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.432003432003432, | |
| "eval_loss": 0.43410712480545044, | |
| "eval_runtime": 2229.8693, | |
| "eval_samples_per_second": 1.749, | |
| "eval_steps_per_second": 0.109, | |
| "eval_wer": 0.2852313339128755, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.4534534534534536, | |
| "grad_norm": 1.9216057062149048, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.1126, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 3.474903474903475, | |
| "grad_norm": 2.277034282684326, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.1088, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.496353496353496, | |
| "grad_norm": 2.0428617000579834, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.1083, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 3.517803517803518, | |
| "grad_norm": 2.5672316551208496, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.1059, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.539253539253539, | |
| "grad_norm": 2.100874185562134, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.1118, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 3.5607035607035606, | |
| "grad_norm": 2.066019296646118, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.1043, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 3.582153582153582, | |
| "grad_norm": 2.2962706089019775, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.1158, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 3.6036036036036037, | |
| "grad_norm": 2.4337549209594727, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.1074, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.625053625053625, | |
| "grad_norm": 2.21247935295105, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.1081, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 3.6465036465036467, | |
| "grad_norm": 1.9160076379776, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.1139, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.667953667953668, | |
| "grad_norm": 2.366440773010254, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.1072, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 3.6894036894036892, | |
| "grad_norm": 2.299656867980957, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.1108, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.7108537108537107, | |
| "grad_norm": 2.0302138328552246, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.1013, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 3.7323037323037322, | |
| "grad_norm": 2.416654586791992, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.1134, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 3.7537537537537538, | |
| "grad_norm": 1.7967655658721924, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.1001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 3.7752037752037753, | |
| "grad_norm": 2.470515251159668, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.1033, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.7966537966537968, | |
| "grad_norm": 2.558213949203491, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.1081, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 3.8181038181038183, | |
| "grad_norm": 2.0769007205963135, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.1027, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.8395538395538393, | |
| "grad_norm": 1.9774547815322876, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.1136, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.861003861003861, | |
| "grad_norm": 2.82588791847229, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.1129, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.8824538824538823, | |
| "grad_norm": 2.3755438327789307, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.1095, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.903903903903904, | |
| "grad_norm": 2.441087007522583, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.1007, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.9253539253539254, | |
| "grad_norm": 2.604708433151245, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.1055, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.946803946803947, | |
| "grad_norm": 2.5220184326171875, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.112, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.9682539682539684, | |
| "grad_norm": 2.3847906589508057, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.1082, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.9897039897039894, | |
| "grad_norm": 1.749206304550171, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.1096, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 4.011154011154011, | |
| "grad_norm": 1.657009482383728, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.084, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 4.032604032604032, | |
| "grad_norm": 1.4502904415130615, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.0624, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 4.054054054054054, | |
| "grad_norm": 1.753985047340393, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.064, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 4.0755040755040755, | |
| "grad_norm": 1.8382900953292847, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.0607, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 4.0969540969540965, | |
| "grad_norm": 2.292609930038452, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.0617, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 4.1184041184041185, | |
| "grad_norm": 1.5938409566879272, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.0643, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 4.1398541398541395, | |
| "grad_norm": 1.7310892343521118, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.0595, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 4.1613041613041615, | |
| "grad_norm": 1.9981136322021484, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.0553, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 4.1827541827541825, | |
| "grad_norm": 1.4661279916763306, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0599, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 4.2042042042042045, | |
| "grad_norm": 2.204902172088623, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.0627, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.2256542256542255, | |
| "grad_norm": 1.7172112464904785, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.0643, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 4.2471042471042475, | |
| "grad_norm": 1.4979685544967651, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.0602, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.268554268554269, | |
| "grad_norm": 1.880300760269165, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.0591, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 4.29000429000429, | |
| "grad_norm": 1.7681268453598022, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.0665, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.29000429000429, | |
| "eval_loss": 0.46873393654823303, | |
| "eval_runtime": 2420.0826, | |
| "eval_samples_per_second": 1.612, | |
| "eval_steps_per_second": 0.101, | |
| "eval_wer": 0.2887085868830291, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.29000429000429, | |
| "step": 5000, | |
| "total_flos": 5.435317790834688e+20, | |
| "train_loss": 0.2924629046201706, | |
| "train_runtime": 60618.8801, | |
| "train_samples_per_second": 2.639, | |
| "train_steps_per_second": 0.082 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "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": 5.435317790834688e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |