| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.8555111364934325, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.014277555682467162, | |
| "grad_norm": 7.571424961090088, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.5088, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.028555111364934323, | |
| "grad_norm": 5.992729187011719, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.2038, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04283266704740148, | |
| "grad_norm": 5.949503421783447, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.8879, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05711022272986865, | |
| "grad_norm": 4.452832221984863, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.7647, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0713877784123358, | |
| "grad_norm": 4.690545558929443, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.6792, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08566533409480297, | |
| "grad_norm": 4.969720840454102, | |
| "learning_rate": 3e-06, | |
| "loss": 0.6549, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09994288977727013, | |
| "grad_norm": 5.184281349182129, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.6376, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1142204454597373, | |
| "grad_norm": 5.00349235534668, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.5982, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.12849800114220444, | |
| "grad_norm": 4.239490032196045, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.6084, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1427755568246716, | |
| "grad_norm": 4.2740068435668945, | |
| "learning_rate": 5e-06, | |
| "loss": 0.58, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.15705311250713877, | |
| "grad_norm": 4.718848705291748, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.5759, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.17133066818960593, | |
| "grad_norm": 4.2935638427734375, | |
| "learning_rate": 6e-06, | |
| "loss": 0.5625, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1856082238720731, | |
| "grad_norm": 4.917020797729492, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.5621, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.19988577955454026, | |
| "grad_norm": 3.9521942138671875, | |
| "learning_rate": 7e-06, | |
| "loss": 0.5644, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.21416333523700742, | |
| "grad_norm": 4.506232738494873, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.5508, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2284408909194746, | |
| "grad_norm": 4.1483540534973145, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.5244, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.24271844660194175, | |
| "grad_norm": 4.077396392822266, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.5051, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.2569960022844089, | |
| "grad_norm": 4.375626087188721, | |
| "learning_rate": 9e-06, | |
| "loss": 0.5222, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2712735579668761, | |
| "grad_norm": 3.5698530673980713, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.5038, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.2855511136493432, | |
| "grad_norm": 4.99509859085083, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5196, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2998286693318104, | |
| "grad_norm": 3.666332721710205, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.5066, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.31410622501427754, | |
| "grad_norm": 3.9203736782073975, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.4822, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.32838378069674473, | |
| "grad_norm": 3.5677530765533447, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.519, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.34266133637921187, | |
| "grad_norm": 3.3873414993286133, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.5205, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.35693889206167906, | |
| "grad_norm": 3.9527816772460938, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.4769, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.3712164477441462, | |
| "grad_norm": 3.3437490463256836, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.4629, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3854940034266134, | |
| "grad_norm": 3.7754790782928467, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.4812, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.3997715591090805, | |
| "grad_norm": 3.744267225265503, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.467, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4140491147915477, | |
| "grad_norm": 3.5076072216033936, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.4454, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.42832667047401485, | |
| "grad_norm": 3.556335687637329, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.4447, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.442604226156482, | |
| "grad_norm": 4.256951332092285, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.4809, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.4568817818389492, | |
| "grad_norm": 3.533447742462158, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.4425, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4711593375214163, | |
| "grad_norm": 4.324098587036133, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.424, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.4854368932038835, | |
| "grad_norm": 2.913189649581909, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.4314, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.49971444888635064, | |
| "grad_norm": 3.432490825653076, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.4355, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.5139920045688178, | |
| "grad_norm": 3.645869255065918, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.4395, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.528269560251285, | |
| "grad_norm": 3.2094240188598633, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.4144, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.5425471159337522, | |
| "grad_norm": 3.4623546600341797, | |
| "learning_rate": 9e-06, | |
| "loss": 0.4277, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5568246716162193, | |
| "grad_norm": 3.640333414077759, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.4246, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.5711022272986864, | |
| "grad_norm": 3.0283167362213135, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.4047, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5711022272986864, | |
| "eval_loss": 0.4848648011684418, | |
| "eval_runtime": 1825.4203, | |
| "eval_samples_per_second": 2.137, | |
| "eval_steps_per_second": 0.134, | |
| "eval_wer": 0.35052641746353713, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5853797829811537, | |
| "grad_norm": 3.7762739658355713, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.4218, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.5996573386636208, | |
| "grad_norm": 3.495347023010254, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.3968, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.613934894346088, | |
| "grad_norm": 3.5088939666748047, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.4108, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.6282124500285551, | |
| "grad_norm": 3.555328845977783, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.4063, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6424900057110223, | |
| "grad_norm": 2.9576587677001953, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.4116, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.6567675613934895, | |
| "grad_norm": 3.280855178833008, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.4083, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.6710451170759566, | |
| "grad_norm": 3.903722047805786, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.411, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.6853226727584237, | |
| "grad_norm": 3.519038438796997, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.3964, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6996002284408909, | |
| "grad_norm": 3.3553972244262695, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.4049, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.7138777841233581, | |
| "grad_norm": 3.3820197582244873, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.4159, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7281553398058253, | |
| "grad_norm": 2.782127857208252, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.3859, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.7424328954882924, | |
| "grad_norm": 3.5839345455169678, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.392, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7567104511707595, | |
| "grad_norm": 3.0308761596679688, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.3899, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.7709880068532268, | |
| "grad_norm": 3.136904001235962, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.3907, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7852655625356939, | |
| "grad_norm": 3.3192756175994873, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.3941, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.799543118218161, | |
| "grad_norm": 4.766107082366943, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3887, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.8138206739006282, | |
| "grad_norm": 4.241744041442871, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.4033, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.8280982295830954, | |
| "grad_norm": 3.1559460163116455, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.3567, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.8423757852655626, | |
| "grad_norm": 3.142645835876465, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.3731, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.8566533409480297, | |
| "grad_norm": 3.1183199882507324, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.3668, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8709308966304968, | |
| "grad_norm": 2.7859325408935547, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.3965, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.885208452312964, | |
| "grad_norm": 3.191088914871216, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.3574, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.8994860079954312, | |
| "grad_norm": 3.0640053749084473, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.3811, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.9137635636778983, | |
| "grad_norm": 3.0769450664520264, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.3788, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.9280411193603655, | |
| "grad_norm": 3.1407933235168457, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3698, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.9423186750428326, | |
| "grad_norm": 3.410187244415283, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.3907, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.9565962307252999, | |
| "grad_norm": 3.3382880687713623, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.3368, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.970873786407767, | |
| "grad_norm": 3.194368600845337, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.369, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.9851513420902341, | |
| "grad_norm": 3.089852809906006, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.3765, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.9994288977727013, | |
| "grad_norm": 3.0002810955047607, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.3705, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.0137064534551685, | |
| "grad_norm": 2.3977696895599365, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.2584, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.0279840091376355, | |
| "grad_norm": 2.3220465183258057, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.2538, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.0422615648201028, | |
| "grad_norm": 2.819687843322754, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.2571, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.05653912050257, | |
| "grad_norm": 2.514644145965576, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2806, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.070816676185037, | |
| "grad_norm": 2.1887128353118896, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.2626, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.0850942318675043, | |
| "grad_norm": 2.592247486114502, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.2509, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.0993717875499716, | |
| "grad_norm": 2.371534824371338, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.2605, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.1136493432324386, | |
| "grad_norm": 3.1825778484344482, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.2495, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.1279268989149058, | |
| "grad_norm": 2.901749849319458, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.261, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.1422044545973729, | |
| "grad_norm": 2.658766984939575, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2476, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.1422044545973729, | |
| "eval_loss": 0.41870468854904175, | |
| "eval_runtime": 1722.2575, | |
| "eval_samples_per_second": 2.265, | |
| "eval_steps_per_second": 0.142, | |
| "eval_wer": 0.3136771950159374, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.15648201027984, | |
| "grad_norm": 2.711312770843506, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.2414, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.1707595659623073, | |
| "grad_norm": 2.9044759273529053, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.2502, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.1850371216447744, | |
| "grad_norm": 2.549725294113159, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.2511, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.1993146773272416, | |
| "grad_norm": 2.95792555809021, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.2427, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.2135922330097086, | |
| "grad_norm": 2.686870574951172, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.2637, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.227869788692176, | |
| "grad_norm": 3.7834455966949463, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.2554, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.2421473443746431, | |
| "grad_norm": 3.0891430377960205, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.2467, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.2564249000571102, | |
| "grad_norm": 2.771472930908203, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.2467, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.2707024557395774, | |
| "grad_norm": 2.6807925701141357, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.2682, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.2849800114220447, | |
| "grad_norm": 2.2320196628570557, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.2408, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.2992575671045117, | |
| "grad_norm": 3.066009759902954, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.2363, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.313535122786979, | |
| "grad_norm": 2.6043167114257812, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2483, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.327812678469446, | |
| "grad_norm": 2.6250624656677246, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.2563, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.3420902341519132, | |
| "grad_norm": 2.508998394012451, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.2581, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.3563677898343802, | |
| "grad_norm": 2.872715473175049, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.2371, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.3706453455168475, | |
| "grad_norm": 3.1910557746887207, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.2515, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.3849229011993147, | |
| "grad_norm": 2.7466485500335693, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.2578, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.3992004568817817, | |
| "grad_norm": 2.388066530227661, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.2541, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.413478012564249, | |
| "grad_norm": 2.688497304916382, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.2514, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.4277555682467162, | |
| "grad_norm": 2.710899591445923, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.2765, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4420331239291833, | |
| "grad_norm": 2.296635389328003, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.2487, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.4563106796116505, | |
| "grad_norm": 2.7988133430480957, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.2499, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.4705882352941178, | |
| "grad_norm": 3.1988582611083984, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.2456, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.4848657909765848, | |
| "grad_norm": 2.657517910003662, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.2613, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.499143346659052, | |
| "grad_norm": 2.5517725944519043, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.2528, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.5134209023415193, | |
| "grad_norm": 2.7166850566864014, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.2476, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.5276984580239863, | |
| "grad_norm": 2.7338292598724365, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.2489, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.5419760137064533, | |
| "grad_norm": 2.1498470306396484, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.2388, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.5562535693889206, | |
| "grad_norm": 2.595247745513916, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.2566, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.5705311250713878, | |
| "grad_norm": 2.652132987976074, | |
| "learning_rate": 5e-06, | |
| "loss": 0.239, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.5848086807538548, | |
| "grad_norm": 2.436605930328369, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.2419, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.599086236436322, | |
| "grad_norm": 2.618035316467285, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.2295, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.6133637921187893, | |
| "grad_norm": 2.2901298999786377, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.2446, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.6276413478012564, | |
| "grad_norm": 2.899315595626831, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.2628, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.6419189034837236, | |
| "grad_norm": 2.616224527359009, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.2273, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.6561964591661908, | |
| "grad_norm": 2.43113112449646, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.2362, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.6704740148486579, | |
| "grad_norm": 2.5203065872192383, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.2428, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.6847515705311251, | |
| "grad_norm": 2.3064985275268555, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.2441, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.6990291262135924, | |
| "grad_norm": 2.201695680618286, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.2324, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.7133066818960594, | |
| "grad_norm": 2.442471981048584, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.2527, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7133066818960594, | |
| "eval_loss": 0.3882293701171875, | |
| "eval_runtime": 1749.1422, | |
| "eval_samples_per_second": 2.23, | |
| "eval_steps_per_second": 0.139, | |
| "eval_wer": 0.2901091471071187, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7275842375785264, | |
| "grad_norm": 2.77786922454834, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.2492, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.7418617932609937, | |
| "grad_norm": 2.5009052753448486, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.2341, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.756139348943461, | |
| "grad_norm": 2.780186176300049, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.2407, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.770416904625928, | |
| "grad_norm": 1.9574618339538574, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.2437, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.7846944603083952, | |
| "grad_norm": 2.151125907897949, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.2341, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.7989720159908624, | |
| "grad_norm": 2.170015811920166, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.2373, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.8132495716733295, | |
| "grad_norm": 3.0467231273651123, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.2317, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.8275271273557967, | |
| "grad_norm": 3.0150015354156494, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.228, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.841804683038264, | |
| "grad_norm": 3.275949001312256, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.2438, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.856082238720731, | |
| "grad_norm": 3.0381839275360107, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.2478, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.8703597944031982, | |
| "grad_norm": 2.770716428756714, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.2312, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.8846373500856655, | |
| "grad_norm": 2.6976678371429443, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.2284, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.8989149057681325, | |
| "grad_norm": 2.8799102306365967, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.2484, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.9131924614505995, | |
| "grad_norm": 2.574629545211792, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.2295, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.927470017133067, | |
| "grad_norm": 2.4746835231781006, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.2335, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.941747572815534, | |
| "grad_norm": 3.084383964538574, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.212, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.956025128498001, | |
| "grad_norm": 2.4441068172454834, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.221, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.9703026841804683, | |
| "grad_norm": 3.031568765640259, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.2341, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.9845802398629355, | |
| "grad_norm": 2.3584327697753906, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.2431, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.9988577955454025, | |
| "grad_norm": 2.1590421199798584, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.2357, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.0131353512278696, | |
| "grad_norm": 2.2845587730407715, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.1576, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.027412906910337, | |
| "grad_norm": 2.033133029937744, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.1422, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.041690462592804, | |
| "grad_norm": 2.2549259662628174, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.1473, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.055968018275271, | |
| "grad_norm": 1.5837754011154175, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.143, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.0702455739577386, | |
| "grad_norm": 1.9988360404968262, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.1416, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.0845231296402056, | |
| "grad_norm": 2.148613929748535, | |
| "learning_rate": 3e-06, | |
| "loss": 0.1338, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.0988006853226726, | |
| "grad_norm": 1.8176393508911133, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.1514, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.11307824100514, | |
| "grad_norm": 2.60271954536438, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.1533, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.127355796687607, | |
| "grad_norm": 2.120281457901001, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.1404, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.141633352370074, | |
| "grad_norm": 2.3522286415100098, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.1511, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.1559109080525416, | |
| "grad_norm": 1.8738924264907837, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.1417, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.1701884637350086, | |
| "grad_norm": 2.255291223526001, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.1437, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.1844660194174756, | |
| "grad_norm": 1.7046154737472534, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.1446, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.198743575099943, | |
| "grad_norm": 2.0543861389160156, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.1504, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.21302113078241, | |
| "grad_norm": 2.139716863632202, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.1345, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.227298686464877, | |
| "grad_norm": 1.7999951839447021, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.1389, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.241576242147344, | |
| "grad_norm": 1.7282090187072754, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.1324, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.2558537978298117, | |
| "grad_norm": 2.6271605491638184, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.1551, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.2701313535122787, | |
| "grad_norm": 2.170382022857666, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.144, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.2844089091947457, | |
| "grad_norm": 1.796635627746582, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.1568, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.2844089091947457, | |
| "eval_loss": 0.3901652991771698, | |
| "eval_runtime": 1765.3609, | |
| "eval_samples_per_second": 2.21, | |
| "eval_steps_per_second": 0.138, | |
| "eval_wer": 0.28160919540229884, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.298686464877213, | |
| "grad_norm": 2.0357980728149414, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.161, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.31296402055968, | |
| "grad_norm": 2.027215003967285, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.1353, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.3272415762421472, | |
| "grad_norm": 2.8169405460357666, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.1449, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.3415191319246147, | |
| "grad_norm": 1.9528751373291016, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.1376, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.3557966876070817, | |
| "grad_norm": 2.5781335830688477, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.1383, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.3700742432895487, | |
| "grad_norm": 2.083077907562256, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.1362, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.384351798972016, | |
| "grad_norm": 2.431272029876709, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.1329, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.3986293546544832, | |
| "grad_norm": 2.157139539718628, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.1377, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.4129069103369503, | |
| "grad_norm": 2.5328071117401123, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.1361, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.4271844660194173, | |
| "grad_norm": 2.433239459991455, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.157, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.4414620217018848, | |
| "grad_norm": 2.5167510509490967, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.132, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.455739577384352, | |
| "grad_norm": 1.9507442712783813, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.1625, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.470017133066819, | |
| "grad_norm": 2.2467007637023926, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.1333, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.4842946887492863, | |
| "grad_norm": 2.4816768169403076, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.1499, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.4985722444317533, | |
| "grad_norm": 2.0616416931152344, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.1508, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.5128498001142203, | |
| "grad_norm": 2.089355230331421, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.1344, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.5271273557966873, | |
| "grad_norm": 2.2235498428344727, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.1717, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.541404911479155, | |
| "grad_norm": 1.9268138408660889, | |
| "learning_rate": 1.2244444444444445e-06, | |
| "loss": 0.143, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.555682467161622, | |
| "grad_norm": 1.8911551237106323, | |
| "learning_rate": 1.168888888888889e-06, | |
| "loss": 0.1439, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.5699600228440893, | |
| "grad_norm": 2.5078868865966797, | |
| "learning_rate": 1.1133333333333334e-06, | |
| "loss": 0.1341, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.5842375785265563, | |
| "grad_norm": 2.1232492923736572, | |
| "learning_rate": 1.0577777777777779e-06, | |
| "loss": 0.1415, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.5985151342090234, | |
| "grad_norm": 1.9214311838150024, | |
| "learning_rate": 1.0022222222222223e-06, | |
| "loss": 0.1301, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.6127926898914904, | |
| "grad_norm": 2.4226858615875244, | |
| "learning_rate": 9.466666666666667e-07, | |
| "loss": 0.1438, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.627070245573958, | |
| "grad_norm": 2.324777126312256, | |
| "learning_rate": 8.911111111111112e-07, | |
| "loss": 0.1306, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.641347801256425, | |
| "grad_norm": 2.427114486694336, | |
| "learning_rate": 8.355555555555556e-07, | |
| "loss": 0.1359, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.655625356938892, | |
| "grad_norm": 1.989882469177246, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.1386, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.6699029126213594, | |
| "grad_norm": 2.6079118251800537, | |
| "learning_rate": 7.244444444444446e-07, | |
| "loss": 0.135, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.6841804683038264, | |
| "grad_norm": 2.3429243564605713, | |
| "learning_rate": 6.68888888888889e-07, | |
| "loss": 0.1356, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.6984580239862934, | |
| "grad_norm": 2.3358540534973145, | |
| "learning_rate": 6.133333333333333e-07, | |
| "loss": 0.1304, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.7127355796687604, | |
| "grad_norm": 1.917809247970581, | |
| "learning_rate": 5.577777777777779e-07, | |
| "loss": 0.1395, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.727013135351228, | |
| "grad_norm": 2.0677952766418457, | |
| "learning_rate": 5.022222222222222e-07, | |
| "loss": 0.1309, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.741290691033695, | |
| "grad_norm": 2.135127305984497, | |
| "learning_rate": 4.466666666666667e-07, | |
| "loss": 0.1424, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.7555682467161624, | |
| "grad_norm": 2.3306682109832764, | |
| "learning_rate": 3.9111111111111115e-07, | |
| "loss": 0.1318, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.7698458023986294, | |
| "grad_norm": 2.0700454711914062, | |
| "learning_rate": 3.3555555555555556e-07, | |
| "loss": 0.1566, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.7841233580810965, | |
| "grad_norm": 1.8561683893203735, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.1453, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.7984009137635635, | |
| "grad_norm": 2.2682347297668457, | |
| "learning_rate": 2.2444444444444445e-07, | |
| "loss": 0.1415, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.812678469446031, | |
| "grad_norm": 2.2898778915405273, | |
| "learning_rate": 1.6888888888888888e-07, | |
| "loss": 0.1427, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.826956025128498, | |
| "grad_norm": 2.328401803970337, | |
| "learning_rate": 1.1333333333333336e-07, | |
| "loss": 0.1357, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.841233580810965, | |
| "grad_norm": 2.2169013023376465, | |
| "learning_rate": 5.777777777777778e-08, | |
| "loss": 0.1343, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.8555111364934325, | |
| "grad_norm": 2.42340350151062, | |
| "learning_rate": 2.2222222222222225e-09, | |
| "loss": 0.1313, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.8555111364934325, | |
| "eval_loss": 0.38383349776268005, | |
| "eval_runtime": 1820.062, | |
| "eval_samples_per_second": 2.143, | |
| "eval_steps_per_second": 0.134, | |
| "eval_wer": 0.27318168646769053, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.8555111364934325, | |
| "step": 5000, | |
| "total_flos": 5.435589590699213e+20, | |
| "train_loss": 0.3002769865989685, | |
| "train_runtime": 59305.2217, | |
| "train_samples_per_second": 2.698, | |
| "train_steps_per_second": 0.084 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 5.435589590699213e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |