| { | |
| "best_metric": 0.8704663212435233, | |
| "best_model_checkpoint": "./vit-base-skin/checkpoint-2504", | |
| "epoch": 4.0, | |
| "global_step": 2504, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00019920127795527157, | |
| "loss": 1.3493, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019840255591054313, | |
| "loss": 1.076, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019760383386581472, | |
| "loss": 0.8377, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019680511182108628, | |
| "loss": 1.0358, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00019600638977635784, | |
| "loss": 0.948, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0001952076677316294, | |
| "loss": 0.9304, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00019440894568690097, | |
| "loss": 0.9603, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00019361022364217253, | |
| "loss": 0.8858, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001928115015974441, | |
| "loss": 0.968, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00019201277955271565, | |
| "loss": 0.8077, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001912140575079872, | |
| "loss": 0.8553, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0001904153354632588, | |
| "loss": 0.8923, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00018961661341853036, | |
| "loss": 0.7868, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00018881789137380192, | |
| "loss": 0.6841, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00018801916932907348, | |
| "loss": 0.8815, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00018722044728434505, | |
| "loss": 0.8815, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00018642172523961663, | |
| "loss": 0.6795, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0001856230031948882, | |
| "loss": 0.8106, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00018482428115015976, | |
| "loss": 0.7214, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00018402555910543132, | |
| "loss": 0.7661, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00018322683706070288, | |
| "loss": 0.8526, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00018242811501597444, | |
| "loss": 0.7321, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000181629392971246, | |
| "loss": 0.7255, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00018083067092651756, | |
| "loss": 0.7053, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00018003194888178913, | |
| "loss": 0.612, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00017923322683706071, | |
| "loss": 0.7517, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00017843450479233228, | |
| "loss": 0.9721, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00017763578274760384, | |
| "loss": 0.7345, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0001768370607028754, | |
| "loss": 0.7104, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.000176038338658147, | |
| "loss": 0.636, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00017523961661341855, | |
| "loss": 0.7007, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0001744408945686901, | |
| "loss": 0.589, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00017364217252396167, | |
| "loss": 0.6807, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00017284345047923323, | |
| "loss": 0.8574, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00017204472843450482, | |
| "loss": 0.8194, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00017124600638977638, | |
| "loss": 0.7184, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00017044728434504792, | |
| "loss": 0.6792, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00016964856230031948, | |
| "loss": 0.6489, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00016884984025559104, | |
| "loss": 0.6107, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00016805111821086263, | |
| "loss": 0.819, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001672523961661342, | |
| "loss": 0.672, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00016645367412140575, | |
| "loss": 0.6564, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0001656549520766773, | |
| "loss": 0.7774, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0001648562300319489, | |
| "loss": 0.7113, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00016405750798722046, | |
| "loss": 0.5898, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00016325878594249202, | |
| "loss": 0.534, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00016246006389776359, | |
| "loss": 0.5966, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00016166134185303515, | |
| "loss": 0.6295, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00016086261980830674, | |
| "loss": 0.8287, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0001600638977635783, | |
| "loss": 0.6419, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00015926517571884986, | |
| "loss": 0.6292, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00015846645367412142, | |
| "loss": 0.6599, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00015766773162939298, | |
| "loss": 0.5267, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00015686900958466454, | |
| "loss": 0.7095, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0001560702875399361, | |
| "loss": 0.6641, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00015527156549520767, | |
| "loss": 0.6191, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00015447284345047923, | |
| "loss": 0.5588, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00015367412140575082, | |
| "loss": 0.4932, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00015287539936102238, | |
| "loss": 0.5364, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00015207667731629394, | |
| "loss": 0.5854, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0001512779552715655, | |
| "loss": 0.6835, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00015047923322683706, | |
| "loss": 0.5676, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8341968911917098, | |
| "eval_f1": 0.8241786506017417, | |
| "eval_loss": 0.508845329284668, | |
| "eval_precision": 0.8626257971303308, | |
| "eval_recall": 0.8341968911917098, | |
| "eval_runtime": 1.3717, | |
| "eval_samples_per_second": 140.7, | |
| "eval_steps_per_second": 9.477, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00014968051118210865, | |
| "loss": 0.5525, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0001488817891373802, | |
| "loss": 0.3613, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00014808306709265177, | |
| "loss": 0.5148, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00014728434504792333, | |
| "loss": 0.5467, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0001464856230031949, | |
| "loss": 0.6535, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00014568690095846646, | |
| "loss": 0.5865, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00014488817891373802, | |
| "loss": 0.4595, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00014408945686900958, | |
| "loss": 0.46, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00014329073482428114, | |
| "loss": 0.5835, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00014249201277955273, | |
| "loss": 0.4806, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001416932907348243, | |
| "loss": 0.4676, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00014089456869009585, | |
| "loss": 0.456, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0001400958466453674, | |
| "loss": 0.4707, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.000139297124600639, | |
| "loss": 0.4849, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013849840255591056, | |
| "loss": 0.3835, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00013769968051118212, | |
| "loss": 0.5962, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00013690095846645369, | |
| "loss": 0.5128, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00013610223642172525, | |
| "loss": 0.4967, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0001353035143769968, | |
| "loss": 0.4347, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00013450479233226837, | |
| "loss": 0.624, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00013370607028753993, | |
| "loss": 0.365, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.0001329073482428115, | |
| "loss": 0.5736, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00013210862619808305, | |
| "loss": 0.4688, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.00013130990415335464, | |
| "loss": 0.5979, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0001305111821086262, | |
| "loss": 0.6397, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00012971246006389777, | |
| "loss": 0.3582, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00012891373801916933, | |
| "loss": 0.5079, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00012811501597444092, | |
| "loss": 0.4426, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.00012731629392971248, | |
| "loss": 0.3925, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00012651757188498404, | |
| "loss": 0.5077, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.0001257188498402556, | |
| "loss": 0.4496, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00012492012779552716, | |
| "loss": 0.5264, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00012412140575079872, | |
| "loss": 0.5064, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00012332268370607028, | |
| "loss": 0.3804, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.00012252396166134185, | |
| "loss": 0.4211, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.00012172523961661342, | |
| "loss": 0.3981, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.000120926517571885, | |
| "loss": 0.4908, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00012012779552715656, | |
| "loss": 0.4697, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.00011932907348242812, | |
| "loss": 0.4234, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00011853035143769968, | |
| "loss": 0.414, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00011773162939297124, | |
| "loss": 0.5027, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00011693290734824283, | |
| "loss": 0.5786, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00011613418530351439, | |
| "loss": 0.489, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00011533546325878595, | |
| "loss": 0.4423, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011453674121405751, | |
| "loss": 0.4269, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011373801916932908, | |
| "loss": 0.4876, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00011293929712460065, | |
| "loss": 0.4422, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011214057507987221, | |
| "loss": 0.3992, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00011134185303514377, | |
| "loss": 0.6031, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00011054313099041533, | |
| "loss": 0.357, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00010974440894568691, | |
| "loss": 0.311, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00010894568690095847, | |
| "loss": 0.3321, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00010814696485623003, | |
| "loss": 0.5434, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.0001073482428115016, | |
| "loss": 0.3562, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00010654952076677316, | |
| "loss": 0.4395, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00010575079872204474, | |
| "loss": 0.557, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0001049520766773163, | |
| "loss": 0.2622, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00010415335463258787, | |
| "loss": 0.414, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00010335463258785943, | |
| "loss": 0.4796, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.000102555910543131, | |
| "loss": 0.2579, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.00010175718849840256, | |
| "loss": 0.3208, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00010095846645367413, | |
| "loss": 0.4819, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00010015974440894569, | |
| "loss": 0.4883, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.844559585492228, | |
| "eval_f1": 0.8469257868227422, | |
| "eval_loss": 0.4589127004146576, | |
| "eval_precision": 0.8511589179252687, | |
| "eval_recall": 0.844559585492228, | |
| "eval_runtime": 1.2587, | |
| "eval_samples_per_second": 153.331, | |
| "eval_steps_per_second": 10.328, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.936102236421726e-05, | |
| "loss": 0.4284, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.856230031948882e-05, | |
| "loss": 0.2118, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.77635782747604e-05, | |
| "loss": 0.3613, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 9.696485623003195e-05, | |
| "loss": 0.2461, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.616613418530351e-05, | |
| "loss": 0.2217, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.536741214057508e-05, | |
| "loss": 0.2119, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.456869009584664e-05, | |
| "loss": 0.3351, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.376996805111822e-05, | |
| "loss": 0.2057, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.297124600638978e-05, | |
| "loss": 0.2362, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.217252396166136e-05, | |
| "loss": 0.1849, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.137380191693292e-05, | |
| "loss": 0.2675, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.057507987220448e-05, | |
| "loss": 0.2027, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.977635782747604e-05, | |
| "loss": 0.3192, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.89776357827476e-05, | |
| "loss": 0.3101, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.817891373801918e-05, | |
| "loss": 0.2286, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.738019169329074e-05, | |
| "loss": 0.2725, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.658146964856231e-05, | |
| "loss": 0.323, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.578274760383387e-05, | |
| "loss": 0.1575, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.498402555910544e-05, | |
| "loss": 0.1961, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 8.4185303514377e-05, | |
| "loss": 0.26, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 8.338658146964856e-05, | |
| "loss": 0.2546, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 8.258785942492013e-05, | |
| "loss": 0.1552, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 8.17891373801917e-05, | |
| "loss": 0.3564, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 8.099041533546327e-05, | |
| "loss": 0.2165, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 8.019169329073483e-05, | |
| "loss": 0.2325, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.939297124600639e-05, | |
| "loss": 0.1835, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.859424920127795e-05, | |
| "loss": 0.2369, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.779552715654951e-05, | |
| "loss": 0.2223, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.699680511182109e-05, | |
| "loss": 0.2646, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 7.619808306709265e-05, | |
| "loss": 0.1929, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.539936102236423e-05, | |
| "loss": 0.2013, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 7.460063897763579e-05, | |
| "loss": 0.2541, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.380191693290735e-05, | |
| "loss": 0.2709, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.300319488817891e-05, | |
| "loss": 0.2275, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.220447284345049e-05, | |
| "loss": 0.2351, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.140575079872205e-05, | |
| "loss": 0.1267, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.060702875399361e-05, | |
| "loss": 0.2509, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.980830670926518e-05, | |
| "loss": 0.3611, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.900958466453674e-05, | |
| "loss": 0.248, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.821086261980832e-05, | |
| "loss": 0.2878, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.741214057507987e-05, | |
| "loss": 0.1851, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 6.661341853035144e-05, | |
| "loss": 0.3701, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 6.5814696485623e-05, | |
| "loss": 0.236, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 6.501597444089457e-05, | |
| "loss": 0.3649, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 6.421725239616614e-05, | |
| "loss": 0.1385, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 6.34185303514377e-05, | |
| "loss": 0.2905, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 6.261980830670928e-05, | |
| "loss": 0.2468, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 6.182108626198084e-05, | |
| "loss": 0.1864, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 6.1022364217252406e-05, | |
| "loss": 0.1971, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 6.022364217252396e-05, | |
| "loss": 0.2589, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 5.942492012779552e-05, | |
| "loss": 0.1774, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 5.86261980830671e-05, | |
| "loss": 0.19, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 5.782747603833866e-05, | |
| "loss": 0.2237, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 5.702875399361023e-05, | |
| "loss": 0.2003, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 5.623003194888179e-05, | |
| "loss": 0.1557, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 5.543130990415336e-05, | |
| "loss": 0.2719, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 5.4632587859424925e-05, | |
| "loss": 0.2797, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.383386581469649e-05, | |
| "loss": 0.1199, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 5.3035143769968054e-05, | |
| "loss": 0.1957, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 5.2236421725239616e-05, | |
| "loss": 0.1413, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.1437699680511184e-05, | |
| "loss": 0.2592, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 5.0638977635782745e-05, | |
| "loss": 0.3582, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8601036269430051, | |
| "eval_f1": 0.8594105048984151, | |
| "eval_loss": 0.38329267501831055, | |
| "eval_precision": 0.8739840495783806, | |
| "eval_recall": 0.8601036269430051, | |
| "eval_runtime": 1.2365, | |
| "eval_samples_per_second": 156.088, | |
| "eval_steps_per_second": 10.514, | |
| "step": 1878 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.984025559105431e-05, | |
| "loss": 0.1211, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 4.904153354632588e-05, | |
| "loss": 0.0816, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 4.824281150159744e-05, | |
| "loss": 0.159, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.744408945686901e-05, | |
| "loss": 0.1156, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 4.664536741214058e-05, | |
| "loss": 0.0798, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.584664536741215e-05, | |
| "loss": 0.0656, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.504792332268371e-05, | |
| "loss": 0.04, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.424920127795527e-05, | |
| "loss": 0.0895, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.345047923322684e-05, | |
| "loss": 0.1021, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.265175718849841e-05, | |
| "loss": 0.0627, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 4.185303514376997e-05, | |
| "loss": 0.0462, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 4.1054313099041536e-05, | |
| "loss": 0.0793, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 4.0255591054313104e-05, | |
| "loss": 0.0458, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.9456869009584666e-05, | |
| "loss": 0.0512, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.8658146964856234e-05, | |
| "loss": 0.0359, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.7859424920127795e-05, | |
| "loss": 0.1464, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.7060702875399364e-05, | |
| "loss": 0.1052, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.6261980830670925e-05, | |
| "loss": 0.065, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.546325878594249e-05, | |
| "loss": 0.078, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 3.466453674121406e-05, | |
| "loss": 0.0754, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.386581469648562e-05, | |
| "loss": 0.0373, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.306709265175719e-05, | |
| "loss": 0.1033, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.226837060702875e-05, | |
| "loss": 0.0428, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.146964856230032e-05, | |
| "loss": 0.1005, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.067092651757188e-05, | |
| "loss": 0.0795, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 2.987220447284345e-05, | |
| "loss": 0.0703, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 2.907348242811502e-05, | |
| "loss": 0.0338, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 2.8274760383386583e-05, | |
| "loss": 0.1082, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 2.747603833865815e-05, | |
| "loss": 0.0696, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 2.6757188498402557e-05, | |
| "loss": 0.1096, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 2.5958466453674125e-05, | |
| "loss": 0.057, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 2.515974440894569e-05, | |
| "loss": 0.0606, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 2.4361022364217255e-05, | |
| "loss": 0.0742, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 2.356230031948882e-05, | |
| "loss": 0.0591, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 2.2763578274760385e-05, | |
| "loss": 0.0453, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.196485623003195e-05, | |
| "loss": 0.062, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 2.1166134185303514e-05, | |
| "loss": 0.1079, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.0367412140575082e-05, | |
| "loss": 0.059, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 1.9568690095846644e-05, | |
| "loss": 0.104, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.8769968051118212e-05, | |
| "loss": 0.0692, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 1.7971246006389777e-05, | |
| "loss": 0.049, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.7172523961661345e-05, | |
| "loss": 0.0299, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.6373801916932906e-05, | |
| "loss": 0.0638, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 1.5575079872204475e-05, | |
| "loss": 0.0954, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.477635782747604e-05, | |
| "loss": 0.1079, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 1.3977635782747606e-05, | |
| "loss": 0.0959, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.3178913738019169e-05, | |
| "loss": 0.0368, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.2380191693290735e-05, | |
| "loss": 0.0453, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.1581469648562302e-05, | |
| "loss": 0.0836, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.0782747603833867e-05, | |
| "loss": 0.0483, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 9.984025559105432e-06, | |
| "loss": 0.0689, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 9.185303514376996e-06, | |
| "loss": 0.0868, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 8.386581469648563e-06, | |
| "loss": 0.0973, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.5878594249201285e-06, | |
| "loss": 0.045, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.789137380191693e-06, | |
| "loss": 0.1082, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.990415335463259e-06, | |
| "loss": 0.0771, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 5.191693290734825e-06, | |
| "loss": 0.015, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 4.39297124600639e-06, | |
| "loss": 0.0441, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 3.5942492012779555e-06, | |
| "loss": 0.057, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 2.7955271565495207e-06, | |
| "loss": 0.0646, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 1.9968051118210863e-06, | |
| "loss": 0.0454, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.1980830670926517e-06, | |
| "loss": 0.0451, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 3.9936102236421723e-07, | |
| "loss": 0.0953, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8704663212435233, | |
| "eval_f1": 0.8724324567519218, | |
| "eval_loss": 0.4405668079853058, | |
| "eval_precision": 0.8770634374261316, | |
| "eval_recall": 0.8704663212435233, | |
| "eval_runtime": 1.2968, | |
| "eval_samples_per_second": 148.824, | |
| "eval_steps_per_second": 10.024, | |
| "step": 2504 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 2504, | |
| "total_flos": 3.104468219559813e+18, | |
| "train_loss": 0.3803590432987712, | |
| "train_runtime": 848.4129, | |
| "train_samples_per_second": 47.218, | |
| "train_steps_per_second": 2.951 | |
| } | |
| ], | |
| "max_steps": 2504, | |
| "num_train_epochs": 4, | |
| "total_flos": 3.104468219559813e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |