{ "best_metric": 0.97, "best_model_checkpoint": "./vit-LungCancerAUG10000img/checkpoint-3768", "epoch": 24.0, "global_step": 3768, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.06, "learning_rate": 4.2462845010615706e-08, "loss": 1.4586, "step": 10 }, { "epoch": 0.13, "learning_rate": 8.492569002123141e-08, "loss": 1.4488, "step": 20 }, { "epoch": 0.19, "learning_rate": 1.2738853503184713e-07, "loss": 1.4723, "step": 30 }, { "epoch": 0.25, "learning_rate": 1.6985138004246283e-07, "loss": 1.4157, "step": 40 }, { "epoch": 0.32, "learning_rate": 2.1231422505307855e-07, "loss": 1.4137, "step": 50 }, { "epoch": 0.38, "learning_rate": 2.5477707006369425e-07, "loss": 1.3989, "step": 60 }, { "epoch": 0.45, "learning_rate": 2.9723991507431e-07, "loss": 1.4097, "step": 70 }, { "epoch": 0.51, "learning_rate": 3.3970276008492565e-07, "loss": 1.4044, "step": 80 }, { "epoch": 0.57, "learning_rate": 3.8216560509554143e-07, "loss": 1.4031, "step": 90 }, { "epoch": 0.64, "learning_rate": 4.246284501061571e-07, "loss": 1.3575, "step": 100 }, { "epoch": 0.7, "learning_rate": 4.6709129511677283e-07, "loss": 1.3662, "step": 110 }, { "epoch": 0.76, "learning_rate": 5.095541401273885e-07, "loss": 1.3338, "step": 120 }, { "epoch": 0.83, "learning_rate": 5.520169851380042e-07, "loss": 1.3164, "step": 130 }, { "epoch": 0.89, "learning_rate": 5.9447983014862e-07, "loss": 1.3017, "step": 140 }, { "epoch": 0.96, "learning_rate": 6.369426751592356e-07, "loss": 1.2903, "step": 150 }, { "epoch": 1.0, "eval_accuracy": 0.445, "eval_loss": 1.260408878326416, "eval_runtime": 129.0129, "eval_samples_per_second": 3.1, "eval_steps_per_second": 0.388, "step": 157 }, { "epoch": 1.02, "learning_rate": 6.794055201698513e-07, "loss": 1.2839, "step": 160 }, { "epoch": 1.08, "learning_rate": 7.21868365180467e-07, "loss": 1.226, "step": 170 }, { "epoch": 1.15, "learning_rate": 7.643312101910829e-07, "loss": 1.2273, "step": 180 }, { "epoch": 1.21, "learning_rate": 8.067940552016985e-07, "loss": 1.2059, "step": 190 }, { "epoch": 1.27, "learning_rate": 8.492569002123142e-07, "loss": 1.1669, "step": 200 }, { "epoch": 1.34, "learning_rate": 8.917197452229299e-07, "loss": 1.1302, "step": 210 }, { "epoch": 1.4, "learning_rate": 9.341825902335457e-07, "loss": 1.1348, "step": 220 }, { "epoch": 1.46, "learning_rate": 9.766454352441613e-07, "loss": 1.0836, "step": 230 }, { "epoch": 1.53, "learning_rate": 1.019108280254777e-06, "loss": 1.0167, "step": 240 }, { "epoch": 1.59, "learning_rate": 1.0615711252653927e-06, "loss": 1.0234, "step": 250 }, { "epoch": 1.66, "learning_rate": 1.1040339702760084e-06, "loss": 1.0272, "step": 260 }, { "epoch": 1.72, "learning_rate": 1.146496815286624e-06, "loss": 0.9718, "step": 270 }, { "epoch": 1.78, "learning_rate": 1.18895966029724e-06, "loss": 0.9686, "step": 280 }, { "epoch": 1.85, "learning_rate": 1.2314225053078556e-06, "loss": 0.9369, "step": 290 }, { "epoch": 1.91, "learning_rate": 1.2738853503184713e-06, "loss": 0.9313, "step": 300 }, { "epoch": 1.97, "learning_rate": 1.316348195329087e-06, "loss": 0.9537, "step": 310 }, { "epoch": 2.0, "eval_accuracy": 0.52, "eval_loss": 0.9655842781066895, "eval_runtime": 5.4121, "eval_samples_per_second": 73.908, "eval_steps_per_second": 9.239, "step": 314 }, { "epoch": 2.04, "learning_rate": 1.3588110403397026e-06, "loss": 0.9027, "step": 320 }, { "epoch": 2.1, "learning_rate": 1.4012738853503183e-06, "loss": 0.8836, "step": 330 }, { "epoch": 2.17, "learning_rate": 1.4394904458598724e-06, "loss": 0.8358, "step": 340 }, { "epoch": 2.23, "learning_rate": 1.481953290870488e-06, "loss": 0.8609, "step": 350 }, { "epoch": 2.29, "learning_rate": 1.5244161358811042e-06, "loss": 0.8037, "step": 360 }, { "epoch": 2.36, "learning_rate": 1.5668789808917198e-06, "loss": 0.7825, "step": 370 }, { "epoch": 2.42, "learning_rate": 1.6093418259023355e-06, "loss": 0.8254, "step": 380 }, { "epoch": 2.48, "learning_rate": 1.6518046709129512e-06, "loss": 0.79, "step": 390 }, { "epoch": 2.55, "learning_rate": 1.6942675159235668e-06, "loss": 0.8032, "step": 400 }, { "epoch": 2.61, "learning_rate": 1.7367303609341825e-06, "loss": 0.7757, "step": 410 }, { "epoch": 2.68, "learning_rate": 1.7791932059447982e-06, "loss": 0.7849, "step": 420 }, { "epoch": 2.74, "learning_rate": 1.8216560509554139e-06, "loss": 0.7449, "step": 430 }, { "epoch": 2.8, "learning_rate": 1.8641188959660298e-06, "loss": 0.7586, "step": 440 }, { "epoch": 2.87, "learning_rate": 1.9065817409766454e-06, "loss": 0.7547, "step": 450 }, { "epoch": 2.93, "learning_rate": 1.949044585987261e-06, "loss": 0.7729, "step": 460 }, { "epoch": 2.99, "learning_rate": 1.987261146496815e-06, "loss": 0.7124, "step": 470 }, { "epoch": 3.0, "eval_accuracy": 0.575, "eval_loss": 0.8280940055847168, "eval_runtime": 5.3205, "eval_samples_per_second": 75.18, "eval_steps_per_second": 9.398, "step": 471 }, { "epoch": 3.06, "learning_rate": 1.996697334276952e-06, "loss": 0.7028, "step": 480 }, { "epoch": 3.12, "learning_rate": 1.9919792403868834e-06, "loss": 0.6914, "step": 490 }, { "epoch": 3.18, "learning_rate": 1.987261146496815e-06, "loss": 0.7107, "step": 500 }, { "epoch": 3.25, "learning_rate": 1.9825430526067466e-06, "loss": 0.7197, "step": 510 }, { "epoch": 3.31, "learning_rate": 1.9778249587166784e-06, "loss": 0.6729, "step": 520 }, { "epoch": 3.38, "learning_rate": 1.97310686482661e-06, "loss": 0.699, "step": 530 }, { "epoch": 3.44, "learning_rate": 1.9683887709365417e-06, "loss": 0.6529, "step": 540 }, { "epoch": 3.5, "learning_rate": 1.963670677046473e-06, "loss": 0.6093, "step": 550 }, { "epoch": 3.57, "learning_rate": 1.9589525831564045e-06, "loss": 0.6709, "step": 560 }, { "epoch": 3.63, "learning_rate": 1.9542344892663363e-06, "loss": 0.6327, "step": 570 }, { "epoch": 3.69, "learning_rate": 1.9495163953762677e-06, "loss": 0.6713, "step": 580 }, { "epoch": 3.76, "learning_rate": 1.9447983014861995e-06, "loss": 0.6506, "step": 590 }, { "epoch": 3.82, "learning_rate": 1.940080207596131e-06, "loss": 0.6314, "step": 600 }, { "epoch": 3.89, "learning_rate": 1.9353621137060628e-06, "loss": 0.6114, "step": 610 }, { "epoch": 3.95, "learning_rate": 1.930644019815994e-06, "loss": 0.6114, "step": 620 }, { "epoch": 4.0, "eval_accuracy": 0.64, "eval_loss": 0.753457248210907, "eval_runtime": 5.5017, "eval_samples_per_second": 72.704, "eval_steps_per_second": 9.088, "step": 628 }, { "epoch": 4.01, "learning_rate": 1.9259259259259256e-06, "loss": 0.6592, "step": 630 }, { "epoch": 4.08, "learning_rate": 1.9212078320358574e-06, "loss": 0.5994, "step": 640 }, { "epoch": 4.14, "learning_rate": 1.916489738145789e-06, "loss": 0.6178, "step": 650 }, { "epoch": 4.2, "learning_rate": 1.9117716442557206e-06, "loss": 0.5675, "step": 660 }, { "epoch": 4.27, "learning_rate": 1.9070535503656522e-06, "loss": 0.6167, "step": 670 }, { "epoch": 4.33, "learning_rate": 1.9023354564755839e-06, "loss": 0.5665, "step": 680 }, { "epoch": 4.39, "learning_rate": 1.8976173625855155e-06, "loss": 0.6061, "step": 690 }, { "epoch": 4.46, "learning_rate": 1.892899268695447e-06, "loss": 0.5977, "step": 700 }, { "epoch": 4.52, "learning_rate": 1.8881811748053787e-06, "loss": 0.5113, "step": 710 }, { "epoch": 4.59, "learning_rate": 1.8834630809153101e-06, "loss": 0.5353, "step": 720 }, { "epoch": 4.65, "learning_rate": 1.8787449870252417e-06, "loss": 0.5925, "step": 730 }, { "epoch": 4.71, "learning_rate": 1.8740268931351733e-06, "loss": 0.5639, "step": 740 }, { "epoch": 4.78, "learning_rate": 1.869308799245105e-06, "loss": 0.5562, "step": 750 }, { "epoch": 4.84, "learning_rate": 1.8645907053550366e-06, "loss": 0.544, "step": 760 }, { "epoch": 4.9, "learning_rate": 1.8598726114649682e-06, "loss": 0.5563, "step": 770 }, { "epoch": 4.97, "learning_rate": 1.8551545175748998e-06, "loss": 0.517, "step": 780 }, { "epoch": 5.0, "eval_accuracy": 0.725, "eval_loss": 0.6396925449371338, "eval_runtime": 5.3831, "eval_samples_per_second": 74.306, "eval_steps_per_second": 9.288, "step": 785 }, { "epoch": 5.03, "learning_rate": 1.8504364236848314e-06, "loss": 0.5165, "step": 790 }, { "epoch": 5.1, "learning_rate": 1.845718329794763e-06, "loss": 0.5175, "step": 800 }, { "epoch": 5.16, "learning_rate": 1.8410002359046944e-06, "loss": 0.5361, "step": 810 }, { "epoch": 5.22, "learning_rate": 1.8362821420146258e-06, "loss": 0.475, "step": 820 }, { "epoch": 5.29, "learning_rate": 1.8315640481245575e-06, "loss": 0.4947, "step": 830 }, { "epoch": 5.35, "learning_rate": 1.826845954234489e-06, "loss": 0.504, "step": 840 }, { "epoch": 5.41, "learning_rate": 1.8221278603444207e-06, "loss": 0.5223, "step": 850 }, { "epoch": 5.48, "learning_rate": 1.8174097664543523e-06, "loss": 0.4913, "step": 860 }, { "epoch": 5.54, "learning_rate": 1.812691672564284e-06, "loss": 0.4583, "step": 870 }, { "epoch": 5.61, "learning_rate": 1.8079735786742155e-06, "loss": 0.4675, "step": 880 }, { "epoch": 5.67, "learning_rate": 1.803255484784147e-06, "loss": 0.4712, "step": 890 }, { "epoch": 5.73, "learning_rate": 1.7985373908940786e-06, "loss": 0.5068, "step": 900 }, { "epoch": 5.8, "learning_rate": 1.7938192970040102e-06, "loss": 0.5049, "step": 910 }, { "epoch": 5.86, "learning_rate": 1.7891012031139418e-06, "loss": 0.4855, "step": 920 }, { "epoch": 5.92, "learning_rate": 1.7843831092238734e-06, "loss": 0.4233, "step": 930 }, { "epoch": 5.99, "learning_rate": 1.779665015333805e-06, "loss": 0.4773, "step": 940 }, { "epoch": 6.0, "eval_accuracy": 0.78, "eval_loss": 0.5554624795913696, "eval_runtime": 5.952, "eval_samples_per_second": 67.204, "eval_steps_per_second": 8.4, "step": 942 }, { "epoch": 6.05, "learning_rate": 1.7749469214437366e-06, "loss": 0.455, "step": 950 }, { "epoch": 6.11, "learning_rate": 1.7702288275536682e-06, "loss": 0.4656, "step": 960 }, { "epoch": 6.18, "learning_rate": 1.7655107336635999e-06, "loss": 0.4254, "step": 970 }, { "epoch": 6.24, "learning_rate": 1.7607926397735313e-06, "loss": 0.4177, "step": 980 }, { "epoch": 6.31, "learning_rate": 1.7560745458834629e-06, "loss": 0.4008, "step": 990 }, { "epoch": 6.37, "learning_rate": 1.7513564519933945e-06, "loss": 0.449, "step": 1000 }, { "epoch": 6.43, "learning_rate": 1.7466383581033261e-06, "loss": 0.4398, "step": 1010 }, { "epoch": 6.5, "learning_rate": 1.7419202642132577e-06, "loss": 0.4012, "step": 1020 }, { "epoch": 6.56, "learning_rate": 1.7372021703231893e-06, "loss": 0.3977, "step": 1030 }, { "epoch": 6.62, "learning_rate": 1.732484076433121e-06, "loss": 0.4268, "step": 1040 }, { "epoch": 6.69, "learning_rate": 1.7277659825430526e-06, "loss": 0.4132, "step": 1050 }, { "epoch": 6.75, "learning_rate": 1.7230478886529842e-06, "loss": 0.3898, "step": 1060 }, { "epoch": 6.82, "learning_rate": 1.7183297947629156e-06, "loss": 0.4145, "step": 1070 }, { "epoch": 6.88, "learning_rate": 1.7136117008728472e-06, "loss": 0.3537, "step": 1080 }, { "epoch": 6.94, "learning_rate": 1.7088936069827788e-06, "loss": 0.402, "step": 1090 }, { "epoch": 7.0, "eval_accuracy": 0.845, "eval_loss": 0.4491625130176544, "eval_runtime": 5.5591, "eval_samples_per_second": 71.954, "eval_steps_per_second": 8.994, "step": 1099 }, { "epoch": 7.01, "learning_rate": 1.7041755130927104e-06, "loss": 0.376, "step": 1100 }, { "epoch": 7.07, "learning_rate": 1.699457419202642e-06, "loss": 0.3519, "step": 1110 }, { "epoch": 7.13, "learning_rate": 1.6947393253125737e-06, "loss": 0.3645, "step": 1120 }, { "epoch": 7.2, "learning_rate": 1.6900212314225053e-06, "loss": 0.3703, "step": 1130 }, { "epoch": 7.26, "learning_rate": 1.685303137532437e-06, "loss": 0.3766, "step": 1140 }, { "epoch": 7.32, "learning_rate": 1.6805850436423683e-06, "loss": 0.3608, "step": 1150 }, { "epoch": 7.39, "learning_rate": 1.6758669497523e-06, "loss": 0.3515, "step": 1160 }, { "epoch": 7.45, "learning_rate": 1.6711488558622315e-06, "loss": 0.3624, "step": 1170 }, { "epoch": 7.52, "learning_rate": 1.6664307619721631e-06, "loss": 0.3434, "step": 1180 }, { "epoch": 7.58, "learning_rate": 1.6617126680820948e-06, "loss": 0.3247, "step": 1190 }, { "epoch": 7.64, "learning_rate": 1.6569945741920264e-06, "loss": 0.3103, "step": 1200 }, { "epoch": 7.71, "learning_rate": 1.652276480301958e-06, "loss": 0.3281, "step": 1210 }, { "epoch": 7.77, "learning_rate": 1.6475583864118896e-06, "loss": 0.3692, "step": 1220 }, { "epoch": 7.83, "learning_rate": 1.6428402925218212e-06, "loss": 0.3082, "step": 1230 }, { "epoch": 7.9, "learning_rate": 1.6381221986317526e-06, "loss": 0.3278, "step": 1240 }, { "epoch": 7.96, "learning_rate": 1.6334041047416842e-06, "loss": 0.3504, "step": 1250 }, { "epoch": 8.0, "eval_accuracy": 0.8775, "eval_loss": 0.37387552857398987, "eval_runtime": 6.4049, "eval_samples_per_second": 62.452, "eval_steps_per_second": 7.807, "step": 1256 }, { "epoch": 8.03, "learning_rate": 1.6286860108516159e-06, "loss": 0.3603, "step": 1260 }, { "epoch": 8.09, "learning_rate": 1.6239679169615475e-06, "loss": 0.2868, "step": 1270 }, { "epoch": 8.15, "learning_rate": 1.619249823071479e-06, "loss": 0.292, "step": 1280 }, { "epoch": 8.22, "learning_rate": 1.6145317291814107e-06, "loss": 0.321, "step": 1290 }, { "epoch": 8.28, "learning_rate": 1.6098136352913423e-06, "loss": 0.3054, "step": 1300 }, { "epoch": 8.34, "learning_rate": 1.605095541401274e-06, "loss": 0.2833, "step": 1310 }, { "epoch": 8.41, "learning_rate": 1.6003774475112053e-06, "loss": 0.3144, "step": 1320 }, { "epoch": 8.47, "learning_rate": 1.595659353621137e-06, "loss": 0.2801, "step": 1330 }, { "epoch": 8.54, "learning_rate": 1.5909412597310686e-06, "loss": 0.2938, "step": 1340 }, { "epoch": 8.6, "learning_rate": 1.5862231658410002e-06, "loss": 0.2794, "step": 1350 }, { "epoch": 8.66, "learning_rate": 1.5815050719509318e-06, "loss": 0.2566, "step": 1360 }, { "epoch": 8.73, "learning_rate": 1.5767869780608634e-06, "loss": 0.2926, "step": 1370 }, { "epoch": 8.79, "learning_rate": 1.572068884170795e-06, "loss": 0.2733, "step": 1380 }, { "epoch": 8.85, "learning_rate": 1.5673507902807266e-06, "loss": 0.2815, "step": 1390 }, { "epoch": 8.92, "learning_rate": 1.5626326963906583e-06, "loss": 0.2973, "step": 1400 }, { "epoch": 8.98, "learning_rate": 1.5579146025005897e-06, "loss": 0.2584, "step": 1410 }, { "epoch": 9.0, "eval_accuracy": 0.89, "eval_loss": 0.3347581624984741, "eval_runtime": 6.0288, "eval_samples_per_second": 66.348, "eval_steps_per_second": 8.294, "step": 1413 }, { "epoch": 9.04, "learning_rate": 1.5531965086105213e-06, "loss": 0.2353, "step": 1420 }, { "epoch": 9.11, "learning_rate": 1.548478414720453e-06, "loss": 0.2659, "step": 1430 }, { "epoch": 9.17, "learning_rate": 1.5437603208303845e-06, "loss": 0.2395, "step": 1440 }, { "epoch": 9.24, "learning_rate": 1.5390422269403161e-06, "loss": 0.2494, "step": 1450 }, { "epoch": 9.3, "learning_rate": 1.5343241330502477e-06, "loss": 0.2551, "step": 1460 }, { "epoch": 9.36, "learning_rate": 1.5296060391601794e-06, "loss": 0.2346, "step": 1470 }, { "epoch": 9.43, "learning_rate": 1.524887945270111e-06, "loss": 0.2311, "step": 1480 }, { "epoch": 9.49, "learning_rate": 1.5201698513800426e-06, "loss": 0.237, "step": 1490 }, { "epoch": 9.55, "learning_rate": 1.515451757489974e-06, "loss": 0.2366, "step": 1500 }, { "epoch": 9.62, "learning_rate": 1.5107336635999056e-06, "loss": 0.242, "step": 1510 }, { "epoch": 9.68, "learning_rate": 1.5060155697098372e-06, "loss": 0.2635, "step": 1520 }, { "epoch": 9.75, "learning_rate": 1.5012974758197688e-06, "loss": 0.2243, "step": 1530 }, { "epoch": 9.81, "learning_rate": 1.4965793819297002e-06, "loss": 0.236, "step": 1540 }, { "epoch": 9.87, "learning_rate": 1.4918612880396319e-06, "loss": 0.2312, "step": 1550 }, { "epoch": 9.94, "learning_rate": 1.4871431941495635e-06, "loss": 0.2357, "step": 1560 }, { "epoch": 10.0, "learning_rate": 1.482425100259495e-06, "loss": 0.2439, "step": 1570 }, { "epoch": 10.0, "eval_accuracy": 0.9, "eval_loss": 0.29536351561546326, "eval_runtime": 6.1912, "eval_samples_per_second": 64.608, "eval_steps_per_second": 8.076, "step": 1570 }, { "epoch": 10.06, "learning_rate": 1.4777070063694265e-06, "loss": 0.2089, "step": 1580 }, { "epoch": 10.13, "learning_rate": 1.4729889124793581e-06, "loss": 0.198, "step": 1590 }, { "epoch": 10.19, "learning_rate": 1.4682708185892897e-06, "loss": 0.2017, "step": 1600 }, { "epoch": 10.25, "learning_rate": 1.4635527246992213e-06, "loss": 0.2091, "step": 1610 }, { "epoch": 10.32, "learning_rate": 1.458834630809153e-06, "loss": 0.2016, "step": 1620 }, { "epoch": 10.38, "learning_rate": 1.4541165369190846e-06, "loss": 0.2036, "step": 1630 }, { "epoch": 10.45, "learning_rate": 1.4493984430290162e-06, "loss": 0.2159, "step": 1640 }, { "epoch": 10.51, "learning_rate": 1.4446803491389478e-06, "loss": 0.2042, "step": 1650 }, { "epoch": 10.57, "learning_rate": 1.4399622552488794e-06, "loss": 0.1891, "step": 1660 }, { "epoch": 10.64, "learning_rate": 1.4352441613588108e-06, "loss": 0.196, "step": 1670 }, { "epoch": 10.7, "learning_rate": 1.4305260674687424e-06, "loss": 0.1919, "step": 1680 }, { "epoch": 10.76, "learning_rate": 1.425807973578674e-06, "loss": 0.1908, "step": 1690 }, { "epoch": 10.83, "learning_rate": 1.4210898796886057e-06, "loss": 0.2065, "step": 1700 }, { "epoch": 10.89, "learning_rate": 1.4163717857985373e-06, "loss": 0.1718, "step": 1710 }, { "epoch": 10.96, "learning_rate": 1.411653691908469e-06, "loss": 0.1925, "step": 1720 }, { "epoch": 11.0, "eval_accuracy": 0.9275, "eval_loss": 0.24589473009109497, "eval_runtime": 9.3749, "eval_samples_per_second": 42.667, "eval_steps_per_second": 5.333, "step": 1727 }, { "epoch": 11.02, "learning_rate": 1.4069355980184005e-06, "loss": 0.2006, "step": 1730 }, { "epoch": 11.08, "learning_rate": 1.4022175041283321e-06, "loss": 0.1676, "step": 1740 }, { "epoch": 11.15, "learning_rate": 1.3974994102382637e-06, "loss": 0.1749, "step": 1750 }, { "epoch": 11.21, "learning_rate": 1.3927813163481951e-06, "loss": 0.1616, "step": 1760 }, { "epoch": 11.27, "learning_rate": 1.3880632224581268e-06, "loss": 0.1564, "step": 1770 }, { "epoch": 11.34, "learning_rate": 1.3833451285680584e-06, "loss": 0.1709, "step": 1780 }, { "epoch": 11.4, "learning_rate": 1.37862703467799e-06, "loss": 0.1612, "step": 1790 }, { "epoch": 11.46, "learning_rate": 1.3739089407879216e-06, "loss": 0.1634, "step": 1800 }, { "epoch": 11.53, "learning_rate": 1.3691908468978532e-06, "loss": 0.1604, "step": 1810 }, { "epoch": 11.59, "learning_rate": 1.3644727530077848e-06, "loss": 0.167, "step": 1820 }, { "epoch": 11.66, "learning_rate": 1.3597546591177165e-06, "loss": 0.1548, "step": 1830 }, { "epoch": 11.72, "learning_rate": 1.3550365652276479e-06, "loss": 0.1718, "step": 1840 }, { "epoch": 11.78, "learning_rate": 1.3503184713375795e-06, "loss": 0.1443, "step": 1850 }, { "epoch": 11.85, "learning_rate": 1.345600377447511e-06, "loss": 0.1606, "step": 1860 }, { "epoch": 11.91, "learning_rate": 1.3408822835574427e-06, "loss": 0.1703, "step": 1870 }, { "epoch": 11.97, "learning_rate": 1.3361641896673743e-06, "loss": 0.1693, "step": 1880 }, { "epoch": 12.0, "eval_accuracy": 0.9325, "eval_loss": 0.21917079389095306, "eval_runtime": 5.5809, "eval_samples_per_second": 71.673, "eval_steps_per_second": 8.959, "step": 1884 }, { "epoch": 12.04, "learning_rate": 1.331446095777306e-06, "loss": 0.1501, "step": 1890 }, { "epoch": 12.1, "learning_rate": 1.3267280018872375e-06, "loss": 0.1339, "step": 1900 }, { "epoch": 12.17, "learning_rate": 1.3220099079971692e-06, "loss": 0.1506, "step": 1910 }, { "epoch": 12.23, "learning_rate": 1.3172918141071008e-06, "loss": 0.1347, "step": 1920 }, { "epoch": 12.29, "learning_rate": 1.3125737202170322e-06, "loss": 0.1378, "step": 1930 }, { "epoch": 12.36, "learning_rate": 1.3078556263269638e-06, "loss": 0.1367, "step": 1940 }, { "epoch": 12.42, "learning_rate": 1.3031375324368954e-06, "loss": 0.1236, "step": 1950 }, { "epoch": 12.48, "learning_rate": 1.298419438546827e-06, "loss": 0.1332, "step": 1960 }, { "epoch": 12.55, "learning_rate": 1.2937013446567586e-06, "loss": 0.1424, "step": 1970 }, { "epoch": 12.61, "learning_rate": 1.2889832507666903e-06, "loss": 0.1426, "step": 1980 }, { "epoch": 12.68, "learning_rate": 1.2842651568766219e-06, "loss": 0.1364, "step": 1990 }, { "epoch": 12.74, "learning_rate": 1.2795470629865535e-06, "loss": 0.1488, "step": 2000 }, { "epoch": 12.8, "learning_rate": 1.274828969096485e-06, "loss": 0.1308, "step": 2010 }, { "epoch": 12.87, "learning_rate": 1.2701108752064165e-06, "loss": 0.1278, "step": 2020 }, { "epoch": 12.93, "learning_rate": 1.2653927813163481e-06, "loss": 0.1134, "step": 2030 }, { "epoch": 12.99, "learning_rate": 1.2606746874262797e-06, "loss": 0.1157, "step": 2040 }, { "epoch": 13.0, "eval_accuracy": 0.9375, "eval_loss": 0.22450672090053558, "eval_runtime": 5.5269, "eval_samples_per_second": 72.373, "eval_steps_per_second": 9.047, "step": 2041 }, { "epoch": 13.06, "learning_rate": 1.2559565935362114e-06, "loss": 0.1235, "step": 2050 }, { "epoch": 13.12, "learning_rate": 1.251238499646143e-06, "loss": 0.117, "step": 2060 }, { "epoch": 13.18, "learning_rate": 1.2465204057560746e-06, "loss": 0.1192, "step": 2070 }, { "epoch": 13.25, "learning_rate": 1.2418023118660062e-06, "loss": 0.1031, "step": 2080 }, { "epoch": 13.31, "learning_rate": 1.2370842179759378e-06, "loss": 0.1111, "step": 2090 }, { "epoch": 13.38, "learning_rate": 1.2323661240858692e-06, "loss": 0.0991, "step": 2100 }, { "epoch": 13.44, "learning_rate": 1.2276480301958008e-06, "loss": 0.1041, "step": 2110 }, { "epoch": 13.5, "learning_rate": 1.2229299363057325e-06, "loss": 0.1251, "step": 2120 }, { "epoch": 13.57, "learning_rate": 1.218211842415664e-06, "loss": 0.0937, "step": 2130 }, { "epoch": 13.63, "learning_rate": 1.2134937485255957e-06, "loss": 0.1133, "step": 2140 }, { "epoch": 13.69, "learning_rate": 1.2087756546355273e-06, "loss": 0.0928, "step": 2150 }, { "epoch": 13.76, "learning_rate": 1.204057560745459e-06, "loss": 0.1286, "step": 2160 }, { "epoch": 13.82, "learning_rate": 1.1993394668553905e-06, "loss": 0.102, "step": 2170 }, { "epoch": 13.89, "learning_rate": 1.1946213729653221e-06, "loss": 0.1028, "step": 2180 }, { "epoch": 13.95, "learning_rate": 1.1899032790752535e-06, "loss": 0.1014, "step": 2190 }, { "epoch": 14.0, "eval_accuracy": 0.9475, "eval_loss": 0.18889574706554413, "eval_runtime": 5.4539, "eval_samples_per_second": 73.342, "eval_steps_per_second": 9.168, "step": 2198 }, { "epoch": 14.01, "learning_rate": 1.1851851851851852e-06, "loss": 0.112, "step": 2200 }, { "epoch": 14.08, "learning_rate": 1.1804670912951168e-06, "loss": 0.0879, "step": 2210 }, { "epoch": 14.14, "learning_rate": 1.1757489974050484e-06, "loss": 0.089, "step": 2220 }, { "epoch": 14.2, "learning_rate": 1.17103090351498e-06, "loss": 0.0844, "step": 2230 }, { "epoch": 14.27, "learning_rate": 1.1663128096249116e-06, "loss": 0.0949, "step": 2240 }, { "epoch": 14.33, "learning_rate": 1.161594715734843e-06, "loss": 0.1087, "step": 2250 }, { "epoch": 14.39, "learning_rate": 1.1568766218447746e-06, "loss": 0.0824, "step": 2260 }, { "epoch": 14.46, "learning_rate": 1.152158527954706e-06, "loss": 0.0885, "step": 2270 }, { "epoch": 14.52, "learning_rate": 1.1474404340646377e-06, "loss": 0.0886, "step": 2280 }, { "epoch": 14.59, "learning_rate": 1.1427223401745693e-06, "loss": 0.1076, "step": 2290 }, { "epoch": 14.65, "learning_rate": 1.1380042462845009e-06, "loss": 0.0756, "step": 2300 }, { "epoch": 14.71, "learning_rate": 1.1332861523944325e-06, "loss": 0.0855, "step": 2310 }, { "epoch": 14.78, "learning_rate": 1.1285680585043641e-06, "loss": 0.0811, "step": 2320 }, { "epoch": 14.84, "learning_rate": 1.1238499646142957e-06, "loss": 0.0884, "step": 2330 }, { "epoch": 14.9, "learning_rate": 1.1191318707242274e-06, "loss": 0.0899, "step": 2340 }, { "epoch": 14.97, "learning_rate": 1.114413776834159e-06, "loss": 0.075, "step": 2350 }, { "epoch": 15.0, "eval_accuracy": 0.9525, "eval_loss": 0.17460297048091888, "eval_runtime": 5.5354, "eval_samples_per_second": 72.262, "eval_steps_per_second": 9.033, "step": 2355 }, { "epoch": 15.03, "learning_rate": 1.1096956829440904e-06, "loss": 0.0766, "step": 2360 }, { "epoch": 15.1, "learning_rate": 1.104977589054022e-06, "loss": 0.0702, "step": 2370 }, { "epoch": 15.16, "learning_rate": 1.1002594951639536e-06, "loss": 0.0888, "step": 2380 }, { "epoch": 15.22, "learning_rate": 1.0955414012738852e-06, "loss": 0.0716, "step": 2390 }, { "epoch": 15.29, "learning_rate": 1.0908233073838168e-06, "loss": 0.0687, "step": 2400 }, { "epoch": 15.35, "learning_rate": 1.0861052134937485e-06, "loss": 0.073, "step": 2410 }, { "epoch": 15.41, "learning_rate": 1.08138711960368e-06, "loss": 0.0695, "step": 2420 }, { "epoch": 15.48, "learning_rate": 1.0766690257136117e-06, "loss": 0.0738, "step": 2430 }, { "epoch": 15.54, "learning_rate": 1.0719509318235433e-06, "loss": 0.0836, "step": 2440 }, { "epoch": 15.61, "learning_rate": 1.0672328379334747e-06, "loss": 0.0622, "step": 2450 }, { "epoch": 15.67, "learning_rate": 1.0625147440434063e-06, "loss": 0.0706, "step": 2460 }, { "epoch": 15.73, "learning_rate": 1.057796650153338e-06, "loss": 0.0688, "step": 2470 }, { "epoch": 15.8, "learning_rate": 1.0530785562632695e-06, "loss": 0.0674, "step": 2480 }, { "epoch": 15.86, "learning_rate": 1.0483604623732012e-06, "loss": 0.0794, "step": 2490 }, { "epoch": 15.92, "learning_rate": 1.0436423684831328e-06, "loss": 0.0604, "step": 2500 }, { "epoch": 15.99, "learning_rate": 1.0389242745930644e-06, "loss": 0.0581, "step": 2510 }, { "epoch": 16.0, "eval_accuracy": 0.95, "eval_loss": 0.17460142076015472, "eval_runtime": 5.4732, "eval_samples_per_second": 73.083, "eval_steps_per_second": 9.135, "step": 2512 }, { "epoch": 16.05, "learning_rate": 1.034206180702996e-06, "loss": 0.0688, "step": 2520 }, { "epoch": 16.11, "learning_rate": 1.0294880868129274e-06, "loss": 0.0559, "step": 2530 }, { "epoch": 16.18, "learning_rate": 1.024769992922859e-06, "loss": 0.0561, "step": 2540 }, { "epoch": 16.24, "learning_rate": 1.0200518990327906e-06, "loss": 0.0558, "step": 2550 }, { "epoch": 16.31, "learning_rate": 1.0153338051427223e-06, "loss": 0.0602, "step": 2560 }, { "epoch": 16.37, "learning_rate": 1.0106157112526539e-06, "loss": 0.0605, "step": 2570 }, { "epoch": 16.43, "learning_rate": 1.0058976173625855e-06, "loss": 0.0516, "step": 2580 }, { "epoch": 16.5, "learning_rate": 1.001179523472517e-06, "loss": 0.0566, "step": 2590 }, { "epoch": 16.56, "learning_rate": 9.964614295824485e-07, "loss": 0.0624, "step": 2600 }, { "epoch": 16.62, "learning_rate": 9.92215145081387e-07, "loss": 0.0522, "step": 2610 }, { "epoch": 16.69, "learning_rate": 9.879688605803254e-07, "loss": 0.0563, "step": 2620 }, { "epoch": 16.75, "learning_rate": 9.83250766690257e-07, "loss": 0.0583, "step": 2630 }, { "epoch": 16.82, "learning_rate": 9.785326728001886e-07, "loss": 0.0587, "step": 2640 }, { "epoch": 16.88, "learning_rate": 9.738145789101202e-07, "loss": 0.0552, "step": 2650 }, { "epoch": 16.94, "learning_rate": 9.690964850200518e-07, "loss": 0.0612, "step": 2660 }, { "epoch": 17.0, "eval_accuracy": 0.9575, "eval_loss": 0.14898547530174255, "eval_runtime": 5.3718, "eval_samples_per_second": 74.462, "eval_steps_per_second": 9.308, "step": 2669 }, { "epoch": 17.01, "learning_rate": 9.643783911299834e-07, "loss": 0.0464, "step": 2670 }, { "epoch": 17.07, "learning_rate": 9.59660297239915e-07, "loss": 0.049, "step": 2680 }, { "epoch": 17.13, "learning_rate": 9.549422033498467e-07, "loss": 0.0484, "step": 2690 }, { "epoch": 17.2, "learning_rate": 9.502241094597782e-07, "loss": 0.0419, "step": 2700 }, { "epoch": 17.26, "learning_rate": 9.455060155697098e-07, "loss": 0.0422, "step": 2710 }, { "epoch": 17.32, "learning_rate": 9.407879216796414e-07, "loss": 0.0476, "step": 2720 }, { "epoch": 17.39, "learning_rate": 9.36069827789573e-07, "loss": 0.054, "step": 2730 }, { "epoch": 17.45, "learning_rate": 9.313517338995047e-07, "loss": 0.0452, "step": 2740 }, { "epoch": 17.52, "learning_rate": 9.266336400094362e-07, "loss": 0.0474, "step": 2750 }, { "epoch": 17.58, "learning_rate": 9.219155461193678e-07, "loss": 0.0406, "step": 2760 }, { "epoch": 17.64, "learning_rate": 9.171974522292994e-07, "loss": 0.0482, "step": 2770 }, { "epoch": 17.71, "learning_rate": 9.124793583392309e-07, "loss": 0.0407, "step": 2780 }, { "epoch": 17.77, "learning_rate": 9.077612644491624e-07, "loss": 0.043, "step": 2790 }, { "epoch": 17.83, "learning_rate": 9.03043170559094e-07, "loss": 0.0468, "step": 2800 }, { "epoch": 17.9, "learning_rate": 8.983250766690256e-07, "loss": 0.0381, "step": 2810 }, { "epoch": 17.96, "learning_rate": 8.936069827789573e-07, "loss": 0.0515, "step": 2820 }, { "epoch": 18.0, "eval_accuracy": 0.955, "eval_loss": 0.1454763561487198, "eval_runtime": 5.4502, "eval_samples_per_second": 73.392, "eval_steps_per_second": 9.174, "step": 2826 }, { "epoch": 18.03, "learning_rate": 8.888888888888888e-07, "loss": 0.0448, "step": 2830 }, { "epoch": 18.09, "learning_rate": 8.841707949988204e-07, "loss": 0.0376, "step": 2840 }, { "epoch": 18.15, "learning_rate": 8.79452701108752e-07, "loss": 0.0367, "step": 2850 }, { "epoch": 18.22, "learning_rate": 8.747346072186836e-07, "loss": 0.0326, "step": 2860 }, { "epoch": 18.28, "learning_rate": 8.700165133286152e-07, "loss": 0.0378, "step": 2870 }, { "epoch": 18.34, "learning_rate": 8.652984194385467e-07, "loss": 0.0427, "step": 2880 }, { "epoch": 18.41, "learning_rate": 8.605803255484784e-07, "loss": 0.0388, "step": 2890 }, { "epoch": 18.47, "learning_rate": 8.5586223165841e-07, "loss": 0.0342, "step": 2900 }, { "epoch": 18.54, "learning_rate": 8.511441377683416e-07, "loss": 0.0373, "step": 2910 }, { "epoch": 18.6, "learning_rate": 8.464260438782731e-07, "loss": 0.032, "step": 2920 }, { "epoch": 18.66, "learning_rate": 8.417079499882047e-07, "loss": 0.0336, "step": 2930 }, { "epoch": 18.73, "learning_rate": 8.369898560981363e-07, "loss": 0.0411, "step": 2940 }, { "epoch": 18.79, "learning_rate": 8.322717622080679e-07, "loss": 0.0412, "step": 2950 }, { "epoch": 18.85, "learning_rate": 8.275536683179994e-07, "loss": 0.036, "step": 2960 }, { "epoch": 18.92, "learning_rate": 8.228355744279311e-07, "loss": 0.0319, "step": 2970 }, { "epoch": 18.98, "learning_rate": 8.181174805378627e-07, "loss": 0.0339, "step": 2980 }, { "epoch": 19.0, "eval_accuracy": 0.965, "eval_loss": 0.13269954919815063, "eval_runtime": 5.908, "eval_samples_per_second": 67.705, "eval_steps_per_second": 8.463, "step": 2983 }, { "epoch": 19.04, "learning_rate": 8.133993866477943e-07, "loss": 0.0311, "step": 2990 }, { "epoch": 19.11, "learning_rate": 8.086812927577259e-07, "loss": 0.0308, "step": 3000 }, { "epoch": 19.17, "learning_rate": 8.039631988676574e-07, "loss": 0.0285, "step": 3010 }, { "epoch": 19.24, "learning_rate": 7.99245104977589e-07, "loss": 0.0307, "step": 3020 }, { "epoch": 19.3, "learning_rate": 7.945270110875207e-07, "loss": 0.0332, "step": 3030 }, { "epoch": 19.36, "learning_rate": 7.898089171974523e-07, "loss": 0.0334, "step": 3040 }, { "epoch": 19.43, "learning_rate": 7.850908233073838e-07, "loss": 0.0337, "step": 3050 }, { "epoch": 19.49, "learning_rate": 7.803727294173154e-07, "loss": 0.0254, "step": 3060 }, { "epoch": 19.55, "learning_rate": 7.75654635527247e-07, "loss": 0.0272, "step": 3070 }, { "epoch": 19.62, "learning_rate": 7.709365416371786e-07, "loss": 0.0256, "step": 3080 }, { "epoch": 19.68, "learning_rate": 7.662184477471101e-07, "loss": 0.0334, "step": 3090 }, { "epoch": 19.75, "learning_rate": 7.615003538570417e-07, "loss": 0.0302, "step": 3100 }, { "epoch": 19.81, "learning_rate": 7.567822599669734e-07, "loss": 0.0243, "step": 3110 }, { "epoch": 19.87, "learning_rate": 7.52064166076905e-07, "loss": 0.0284, "step": 3120 }, { "epoch": 19.94, "learning_rate": 7.473460721868364e-07, "loss": 0.0291, "step": 3130 }, { "epoch": 20.0, "learning_rate": 7.42627978296768e-07, "loss": 0.0251, "step": 3140 }, { "epoch": 20.0, "eval_accuracy": 0.955, "eval_loss": 0.13997198641300201, "eval_runtime": 6.1227, "eval_samples_per_second": 65.33, "eval_steps_per_second": 8.166, "step": 3140 }, { "epoch": 20.06, "learning_rate": 7.379098844066996e-07, "loss": 0.026, "step": 3150 }, { "epoch": 20.13, "learning_rate": 7.331917905166312e-07, "loss": 0.0243, "step": 3160 }, { "epoch": 20.19, "learning_rate": 7.284736966265628e-07, "loss": 0.0299, "step": 3170 }, { "epoch": 20.25, "learning_rate": 7.237556027364944e-07, "loss": 0.0253, "step": 3180 }, { "epoch": 20.32, "learning_rate": 7.19037508846426e-07, "loss": 0.0266, "step": 3190 }, { "epoch": 20.38, "learning_rate": 7.143194149563576e-07, "loss": 0.0227, "step": 3200 }, { "epoch": 20.45, "learning_rate": 7.096013210662892e-07, "loss": 0.0227, "step": 3210 }, { "epoch": 20.51, "learning_rate": 7.048832271762207e-07, "loss": 0.0234, "step": 3220 }, { "epoch": 20.57, "learning_rate": 7.001651332861523e-07, "loss": 0.0238, "step": 3230 }, { "epoch": 20.64, "learning_rate": 6.954470393960839e-07, "loss": 0.0278, "step": 3240 }, { "epoch": 20.7, "learning_rate": 6.907289455060156e-07, "loss": 0.0211, "step": 3250 }, { "epoch": 20.76, "learning_rate": 6.860108516159471e-07, "loss": 0.0196, "step": 3260 }, { "epoch": 20.83, "learning_rate": 6.812927577258787e-07, "loss": 0.019, "step": 3270 }, { "epoch": 20.89, "learning_rate": 6.765746638358103e-07, "loss": 0.0213, "step": 3280 }, { "epoch": 20.96, "learning_rate": 6.718565699457419e-07, "loss": 0.0216, "step": 3290 }, { "epoch": 21.0, "eval_accuracy": 0.96, "eval_loss": 0.13987915217876434, "eval_runtime": 6.4007, "eval_samples_per_second": 62.494, "eval_steps_per_second": 7.812, "step": 3297 }, { "epoch": 21.02, "learning_rate": 6.671384760556735e-07, "loss": 0.023, "step": 3300 }, { "epoch": 21.08, "learning_rate": 6.62420382165605e-07, "loss": 0.022, "step": 3310 }, { "epoch": 21.15, "learning_rate": 6.577022882755366e-07, "loss": 0.02, "step": 3320 }, { "epoch": 21.21, "learning_rate": 6.529841943854683e-07, "loss": 0.019, "step": 3330 }, { "epoch": 21.27, "learning_rate": 6.482661004953999e-07, "loss": 0.0188, "step": 3340 }, { "epoch": 21.34, "learning_rate": 6.435480066053314e-07, "loss": 0.0186, "step": 3350 }, { "epoch": 21.4, "learning_rate": 6.38829912715263e-07, "loss": 0.0182, "step": 3360 }, { "epoch": 21.46, "learning_rate": 6.341118188251946e-07, "loss": 0.0182, "step": 3370 }, { "epoch": 21.53, "learning_rate": 6.293937249351262e-07, "loss": 0.0233, "step": 3380 }, { "epoch": 21.59, "learning_rate": 6.246756310450577e-07, "loss": 0.0234, "step": 3390 }, { "epoch": 21.66, "learning_rate": 6.199575371549894e-07, "loss": 0.0213, "step": 3400 }, { "epoch": 21.72, "learning_rate": 6.15239443264921e-07, "loss": 0.0226, "step": 3410 }, { "epoch": 21.78, "learning_rate": 6.105213493748526e-07, "loss": 0.016, "step": 3420 }, { "epoch": 21.85, "learning_rate": 6.058032554847842e-07, "loss": 0.0177, "step": 3430 }, { "epoch": 21.91, "learning_rate": 6.010851615947157e-07, "loss": 0.0183, "step": 3440 }, { "epoch": 21.97, "learning_rate": 5.963670677046473e-07, "loss": 0.0159, "step": 3450 }, { "epoch": 22.0, "eval_accuracy": 0.96, "eval_loss": 0.12578421831130981, "eval_runtime": 6.9215, "eval_samples_per_second": 57.791, "eval_steps_per_second": 7.224, "step": 3454 }, { "epoch": 22.04, "learning_rate": 5.916489738145789e-07, "loss": 0.0183, "step": 3460 }, { "epoch": 22.1, "learning_rate": 5.869308799245106e-07, "loss": 0.0182, "step": 3470 }, { "epoch": 22.17, "learning_rate": 5.82212786034442e-07, "loss": 0.0163, "step": 3480 }, { "epoch": 22.23, "learning_rate": 5.774946921443736e-07, "loss": 0.0193, "step": 3490 }, { "epoch": 22.29, "learning_rate": 5.727765982543052e-07, "loss": 0.0176, "step": 3500 }, { "epoch": 22.36, "learning_rate": 5.680585043642368e-07, "loss": 0.0156, "step": 3510 }, { "epoch": 22.42, "learning_rate": 5.633404104741683e-07, "loss": 0.016, "step": 3520 }, { "epoch": 22.48, "learning_rate": 5.586223165840999e-07, "loss": 0.015, "step": 3530 }, { "epoch": 22.55, "learning_rate": 5.539042226940316e-07, "loss": 0.0166, "step": 3540 }, { "epoch": 22.61, "learning_rate": 5.491861288039632e-07, "loss": 0.0156, "step": 3550 }, { "epoch": 22.68, "learning_rate": 5.444680349138948e-07, "loss": 0.0152, "step": 3560 }, { "epoch": 22.74, "learning_rate": 5.397499410238263e-07, "loss": 0.0179, "step": 3570 }, { "epoch": 22.8, "learning_rate": 5.350318471337579e-07, "loss": 0.0148, "step": 3580 }, { "epoch": 22.87, "learning_rate": 5.303137532436895e-07, "loss": 0.0157, "step": 3590 }, { "epoch": 22.93, "learning_rate": 5.255956593536211e-07, "loss": 0.016, "step": 3600 }, { "epoch": 22.99, "learning_rate": 5.208775654635526e-07, "loss": 0.0137, "step": 3610 }, { "epoch": 23.0, "eval_accuracy": 0.9625, "eval_loss": 0.12316268682479858, "eval_runtime": 6.7883, "eval_samples_per_second": 58.925, "eval_steps_per_second": 7.366, "step": 3611 }, { "epoch": 23.06, "learning_rate": 5.161594715734843e-07, "loss": 0.0141, "step": 3620 }, { "epoch": 23.12, "learning_rate": 5.114413776834159e-07, "loss": 0.0116, "step": 3630 }, { "epoch": 23.18, "learning_rate": 5.067232837933475e-07, "loss": 0.0142, "step": 3640 }, { "epoch": 23.25, "learning_rate": 5.02005189903279e-07, "loss": 0.0126, "step": 3650 }, { "epoch": 23.31, "learning_rate": 4.972870960132106e-07, "loss": 0.0138, "step": 3660 }, { "epoch": 23.38, "learning_rate": 4.925690021231422e-07, "loss": 0.0135, "step": 3670 }, { "epoch": 23.44, "learning_rate": 4.878509082330738e-07, "loss": 0.0144, "step": 3680 }, { "epoch": 23.5, "learning_rate": 4.831328143430054e-07, "loss": 0.0147, "step": 3690 }, { "epoch": 23.57, "learning_rate": 4.78414720452937e-07, "loss": 0.0131, "step": 3700 }, { "epoch": 23.63, "learning_rate": 4.736966265628686e-07, "loss": 0.0142, "step": 3710 }, { "epoch": 23.69, "learning_rate": 4.689785326728002e-07, "loss": 0.0122, "step": 3720 }, { "epoch": 23.76, "learning_rate": 4.6426043878273177e-07, "loss": 0.0179, "step": 3730 }, { "epoch": 23.82, "learning_rate": 4.595423448926634e-07, "loss": 0.0152, "step": 3740 }, { "epoch": 23.89, "learning_rate": 4.548242510025949e-07, "loss": 0.015, "step": 3750 }, { "epoch": 23.95, "learning_rate": 4.501061571125265e-07, "loss": 0.0135, "step": 3760 }, { "epoch": 24.0, "eval_accuracy": 0.97, "eval_loss": 0.12535163760185242, "eval_runtime": 6.4232, "eval_samples_per_second": 62.274, "eval_steps_per_second": 7.784, "step": 3768 } ], "max_steps": 4710, "num_train_epochs": 30, "total_flos": 1.859841088487424e+19, "trial_name": null, "trial_params": null }