| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.1006164902911021, | |
| "eval_steps": 500, | |
| "global_step": 3052, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010219892526291499, | |
| "grad_norm": 52.572940826416016, | |
| "learning_rate": 1.0157273918741808e-06, | |
| "loss": 9.6136, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0020439785052582997, | |
| "grad_norm": 26.900487899780273, | |
| "learning_rate": 2.0314547837483616e-06, | |
| "loss": 7.35, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.003065967757887449, | |
| "grad_norm": 15.413382530212402, | |
| "learning_rate": 3.0471821756225426e-06, | |
| "loss": 5.872, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0040879570105165994, | |
| "grad_norm": 9.475220680236816, | |
| "learning_rate": 4.062909567496723e-06, | |
| "loss": 4.8148, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.005109946263145749, | |
| "grad_norm": 16.33511734008789, | |
| "learning_rate": 5.078636959370905e-06, | |
| "loss": 4.2357, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.006131935515774898, | |
| "grad_norm": 17.79788589477539, | |
| "learning_rate": 6.094364351245085e-06, | |
| "loss": 3.9005, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.007153924768404049, | |
| "grad_norm": 7.385252952575684, | |
| "learning_rate": 7.110091743119267e-06, | |
| "loss": 3.6349, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.008175914021033199, | |
| "grad_norm": 7.6574835777282715, | |
| "learning_rate": 8.125819134993446e-06, | |
| "loss": 3.4197, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.009197903273662348, | |
| "grad_norm": 10.027671813964844, | |
| "learning_rate": 9.141546526867629e-06, | |
| "loss": 3.2883, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.010219892526291498, | |
| "grad_norm": 7.447092056274414, | |
| "learning_rate": 1.015727391874181e-05, | |
| "loss": 3.1606, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.011241881778920647, | |
| "grad_norm": 8.796594619750977, | |
| "learning_rate": 1.117300131061599e-05, | |
| "loss": 3.0724, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.012263871031549797, | |
| "grad_norm": 8.47999382019043, | |
| "learning_rate": 1.218872870249017e-05, | |
| "loss": 3.0249, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.013285860284178948, | |
| "grad_norm": 11.703062057495117, | |
| "learning_rate": 1.3204456094364351e-05, | |
| "loss": 2.9383, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.014307849536808097, | |
| "grad_norm": 7.270478248596191, | |
| "learning_rate": 1.4220183486238533e-05, | |
| "loss": 2.8753, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.015329838789437247, | |
| "grad_norm": 5.699134349822998, | |
| "learning_rate": 1.5235910878112714e-05, | |
| "loss": 2.8226, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.016351828042066398, | |
| "grad_norm": 6.3912529945373535, | |
| "learning_rate": 1.6251638269986893e-05, | |
| "loss": 2.7754, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.017373817294695545, | |
| "grad_norm": 8.240257263183594, | |
| "learning_rate": 1.7267365661861077e-05, | |
| "loss": 2.7351, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.018395806547324697, | |
| "grad_norm": 9.017704010009766, | |
| "learning_rate": 1.8283093053735257e-05, | |
| "loss": 2.7084, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.019417795799953844, | |
| "grad_norm": 6.017426013946533, | |
| "learning_rate": 1.9298820445609438e-05, | |
| "loss": 2.6741, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.020439785052582995, | |
| "grad_norm": 3.789416790008545, | |
| "learning_rate": 2.031454783748362e-05, | |
| "loss": 2.6605, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.021461774305212147, | |
| "grad_norm": 4.747366428375244, | |
| "learning_rate": 2.13302752293578e-05, | |
| "loss": 2.6245, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.022483763557841294, | |
| "grad_norm": 3.8344268798828125, | |
| "learning_rate": 2.234600262123198e-05, | |
| "loss": 2.5791, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.023505752810470446, | |
| "grad_norm": 4.5539116859436035, | |
| "learning_rate": 2.336173001310616e-05, | |
| "loss": 2.5648, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.024527742063099593, | |
| "grad_norm": 4.216089248657227, | |
| "learning_rate": 2.437745740498034e-05, | |
| "loss": 2.5125, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.025549731315728744, | |
| "grad_norm": 4.225554466247559, | |
| "learning_rate": 2.5393184796854525e-05, | |
| "loss": 2.4973, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.026571720568357896, | |
| "grad_norm": 3.357351541519165, | |
| "learning_rate": 2.6408912188728702e-05, | |
| "loss": 2.4532, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.027593709820987043, | |
| "grad_norm": 4.109170436859131, | |
| "learning_rate": 2.7424639580602886e-05, | |
| "loss": 2.4504, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.028615699073616194, | |
| "grad_norm": 17.247940063476562, | |
| "learning_rate": 2.8440366972477066e-05, | |
| "loss": 2.4226, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.029637688326245342, | |
| "grad_norm": 3.2851662635803223, | |
| "learning_rate": 2.9456094364351244e-05, | |
| "loss": 2.4053, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.030659677578874493, | |
| "grad_norm": 3.3796141147613525, | |
| "learning_rate": 3.0471821756225428e-05, | |
| "loss": 2.3977, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.03168166683150364, | |
| "grad_norm": 3.0984764099121094, | |
| "learning_rate": 3.148754914809961e-05, | |
| "loss": 2.3622, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.032703656084132796, | |
| "grad_norm": 2.5464305877685547, | |
| "learning_rate": 3.2503276539973785e-05, | |
| "loss": 2.348, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.03372564533676194, | |
| "grad_norm": 2.848860263824463, | |
| "learning_rate": 3.351900393184797e-05, | |
| "loss": 2.3208, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.03474763458939109, | |
| "grad_norm": 2.5870606899261475, | |
| "learning_rate": 3.453473132372215e-05, | |
| "loss": 2.3023, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.03576962384202024, | |
| "grad_norm": 2.6556804180145264, | |
| "learning_rate": 3.555045871559633e-05, | |
| "loss": 2.2698, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.03679161309464939, | |
| "grad_norm": 2.779650926589966, | |
| "learning_rate": 3.6566186107470514e-05, | |
| "loss": 2.2614, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.03781360234727854, | |
| "grad_norm": 2.624191999435425, | |
| "learning_rate": 3.7581913499344695e-05, | |
| "loss": 2.2422, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.03883559159990769, | |
| "grad_norm": 2.5255484580993652, | |
| "learning_rate": 3.8597640891218876e-05, | |
| "loss": 2.2419, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.03985758085253684, | |
| "grad_norm": 2.3195745944976807, | |
| "learning_rate": 3.9613368283093056e-05, | |
| "loss": 2.2468, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.04087957010516599, | |
| "grad_norm": 2.4712162017822266, | |
| "learning_rate": 4.062909567496724e-05, | |
| "loss": 2.2072, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.04190155935779514, | |
| "grad_norm": 4.451296806335449, | |
| "learning_rate": 4.164482306684142e-05, | |
| "loss": 2.2136, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.04292354861042429, | |
| "grad_norm": 2.4180150032043457, | |
| "learning_rate": 4.26605504587156e-05, | |
| "loss": 2.1843, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.04394553786305344, | |
| "grad_norm": 2.395840883255005, | |
| "learning_rate": 4.367627785058978e-05, | |
| "loss": 2.1447, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.04496752711568259, | |
| "grad_norm": 3.072429895401001, | |
| "learning_rate": 4.469200524246396e-05, | |
| "loss": 2.1405, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 0.045989516368311736, | |
| "grad_norm": 1.9566724300384521, | |
| "learning_rate": 4.570773263433814e-05, | |
| "loss": 2.1321, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.04701150562094089, | |
| "grad_norm": 23.60377311706543, | |
| "learning_rate": 4.672346002621232e-05, | |
| "loss": 2.1289, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 0.04803349487357004, | |
| "grad_norm": 2.117250919342041, | |
| "learning_rate": 4.77391874180865e-05, | |
| "loss": 2.1239, | |
| "step": 1457 | |
| }, | |
| { | |
| "epoch": 0.049055484126199186, | |
| "grad_norm": 2.361362934112549, | |
| "learning_rate": 4.875491480996068e-05, | |
| "loss": 2.0838, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 0.05007747337882834, | |
| "grad_norm": 2.1867992877960205, | |
| "learning_rate": 4.977064220183487e-05, | |
| "loss": 2.0622, | |
| "step": 1519 | |
| }, | |
| { | |
| "epoch": 0.05109946263145749, | |
| "grad_norm": 1.9326456785202026, | |
| "learning_rate": 4.9999915451558777e-05, | |
| "loss": 2.0441, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.052121451884086636, | |
| "grad_norm": 2.076503038406372, | |
| "learning_rate": 4.999955597496219e-05, | |
| "loss": 2.0503, | |
| "step": 1581 | |
| }, | |
| { | |
| "epoch": 0.05314344113671579, | |
| "grad_norm": 3.051212787628174, | |
| "learning_rate": 4.9998914381774255e-05, | |
| "loss": 2.0357, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 0.05416543038934494, | |
| "grad_norm": 1.7924102544784546, | |
| "learning_rate": 4.999799067923527e-05, | |
| "loss": 2.0326, | |
| "step": 1643 | |
| }, | |
| { | |
| "epoch": 0.055187419641974086, | |
| "grad_norm": 1.8103365898132324, | |
| "learning_rate": 4.999678487776908e-05, | |
| "loss": 2.0023, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 0.056209408894603234, | |
| "grad_norm": 1.913725733757019, | |
| "learning_rate": 4.9995296990983006e-05, | |
| "loss": 2.0035, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 0.05723139814723239, | |
| "grad_norm": 1.7415859699249268, | |
| "learning_rate": 4.999352703566763e-05, | |
| "loss": 2.0063, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 0.058253387399861536, | |
| "grad_norm": 1.8145827054977417, | |
| "learning_rate": 4.999147503179668e-05, | |
| "loss": 1.9731, | |
| "step": 1767 | |
| }, | |
| { | |
| "epoch": 0.059275376652490684, | |
| "grad_norm": 1.8761731386184692, | |
| "learning_rate": 4.998914100252672e-05, | |
| "loss": 1.9883, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 0.06029736590511984, | |
| "grad_norm": 6.824073791503906, | |
| "learning_rate": 4.998652497419696e-05, | |
| "loss": 1.9713, | |
| "step": 1829 | |
| }, | |
| { | |
| "epoch": 0.061319355157748986, | |
| "grad_norm": 1.7231906652450562, | |
| "learning_rate": 4.9983626976328927e-05, | |
| "loss": 1.9492, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.062341344410378134, | |
| "grad_norm": 1.6888827085494995, | |
| "learning_rate": 4.998044704162613e-05, | |
| "loss": 1.9449, | |
| "step": 1891 | |
| }, | |
| { | |
| "epoch": 0.06336333366300728, | |
| "grad_norm": 1.5924124717712402, | |
| "learning_rate": 4.9976985205973705e-05, | |
| "loss": 1.938, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 0.06438532291563644, | |
| "grad_norm": 1.7029409408569336, | |
| "learning_rate": 4.997324150843799e-05, | |
| "loss": 1.9285, | |
| "step": 1953 | |
| }, | |
| { | |
| "epoch": 0.06540731216826559, | |
| "grad_norm": 1.5554330348968506, | |
| "learning_rate": 4.99692159912661e-05, | |
| "loss": 1.9091, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 0.06642930142089473, | |
| "grad_norm": 1.5639121532440186, | |
| "learning_rate": 4.996490869988546e-05, | |
| "loss": 1.9105, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 0.06745129067352389, | |
| "grad_norm": 1.6262747049331665, | |
| "learning_rate": 4.996031968290326e-05, | |
| "loss": 1.9164, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 0.06847327992615304, | |
| "grad_norm": 1.5408495664596558, | |
| "learning_rate": 4.995544899210594e-05, | |
| "loss": 1.8728, | |
| "step": 2077 | |
| }, | |
| { | |
| "epoch": 0.06949526917878218, | |
| "grad_norm": 1.5183970928192139, | |
| "learning_rate": 4.9950296682458583e-05, | |
| "loss": 1.8854, | |
| "step": 2108 | |
| }, | |
| { | |
| "epoch": 0.07051725843141134, | |
| "grad_norm": 1.9362810850143433, | |
| "learning_rate": 4.994486281210429e-05, | |
| "loss": 1.8811, | |
| "step": 2139 | |
| }, | |
| { | |
| "epoch": 0.07153924768404048, | |
| "grad_norm": 1.5646640062332153, | |
| "learning_rate": 4.9939147442363566e-05, | |
| "loss": 1.8744, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.07256123693666963, | |
| "grad_norm": 1.5929124355316162, | |
| "learning_rate": 4.9933150637733574e-05, | |
| "loss": 1.867, | |
| "step": 2201 | |
| }, | |
| { | |
| "epoch": 0.07358322618929879, | |
| "grad_norm": 1.7689995765686035, | |
| "learning_rate": 4.992687246588743e-05, | |
| "loss": 1.8659, | |
| "step": 2232 | |
| }, | |
| { | |
| "epoch": 0.07460521544192793, | |
| "grad_norm": 1.375406265258789, | |
| "learning_rate": 4.992031299767347e-05, | |
| "loss": 1.8477, | |
| "step": 2263 | |
| }, | |
| { | |
| "epoch": 0.07562720469455708, | |
| "grad_norm": 1.3596042394638062, | |
| "learning_rate": 4.9913472307114386e-05, | |
| "loss": 1.8498, | |
| "step": 2294 | |
| }, | |
| { | |
| "epoch": 0.07664919394718624, | |
| "grad_norm": 1.3918544054031372, | |
| "learning_rate": 4.9906350471406446e-05, | |
| "loss": 1.8538, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.07767118319981538, | |
| "grad_norm": 1.4614112377166748, | |
| "learning_rate": 4.989894757091861e-05, | |
| "loss": 1.8359, | |
| "step": 2356 | |
| }, | |
| { | |
| "epoch": 0.07869317245244453, | |
| "grad_norm": 1.563186764717102, | |
| "learning_rate": 4.989126368919158e-05, | |
| "loss": 1.821, | |
| "step": 2387 | |
| }, | |
| { | |
| "epoch": 0.07971516170507369, | |
| "grad_norm": 1.3217226266860962, | |
| "learning_rate": 4.988329891293693e-05, | |
| "loss": 1.8345, | |
| "step": 2418 | |
| }, | |
| { | |
| "epoch": 0.08073715095770283, | |
| "grad_norm": 1.3906276226043701, | |
| "learning_rate": 4.987505333203608e-05, | |
| "loss": 1.8313, | |
| "step": 2449 | |
| }, | |
| { | |
| "epoch": 0.08175914021033198, | |
| "grad_norm": 1.2736890316009521, | |
| "learning_rate": 4.9866527039539276e-05, | |
| "loss": 1.8234, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.08278112946296114, | |
| "grad_norm": 1.3050850629806519, | |
| "learning_rate": 4.9857720131664594e-05, | |
| "loss": 1.8181, | |
| "step": 2511 | |
| }, | |
| { | |
| "epoch": 0.08380311871559028, | |
| "grad_norm": 1.3847019672393799, | |
| "learning_rate": 4.9848632707796773e-05, | |
| "loss": 1.8128, | |
| "step": 2542 | |
| }, | |
| { | |
| "epoch": 0.08482510796821943, | |
| "grad_norm": 1.3893433809280396, | |
| "learning_rate": 4.9839264870486155e-05, | |
| "loss": 1.7871, | |
| "step": 2573 | |
| }, | |
| { | |
| "epoch": 0.08584709722084859, | |
| "grad_norm": 1.3469080924987793, | |
| "learning_rate": 4.9829616725447526e-05, | |
| "loss": 1.8212, | |
| "step": 2604 | |
| }, | |
| { | |
| "epoch": 0.08686908647347773, | |
| "grad_norm": 1.269865870475769, | |
| "learning_rate": 4.981968838155888e-05, | |
| "loss": 1.7868, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 0.08789107572610688, | |
| "grad_norm": 1.2972242832183838, | |
| "learning_rate": 4.980947995086024e-05, | |
| "loss": 1.7987, | |
| "step": 2666 | |
| }, | |
| { | |
| "epoch": 0.08891306497873604, | |
| "grad_norm": 1.3488340377807617, | |
| "learning_rate": 4.979899154855234e-05, | |
| "loss": 1.7859, | |
| "step": 2697 | |
| }, | |
| { | |
| "epoch": 0.08993505423136518, | |
| "grad_norm": 1.3487133979797363, | |
| "learning_rate": 4.9788223292995386e-05, | |
| "loss": 1.7851, | |
| "step": 2728 | |
| }, | |
| { | |
| "epoch": 0.09095704348399433, | |
| "grad_norm": 1.2412410974502563, | |
| "learning_rate": 4.977717530570768e-05, | |
| "loss": 1.7802, | |
| "step": 2759 | |
| }, | |
| { | |
| "epoch": 0.09197903273662347, | |
| "grad_norm": 1.3017562627792358, | |
| "learning_rate": 4.976584771136425e-05, | |
| "loss": 1.775, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.09300102198925263, | |
| "grad_norm": 1.2923076152801514, | |
| "learning_rate": 4.975424063779547e-05, | |
| "loss": 1.76, | |
| "step": 2821 | |
| }, | |
| { | |
| "epoch": 0.09402301124188178, | |
| "grad_norm": 1.2735223770141602, | |
| "learning_rate": 4.974235421598557e-05, | |
| "loss": 1.7604, | |
| "step": 2852 | |
| }, | |
| { | |
| "epoch": 0.09504500049451092, | |
| "grad_norm": 1.303673505783081, | |
| "learning_rate": 4.973018858007122e-05, | |
| "loss": 1.7685, | |
| "step": 2883 | |
| }, | |
| { | |
| "epoch": 0.09606698974714008, | |
| "grad_norm": 1.5336410999298096, | |
| "learning_rate": 4.9717743867339963e-05, | |
| "loss": 1.7682, | |
| "step": 2914 | |
| }, | |
| { | |
| "epoch": 0.09708897899976923, | |
| "grad_norm": 5.014227867126465, | |
| "learning_rate": 4.9705020218228695e-05, | |
| "loss": 1.7617, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 0.09811096825239837, | |
| "grad_norm": 1.2603938579559326, | |
| "learning_rate": 4.969201777632205e-05, | |
| "loss": 1.7621, | |
| "step": 2976 | |
| }, | |
| { | |
| "epoch": 0.09913295750502753, | |
| "grad_norm": 1.32491934299469, | |
| "learning_rate": 4.9678736688350846e-05, | |
| "loss": 1.7411, | |
| "step": 3007 | |
| }, | |
| { | |
| "epoch": 0.10015494675765668, | |
| "grad_norm": 1.206735372543335, | |
| "learning_rate": 4.966517710419033e-05, | |
| "loss": 1.7262, | |
| "step": 3038 | |
| } | |
| ], | |
| "logging_steps": 31, | |
| "max_steps": 30517, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 3052, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.131685419530584e+18, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |