{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.46643603995802074, "eval_steps": 500, "global_step": 1500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 0.47774845361709595, "learning_rate": 4.999970160815579e-05, "loss": 2.0765, "step": 5 }, { "epoch": 0.0, "grad_norm": 0.6051416397094727, "learning_rate": 4.999880643974619e-05, "loss": 2.2297, "step": 10 }, { "epoch": 0.0, "grad_norm": 0.6161717772483826, "learning_rate": 4.9997314516140056e-05, "loss": 2.1103, "step": 15 }, { "epoch": 0.01, "grad_norm": 0.4686434268951416, "learning_rate": 4.999522587295162e-05, "loss": 2.0057, "step": 20 }, { "epoch": 0.01, "grad_norm": 0.8412289023399353, "learning_rate": 4.999254056003963e-05, "loss": 2.1778, "step": 25 }, { "epoch": 0.01, "grad_norm": 0.5333625078201294, "learning_rate": 4.99892586415061e-05, "loss": 2.2399, "step": 30 }, { "epoch": 0.01, "grad_norm": 0.821148157119751, "learning_rate": 4.9985380195694856e-05, "loss": 2.3215, "step": 35 }, { "epoch": 0.01, "grad_norm": 0.8403909206390381, "learning_rate": 4.998090531518962e-05, "loss": 1.8295, "step": 40 }, { "epoch": 0.01, "grad_norm": 0.6633398532867432, "learning_rate": 4.9975834106811834e-05, "loss": 2.0195, "step": 45 }, { "epoch": 0.02, "grad_norm": 0.6386868357658386, "learning_rate": 4.997016669161806e-05, "loss": 2.1257, "step": 50 }, { "epoch": 0.02, "grad_norm": 0.7762248516082764, "learning_rate": 4.996390320489715e-05, "loss": 2.057, "step": 55 }, { "epoch": 0.02, "grad_norm": 1.3192856311798096, "learning_rate": 4.9957043796166966e-05, "loss": 2.0753, "step": 60 }, { "epoch": 0.02, "grad_norm": 0.9797518849372864, "learning_rate": 4.994958862917083e-05, "loss": 1.9736, "step": 65 }, { "epoch": 0.02, "grad_norm": 1.000693440437317, "learning_rate": 4.994153788187363e-05, "loss": 2.1572, "step": 70 }, { "epoch": 0.02, "grad_norm": 0.6852813959121704, "learning_rate": 4.993289174645757e-05, "loss": 2.1491, "step": 75 }, { "epoch": 0.02, "grad_norm": 1.0075691938400269, "learning_rate": 4.992365042931752e-05, "loss": 1.945, "step": 80 }, { "epoch": 0.03, "grad_norm": 1.1973133087158203, "learning_rate": 4.991381415105619e-05, "loss": 2.0811, "step": 85 }, { "epoch": 0.03, "grad_norm": 0.9927239418029785, "learning_rate": 4.990338314647881e-05, "loss": 1.961, "step": 90 }, { "epoch": 0.03, "grad_norm": 0.9499759674072266, "learning_rate": 4.98923576645875e-05, "loss": 2.0653, "step": 95 }, { "epoch": 0.03, "grad_norm": 0.7233040928840637, "learning_rate": 4.9880737968575365e-05, "loss": 1.9999, "step": 100 }, { "epoch": 0.03, "grad_norm": 1.55235755443573, "learning_rate": 4.986852433582022e-05, "loss": 2.2258, "step": 105 }, { "epoch": 0.03, "grad_norm": 0.9007890820503235, "learning_rate": 4.985571705787793e-05, "loss": 2.1034, "step": 110 }, { "epoch": 0.04, "grad_norm": 0.6774860620498657, "learning_rate": 4.9842316440475475e-05, "loss": 2.1753, "step": 115 }, { "epoch": 0.04, "grad_norm": 0.7676737308502197, "learning_rate": 4.9828322803503665e-05, "loss": 2.1384, "step": 120 }, { "epoch": 0.04, "grad_norm": 0.9624544978141785, "learning_rate": 4.981373648100946e-05, "loss": 2.0521, "step": 125 }, { "epoch": 0.04, "grad_norm": 0.9315722584724426, "learning_rate": 4.979855782118802e-05, "loss": 1.9256, "step": 130 }, { "epoch": 0.04, "grad_norm": 0.9035864472389221, "learning_rate": 4.978278718637443e-05, "loss": 2.0882, "step": 135 }, { "epoch": 0.04, "grad_norm": 0.7997236251831055, "learning_rate": 4.9766424953035e-05, "loss": 2.0724, "step": 140 }, { "epoch": 0.05, "grad_norm": 1.0692921876907349, "learning_rate": 4.974947151175826e-05, "loss": 2.1329, "step": 145 }, { "epoch": 0.05, "grad_norm": 0.9506180286407471, "learning_rate": 4.973192726724572e-05, "loss": 2.082, "step": 150 }, { "epoch": 0.05, "grad_norm": 0.8647387027740479, "learning_rate": 4.9713792638302145e-05, "loss": 2.0366, "step": 155 }, { "epoch": 0.05, "grad_norm": 1.105302095413208, "learning_rate": 4.969506805782555e-05, "loss": 2.1481, "step": 160 }, { "epoch": 0.05, "grad_norm": 0.7593303918838501, "learning_rate": 4.967575397279689e-05, "loss": 2.032, "step": 165 }, { "epoch": 0.05, "grad_norm": 0.7521979808807373, "learning_rate": 4.965585084426943e-05, "loss": 2.0379, "step": 170 }, { "epoch": 0.05, "grad_norm": 0.947120726108551, "learning_rate": 4.9635359147357655e-05, "loss": 2.1444, "step": 175 }, { "epoch": 0.06, "grad_norm": 1.2184454202651978, "learning_rate": 4.961427937122598e-05, "loss": 1.9164, "step": 180 }, { "epoch": 0.06, "grad_norm": 1.221663475036621, "learning_rate": 4.959261201907707e-05, "loss": 2.0084, "step": 185 }, { "epoch": 0.06, "grad_norm": 1.0457361936569214, "learning_rate": 4.957035760813982e-05, "loss": 2.2032, "step": 190 }, { "epoch": 0.06, "grad_norm": 0.8834909200668335, "learning_rate": 4.954751666965701e-05, "loss": 2.2101, "step": 195 }, { "epoch": 0.06, "grad_norm": 0.791902482509613, "learning_rate": 4.9524089748872615e-05, "loss": 2.0472, "step": 200 }, { "epoch": 0.06, "grad_norm": 1.2905739545822144, "learning_rate": 4.9500077405018807e-05, "loss": 2.0987, "step": 205 }, { "epoch": 0.07, "grad_norm": 0.8612006306648254, "learning_rate": 4.9475480211302583e-05, "loss": 2.1765, "step": 210 }, { "epoch": 0.07, "grad_norm": 1.3128459453582764, "learning_rate": 4.945029875489212e-05, "loss": 1.9926, "step": 215 }, { "epoch": 0.07, "grad_norm": 0.9610918164253235, "learning_rate": 4.94245336369027e-05, "loss": 2.0124, "step": 220 }, { "epoch": 0.07, "grad_norm": 0.873160183429718, "learning_rate": 4.939818547238241e-05, "loss": 2.2229, "step": 225 }, { "epoch": 0.07, "grad_norm": 1.5535285472869873, "learning_rate": 4.9371254890297446e-05, "loss": 2.2013, "step": 230 }, { "epoch": 0.07, "grad_norm": 1.1951836347579956, "learning_rate": 4.93437425335171e-05, "loss": 2.014, "step": 235 }, { "epoch": 0.07, "grad_norm": 0.7874170541763306, "learning_rate": 4.9315649058798384e-05, "loss": 2.1701, "step": 240 }, { "epoch": 0.08, "grad_norm": 1.3503323793411255, "learning_rate": 4.928697513677042e-05, "loss": 2.1681, "step": 245 }, { "epoch": 0.08, "grad_norm": 1.3091179132461548, "learning_rate": 4.925772145191834e-05, "loss": 2.1224, "step": 250 }, { "epoch": 0.08, "grad_norm": 1.4428555965423584, "learning_rate": 4.9227888702567044e-05, "loss": 2.0512, "step": 255 }, { "epoch": 0.08, "grad_norm": 0.8234395980834961, "learning_rate": 4.9197477600864446e-05, "loss": 2.1067, "step": 260 }, { "epoch": 0.08, "grad_norm": 1.9094969034194946, "learning_rate": 4.9166488872764526e-05, "loss": 1.8884, "step": 265 }, { "epoch": 0.08, "grad_norm": 1.0074087381362915, "learning_rate": 4.913492325800999e-05, "loss": 1.9345, "step": 270 }, { "epoch": 0.09, "grad_norm": 1.0867297649383545, "learning_rate": 4.910278151011458e-05, "loss": 2.1928, "step": 275 }, { "epoch": 0.09, "grad_norm": 0.6842357516288757, "learning_rate": 4.907006439634516e-05, "loss": 2.0407, "step": 280 }, { "epoch": 0.09, "grad_norm": 0.8409023284912109, "learning_rate": 4.903677269770329e-05, "loss": 2.2344, "step": 285 }, { "epoch": 0.09, "grad_norm": 0.8119503259658813, "learning_rate": 4.900290720890671e-05, "loss": 2.1296, "step": 290 }, { "epoch": 0.09, "grad_norm": 0.9938147068023682, "learning_rate": 4.8968468738370244e-05, "loss": 2.152, "step": 295 }, { "epoch": 0.09, "grad_norm": 0.9865244030952454, "learning_rate": 4.8933458108186606e-05, "loss": 1.9623, "step": 300 }, { "epoch": 0.09, "grad_norm": 1.3944802284240723, "learning_rate": 4.889787615410672e-05, "loss": 1.915, "step": 305 }, { "epoch": 0.1, "grad_norm": 1.3749767541885376, "learning_rate": 4.886172372551977e-05, "loss": 1.9934, "step": 310 }, { "epoch": 0.1, "grad_norm": 0.9024938941001892, "learning_rate": 4.882500168543294e-05, "loss": 2.1541, "step": 315 }, { "epoch": 0.1, "grad_norm": 1.1978263854980469, "learning_rate": 4.878771091045082e-05, "loss": 2.1688, "step": 320 }, { "epoch": 0.1, "grad_norm": 0.8360010981559753, "learning_rate": 4.874985229075446e-05, "loss": 2.1387, "step": 325 }, { "epoch": 0.1, "grad_norm": 0.7683364152908325, "learning_rate": 4.871142673008012e-05, "loss": 2.0215, "step": 330 }, { "epoch": 0.1, "grad_norm": 1.4230670928955078, "learning_rate": 4.867243514569772e-05, "loss": 1.9491, "step": 335 }, { "epoch": 0.11, "grad_norm": 0.8198773860931396, "learning_rate": 4.863287846838891e-05, "loss": 2.0151, "step": 340 }, { "epoch": 0.11, "grad_norm": 1.467207908630371, "learning_rate": 4.85927576424249e-05, "loss": 1.8906, "step": 345 }, { "epoch": 0.11, "grad_norm": 0.9537095427513123, "learning_rate": 4.855207362554385e-05, "loss": 2.1844, "step": 350 }, { "epoch": 0.11, "grad_norm": 1.0757155418395996, "learning_rate": 4.851082738892809e-05, "loss": 2.048, "step": 355 }, { "epoch": 0.11, "grad_norm": 1.6884938478469849, "learning_rate": 4.8469019917180846e-05, "loss": 1.9537, "step": 360 }, { "epoch": 0.11, "grad_norm": 1.4680182933807373, "learning_rate": 4.8426652208302814e-05, "loss": 1.9731, "step": 365 }, { "epoch": 0.12, "grad_norm": 1.1778632402420044, "learning_rate": 4.83837252736683e-05, "loss": 2.1395, "step": 370 }, { "epoch": 0.12, "grad_norm": 1.2865056991577148, "learning_rate": 4.834024013800108e-05, "loss": 2.0016, "step": 375 }, { "epoch": 0.12, "grad_norm": 1.055177092552185, "learning_rate": 4.8296197839349944e-05, "loss": 1.9632, "step": 380 }, { "epoch": 0.12, "grad_norm": 1.0041871070861816, "learning_rate": 4.825159942906389e-05, "loss": 2.3302, "step": 385 }, { "epoch": 0.12, "grad_norm": 1.0026438236236572, "learning_rate": 4.820644597176709e-05, "loss": 2.1517, "step": 390 }, { "epoch": 0.12, "grad_norm": 1.3532180786132812, "learning_rate": 4.81607385453334e-05, "loss": 2.1229, "step": 395 }, { "epoch": 0.12, "grad_norm": 0.7670988440513611, "learning_rate": 4.81144782408607e-05, "loss": 2.1382, "step": 400 }, { "epoch": 0.13, "grad_norm": 1.0405700206756592, "learning_rate": 4.8067666162644774e-05, "loss": 1.9614, "step": 405 }, { "epoch": 0.13, "grad_norm": 1.2252662181854248, "learning_rate": 4.802030342815304e-05, "loss": 2.1399, "step": 410 }, { "epoch": 0.13, "grad_norm": 1.237946629524231, "learning_rate": 4.7972391167997754e-05, "loss": 1.9034, "step": 415 }, { "epoch": 0.13, "grad_norm": 0.8064705729484558, "learning_rate": 4.7923930525909156e-05, "loss": 2.0075, "step": 420 }, { "epoch": 0.13, "grad_norm": 0.8717565536499023, "learning_rate": 4.7874922658708065e-05, "loss": 2.0105, "step": 425 }, { "epoch": 0.13, "grad_norm": 1.6693098545074463, "learning_rate": 4.782536873627832e-05, "loss": 2.0242, "step": 430 }, { "epoch": 0.14, "grad_norm": 0.82447350025177, "learning_rate": 4.777526994153882e-05, "loss": 2.0267, "step": 435 }, { "epoch": 0.14, "grad_norm": 0.9926588535308838, "learning_rate": 4.7724627470415307e-05, "loss": 1.9119, "step": 440 }, { "epoch": 0.14, "grad_norm": 1.0924450159072876, "learning_rate": 4.7673442531811796e-05, "loss": 2.2653, "step": 445 }, { "epoch": 0.14, "grad_norm": 1.1592103242874146, "learning_rate": 4.762171634758177e-05, "loss": 2.0017, "step": 450 }, { "epoch": 0.14, "grad_norm": 0.9172110557556152, "learning_rate": 4.7569450152498927e-05, "loss": 2.1408, "step": 455 }, { "epoch": 0.14, "grad_norm": 1.1897525787353516, "learning_rate": 4.751664519422778e-05, "loss": 2.0935, "step": 460 }, { "epoch": 0.14, "grad_norm": 0.8793094158172607, "learning_rate": 4.746330273329386e-05, "loss": 2.1142, "step": 465 }, { "epoch": 0.15, "grad_norm": 1.4337489604949951, "learning_rate": 4.740942404305356e-05, "loss": 2.1289, "step": 470 }, { "epoch": 0.15, "grad_norm": 1.0251764059066772, "learning_rate": 4.735501040966383e-05, "loss": 1.9741, "step": 475 }, { "epoch": 0.15, "grad_norm": 1.2659822702407837, "learning_rate": 4.730006313205143e-05, "loss": 2.088, "step": 480 }, { "epoch": 0.15, "grad_norm": 0.8884140849113464, "learning_rate": 4.724458352188192e-05, "loss": 2.2079, "step": 485 }, { "epoch": 0.15, "grad_norm": 1.1937768459320068, "learning_rate": 4.718857290352835e-05, "loss": 2.048, "step": 490 }, { "epoch": 0.15, "grad_norm": 0.9741552472114563, "learning_rate": 4.713203261403966e-05, "loss": 2.2569, "step": 495 }, { "epoch": 0.16, "grad_norm": 0.7996780872344971, "learning_rate": 4.707496400310874e-05, "loss": 1.9574, "step": 500 }, { "epoch": 0.16, "grad_norm": 1.8182051181793213, "learning_rate": 4.701736843304025e-05, "loss": 2.0951, "step": 505 }, { "epoch": 0.16, "grad_norm": 1.507320761680603, "learning_rate": 4.695924727871805e-05, "loss": 2.0253, "step": 510 }, { "epoch": 0.16, "grad_norm": 0.759121835231781, "learning_rate": 4.690060192757242e-05, "loss": 2.0602, "step": 515 }, { "epoch": 0.16, "grad_norm": 1.5943195819854736, "learning_rate": 4.684143377954691e-05, "loss": 2.0386, "step": 520 }, { "epoch": 0.16, "grad_norm": 0.8568710088729858, "learning_rate": 4.6781744247064955e-05, "loss": 2.073, "step": 525 }, { "epoch": 0.16, "grad_norm": 1.3352620601654053, "learning_rate": 4.6721534754996125e-05, "loss": 2.1443, "step": 530 }, { "epoch": 0.17, "grad_norm": 1.3417474031448364, "learning_rate": 4.666080674062213e-05, "loss": 2.0288, "step": 535 }, { "epoch": 0.17, "grad_norm": 1.5334464311599731, "learning_rate": 4.659956165360251e-05, "loss": 2.0609, "step": 540 }, { "epoch": 0.17, "grad_norm": 0.9658721089363098, "learning_rate": 4.6537800955940005e-05, "loss": 1.9539, "step": 545 }, { "epoch": 0.17, "grad_norm": 1.9197947978973389, "learning_rate": 4.647552612194572e-05, "loss": 2.149, "step": 550 }, { "epoch": 0.17, "grad_norm": 0.8512137532234192, "learning_rate": 4.641273863820383e-05, "loss": 1.9722, "step": 555 }, { "epoch": 0.17, "grad_norm": 1.827289342880249, "learning_rate": 4.634944000353622e-05, "loss": 2.0729, "step": 560 }, { "epoch": 0.18, "grad_norm": 1.088416337966919, "learning_rate": 4.628563172896655e-05, "loss": 1.9507, "step": 565 }, { "epoch": 0.18, "grad_norm": 1.3566908836364746, "learning_rate": 4.6221315337684353e-05, "loss": 2.1643, "step": 570 }, { "epoch": 0.18, "grad_norm": 1.3541293144226074, "learning_rate": 4.615649236500854e-05, "loss": 2.1839, "step": 575 }, { "epoch": 0.18, "grad_norm": 0.991269588470459, "learning_rate": 4.609116435835083e-05, "loss": 2.0976, "step": 580 }, { "epoch": 0.18, "grad_norm": 1.0280535221099854, "learning_rate": 4.602533287717877e-05, "loss": 2.1474, "step": 585 }, { "epoch": 0.18, "grad_norm": 1.013123631477356, "learning_rate": 4.5958999492978524e-05, "loss": 2.1873, "step": 590 }, { "epoch": 0.19, "grad_norm": 1.1753040552139282, "learning_rate": 4.589216578921737e-05, "loss": 2.1744, "step": 595 }, { "epoch": 0.19, "grad_norm": 1.1839090585708618, "learning_rate": 4.582483336130586e-05, "loss": 1.9982, "step": 600 }, { "epoch": 0.19, "grad_norm": 1.0724798440933228, "learning_rate": 4.575700381655979e-05, "loss": 2.1234, "step": 605 }, { "epoch": 0.19, "grad_norm": 2.009913682937622, "learning_rate": 4.5688678774161796e-05, "loss": 1.9478, "step": 610 }, { "epoch": 0.19, "grad_norm": 0.9897060394287109, "learning_rate": 4.561985986512271e-05, "loss": 1.8268, "step": 615 }, { "epoch": 0.19, "grad_norm": 0.8881808519363403, "learning_rate": 4.555054873224263e-05, "loss": 1.9887, "step": 620 }, { "epoch": 0.19, "grad_norm": 1.155900001525879, "learning_rate": 4.54807470300717e-05, "loss": 2.0777, "step": 625 }, { "epoch": 0.2, "grad_norm": 0.8782421350479126, "learning_rate": 4.5410456424870596e-05, "loss": 2.0566, "step": 630 }, { "epoch": 0.2, "grad_norm": 1.3324674367904663, "learning_rate": 4.5339678594570795e-05, "loss": 2.047, "step": 635 }, { "epoch": 0.2, "grad_norm": 1.9805939197540283, "learning_rate": 4.526841522873449e-05, "loss": 1.962, "step": 640 }, { "epoch": 0.2, "grad_norm": 1.4999943971633911, "learning_rate": 4.519666802851422e-05, "loss": 2.0972, "step": 645 }, { "epoch": 0.2, "grad_norm": 1.4504961967468262, "learning_rate": 4.5124438706612376e-05, "loss": 2.0041, "step": 650 }, { "epoch": 0.2, "grad_norm": 0.9078169465065002, "learning_rate": 4.505172898724018e-05, "loss": 2.1229, "step": 655 }, { "epoch": 0.21, "grad_norm": 1.1635804176330566, "learning_rate": 4.497854060607662e-05, "loss": 2.0195, "step": 660 }, { "epoch": 0.21, "grad_norm": 1.46576726436615, "learning_rate": 4.490487531022699e-05, "loss": 2.0745, "step": 665 }, { "epoch": 0.21, "grad_norm": 1.2094652652740479, "learning_rate": 4.4830734858181145e-05, "loss": 2.1068, "step": 670 }, { "epoch": 0.21, "grad_norm": 1.4738895893096924, "learning_rate": 4.47561210197716e-05, "loss": 1.8088, "step": 675 }, { "epoch": 0.21, "grad_norm": 1.23384690284729, "learning_rate": 4.4681035576131215e-05, "loss": 2.0995, "step": 680 }, { "epoch": 0.21, "grad_norm": 0.8332946300506592, "learning_rate": 4.46054803196507e-05, "loss": 2.0541, "step": 685 }, { "epoch": 0.21, "grad_norm": 0.9207485318183899, "learning_rate": 4.452945705393586e-05, "loss": 2.166, "step": 690 }, { "epoch": 0.22, "grad_norm": 1.292945146560669, "learning_rate": 4.445296759376449e-05, "loss": 2.0784, "step": 695 }, { "epoch": 0.22, "grad_norm": 0.9874763488769531, "learning_rate": 4.437601376504307e-05, "loss": 2.2087, "step": 700 }, { "epoch": 0.22, "grad_norm": 0.9427415132522583, "learning_rate": 4.4298597404763186e-05, "loss": 2.1199, "step": 705 }, { "epoch": 0.22, "grad_norm": 1.7369529008865356, "learning_rate": 4.422072036095768e-05, "loss": 2.0355, "step": 710 }, { "epoch": 0.22, "grad_norm": 1.2423696517944336, "learning_rate": 4.414238449265654e-05, "loss": 2.0011, "step": 715 }, { "epoch": 0.22, "grad_norm": 1.2304831743240356, "learning_rate": 4.406359166984249e-05, "loss": 2.0368, "step": 720 }, { "epoch": 0.23, "grad_norm": 0.9090413451194763, "learning_rate": 4.39843437734064e-05, "loss": 1.9983, "step": 725 }, { "epoch": 0.23, "grad_norm": 1.2729507684707642, "learning_rate": 4.390464269510233e-05, "loss": 2.021, "step": 730 }, { "epoch": 0.23, "grad_norm": 1.3009227514266968, "learning_rate": 4.382449033750244e-05, "loss": 1.9743, "step": 735 }, { "epoch": 0.23, "grad_norm": 1.5456056594848633, "learning_rate": 4.37438886139515e-05, "loss": 2.0689, "step": 740 }, { "epoch": 0.23, "grad_norm": 1.3235007524490356, "learning_rate": 4.3662839448521264e-05, "loss": 2.0838, "step": 745 }, { "epoch": 0.23, "grad_norm": 2.2074007987976074, "learning_rate": 4.358134477596454e-05, "loss": 2.0835, "step": 750 }, { "epoch": 0.23, "grad_norm": 1.403738021850586, "learning_rate": 4.3499406541668966e-05, "loss": 2.0916, "step": 755 }, { "epoch": 0.24, "grad_norm": 1.0940325260162354, "learning_rate": 4.3417026701610616e-05, "loss": 1.972, "step": 760 }, { "epoch": 0.24, "grad_norm": 1.666353702545166, "learning_rate": 4.3334207222307275e-05, "loss": 1.927, "step": 765 }, { "epoch": 0.24, "grad_norm": 1.0777515172958374, "learning_rate": 4.325095008077154e-05, "loss": 2.1192, "step": 770 }, { "epoch": 0.24, "grad_norm": 1.7218186855316162, "learning_rate": 4.316725726446353e-05, "loss": 2.0774, "step": 775 }, { "epoch": 0.24, "grad_norm": 1.356753945350647, "learning_rate": 4.3083130771243586e-05, "loss": 2.0847, "step": 780 }, { "epoch": 0.24, "grad_norm": 0.9967429637908936, "learning_rate": 4.299857260932445e-05, "loss": 2.0485, "step": 785 }, { "epoch": 0.25, "grad_norm": 1.6216442584991455, "learning_rate": 4.2913584797223397e-05, "loss": 2.1008, "step": 790 }, { "epoch": 0.25, "grad_norm": 1.2556742429733276, "learning_rate": 4.2828169363714016e-05, "loss": 1.9209, "step": 795 }, { "epoch": 0.25, "grad_norm": 1.1800439357757568, "learning_rate": 4.274232834777782e-05, "loss": 1.9722, "step": 800 }, { "epoch": 0.25, "grad_norm": 1.1313499212265015, "learning_rate": 4.2656063798555515e-05, "loss": 1.9176, "step": 805 }, { "epoch": 0.25, "grad_norm": 1.137534737586975, "learning_rate": 4.256937777529815e-05, "loss": 1.9929, "step": 810 }, { "epoch": 0.25, "grad_norm": 1.0575093030929565, "learning_rate": 4.2482272347317906e-05, "loss": 2.166, "step": 815 }, { "epoch": 0.25, "grad_norm": 1.5939594507217407, "learning_rate": 4.2394749593938733e-05, "loss": 2.1334, "step": 820 }, { "epoch": 0.26, "grad_norm": 1.1045507192611694, "learning_rate": 4.230681160444669e-05, "loss": 2.0853, "step": 825 }, { "epoch": 0.26, "grad_norm": 1.3480136394500732, "learning_rate": 4.221846047804009e-05, "loss": 2.1802, "step": 830 }, { "epoch": 0.26, "grad_norm": 1.1822657585144043, "learning_rate": 4.2129698323779366e-05, "loss": 2.0739, "step": 835 }, { "epoch": 0.26, "grad_norm": 1.1771117448806763, "learning_rate": 4.204052726053676e-05, "loss": 2.0238, "step": 840 }, { "epoch": 0.26, "grad_norm": 1.4757814407348633, "learning_rate": 4.195094941694571e-05, "loss": 2.1557, "step": 845 }, { "epoch": 0.26, "grad_norm": 0.9095075726509094, "learning_rate": 4.1860966931350054e-05, "loss": 2.1666, "step": 850 }, { "epoch": 0.27, "grad_norm": 1.1039543151855469, "learning_rate": 4.1770581951752976e-05, "loss": 2.105, "step": 855 }, { "epoch": 0.27, "grad_norm": 0.8517205119132996, "learning_rate": 4.1679796635765735e-05, "loss": 1.9656, "step": 860 }, { "epoch": 0.27, "grad_norm": 1.239492654800415, "learning_rate": 4.158861315055617e-05, "loss": 2.0166, "step": 865 }, { "epoch": 0.27, "grad_norm": 1.1358321905136108, "learning_rate": 4.1497033672796924e-05, "loss": 2.0076, "step": 870 }, { "epoch": 0.27, "grad_norm": 1.6215249300003052, "learning_rate": 4.140506038861356e-05, "loss": 2.1594, "step": 875 }, { "epoch": 0.27, "grad_norm": 1.0528080463409424, "learning_rate": 4.131269549353229e-05, "loss": 2.1416, "step": 880 }, { "epoch": 0.28, "grad_norm": 0.8976901769638062, "learning_rate": 4.1219941192427644e-05, "loss": 2.1242, "step": 885 }, { "epoch": 0.28, "grad_norm": 1.263594388961792, "learning_rate": 4.112679969946977e-05, "loss": 2.02, "step": 890 }, { "epoch": 0.28, "grad_norm": 1.4173017740249634, "learning_rate": 4.103327323807162e-05, "loss": 2.0438, "step": 895 }, { "epoch": 0.28, "grad_norm": 1.876170039176941, "learning_rate": 4.093936404083585e-05, "loss": 1.9806, "step": 900 }, { "epoch": 0.28, "grad_norm": 1.4649231433868408, "learning_rate": 4.0845074349501544e-05, "loss": 2.1476, "step": 905 }, { "epoch": 0.28, "grad_norm": 1.0446043014526367, "learning_rate": 4.0750406414890695e-05, "loss": 1.9672, "step": 910 }, { "epoch": 0.28, "grad_norm": 1.0225305557250977, "learning_rate": 4.065536249685448e-05, "loss": 1.9984, "step": 915 }, { "epoch": 0.29, "grad_norm": 1.0120617151260376, "learning_rate": 4.055994486421929e-05, "loss": 2.1162, "step": 920 }, { "epoch": 0.29, "grad_norm": 1.0469881296157837, "learning_rate": 4.04641557947326e-05, "loss": 2.0435, "step": 925 }, { "epoch": 0.29, "grad_norm": 1.2435941696166992, "learning_rate": 4.036799757500856e-05, "loss": 2.0431, "step": 930 }, { "epoch": 0.29, "grad_norm": 1.0055103302001953, "learning_rate": 4.027147250047348e-05, "loss": 2.2021, "step": 935 }, { "epoch": 0.29, "grad_norm": 1.1212949752807617, "learning_rate": 4.017458287531094e-05, "loss": 1.997, "step": 940 }, { "epoch": 0.29, "grad_norm": 1.1048357486724854, "learning_rate": 4.007733101240685e-05, "loss": 1.946, "step": 945 }, { "epoch": 0.3, "grad_norm": 1.4721689224243164, "learning_rate": 3.997971923329426e-05, "loss": 2.0723, "step": 950 }, { "epoch": 0.3, "grad_norm": 1.3793156147003174, "learning_rate": 3.988174986809783e-05, "loss": 2.034, "step": 955 }, { "epoch": 0.3, "grad_norm": 0.9013482928276062, "learning_rate": 3.9783425255478355e-05, "loss": 1.9736, "step": 960 }, { "epoch": 0.3, "grad_norm": 0.9192422032356262, "learning_rate": 3.968474774257682e-05, "loss": 1.9878, "step": 965 }, { "epoch": 0.3, "grad_norm": 1.9304206371307373, "learning_rate": 3.9585719684958446e-05, "loss": 2.117, "step": 970 }, { "epoch": 0.3, "grad_norm": 1.0435137748718262, "learning_rate": 3.948634344655639e-05, "loss": 2.0585, "step": 975 }, { "epoch": 0.3, "grad_norm": 1.4636590480804443, "learning_rate": 3.938662139961538e-05, "loss": 2.0409, "step": 980 }, { "epoch": 0.31, "grad_norm": 1.8014529943466187, "learning_rate": 3.928655592463508e-05, "loss": 2.0369, "step": 985 }, { "epoch": 0.31, "grad_norm": 1.2412620782852173, "learning_rate": 3.918614941031319e-05, "loss": 1.967, "step": 990 }, { "epoch": 0.31, "grad_norm": 1.3581103086471558, "learning_rate": 3.908540425348852e-05, "loss": 2.0037, "step": 995 }, { "epoch": 0.31, "grad_norm": 1.2377780675888062, "learning_rate": 3.8984322859083725e-05, "loss": 1.9991, "step": 1000 }, { "epoch": 0.31, "grad_norm": 0.9209259748458862, "learning_rate": 3.8882907640047896e-05, "loss": 2.0448, "step": 1005 }, { "epoch": 0.31, "grad_norm": 1.0150959491729736, "learning_rate": 3.878116101729897e-05, "loss": 2.0791, "step": 1010 }, { "epoch": 0.32, "grad_norm": 1.5959141254425049, "learning_rate": 3.867908541966594e-05, "loss": 1.9997, "step": 1015 }, { "epoch": 0.32, "grad_norm": 1.3945012092590332, "learning_rate": 3.857668328383088e-05, "loss": 2.0481, "step": 1020 }, { "epoch": 0.32, "grad_norm": 1.2361671924591064, "learning_rate": 3.847395705427075e-05, "loss": 2.2664, "step": 1025 }, { "epoch": 0.32, "grad_norm": 1.9661719799041748, "learning_rate": 3.837090918319909e-05, "loss": 1.9752, "step": 1030 }, { "epoch": 0.32, "grad_norm": 1.6995949745178223, "learning_rate": 3.8267542130507436e-05, "loss": 2.1332, "step": 1035 }, { "epoch": 0.32, "grad_norm": 1.1248412132263184, "learning_rate": 3.816385836370663e-05, "loss": 2.0432, "step": 1040 }, { "epoch": 0.32, "grad_norm": 0.8734235763549805, "learning_rate": 3.805986035786789e-05, "loss": 1.9618, "step": 1045 }, { "epoch": 0.33, "grad_norm": 1.322766661643982, "learning_rate": 3.795555059556378e-05, "loss": 2.0267, "step": 1050 }, { "epoch": 0.33, "grad_norm": 1.0396028757095337, "learning_rate": 3.7850931566808866e-05, "loss": 2.1075, "step": 1055 }, { "epoch": 0.33, "grad_norm": 0.9574625492095947, "learning_rate": 3.7746005769000363e-05, "loss": 2.156, "step": 1060 }, { "epoch": 0.33, "grad_norm": 1.4480133056640625, "learning_rate": 3.764077570685844e-05, "loss": 1.9615, "step": 1065 }, { "epoch": 0.33, "grad_norm": 1.5908560752868652, "learning_rate": 3.753524389236648e-05, "loss": 2.0928, "step": 1070 }, { "epoch": 0.33, "grad_norm": 1.2628813982009888, "learning_rate": 3.742941284471111e-05, "loss": 2.1074, "step": 1075 }, { "epoch": 0.34, "grad_norm": 1.2687503099441528, "learning_rate": 3.7323285090222054e-05, "loss": 1.9666, "step": 1080 }, { "epoch": 0.34, "grad_norm": 1.2571731805801392, "learning_rate": 3.721686316231181e-05, "loss": 2.0468, "step": 1085 }, { "epoch": 0.34, "grad_norm": 1.007453441619873, "learning_rate": 3.7110149601415215e-05, "loss": 2.0624, "step": 1090 }, { "epoch": 0.34, "grad_norm": 1.2390377521514893, "learning_rate": 3.700314695492876e-05, "loss": 1.9888, "step": 1095 }, { "epoch": 0.34, "grad_norm": 1.0878371000289917, "learning_rate": 3.6895857777149825e-05, "loss": 2.1013, "step": 1100 }, { "epoch": 0.34, "grad_norm": 0.8759217262268066, "learning_rate": 3.6788284629215624e-05, "loss": 1.875, "step": 1105 }, { "epoch": 0.35, "grad_norm": 1.1345970630645752, "learning_rate": 3.668043007904219e-05, "loss": 1.9096, "step": 1110 }, { "epoch": 0.35, "grad_norm": 1.253629446029663, "learning_rate": 3.6572296701262966e-05, "loss": 2.1859, "step": 1115 }, { "epoch": 0.35, "grad_norm": 0.9796190857887268, "learning_rate": 3.646388707716738e-05, "loss": 2.2092, "step": 1120 }, { "epoch": 0.35, "grad_norm": 1.3893767595291138, "learning_rate": 3.635520379463926e-05, "loss": 2.0026, "step": 1125 }, { "epoch": 0.35, "grad_norm": 0.8778309226036072, "learning_rate": 3.6246249448095004e-05, "loss": 2.2112, "step": 1130 }, { "epoch": 0.35, "grad_norm": 1.2479698657989502, "learning_rate": 3.6137026638421696e-05, "loss": 2.0221, "step": 1135 }, { "epoch": 0.35, "grad_norm": 1.3813824653625488, "learning_rate": 3.6027537972914974e-05, "loss": 1.9106, "step": 1140 }, { "epoch": 0.36, "grad_norm": 1.2043218612670898, "learning_rate": 3.5917786065216826e-05, "loss": 2.0673, "step": 1145 }, { "epoch": 0.36, "grad_norm": 1.5337340831756592, "learning_rate": 3.580777353525318e-05, "loss": 2.1463, "step": 1150 }, { "epoch": 0.36, "grad_norm": 1.155813455581665, "learning_rate": 3.5697503009171385e-05, "loss": 2.0255, "step": 1155 }, { "epoch": 0.36, "grad_norm": 1.034644365310669, "learning_rate": 3.558697711927748e-05, "loss": 2.1348, "step": 1160 }, { "epoch": 0.36, "grad_norm": 1.0959795713424683, "learning_rate": 3.54761985039734e-05, "loss": 2.1457, "step": 1165 }, { "epoch": 0.36, "grad_norm": 1.1938838958740234, "learning_rate": 3.5365169807693966e-05, "loss": 2.1256, "step": 1170 }, { "epoch": 0.37, "grad_norm": 0.8162047863006592, "learning_rate": 3.525389368084379e-05, "loss": 1.9587, "step": 1175 }, { "epoch": 0.37, "grad_norm": 0.9358930587768555, "learning_rate": 3.514237277973393e-05, "loss": 1.8965, "step": 1180 }, { "epoch": 0.37, "grad_norm": 0.9210988879203796, "learning_rate": 3.503060976651862e-05, "loss": 1.9669, "step": 1185 }, { "epoch": 0.37, "grad_norm": 1.4641343355178833, "learning_rate": 3.491860730913156e-05, "loss": 2.003, "step": 1190 }, { "epoch": 0.37, "grad_norm": 1.2458257675170898, "learning_rate": 3.480636808122235e-05, "loss": 2.1487, "step": 1195 }, { "epoch": 0.37, "grad_norm": 1.6770122051239014, "learning_rate": 3.469389476209259e-05, "loss": 2.0686, "step": 1200 }, { "epoch": 0.37, "grad_norm": 0.9083845019340515, "learning_rate": 3.458119003663199e-05, "loss": 2.0284, "step": 1205 }, { "epoch": 0.38, "grad_norm": 1.2679696083068848, "learning_rate": 3.446825659525421e-05, "loss": 2.0555, "step": 1210 }, { "epoch": 0.38, "grad_norm": 1.3823720216751099, "learning_rate": 3.435509713383268e-05, "loss": 1.9375, "step": 1215 }, { "epoch": 0.38, "grad_norm": 1.5862077474594116, "learning_rate": 3.424171435363623e-05, "loss": 2.0271, "step": 1220 }, { "epoch": 0.38, "grad_norm": 2.0107533931732178, "learning_rate": 3.412811096126461e-05, "loss": 2.1897, "step": 1225 }, { "epoch": 0.38, "grad_norm": 1.4544458389282227, "learning_rate": 3.401428966858387e-05, "loss": 1.9978, "step": 1230 }, { "epoch": 0.38, "grad_norm": 1.188170075416565, "learning_rate": 3.390025319266167e-05, "loss": 2.0688, "step": 1235 }, { "epoch": 0.39, "grad_norm": 1.1016322374343872, "learning_rate": 3.3786004255702336e-05, "loss": 2.0396, "step": 1240 }, { "epoch": 0.39, "grad_norm": 1.6623334884643555, "learning_rate": 3.3671545584981954e-05, "loss": 1.9566, "step": 1245 }, { "epoch": 0.39, "grad_norm": 0.9161584377288818, "learning_rate": 3.355687991278324e-05, "loss": 2.0474, "step": 1250 }, { "epoch": 0.39, "grad_norm": 0.9911025166511536, "learning_rate": 3.3442009976330305e-05, "loss": 2.2163, "step": 1255 }, { "epoch": 0.39, "grad_norm": 1.1504255533218384, "learning_rate": 3.332693851772331e-05, "loss": 2.1088, "step": 1260 }, { "epoch": 0.39, "grad_norm": 0.9544184803962708, "learning_rate": 3.3211668283873035e-05, "loss": 1.8947, "step": 1265 }, { "epoch": 0.39, "grad_norm": 1.4625756740570068, "learning_rate": 3.3096202026435304e-05, "loss": 2.1748, "step": 1270 }, { "epoch": 0.4, "grad_norm": 1.3267475366592407, "learning_rate": 3.298054250174527e-05, "loss": 1.9218, "step": 1275 }, { "epoch": 0.4, "grad_norm": 0.9869363903999329, "learning_rate": 3.2864692470751654e-05, "loss": 2.2723, "step": 1280 }, { "epoch": 0.4, "grad_norm": 1.5177838802337646, "learning_rate": 3.27486546989508e-05, "loss": 2.1456, "step": 1285 }, { "epoch": 0.4, "grad_norm": 1.1998714208602905, "learning_rate": 3.263243195632068e-05, "loss": 1.8877, "step": 1290 }, { "epoch": 0.4, "grad_norm": 1.2112164497375488, "learning_rate": 3.2516027017254785e-05, "loss": 2.0615, "step": 1295 }, { "epoch": 0.4, "grad_norm": 1.0616129636764526, "learning_rate": 3.239944266049587e-05, "loss": 2.0402, "step": 1300 }, { "epoch": 0.41, "grad_norm": 1.4537287950515747, "learning_rate": 3.228268166906962e-05, "loss": 2.0728, "step": 1305 }, { "epoch": 0.41, "grad_norm": 1.3899391889572144, "learning_rate": 3.2165746830218254e-05, "loss": 2.1815, "step": 1310 }, { "epoch": 0.41, "grad_norm": 1.332529067993164, "learning_rate": 3.204864093533394e-05, "loss": 1.8935, "step": 1315 }, { "epoch": 0.41, "grad_norm": 1.4466496706008911, "learning_rate": 3.193136677989221e-05, "loss": 1.9567, "step": 1320 }, { "epoch": 0.41, "grad_norm": 1.1781721115112305, "learning_rate": 3.181392716338516e-05, "loss": 2.055, "step": 1325 }, { "epoch": 0.41, "grad_norm": 0.9411901831626892, "learning_rate": 3.1696324889254716e-05, "loss": 1.8794, "step": 1330 }, { "epoch": 0.42, "grad_norm": 1.2628341913223267, "learning_rate": 3.15785627648256e-05, "loss": 2.0299, "step": 1335 }, { "epoch": 0.42, "grad_norm": 1.4857370853424072, "learning_rate": 3.146064360123846e-05, "loss": 1.9342, "step": 1340 }, { "epoch": 0.42, "grad_norm": 1.661470651626587, "learning_rate": 3.1342570213382594e-05, "loss": 2.0399, "step": 1345 }, { "epoch": 0.42, "grad_norm": 1.522845983505249, "learning_rate": 3.122434541982888e-05, "loss": 2.1419, "step": 1350 }, { "epoch": 0.42, "grad_norm": 1.5679118633270264, "learning_rate": 3.110597204276247e-05, "loss": 2.2932, "step": 1355 }, { "epoch": 0.42, "grad_norm": 1.3367788791656494, "learning_rate": 3.098745290791539e-05, "loss": 1.8989, "step": 1360 }, { "epoch": 0.42, "grad_norm": 1.3873472213745117, "learning_rate": 3.086879084449907e-05, "loss": 2.1214, "step": 1365 }, { "epoch": 0.43, "grad_norm": 1.2957035303115845, "learning_rate": 3.074998868513688e-05, "loss": 2.2538, "step": 1370 }, { "epoch": 0.43, "grad_norm": 1.122176170349121, "learning_rate": 3.0631049265796465e-05, "loss": 2.0974, "step": 1375 }, { "epoch": 0.43, "grad_norm": 1.0422618389129639, "learning_rate": 3.051197542572203e-05, "loss": 2.054, "step": 1380 }, { "epoch": 0.43, "grad_norm": 1.1926140785217285, "learning_rate": 3.0392770007366584e-05, "loss": 1.9798, "step": 1385 }, { "epoch": 0.43, "grad_norm": 0.8764025568962097, "learning_rate": 3.0273435856324112e-05, "loss": 2.0796, "step": 1390 }, { "epoch": 0.43, "grad_norm": 0.8200764656066895, "learning_rate": 3.0153975821261605e-05, "loss": 1.9116, "step": 1395 }, { "epoch": 0.44, "grad_norm": 1.0340498685836792, "learning_rate": 3.0034392753851066e-05, "loss": 2.0235, "step": 1400 }, { "epoch": 0.44, "grad_norm": 1.0799012184143066, "learning_rate": 2.9914689508701476e-05, "loss": 2.1455, "step": 1405 }, { "epoch": 0.44, "grad_norm": 1.301015853881836, "learning_rate": 2.979486894329058e-05, "loss": 2.0355, "step": 1410 }, { "epoch": 0.44, "grad_norm": 1.2926914691925049, "learning_rate": 2.9674933917896747e-05, "loss": 2.0379, "step": 1415 }, { "epoch": 0.44, "grad_norm": 1.4712942838668823, "learning_rate": 2.9554887295530647e-05, "loss": 2.0802, "step": 1420 }, { "epoch": 0.44, "grad_norm": 1.1957335472106934, "learning_rate": 2.943473194186693e-05, "loss": 2.1044, "step": 1425 }, { "epoch": 0.44, "grad_norm": 1.568293571472168, "learning_rate": 2.9314470725175792e-05, "loss": 2.0121, "step": 1430 }, { "epoch": 0.45, "grad_norm": 1.4844893217086792, "learning_rate": 2.919410651625455e-05, "loss": 2.0717, "step": 1435 }, { "epoch": 0.45, "grad_norm": 1.3942641019821167, "learning_rate": 2.907364218835904e-05, "loss": 1.9522, "step": 1440 }, { "epoch": 0.45, "grad_norm": 0.7795314788818359, "learning_rate": 2.8953080617135115e-05, "loss": 1.9593, "step": 1445 }, { "epoch": 0.45, "grad_norm": 1.751107931137085, "learning_rate": 2.8832424680549937e-05, "loss": 1.8073, "step": 1450 }, { "epoch": 0.45, "grad_norm": 1.2202279567718506, "learning_rate": 2.8711677258823306e-05, "loss": 2.0042, "step": 1455 }, { "epoch": 0.45, "grad_norm": 1.5163853168487549, "learning_rate": 2.859084123435887e-05, "loss": 1.9931, "step": 1460 }, { "epoch": 0.46, "grad_norm": 0.94038987159729, "learning_rate": 2.84699194916754e-05, "loss": 2.1533, "step": 1465 }, { "epoch": 0.46, "grad_norm": 1.4618102312088013, "learning_rate": 2.834891491733781e-05, "loss": 2.029, "step": 1470 }, { "epoch": 0.46, "grad_norm": 0.9747155904769897, "learning_rate": 2.822783039988836e-05, "loss": 2.0241, "step": 1475 }, { "epoch": 0.46, "grad_norm": 1.0887038707733154, "learning_rate": 2.8106668829777645e-05, "loss": 2.0959, "step": 1480 }, { "epoch": 0.46, "grad_norm": 1.2170171737670898, "learning_rate": 2.7985433099295618e-05, "loss": 1.8718, "step": 1485 }, { "epoch": 0.46, "grad_norm": 1.1366883516311646, "learning_rate": 2.7864126102502524e-05, "loss": 2.2397, "step": 1490 }, { "epoch": 0.46, "grad_norm": 1.1206785440444946, "learning_rate": 2.774275073515985e-05, "loss": 2.1083, "step": 1495 }, { "epoch": 0.47, "grad_norm": 1.126807451248169, "learning_rate": 2.7621309894661167e-05, "loss": 2.0764, "step": 1500 } ], "logging_steps": 5, "max_steps": 3215, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "total_flos": 2.01535997018112e+16, "train_batch_size": 2, "trial_name": null, "trial_params": null }