{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.2, "eval_steps": 500, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "grad_norm": 0.15373489260673523, "learning_rate": 9e-07, "loss": 1.2116, "step": 10 }, { "grad_norm": 0.12143079191446304, "learning_rate": 1.9e-06, "loss": 1.2054, "step": 20 }, { "grad_norm": 0.11240886896848679, "learning_rate": 2.9e-06, "loss": 1.2027, "step": 30 }, { "grad_norm": 0.21381829679012299, "learning_rate": 3.9e-06, "loss": 1.1886, "step": 40 }, { "grad_norm": 0.2458868771791458, "learning_rate": 4.9000000000000005e-06, "loss": 1.1589, "step": 50 }, { "grad_norm": 0.2544420659542084, "learning_rate": 5.9e-06, "loss": 1.1296, "step": 60 }, { "grad_norm": 0.23853255808353424, "learning_rate": 6.900000000000001e-06, "loss": 1.1103, "step": 70 }, { "grad_norm": 0.31489869952201843, "learning_rate": 7.9e-06, "loss": 1.0948, "step": 80 }, { "grad_norm": 0.2596352696418762, "learning_rate": 8.9e-06, "loss": 1.0914, "step": 90 }, { "grad_norm": 0.2258705496788025, "learning_rate": 9.900000000000002e-06, "loss": 1.0912, "step": 100 }, { "grad_norm": 0.5682782530784607, "learning_rate": 1.09e-05, "loss": 1.0893, "step": 110 }, { "grad_norm": 0.3197626769542694, "learning_rate": 1.19e-05, "loss": 1.081, "step": 120 }, { "grad_norm": 0.18166182935237885, "learning_rate": 1.29e-05, "loss": 1.073, "step": 130 }, { "grad_norm": 0.4554118812084198, "learning_rate": 1.3900000000000002e-05, "loss": 1.0634, "step": 140 }, { "grad_norm": 0.42795446515083313, "learning_rate": 1.49e-05, "loss": 1.047, "step": 150 }, { "grad_norm": 0.5283658504486084, "learning_rate": 1.59e-05, "loss": 1.013, "step": 160 }, { "grad_norm": 0.3358186185359955, "learning_rate": 1.69e-05, "loss": 0.9804, "step": 170 }, { "grad_norm": 0.5145778656005859, "learning_rate": 1.79e-05, "loss": 0.936, "step": 180 }, { "grad_norm": 0.5755268335342407, "learning_rate": 1.8900000000000002e-05, "loss": 0.8896, "step": 190 }, { "grad_norm": 0.6188399791717529, "learning_rate": 1.9900000000000003e-05, "loss": 0.8577, "step": 200 }, { "grad_norm": 0.9716882109642029, "learning_rate": 2.09e-05, "loss": 0.8271, "step": 210 }, { "grad_norm": 0.918986976146698, "learning_rate": 2.19e-05, "loss": 0.7906, "step": 220 }, { "grad_norm": 0.8000361919403076, "learning_rate": 2.29e-05, "loss": 0.751, "step": 230 }, { "grad_norm": 1.0514482259750366, "learning_rate": 2.39e-05, "loss": 0.7155, "step": 240 }, { "grad_norm": 0.7911136746406555, "learning_rate": 2.4900000000000002e-05, "loss": 0.6906, "step": 250 }, { "grad_norm": 0.92540043592453, "learning_rate": 2.5900000000000003e-05, "loss": 0.6638, "step": 260 }, { "grad_norm": 1.0331710577011108, "learning_rate": 2.6900000000000003e-05, "loss": 0.6444, "step": 270 }, { "grad_norm": 1.2573391199111938, "learning_rate": 2.7900000000000004e-05, "loss": 0.618, "step": 280 }, { "grad_norm": 1.2672662734985352, "learning_rate": 2.8899999999999998e-05, "loss": 0.6095, "step": 290 }, { "grad_norm": 1.0279985666275024, "learning_rate": 2.9900000000000002e-05, "loss": 0.5921, "step": 300 }, { "grad_norm": 0.9896909594535828, "learning_rate": 3.09e-05, "loss": 0.5765, "step": 310 }, { "grad_norm": 1.0756268501281738, "learning_rate": 3.19e-05, "loss": 0.5679, "step": 320 }, { "grad_norm": 1.3084145784378052, "learning_rate": 3.29e-05, "loss": 0.5569, "step": 330 }, { "grad_norm": 1.2895151376724243, "learning_rate": 3.3900000000000004e-05, "loss": 0.5475, "step": 340 }, { "grad_norm": 1.1118186712265015, "learning_rate": 3.49e-05, "loss": 0.5367, "step": 350 }, { "grad_norm": 0.9377628564834595, "learning_rate": 3.59e-05, "loss": 0.5192, "step": 360 }, { "grad_norm": 1.1136053800582886, "learning_rate": 3.69e-05, "loss": 0.5096, "step": 370 }, { "grad_norm": 0.8958713412284851, "learning_rate": 3.79e-05, "loss": 0.4938, "step": 380 }, { "grad_norm": 1.4132113456726074, "learning_rate": 3.8900000000000004e-05, "loss": 0.4777, "step": 390 }, { "grad_norm": 1.250508189201355, "learning_rate": 3.99e-05, "loss": 0.4575, "step": 400 }, { "grad_norm": 1.2522258758544922, "learning_rate": 4.09e-05, "loss": 0.4386, "step": 410 }, { "grad_norm": 1.4475443363189697, "learning_rate": 4.19e-05, "loss": 0.4295, "step": 420 }, { "grad_norm": 0.925055742263794, "learning_rate": 4.29e-05, "loss": 0.4157, "step": 430 }, { "grad_norm": 1.2969681024551392, "learning_rate": 4.39e-05, "loss": 0.4052, "step": 440 }, { "grad_norm": 1.3583041429519653, "learning_rate": 4.49e-05, "loss": 0.3953, "step": 450 }, { "grad_norm": 1.028383493423462, "learning_rate": 4.5900000000000004e-05, "loss": 0.375, "step": 460 }, { "grad_norm": 1.255179762840271, "learning_rate": 4.69e-05, "loss": 0.3699, "step": 470 }, { "grad_norm": 1.326106309890747, "learning_rate": 4.79e-05, "loss": 0.3579, "step": 480 }, { "grad_norm": 1.2961987257003784, "learning_rate": 4.89e-05, "loss": 0.3478, "step": 490 }, { "grad_norm": 1.2104370594024658, "learning_rate": 4.99e-05, "loss": 0.339, "step": 500 }, { "grad_norm": 1.4508124589920044, "learning_rate": 5.0900000000000004e-05, "loss": 0.3219, "step": 510 }, { "grad_norm": 1.4312357902526855, "learning_rate": 5.19e-05, "loss": 0.3042, "step": 520 }, { "grad_norm": 1.7213497161865234, "learning_rate": 5.2900000000000005e-05, "loss": 0.2887, "step": 530 }, { "grad_norm": 1.7162240743637085, "learning_rate": 5.390000000000001e-05, "loss": 0.2856, "step": 540 }, { "grad_norm": 1.1437758207321167, "learning_rate": 5.4900000000000006e-05, "loss": 0.2642, "step": 550 }, { "grad_norm": 1.4461580514907837, "learning_rate": 5.590000000000001e-05, "loss": 0.2508, "step": 560 }, { "grad_norm": 1.4058153629302979, "learning_rate": 5.69e-05, "loss": 0.2437, "step": 570 }, { "grad_norm": 1.3282971382141113, "learning_rate": 5.79e-05, "loss": 0.2275, "step": 580 }, { "grad_norm": 1.6089696884155273, "learning_rate": 5.89e-05, "loss": 0.2166, "step": 590 }, { "grad_norm": 1.5542049407958984, "learning_rate": 5.99e-05, "loss": 0.2008, "step": 600 }, { "grad_norm": 1.372789740562439, "learning_rate": 6.09e-05, "loss": 0.1888, "step": 610 }, { "grad_norm": 1.3900244235992432, "learning_rate": 6.19e-05, "loss": 0.1825, "step": 620 }, { "grad_norm": 1.1616129875183105, "learning_rate": 6.29e-05, "loss": 0.1673, "step": 630 }, { "grad_norm": 1.3904050588607788, "learning_rate": 6.390000000000001e-05, "loss": 0.1631, "step": 640 }, { "grad_norm": 1.6666728258132935, "learning_rate": 6.49e-05, "loss": 0.1508, "step": 650 }, { "grad_norm": 1.458060383796692, "learning_rate": 6.59e-05, "loss": 0.134, "step": 660 }, { "grad_norm": 1.2632263898849487, "learning_rate": 6.690000000000001e-05, "loss": 0.1435, "step": 670 }, { "grad_norm": 1.3523911237716675, "learning_rate": 6.790000000000001e-05, "loss": 0.1383, "step": 680 }, { "grad_norm": 1.3925788402557373, "learning_rate": 6.89e-05, "loss": 0.1317, "step": 690 }, { "grad_norm": 1.6858913898468018, "learning_rate": 6.99e-05, "loss": 0.1309, "step": 700 }, { "grad_norm": 1.2360647916793823, "learning_rate": 7.09e-05, "loss": 0.1267, "step": 710 }, { "grad_norm": 1.1521002054214478, "learning_rate": 7.19e-05, "loss": 0.1259, "step": 720 }, { "grad_norm": 1.232092022895813, "learning_rate": 7.29e-05, "loss": 0.1333, "step": 730 }, { "grad_norm": 1.4635534286499023, "learning_rate": 7.390000000000001e-05, "loss": 0.129, "step": 740 }, { "grad_norm": 1.108782172203064, "learning_rate": 7.49e-05, "loss": 0.1285, "step": 750 }, { "grad_norm": 1.3770654201507568, "learning_rate": 7.59e-05, "loss": 0.1247, "step": 760 }, { "grad_norm": 1.1716821193695068, "learning_rate": 7.69e-05, "loss": 0.1297, "step": 770 }, { "grad_norm": 1.185289740562439, "learning_rate": 7.790000000000001e-05, "loss": 0.1246, "step": 780 }, { "grad_norm": 1.3426638841629028, "learning_rate": 7.890000000000001e-05, "loss": 0.1209, "step": 790 }, { "grad_norm": 1.0914347171783447, "learning_rate": 7.99e-05, "loss": 0.1226, "step": 800 }, { "grad_norm": 1.47992742061615, "learning_rate": 8.090000000000001e-05, "loss": 0.1116, "step": 810 }, { "grad_norm": 1.189318299293518, "learning_rate": 8.19e-05, "loss": 0.1155, "step": 820 }, { "grad_norm": 1.218897819519043, "learning_rate": 8.29e-05, "loss": 0.1106, "step": 830 }, { "grad_norm": 1.1049423217773438, "learning_rate": 8.39e-05, "loss": 0.1143, "step": 840 }, { "grad_norm": 1.3933051824569702, "learning_rate": 8.49e-05, "loss": 0.1241, "step": 850 }, { "grad_norm": 1.0603047609329224, "learning_rate": 8.59e-05, "loss": 0.1111, "step": 860 }, { "grad_norm": 1.1517871618270874, "learning_rate": 8.69e-05, "loss": 0.1114, "step": 870 }, { "grad_norm": 1.2335821390151978, "learning_rate": 8.790000000000001e-05, "loss": 0.1135, "step": 880 }, { "grad_norm": 1.471954107284546, "learning_rate": 8.89e-05, "loss": 0.1149, "step": 890 }, { "grad_norm": 1.1479299068450928, "learning_rate": 8.99e-05, "loss": 0.1144, "step": 900 }, { "grad_norm": 1.1517722606658936, "learning_rate": 9.090000000000001e-05, "loss": 0.1097, "step": 910 }, { "grad_norm": 1.2328661680221558, "learning_rate": 9.190000000000001e-05, "loss": 0.11, "step": 920 }, { "grad_norm": 1.227306842803955, "learning_rate": 9.290000000000001e-05, "loss": 0.1144, "step": 930 }, { "grad_norm": 1.1643450260162354, "learning_rate": 9.39e-05, "loss": 0.1074, "step": 940 }, { "grad_norm": 1.1138417720794678, "learning_rate": 9.49e-05, "loss": 0.113, "step": 950 }, { "grad_norm": 1.0035343170166016, "learning_rate": 9.59e-05, "loss": 0.1102, "step": 960 }, { "grad_norm": 1.089536190032959, "learning_rate": 9.69e-05, "loss": 0.1089, "step": 970 }, { "grad_norm": 1.1418471336364746, "learning_rate": 9.790000000000001e-05, "loss": 0.1075, "step": 980 }, { "grad_norm": 1.0369817018508911, "learning_rate": 9.89e-05, "loss": 0.1094, "step": 990 }, { "grad_norm": 1.081180214881897, "learning_rate": 9.99e-05, "loss": 0.1069, "step": 1000 }, { "grad_norm": 1.1965157985687256, "learning_rate": 9.999994463727085e-05, "loss": 0.1078, "step": 1010 }, { "grad_norm": 1.0836145877838135, "learning_rate": 9.999975326009292e-05, "loss": 0.1034, "step": 1020 }, { "grad_norm": 1.0701448917388916, "learning_rate": 9.999942518549879e-05, "loss": 0.1063, "step": 1030 }, { "grad_norm": 1.0245521068572998, "learning_rate": 9.999896041438544e-05, "loss": 0.1003, "step": 1040 }, { "grad_norm": 0.9008662104606628, "learning_rate": 9.999835894802353e-05, "loss": 0.1007, "step": 1050 }, { "grad_norm": 1.405444860458374, "learning_rate": 9.999762078805743e-05, "loss": 0.0983, "step": 1060 }, { "grad_norm": 0.9956735968589783, "learning_rate": 9.999674593650526e-05, "loss": 0.1015, "step": 1070 }, { "grad_norm": 0.9828646779060364, "learning_rate": 9.99957343957588e-05, "loss": 0.099, "step": 1080 }, { "grad_norm": 1.3071095943450928, "learning_rate": 9.99945861685836e-05, "loss": 0.0949, "step": 1090 }, { "grad_norm": 0.9832029342651367, "learning_rate": 9.999330125811884e-05, "loss": 0.102, "step": 1100 }, { "grad_norm": 1.0148258209228516, "learning_rate": 9.999187966787744e-05, "loss": 0.103, "step": 1110 }, { "grad_norm": 1.0134457349777222, "learning_rate": 9.999032140174595e-05, "loss": 0.0949, "step": 1120 }, { "grad_norm": 0.9154807329177856, "learning_rate": 9.998862646398464e-05, "loss": 0.0989, "step": 1130 }, { "grad_norm": 1.0683287382125854, "learning_rate": 9.998679485922739e-05, "loss": 0.0937, "step": 1140 }, { "grad_norm": 0.9443909525871277, "learning_rate": 9.998482659248174e-05, "loss": 0.0964, "step": 1150 }, { "grad_norm": 0.8048564195632935, "learning_rate": 9.998272166912883e-05, "loss": 0.0978, "step": 1160 }, { "grad_norm": 1.0226809978485107, "learning_rate": 9.998048009492347e-05, "loss": 0.0979, "step": 1170 }, { "grad_norm": 1.0117919445037842, "learning_rate": 9.997810187599403e-05, "loss": 0.0993, "step": 1180 }, { "grad_norm": 0.8559236526489258, "learning_rate": 9.997558701884249e-05, "loss": 0.095, "step": 1190 }, { "grad_norm": 1.009264349937439, "learning_rate": 9.997293553034433e-05, "loss": 0.0947, "step": 1200 }, { "grad_norm": 1.0090675354003906, "learning_rate": 9.997014741774866e-05, "loss": 0.0991, "step": 1210 }, { "grad_norm": 0.9128223657608032, "learning_rate": 9.996722268867803e-05, "loss": 0.0973, "step": 1220 }, { "grad_norm": 0.8223145008087158, "learning_rate": 9.996416135112858e-05, "loss": 0.0943, "step": 1230 }, { "grad_norm": 0.6892135143280029, "learning_rate": 9.996096341346988e-05, "loss": 0.0914, "step": 1240 }, { "grad_norm": 0.8441371321678162, "learning_rate": 9.995762888444495e-05, "loss": 0.095, "step": 1250 }, { "grad_norm": 0.8410964608192444, "learning_rate": 9.995415777317027e-05, "loss": 0.0905, "step": 1260 }, { "grad_norm": 0.9202972650527954, "learning_rate": 9.995055008913574e-05, "loss": 0.09, "step": 1270 }, { "grad_norm": 0.969850480556488, "learning_rate": 9.994680584220463e-05, "loss": 0.0977, "step": 1280 }, { "grad_norm": 0.9843644499778748, "learning_rate": 9.994292504261355e-05, "loss": 0.0903, "step": 1290 }, { "grad_norm": 0.8046810030937195, "learning_rate": 9.993890770097247e-05, "loss": 0.0885, "step": 1300 }, { "grad_norm": 1.0213329792022705, "learning_rate": 9.993475382826467e-05, "loss": 0.0906, "step": 1310 }, { "grad_norm": 0.9399948120117188, "learning_rate": 9.993046343584664e-05, "loss": 0.0885, "step": 1320 }, { "grad_norm": 0.8615891337394714, "learning_rate": 9.992603653544816e-05, "loss": 0.0898, "step": 1330 }, { "grad_norm": 0.7804251909255981, "learning_rate": 9.992147313917222e-05, "loss": 0.0879, "step": 1340 }, { "grad_norm": 0.7505974173545837, "learning_rate": 9.991677325949497e-05, "loss": 0.0877, "step": 1350 }, { "grad_norm": 0.8564302921295166, "learning_rate": 9.991193690926568e-05, "loss": 0.0853, "step": 1360 }, { "grad_norm": 0.921603798866272, "learning_rate": 9.990696410170678e-05, "loss": 0.0868, "step": 1370 }, { "grad_norm": 0.905284583568573, "learning_rate": 9.990185485041371e-05, "loss": 0.0812, "step": 1380 }, { "grad_norm": 0.8342713713645935, "learning_rate": 9.989660916935498e-05, "loss": 0.0845, "step": 1390 }, { "grad_norm": 0.8646072149276733, "learning_rate": 9.989122707287208e-05, "loss": 0.0863, "step": 1400 }, { "grad_norm": 1.0666006803512573, "learning_rate": 9.988570857567945e-05, "loss": 0.0887, "step": 1410 }, { "grad_norm": 0.8190611004829407, "learning_rate": 9.988005369286446e-05, "loss": 0.0845, "step": 1420 }, { "grad_norm": 0.8214516639709473, "learning_rate": 9.987426243988734e-05, "loss": 0.0883, "step": 1430 }, { "grad_norm": 0.7905929684638977, "learning_rate": 9.986833483258114e-05, "loss": 0.0861, "step": 1440 }, { "grad_norm": 0.8426095247268677, "learning_rate": 9.986227088715173e-05, "loss": 0.0823, "step": 1450 }, { "grad_norm": 0.9610787630081177, "learning_rate": 9.98560706201777e-05, "loss": 0.0835, "step": 1460 }, { "grad_norm": 0.8222280740737915, "learning_rate": 9.984973404861036e-05, "loss": 0.0878, "step": 1470 }, { "grad_norm": 0.840451717376709, "learning_rate": 9.984326118977361e-05, "loss": 0.0812, "step": 1480 }, { "grad_norm": 0.7655553817749023, "learning_rate": 9.983665206136406e-05, "loss": 0.0784, "step": 1490 }, { "grad_norm": 0.9331004023551941, "learning_rate": 9.982990668145075e-05, "loss": 0.0779, "step": 1500 }, { "grad_norm": 0.7345725297927856, "learning_rate": 9.982302506847534e-05, "loss": 0.0792, "step": 1510 }, { "grad_norm": 0.8967569470405579, "learning_rate": 9.981600724125189e-05, "loss": 0.0782, "step": 1520 }, { "grad_norm": 0.7933437824249268, "learning_rate": 9.980885321896685e-05, "loss": 0.0773, "step": 1530 }, { "grad_norm": 0.6864101886749268, "learning_rate": 9.980156302117905e-05, "loss": 0.0773, "step": 1540 }, { "grad_norm": 0.689123272895813, "learning_rate": 9.979413666781963e-05, "loss": 0.0824, "step": 1550 }, { "grad_norm": 0.8207864165306091, "learning_rate": 9.978657417919193e-05, "loss": 0.0803, "step": 1560 }, { "grad_norm": 0.7813171744346619, "learning_rate": 9.977887557597153e-05, "loss": 0.0817, "step": 1570 }, { "grad_norm": 0.695106565952301, "learning_rate": 9.97710408792061e-05, "loss": 0.0783, "step": 1580 }, { "grad_norm": 0.9048652052879333, "learning_rate": 9.976307011031542e-05, "loss": 0.0788, "step": 1590 }, { "grad_norm": 0.8499309420585632, "learning_rate": 9.975496329109126e-05, "loss": 0.081, "step": 1600 }, { "grad_norm": 0.8718334436416626, "learning_rate": 9.974672044369732e-05, "loss": 0.0777, "step": 1610 }, { "grad_norm": 0.6900801658630371, "learning_rate": 9.97383415906693e-05, "loss": 0.077, "step": 1620 }, { "grad_norm": 0.8092792630195618, "learning_rate": 9.97298267549146e-05, "loss": 0.0788, "step": 1630 }, { "grad_norm": 0.8192611932754517, "learning_rate": 9.972117595971249e-05, "loss": 0.0799, "step": 1640 }, { "grad_norm": 0.8722266554832458, "learning_rate": 9.971238922871391e-05, "loss": 0.0819, "step": 1650 }, { "grad_norm": 0.914159893989563, "learning_rate": 9.970346658594142e-05, "loss": 0.0838, "step": 1660 }, { "grad_norm": 0.8131964206695557, "learning_rate": 9.969440805578923e-05, "loss": 0.0797, "step": 1670 }, { "grad_norm": 0.7147591710090637, "learning_rate": 9.968521366302298e-05, "loss": 0.0799, "step": 1680 }, { "grad_norm": 0.8249319791793823, "learning_rate": 9.967588343277981e-05, "loss": 0.0783, "step": 1690 }, { "grad_norm": 0.7327215075492859, "learning_rate": 9.966641739056818e-05, "loss": 0.0787, "step": 1700 }, { "grad_norm": 0.7018980979919434, "learning_rate": 9.965681556226793e-05, "loss": 0.0705, "step": 1710 }, { "grad_norm": 0.8972001671791077, "learning_rate": 9.964707797413006e-05, "loss": 0.0796, "step": 1720 }, { "grad_norm": 0.8128800392150879, "learning_rate": 9.963720465277679e-05, "loss": 0.0818, "step": 1730 }, { "grad_norm": 0.9391877055168152, "learning_rate": 9.96271956252014e-05, "loss": 0.0758, "step": 1740 }, { "grad_norm": 0.8188132047653198, "learning_rate": 9.961705091876816e-05, "loss": 0.0756, "step": 1750 }, { "grad_norm": 0.8037185668945312, "learning_rate": 9.960677056121235e-05, "loss": 0.0783, "step": 1760 }, { "grad_norm": 0.6627317070960999, "learning_rate": 9.959635458064005e-05, "loss": 0.0773, "step": 1770 }, { "grad_norm": 0.6626058220863342, "learning_rate": 9.958580300552815e-05, "loss": 0.0775, "step": 1780 }, { "grad_norm": 0.7204414010047913, "learning_rate": 9.957511586472426e-05, "loss": 0.0762, "step": 1790 }, { "grad_norm": 0.6517882347106934, "learning_rate": 9.956429318744662e-05, "loss": 0.0776, "step": 1800 }, { "grad_norm": 0.7082467675209045, "learning_rate": 9.955333500328404e-05, "loss": 0.0737, "step": 1810 }, { "grad_norm": 0.7636476159095764, "learning_rate": 9.95422413421957e-05, "loss": 0.0733, "step": 1820 }, { "grad_norm": 0.72133868932724, "learning_rate": 9.953101223451133e-05, "loss": 0.0749, "step": 1830 }, { "grad_norm": 0.7455642223358154, "learning_rate": 9.951964771093085e-05, "loss": 0.074, "step": 1840 }, { "grad_norm": 0.6844678521156311, "learning_rate": 9.950814780252442e-05, "loss": 0.0773, "step": 1850 }, { "grad_norm": 0.49062681198120117, "learning_rate": 9.949651254073236e-05, "loss": 0.0732, "step": 1860 }, { "grad_norm": 0.789148211479187, "learning_rate": 9.948474195736504e-05, "loss": 0.0718, "step": 1870 }, { "grad_norm": 0.6764615178108215, "learning_rate": 9.947283608460277e-05, "loss": 0.0742, "step": 1880 }, { "grad_norm": 0.6999238133430481, "learning_rate": 9.946079495499577e-05, "loss": 0.0747, "step": 1890 }, { "grad_norm": 0.7517783045768738, "learning_rate": 9.944861860146401e-05, "loss": 0.0753, "step": 1900 }, { "grad_norm": 0.7183238863945007, "learning_rate": 9.943630705729719e-05, "loss": 0.0697, "step": 1910 }, { "grad_norm": 0.5847482681274414, "learning_rate": 9.942386035615459e-05, "loss": 0.072, "step": 1920 }, { "grad_norm": 0.5976929068565369, "learning_rate": 9.941127853206503e-05, "loss": 0.0694, "step": 1930 }, { "grad_norm": 0.5992315411567688, "learning_rate": 9.939856161942673e-05, "loss": 0.0729, "step": 1940 }, { "grad_norm": 0.6920360922813416, "learning_rate": 9.938570965300724e-05, "loss": 0.0693, "step": 1950 }, { "grad_norm": 0.7227588891983032, "learning_rate": 9.937272266794335e-05, "loss": 0.0779, "step": 1960 }, { "grad_norm": 0.6451845169067383, "learning_rate": 9.935960069974096e-05, "loss": 0.0705, "step": 1970 }, { "grad_norm": 0.7243984341621399, "learning_rate": 9.934634378427506e-05, "loss": 0.0708, "step": 1980 }, { "grad_norm": 0.6210780143737793, "learning_rate": 9.933295195778954e-05, "loss": 0.0772, "step": 1990 }, { "grad_norm": 0.5807631611824036, "learning_rate": 9.931942525689715e-05, "loss": 0.0745, "step": 2000 }, { "grad_norm": 0.6609013676643372, "learning_rate": 9.930576371857936e-05, "loss": 0.0723, "step": 2010 }, { "grad_norm": 0.7919302582740784, "learning_rate": 9.929196738018629e-05, "loss": 0.0702, "step": 2020 }, { "grad_norm": 0.6295292973518372, "learning_rate": 9.927803627943662e-05, "loss": 0.0674, "step": 2030 }, { "grad_norm": 0.756155788898468, "learning_rate": 9.926397045441744e-05, "loss": 0.0687, "step": 2040 }, { "grad_norm": 0.7688308954238892, "learning_rate": 9.924976994358417e-05, "loss": 0.0718, "step": 2050 }, { "grad_norm": 0.8509033918380737, "learning_rate": 9.923543478576048e-05, "loss": 0.0727, "step": 2060 }, { "grad_norm": 0.7590917944908142, "learning_rate": 9.922096502013813e-05, "loss": 0.074, "step": 2070 }, { "grad_norm": 0.8109246492385864, "learning_rate": 9.92063606862769e-05, "loss": 0.0721, "step": 2080 }, { "grad_norm": 0.6400628685951233, "learning_rate": 9.919162182410453e-05, "loss": 0.0707, "step": 2090 }, { "grad_norm": 0.5556660890579224, "learning_rate": 9.917674847391645e-05, "loss": 0.0663, "step": 2100 }, { "grad_norm": 0.7919337749481201, "learning_rate": 9.916174067637584e-05, "loss": 0.0675, "step": 2110 }, { "grad_norm": 0.6571515798568726, "learning_rate": 9.914659847251348e-05, "loss": 0.0683, "step": 2120 }, { "grad_norm": 0.5666832327842712, "learning_rate": 9.913132190372753e-05, "loss": 0.0706, "step": 2130 }, { "grad_norm": 0.6068550944328308, "learning_rate": 9.911591101178359e-05, "loss": 0.067, "step": 2140 }, { "grad_norm": 0.5472414493560791, "learning_rate": 9.910036583881443e-05, "loss": 0.0686, "step": 2150 }, { "grad_norm": 0.6243656277656555, "learning_rate": 9.908468642731995e-05, "loss": 0.0663, "step": 2160 }, { "grad_norm": 0.685313880443573, "learning_rate": 9.906887282016707e-05, "loss": 0.0704, "step": 2170 }, { "grad_norm": 0.6772751212120056, "learning_rate": 9.90529250605896e-05, "loss": 0.0651, "step": 2180 }, { "grad_norm": 0.6815426349639893, "learning_rate": 9.903684319218809e-05, "loss": 0.0728, "step": 2190 }, { "grad_norm": 0.822087287902832, "learning_rate": 9.902062725892976e-05, "loss": 0.0723, "step": 2200 }, { "grad_norm": 0.6740921139717102, "learning_rate": 9.900427730514834e-05, "loss": 0.0722, "step": 2210 }, { "grad_norm": 0.730047345161438, "learning_rate": 9.8987793375544e-05, "loss": 0.0717, "step": 2220 }, { "grad_norm": 0.8769873976707458, "learning_rate": 9.897117551518318e-05, "loss": 0.0672, "step": 2230 }, { "grad_norm": 0.5988404750823975, "learning_rate": 9.895442376949844e-05, "loss": 0.0691, "step": 2240 }, { "grad_norm": 0.7594494223594666, "learning_rate": 9.893753818428845e-05, "loss": 0.0703, "step": 2250 }, { "grad_norm": 0.7655356526374817, "learning_rate": 9.892051880571773e-05, "loss": 0.0643, "step": 2260 }, { "grad_norm": 0.6700913906097412, "learning_rate": 9.890336568031663e-05, "loss": 0.066, "step": 2270 }, { "grad_norm": 0.6776233315467834, "learning_rate": 9.888607885498113e-05, "loss": 0.0713, "step": 2280 }, { "grad_norm": 0.6251364350318909, "learning_rate": 9.886865837697275e-05, "loss": 0.0675, "step": 2290 }, { "grad_norm": 0.6322036981582642, "learning_rate": 9.88511042939184e-05, "loss": 0.07, "step": 2300 }, { "grad_norm": 0.6706236600875854, "learning_rate": 9.883341665381028e-05, "loss": 0.0707, "step": 2310 }, { "grad_norm": 0.7654581069946289, "learning_rate": 9.881559550500575e-05, "loss": 0.0684, "step": 2320 }, { "grad_norm": 0.6615346074104309, "learning_rate": 9.879764089622712e-05, "loss": 0.0664, "step": 2330 }, { "grad_norm": 0.6900296807289124, "learning_rate": 9.87795528765616e-05, "loss": 0.0703, "step": 2340 }, { "grad_norm": 0.6962850689888, "learning_rate": 9.876133149546118e-05, "loss": 0.0712, "step": 2350 }, { "grad_norm": 0.5966289639472961, "learning_rate": 9.874297680274238e-05, "loss": 0.0705, "step": 2360 }, { "grad_norm": 0.5869346261024475, "learning_rate": 9.872448884858624e-05, "loss": 0.0673, "step": 2370 }, { "grad_norm": 0.6729479432106018, "learning_rate": 9.870586768353815e-05, "loss": 0.0698, "step": 2380 }, { "grad_norm": 0.4968750476837158, "learning_rate": 9.868711335850764e-05, "loss": 0.071, "step": 2390 }, { "grad_norm": 0.5750929713249207, "learning_rate": 9.866822592476833e-05, "loss": 0.066, "step": 2400 }, { "grad_norm": 0.6520802974700928, "learning_rate": 9.86492054339577e-05, "loss": 0.0704, "step": 2410 }, { "grad_norm": 0.6432913541793823, "learning_rate": 9.863005193807711e-05, "loss": 0.0669, "step": 2420 }, { "grad_norm": 0.7633799314498901, "learning_rate": 9.861076548949143e-05, "loss": 0.0641, "step": 2430 }, { "grad_norm": 0.727251410484314, "learning_rate": 9.859134614092912e-05, "loss": 0.0666, "step": 2440 }, { "grad_norm": 0.6507017016410828, "learning_rate": 9.857179394548191e-05, "loss": 0.0631, "step": 2450 }, { "grad_norm": 0.802526593208313, "learning_rate": 9.855210895660477e-05, "loss": 0.0695, "step": 2460 }, { "grad_norm": 0.4968661069869995, "learning_rate": 9.853229122811568e-05, "loss": 0.0668, "step": 2470 }, { "grad_norm": 0.5049183964729309, "learning_rate": 9.851234081419559e-05, "loss": 0.0679, "step": 2480 }, { "grad_norm": 0.5562856793403625, "learning_rate": 9.849225776938814e-05, "loss": 0.067, "step": 2490 }, { "grad_norm": 0.578382134437561, "learning_rate": 9.847204214859964e-05, "loss": 0.0641, "step": 2500 }, { "grad_norm": 0.6934925317764282, "learning_rate": 9.845169400709879e-05, "loss": 0.0677, "step": 2510 }, { "grad_norm": 0.729125440120697, "learning_rate": 9.843121340051664e-05, "loss": 0.0637, "step": 2520 }, { "grad_norm": 0.6229711174964905, "learning_rate": 9.841060038484641e-05, "loss": 0.064, "step": 2530 }, { "grad_norm": 0.6090768575668335, "learning_rate": 9.838985501644328e-05, "loss": 0.0623, "step": 2540 }, { "grad_norm": 0.8319109082221985, "learning_rate": 9.83689773520243e-05, "loss": 0.0669, "step": 2550 }, { "grad_norm": 0.6815285086631775, "learning_rate": 9.834796744866819e-05, "loss": 0.0661, "step": 2560 }, { "grad_norm": 0.49073100090026855, "learning_rate": 9.832682536381525e-05, "loss": 0.0686, "step": 2570 }, { "grad_norm": 0.5500509142875671, "learning_rate": 9.830555115526711e-05, "loss": 0.0663, "step": 2580 }, { "grad_norm": 0.5543439388275146, "learning_rate": 9.828414488118667e-05, "loss": 0.0586, "step": 2590 }, { "grad_norm": 0.5412293672561646, "learning_rate": 9.826260660009785e-05, "loss": 0.065, "step": 2600 }, { "grad_norm": 0.46607404947280884, "learning_rate": 9.824093637088547e-05, "loss": 0.0643, "step": 2610 }, { "grad_norm": 0.5089811682701111, "learning_rate": 9.821913425279514e-05, "loss": 0.0627, "step": 2620 }, { "grad_norm": 0.6004795432090759, "learning_rate": 9.8197200305433e-05, "loss": 0.0648, "step": 2630 }, { "grad_norm": 0.5946650505065918, "learning_rate": 9.817513458876564e-05, "loss": 0.0634, "step": 2640 }, { "grad_norm": 0.7084515690803528, "learning_rate": 9.815293716311987e-05, "loss": 0.0653, "step": 2650 }, { "grad_norm": 0.6806345582008362, "learning_rate": 9.813060808918262e-05, "loss": 0.0639, "step": 2660 }, { "grad_norm": 0.6408096551895142, "learning_rate": 9.810814742800069e-05, "loss": 0.0641, "step": 2670 }, { "grad_norm": 0.5025924444198608, "learning_rate": 9.808555524098074e-05, "loss": 0.063, "step": 2680 }, { "grad_norm": 0.7290829420089722, "learning_rate": 9.806283158988887e-05, "loss": 0.0647, "step": 2690 }, { "grad_norm": 0.5606433153152466, "learning_rate": 9.803997653685072e-05, "loss": 0.062, "step": 2700 }, { "grad_norm": 0.5064647793769836, "learning_rate": 9.801699014435112e-05, "loss": 0.0643, "step": 2710 }, { "grad_norm": 0.5947048664093018, "learning_rate": 9.799387247523398e-05, "loss": 0.06, "step": 2720 }, { "grad_norm": 0.6373162269592285, "learning_rate": 9.797062359270215e-05, "loss": 0.0596, "step": 2730 }, { "grad_norm": 0.683765709400177, "learning_rate": 9.794724356031715e-05, "loss": 0.0641, "step": 2740 }, { "grad_norm": 0.6411890983581543, "learning_rate": 9.792373244199913e-05, "loss": 0.0687, "step": 2750 }, { "grad_norm": 0.4775662422180176, "learning_rate": 9.790009030202658e-05, "loss": 0.0622, "step": 2760 }, { "grad_norm": 0.651835560798645, "learning_rate": 9.78763172050362e-05, "loss": 0.0621, "step": 2770 }, { "grad_norm": 0.6854885220527649, "learning_rate": 9.785241321602274e-05, "loss": 0.0667, "step": 2780 }, { "grad_norm": 0.6907970309257507, "learning_rate": 9.782837840033879e-05, "loss": 0.0669, "step": 2790 }, { "grad_norm": 0.5823140740394592, "learning_rate": 9.780421282369461e-05, "loss": 0.0678, "step": 2800 }, { "grad_norm": 0.6364890933036804, "learning_rate": 9.777991655215797e-05, "loss": 0.0663, "step": 2810 }, { "grad_norm": 0.5718041658401489, "learning_rate": 9.775548965215394e-05, "loss": 0.0626, "step": 2820 }, { "grad_norm": 0.4957229793071747, "learning_rate": 9.773093219046474e-05, "loss": 0.0634, "step": 2830 }, { "grad_norm": 0.6113382577896118, "learning_rate": 9.770624423422954e-05, "loss": 0.0665, "step": 2840 }, { "grad_norm": 0.6246403455734253, "learning_rate": 9.768142585094426e-05, "loss": 0.0626, "step": 2850 }, { "grad_norm": 0.6030095219612122, "learning_rate": 9.765647710846142e-05, "loss": 0.0662, "step": 2860 }, { "grad_norm": 0.5451913475990295, "learning_rate": 9.763139807498991e-05, "loss": 0.0692, "step": 2870 }, { "grad_norm": 0.5483195185661316, "learning_rate": 9.760618881909487e-05, "loss": 0.0599, "step": 2880 }, { "grad_norm": 0.47611215710639954, "learning_rate": 9.758084940969744e-05, "loss": 0.0635, "step": 2890 }, { "grad_norm": 0.5544406175613403, "learning_rate": 9.755537991607459e-05, "loss": 0.0592, "step": 2900 }, { "grad_norm": 0.6410639882087708, "learning_rate": 9.752978040785895e-05, "loss": 0.0613, "step": 2910 }, { "grad_norm": 0.6519464254379272, "learning_rate": 9.750405095503859e-05, "loss": 0.0584, "step": 2920 }, { "grad_norm": 0.6715825200080872, "learning_rate": 9.747819162795686e-05, "loss": 0.064, "step": 2930 }, { "grad_norm": 0.5294116139411926, "learning_rate": 9.745220249731217e-05, "loss": 0.0617, "step": 2940 }, { "grad_norm": 0.5013177394866943, "learning_rate": 9.742608363415781e-05, "loss": 0.0579, "step": 2950 }, { "grad_norm": 0.7086690664291382, "learning_rate": 9.739983510990176e-05, "loss": 0.0627, "step": 2960 }, { "grad_norm": 0.6349759697914124, "learning_rate": 9.737345699630647e-05, "loss": 0.0676, "step": 2970 }, { "grad_norm": 0.6648471355438232, "learning_rate": 9.734694936548869e-05, "loss": 0.0642, "step": 2980 }, { "grad_norm": 0.4997161328792572, "learning_rate": 9.732031228991932e-05, "loss": 0.0603, "step": 2990 }, { "grad_norm": 0.4930708408355713, "learning_rate": 9.729354584242302e-05, "loss": 0.0596, "step": 3000 }, { "grad_norm": 0.6939645409584045, "learning_rate": 9.726665009617832e-05, "loss": 0.0627, "step": 3010 }, { "grad_norm": 0.6745036244392395, "learning_rate": 9.723962512471714e-05, "loss": 0.0622, "step": 3020 }, { "grad_norm": 0.6208877563476562, "learning_rate": 9.72124710019247e-05, "loss": 0.0599, "step": 3030 }, { "grad_norm": 0.615725576877594, "learning_rate": 9.718518780203934e-05, "loss": 0.0611, "step": 3040 }, { "grad_norm": 0.5784436464309692, "learning_rate": 9.715777559965228e-05, "loss": 0.0598, "step": 3050 }, { "grad_norm": 0.6067401766777039, "learning_rate": 9.713023446970746e-05, "loss": 0.0647, "step": 3060 }, { "grad_norm": 0.5806649327278137, "learning_rate": 9.710256448750126e-05, "loss": 0.0625, "step": 3070 }, { "grad_norm": 0.6504300236701965, "learning_rate": 9.707476572868235e-05, "loss": 0.058, "step": 3080 }, { "grad_norm": 0.500622570514679, "learning_rate": 9.704683826925149e-05, "loss": 0.0577, "step": 3090 }, { "grad_norm": 0.6325183510780334, "learning_rate": 9.701878218556129e-05, "loss": 0.0603, "step": 3100 }, { "grad_norm": 0.555566132068634, "learning_rate": 9.699059755431598e-05, "loss": 0.0616, "step": 3110 }, { "grad_norm": 0.4982530474662781, "learning_rate": 9.696228445257132e-05, "loss": 0.0594, "step": 3120 }, { "grad_norm": 0.5743557214736938, "learning_rate": 9.693384295773419e-05, "loss": 0.056, "step": 3130 }, { "grad_norm": 0.4599805772304535, "learning_rate": 9.690527314756259e-05, "loss": 0.0551, "step": 3140 }, { "grad_norm": 0.6725175976753235, "learning_rate": 9.687657510016527e-05, "loss": 0.0586, "step": 3150 }, { "grad_norm": 0.5373369455337524, "learning_rate": 9.684774889400161e-05, "loss": 0.0576, "step": 3160 }, { "grad_norm": 0.6253339648246765, "learning_rate": 9.681879460788135e-05, "loss": 0.0586, "step": 3170 }, { "grad_norm": 0.6447578072547913, "learning_rate": 9.67897123209644e-05, "loss": 0.0639, "step": 3180 }, { "grad_norm": 0.44229140877723694, "learning_rate": 9.676050211276062e-05, "loss": 0.0606, "step": 3190 }, { "grad_norm": 0.6652916669845581, "learning_rate": 9.673116406312962e-05, "loss": 0.0629, "step": 3200 }, { "grad_norm": 0.7552599906921387, "learning_rate": 9.67016982522805e-05, "loss": 0.0666, "step": 3210 }, { "grad_norm": 0.6908206939697266, "learning_rate": 9.667210476077164e-05, "loss": 0.0606, "step": 3220 }, { "grad_norm": 0.5687106251716614, "learning_rate": 9.664238366951055e-05, "loss": 0.0599, "step": 3230 }, { "grad_norm": 0.563493549823761, "learning_rate": 9.661253505975355e-05, "loss": 0.0569, "step": 3240 }, { "grad_norm": 0.6353240609169006, "learning_rate": 9.658255901310557e-05, "loss": 0.0593, "step": 3250 }, { "grad_norm": 0.5277099609375, "learning_rate": 9.655245561152e-05, "loss": 0.0608, "step": 3260 }, { "grad_norm": 0.5423051118850708, "learning_rate": 9.65222249372984e-05, "loss": 0.0549, "step": 3270 }, { "grad_norm": 0.474439412355423, "learning_rate": 9.649186707309026e-05, "loss": 0.0592, "step": 3280 }, { "grad_norm": 0.5447307825088501, "learning_rate": 9.646138210189283e-05, "loss": 0.0573, "step": 3290 }, { "grad_norm": 0.46958911418914795, "learning_rate": 9.643077010705087e-05, "loss": 0.0648, "step": 3300 }, { "grad_norm": 0.6247355341911316, "learning_rate": 9.640003117225637e-05, "loss": 0.0649, "step": 3310 }, { "grad_norm": 0.622942328453064, "learning_rate": 9.636916538154846e-05, "loss": 0.0669, "step": 3320 }, { "grad_norm": 0.46045130491256714, "learning_rate": 9.633817281931296e-05, "loss": 0.0622, "step": 3330 }, { "grad_norm": 0.5312515497207642, "learning_rate": 9.630705357028242e-05, "loss": 0.0597, "step": 3340 }, { "grad_norm": 0.4561923146247864, "learning_rate": 9.627580771953563e-05, "loss": 0.0564, "step": 3350 }, { "grad_norm": 0.4700702428817749, "learning_rate": 9.624443535249759e-05, "loss": 0.0633, "step": 3360 }, { "grad_norm": 0.5625237226486206, "learning_rate": 9.621293655493913e-05, "loss": 0.0578, "step": 3370 }, { "grad_norm": 0.5612451434135437, "learning_rate": 9.618131141297675e-05, "loss": 0.0649, "step": 3380 }, { "grad_norm": 0.5531405210494995, "learning_rate": 9.614956001307242e-05, "loss": 0.0636, "step": 3390 }, { "grad_norm": 0.5203084945678711, "learning_rate": 9.611768244203321e-05, "loss": 0.0597, "step": 3400 }, { "grad_norm": 0.48514410853385925, "learning_rate": 9.60856787870112e-05, "loss": 0.0618, "step": 3410 }, { "grad_norm": 0.42436739802360535, "learning_rate": 9.605354913550318e-05, "loss": 0.0594, "step": 3420 }, { "grad_norm": 0.4707908630371094, "learning_rate": 9.602129357535037e-05, "loss": 0.0581, "step": 3430 }, { "grad_norm": 0.6007688045501709, "learning_rate": 9.598891219473825e-05, "loss": 0.0583, "step": 3440 }, { "grad_norm": 0.58519446849823, "learning_rate": 9.595640508219625e-05, "loss": 0.0582, "step": 3450 }, { "grad_norm": 0.6219605207443237, "learning_rate": 9.592377232659761e-05, "loss": 0.0607, "step": 3460 }, { "grad_norm": 0.5018323659896851, "learning_rate": 9.589101401715904e-05, "loss": 0.0551, "step": 3470 }, { "grad_norm": 0.5631371140480042, "learning_rate": 9.585813024344045e-05, "loss": 0.0579, "step": 3480 }, { "grad_norm": 0.7580733299255371, "learning_rate": 9.58251210953449e-05, "loss": 0.0571, "step": 3490 }, { "grad_norm": 0.6364933252334595, "learning_rate": 9.579198666311809e-05, "loss": 0.0545, "step": 3500 }, { "grad_norm": 0.6756724715232849, "learning_rate": 9.575872703734832e-05, "loss": 0.0589, "step": 3510 }, { "grad_norm": 0.4428686499595642, "learning_rate": 9.572534230896611e-05, "loss": 0.0565, "step": 3520 }, { "grad_norm": 0.5265988707542419, "learning_rate": 9.569183256924403e-05, "loss": 0.0531, "step": 3530 }, { "grad_norm": 0.6523367762565613, "learning_rate": 9.565819790979646e-05, "loss": 0.058, "step": 3540 }, { "grad_norm": 0.5738118886947632, "learning_rate": 9.562443842257925e-05, "loss": 0.0607, "step": 3550 }, { "grad_norm": 0.5877414345741272, "learning_rate": 9.559055419988956e-05, "loss": 0.0599, "step": 3560 }, { "grad_norm": 0.6353724598884583, "learning_rate": 9.555654533436557e-05, "loss": 0.0581, "step": 3570 }, { "grad_norm": 0.42666953802108765, "learning_rate": 9.552241191898621e-05, "loss": 0.0556, "step": 3580 }, { "grad_norm": 0.5157475471496582, "learning_rate": 9.548815404707092e-05, "loss": 0.0604, "step": 3590 }, { "grad_norm": 0.7215608954429626, "learning_rate": 9.545377181227942e-05, "loss": 0.0601, "step": 3600 }, { "grad_norm": 0.5162255764007568, "learning_rate": 9.541926530861145e-05, "loss": 0.055, "step": 3610 }, { "grad_norm": 0.4929755628108978, "learning_rate": 9.538463463040645e-05, "loss": 0.0624, "step": 3620 }, { "grad_norm": 0.4630160331726074, "learning_rate": 9.534987987234337e-05, "loss": 0.0531, "step": 3630 }, { "grad_norm": 0.6550790071487427, "learning_rate": 9.53150011294404e-05, "loss": 0.06, "step": 3640 }, { "grad_norm": 0.6224126815795898, "learning_rate": 9.527999849705471e-05, "loss": 0.058, "step": 3650 }, { "grad_norm": 0.5941552519798279, "learning_rate": 9.524487207088213e-05, "loss": 0.0605, "step": 3660 }, { "grad_norm": 0.5191814303398132, "learning_rate": 9.520962194695698e-05, "loss": 0.0554, "step": 3670 }, { "grad_norm": 0.5197587013244629, "learning_rate": 9.517424822165175e-05, "loss": 0.0604, "step": 3680 }, { "grad_norm": 0.6364939212799072, "learning_rate": 9.513875099167685e-05, "loss": 0.0607, "step": 3690 }, { "grad_norm": 0.6344776153564453, "learning_rate": 9.510313035408035e-05, "loss": 0.0591, "step": 3700 }, { "grad_norm": 0.5283899903297424, "learning_rate": 9.506738640624775e-05, "loss": 0.0548, "step": 3710 }, { "grad_norm": 0.5169802308082581, "learning_rate": 9.50315192459016e-05, "loss": 0.0568, "step": 3720 }, { "grad_norm": 0.5627011656761169, "learning_rate": 9.499552897110136e-05, "loss": 0.0575, "step": 3730 }, { "grad_norm": 0.5730915665626526, "learning_rate": 9.495941568024304e-05, "loss": 0.0544, "step": 3740 }, { "grad_norm": 0.5705208778381348, "learning_rate": 9.492317947205904e-05, "loss": 0.0554, "step": 3750 }, { "grad_norm": 0.48615023493766785, "learning_rate": 9.488682044561775e-05, "loss": 0.054, "step": 3760 }, { "grad_norm": 0.5147746205329895, "learning_rate": 9.485033870032335e-05, "loss": 0.0559, "step": 3770 }, { "grad_norm": 0.5804494023323059, "learning_rate": 9.481373433591556e-05, "loss": 0.0593, "step": 3780 }, { "grad_norm": 0.5439186096191406, "learning_rate": 9.47770074524693e-05, "loss": 0.0551, "step": 3790 }, { "grad_norm": 0.529241681098938, "learning_rate": 9.474015815039446e-05, "loss": 0.0579, "step": 3800 }, { "grad_norm": 0.42327943444252014, "learning_rate": 9.470318653043565e-05, "loss": 0.0551, "step": 3810 }, { "grad_norm": 0.4692992568016052, "learning_rate": 9.466609269367185e-05, "loss": 0.0577, "step": 3820 }, { "grad_norm": 0.6473419070243835, "learning_rate": 9.46288767415162e-05, "loss": 0.0609, "step": 3830 }, { "grad_norm": 0.624902069568634, "learning_rate": 9.459153877571567e-05, "loss": 0.0592, "step": 3840 }, { "grad_norm": 0.6024203896522522, "learning_rate": 9.455407889835087e-05, "loss": 0.0575, "step": 3850 }, { "grad_norm": 0.4566102623939514, "learning_rate": 9.451649721183564e-05, "loss": 0.0536, "step": 3860 }, { "grad_norm": 0.5603846907615662, "learning_rate": 9.447879381891692e-05, "loss": 0.0541, "step": 3870 }, { "grad_norm": 0.4703678488731384, "learning_rate": 9.444096882267428e-05, "loss": 0.0544, "step": 3880 }, { "grad_norm": 0.5888656377792358, "learning_rate": 9.440302232651988e-05, "loss": 0.0562, "step": 3890 }, { "grad_norm": 0.541254460811615, "learning_rate": 9.436495443419795e-05, "loss": 0.0531, "step": 3900 }, { "grad_norm": 0.48808667063713074, "learning_rate": 9.432676524978466e-05, "loss": 0.0515, "step": 3910 }, { "grad_norm": 0.6694585084915161, "learning_rate": 9.42884548776878e-05, "loss": 0.0552, "step": 3920 }, { "grad_norm": 0.683584451675415, "learning_rate": 9.425002342264646e-05, "loss": 0.0565, "step": 3930 }, { "grad_norm": 0.5609367489814758, "learning_rate": 9.421147098973077e-05, "loss": 0.0547, "step": 3940 }, { "grad_norm": 0.6370012164115906, "learning_rate": 9.41727976843416e-05, "loss": 0.0616, "step": 3950 }, { "grad_norm": 0.5923592448234558, "learning_rate": 9.413400361221029e-05, "loss": 0.0583, "step": 3960 }, { "grad_norm": 0.6023064255714417, "learning_rate": 9.409508887939835e-05, "loss": 0.0588, "step": 3970 }, { "grad_norm": 0.6249566078186035, "learning_rate": 9.40560535922972e-05, "loss": 0.0587, "step": 3980 }, { "grad_norm": 0.4997934103012085, "learning_rate": 9.40168978576278e-05, "loss": 0.0654, "step": 3990 }, { "grad_norm": 0.5568627119064331, "learning_rate": 9.397762178244043e-05, "loss": 0.055, "step": 4000 } ], "logging_steps": 10, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 9223372036854775807, "save_steps": 4000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 128, "trial_name": null, "trial_params": null }