| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.34325, | |
| "eval_steps": 500, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 10.3125, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.2492, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 11.0, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.3168, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 8.9375, | |
| "learning_rate": 3e-06, | |
| "loss": 1.3578, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 10.1875, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.2434, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 9.5625, | |
| "learning_rate": 5e-06, | |
| "loss": 1.2461, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 6e-06, | |
| "loss": 1.1131, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 11.6875, | |
| "learning_rate": 7e-06, | |
| "loss": 1.1562, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 8.6875, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.2025, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 6.875, | |
| "learning_rate": 9e-06, | |
| "loss": 1.1477, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0277, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 8.25, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 1.0428, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.78125, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.9688, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 6.71875, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 0.9908, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 8.125, | |
| "learning_rate": 1.4e-05, | |
| "loss": 1.0992, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.9598, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 9.25, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.9885, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 13.875, | |
| "learning_rate": 1.7e-05, | |
| "loss": 1.2307, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 9.3125, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.9137, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 9.25, | |
| "learning_rate": 1.9e-05, | |
| "loss": 1.5107, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 96.0, | |
| "learning_rate": 2e-05, | |
| "loss": 0.9955, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 26.25, | |
| "learning_rate": 1.9998476951563914e-05, | |
| "loss": 1.6652, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 1.999390827019096e-05, | |
| "loss": 0.86, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 56.0, | |
| "learning_rate": 1.9986295347545738e-05, | |
| "loss": 1.9418, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 11.625, | |
| "learning_rate": 1.9975640502598243e-05, | |
| "loss": 0.8643, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 52.0, | |
| "learning_rate": 1.9961946980917457e-05, | |
| "loss": 1.3936, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 28.125, | |
| "learning_rate": 1.9945218953682736e-05, | |
| "loss": 0.8059, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 12.125, | |
| "learning_rate": 1.9925461516413224e-05, | |
| "loss": 0.8197, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 30.5, | |
| "learning_rate": 1.9902680687415704e-05, | |
| "loss": 1.3477, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 24.5, | |
| "learning_rate": 1.9876883405951378e-05, | |
| "loss": 1.0397, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 7.15625, | |
| "learning_rate": 1.9848077530122083e-05, | |
| "loss": 0.8006, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 87.0, | |
| "learning_rate": 1.9816271834476642e-05, | |
| "loss": 1.0712, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 6.375, | |
| "learning_rate": 1.9781476007338058e-05, | |
| "loss": 0.6074, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 74.0, | |
| "learning_rate": 1.9743700647852356e-05, | |
| "loss": 0.8439, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.9702957262759964e-05, | |
| "loss": 0.6888, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 59.25, | |
| "learning_rate": 1.9659258262890683e-05, | |
| "loss": 0.7408, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 24.0, | |
| "learning_rate": 1.961261695938319e-05, | |
| "loss": 0.6303, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.9563047559630356e-05, | |
| "loss": 2.4628, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 20.125, | |
| "learning_rate": 1.9510565162951538e-05, | |
| "loss": 0.7479, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 9.0, | |
| "learning_rate": 1.945518575599317e-05, | |
| "loss": 0.6927, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 11.5625, | |
| "learning_rate": 1.9396926207859085e-05, | |
| "loss": 0.5542, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 6.9375, | |
| "learning_rate": 1.9335804264972018e-05, | |
| "loss": 1.097, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 118.5, | |
| "learning_rate": 1.9271838545667876e-05, | |
| "loss": 1.0457, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 9.0, | |
| "learning_rate": 1.9205048534524405e-05, | |
| "loss": 0.9044, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 8.3125, | |
| "learning_rate": 1.913545457642601e-05, | |
| "loss": 1.1971, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 10.625, | |
| "learning_rate": 1.9063077870366504e-05, | |
| "loss": 0.9662, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 68.5, | |
| "learning_rate": 1.8987940462991673e-05, | |
| "loss": 0.5489, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 21.625, | |
| "learning_rate": 1.891006524188368e-05, | |
| "loss": 0.3805, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 22.125, | |
| "learning_rate": 1.8829475928589272e-05, | |
| "loss": 0.4361, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 8.6875, | |
| "learning_rate": 1.874619707139396e-05, | |
| "loss": 0.4573, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 85.0, | |
| "learning_rate": 1.866025403784439e-05, | |
| "loss": 1.7953, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 1.8571673007021124e-05, | |
| "loss": 0.319, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 5.875, | |
| "learning_rate": 1.848048096156426e-05, | |
| "loss": 0.778, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 85.0, | |
| "learning_rate": 1.8386705679454243e-05, | |
| "loss": 1.6186, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 7.875, | |
| "learning_rate": 1.8290375725550417e-05, | |
| "loss": 1.8343, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 1.819152044288992e-05, | |
| "loss": 0.4746, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 1.8090169943749477e-05, | |
| "loss": 1.074, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.798635510047293e-05, | |
| "loss": 0.2415, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 14.25, | |
| "learning_rate": 1.788010753606722e-05, | |
| "loss": 0.2946, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 12.125, | |
| "learning_rate": 1.777145961456971e-05, | |
| "loss": 0.3634, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 5.3125, | |
| "learning_rate": 1.766044443118978e-05, | |
| "loss": 0.7839, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 1.7547095802227723e-05, | |
| "loss": 0.502, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 91.0, | |
| "learning_rate": 1.7431448254773943e-05, | |
| "loss": 0.3595, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 26.875, | |
| "learning_rate": 1.7313537016191706e-05, | |
| "loss": 0.5093, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.7193398003386514e-05, | |
| "loss": 0.2864, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 6.0, | |
| "learning_rate": 1.7071067811865477e-05, | |
| "loss": 0.4504, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 1.6946583704589973e-05, | |
| "loss": 0.2695, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 1.6819983600624986e-05, | |
| "loss": 0.479, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 106.0, | |
| "learning_rate": 1.6691306063588583e-05, | |
| "loss": 0.7245, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 5.90625, | |
| "learning_rate": 1.6560590289905074e-05, | |
| "loss": 0.4877, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.6427876096865394e-05, | |
| "loss": 0.571, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 11.8125, | |
| "learning_rate": 1.6293203910498375e-05, | |
| "loss": 0.2269, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 6.75, | |
| "learning_rate": 1.6156614753256583e-05, | |
| "loss": 0.575, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 3.875, | |
| "learning_rate": 1.6018150231520486e-05, | |
| "loss": 0.1943, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 4.15625, | |
| "learning_rate": 1.5877852522924733e-05, | |
| "loss": 0.2041, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 3.515625, | |
| "learning_rate": 1.573576436351046e-05, | |
| "loss": 0.9804, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 3.296875, | |
| "learning_rate": 1.5591929034707468e-05, | |
| "loss": 1.1265, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 1.5446390350150272e-05, | |
| "loss": 0.752, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.529919264233205e-05, | |
| "loss": 0.1821, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 1.5150380749100545e-05, | |
| "loss": 0.1494, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 3.96875, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.2066, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 6.0, | |
| "learning_rate": 1.4848096202463373e-05, | |
| "loss": 0.1645, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 11.4375, | |
| "learning_rate": 1.469471562785891e-05, | |
| "loss": 0.1673, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 9.625, | |
| "learning_rate": 1.4539904997395468e-05, | |
| "loss": 0.7342, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 2.375, | |
| "learning_rate": 1.4383711467890776e-05, | |
| "loss": 0.4579, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.4226182617406996e-05, | |
| "loss": 0.3541, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 2.078125, | |
| "learning_rate": 1.4067366430758004e-05, | |
| "loss": 0.1596, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 3.21875, | |
| "learning_rate": 1.3907311284892737e-05, | |
| "loss": 0.1419, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 3.8125, | |
| "learning_rate": 1.3746065934159123e-05, | |
| "loss": 0.1184, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 37.75, | |
| "learning_rate": 1.3583679495453e-05, | |
| "loss": 0.4206, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 67.0, | |
| "learning_rate": 1.342020143325669e-05, | |
| "loss": 1.6792, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 4.375, | |
| "learning_rate": 1.3255681544571568e-05, | |
| "loss": 0.4549, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 10.625, | |
| "learning_rate": 1.3090169943749475e-05, | |
| "loss": 0.1843, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 3.796875, | |
| "learning_rate": 1.2923717047227368e-05, | |
| "loss": 0.1655, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 2.296875, | |
| "learning_rate": 1.2756373558169992e-05, | |
| "loss": 0.1113, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 2.84375, | |
| "learning_rate": 1.2588190451025209e-05, | |
| "loss": 0.1282, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.8203125, | |
| "learning_rate": 1.2419218955996677e-05, | |
| "loss": 0.1195, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 1.6015625, | |
| "learning_rate": 1.2249510543438652e-05, | |
| "loss": 0.1383, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 2.875, | |
| "learning_rate": 1.2079116908177592e-05, | |
| "loss": 0.1135, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 2.359375, | |
| "learning_rate": 1.190808995376545e-05, | |
| "loss": 0.1054, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.671875, | |
| "learning_rate": 1.1736481776669307e-05, | |
| "loss": 0.0947, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 1.9140625, | |
| "learning_rate": 1.156434465040231e-05, | |
| "loss": 0.0942, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.703125, | |
| "learning_rate": 1.1391731009600655e-05, | |
| "loss": 0.088, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 2.671875, | |
| "learning_rate": 1.1218693434051475e-05, | |
| "loss": 0.0831, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 76.5, | |
| "learning_rate": 1.1045284632676535e-05, | |
| "loss": 0.9448, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 2.0625, | |
| "learning_rate": 1.0871557427476585e-05, | |
| "loss": 0.0911, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 4.0, | |
| "learning_rate": 1.0697564737441254e-05, | |
| "loss": 0.0968, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 2.796875, | |
| "learning_rate": 1.0523359562429441e-05, | |
| "loss": 0.1019, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 2.125, | |
| "learning_rate": 1.0348994967025012e-05, | |
| "loss": 0.266, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 1.46875, | |
| "learning_rate": 1.0174524064372837e-05, | |
| "loss": 0.095, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 1.2109375, | |
| "learning_rate": 1e-05, | |
| "loss": 0.1129, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 2.25, | |
| "learning_rate": 9.825475935627165e-06, | |
| "loss": 0.0981, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 2.234375, | |
| "learning_rate": 9.651005032974994e-06, | |
| "loss": 0.0866, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 1.703125, | |
| "learning_rate": 9.476640437570562e-06, | |
| "loss": 0.0988, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 1.5703125, | |
| "learning_rate": 9.302435262558748e-06, | |
| "loss": 0.0764, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 1.59375, | |
| "learning_rate": 9.128442572523418e-06, | |
| "loss": 0.0844, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.703125, | |
| "learning_rate": 8.954715367323468e-06, | |
| "loss": 0.0742, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 2.0, | |
| "learning_rate": 8.781306565948528e-06, | |
| "loss": 0.0713, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 1.3359375, | |
| "learning_rate": 8.60826899039935e-06, | |
| "loss": 0.0711, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 1.53125, | |
| "learning_rate": 8.43565534959769e-06, | |
| "loss": 0.0688, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 1.4296875, | |
| "learning_rate": 8.263518223330698e-06, | |
| "loss": 0.0661, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 2.8125, | |
| "learning_rate": 8.091910046234552e-06, | |
| "loss": 0.0674, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 1.828125, | |
| "learning_rate": 7.92088309182241e-06, | |
| "loss": 0.0646, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 1.5390625, | |
| "learning_rate": 7.750489456561351e-06, | |
| "loss": 0.0703, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.7109375, | |
| "learning_rate": 7.580781044003324e-06, | |
| "loss": 0.0661, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 1.3984375, | |
| "learning_rate": 7.411809548974792e-06, | |
| "loss": 0.066, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 1.59375, | |
| "learning_rate": 7.243626441830009e-06, | |
| "loss": 0.0674, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 2.109375, | |
| "learning_rate": 7.076282952772634e-06, | |
| "loss": 0.0687, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.25, | |
| "learning_rate": 6.909830056250527e-06, | |
| "loss": 0.0682, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 1.53125, | |
| "learning_rate": 6.744318455428436e-06, | |
| "loss": 0.0629, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 6.579798566743314e-06, | |
| "loss": 0.0633, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 6.4163205045469975e-06, | |
| "loss": 0.1746, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.00325, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 6.25393406584088e-06, | |
| "loss": 0.0615, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.00825, | |
| "grad_norm": 1.984375, | |
| "learning_rate": 6.092688715107265e-06, | |
| "loss": 0.054, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.01325, | |
| "grad_norm": 1.15625, | |
| "learning_rate": 5.932633569242e-06, | |
| "loss": 0.0515, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.01825, | |
| "grad_norm": 1.3515625, | |
| "learning_rate": 5.773817382593008e-06, | |
| "loss": 0.0585, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.02325, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 5.616288532109225e-06, | |
| "loss": 0.0638, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.02825, | |
| "grad_norm": 0.9453125, | |
| "learning_rate": 5.460095002604533e-06, | |
| "loss": 0.0591, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.03325, | |
| "grad_norm": 1.15625, | |
| "learning_rate": 5.305284372141095e-06, | |
| "loss": 0.056, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.03825, | |
| "grad_norm": 61.25, | |
| "learning_rate": 5.151903797536631e-06, | |
| "loss": 0.075, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.04325, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.0569, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.04825, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 4.849619250899458e-06, | |
| "loss": 0.0593, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.05325, | |
| "grad_norm": 1.3515625, | |
| "learning_rate": 4.700807357667953e-06, | |
| "loss": 0.0586, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.05825, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 4.5536096498497295e-06, | |
| "loss": 0.0553, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.06325, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 4.408070965292534e-06, | |
| "loss": 0.0591, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.06825, | |
| "grad_norm": 0.75, | |
| "learning_rate": 4.264235636489542e-06, | |
| "loss": 0.0518, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.07325, | |
| "grad_norm": 1.171875, | |
| "learning_rate": 4.12214747707527e-06, | |
| "loss": 0.0552, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.07825, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 3.981849768479516e-06, | |
| "loss": 0.0503, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.08325, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 3.8433852467434175e-06, | |
| "loss": 0.0493, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.08825, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 3.7067960895016277e-06, | |
| "loss": 0.0511, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.09325, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 3.5721239031346067e-06, | |
| "loss": 0.052, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.09825, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 3.4394097100949286e-06, | |
| "loss": 0.0508, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.10325, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 3.308693936411421e-06, | |
| "loss": 0.0518, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.10825, | |
| "grad_norm": 1.0, | |
| "learning_rate": 3.1800163993750166e-06, | |
| "loss": 0.054, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.11325, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 3.0534162954100264e-06, | |
| "loss": 0.0524, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.11825, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 2.9289321881345257e-06, | |
| "loss": 0.0544, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.12325, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 2.8066019966134907e-06, | |
| "loss": 0.0589, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.12825, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 2.6864629838082957e-06, | |
| "loss": 0.0529, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.13325, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 2.5685517452260566e-06, | |
| "loss": 0.0531, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.13825, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 2.45290419777228e-06, | |
| "loss": 0.0553, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.14325, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 2.339555568810221e-06, | |
| "loss": 0.0542, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.14825, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 2.2285403854302912e-06, | |
| "loss": 0.0528, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.1532499999999999, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 2.119892463932781e-06, | |
| "loss": 0.0495, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.15825, | |
| "grad_norm": 1.2734375, | |
| "learning_rate": 2.013644899527074e-06, | |
| "loss": 0.0529, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.1632500000000001, | |
| "grad_norm": 2.09375, | |
| "learning_rate": 1.9098300562505266e-06, | |
| "loss": 0.0835, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.16825, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 1.808479557110081e-06, | |
| "loss": 0.0514, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.17325, | |
| "grad_norm": 1.125, | |
| "learning_rate": 1.709624274449584e-06, | |
| "loss": 0.0484, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.17825, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 1.6132943205457607e-06, | |
| "loss": 0.0496, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.18325, | |
| "grad_norm": 0.9453125, | |
| "learning_rate": 1.5195190384357405e-06, | |
| "loss": 0.0501, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.18825, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 1.4283269929788779e-06, | |
| "loss": 0.0515, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.19325, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 1.339745962155613e-06, | |
| "loss": 0.0528, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.19825, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 1.2538029286060428e-06, | |
| "loss": 0.0499, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.20325, | |
| "grad_norm": 2.21875, | |
| "learning_rate": 1.1705240714107301e-06, | |
| "loss": 0.0546, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.20825, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 1.0899347581163222e-06, | |
| "loss": 0.0541, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.21325, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 1.012059537008332e-06, | |
| "loss": 0.0523, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.21825, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 9.369221296335007e-07, | |
| "loss": 0.0492, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.22325, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 8.645454235739903e-07, | |
| "loss": 0.0544, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.22825, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 7.949514654755963e-07, | |
| "loss": 0.0542, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.23325, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 7.281614543321269e-07, | |
| "loss": 0.0522, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.23825, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 6.641957350279838e-07, | |
| "loss": 0.0492, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.24325, | |
| "grad_norm": 0.921875, | |
| "learning_rate": 6.030737921409169e-07, | |
| "loss": 0.05, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.24825, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 5.448142440068316e-07, | |
| "loss": 0.0484, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.25325, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 4.894348370484648e-07, | |
| "loss": 0.0481, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.2582499999999999, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 4.3695244036964567e-07, | |
| "loss": 0.0492, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.26325, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 3.8738304061681107e-07, | |
| "loss": 0.0492, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.26825, | |
| "grad_norm": 1.7109375, | |
| "learning_rate": 3.4074173710931804e-07, | |
| "loss": 0.0501, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.27325, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 2.970427372400353e-07, | |
| "loss": 0.0488, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.2782499999999999, | |
| "grad_norm": 0.95703125, | |
| "learning_rate": 2.5629935214764866e-07, | |
| "loss": 0.0499, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.28325, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 2.1852399266194312e-07, | |
| "loss": 0.0503, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.2882500000000001, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 1.8372816552336025e-07, | |
| "loss": 0.0515, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.29325, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 1.519224698779198e-07, | |
| "loss": 0.0546, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.29825, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 1.231165940486234e-07, | |
| "loss": 0.054, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.30325, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 9.731931258429638e-08, | |
| "loss": 0.0537, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.3082500000000001, | |
| "grad_norm": 0.875, | |
| "learning_rate": 7.453848358678018e-08, | |
| "loss": 0.0515, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.31325, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 5.4781046317267103e-08, | |
| "loss": 0.0538, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.31825, | |
| "grad_norm": 0.98046875, | |
| "learning_rate": 3.805301908254455e-08, | |
| "loss": 0.0497, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.32325, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 2.4359497401758026e-08, | |
| "loss": 0.0531, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.32825, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 1.370465245426167e-08, | |
| "loss": 0.0506, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.33325, | |
| "grad_norm": 2.0625, | |
| "learning_rate": 6.091729809042379e-09, | |
| "loss": 0.0499, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.33825, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.5230484360873043e-09, | |
| "loss": 0.0466, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.34325, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.0, | |
| "loss": 0.0484, | |
| "step": 2000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.9130140524439142e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |