| { | |
| "best_metric": 0.772007503464447, | |
| "best_model_checkpoint": "./results/models/nusaparagraph_emot/nusabert-bigru-concate-8-mean\\checkpoint-12222", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 12222, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0286368843069874, | |
| "grad_norm": 4.918474197387695, | |
| "learning_rate": 9.985681557846507e-06, | |
| "loss": 1.918, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0572737686139748, | |
| "grad_norm": 5.332164287567139, | |
| "learning_rate": 9.971363115693013e-06, | |
| "loss": 1.8503, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0859106529209622, | |
| "grad_norm": 13.817869186401367, | |
| "learning_rate": 9.95704467353952e-06, | |
| "loss": 1.7392, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1145475372279496, | |
| "grad_norm": 12.340777397155762, | |
| "learning_rate": 9.942726231386026e-06, | |
| "loss": 1.4836, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.143184421534937, | |
| "grad_norm": 17.78537940979004, | |
| "learning_rate": 9.928407789232532e-06, | |
| "loss": 1.4447, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1718213058419244, | |
| "grad_norm": 15.181681632995605, | |
| "learning_rate": 9.914089347079038e-06, | |
| "loss": 1.1689, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2004581901489118, | |
| "grad_norm": 12.248451232910156, | |
| "learning_rate": 9.899770904925546e-06, | |
| "loss": 1.229, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2290950744558992, | |
| "grad_norm": 15.683427810668945, | |
| "learning_rate": 9.885452462772052e-06, | |
| "loss": 1.0561, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.25773195876288657, | |
| "grad_norm": 17.413002014160156, | |
| "learning_rate": 9.871134020618558e-06, | |
| "loss": 0.9704, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.286368843069874, | |
| "grad_norm": 21.993549346923828, | |
| "learning_rate": 9.856815578465064e-06, | |
| "loss": 1.0062, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3150057273768614, | |
| "grad_norm": 15.140169143676758, | |
| "learning_rate": 9.84249713631157e-06, | |
| "loss": 0.9102, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3436426116838488, | |
| "grad_norm": 25.413042068481445, | |
| "learning_rate": 9.828178694158076e-06, | |
| "loss": 0.8857, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3722794959908362, | |
| "grad_norm": 13.888834953308105, | |
| "learning_rate": 9.813860252004582e-06, | |
| "loss": 0.9794, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4009163802978236, | |
| "grad_norm": 18.061304092407227, | |
| "learning_rate": 9.799541809851088e-06, | |
| "loss": 0.9635, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.42955326460481097, | |
| "grad_norm": 16.933076858520508, | |
| "learning_rate": 9.785223367697596e-06, | |
| "loss": 0.8568, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.4581901489117984, | |
| "grad_norm": 15.249773979187012, | |
| "learning_rate": 9.770904925544102e-06, | |
| "loss": 0.8865, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4868270332187858, | |
| "grad_norm": 16.965801239013672, | |
| "learning_rate": 9.756586483390608e-06, | |
| "loss": 0.776, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5154639175257731, | |
| "grad_norm": 6.5678791999816895, | |
| "learning_rate": 9.742268041237114e-06, | |
| "loss": 0.8698, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.5441008018327605, | |
| "grad_norm": 11.134130477905273, | |
| "learning_rate": 9.72794959908362e-06, | |
| "loss": 0.7746, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.572737686139748, | |
| "grad_norm": 13.739683151245117, | |
| "learning_rate": 9.713631156930127e-06, | |
| "loss": 0.8256, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6013745704467354, | |
| "grad_norm": 13.819252967834473, | |
| "learning_rate": 9.699312714776633e-06, | |
| "loss": 0.8959, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6300114547537228, | |
| "grad_norm": 21.550695419311523, | |
| "learning_rate": 9.68499427262314e-06, | |
| "loss": 0.8417, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6586483390607102, | |
| "grad_norm": 15.80966854095459, | |
| "learning_rate": 9.670675830469645e-06, | |
| "loss": 0.9138, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.6872852233676976, | |
| "grad_norm": 9.978912353515625, | |
| "learning_rate": 9.656357388316153e-06, | |
| "loss": 0.7363, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.715922107674685, | |
| "grad_norm": 15.32123851776123, | |
| "learning_rate": 9.642038946162659e-06, | |
| "loss": 0.849, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7445589919816724, | |
| "grad_norm": 21.705232620239258, | |
| "learning_rate": 9.627720504009165e-06, | |
| "loss": 0.8068, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7731958762886598, | |
| "grad_norm": 23.230682373046875, | |
| "learning_rate": 9.613402061855671e-06, | |
| "loss": 0.7546, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8018327605956472, | |
| "grad_norm": 18.663450241088867, | |
| "learning_rate": 9.599083619702177e-06, | |
| "loss": 0.7417, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.8304696449026346, | |
| "grad_norm": 17.77196502685547, | |
| "learning_rate": 9.584765177548685e-06, | |
| "loss": 0.7517, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.8591065292096219, | |
| "grad_norm": 7.162426471710205, | |
| "learning_rate": 9.57044673539519e-06, | |
| "loss": 0.6956, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8877434135166093, | |
| "grad_norm": 11.551986694335938, | |
| "learning_rate": 9.556128293241697e-06, | |
| "loss": 0.7805, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.9163802978235968, | |
| "grad_norm": 28.649534225463867, | |
| "learning_rate": 9.541809851088203e-06, | |
| "loss": 0.7907, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.9450171821305842, | |
| "grad_norm": 24.329723358154297, | |
| "learning_rate": 9.527491408934708e-06, | |
| "loss": 0.662, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.9736540664375716, | |
| "grad_norm": 8.512353897094727, | |
| "learning_rate": 9.513172966781214e-06, | |
| "loss": 0.7086, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7155204810221721, | |
| "eval_f1": 0.7112801498610356, | |
| "eval_loss": 0.8033041954040527, | |
| "eval_runtime": 36.3098, | |
| "eval_samples_per_second": 73.286, | |
| "eval_steps_per_second": 9.171, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 1.002290950744559, | |
| "grad_norm": 19.602813720703125, | |
| "learning_rate": 9.49885452462772e-06, | |
| "loss": 0.7832, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.0309278350515463, | |
| "grad_norm": 8.915902137756348, | |
| "learning_rate": 9.484536082474226e-06, | |
| "loss": 0.5571, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.0595647193585338, | |
| "grad_norm": 44.88418960571289, | |
| "learning_rate": 9.470217640320734e-06, | |
| "loss": 0.5963, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.088201603665521, | |
| "grad_norm": 7.115338325500488, | |
| "learning_rate": 9.45589919816724e-06, | |
| "loss": 0.4956, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.1168384879725086, | |
| "grad_norm": 16.892269134521484, | |
| "learning_rate": 9.441580756013746e-06, | |
| "loss": 0.4767, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.145475372279496, | |
| "grad_norm": 30.78487205505371, | |
| "learning_rate": 9.427262313860252e-06, | |
| "loss": 0.5456, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.1741122565864834, | |
| "grad_norm": 12.62244987487793, | |
| "learning_rate": 9.41294387170676e-06, | |
| "loss": 0.531, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.2027491408934707, | |
| "grad_norm": 20.651899337768555, | |
| "learning_rate": 9.398625429553266e-06, | |
| "loss": 0.529, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.2313860252004583, | |
| "grad_norm": 21.23448371887207, | |
| "learning_rate": 9.384306987399772e-06, | |
| "loss": 0.5307, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.2600229095074456, | |
| "grad_norm": 6.756899356842041, | |
| "learning_rate": 9.369988545246278e-06, | |
| "loss": 0.5302, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.2886597938144329, | |
| "grad_norm": 11.766059875488281, | |
| "learning_rate": 9.355670103092784e-06, | |
| "loss": 0.5169, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.3172966781214204, | |
| "grad_norm": 12.042119026184082, | |
| "learning_rate": 9.341351660939291e-06, | |
| "loss": 0.4705, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.345933562428408, | |
| "grad_norm": 13.262438774108887, | |
| "learning_rate": 9.327033218785797e-06, | |
| "loss": 0.5263, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.3745704467353952, | |
| "grad_norm": 24.665464401245117, | |
| "learning_rate": 9.312714776632303e-06, | |
| "loss": 0.5039, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.4032073310423825, | |
| "grad_norm": 24.94556427001953, | |
| "learning_rate": 9.29839633447881e-06, | |
| "loss": 0.5736, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.43184421534937, | |
| "grad_norm": 18.88966941833496, | |
| "learning_rate": 9.284077892325315e-06, | |
| "loss": 0.5569, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4604810996563573, | |
| "grad_norm": 8.994138717651367, | |
| "learning_rate": 9.269759450171823e-06, | |
| "loss": 0.4279, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.4891179839633448, | |
| "grad_norm": 8.518653869628906, | |
| "learning_rate": 9.255441008018329e-06, | |
| "loss": 0.4764, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.5177548682703321, | |
| "grad_norm": 35.21455764770508, | |
| "learning_rate": 9.241122565864835e-06, | |
| "loss": 0.5318, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.5463917525773194, | |
| "grad_norm": 13.781972885131836, | |
| "learning_rate": 9.226804123711341e-06, | |
| "loss": 0.4229, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.575028636884307, | |
| "grad_norm": 7.166682720184326, | |
| "learning_rate": 9.212485681557847e-06, | |
| "loss": 0.511, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.6036655211912945, | |
| "grad_norm": 20.155414581298828, | |
| "learning_rate": 9.198167239404353e-06, | |
| "loss": 0.5216, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.6323024054982818, | |
| "grad_norm": 13.758686065673828, | |
| "learning_rate": 9.183848797250859e-06, | |
| "loss": 0.5359, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.660939289805269, | |
| "grad_norm": 17.821603775024414, | |
| "learning_rate": 9.169530355097367e-06, | |
| "loss": 0.5762, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.6895761741122566, | |
| "grad_norm": 22.8956356048584, | |
| "learning_rate": 9.155211912943873e-06, | |
| "loss": 0.5629, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.718213058419244, | |
| "grad_norm": 3.841970443725586, | |
| "learning_rate": 9.140893470790379e-06, | |
| "loss": 0.4847, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7468499427262314, | |
| "grad_norm": 22.353851318359375, | |
| "learning_rate": 9.126575028636885e-06, | |
| "loss": 0.499, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.7754868270332187, | |
| "grad_norm": 6.219491481781006, | |
| "learning_rate": 9.11225658648339e-06, | |
| "loss": 0.6343, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.8041237113402062, | |
| "grad_norm": 26.25059700012207, | |
| "learning_rate": 9.097938144329898e-06, | |
| "loss": 0.5259, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.8327605956471937, | |
| "grad_norm": 14.963705062866211, | |
| "learning_rate": 9.083619702176404e-06, | |
| "loss": 0.5091, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.861397479954181, | |
| "grad_norm": 4.209954261779785, | |
| "learning_rate": 9.06930126002291e-06, | |
| "loss": 0.5591, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.8900343642611683, | |
| "grad_norm": 12.075881958007812, | |
| "learning_rate": 9.054982817869416e-06, | |
| "loss": 0.4983, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.9186712485681556, | |
| "grad_norm": 14.535738945007324, | |
| "learning_rate": 9.040664375715922e-06, | |
| "loss": 0.4013, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.9473081328751431, | |
| "grad_norm": 12.017505645751953, | |
| "learning_rate": 9.02634593356243e-06, | |
| "loss": 0.5307, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.9759450171821307, | |
| "grad_norm": 24.822385787963867, | |
| "learning_rate": 9.012027491408936e-06, | |
| "loss": 0.4848, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7576099210822999, | |
| "eval_f1": 0.7572228179173918, | |
| "eval_loss": 0.7472622394561768, | |
| "eval_runtime": 36.2957, | |
| "eval_samples_per_second": 73.314, | |
| "eval_steps_per_second": 9.175, | |
| "step": 3492 | |
| }, | |
| { | |
| "epoch": 2.004581901489118, | |
| "grad_norm": 0.584539532661438, | |
| "learning_rate": 8.997709049255442e-06, | |
| "loss": 0.4442, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.0332187857961053, | |
| "grad_norm": 48.69373321533203, | |
| "learning_rate": 8.983390607101948e-06, | |
| "loss": 0.3488, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.0618556701030926, | |
| "grad_norm": 6.41154670715332, | |
| "learning_rate": 8.969072164948455e-06, | |
| "loss": 0.3281, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.0904925544100803, | |
| "grad_norm": 29.08100700378418, | |
| "learning_rate": 8.954753722794961e-06, | |
| "loss": 0.2916, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.1191294387170676, | |
| "grad_norm": 9.118326187133789, | |
| "learning_rate": 8.940435280641467e-06, | |
| "loss": 0.2665, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.147766323024055, | |
| "grad_norm": 3.8800840377807617, | |
| "learning_rate": 8.926116838487973e-06, | |
| "loss": 0.2697, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.176403207331042, | |
| "grad_norm": 14.715743064880371, | |
| "learning_rate": 8.91179839633448e-06, | |
| "loss": 0.3374, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.20504009163803, | |
| "grad_norm": 46.47842025756836, | |
| "learning_rate": 8.897479954180985e-06, | |
| "loss": 0.2725, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.2336769759450172, | |
| "grad_norm": 25.640724182128906, | |
| "learning_rate": 8.883161512027491e-06, | |
| "loss": 0.2656, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.2623138602520045, | |
| "grad_norm": 33.861568450927734, | |
| "learning_rate": 8.868843069873997e-06, | |
| "loss": 0.3202, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.290950744558992, | |
| "grad_norm": 1.642096996307373, | |
| "learning_rate": 8.854524627720505e-06, | |
| "loss": 0.2686, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.319587628865979, | |
| "grad_norm": 10.57680606842041, | |
| "learning_rate": 8.840206185567011e-06, | |
| "loss": 0.2817, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.348224513172967, | |
| "grad_norm": 22.718547821044922, | |
| "learning_rate": 8.825887743413517e-06, | |
| "loss": 0.3209, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.376861397479954, | |
| "grad_norm": 39.80268859863281, | |
| "learning_rate": 8.811569301260023e-06, | |
| "loss": 0.3598, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.4054982817869415, | |
| "grad_norm": 13.368090629577637, | |
| "learning_rate": 8.797250859106529e-06, | |
| "loss": 0.3448, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.434135166093929, | |
| "grad_norm": 55.12387466430664, | |
| "learning_rate": 8.782932416953037e-06, | |
| "loss": 0.2845, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.4627720504009165, | |
| "grad_norm": 8.037142753601074, | |
| "learning_rate": 8.768613974799543e-06, | |
| "loss": 0.2482, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.491408934707904, | |
| "grad_norm": 14.919239044189453, | |
| "learning_rate": 8.754295532646049e-06, | |
| "loss": 0.2222, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.520045819014891, | |
| "grad_norm": 30.16111946105957, | |
| "learning_rate": 8.739977090492555e-06, | |
| "loss": 0.2577, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.5486827033218784, | |
| "grad_norm": 18.18327522277832, | |
| "learning_rate": 8.725658648339062e-06, | |
| "loss": 0.2802, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.5773195876288657, | |
| "grad_norm": 14.579641342163086, | |
| "learning_rate": 8.711340206185568e-06, | |
| "loss": 0.2528, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.6059564719358534, | |
| "grad_norm": 17.266183853149414, | |
| "learning_rate": 8.697021764032074e-06, | |
| "loss": 0.3592, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.6345933562428407, | |
| "grad_norm": 16.718862533569336, | |
| "learning_rate": 8.68270332187858e-06, | |
| "loss": 0.3184, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.663230240549828, | |
| "grad_norm": 25.249649047851562, | |
| "learning_rate": 8.668384879725086e-06, | |
| "loss": 0.2697, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.691867124856816, | |
| "grad_norm": 11.022406578063965, | |
| "learning_rate": 8.654066437571594e-06, | |
| "loss": 0.2776, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.720504009163803, | |
| "grad_norm": 8.15123176574707, | |
| "learning_rate": 8.6397479954181e-06, | |
| "loss": 0.3462, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.7491408934707904, | |
| "grad_norm": 11.458098411560059, | |
| "learning_rate": 8.625429553264606e-06, | |
| "loss": 0.3657, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 3.9528236389160156, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.3086, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.806414662084765, | |
| "grad_norm": 39.1205940246582, | |
| "learning_rate": 8.596792668957618e-06, | |
| "loss": 0.3528, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.8350515463917527, | |
| "grad_norm": 53.56584167480469, | |
| "learning_rate": 8.582474226804124e-06, | |
| "loss": 0.2909, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.86368843069874, | |
| "grad_norm": 51.294132232666016, | |
| "learning_rate": 8.56815578465063e-06, | |
| "loss": 0.2571, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.8923253150057273, | |
| "grad_norm": 48.5758056640625, | |
| "learning_rate": 8.553837342497136e-06, | |
| "loss": 0.2812, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.9209621993127146, | |
| "grad_norm": 9.625508308410645, | |
| "learning_rate": 8.539518900343643e-06, | |
| "loss": 0.2606, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.9495990836197024, | |
| "grad_norm": 22.97384262084961, | |
| "learning_rate": 8.52520045819015e-06, | |
| "loss": 0.3566, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.9782359679266897, | |
| "grad_norm": 34.34775161743164, | |
| "learning_rate": 8.510882016036655e-06, | |
| "loss": 0.3251, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7493423524990604, | |
| "eval_f1": 0.7488536960670391, | |
| "eval_loss": 0.9042189121246338, | |
| "eval_runtime": 36.3022, | |
| "eval_samples_per_second": 73.301, | |
| "eval_steps_per_second": 9.173, | |
| "step": 5238 | |
| }, | |
| { | |
| "epoch": 3.006872852233677, | |
| "grad_norm": 0.47577229142189026, | |
| "learning_rate": 8.496563573883161e-06, | |
| "loss": 0.2671, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.0355097365406642, | |
| "grad_norm": 36.310089111328125, | |
| "learning_rate": 8.482245131729669e-06, | |
| "loss": 0.1493, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.0641466208476515, | |
| "grad_norm": 29.30960464477539, | |
| "learning_rate": 8.467926689576175e-06, | |
| "loss": 0.1603, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 3.0927835051546393, | |
| "grad_norm": 2.112786054611206, | |
| "learning_rate": 8.453608247422681e-06, | |
| "loss": 0.1363, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.1214203894616266, | |
| "grad_norm": 3.0346627235412598, | |
| "learning_rate": 8.439289805269187e-06, | |
| "loss": 0.1359, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 3.150057273768614, | |
| "grad_norm": 22.331645965576172, | |
| "learning_rate": 8.424971363115693e-06, | |
| "loss": 0.1479, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.178694158075601, | |
| "grad_norm": 13.065728187561035, | |
| "learning_rate": 8.4106529209622e-06, | |
| "loss": 0.1311, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 3.207331042382589, | |
| "grad_norm": 3.163928270339966, | |
| "learning_rate": 8.396334478808707e-06, | |
| "loss": 0.1364, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.2359679266895762, | |
| "grad_norm": 33.371849060058594, | |
| "learning_rate": 8.382016036655213e-06, | |
| "loss": 0.1485, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.2646048109965635, | |
| "grad_norm": 24.721717834472656, | |
| "learning_rate": 8.367697594501719e-06, | |
| "loss": 0.155, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.293241695303551, | |
| "grad_norm": 15.719419479370117, | |
| "learning_rate": 8.353379152348225e-06, | |
| "loss": 0.1539, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.3218785796105386, | |
| "grad_norm": 41.62794876098633, | |
| "learning_rate": 8.339060710194732e-06, | |
| "loss": 0.1695, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.350515463917526, | |
| "grad_norm": 1.7000266313552856, | |
| "learning_rate": 8.324742268041238e-06, | |
| "loss": 0.187, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.379152348224513, | |
| "grad_norm": 9.574384689331055, | |
| "learning_rate": 8.310423825887744e-06, | |
| "loss": 0.1909, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.4077892325315005, | |
| "grad_norm": 19.80113410949707, | |
| "learning_rate": 8.29610538373425e-06, | |
| "loss": 0.1656, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 3.436426116838488, | |
| "grad_norm": 71.53827667236328, | |
| "learning_rate": 8.281786941580758e-06, | |
| "loss": 0.1116, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.4650630011454755, | |
| "grad_norm": 65.89808654785156, | |
| "learning_rate": 8.267468499427262e-06, | |
| "loss": 0.188, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 3.493699885452463, | |
| "grad_norm": 60.68288803100586, | |
| "learning_rate": 8.253150057273768e-06, | |
| "loss": 0.1315, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.52233676975945, | |
| "grad_norm": 25.550884246826172, | |
| "learning_rate": 8.238831615120276e-06, | |
| "loss": 0.1194, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.5509736540664374, | |
| "grad_norm": 6.416481971740723, | |
| "learning_rate": 8.224513172966782e-06, | |
| "loss": 0.134, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.579610538373425, | |
| "grad_norm": 1.5923579931259155, | |
| "learning_rate": 8.210194730813288e-06, | |
| "loss": 0.1672, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.6082474226804124, | |
| "grad_norm": 19.590898513793945, | |
| "learning_rate": 8.195876288659794e-06, | |
| "loss": 0.14, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.6368843069873997, | |
| "grad_norm": 0.4376499652862549, | |
| "learning_rate": 8.1815578465063e-06, | |
| "loss": 0.1425, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.665521191294387, | |
| "grad_norm": 11.566116333007812, | |
| "learning_rate": 8.167239404352808e-06, | |
| "loss": 0.205, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.6941580756013748, | |
| "grad_norm": 62.71388626098633, | |
| "learning_rate": 8.152920962199314e-06, | |
| "loss": 0.184, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.722794959908362, | |
| "grad_norm": 0.32115602493286133, | |
| "learning_rate": 8.13860252004582e-06, | |
| "loss": 0.1267, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.7514318442153494, | |
| "grad_norm": 53.07960891723633, | |
| "learning_rate": 8.124284077892326e-06, | |
| "loss": 0.1796, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.7800687285223367, | |
| "grad_norm": 2.265838861465454, | |
| "learning_rate": 8.109965635738832e-06, | |
| "loss": 0.2187, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.808705612829324, | |
| "grad_norm": 0.6725891828536987, | |
| "learning_rate": 8.09564719358534e-06, | |
| "loss": 0.1518, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.8373424971363117, | |
| "grad_norm": 53.60797882080078, | |
| "learning_rate": 8.081328751431845e-06, | |
| "loss": 0.1619, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.865979381443299, | |
| "grad_norm": 22.341100692749023, | |
| "learning_rate": 8.067010309278351e-06, | |
| "loss": 0.1378, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.8946162657502863, | |
| "grad_norm": 64.15727996826172, | |
| "learning_rate": 8.052691867124857e-06, | |
| "loss": 0.1445, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.923253150057274, | |
| "grad_norm": 36.292484283447266, | |
| "learning_rate": 8.038373424971365e-06, | |
| "loss": 0.1612, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.9518900343642613, | |
| "grad_norm": 4.331699848175049, | |
| "learning_rate": 8.02405498281787e-06, | |
| "loss": 0.2337, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.9805269186712486, | |
| "grad_norm": 0.28292712569236755, | |
| "learning_rate": 8.009736540664377e-06, | |
| "loss": 0.1779, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7606163096580233, | |
| "eval_f1": 0.7597313264211647, | |
| "eval_loss": 0.9818114042282104, | |
| "eval_runtime": 36.2859, | |
| "eval_samples_per_second": 73.334, | |
| "eval_steps_per_second": 9.177, | |
| "step": 6984 | |
| }, | |
| { | |
| "epoch": 4.009163802978236, | |
| "grad_norm": 14.622065544128418, | |
| "learning_rate": 7.995418098510883e-06, | |
| "loss": 0.1005, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.037800687285223, | |
| "grad_norm": 0.4035070836544037, | |
| "learning_rate": 7.981099656357389e-06, | |
| "loss": 0.0777, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 4.0664375715922105, | |
| "grad_norm": 31.626924514770508, | |
| "learning_rate": 7.966781214203895e-06, | |
| "loss": 0.0581, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 4.095074455899198, | |
| "grad_norm": 0.15928135812282562, | |
| "learning_rate": 7.9524627720504e-06, | |
| "loss": 0.0477, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 4.123711340206185, | |
| "grad_norm": 49.666709899902344, | |
| "learning_rate": 7.938144329896907e-06, | |
| "loss": 0.0752, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 4.152348224513173, | |
| "grad_norm": 0.6751652956008911, | |
| "learning_rate": 7.923825887743414e-06, | |
| "loss": 0.0807, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 4.180985108820161, | |
| "grad_norm": 0.623131513595581, | |
| "learning_rate": 7.90950744558992e-06, | |
| "loss": 0.1205, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 4.209621993127148, | |
| "grad_norm": 0.47417309880256653, | |
| "learning_rate": 7.895189003436426e-06, | |
| "loss": 0.0286, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 4.238258877434135, | |
| "grad_norm": 0.37885400652885437, | |
| "learning_rate": 7.880870561282932e-06, | |
| "loss": 0.0747, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 4.2668957617411225, | |
| "grad_norm": 1.7955352067947388, | |
| "learning_rate": 7.866552119129438e-06, | |
| "loss": 0.0544, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 4.29553264604811, | |
| "grad_norm": 46.27180862426758, | |
| "learning_rate": 7.852233676975946e-06, | |
| "loss": 0.0447, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 4.324169530355097, | |
| "grad_norm": 2.345630645751953, | |
| "learning_rate": 7.837915234822452e-06, | |
| "loss": 0.0599, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 4.352806414662084, | |
| "grad_norm": 0.334553986787796, | |
| "learning_rate": 7.823596792668958e-06, | |
| "loss": 0.0494, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 4.381443298969073, | |
| "grad_norm": 8.703739166259766, | |
| "learning_rate": 7.809278350515464e-06, | |
| "loss": 0.0523, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 4.41008018327606, | |
| "grad_norm": 36.21126937866211, | |
| "learning_rate": 7.794959908361972e-06, | |
| "loss": 0.115, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 4.438717067583047, | |
| "grad_norm": 0.7834786772727966, | |
| "learning_rate": 7.780641466208478e-06, | |
| "loss": 0.0602, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 4.4673539518900345, | |
| "grad_norm": 1.5702382326126099, | |
| "learning_rate": 7.766323024054984e-06, | |
| "loss": 0.1341, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 4.495990836197022, | |
| "grad_norm": 0.2708655595779419, | |
| "learning_rate": 7.75200458190149e-06, | |
| "loss": 0.1514, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 4.524627720504009, | |
| "grad_norm": 0.48853299021720886, | |
| "learning_rate": 7.737686139747996e-06, | |
| "loss": 0.1418, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 4.553264604810996, | |
| "grad_norm": 36.37514114379883, | |
| "learning_rate": 7.723367697594503e-06, | |
| "loss": 0.0663, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 4.581901489117984, | |
| "grad_norm": 30.906522750854492, | |
| "learning_rate": 7.70904925544101e-06, | |
| "loss": 0.0655, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.610538373424971, | |
| "grad_norm": 3.7403993606567383, | |
| "learning_rate": 7.694730813287515e-06, | |
| "loss": 0.0496, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 4.639175257731958, | |
| "grad_norm": 0.10339858382940292, | |
| "learning_rate": 7.680412371134021e-06, | |
| "loss": 0.1213, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 4.6678121420389465, | |
| "grad_norm": 0.03920993208885193, | |
| "learning_rate": 7.666093928980529e-06, | |
| "loss": 0.1407, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.696449026345934, | |
| "grad_norm": 70.78038024902344, | |
| "learning_rate": 7.651775486827033e-06, | |
| "loss": 0.0922, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.725085910652921, | |
| "grad_norm": 20.5271053314209, | |
| "learning_rate": 7.63745704467354e-06, | |
| "loss": 0.1784, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.753722794959908, | |
| "grad_norm": 0.17482055723667145, | |
| "learning_rate": 7.623138602520046e-06, | |
| "loss": 0.1508, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.782359679266896, | |
| "grad_norm": 33.39112091064453, | |
| "learning_rate": 7.608820160366552e-06, | |
| "loss": 0.1115, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.810996563573883, | |
| "grad_norm": 0.4592705965042114, | |
| "learning_rate": 7.594501718213059e-06, | |
| "loss": 0.1188, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.83963344788087, | |
| "grad_norm": 38.45854187011719, | |
| "learning_rate": 7.580183276059565e-06, | |
| "loss": 0.0788, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.868270332187858, | |
| "grad_norm": 60.54808044433594, | |
| "learning_rate": 7.565864833906072e-06, | |
| "loss": 0.1159, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.896907216494846, | |
| "grad_norm": 0.5872980952262878, | |
| "learning_rate": 7.551546391752578e-06, | |
| "loss": 0.0984, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.925544100801833, | |
| "grad_norm": 0.09889842569828033, | |
| "learning_rate": 7.5372279495990845e-06, | |
| "loss": 0.1722, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.95418098510882, | |
| "grad_norm": 0.07930275052785873, | |
| "learning_rate": 7.5229095074455904e-06, | |
| "loss": 0.0861, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 4.982817869415808, | |
| "grad_norm": 9.635697364807129, | |
| "learning_rate": 7.5085910652920964e-06, | |
| "loss": 0.1452, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7602405110860578, | |
| "eval_f1": 0.7599417916848927, | |
| "eval_loss": 1.1236233711242676, | |
| "eval_runtime": 36.3176, | |
| "eval_samples_per_second": 73.27, | |
| "eval_steps_per_second": 9.169, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 5.011454753722795, | |
| "grad_norm": 0.5619956254959106, | |
| "learning_rate": 7.494272623138603e-06, | |
| "loss": 0.0687, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 5.040091638029782, | |
| "grad_norm": 85.73548126220703, | |
| "learning_rate": 7.479954180985109e-06, | |
| "loss": 0.0478, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 5.0687285223367695, | |
| "grad_norm": 13.715363502502441, | |
| "learning_rate": 7.465635738831616e-06, | |
| "loss": 0.0396, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 5.097365406643757, | |
| "grad_norm": 70.08007049560547, | |
| "learning_rate": 7.451317296678122e-06, | |
| "loss": 0.0562, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 5.126002290950744, | |
| "grad_norm": 0.8938388824462891, | |
| "learning_rate": 7.436998854524629e-06, | |
| "loss": 0.0974, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 5.154639175257732, | |
| "grad_norm": 0.3734937012195587, | |
| "learning_rate": 7.422680412371135e-06, | |
| "loss": 0.0742, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 5.18327605956472, | |
| "grad_norm": 10.571038246154785, | |
| "learning_rate": 7.408361970217641e-06, | |
| "loss": 0.021, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 5.211912943871707, | |
| "grad_norm": 0.09976017475128174, | |
| "learning_rate": 7.394043528064148e-06, | |
| "loss": 0.0546, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 5.240549828178694, | |
| "grad_norm": 0.07766138762235641, | |
| "learning_rate": 7.379725085910654e-06, | |
| "loss": 0.0827, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 5.2691867124856815, | |
| "grad_norm": 0.021182745695114136, | |
| "learning_rate": 7.3654066437571605e-06, | |
| "loss": 0.0508, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 5.297823596792669, | |
| "grad_norm": 0.08311637490987778, | |
| "learning_rate": 7.3510882016036665e-06, | |
| "loss": 0.0359, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 5.326460481099656, | |
| "grad_norm": 10.934857368469238, | |
| "learning_rate": 7.336769759450172e-06, | |
| "loss": 0.0431, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 5.355097365406643, | |
| "grad_norm": 0.061881110072135925, | |
| "learning_rate": 7.3224513172966785e-06, | |
| "loss": 0.0245, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 5.383734249713632, | |
| "grad_norm": 0.09978004544973373, | |
| "learning_rate": 7.3081328751431845e-06, | |
| "loss": 0.0201, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 5.412371134020619, | |
| "grad_norm": 0.04522474855184555, | |
| "learning_rate": 7.293814432989691e-06, | |
| "loss": 0.0554, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 5.441008018327606, | |
| "grad_norm": 27.210500717163086, | |
| "learning_rate": 7.279495990836197e-06, | |
| "loss": 0.0289, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 5.4696449026345935, | |
| "grad_norm": 88.04645538330078, | |
| "learning_rate": 7.265177548682703e-06, | |
| "loss": 0.0786, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 5.498281786941581, | |
| "grad_norm": 9.045210838317871, | |
| "learning_rate": 7.25085910652921e-06, | |
| "loss": 0.0776, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 5.526918671248568, | |
| "grad_norm": 0.031076449900865555, | |
| "learning_rate": 7.236540664375716e-06, | |
| "loss": 0.0917, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 0.07342424243688583, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.067, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 5.584192439862543, | |
| "grad_norm": 0.07349126785993576, | |
| "learning_rate": 7.207903780068729e-06, | |
| "loss": 0.0275, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 5.61282932416953, | |
| "grad_norm": 0.4687565267086029, | |
| "learning_rate": 7.193585337915236e-06, | |
| "loss": 0.0308, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 5.641466208476518, | |
| "grad_norm": 0.03109198622405529, | |
| "learning_rate": 7.179266895761742e-06, | |
| "loss": 0.0512, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 5.670103092783505, | |
| "grad_norm": 2.063464403152466, | |
| "learning_rate": 7.164948453608248e-06, | |
| "loss": 0.0436, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 5.698739977090493, | |
| "grad_norm": 0.7259678244590759, | |
| "learning_rate": 7.1506300114547545e-06, | |
| "loss": 0.0452, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 5.72737686139748, | |
| "grad_norm": 22.87729263305664, | |
| "learning_rate": 7.1363115693012605e-06, | |
| "loss": 0.0871, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.756013745704467, | |
| "grad_norm": 0.6141005754470825, | |
| "learning_rate": 7.121993127147767e-06, | |
| "loss": 0.0432, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 5.784650630011455, | |
| "grad_norm": 0.29292017221450806, | |
| "learning_rate": 7.107674684994273e-06, | |
| "loss": 0.0424, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 5.813287514318442, | |
| "grad_norm": 0.059430770576000214, | |
| "learning_rate": 7.09335624284078e-06, | |
| "loss": 0.0751, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 5.841924398625429, | |
| "grad_norm": 10.410508155822754, | |
| "learning_rate": 7.079037800687286e-06, | |
| "loss": 0.1061, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 5.870561282932417, | |
| "grad_norm": 1.1330084800720215, | |
| "learning_rate": 7.064719358533792e-06, | |
| "loss": 0.0421, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 5.899198167239405, | |
| "grad_norm": 0.1360820233821869, | |
| "learning_rate": 7.050400916380299e-06, | |
| "loss": 0.0664, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 5.927835051546392, | |
| "grad_norm": 1.3874069452285767, | |
| "learning_rate": 7.036082474226805e-06, | |
| "loss": 0.1134, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 5.956471935853379, | |
| "grad_norm": 0.0816207155585289, | |
| "learning_rate": 7.02176403207331e-06, | |
| "loss": 0.052, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 5.985108820160367, | |
| "grad_norm": 0.0460013747215271, | |
| "learning_rate": 7.007445589919817e-06, | |
| "loss": 0.0362, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.7591131153701616, | |
| "eval_f1": 0.7565525682738199, | |
| "eval_loss": 1.2660281658172607, | |
| "eval_runtime": 36.2808, | |
| "eval_samples_per_second": 73.345, | |
| "eval_steps_per_second": 9.178, | |
| "step": 10476 | |
| }, | |
| { | |
| "epoch": 6.013745704467354, | |
| "grad_norm": 0.09728560596704483, | |
| "learning_rate": 6.993127147766323e-06, | |
| "loss": 0.0709, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 6.042382588774341, | |
| "grad_norm": 0.10100292414426804, | |
| "learning_rate": 6.97880870561283e-06, | |
| "loss": 0.0754, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 6.0710194730813285, | |
| "grad_norm": 0.19107265770435333, | |
| "learning_rate": 6.964490263459336e-06, | |
| "loss": 0.0408, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 6.099656357388316, | |
| "grad_norm": 0.0762198194861412, | |
| "learning_rate": 6.9501718213058426e-06, | |
| "loss": 0.0228, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 6.128293241695303, | |
| "grad_norm": 0.023416342213749886, | |
| "learning_rate": 6.9358533791523485e-06, | |
| "loss": 0.0326, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 6.156930126002291, | |
| "grad_norm": 0.13443826138973236, | |
| "learning_rate": 6.9215349369988545e-06, | |
| "loss": 0.021, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 6.185567010309279, | |
| "grad_norm": 47.05551528930664, | |
| "learning_rate": 6.907216494845361e-06, | |
| "loss": 0.0505, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 6.214203894616266, | |
| "grad_norm": 0.19994887709617615, | |
| "learning_rate": 6.892898052691867e-06, | |
| "loss": 0.044, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 6.242840778923253, | |
| "grad_norm": 6.800403118133545, | |
| "learning_rate": 6.878579610538374e-06, | |
| "loss": 0.0212, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 6.2714776632302405, | |
| "grad_norm": 0.18680407106876373, | |
| "learning_rate": 6.86426116838488e-06, | |
| "loss": 0.04, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 6.300114547537228, | |
| "grad_norm": 2.269585132598877, | |
| "learning_rate": 6.849942726231387e-06, | |
| "loss": 0.0385, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 6.328751431844215, | |
| "grad_norm": 0.05255923420190811, | |
| "learning_rate": 6.835624284077893e-06, | |
| "loss": 0.0364, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 6.357388316151202, | |
| "grad_norm": 0.026086222380399704, | |
| "learning_rate": 6.821305841924399e-06, | |
| "loss": 0.0462, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 6.3860252004581906, | |
| "grad_norm": 0.026915445923805237, | |
| "learning_rate": 6.806987399770906e-06, | |
| "loss": 0.0473, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 6.414662084765178, | |
| "grad_norm": 30.492359161376953, | |
| "learning_rate": 6.792668957617412e-06, | |
| "loss": 0.0508, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 6.443298969072165, | |
| "grad_norm": 0.030431082472205162, | |
| "learning_rate": 6.778350515463919e-06, | |
| "loss": 0.0555, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 6.4719358533791524, | |
| "grad_norm": 0.19676260650157928, | |
| "learning_rate": 6.764032073310425e-06, | |
| "loss": 0.0435, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 6.50057273768614, | |
| "grad_norm": 0.3724329471588135, | |
| "learning_rate": 6.7497136311569314e-06, | |
| "loss": 0.0209, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 6.529209621993127, | |
| "grad_norm": 0.07832983881235123, | |
| "learning_rate": 6.735395189003437e-06, | |
| "loss": 0.0368, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 6.557846506300114, | |
| "grad_norm": 1.6286810636520386, | |
| "learning_rate": 6.721076746849944e-06, | |
| "loss": 0.0546, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 6.586483390607102, | |
| "grad_norm": 10.142606735229492, | |
| "learning_rate": 6.706758304696449e-06, | |
| "loss": 0.0537, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 6.615120274914089, | |
| "grad_norm": 0.03260861337184906, | |
| "learning_rate": 6.692439862542955e-06, | |
| "loss": 0.0801, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 6.643757159221077, | |
| "grad_norm": 0.06774479895830154, | |
| "learning_rate": 6.678121420389461e-06, | |
| "loss": 0.0276, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 6.672394043528064, | |
| "grad_norm": 0.10083146393299103, | |
| "learning_rate": 6.663802978235968e-06, | |
| "loss": 0.0355, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 6.701030927835052, | |
| "grad_norm": 0.5465721487998962, | |
| "learning_rate": 6.649484536082474e-06, | |
| "loss": 0.0531, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 6.729667812142039, | |
| "grad_norm": 0.36361464858055115, | |
| "learning_rate": 6.635166093928981e-06, | |
| "loss": 0.0534, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 6.758304696449026, | |
| "grad_norm": 0.3213222622871399, | |
| "learning_rate": 6.620847651775487e-06, | |
| "loss": 0.021, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 6.786941580756014, | |
| "grad_norm": 1.3060314655303955, | |
| "learning_rate": 6.606529209621994e-06, | |
| "loss": 0.0371, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 6.815578465063001, | |
| "grad_norm": 15.44947338104248, | |
| "learning_rate": 6.5922107674685e-06, | |
| "loss": 0.0475, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 6.844215349369988, | |
| "grad_norm": 0.1092597097158432, | |
| "learning_rate": 6.577892325315006e-06, | |
| "loss": 0.0549, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 6.872852233676976, | |
| "grad_norm": 0.24638314545154572, | |
| "learning_rate": 6.563573883161513e-06, | |
| "loss": 0.044, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 6.901489117983964, | |
| "grad_norm": 0.17163485288619995, | |
| "learning_rate": 6.549255441008019e-06, | |
| "loss": 0.0478, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 6.930126002290951, | |
| "grad_norm": 0.5006637573242188, | |
| "learning_rate": 6.5349369988545254e-06, | |
| "loss": 0.0595, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 6.958762886597938, | |
| "grad_norm": 0.09898879379034042, | |
| "learning_rate": 6.520618556701031e-06, | |
| "loss": 0.0415, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 6.987399770904926, | |
| "grad_norm": 0.8853304386138916, | |
| "learning_rate": 6.506300114547538e-06, | |
| "loss": 0.023, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.7718902668169861, | |
| "eval_f1": 0.772007503464447, | |
| "eval_loss": 1.2936097383499146, | |
| "eval_runtime": 36.2937, | |
| "eval_samples_per_second": 73.319, | |
| "eval_steps_per_second": 9.175, | |
| "step": 12222 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 34920, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.283848509009536e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |