diff --git "a/checkpoint-5500/trainer_state.json" "b/checkpoint-5500/trainer_state.json" --- "a/checkpoint-5500/trainer_state.json" +++ "b/checkpoint-5500/trainer_state.json" @@ -1,7 +1,7 @@ { "best_global_step": 5500, - "best_metric": 0.9099580645561218, - "best_model_checkpoint": "models/MNLP_M3_rag_model_test/checkpoint-5500", + "best_metric": 0.9267139434814453, + "best_model_checkpoint": "models/MNLP_M3_rag_model_new/checkpoint-5500", "epoch": 2.8732210471340904, "eval_steps": 500, "global_step": 5500, @@ -11,3940 +11,3940 @@ "log_history": [ { "epoch": 0.005222613918266093, - "grad_norm": 17.1588077545166, + "grad_norm": 19.70277214050293, "learning_rate": 3.757828810020877e-07, - "loss": 3.1712, + "loss": 3.2075, "step": 10 }, { "epoch": 0.010445227836532185, - "grad_norm": 12.760807991027832, + "grad_norm": 13.881381034851074, "learning_rate": 7.933194154488519e-07, - "loss": 3.1084, + "loss": 3.1094, "step": 20 }, { "epoch": 0.015667841754798278, - "grad_norm": 9.724287986755371, + "grad_norm": 11.773683547973633, "learning_rate": 1.210855949895616e-06, - "loss": 2.7273, + "loss": 2.7336, "step": 30 }, { "epoch": 0.02089045567306437, - "grad_norm": 6.329061031341553, + "grad_norm": 7.1139326095581055, "learning_rate": 1.62839248434238e-06, - "loss": 2.5386, + "loss": 2.5131, "step": 40 }, { "epoch": 0.02611306959133046, - "grad_norm": 8.292409896850586, + "grad_norm": 6.525805950164795, "learning_rate": 2.0459290187891444e-06, - "loss": 2.4021, + "loss": 2.4319, "step": 50 }, { "epoch": 0.031335683509596556, - "grad_norm": 7.1383161544799805, + "grad_norm": 6.677958011627197, "learning_rate": 2.4634655532359085e-06, - "loss": 2.3112, + "loss": 2.3984, "step": 60 }, { "epoch": 0.036558297427862645, - "grad_norm": 6.580810546875, + "grad_norm": 5.940732479095459, "learning_rate": 2.8810020876826722e-06, - "loss": 2.3126, + "loss": 2.3708, "step": 70 }, { "epoch": 0.04178091134612874, - "grad_norm": 7.729620456695557, + "grad_norm": 6.083739757537842, "learning_rate": 3.2985386221294363e-06, - "loss": 2.3074, + "loss": 2.3667, "step": 80 }, { "epoch": 0.04700352526439483, - "grad_norm": 6.321948528289795, + "grad_norm": 6.817051410675049, "learning_rate": 3.716075156576201e-06, - "loss": 2.2807, + "loss": 2.2422, "step": 90 }, { "epoch": 0.05222613918266092, - "grad_norm": 6.126489162445068, + "grad_norm": 6.625982284545898, "learning_rate": 4.1336116910229655e-06, - "loss": 2.2118, + "loss": 2.2547, "step": 100 }, { "epoch": 0.057448753100927015, - "grad_norm": 6.092052936553955, + "grad_norm": 6.802444934844971, "learning_rate": 4.551148225469729e-06, - "loss": 2.2692, + "loss": 2.1624, "step": 110 }, { "epoch": 0.06267136701919311, - "grad_norm": 6.273594856262207, + "grad_norm": 6.426488399505615, "learning_rate": 4.968684759916494e-06, - "loss": 2.2302, + "loss": 2.1256, "step": 120 }, { "epoch": 0.0678939809374592, - "grad_norm": 6.156396389007568, + "grad_norm": 6.226603984832764, "learning_rate": 5.3862212943632574e-06, - "loss": 2.1471, + "loss": 2.295, "step": 130 }, { "epoch": 0.07311659485572529, - "grad_norm": 6.262785911560059, + "grad_norm": 6.958126068115234, "learning_rate": 5.803757828810022e-06, - "loss": 2.2413, + "loss": 2.1724, "step": 140 }, { "epoch": 0.07833920877399138, - "grad_norm": 9.364627838134766, + "grad_norm": 6.839420795440674, "learning_rate": 6.221294363256786e-06, - "loss": 2.0653, + "loss": 2.099, "step": 150 }, { "epoch": 0.08356182269225748, - "grad_norm": 5.752789497375488, + "grad_norm": 5.962235927581787, "learning_rate": 6.638830897703549e-06, - "loss": 2.2157, + "loss": 2.2467, "step": 160 }, { "epoch": 0.08878443661052357, - "grad_norm": 6.819472789764404, + "grad_norm": 6.217709541320801, "learning_rate": 7.056367432150314e-06, - "loss": 2.1066, + "loss": 2.0834, "step": 170 }, { "epoch": 0.09400705052878966, - "grad_norm": 7.4179182052612305, - "learning_rate": 7.473903966597078e-06, - "loss": 2.1168, + "grad_norm": 8.158942222595215, + "learning_rate": 7.432150313152402e-06, + "loss": 2.09, "step": 180 }, { "epoch": 0.09922966444705575, - "grad_norm": 7.955780029296875, - "learning_rate": 7.891440501043842e-06, - "loss": 2.0347, + "grad_norm": 7.249250411987305, + "learning_rate": 7.849686847599166e-06, + "loss": 1.9052, "step": 190 }, { "epoch": 0.10445227836532184, - "grad_norm": 6.121779441833496, - "learning_rate": 8.308977035490606e-06, - "loss": 2.0351, + "grad_norm": 6.530272960662842, + "learning_rate": 8.267223382045931e-06, + "loss": 1.9663, "step": 200 }, { "epoch": 0.10967489228358794, - "grad_norm": 7.683338642120361, - "learning_rate": 8.72651356993737e-06, - "loss": 2.0195, + "grad_norm": 7.011912822723389, + "learning_rate": 8.684759916492695e-06, + "loss": 2.0027, "step": 210 }, { "epoch": 0.11489750620185403, - "grad_norm": 6.982047080993652, - "learning_rate": 9.144050104384135e-06, - "loss": 1.9903, + "grad_norm": 7.482737064361572, + "learning_rate": 9.102296450939458e-06, + "loss": 1.9008, "step": 220 }, { "epoch": 0.12012012012012012, - "grad_norm": 6.351634502410889, - "learning_rate": 9.561586638830899e-06, - "loss": 1.7746, + "grad_norm": 6.680603981018066, + "learning_rate": 9.519832985386222e-06, + "loss": 2.0743, "step": 230 }, { "epoch": 0.12534273403838622, - "grad_norm": 6.279501438140869, - "learning_rate": 9.979123173277662e-06, - "loss": 2.1087, + "grad_norm": 9.845396041870117, + "learning_rate": 9.937369519832987e-06, + "loss": 1.835, "step": 240 }, { "epoch": 0.1305653479566523, - "grad_norm": 6.24404239654541, - "learning_rate": 1.0396659707724426e-05, - "loss": 1.9282, + "grad_norm": 6.5602240562438965, + "learning_rate": 1.035490605427975e-05, + "loss": 1.8904, "step": 250 }, { "epoch": 0.1357879618749184, - "grad_norm": 7.28134298324585, - "learning_rate": 1.0814196242171192e-05, - "loss": 1.8358, + "grad_norm": 6.729679107666016, + "learning_rate": 1.0772442588726515e-05, + "loss": 1.7183, "step": 260 }, { "epoch": 0.1410105757931845, - "grad_norm": 7.017399787902832, - "learning_rate": 1.1231732776617954e-05, - "loss": 1.8979, + "grad_norm": 6.53078031539917, + "learning_rate": 1.1189979123173279e-05, + "loss": 1.7225, "step": 270 }, { "epoch": 0.14623318971145058, - "grad_norm": 6.723785400390625, - "learning_rate": 1.1649269311064719e-05, - "loss": 1.8811, + "grad_norm": Infinity, + "learning_rate": 1.1607515657620044e-05, + "loss": 1.6489, "step": 280 }, { "epoch": 0.15145580362971667, - "grad_norm": 6.5550360679626465, - "learning_rate": 1.2066805845511483e-05, - "loss": 1.7359, + "grad_norm": 6.237346649169922, + "learning_rate": 1.198329853862213e-05, + "loss": 1.892, "step": 290 }, { "epoch": 0.15667841754798276, - "grad_norm": 6.1090240478515625, - "learning_rate": 1.2484342379958248e-05, - "loss": 1.8842, + "grad_norm": 7.163219928741455, + "learning_rate": 1.2400835073068895e-05, + "loss": 1.8219, "step": 300 }, { "epoch": 0.16190103146624885, - "grad_norm": 7.078915596008301, - "learning_rate": 1.290187891440501e-05, - "loss": 1.9297, + "grad_norm": 6.4514360427856445, + "learning_rate": 1.2818371607515658e-05, + "loss": 1.9048, "step": 310 }, { "epoch": 0.16712364538451496, - "grad_norm": 7.479031085968018, - "learning_rate": 1.3319415448851776e-05, - "loss": 1.8578, + "grad_norm": 6.991766929626465, + "learning_rate": 1.3235908141962424e-05, + "loss": 1.7308, "step": 320 }, { "epoch": 0.17234625930278105, - "grad_norm": 6.292983531951904, - "learning_rate": 1.373695198329854e-05, - "loss": 1.7184, + "grad_norm": 7.481961727142334, + "learning_rate": 1.3653444676409186e-05, + "loss": 1.8498, "step": 330 }, { "epoch": 0.17756887322104714, - "grad_norm": 7.733270168304443, - "learning_rate": 1.4154488517745305e-05, - "loss": 1.8098, + "grad_norm": 6.713171482086182, + "learning_rate": 1.4070981210855951e-05, + "loss": 1.6476, "step": 340 }, { "epoch": 0.18279148713931323, - "grad_norm": 8.41698169708252, - "learning_rate": 1.4572025052192067e-05, - "loss": 1.8588, + "grad_norm": 7.937409400939941, + "learning_rate": 1.4488517745302715e-05, + "loss": 1.8054, "step": 350 }, { "epoch": 0.18801410105757932, - "grad_norm": 7.4168243408203125, - "learning_rate": 1.4989561586638832e-05, - "loss": 1.6814, + "grad_norm": 6.989377498626709, + "learning_rate": 1.490605427974948e-05, + "loss": 1.667, "step": 360 }, { "epoch": 0.1932367149758454, - "grad_norm": 6.406525611877441, - "learning_rate": 1.5407098121085594e-05, - "loss": 1.805, + "grad_norm": 6.625256061553955, + "learning_rate": 1.5323590814196244e-05, + "loss": 1.5924, "step": 370 }, { "epoch": 0.1984593288941115, - "grad_norm": 7.147383689880371, - "learning_rate": 1.582463465553236e-05, - "loss": 1.7771, + "grad_norm": 6.996474266052246, + "learning_rate": 1.5741127348643008e-05, + "loss": 1.6715, "step": 380 }, { "epoch": 0.2036819428123776, - "grad_norm": 6.644751071929932, - "learning_rate": 1.6242171189979125e-05, - "loss": 1.7242, + "grad_norm": 6.7831196784973145, + "learning_rate": 1.615866388308977e-05, + "loss": 1.6487, "step": 390 }, { "epoch": 0.20890455673064368, - "grad_norm": 6.507500648498535, - "learning_rate": 1.665970772442589e-05, - "loss": 1.8213, + "grad_norm": 5.994639873504639, + "learning_rate": 1.6576200417536535e-05, + "loss": 1.7635, "step": 400 }, { "epoch": 0.2141271706489098, - "grad_norm": 6.171652317047119, - "learning_rate": 1.7077244258872652e-05, - "loss": 1.6563, + "grad_norm": 6.5115580558776855, + "learning_rate": 1.69937369519833e-05, + "loss": 1.6552, "step": 410 }, { "epoch": 0.21934978456717588, - "grad_norm": 6.022638320922852, - "learning_rate": 1.745302713987474e-05, - "loss": 1.7445, + "grad_norm": 6.801713943481445, + "learning_rate": 1.7411273486430066e-05, + "loss": 1.5802, "step": 420 }, { "epoch": 0.22457239848544197, - "grad_norm": 6.484378814697266, - "learning_rate": 1.7870563674321503e-05, - "loss": 1.3785, + "grad_norm": 8.447670936584473, + "learning_rate": 1.7828810020876826e-05, + "loss": 1.3681, "step": 430 }, { "epoch": 0.22979501240370806, - "grad_norm": 6.780409336090088, - "learning_rate": 1.828810020876827e-05, - "loss": 1.6468, + "grad_norm": 7.401574611663818, + "learning_rate": 1.8246346555323593e-05, + "loss": 1.5562, "step": 440 }, { "epoch": 0.23501762632197415, - "grad_norm": 6.643301963806152, - "learning_rate": 1.870563674321503e-05, - "loss": 1.6664, + "grad_norm": 6.971620559692383, + "learning_rate": 1.8663883089770357e-05, + "loss": 1.5721, "step": 450 }, { "epoch": 0.24024024024024024, - "grad_norm": 6.810093879699707, - "learning_rate": 1.9123173277661798e-05, - "loss": 1.4001, + "grad_norm": 7.4317755699157715, + "learning_rate": 1.908141962421712e-05, + "loss": 1.4284, "step": 460 }, { "epoch": 0.24546285415850633, - "grad_norm": 7.195308208465576, - "learning_rate": 1.954070981210856e-05, - "loss": 1.5031, + "grad_norm": 6.191842555999756, + "learning_rate": 1.9498956158663885e-05, + "loss": 1.6009, "step": 470 }, { "epoch": 0.25068546807677244, - "grad_norm": 7.472316741943359, - "learning_rate": 1.9958246346555325e-05, - "loss": 1.5853, + "grad_norm": 6.225357532501221, + "learning_rate": 1.9916492693110648e-05, + "loss": 1.5541, "step": 480 }, { "epoch": 0.2559080819950385, - "grad_norm": 6.748635768890381, - "learning_rate": 1.998020019799802e-05, - "loss": 1.5791, + "grad_norm": 5.622026443481445, + "learning_rate": 1.998240017599824e-05, + "loss": 1.6976, "step": 490 }, { "epoch": 0.2611306959133046, - "grad_norm": 6.5012431144714355, - "learning_rate": 1.9958200417995823e-05, - "loss": 1.4822, + "grad_norm": 7.092662811279297, + "learning_rate": 1.9960400395996044e-05, + "loss": 1.5982, "step": 500 }, { "epoch": 0.2611306959133046, - "eval_loss": 1.5763118267059326, - "eval_runtime": 46.6118, - "eval_samples_per_second": 36.514, - "eval_steps_per_second": 4.57, + "eval_loss": 1.5514968633651733, + "eval_runtime": 46.3391, + "eval_samples_per_second": 36.729, + "eval_steps_per_second": 4.597, "step": 500 }, { "epoch": 0.2663533098315707, - "grad_norm": 6.5022149085998535, - "learning_rate": 1.9936200637993622e-05, - "loss": 1.5332, + "grad_norm": 7.0048136711120605, + "learning_rate": 1.9938400615993844e-05, + "loss": 1.6108, "step": 510 }, { "epoch": 0.2715759237498368, - "grad_norm": 8.452847480773926, - "learning_rate": 1.9914200857991422e-05, - "loss": 1.547, + "grad_norm": 6.627659797668457, + "learning_rate": 1.9916400835991644e-05, + "loss": 1.5646, "step": 520 }, { "epoch": 0.27679853766810286, - "grad_norm": 6.139389991760254, - "learning_rate": 1.9892201077989222e-05, - "loss": 1.6269, + "grad_norm": 6.931585311889648, + "learning_rate": 1.989440105598944e-05, + "loss": 1.5008, "step": 530 }, { "epoch": 0.282021151586369, - "grad_norm": 6.797521114349365, - "learning_rate": 1.9870201297987022e-05, - "loss": 1.5635, + "grad_norm": 6.761887550354004, + "learning_rate": 1.987240127598724e-05, + "loss": 1.3353, "step": 540 }, { "epoch": 0.2872437655046351, - "grad_norm": 7.221142292022705, - "learning_rate": 1.9848201517984822e-05, - "loss": 1.3621, + "grad_norm": 5.9434123039245605, + "learning_rate": 1.985040149598504e-05, + "loss": 1.489, "step": 550 }, { "epoch": 0.29246637942290116, - "grad_norm": 6.7414984703063965, - "learning_rate": 1.982620173798262e-05, - "loss": 1.4587, + "grad_norm": 5.751526355743408, + "learning_rate": 1.982840171598284e-05, + "loss": 1.3439, "step": 560 }, { "epoch": 0.2976889933411673, - "grad_norm": 6.582213401794434, - "learning_rate": 1.980420195798042e-05, - "loss": 1.5187, + "grad_norm": 5.684656620025635, + "learning_rate": 1.9806401935980643e-05, + "loss": 1.4879, "step": 570 }, { "epoch": 0.30291160725943334, - "grad_norm": 6.121609210968018, - "learning_rate": 1.978220217797822e-05, - "loss": 1.7261, + "grad_norm": 5.907355785369873, + "learning_rate": 1.9784402155978443e-05, + "loss": 1.3908, "step": 580 }, { "epoch": 0.30813422117769945, - "grad_norm": 7.791438579559326, - "learning_rate": 1.976020239797602e-05, - "loss": 1.6164, + "grad_norm": 6.36244010925293, + "learning_rate": 1.9762402375976242e-05, + "loss": 1.4947, "step": 590 }, { "epoch": 0.3133568350959655, - "grad_norm": 8.201029777526855, - "learning_rate": 1.9738202617973824e-05, - "loss": 1.4859, + "grad_norm": 10.931468963623047, + "learning_rate": 1.9740402595974042e-05, + "loss": 1.5419, "step": 600 }, { "epoch": 0.31857944901423163, - "grad_norm": 5.579436302185059, - "learning_rate": 1.971620283797162e-05, - "loss": 1.2939, + "grad_norm": 5.87261438369751, + "learning_rate": 1.9718402815971842e-05, + "loss": 1.5076, "step": 610 }, { "epoch": 0.3238020629324977, - "grad_norm": 6.592574119567871, - "learning_rate": 1.969420305796942e-05, - "loss": 1.3201, + "grad_norm": 7.896998405456543, + "learning_rate": 1.9696403035969642e-05, + "loss": 1.4993, "step": 620 }, { "epoch": 0.3290246768507638, - "grad_norm": 5.446815490722656, - "learning_rate": 1.967220327796722e-05, - "loss": 1.5061, + "grad_norm": 5.621824264526367, + "learning_rate": 1.9674403255967442e-05, + "loss": 1.4226, "step": 630 }, { "epoch": 0.3342472907690299, - "grad_norm": 5.549929141998291, - "learning_rate": 1.965020349796502e-05, - "loss": 1.5843, + "grad_norm": 6.363002300262451, + "learning_rate": 1.965240347596524e-05, + "loss": 1.386, "step": 640 }, { "epoch": 0.339469904687296, - "grad_norm": 5.987018585205078, - "learning_rate": 1.962820371796282e-05, - "loss": 1.3113, + "grad_norm": 7.999145030975342, + "learning_rate": 1.963040369596304e-05, + "loss": 1.5555, "step": 650 }, { "epoch": 0.3446925186055621, - "grad_norm": 5.247632026672363, - "learning_rate": 1.9606203937960623e-05, - "loss": 1.476, + "grad_norm": 8.901324272155762, + "learning_rate": 1.9608403915960845e-05, + "loss": 1.3187, "step": 660 }, { "epoch": 0.34991513252382817, - "grad_norm": 6.408908367156982, - "learning_rate": 1.9584204157958423e-05, - "loss": 1.67, + "grad_norm": 5.6642913818359375, + "learning_rate": 1.958640413595864e-05, + "loss": 1.2214, "step": 670 }, { "epoch": 0.3551377464420943, - "grad_norm": 7.674783229827881, - "learning_rate": 1.9562204377956223e-05, - "loss": 1.4345, + "grad_norm": 5.334066867828369, + "learning_rate": 1.956440435595644e-05, + "loss": 1.4597, "step": 680 }, { "epoch": 0.36036036036036034, - "grad_norm": 5.847611427307129, - "learning_rate": 1.9540204597954023e-05, - "loss": 1.4084, + "grad_norm": 5.933049201965332, + "learning_rate": 1.954240457595424e-05, + "loss": 1.3489, "step": 690 }, { "epoch": 0.36558297427862646, - "grad_norm": 7.193668365478516, - "learning_rate": 1.9518204817951823e-05, - "loss": 1.4818, + "grad_norm": 5.7526469230651855, + "learning_rate": 1.952040479595204e-05, + "loss": 1.5943, "step": 700 }, { "epoch": 0.3708055881968925, - "grad_norm": 5.585901260375977, - "learning_rate": 1.9496205037949622e-05, - "loss": 1.3425, + "grad_norm": 5.9043378829956055, + "learning_rate": 1.949840501594984e-05, + "loss": 1.2598, "step": 710 }, { "epoch": 0.37602820211515864, - "grad_norm": 5.371877193450928, - "learning_rate": 1.9474205257947422e-05, - "loss": 1.4872, + "grad_norm": 5.5188751220703125, + "learning_rate": 1.9476405235947644e-05, + "loss": 1.2895, "step": 720 }, { "epoch": 0.38125081603342476, - "grad_norm": 7.625179767608643, - "learning_rate": 1.9452205477945222e-05, - "loss": 1.4589, + "grad_norm": 6.718273162841797, + "learning_rate": 1.9454405455945443e-05, + "loss": 1.4436, "step": 730 }, { "epoch": 0.3864734299516908, - "grad_norm": 6.194158554077148, - "learning_rate": 1.9430205697943022e-05, - "loss": 1.3672, + "grad_norm": 5.539391994476318, + "learning_rate": 1.9432405675943243e-05, + "loss": 1.2003, "step": 740 }, { "epoch": 0.39169604386995693, - "grad_norm": 5.667837619781494, - "learning_rate": 1.940820591794082e-05, - "loss": 1.2324, + "grad_norm": 6.471625328063965, + "learning_rate": 1.9410405895941043e-05, + "loss": 1.2729, "step": 750 }, { "epoch": 0.396918657788223, - "grad_norm": 5.823015213012695, - "learning_rate": 1.938620613793862e-05, - "loss": 1.5731, + "grad_norm": 6.520134449005127, + "learning_rate": 1.9388406115938843e-05, + "loss": 1.1503, "step": 760 }, { "epoch": 0.4021412717064891, - "grad_norm": 5.3874945640563965, - "learning_rate": 1.936420635793642e-05, - "loss": 1.456, + "grad_norm": 6.594412326812744, + "learning_rate": 1.9366406335936643e-05, + "loss": 1.1923, "step": 770 }, { "epoch": 0.4073638856247552, - "grad_norm": 9.306315422058105, - "learning_rate": 1.934220657793422e-05, - "loss": 1.3355, + "grad_norm": 6.5402021408081055, + "learning_rate": 1.9344406555934442e-05, + "loss": 1.348, "step": 780 }, { "epoch": 0.4125864995430213, - "grad_norm": 7.826662540435791, - "learning_rate": 1.932020679793202e-05, - "loss": 1.2842, + "grad_norm": 5.678527355194092, + "learning_rate": 1.9322406775932242e-05, + "loss": 1.2498, "step": 790 }, { "epoch": 0.41780911346128735, - "grad_norm": 5.670297145843506, - "learning_rate": 1.929820701792982e-05, - "loss": 1.3126, + "grad_norm": 4.883053302764893, + "learning_rate": 1.9300406995930042e-05, + "loss": 1.1495, "step": 800 }, { "epoch": 0.42303172737955347, - "grad_norm": 5.819605827331543, - "learning_rate": 1.927620723792762e-05, - "loss": 1.4907, + "grad_norm": 6.061793804168701, + "learning_rate": 1.9278407215927842e-05, + "loss": 1.4883, "step": 810 }, { "epoch": 0.4282543412978196, - "grad_norm": 5.6237335205078125, - "learning_rate": 1.9254207457925424e-05, - "loss": 1.3278, + "grad_norm": 6.2135233879089355, + "learning_rate": 1.9256407435925642e-05, + "loss": 1.532, "step": 820 }, { "epoch": 0.43347695521608565, - "grad_norm": 8.079651832580566, - "learning_rate": 1.9232207677923224e-05, - "loss": 1.468, + "grad_norm": 5.913623809814453, + "learning_rate": 1.923440765592344e-05, + "loss": 1.3066, "step": 830 }, { "epoch": 0.43869956913435176, - "grad_norm": 5.166318416595459, - "learning_rate": 1.9210207897921023e-05, - "loss": 1.3633, + "grad_norm": 6.234282493591309, + "learning_rate": 1.921240787592124e-05, + "loss": 0.9888, "step": 840 }, { "epoch": 0.4439221830526178, - "grad_norm": 5.308135509490967, - "learning_rate": 1.9188208117918823e-05, - "loss": 1.3076, + "grad_norm": 5.181183338165283, + "learning_rate": 1.919040809591904e-05, + "loss": 1.3848, "step": 850 }, { "epoch": 0.44914479697088394, - "grad_norm": 5.224324703216553, - "learning_rate": 1.9166208337916623e-05, - "loss": 1.3712, + "grad_norm": 6.581812381744385, + "learning_rate": 1.916840831591684e-05, + "loss": 1.3326, "step": 860 }, { "epoch": 0.45436741088915, - "grad_norm": 5.71431827545166, - "learning_rate": 1.9144208557914423e-05, - "loss": 1.4854, + "grad_norm": 5.040493488311768, + "learning_rate": 1.914640853591464e-05, + "loss": 1.2812, "step": 870 }, { "epoch": 0.4595900248074161, - "grad_norm": 8.291099548339844, - "learning_rate": 1.9122208777912223e-05, - "loss": 1.3062, + "grad_norm": 6.366310119628906, + "learning_rate": 1.9124408755912444e-05, + "loss": 1.2712, "step": 880 }, { "epoch": 0.4648126387256822, - "grad_norm": 6.186370849609375, - "learning_rate": 1.9100208997910023e-05, - "loss": 1.3916, + "grad_norm": 5.0565409660339355, + "learning_rate": 1.9102408975910244e-05, + "loss": 1.1504, "step": 890 }, { "epoch": 0.4700352526439483, - "grad_norm": 5.299021244049072, - "learning_rate": 1.9078209217907822e-05, - "loss": 1.3207, + "grad_norm": 6.553602695465088, + "learning_rate": 1.9080409195908044e-05, + "loss": 1.1751, "step": 900 }, { "epoch": 0.4752578665622144, - "grad_norm": 5.656353950500488, - "learning_rate": 1.9056209437905622e-05, - "loss": 1.2343, + "grad_norm": 6.258462429046631, + "learning_rate": 1.9058409415905844e-05, + "loss": 1.2328, "step": 910 }, { "epoch": 0.4804804804804805, - "grad_norm": 5.802785396575928, - "learning_rate": 1.9034209657903422e-05, - "loss": 1.4532, + "grad_norm": 5.85305118560791, + "learning_rate": 1.9036409635903643e-05, + "loss": 1.1557, "step": 920 }, { "epoch": 0.4857030943987466, - "grad_norm": 6.065880298614502, - "learning_rate": 1.9012209877901222e-05, - "loss": 1.3908, + "grad_norm": 4.799736499786377, + "learning_rate": 1.9014409855901443e-05, + "loss": 1.3247, "step": 930 }, { "epoch": 0.49092570831701265, - "grad_norm": 5.135191440582275, - "learning_rate": 1.8990210097899022e-05, - "loss": 1.3211, + "grad_norm": 5.7678728103637695, + "learning_rate": 1.8992410075899243e-05, + "loss": 1.1796, "step": 940 }, { "epoch": 0.49614832223527877, - "grad_norm": 5.89790153503418, - "learning_rate": 1.896821031789682e-05, - "loss": 1.4183, + "grad_norm": 5.549786567687988, + "learning_rate": 1.8970410295897043e-05, + "loss": 1.2742, "step": 950 }, { "epoch": 0.5013709361535449, - "grad_norm": 5.3774566650390625, - "learning_rate": 1.894621053789462e-05, - "loss": 1.3267, + "grad_norm": 6.545316696166992, + "learning_rate": 1.8948410515894843e-05, + "loss": 1.2211, "step": 960 }, { "epoch": 0.506593550071811, - "grad_norm": 6.830160617828369, - "learning_rate": 1.892421075789242e-05, - "loss": 1.3054, + "grad_norm": 5.516853332519531, + "learning_rate": 1.8926410735892643e-05, + "loss": 1.1495, "step": 970 }, { "epoch": 0.511816163990077, - "grad_norm": 5.20271110534668, - "learning_rate": 1.8902210977890224e-05, - "loss": 1.3637, + "grad_norm": 5.666198253631592, + "learning_rate": 1.8904410955890442e-05, + "loss": 1.3235, "step": 980 }, { "epoch": 0.5170387779083431, - "grad_norm": 7.338613986968994, - "learning_rate": 1.8880211197888024e-05, - "loss": 1.2787, + "grad_norm": 6.691445827484131, + "learning_rate": 1.8882411175888242e-05, + "loss": 1.1672, "step": 990 }, { "epoch": 0.5222613918266092, - "grad_norm": 5.433679580688477, - "learning_rate": 1.8858211417885824e-05, - "loss": 1.2041, + "grad_norm": 5.887513637542725, + "learning_rate": 1.8860411395886042e-05, + "loss": 1.2825, "step": 1000 }, { "epoch": 0.5222613918266092, - "eval_loss": 1.2922958135604858, - "eval_runtime": 46.4161, - "eval_samples_per_second": 36.668, - "eval_steps_per_second": 4.589, + "eval_loss": 1.2772217988967896, + "eval_runtime": 46.2144, + "eval_samples_per_second": 36.828, + "eval_steps_per_second": 4.609, "step": 1000 }, { "epoch": 0.5274840057448753, - "grad_norm": 6.289940357208252, - "learning_rate": 1.8836211637883624e-05, - "loss": 1.4183, + "grad_norm": 6.7056097984313965, + "learning_rate": 1.8838411615883842e-05, + "loss": 1.3954, "step": 1010 }, { "epoch": 0.5327066196631414, - "grad_norm": 4.168149471282959, - "learning_rate": 1.8814211857881424e-05, - "loss": 1.209, + "grad_norm": 5.650666236877441, + "learning_rate": 1.881641183588164e-05, + "loss": 1.1785, "step": 1020 }, { "epoch": 0.5379292335814075, - "grad_norm": 7.0136189460754395, - "learning_rate": 1.879221207787922e-05, - "loss": 1.38, + "grad_norm": 6.400293350219727, + "learning_rate": 1.879441205587944e-05, + "loss": 1.2406, "step": 1030 }, { "epoch": 0.5431518474996736, - "grad_norm": 4.89089822769165, - "learning_rate": 1.8770212297877023e-05, - "loss": 1.5683, + "grad_norm": 4.916640281677246, + "learning_rate": 1.8772412275877245e-05, + "loss": 1.2767, "step": 1040 }, { "epoch": 0.5483744614179397, - "grad_norm": 6.398843288421631, - "learning_rate": 1.8748212517874823e-05, - "loss": 1.3035, + "grad_norm": 5.86029052734375, + "learning_rate": 1.8750412495875045e-05, + "loss": 1.3341, "step": 1050 }, { "epoch": 0.5535970753362057, - "grad_norm": 5.6711859703063965, - "learning_rate": 1.8726212737872623e-05, - "loss": 1.2262, + "grad_norm": 4.692739009857178, + "learning_rate": 1.8728412715872844e-05, + "loss": 1.1973, "step": 1060 }, { "epoch": 0.5588196892544719, - "grad_norm": 5.641086578369141, - "learning_rate": 1.8704212957870423e-05, - "loss": 1.442, + "grad_norm": 5.469728946685791, + "learning_rate": 1.8706412935870644e-05, + "loss": 1.129, "step": 1070 }, { "epoch": 0.564042303172738, - "grad_norm": 5.436042308807373, - "learning_rate": 1.8682213177868223e-05, - "loss": 1.2896, + "grad_norm": 4.398756504058838, + "learning_rate": 1.8684413155868444e-05, + "loss": 1.1981, "step": 1080 }, { "epoch": 0.569264917091004, - "grad_norm": 5.110105514526367, - "learning_rate": 1.8660213397866022e-05, - "loss": 1.2142, + "grad_norm": 7.005917549133301, + "learning_rate": 1.866241337586624e-05, + "loss": 1.4209, "step": 1090 }, { "epoch": 0.5744875310092702, - "grad_norm": 4.662496089935303, - "learning_rate": 1.8638213617863822e-05, - "loss": 1.4738, + "grad_norm": 5.909639835357666, + "learning_rate": 1.8640413595864044e-05, + "loss": 1.3596, "step": 1100 }, { "epoch": 0.5797101449275363, - "grad_norm": 6.685956954956055, - "learning_rate": 1.8616213837861622e-05, - "loss": 1.1842, + "grad_norm": 4.883074760437012, + "learning_rate": 1.8618413815861843e-05, + "loss": 1.2711, "step": 1110 }, { "epoch": 0.5849327588458023, - "grad_norm": 4.521024227142334, - "learning_rate": 1.8594214057859422e-05, - "loss": 1.2215, + "grad_norm": 4.741674900054932, + "learning_rate": 1.8596414035859643e-05, + "loss": 1.1963, "step": 1120 }, { "epoch": 0.5901553727640684, - "grad_norm": 5.817269802093506, - "learning_rate": 1.8572214277857222e-05, - "loss": 1.4296, + "grad_norm": 5.9531354904174805, + "learning_rate": 1.8574414255857443e-05, + "loss": 1.0604, "step": 1130 }, { "epoch": 0.5953779866823345, - "grad_norm": 6.144318103790283, - "learning_rate": 1.8550214497855025e-05, - "loss": 1.3538, + "grad_norm": 5.020533084869385, + "learning_rate": 1.8552414475855243e-05, + "loss": 1.2055, "step": 1140 }, { "epoch": 0.6006006006006006, - "grad_norm": 5.195668697357178, - "learning_rate": 1.8528214717852825e-05, - "loss": 1.3292, + "grad_norm": 6.009571552276611, + "learning_rate": 1.8530414695853043e-05, + "loss": 1.1205, "step": 1150 }, { "epoch": 0.6058232145188667, - "grad_norm": 5.556060314178467, - "learning_rate": 1.8506214937850625e-05, - "loss": 1.3371, + "grad_norm": 5.907603740692139, + "learning_rate": 1.8508414915850843e-05, + "loss": 1.1337, "step": 1160 }, { "epoch": 0.6110458284371327, - "grad_norm": 5.5624613761901855, - "learning_rate": 1.8484215157848424e-05, - "loss": 1.3969, + "grad_norm": 5.728344917297363, + "learning_rate": 1.8486415135848642e-05, + "loss": 1.1401, "step": 1170 }, { "epoch": 0.6162684423553989, - "grad_norm": 4.879144668579102, - "learning_rate": 1.846221537784622e-05, - "loss": 1.3443, + "grad_norm": 7.001476764678955, + "learning_rate": 1.8464415355846442e-05, + "loss": 1.1664, "step": 1180 }, { "epoch": 0.621491056273665, - "grad_norm": 5.067091941833496, - "learning_rate": 1.844021559784402e-05, - "loss": 1.2455, + "grad_norm": 4.943700313568115, + "learning_rate": 1.8442415575844242e-05, + "loss": 1.1148, "step": 1190 }, { "epoch": 0.626713670191931, - "grad_norm": 5.5850443840026855, - "learning_rate": 1.8418215817841824e-05, - "loss": 1.0956, + "grad_norm": 5.676549434661865, + "learning_rate": 1.8420415795842045e-05, + "loss": 1.3097, "step": 1200 }, { "epoch": 0.6319362841101972, - "grad_norm": 4.342463493347168, - "learning_rate": 1.8396216037839624e-05, - "loss": 1.2106, + "grad_norm": 4.349759578704834, + "learning_rate": 1.8398416015839845e-05, + "loss": 1.3017, "step": 1210 }, { "epoch": 0.6371588980284633, - "grad_norm": 5.82832670211792, - "learning_rate": 1.8374216257837424e-05, - "loss": 1.3382, + "grad_norm": 4.567165374755859, + "learning_rate": 1.8376416235837645e-05, + "loss": 1.2503, "step": 1220 }, { "epoch": 0.6423815119467293, - "grad_norm": 5.469226360321045, - "learning_rate": 1.8352216477835223e-05, - "loss": 1.4335, + "grad_norm": 6.539005756378174, + "learning_rate": 1.835441645583544e-05, + "loss": 1.3088, "step": 1230 }, { "epoch": 0.6476041258649954, - "grad_norm": 5.302371978759766, - "learning_rate": 1.8330216697833023e-05, - "loss": 1.312, + "grad_norm": 5.681617736816406, + "learning_rate": 1.833241667583324e-05, + "loss": 1.1112, "step": 1240 }, { "epoch": 0.6528267397832616, - "grad_norm": 5.672764778137207, - "learning_rate": 1.8308216917830823e-05, - "loss": 1.1967, + "grad_norm": 6.110945224761963, + "learning_rate": 1.831041689583104e-05, + "loss": 1.1699, "step": 1250 }, { "epoch": 0.6580493537015276, - "grad_norm": 4.915512561798096, - "learning_rate": 1.8286217137828623e-05, - "loss": 1.4107, + "grad_norm": 7.160697937011719, + "learning_rate": 1.8288417115828844e-05, + "loss": 1.0633, "step": 1260 }, { "epoch": 0.6632719676197937, - "grad_norm": 5.713113784790039, - "learning_rate": 1.8264217357826423e-05, - "loss": 1.2576, + "grad_norm": 4.34064245223999, + "learning_rate": 1.8266417335826644e-05, + "loss": 1.0341, "step": 1270 }, { "epoch": 0.6684945815380599, - "grad_norm": 6.491368293762207, - "learning_rate": 1.8242217577824222e-05, - "loss": 1.3966, + "grad_norm": 4.590273857116699, + "learning_rate": 1.8244417555824444e-05, + "loss": 1.1711, "step": 1280 }, { "epoch": 0.6737171954563259, - "grad_norm": 5.070196628570557, - "learning_rate": 1.8220217797822026e-05, - "loss": 1.3004, + "grad_norm": 6.913531303405762, + "learning_rate": 1.8222417775822244e-05, + "loss": 1.1191, "step": 1290 }, { "epoch": 0.678939809374592, - "grad_norm": 5.9302287101745605, - "learning_rate": 1.8198218017819826e-05, - "loss": 1.2584, + "grad_norm": 5.066277980804443, + "learning_rate": 1.8200417995820044e-05, + "loss": 1.2043, "step": 1300 }, { "epoch": 0.684162423292858, - "grad_norm": 4.848388671875, - "learning_rate": 1.8176218237817625e-05, - "loss": 1.2861, + "grad_norm": 4.551504611968994, + "learning_rate": 1.8178418215817843e-05, + "loss": 1.1502, "step": 1310 }, { "epoch": 0.6893850372111242, - "grad_norm": 5.589436054229736, - "learning_rate": 1.8154218457815422e-05, - "loss": 1.1153, + "grad_norm": 5.54154109954834, + "learning_rate": 1.8156418435815643e-05, + "loss": 1.1562, "step": 1320 }, { "epoch": 0.6946076511293903, - "grad_norm": 4.449565887451172, - "learning_rate": 1.813221867781322e-05, - "loss": 1.2184, + "grad_norm": 5.991913795471191, + "learning_rate": 1.8134418655813443e-05, + "loss": 1.1932, "step": 1330 }, { "epoch": 0.6998302650476563, - "grad_norm": 5.761178970336914, - "learning_rate": 1.811021889781102e-05, - "loss": 1.0727, + "grad_norm": 5.742013454437256, + "learning_rate": 1.8112418875811243e-05, + "loss": 1.2491, "step": 1340 }, { "epoch": 0.7050528789659224, - "grad_norm": 6.625110149383545, - "learning_rate": 1.808821911780882e-05, - "loss": 1.2343, + "grad_norm": 5.051563739776611, + "learning_rate": 1.8090419095809043e-05, + "loss": 1.2941, "step": 1350 }, { "epoch": 0.7102754928841886, - "grad_norm": 6.265412330627441, - "learning_rate": 1.8066219337806624e-05, - "loss": 1.2446, + "grad_norm": 6.9864630699157715, + "learning_rate": 1.8068419315806846e-05, + "loss": 1.1642, "step": 1360 }, { "epoch": 0.7154981068024546, - "grad_norm": 6.213257312774658, - "learning_rate": 1.8044219557804424e-05, - "loss": 1.2587, + "grad_norm": 4.766818523406982, + "learning_rate": 1.8046419535804642e-05, + "loss": 1.1305, "step": 1370 }, { "epoch": 0.7207207207207207, - "grad_norm": 6.336606979370117, - "learning_rate": 1.8022219777802224e-05, - "loss": 1.2753, + "grad_norm": 5.233489990234375, + "learning_rate": 1.8024419755802442e-05, + "loss": 1.1713, "step": 1380 }, { "epoch": 0.7259433346389869, - "grad_norm": 5.962548732757568, - "learning_rate": 1.8000219997800024e-05, - "loss": 1.2126, + "grad_norm": 5.351912975311279, + "learning_rate": 1.8002419975800242e-05, + "loss": 1.2113, "step": 1390 }, { "epoch": 0.7311659485572529, - "grad_norm": 4.752743721008301, - "learning_rate": 1.7978220217797824e-05, - "loss": 1.1085, + "grad_norm": 4.9511542320251465, + "learning_rate": 1.7980420195798042e-05, + "loss": 1.1288, "step": 1400 }, { "epoch": 0.736388562475519, - "grad_norm": 4.578768253326416, - "learning_rate": 1.7956220437795624e-05, - "loss": 0.9734, + "grad_norm": 5.876948356628418, + "learning_rate": 1.795842041579584e-05, + "loss": 1.197, "step": 1410 }, { "epoch": 0.741611176393785, - "grad_norm": 5.62348747253418, - "learning_rate": 1.7934220657793423e-05, - "loss": 1.1462, + "grad_norm": 6.050612926483154, + "learning_rate": 1.7936420635793645e-05, + "loss": 1.2492, "step": 1420 }, { "epoch": 0.7468337903120512, - "grad_norm": 5.4975738525390625, - "learning_rate": 1.7912220877791223e-05, - "loss": 1.0505, + "grad_norm": 5.208220481872559, + "learning_rate": 1.7914420855791445e-05, + "loss": 1.173, "step": 1430 }, { "epoch": 0.7520564042303173, - "grad_norm": 5.553978443145752, - "learning_rate": 1.7890221097789023e-05, - "loss": 1.1606, + "grad_norm": 5.3143134117126465, + "learning_rate": 1.7892421075789244e-05, + "loss": 1.3147, "step": 1440 }, { "epoch": 0.7572790181485833, - "grad_norm": 5.423792362213135, - "learning_rate": 1.7868221317786826e-05, - "loss": 1.2759, + "grad_norm": 4.87001895904541, + "learning_rate": 1.7870421295787044e-05, + "loss": 1.1495, "step": 1450 }, { "epoch": 0.7625016320668495, - "grad_norm": 3.9736289978027344, - "learning_rate": 1.7846221537784623e-05, - "loss": 1.1232, + "grad_norm": 5.324371337890625, + "learning_rate": 1.7848421515784844e-05, + "loss": 1.244, "step": 1460 }, { "epoch": 0.7677242459851156, - "grad_norm": 4.78857421875, + "grad_norm": 5.020702362060547, "learning_rate": 1.7826421735782644e-05, - "loss": 1.2644, + "loss": 1.0374, "step": 1470 }, { "epoch": 0.7729468599033816, - "grad_norm": 4.863739967346191, + "grad_norm": 3.847534418106079, "learning_rate": 1.7804421955780444e-05, - "loss": 1.1593, + "loss": 1.1141, "step": 1480 }, { "epoch": 0.7781694738216477, - "grad_norm": 5.5411882400512695, + "grad_norm": 6.3277668952941895, "learning_rate": 1.7782422175778244e-05, - "loss": 1.1533, + "loss": 1.1015, "step": 1490 }, { "epoch": 0.7833920877399139, - "grad_norm": 4.7969584465026855, + "grad_norm": 4.962647914886475, "learning_rate": 1.7760422395776043e-05, - "loss": 1.2054, + "loss": 1.2934, "step": 1500 }, { "epoch": 0.7833920877399139, - "eval_loss": 1.152809739112854, - "eval_runtime": 46.4044, - "eval_samples_per_second": 36.678, - "eval_steps_per_second": 4.59, + "eval_loss": 1.134088158607483, + "eval_runtime": 46.1817, + "eval_samples_per_second": 36.854, + "eval_steps_per_second": 4.612, "step": 1500 }, { "epoch": 0.7886147016581799, - "grad_norm": 5.786462306976318, + "grad_norm": 10.081462860107422, "learning_rate": 1.7738422615773843e-05, - "loss": 1.1893, + "loss": 1.0337, "step": 1510 }, { "epoch": 0.793837315576446, - "grad_norm": 5.743650913238525, + "grad_norm": 5.772327423095703, "learning_rate": 1.7716422835771643e-05, - "loss": 1.1873, + "loss": 1.2605, "step": 1520 }, { "epoch": 0.799059929494712, - "grad_norm": 5.861995220184326, + "grad_norm": 5.130199909210205, "learning_rate": 1.7694423055769443e-05, - "loss": 1.2693, + "loss": 1.2705, "step": 1530 }, { "epoch": 0.8042825434129782, - "grad_norm": 5.11861515045166, + "grad_norm": 4.830016613006592, "learning_rate": 1.7672423275767243e-05, - "loss": 1.1807, + "loss": 1.0932, "step": 1540 }, { "epoch": 0.8095051573312443, - "grad_norm": 5.063669204711914, + "grad_norm": 3.770686388015747, "learning_rate": 1.7650423495765043e-05, - "loss": 1.1939, + "loss": 1.0961, "step": 1550 }, { "epoch": 0.8147277712495103, - "grad_norm": 6.106082439422607, + "grad_norm": 5.477590560913086, "learning_rate": 1.7628423715762842e-05, - "loss": 1.0904, + "loss": 0.9974, "step": 1560 }, { "epoch": 0.8199503851677765, - "grad_norm": 5.699225902557373, + "grad_norm": 4.328277587890625, "learning_rate": 1.7606423935760642e-05, - "loss": 1.0069, + "loss": 1.1103, "step": 1570 }, { "epoch": 0.8251729990860426, - "grad_norm": 5.60951566696167, + "grad_norm": 5.159764289855957, "learning_rate": 1.7584424155758445e-05, - "loss": 1.196, + "loss": 1.1238, "step": 1580 }, { "epoch": 0.8303956130043086, - "grad_norm": 6.9652099609375, + "grad_norm": 6.441975116729736, "learning_rate": 1.7562424375756245e-05, - "loss": 1.2167, + "loss": 0.9822, "step": 1590 }, { "epoch": 0.8356182269225747, - "grad_norm": 6.079868316650391, + "grad_norm": 4.554153919219971, "learning_rate": 1.7540424595754045e-05, - "loss": 1.296, + "loss": 1.1166, "step": 1600 }, { "epoch": 0.8408408408408409, - "grad_norm": 5.60435152053833, + "grad_norm": 6.087958812713623, "learning_rate": 1.7518424815751845e-05, - "loss": 1.2494, + "loss": 1.0791, "step": 1610 }, { "epoch": 0.8460634547591069, - "grad_norm": 6.481119632720947, + "grad_norm": 6.553117752075195, "learning_rate": 1.7496425035749645e-05, - "loss": 1.2123, + "loss": 1.1168, "step": 1620 }, { "epoch": 0.851286068677373, - "grad_norm": 4.173206806182861, + "grad_norm": 5.731266975402832, "learning_rate": 1.7474425255747444e-05, - "loss": 1.1377, + "loss": 0.9788, "step": 1630 }, { "epoch": 0.8565086825956392, - "grad_norm": 4.763253688812256, + "grad_norm": 5.815290451049805, "learning_rate": 1.7452425475745244e-05, - "loss": 1.0706, + "loss": 1.1205, "step": 1640 }, { "epoch": 0.8617312965139052, - "grad_norm": 5.0582709312438965, + "grad_norm": 7.379497051239014, "learning_rate": 1.7430425695743044e-05, - "loss": 1.0366, + "loss": 1.1138, "step": 1650 }, { "epoch": 0.8669539104321713, - "grad_norm": 4.170965194702148, + "grad_norm": 4.018825531005859, "learning_rate": 1.7408425915740844e-05, - "loss": 1.1789, + "loss": 1.1798, "step": 1660 }, { "epoch": 0.8721765243504374, - "grad_norm": 3.9636096954345703, + "grad_norm": 6.738025188446045, "learning_rate": 1.7386426135738644e-05, - "loss": 0.9923, + "loss": 1.0784, "step": 1670 }, { "epoch": 0.8773991382687035, - "grad_norm": 4.601221561431885, + "grad_norm": 4.209005355834961, "learning_rate": 1.7364426355736444e-05, - "loss": 1.2057, + "loss": 1.0568, "step": 1680 }, { "epoch": 0.8826217521869696, - "grad_norm": 5.6236114501953125, + "grad_norm": 5.212493896484375, "learning_rate": 1.7342426575734243e-05, - "loss": 1.0889, + "loss": 1.0919, "step": 1690 }, { "epoch": 0.8878443661052356, - "grad_norm": 5.330966472625732, + "grad_norm": 4.836096286773682, "learning_rate": 1.7320426795732043e-05, - "loss": 1.1021, + "loss": 1.2106, "step": 1700 }, { "epoch": 0.8930669800235017, - "grad_norm": 5.453662872314453, + "grad_norm": 4.01015567779541, "learning_rate": 1.7298427015729843e-05, - "loss": 1.2257, + "loss": 1.1171, "step": 1710 }, { "epoch": 0.8982895939417679, - "grad_norm": 5.54590368270874, + "grad_norm": 4.925267219543457, "learning_rate": 1.7276427235727643e-05, - "loss": 1.2001, + "loss": 1.1519, "step": 1720 }, { "epoch": 0.903512207860034, - "grad_norm": 5.4851508140563965, + "grad_norm": 5.960758209228516, "learning_rate": 1.7254427455725443e-05, - "loss": 1.2087, + "loss": 0.869, "step": 1730 }, { "epoch": 0.9087348217783, - "grad_norm": 5.15001106262207, + "grad_norm": 3.004330635070801, "learning_rate": 1.7232427675723246e-05, - "loss": 1.178, + "loss": 0.9295, "step": 1740 }, { "epoch": 0.9139574356965662, - "grad_norm": 5.020768165588379, + "grad_norm": 5.189356803894043, "learning_rate": 1.7210427895721046e-05, - "loss": 1.1614, + "loss": 1.09, "step": 1750 }, { "epoch": 0.9191800496148322, - "grad_norm": 4.394384384155273, + "grad_norm": 5.3001298904418945, "learning_rate": 1.7188428115718846e-05, - "loss": 1.1276, + "loss": 1.17, "step": 1760 }, { "epoch": 0.9244026635330983, - "grad_norm": 4.4511237144470215, + "grad_norm": 5.681934356689453, "learning_rate": 1.7166428335716645e-05, - "loss": 1.0134, + "loss": 1.0682, "step": 1770 }, { "epoch": 0.9296252774513644, - "grad_norm": 5.19103479385376, + "grad_norm": 5.24437141418457, "learning_rate": 1.7144428555714445e-05, - "loss": 1.1161, + "loss": 1.0459, "step": 1780 }, { "epoch": 0.9348478913696305, - "grad_norm": 5.5525126457214355, + "grad_norm": 5.576322555541992, "learning_rate": 1.7122428775712245e-05, - "loss": 1.0225, + "loss": 1.0456, "step": 1790 }, { "epoch": 0.9400705052878966, - "grad_norm": 6.351354598999023, + "grad_norm": 5.690619468688965, "learning_rate": 1.7100428995710045e-05, - "loss": 1.0203, + "loss": 1.1784, "step": 1800 }, { "epoch": 0.9452931192061627, - "grad_norm": 4.5916666984558105, + "grad_norm": 4.465052127838135, "learning_rate": 1.7078429215707845e-05, - "loss": 1.1395, + "loss": 1.1642, "step": 1810 }, { "epoch": 0.9505157331244288, - "grad_norm": 5.519266128540039, + "grad_norm": 5.3224992752075195, "learning_rate": 1.7056429435705645e-05, - "loss": 0.9899, + "loss": 1.3685, "step": 1820 }, { "epoch": 0.9557383470426949, - "grad_norm": 5.209677696228027, + "grad_norm": 5.342929840087891, "learning_rate": 1.7034429655703444e-05, - "loss": 1.3285, + "loss": 1.0892, "step": 1830 }, { "epoch": 0.960960960960961, - "grad_norm": 5.857298374176025, + "grad_norm": 6.5859293937683105, "learning_rate": 1.7012429875701244e-05, - "loss": 1.2475, + "loss": 1.0014, "step": 1840 }, { "epoch": 0.966183574879227, - "grad_norm": 6.121822834014893, + "grad_norm": 4.721777439117432, "learning_rate": 1.6990430095699044e-05, - "loss": 1.0665, + "loss": 1.0088, "step": 1850 }, { "epoch": 0.9714061887974932, - "grad_norm": 4.8332295417785645, + "grad_norm": 4.763739585876465, "learning_rate": 1.6968430315696844e-05, - "loss": 1.179, + "loss": 1.1575, "step": 1860 }, { "epoch": 0.9766288027157592, - "grad_norm": 5.4480695724487305, + "grad_norm": 5.839937686920166, "learning_rate": 1.6946430535694644e-05, - "loss": 0.9849, + "loss": 1.0185, "step": 1870 }, { "epoch": 0.9818514166340253, - "grad_norm": 5.037874698638916, + "grad_norm": 5.6197357177734375, "learning_rate": 1.6924430755692443e-05, - "loss": 1.0214, + "loss": 1.0802, "step": 1880 }, { "epoch": 0.9870740305522914, - "grad_norm": 4.6786112785339355, + "grad_norm": 5.277629852294922, "learning_rate": 1.6902430975690243e-05, - "loss": 1.1205, + "loss": 1.0029, "step": 1890 }, { "epoch": 0.9922966444705575, - "grad_norm": 5.271369457244873, + "grad_norm": 5.552341938018799, "learning_rate": 1.6880431195688047e-05, - "loss": 1.1756, + "loss": 1.0826, "step": 1900 }, { "epoch": 0.9975192583888236, - "grad_norm": 5.059353828430176, + "grad_norm": 6.93430233001709, "learning_rate": 1.6858431415685846e-05, - "loss": 1.1025, + "loss": 1.2054, "step": 1910 }, { "epoch": 1.0031335683509597, - "grad_norm": 4.742678165435791, + "grad_norm": 5.64298152923584, "learning_rate": 1.6836431635683646e-05, - "loss": 0.9805, + "loss": 0.9853, "step": 1920 }, { "epoch": 1.0083561822692257, - "grad_norm": 4.218075275421143, + "grad_norm": 4.246952533721924, "learning_rate": 1.6814431855681446e-05, - "loss": 0.8862, + "loss": 1.0021, "step": 1930 }, { "epoch": 1.0135787961874918, - "grad_norm": 7.259308338165283, + "grad_norm": 4.860617160797119, "learning_rate": 1.6792432075679242e-05, - "loss": 0.7837, + "loss": 0.7337, "step": 1940 }, { "epoch": 1.0188014101057579, - "grad_norm": 3.9883408546447754, + "grad_norm": 4.53707218170166, "learning_rate": 1.6770432295677042e-05, - "loss": 0.7887, + "loss": 0.8683, "step": 1950 }, { "epoch": 1.024024024024024, - "grad_norm": 5.002141952514648, + "grad_norm": 5.660172939300537, "learning_rate": 1.6748432515674845e-05, - "loss": 0.8512, + "loss": 0.9018, "step": 1960 }, { "epoch": 1.0292466379422902, - "grad_norm": 4.857590198516846, + "grad_norm": 6.140296459197998, "learning_rate": 1.6726432735672645e-05, - "loss": 0.8973, + "loss": 0.7245, "step": 1970 }, { "epoch": 1.0344692518605563, - "grad_norm": 5.385519981384277, + "grad_norm": 4.372763633728027, "learning_rate": 1.6704432955670445e-05, - "loss": 0.8722, + "loss": 0.7623, "step": 1980 }, { "epoch": 1.0396918657788223, - "grad_norm": 4.655357360839844, + "grad_norm": 6.61103630065918, "learning_rate": 1.6682433175668245e-05, - "loss": 0.9631, + "loss": 0.9878, "step": 1990 }, { "epoch": 1.0449144796970884, - "grad_norm": 4.768300533294678, + "grad_norm": 6.071864604949951, "learning_rate": 1.6660433395666045e-05, - "loss": 0.9284, + "loss": 0.9403, "step": 2000 }, { "epoch": 1.0449144796970884, - "eval_loss": 1.062979817390442, - "eval_runtime": 46.384, - "eval_samples_per_second": 36.694, - "eval_steps_per_second": 4.592, + "eval_loss": 1.060141682624817, + "eval_runtime": 46.3373, + "eval_samples_per_second": 36.731, + "eval_steps_per_second": 4.597, "step": 2000 }, { "epoch": 1.0501370936153545, - "grad_norm": 5.307561874389648, + "grad_norm": 4.46028995513916, "learning_rate": 1.6638433615663845e-05, - "loss": 1.1394, + "loss": 0.8398, "step": 2010 }, { "epoch": 1.0553597075336205, - "grad_norm": 6.255516529083252, + "grad_norm": 5.496013164520264, "learning_rate": 1.6616433835661644e-05, - "loss": 0.933, + "loss": 0.7149, "step": 2020 }, { "epoch": 1.0605823214518866, - "grad_norm": 4.4736433029174805, + "grad_norm": 4.552042007446289, "learning_rate": 1.6594434055659444e-05, - "loss": 0.8371, + "loss": 0.7761, "step": 2030 }, { "epoch": 1.0658049353701529, - "grad_norm": 3.445369005203247, + "grad_norm": 5.607778072357178, "learning_rate": 1.6572434275657244e-05, - "loss": 0.8323, + "loss": 0.7518, "step": 2040 }, { "epoch": 1.071027549288419, - "grad_norm": 3.695296287536621, + "grad_norm": 5.517201900482178, "learning_rate": 1.6550434495655044e-05, - "loss": 0.8505, + "loss": 0.8282, "step": 2050 }, { "epoch": 1.076250163206685, - "grad_norm": 5.231265544891357, + "grad_norm": 4.941043376922607, "learning_rate": 1.6528434715652847e-05, - "loss": 0.8781, + "loss": 0.8358, "step": 2060 }, { "epoch": 1.081472777124951, - "grad_norm": 4.590741157531738, + "grad_norm": 5.169344425201416, "learning_rate": 1.6506434935650647e-05, - "loss": 0.8584, + "loss": 0.8934, "step": 2070 }, { "epoch": 1.086695391043217, - "grad_norm": 5.293918132781982, + "grad_norm": 5.423684120178223, "learning_rate": 1.6484435155648443e-05, - "loss": 0.7935, + "loss": 0.7022, "step": 2080 }, { "epoch": 1.0919180049614832, - "grad_norm": 5.582162380218506, + "grad_norm": 7.237351417541504, "learning_rate": 1.6462435375646243e-05, - "loss": 0.8391, + "loss": 0.9383, "step": 2090 }, { "epoch": 1.0971406188797492, - "grad_norm": 4.409743785858154, + "grad_norm": 6.69820499420166, "learning_rate": 1.6440435595644043e-05, - "loss": 0.8929, + "loss": 0.8252, "step": 2100 }, { "epoch": 1.1023632327980155, - "grad_norm": 6.849720001220703, + "grad_norm": 5.496448516845703, "learning_rate": 1.6418435815641843e-05, - "loss": 1.0301, + "loss": 0.8748, "step": 2110 }, { "epoch": 1.1075858467162816, - "grad_norm": 4.769338607788086, + "grad_norm": 5.05250358581543, "learning_rate": 1.6396436035639646e-05, - "loss": 0.8172, + "loss": 0.8136, "step": 2120 }, { "epoch": 1.1128084606345476, - "grad_norm": 5.646458625793457, + "grad_norm": 6.312356472015381, "learning_rate": 1.6374436255637446e-05, - "loss": 0.8616, + "loss": 0.8099, "step": 2130 }, { "epoch": 1.1180310745528137, - "grad_norm": 5.374930381774902, + "grad_norm": 4.283848762512207, "learning_rate": 1.6352436475635246e-05, - "loss": 0.8015, + "loss": 0.7964, "step": 2140 }, { "epoch": 1.1232536884710798, - "grad_norm": 3.9189627170562744, + "grad_norm": 7.656008243560791, "learning_rate": 1.6330436695633046e-05, - "loss": 0.8825, + "loss": 0.7453, "step": 2150 }, { "epoch": 1.1284763023893458, - "grad_norm": 4.851017951965332, + "grad_norm": 4.6935930252075195, "learning_rate": 1.6308436915630845e-05, - "loss": 0.7428, + "loss": 0.7514, "step": 2160 }, { "epoch": 1.1336989163076119, - "grad_norm": 5.55581521987915, + "grad_norm": 6.587338447570801, "learning_rate": 1.6286437135628645e-05, - "loss": 0.8404, + "loss": 0.7144, "step": 2170 }, { "epoch": 1.1389215302258782, - "grad_norm": 6.765121936798096, + "grad_norm": 6.313709259033203, "learning_rate": 1.6264437355626445e-05, - "loss": 0.9473, + "loss": 0.8747, "step": 2180 }, { "epoch": 1.1441441441441442, - "grad_norm": 4.86257266998291, + "grad_norm": 4.5388875007629395, "learning_rate": 1.6242437575624245e-05, - "loss": 0.8989, + "loss": 0.9627, "step": 2190 }, { "epoch": 1.1493667580624103, - "grad_norm": 5.461528301239014, + "grad_norm": 3.937307357788086, "learning_rate": 1.6220437795622045e-05, - "loss": 0.7735, + "loss": 0.7915, "step": 2200 }, { "epoch": 1.1545893719806763, - "grad_norm": 5.794458389282227, + "grad_norm": 5.025647163391113, "learning_rate": 1.6198438015619848e-05, - "loss": 0.8712, + "loss": 0.6754, "step": 2210 }, { "epoch": 1.1598119858989424, - "grad_norm": 5.1557841300964355, + "grad_norm": 5.299012660980225, "learning_rate": 1.6176438235617644e-05, - "loss": 0.9436, + "loss": 0.846, "step": 2220 }, { "epoch": 1.1650345998172085, - "grad_norm": 5.397255897521973, + "grad_norm": 5.426839828491211, "learning_rate": 1.6154438455615444e-05, - "loss": 0.9447, + "loss": 0.7994, "step": 2230 }, { "epoch": 1.1702572137354745, - "grad_norm": 4.012696743011475, + "grad_norm": 4.719479084014893, "learning_rate": 1.6132438675613244e-05, - "loss": 0.7914, + "loss": 0.7898, "step": 2240 }, { "epoch": 1.1754798276537408, - "grad_norm": 4.5587286949157715, + "grad_norm": 5.899985313415527, "learning_rate": 1.6110438895611044e-05, - "loss": 0.849, + "loss": 0.8208, "step": 2250 }, { "epoch": 1.1807024415720069, - "grad_norm": 5.750669956207275, + "grad_norm": 3.5655477046966553, "learning_rate": 1.6088439115608844e-05, - "loss": 0.9119, + "loss": 0.7577, "step": 2260 }, { "epoch": 1.185925055490273, - "grad_norm": 4.828431606292725, + "grad_norm": 4.261383056640625, "learning_rate": 1.6066439335606643e-05, - "loss": 0.8488, + "loss": 0.8859, "step": 2270 }, { "epoch": 1.191147669408539, - "grad_norm": 4.56440544128418, + "grad_norm": 5.385866165161133, "learning_rate": 1.6044439555604447e-05, - "loss": 0.8229, + "loss": 0.8545, "step": 2280 }, { "epoch": 1.196370283326805, - "grad_norm": 4.551214218139648, + "grad_norm": 5.758038520812988, "learning_rate": 1.6022439775602246e-05, - "loss": 0.8505, + "loss": 0.7794, "step": 2290 }, { "epoch": 1.2015928972450711, - "grad_norm": 5.286618232727051, + "grad_norm": 5.3475422859191895, "learning_rate": 1.6000439995600046e-05, - "loss": 0.8631, + "loss": 0.7089, "step": 2300 }, { "epoch": 1.2068155111633372, - "grad_norm": 5.9425482749938965, + "grad_norm": 4.732235908508301, "learning_rate": 1.5978440215597846e-05, - "loss": 0.6281, + "loss": 0.7492, "step": 2310 }, { "epoch": 1.2120381250816032, - "grad_norm": 3.729724884033203, + "grad_norm": 7.051826000213623, "learning_rate": 1.5956440435595646e-05, - "loss": 0.8309, + "loss": 0.7422, "step": 2320 }, { "epoch": 1.2172607389998695, - "grad_norm": 8.449934959411621, + "grad_norm": 5.329110622406006, "learning_rate": 1.5934440655593446e-05, - "loss": 0.9191, + "loss": 0.7942, "step": 2330 }, { "epoch": 1.2224833529181356, - "grad_norm": 5.1881842613220215, + "grad_norm": 4.756658554077148, "learning_rate": 1.5912440875591246e-05, - "loss": 0.8585, + "loss": 0.6868, "step": 2340 }, { "epoch": 1.2277059668364017, - "grad_norm": 6.055355548858643, + "grad_norm": 3.588118314743042, "learning_rate": 1.5890441095589045e-05, - "loss": 0.8272, + "loss": 0.776, "step": 2350 }, { "epoch": 1.2329285807546677, - "grad_norm": 4.473712921142578, + "grad_norm": 4.676788330078125, "learning_rate": 1.5868441315586845e-05, - "loss": 0.871, + "loss": 0.8421, "step": 2360 }, { "epoch": 1.2381511946729338, - "grad_norm": 3.1991331577301025, + "grad_norm": 6.036731719970703, "learning_rate": 1.5846441535584645e-05, - "loss": 0.8593, + "loss": 0.7423, "step": 2370 }, { "epoch": 1.2433738085911998, - "grad_norm": 5.2965288162231445, + "grad_norm": 3.8142011165618896, "learning_rate": 1.5824441755582445e-05, - "loss": 0.8573, + "loss": 0.6844, "step": 2380 }, { "epoch": 1.248596422509466, - "grad_norm": 5.760184288024902, + "grad_norm": 5.518984794616699, "learning_rate": 1.5802441975580245e-05, - "loss": 0.9083, + "loss": 0.8246, "step": 2390 }, { "epoch": 1.253819036427732, - "grad_norm": 4.14992618560791, + "grad_norm": 5.119243144989014, "learning_rate": 1.5780442195578045e-05, - "loss": 0.8069, + "loss": 0.7811, "step": 2400 }, { "epoch": 1.2590416503459982, - "grad_norm": 5.517070770263672, + "grad_norm": 6.098830223083496, "learning_rate": 1.5758442415575844e-05, - "loss": 0.8448, + "loss": 0.734, "step": 2410 }, { "epoch": 1.2642642642642643, - "grad_norm": 4.883249282836914, + "grad_norm": 3.1341919898986816, "learning_rate": 1.5736442635573644e-05, - "loss": 0.9297, + "loss": 0.925, "step": 2420 }, { "epoch": 1.2694868781825304, - "grad_norm": 4.5844950675964355, + "grad_norm": 4.657491207122803, "learning_rate": 1.5714442855571444e-05, - "loss": 0.7879, + "loss": 0.734, "step": 2430 }, { "epoch": 1.2747094921007964, - "grad_norm": 6.477102279663086, - "learning_rate": 1.5692443075569247e-05, - "loss": 0.9081, + "grad_norm": 4.789829254150391, + "learning_rate": 1.5694643053569465e-05, + "loss": 0.8655, "step": 2440 }, { "epoch": 1.2799321060190625, - "grad_norm": 4.495989799499512, - "learning_rate": 1.5670443295567047e-05, - "loss": 0.9186, + "grad_norm": 5.776912689208984, + "learning_rate": 1.5672643273567265e-05, + "loss": 0.6572, "step": 2450 }, { "epoch": 1.2851547199373288, - "grad_norm": 5.04586124420166, - "learning_rate": 1.5648443515564847e-05, - "loss": 0.8438, + "grad_norm": 5.746379375457764, + "learning_rate": 1.5650643493565065e-05, + "loss": 0.842, "step": 2460 }, { "epoch": 1.2903773338555946, - "grad_norm": 4.635850429534912, - "learning_rate": 1.5626443735562647e-05, - "loss": 0.9261, + "grad_norm": 7.10046911239624, + "learning_rate": 1.5628643713562865e-05, + "loss": 0.7968, "step": 2470 }, { "epoch": 1.295599947773861, - "grad_norm": 6.3529253005981445, - "learning_rate": 1.5604443955560446e-05, - "loss": 0.9077, + "grad_norm": 4.6730217933654785, + "learning_rate": 1.5606643933560664e-05, + "loss": 0.6907, "step": 2480 }, { "epoch": 1.300822561692127, - "grad_norm": 5.649443626403809, - "learning_rate": 1.5582444175558246e-05, - "loss": 0.8207, + "grad_norm": 5.4030632972717285, + "learning_rate": 1.5584644153558464e-05, + "loss": 0.8275, "step": 2490 }, { "epoch": 1.306045175610393, - "grad_norm": 4.3899760246276855, - "learning_rate": 1.5560444395556046e-05, - "loss": 0.8617, + "grad_norm": 5.2555999755859375, + "learning_rate": 1.5562644373556268e-05, + "loss": 0.7337, "step": 2500 }, { "epoch": 1.306045175610393, - "eval_loss": 1.0080269575119019, - "eval_runtime": 46.3858, - "eval_samples_per_second": 36.692, - "eval_steps_per_second": 4.592, + "eval_loss": 1.0199942588806152, + "eval_runtime": 46.2946, + "eval_samples_per_second": 36.765, + "eval_steps_per_second": 4.601, "step": 2500 }, { "epoch": 1.311267789528659, - "grad_norm": 4.011806011199951, - "learning_rate": 1.5538444615553846e-05, - "loss": 0.7535, + "grad_norm": 5.632393836975098, + "learning_rate": 1.5540644593554067e-05, + "loss": 0.7475, "step": 2510 }, { "epoch": 1.3164904034469251, - "grad_norm": 6.140223026275635, - "learning_rate": 1.5516444835551646e-05, - "loss": 0.7793, + "grad_norm": 6.044793605804443, + "learning_rate": 1.5518644813551867e-05, + "loss": 0.8963, "step": 2520 }, { "epoch": 1.3217130173651912, - "grad_norm": 3.5991857051849365, - "learning_rate": 1.5494445055549446e-05, - "loss": 0.8774, + "grad_norm": 4.846126556396484, + "learning_rate": 1.5496645033549667e-05, + "loss": 0.8615, "step": 2530 }, { "epoch": 1.3269356312834573, - "grad_norm": 4.690261363983154, - "learning_rate": 1.5472445275547245e-05, - "loss": 0.8758, + "grad_norm": 4.63947868347168, + "learning_rate": 1.5474645253547467e-05, + "loss": 0.7742, "step": 2540 }, { "epoch": 1.3321582452017235, - "grad_norm": 4.750207424163818, - "learning_rate": 1.5450445495545045e-05, - "loss": 0.9743, + "grad_norm": 3.978665590286255, + "learning_rate": 1.5452645473545267e-05, + "loss": 0.7666, "step": 2550 }, { "epoch": 1.3373808591199896, - "grad_norm": 6.83930778503418, - "learning_rate": 1.5428445715542845e-05, - "loss": 0.8209, + "grad_norm": 3.455286741256714, + "learning_rate": 1.5430645693543066e-05, + "loss": 0.7028, "step": 2560 }, { "epoch": 1.3426034730382557, - "grad_norm": 5.2934370040893555, - "learning_rate": 1.5406445935540645e-05, - "loss": 0.8453, + "grad_norm": 4.03436803817749, + "learning_rate": 1.5408645913540866e-05, + "loss": 0.682, "step": 2570 }, { "epoch": 1.3478260869565217, - "grad_norm": 5.174552917480469, - "learning_rate": 1.5384446155538445e-05, - "loss": 0.7369, + "grad_norm": 3.467874526977539, + "learning_rate": 1.5386646133538666e-05, + "loss": 0.7811, "step": 2580 }, { "epoch": 1.3530487008747878, - "grad_norm": 4.970681667327881, - "learning_rate": 1.5362446375536248e-05, - "loss": 0.7854, + "grad_norm": 4.449886798858643, + "learning_rate": 1.5364646353536466e-05, + "loss": 0.6624, "step": 2590 }, { "epoch": 1.3582713147930539, - "grad_norm": 4.746755599975586, - "learning_rate": 1.5340446595534048e-05, - "loss": 0.7734, + "grad_norm": 2.9593687057495117, + "learning_rate": 1.5342646573534266e-05, + "loss": 0.8729, "step": 2600 }, { "epoch": 1.36349392871132, - "grad_norm": 2.6767237186431885, - "learning_rate": 1.5318446815531848e-05, - "loss": 0.6543, + "grad_norm": 4.363494396209717, + "learning_rate": 1.5320646793532066e-05, + "loss": 0.7399, "step": 2610 }, { "epoch": 1.3687165426295862, - "grad_norm": 5.292436122894287, - "learning_rate": 1.5296447035529647e-05, - "loss": 0.8073, + "grad_norm": 3.814774513244629, + "learning_rate": 1.5298647013529865e-05, + "loss": 0.7642, "step": 2620 }, { "epoch": 1.3739391565478523, - "grad_norm": 5.2635626792907715, - "learning_rate": 1.5274447255527447e-05, - "loss": 0.7409, + "grad_norm": 4.461428642272949, + "learning_rate": 1.5276647233527665e-05, + "loss": 0.8218, "step": 2630 }, { "epoch": 1.3791617704661183, - "grad_norm": 5.708250522613525, - "learning_rate": 1.5252447475525245e-05, - "loss": 0.9656, + "grad_norm": 4.256518363952637, + "learning_rate": 1.5254647453525467e-05, + "loss": 0.891, "step": 2640 }, { "epoch": 1.3843843843843844, - "grad_norm": 5.420493125915527, - "learning_rate": 1.5230447695523045e-05, - "loss": 0.8856, + "grad_norm": 6.17589807510376, + "learning_rate": 1.5232647673523267e-05, + "loss": 0.7736, "step": 2650 }, { "epoch": 1.3896069983026504, - "grad_norm": 4.980160713195801, - "learning_rate": 1.5208447915520845e-05, - "loss": 0.8956, + "grad_norm": 5.511795520782471, + "learning_rate": 1.5210647893521066e-05, + "loss": 0.8211, "step": 2660 }, { "epoch": 1.3948296122209165, - "grad_norm": 6.127252101898193, - "learning_rate": 1.5186448135518645e-05, - "loss": 0.8999, + "grad_norm": 4.086550712585449, + "learning_rate": 1.5188648113518866e-05, + "loss": 0.7784, "step": 2670 }, { "epoch": 1.4000522261391826, - "grad_norm": 5.1474080085754395, - "learning_rate": 1.5164448355516446e-05, - "loss": 0.8222, + "grad_norm": 8.598963737487793, + "learning_rate": 1.5166648333516668e-05, + "loss": 0.8549, "step": 2680 }, { "epoch": 1.4052748400574488, - "grad_norm": 4.6120476722717285, - "learning_rate": 1.5142448575514246e-05, - "loss": 0.8835, + "grad_norm": 5.190873146057129, + "learning_rate": 1.5144648553514468e-05, + "loss": 0.9681, "step": 2690 }, { "epoch": 1.410497453975715, - "grad_norm": 4.124380111694336, - "learning_rate": 1.5120448795512046e-05, - "loss": 0.7693, + "grad_norm": 3.509096145629883, + "learning_rate": 1.5122648773512266e-05, + "loss": 0.7279, "step": 2700 }, { "epoch": 1.415720067893981, - "grad_norm": 5.343372821807861, - "learning_rate": 1.5098449015509846e-05, - "loss": 0.7564, + "grad_norm": 4.427550792694092, + "learning_rate": 1.5100648993510066e-05, + "loss": 0.7962, "step": 2710 }, { "epoch": 1.420942681812247, - "grad_norm": 4.66859245300293, - "learning_rate": 1.5076449235507646e-05, - "loss": 0.9316, + "grad_norm": 4.1175312995910645, + "learning_rate": 1.5078649213507865e-05, + "loss": 0.7329, "step": 2720 }, { "epoch": 1.426165295730513, - "grad_norm": 5.207456111907959, - "learning_rate": 1.5054449455505447e-05, - "loss": 0.8185, + "grad_norm": 4.431763648986816, + "learning_rate": 1.5056649433505665e-05, + "loss": 0.7636, "step": 2730 }, { "epoch": 1.4313879096487792, - "grad_norm": 5.222368240356445, - "learning_rate": 1.5032449675503247e-05, - "loss": 0.763, + "grad_norm": 5.301019191741943, + "learning_rate": 1.5034649653503467e-05, + "loss": 0.6933, "step": 2740 }, { "epoch": 1.4366105235670452, - "grad_norm": 4.925426959991455, - "learning_rate": 1.5010449895501047e-05, - "loss": 0.843, + "grad_norm": 5.713370323181152, + "learning_rate": 1.5012649873501266e-05, + "loss": 0.8076, "step": 2750 }, { "epoch": 1.4418331374853115, - "grad_norm": 4.546454429626465, - "learning_rate": 1.4988450115498847e-05, - "loss": 0.8244, + "grad_norm": 5.600203037261963, + "learning_rate": 1.4990650093499066e-05, + "loss": 0.8204, "step": 2760 }, { "epoch": 1.4470557514035776, - "grad_norm": 5.662976264953613, - "learning_rate": 1.4966450335496646e-05, - "loss": 0.7424, + "grad_norm": 5.35227108001709, + "learning_rate": 1.4968650313496866e-05, + "loss": 0.7246, "step": 2770 }, { "epoch": 1.4522783653218436, - "grad_norm": 6.152126789093018, - "learning_rate": 1.4944450555494445e-05, - "loss": 0.685, + "grad_norm": 6.093803405761719, + "learning_rate": 1.4946650533494666e-05, + "loss": 0.7841, "step": 2780 }, { "epoch": 1.4575009792401097, - "grad_norm": 4.776817321777344, - "learning_rate": 1.4922450775492246e-05, - "loss": 0.8083, + "grad_norm": 5.49162483215332, + "learning_rate": 1.4924650753492467e-05, + "loss": 0.7709, "step": 2790 }, { "epoch": 1.4627235931583757, - "grad_norm": 3.0789973735809326, - "learning_rate": 1.4900450995490046e-05, - "loss": 0.7356, + "grad_norm": 5.1611247062683105, + "learning_rate": 1.4902650973490267e-05, + "loss": 0.7498, "step": 2800 }, { "epoch": 1.4679462070766418, - "grad_norm": 4.335427284240723, - "learning_rate": 1.4878451215487846e-05, - "loss": 0.8127, + "grad_norm": 5.545971393585205, + "learning_rate": 1.4880651193488067e-05, + "loss": 0.7823, "step": 2810 }, { "epoch": 1.4731688209949079, - "grad_norm": 6.07483434677124, - "learning_rate": 1.4856451435485646e-05, - "loss": 0.7791, + "grad_norm": 5.807368755340576, + "learning_rate": 1.4858651413485867e-05, + "loss": 0.8116, "step": 2820 }, { "epoch": 1.4783914349131742, - "grad_norm": 5.220136642456055, - "learning_rate": 1.4834451655483445e-05, - "loss": 0.715, + "grad_norm": 5.002223968505859, + "learning_rate": 1.4836651633483667e-05, + "loss": 0.6767, "step": 2830 }, { "epoch": 1.4836140488314402, - "grad_norm": 5.2860426902771, - "learning_rate": 1.4812451875481247e-05, - "loss": 0.8689, + "grad_norm": 4.863631725311279, + "learning_rate": 1.4814651853481465e-05, + "loss": 0.97, "step": 2840 }, { "epoch": 1.4888366627497063, - "grad_norm": 4.845248222351074, - "learning_rate": 1.4790452095479047e-05, - "loss": 0.8165, + "grad_norm": 3.5901687145233154, + "learning_rate": 1.4792652073479265e-05, + "loss": 0.7176, "step": 2850 }, { "epoch": 1.4940592766679723, - "grad_norm": 5.356618404388428, - "learning_rate": 1.4768452315476847e-05, - "loss": 0.779, + "grad_norm": 6.4546966552734375, + "learning_rate": 1.4770652293477066e-05, + "loss": 0.7844, "step": 2860 }, { "epoch": 1.4992818905862384, - "grad_norm": 5.140944957733154, - "learning_rate": 1.4746452535474646e-05, - "loss": 0.7765, + "grad_norm": 4.67232608795166, + "learning_rate": 1.4748652513474866e-05, + "loss": 0.59, "step": 2870 }, { "epoch": 1.5045045045045045, - "grad_norm": 4.167046070098877, - "learning_rate": 1.4724452755472446e-05, - "loss": 0.7353, + "grad_norm": 5.926489353179932, + "learning_rate": 1.4726652733472666e-05, + "loss": 0.7869, "step": 2880 }, { "epoch": 1.5097271184227705, - "grad_norm": 6.490098476409912, - "learning_rate": 1.4702452975470248e-05, - "loss": 0.8059, + "grad_norm": 5.038000106811523, + "learning_rate": 1.4704652953470466e-05, + "loss": 0.8541, "step": 2890 }, { "epoch": 1.5149497323410368, - "grad_norm": 5.053187847137451, - "learning_rate": 1.4680453195468048e-05, - "loss": 0.8564, + "grad_norm": 5.867493629455566, + "learning_rate": 1.4682653173468267e-05, + "loss": 0.8335, "step": 2900 }, { "epoch": 1.5201723462593026, - "grad_norm": 4.322595119476318, - "learning_rate": 1.4658453415465847e-05, - "loss": 0.8182, + "grad_norm": 4.438129425048828, + "learning_rate": 1.4660653393466067e-05, + "loss": 0.7051, "step": 2910 }, { "epoch": 1.525394960177569, - "grad_norm": 3.603776454925537, - "learning_rate": 1.4636453635463646e-05, - "loss": 0.9616, + "grad_norm": 4.732497692108154, + "learning_rate": 1.4638653613463867e-05, + "loss": 0.6931, "step": 2920 }, { "epoch": 1.530617574095835, - "grad_norm": 4.849047660827637, - "learning_rate": 1.4614453855461445e-05, - "loss": 0.7226, + "grad_norm": 7.7729268074035645, + "learning_rate": 1.4616653833461667e-05, + "loss": 0.7065, "step": 2930 }, { "epoch": 1.535840188014101, - "grad_norm": 6.261702537536621, - "learning_rate": 1.4592454075459245e-05, - "loss": 0.7724, + "grad_norm": 4.613509654998779, + "learning_rate": 1.4594654053459467e-05, + "loss": 0.7972, "step": 2940 }, { "epoch": 1.541062801932367, - "grad_norm": 4.070964336395264, - "learning_rate": 1.4570454295457047e-05, - "loss": 0.6157, + "grad_norm": 5.332968235015869, + "learning_rate": 1.4572654273457268e-05, + "loss": 0.692, "step": 2950 }, { "epoch": 1.5462854158506332, - "grad_norm": 5.896931171417236, - "learning_rate": 1.4548454515454847e-05, - "loss": 0.7687, + "grad_norm": 5.396950721740723, + "learning_rate": 1.4550654493455068e-05, + "loss": 0.6387, "step": 2960 }, { "epoch": 1.5515080297688995, - "grad_norm": 4.215420722961426, - "learning_rate": 1.4526454735452646e-05, - "loss": 0.7984, + "grad_norm": 6.194580554962158, + "learning_rate": 1.4528654713452868e-05, + "loss": 0.7482, "step": 2970 }, { "epoch": 1.5567306436871653, - "grad_norm": 5.551188945770264, - "learning_rate": 1.4504454955450446e-05, - "loss": 0.7284, + "grad_norm": 2.6593708992004395, + "learning_rate": 1.4506654933450666e-05, + "loss": 0.8072, "step": 2980 }, { "epoch": 1.5619532576054316, - "grad_norm": 7.92472505569458, - "learning_rate": 1.4482455175448246e-05, - "loss": 0.7007, + "grad_norm": 4.231916904449463, + "learning_rate": 1.4484655153448466e-05, + "loss": 0.7337, "step": 2990 }, { "epoch": 1.5671758715236976, - "grad_norm": 4.330050468444824, - "learning_rate": 1.4460455395446048e-05, - "loss": 0.948, + "grad_norm": 5.672402858734131, + "learning_rate": 1.4462655373446265e-05, + "loss": 0.8569, "step": 3000 }, { "epoch": 1.5671758715236976, - "eval_loss": 0.9684945344924927, - "eval_runtime": 46.3792, - "eval_samples_per_second": 36.697, - "eval_steps_per_second": 4.593, + "eval_loss": 0.9739871621131897, + "eval_runtime": 46.4553, + "eval_samples_per_second": 36.637, + "eval_steps_per_second": 4.585, "step": 3000 }, { "epoch": 1.5723984854419637, - "grad_norm": 4.803574085235596, - "learning_rate": 1.4438455615443847e-05, - "loss": 0.7733, + "grad_norm": 5.696955680847168, + "learning_rate": 1.4440655593444067e-05, + "loss": 0.8267, "step": 3010 }, { "epoch": 1.5776210993602298, - "grad_norm": 3.641155481338501, - "learning_rate": 1.4416455835441647e-05, - "loss": 0.7893, + "grad_norm": 4.695171356201172, + "learning_rate": 1.4418655813441867e-05, + "loss": 0.8341, "step": 3020 }, { "epoch": 1.5828437132784958, - "grad_norm": 4.811826705932617, - "learning_rate": 1.4394456055439447e-05, - "loss": 0.9068, + "grad_norm": 5.929087162017822, + "learning_rate": 1.4396656033439667e-05, + "loss": 0.6473, "step": 3030 }, { "epoch": 1.588066327196762, - "grad_norm": 4.911914348602295, - "learning_rate": 1.4372456275437247e-05, - "loss": 0.8413, + "grad_norm": 5.73638916015625, + "learning_rate": 1.4374656253437466e-05, + "loss": 0.6306, "step": 3040 }, { "epoch": 1.593288941115028, - "grad_norm": 6.21580171585083, - "learning_rate": 1.4350456495435048e-05, - "loss": 0.7968, + "grad_norm": 2.860508680343628, + "learning_rate": 1.4352656473435266e-05, + "loss": 0.6554, "step": 3050 }, { "epoch": 1.5985115550332942, - "grad_norm": 3.3916568756103516, - "learning_rate": 1.4328456715432846e-05, - "loss": 0.7357, + "grad_norm": 4.515196323394775, + "learning_rate": 1.4330656693433068e-05, + "loss": 0.7652, "step": 3060 }, { "epoch": 1.6037341689515603, - "grad_norm": 4.651008605957031, - "learning_rate": 1.4306456935430646e-05, - "loss": 0.823, + "grad_norm": 5.388582229614258, + "learning_rate": 1.4308656913430868e-05, + "loss": 0.7102, "step": 3070 }, { "epoch": 1.6089567828698264, - "grad_norm": 3.533646583557129, - "learning_rate": 1.4284457155428446e-05, - "loss": 0.8333, + "grad_norm": 4.253854274749756, + "learning_rate": 1.4286657133428667e-05, + "loss": 0.7916, "step": 3080 }, { "epoch": 1.6141793967880924, - "grad_norm": 5.906960487365723, - "learning_rate": 1.4262457375426246e-05, - "loss": 0.9569, + "grad_norm": 5.374643802642822, + "learning_rate": 1.4264657353426467e-05, + "loss": 0.7536, "step": 3090 }, { "epoch": 1.6194020107063585, - "grad_norm": 5.581897258758545, - "learning_rate": 1.4240457595424046e-05, - "loss": 0.7276, + "grad_norm": 7.641873359680176, + "learning_rate": 1.4242657573424267e-05, + "loss": 0.6292, "step": 3100 }, { "epoch": 1.6246246246246248, - "grad_norm": 5.3108439445495605, - "learning_rate": 1.4218457815421847e-05, - "loss": 0.7249, + "grad_norm": 5.747442245483398, + "learning_rate": 1.4220657793422069e-05, + "loss": 0.7622, "step": 3110 }, { "epoch": 1.6298472385428906, - "grad_norm": 9.554184913635254, - "learning_rate": 1.4196458035419647e-05, - "loss": 0.8382, + "grad_norm": 5.221161842346191, + "learning_rate": 1.4198658013419867e-05, + "loss": 0.5696, "step": 3120 }, { "epoch": 1.6350698524611569, - "grad_norm": 5.487770080566406, - "learning_rate": 1.4174458255417447e-05, - "loss": 0.8928, + "grad_norm": 4.334496974945068, + "learning_rate": 1.4176658233417667e-05, + "loss": 0.7446, "step": 3130 }, { "epoch": 1.640292466379423, - "grad_norm": 4.557746887207031, - "learning_rate": 1.4152458475415247e-05, - "loss": 0.7369, + "grad_norm": 3.933018922805786, + "learning_rate": 1.4154658453415466e-05, + "loss": 0.7325, "step": 3140 }, { "epoch": 1.645515080297689, - "grad_norm": 4.392518043518066, - "learning_rate": 1.4130458695413047e-05, - "loss": 0.8775, + "grad_norm": 3.7359838485717773, + "learning_rate": 1.4132658673413266e-05, + "loss": 0.8214, "step": 3150 }, { "epoch": 1.650737694215955, - "grad_norm": 3.4565000534057617, - "learning_rate": 1.4108458915410848e-05, - "loss": 0.7641, + "grad_norm": 5.430748462677002, + "learning_rate": 1.4110658893411066e-05, + "loss": 0.7947, "step": 3160 }, { "epoch": 1.6559603081342211, - "grad_norm": 3.457012891769409, - "learning_rate": 1.4086459135408648e-05, - "loss": 0.8069, + "grad_norm": 5.932344913482666, + "learning_rate": 1.4088659113408868e-05, + "loss": 0.6978, "step": 3170 }, { "epoch": 1.6611829220524874, - "grad_norm": 4.115066051483154, - "learning_rate": 1.4064459355406448e-05, - "loss": 0.8344, + "grad_norm": 7.3491435050964355, + "learning_rate": 1.4066659333406667e-05, + "loss": 0.7685, "step": 3180 }, { "epoch": 1.6664055359707532, - "grad_norm": 4.821453094482422, - "learning_rate": 1.4042459575404248e-05, - "loss": 0.6791, + "grad_norm": 5.5517425537109375, + "learning_rate": 1.4044659553404467e-05, + "loss": 0.7526, "step": 3190 }, { "epoch": 1.6716281498890195, - "grad_norm": 4.313604354858398, - "learning_rate": 1.4020459795402046e-05, - "loss": 0.7312, + "grad_norm": 5.074345111846924, + "learning_rate": 1.4022659773402267e-05, + "loss": 0.8166, "step": 3200 }, { "epoch": 1.6768507638072856, - "grad_norm": 5.048585891723633, - "learning_rate": 1.3998460015399846e-05, - "loss": 0.8069, + "grad_norm": 10.781622886657715, + "learning_rate": 1.4000659993400067e-05, + "loss": 0.7348, "step": 3210 }, { "epoch": 1.6820733777255517, - "grad_norm": 4.065299987792969, - "learning_rate": 1.3976460235397647e-05, - "loss": 0.8265, + "grad_norm": 4.66840934753418, + "learning_rate": 1.3978660213397868e-05, + "loss": 0.7932, "step": 3220 }, { "epoch": 1.6872959916438177, - "grad_norm": 3.214991569519043, - "learning_rate": 1.3954460455395447e-05, - "loss": 0.7092, + "grad_norm": 3.1479058265686035, + "learning_rate": 1.3956660433395668e-05, + "loss": 0.6879, "step": 3230 }, { "epoch": 1.6925186055620838, - "grad_norm": 7.845371723175049, - "learning_rate": 1.3932460675393247e-05, - "loss": 0.7499, + "grad_norm": 2.8037853240966797, + "learning_rate": 1.3934660653393468e-05, + "loss": 0.77, "step": 3240 }, { "epoch": 1.69774121948035, - "grad_norm": 5.4474921226501465, - "learning_rate": 1.3910460895391046e-05, - "loss": 0.781, + "grad_norm": 5.04011869430542, + "learning_rate": 1.3912660873391268e-05, + "loss": 0.7741, "step": 3250 }, { "epoch": 1.702963833398616, - "grad_norm": 7.43993616104126, - "learning_rate": 1.3888461115388846e-05, - "loss": 0.8104, + "grad_norm": 4.707029819488525, + "learning_rate": 1.3890661093389066e-05, + "loss": 0.682, "step": 3260 }, { "epoch": 1.7081864473168822, - "grad_norm": 5.410302639007568, - "learning_rate": 1.3866461335386648e-05, - "loss": 0.909, + "grad_norm": 4.292649745941162, + "learning_rate": 1.3868661313386866e-05, + "loss": 0.8103, "step": 3270 }, { "epoch": 1.7134090612351482, - "grad_norm": 3.9956214427948, - "learning_rate": 1.3844461555384448e-05, - "loss": 0.7744, + "grad_norm": 4.101726531982422, + "learning_rate": 1.3846661533384667e-05, + "loss": 0.7548, "step": 3280 }, { "epoch": 1.7186316751534143, - "grad_norm": 5.384256839752197, - "learning_rate": 1.3822461775382247e-05, - "loss": 0.7898, + "grad_norm": 3.1450841426849365, + "learning_rate": 1.3824661753382467e-05, + "loss": 0.6595, "step": 3290 }, { "epoch": 1.7238542890716804, - "grad_norm": 4.181092739105225, - "learning_rate": 1.3800461995380047e-05, - "loss": 0.7584, + "grad_norm": 3.2959680557250977, + "learning_rate": 1.3802661973380267e-05, + "loss": 0.5613, "step": 3300 }, { "epoch": 1.7290769029899464, - "grad_norm": 4.823882102966309, - "learning_rate": 1.3778462215377849e-05, - "loss": 0.9165, + "grad_norm": 3.8981027603149414, + "learning_rate": 1.3780662193378067e-05, + "loss": 0.569, "step": 3310 }, { "epoch": 1.7342995169082127, - "grad_norm": 5.339610576629639, - "learning_rate": 1.3756462435375649e-05, - "loss": 0.6867, + "grad_norm": 6.0319743156433105, + "learning_rate": 1.3758662413375867e-05, + "loss": 0.7547, "step": 3320 }, { "epoch": 1.7395221308264786, - "grad_norm": 3.4958078861236572, - "learning_rate": 1.3734462655373448e-05, - "loss": 0.9282, + "grad_norm": 5.254234313964844, + "learning_rate": 1.3736662633373668e-05, + "loss": 0.8363, "step": 3330 }, { "epoch": 1.7447447447447448, - "grad_norm": 4.378716945648193, - "learning_rate": 1.3712462875371247e-05, - "loss": 0.5787, + "grad_norm": 5.830689430236816, + "learning_rate": 1.3714662853371468e-05, + "loss": 0.6791, "step": 3340 }, { "epoch": 1.7499673586630107, - "grad_norm": 9.347067832946777, - "learning_rate": 1.3690463095369046e-05, - "loss": 0.7691, + "grad_norm": 5.3443217277526855, + "learning_rate": 1.3692663073369268e-05, + "loss": 0.5974, "step": 3350 }, { "epoch": 1.755189972581277, - "grad_norm": 3.417569398880005, - "learning_rate": 1.3668463315366846e-05, - "loss": 0.7411, + "grad_norm": 6.489872932434082, + "learning_rate": 1.3670663293367068e-05, + "loss": 0.721, "step": 3360 }, { "epoch": 1.760412586499543, - "grad_norm": 4.640798568725586, - "learning_rate": 1.3646463535364646e-05, - "loss": 0.7407, + "grad_norm": 4.440582275390625, + "learning_rate": 1.3648663513364867e-05, + "loss": 0.7012, "step": 3370 }, { "epoch": 1.765635200417809, - "grad_norm": 5.5894341468811035, - "learning_rate": 1.3624463755362448e-05, - "loss": 0.9005, + "grad_norm": 5.982915878295898, + "learning_rate": 1.3626663733362669e-05, + "loss": 0.8002, "step": 3380 }, { "epoch": 1.7708578143360754, - "grad_norm": 5.964491844177246, - "learning_rate": 1.3602463975360247e-05, - "loss": 0.8281, + "grad_norm": 4.627645969390869, + "learning_rate": 1.3604663953360469e-05, + "loss": 0.7377, "step": 3390 }, { "epoch": 1.7760804282543412, - "grad_norm": 5.019880294799805, - "learning_rate": 1.3580464195358047e-05, - "loss": 0.7772, + "grad_norm": 4.162827014923096, + "learning_rate": 1.3582664173358267e-05, + "loss": 0.6035, "step": 3400 }, { "epoch": 1.7813030421726075, - "grad_norm": 7.011587619781494, - "learning_rate": 1.3558464415355847e-05, - "loss": 0.6852, + "grad_norm": 2.8144357204437256, + "learning_rate": 1.3560664393356067e-05, + "loss": 0.7341, "step": 3410 }, { "epoch": 1.7865256560908733, - "grad_norm": 5.2039055824279785, - "learning_rate": 1.3536464635353647e-05, - "loss": 0.8141, + "grad_norm": 4.323256969451904, + "learning_rate": 1.3538664613353867e-05, + "loss": 0.6689, "step": 3420 }, { "epoch": 1.7917482700091396, - "grad_norm": 5.673884868621826, - "learning_rate": 1.3514464855351448e-05, - "loss": 0.7204, + "grad_norm": 4.568783283233643, + "learning_rate": 1.3516664833351666e-05, + "loss": 0.8375, "step": 3430 }, { "epoch": 1.7969708839274057, - "grad_norm": 4.41068172454834, - "learning_rate": 1.3492465075349248e-05, - "loss": 0.8504, + "grad_norm": 4.666477203369141, + "learning_rate": 1.3494665053349468e-05, + "loss": 0.7715, "step": 3440 }, { "epoch": 1.8021934978456717, - "grad_norm": 4.449737071990967, - "learning_rate": 1.3470465295347048e-05, - "loss": 0.7652, + "grad_norm": 4.528486251831055, + "learning_rate": 1.3472665273347268e-05, + "loss": 0.8249, "step": 3450 }, { "epoch": 1.807416111763938, - "grad_norm": 5.095055103302002, - "learning_rate": 1.3448465515344848e-05, - "loss": 0.8654, + "grad_norm": 3.496216058731079, + "learning_rate": 1.3450665493345068e-05, + "loss": 0.7945, "step": 3460 }, { "epoch": 1.8126387256822039, - "grad_norm": 4.90816068649292, - "learning_rate": 1.342646573534265e-05, - "loss": 0.6933, + "grad_norm": 4.029898166656494, + "learning_rate": 1.3428665713342867e-05, + "loss": 0.6752, "step": 3470 }, { "epoch": 1.8178613396004701, - "grad_norm": 3.835984468460083, + "grad_norm": 6.3212890625, "learning_rate": 1.3406665933340667e-05, - "loss": 0.6106, + "loss": 0.6452, "step": 3480 }, { "epoch": 1.823083953518736, - "grad_norm": 4.385377407073975, + "grad_norm": 3.029170274734497, "learning_rate": 1.3384666153338469e-05, - "loss": 0.797, + "loss": 0.6989, "step": 3490 }, { "epoch": 1.8283065674370023, - "grad_norm": 5.261483192443848, + "grad_norm": 5.18508243560791, "learning_rate": 1.3362666373336269e-05, - "loss": 0.7343, + "loss": 0.6875, "step": 3500 }, { "epoch": 1.8283065674370023, - "eval_loss": 0.9347423911094666, - "eval_runtime": 46.3797, - "eval_samples_per_second": 36.697, - "eval_steps_per_second": 4.593, + "eval_loss": 0.9430766701698303, + "eval_runtime": 46.4982, + "eval_samples_per_second": 36.604, + "eval_steps_per_second": 4.581, "step": 3500 }, { "epoch": 1.8335291813552683, - "grad_norm": 5.269721508026123, + "grad_norm": 4.71134090423584, "learning_rate": 1.3340666593334068e-05, - "loss": 0.7558, + "loss": 0.6619, "step": 3510 }, { "epoch": 1.8387517952735344, - "grad_norm": 4.808032512664795, + "grad_norm": 5.293645858764648, "learning_rate": 1.3318666813331868e-05, - "loss": 0.7597, + "loss": 0.8082, "step": 3520 }, { "epoch": 1.8439744091918004, - "grad_norm": 5.560021877288818, + "grad_norm": 4.22677755355835, "learning_rate": 1.3296667033329668e-05, - "loss": 0.6428, + "loss": 0.632, "step": 3530 }, { "epoch": 1.8491970231100665, - "grad_norm": 4.8384013175964355, + "grad_norm": 6.305282115936279, "learning_rate": 1.3274667253327466e-05, - "loss": 0.8231, + "loss": 0.7162, "step": 3540 }, { "epoch": 1.8544196370283328, - "grad_norm": 5.98041296005249, + "grad_norm": 4.559315204620361, "learning_rate": 1.3252667473325268e-05, - "loss": 0.8127, + "loss": 0.7988, "step": 3550 }, { "epoch": 1.8596422509465986, - "grad_norm": 5.745590686798096, + "grad_norm": 4.691094875335693, "learning_rate": 1.3230667693323068e-05, - "loss": 0.7488, + "loss": 0.7266, "step": 3560 }, { "epoch": 1.864864864864865, - "grad_norm": 6.292532920837402, + "grad_norm": 5.862813949584961, "learning_rate": 1.3208667913320867e-05, - "loss": 0.8194, + "loss": 0.7737, "step": 3570 }, { "epoch": 1.870087478783131, - "grad_norm": 4.489058971405029, + "grad_norm": 5.876034736633301, "learning_rate": 1.3186668133318667e-05, - "loss": 0.8918, + "loss": 0.7683, "step": 3580 }, { "epoch": 1.875310092701397, - "grad_norm": 7.152320861816406, + "grad_norm": 3.93103289604187, "learning_rate": 1.3164668353316467e-05, - "loss": 0.8746, + "loss": 0.6797, "step": 3590 }, { "epoch": 1.880532706619663, - "grad_norm": 5.863579273223877, + "grad_norm": 5.028536796569824, "learning_rate": 1.3142668573314268e-05, - "loss": 0.6692, + "loss": 0.7268, "step": 3600 }, { "epoch": 1.8857553205379292, - "grad_norm": 5.481932163238525, + "grad_norm": 5.717294692993164, "learning_rate": 1.3120668793312068e-05, - "loss": 0.9529, + "loss": 0.7067, "step": 3610 }, { "epoch": 1.8909779344561954, - "grad_norm": 5.673171043395996, + "grad_norm": 4.150835037231445, "learning_rate": 1.3098669013309868e-05, - "loss": 0.7145, + "loss": 0.809, "step": 3620 }, { "epoch": 1.8962005483744613, - "grad_norm": 3.1669821739196777, + "grad_norm": 7.41292667388916, "learning_rate": 1.3076669233307668e-05, - "loss": 0.6498, + "loss": 0.7238, "step": 3630 }, { "epoch": 1.9014231622927276, - "grad_norm": 5.213048934936523, + "grad_norm": 5.053524494171143, "learning_rate": 1.3054669453305468e-05, - "loss": 0.8128, + "loss": 0.8901, "step": 3640 }, { "epoch": 1.9066457762109936, - "grad_norm": 3.5980887413024902, + "grad_norm": 4.0097832679748535, "learning_rate": 1.303266967330327e-05, - "loss": 0.692, + "loss": 0.5934, "step": 3650 }, { "epoch": 1.9118683901292597, - "grad_norm": 3.2975611686706543, + "grad_norm": 4.046881198883057, "learning_rate": 1.3010669893301069e-05, - "loss": 0.6698, + "loss": 0.5674, "step": 3660 }, { "epoch": 1.9170910040475257, - "grad_norm": 4.254265308380127, + "grad_norm": 5.750004768371582, "learning_rate": 1.2988670113298869e-05, - "loss": 0.7827, + "loss": 0.7995, "step": 3670 }, { "epoch": 1.9223136179657918, - "grad_norm": 4.849771022796631, + "grad_norm": 5.423394203186035, "learning_rate": 1.2966670333296667e-05, - "loss": 0.6954, + "loss": 0.8165, "step": 3680 }, { "epoch": 1.927536231884058, - "grad_norm": 4.976161956787109, + "grad_norm": 4.755868434906006, "learning_rate": 1.2944670553294467e-05, - "loss": 0.6382, + "loss": 0.7073, "step": 3690 }, { "epoch": 1.932758845802324, - "grad_norm": 5.311154365539551, + "grad_norm": 4.774918079376221, "learning_rate": 1.2922670773292267e-05, - "loss": 0.6891, + "loss": 0.7146, "step": 3700 }, { "epoch": 1.9379814597205902, - "grad_norm": 6.911156177520752, + "grad_norm": 4.875230312347412, "learning_rate": 1.2900670993290068e-05, - "loss": 0.861, + "loss": 0.6761, "step": 3710 }, { "epoch": 1.9432040736388563, - "grad_norm": 4.014489650726318, + "grad_norm": 5.57156229019165, "learning_rate": 1.2878671213287868e-05, - "loss": 0.8185, + "loss": 0.7728, "step": 3720 }, { "epoch": 1.9484266875571223, - "grad_norm": 5.282459259033203, + "grad_norm": 5.970896244049072, "learning_rate": 1.2856671433285668e-05, - "loss": 0.7983, + "loss": 0.729, "step": 3730 }, { "epoch": 1.9536493014753884, - "grad_norm": 4.00164794921875, + "grad_norm": 4.957197189331055, "learning_rate": 1.2834671653283468e-05, - "loss": 0.7628, + "loss": 0.7734, "step": 3740 }, { "epoch": 1.9588719153936545, - "grad_norm": 5.199289798736572, + "grad_norm": 6.597054481506348, "learning_rate": 1.2812671873281268e-05, - "loss": 0.8203, + "loss": 0.7263, "step": 3750 }, { "epoch": 1.9640945293119207, - "grad_norm": 4.595690727233887, + "grad_norm": 5.362037658691406, "learning_rate": 1.2790672093279069e-05, - "loss": 0.6234, + "loss": 0.7644, "step": 3760 }, { "epoch": 1.9693171432301866, - "grad_norm": 4.965597152709961, + "grad_norm": 4.6790337562561035, "learning_rate": 1.2768672313276869e-05, - "loss": 0.7896, + "loss": 0.7415, "step": 3770 }, { "epoch": 1.9745397571484529, - "grad_norm": 6.135955810546875, + "grad_norm": 6.111202239990234, "learning_rate": 1.2746672533274669e-05, - "loss": 0.7313, + "loss": 0.6979, "step": 3780 }, { "epoch": 1.979762371066719, - "grad_norm": 3.6871345043182373, + "grad_norm": 3.712984085083008, "learning_rate": 1.2724672753272469e-05, - "loss": 0.9234, + "loss": 0.7742, "step": 3790 }, { "epoch": 1.984984984984985, - "grad_norm": 4.930665493011475, + "grad_norm": 5.395435810089111, "learning_rate": 1.2702672973270268e-05, - "loss": 0.9072, + "loss": 0.7578, "step": 3800 }, { "epoch": 1.990207598903251, - "grad_norm": 4.443655014038086, + "grad_norm": 7.7390289306640625, "learning_rate": 1.268067319326807e-05, - "loss": 0.7923, + "loss": 0.7587, "step": 3810 }, { "epoch": 1.9954302128215171, - "grad_norm": 4.998082637786865, + "grad_norm": 4.17903995513916, "learning_rate": 1.2658673413265868e-05, - "loss": 0.7933, + "loss": 0.7134, "step": 3820 }, { "epoch": 2.001044522783653, - "grad_norm": 2.840280294418335, + "grad_norm": 5.0334625244140625, "learning_rate": 1.2636673633263668e-05, - "loss": 0.7225, + "loss": 0.6952, "step": 3830 }, { "epoch": 2.0062671367019194, - "grad_norm": 5.853653907775879, + "grad_norm": 4.7001261711120605, "learning_rate": 1.2614673853261468e-05, - "loss": 0.548, + "loss": 0.4677, "step": 3840 }, { "epoch": 2.011489750620185, - "grad_norm": 3.085155725479126, + "grad_norm": 5.617419242858887, "learning_rate": 1.2592674073259267e-05, - "loss": 0.5502, + "loss": 0.4932, "step": 3850 }, { "epoch": 2.0167123645384515, - "grad_norm": 5.955913543701172, + "grad_norm": 5.361858367919922, "learning_rate": 1.2570674293257067e-05, - "loss": 0.5229, + "loss": 0.5958, "step": 3860 }, { "epoch": 2.0219349784567178, - "grad_norm": 5.1260905265808105, + "grad_norm": 5.064599514007568, "learning_rate": 1.2548674513254869e-05, - "loss": 0.6059, + "loss": 0.5058, "step": 3870 }, { "epoch": 2.0271575923749836, - "grad_norm": 2.684438943862915, + "grad_norm": 6.71765661239624, "learning_rate": 1.2526674733252669e-05, - "loss": 0.5465, + "loss": 0.5446, "step": 3880 }, { "epoch": 2.03238020629325, - "grad_norm": 4.3153533935546875, - "learning_rate": 1.2504674953250468e-05, - "loss": 0.5319, + "grad_norm": 3.904301643371582, + "learning_rate": 1.2506874931250688e-05, + "loss": 0.4895, "step": 3890 }, { "epoch": 2.0376028202115157, - "grad_norm": 5.260719299316406, - "learning_rate": 1.2482675173248268e-05, - "loss": 0.4983, + "grad_norm": 6.343670845031738, + "learning_rate": 1.2484875151248488e-05, + "loss": 0.4937, "step": 3900 }, { "epoch": 2.042825434129782, - "grad_norm": 4.053082466125488, - "learning_rate": 1.2460675393246068e-05, - "loss": 0.6385, + "grad_norm": 3.2094414234161377, + "learning_rate": 1.2462875371246288e-05, + "loss": 0.4955, "step": 3910 }, { "epoch": 2.048048048048048, - "grad_norm": 4.619148254394531, - "learning_rate": 1.243867561324387e-05, - "loss": 0.7165, + "grad_norm": 4.333749771118164, + "learning_rate": 1.2440875591244088e-05, + "loss": 0.4895, "step": 3920 }, { "epoch": 2.053270661966314, - "grad_norm": 6.127100944519043, - "learning_rate": 1.241667583324167e-05, - "loss": 0.5853, + "grad_norm": 3.120941162109375, + "learning_rate": 1.2418875811241889e-05, + "loss": 0.4815, "step": 3930 }, { "epoch": 2.0584932758845804, - "grad_norm": 6.0645432472229, - "learning_rate": 1.239467605323947e-05, - "loss": 0.6654, + "grad_norm": 6.374571323394775, + "learning_rate": 1.2396876031239689e-05, + "loss": 0.5032, "step": 3940 }, { "epoch": 2.0637158898028463, - "grad_norm": 4.367906093597412, - "learning_rate": 1.2372676273237269e-05, - "loss": 0.5563, + "grad_norm": 3.659161329269409, + "learning_rate": 1.2374876251237489e-05, + "loss": 0.4167, "step": 3950 }, { "epoch": 2.0689385037211125, - "grad_norm": 6.469724178314209, - "learning_rate": 1.2350676493235067e-05, - "loss": 0.5929, + "grad_norm": 5.340208053588867, + "learning_rate": 1.2352876471235289e-05, + "loss": 0.5707, "step": 3960 }, { "epoch": 2.0741611176393784, - "grad_norm": 3.372661828994751, - "learning_rate": 1.2328676713232867e-05, - "loss": 0.5706, + "grad_norm": 4.3111186027526855, + "learning_rate": 1.2330876691233088e-05, + "loss": 0.4974, "step": 3970 }, { "epoch": 2.0793837315576447, - "grad_norm": 6.373135566711426, - "learning_rate": 1.2306676933230669e-05, - "loss": 0.5192, + "grad_norm": 5.229679584503174, + "learning_rate": 1.230887691123089e-05, + "loss": 0.4687, "step": 3980 }, { "epoch": 2.0846063454759105, - "grad_norm": 5.228106498718262, - "learning_rate": 1.2284677153228468e-05, - "loss": 0.6097, + "grad_norm": 4.563955307006836, + "learning_rate": 1.228687713122869e-05, + "loss": 0.5005, "step": 3990 }, { "epoch": 2.089828959394177, - "grad_norm": 6.330024719238281, - "learning_rate": 1.2262677373226268e-05, - "loss": 0.5799, + "grad_norm": 5.317227363586426, + "learning_rate": 1.226487735122649e-05, + "loss": 0.5475, "step": 4000 }, { "epoch": 2.089828959394177, - "eval_loss": 0.9453912377357483, - "eval_runtime": 46.3882, - "eval_samples_per_second": 36.69, - "eval_steps_per_second": 4.592, + "eval_loss": 0.9593453407287598, + "eval_runtime": 46.3145, + "eval_samples_per_second": 36.749, + "eval_steps_per_second": 4.599, "step": 4000 }, { "epoch": 2.095051573312443, - "grad_norm": 3.039299964904785, - "learning_rate": 1.2240677593224068e-05, - "loss": 0.6099, + "grad_norm": 4.345738410949707, + "learning_rate": 1.224287757122429e-05, + "loss": 0.4033, "step": 4010 }, { "epoch": 2.100274187230709, - "grad_norm": 5.28336763381958, - "learning_rate": 1.2218677813221868e-05, - "loss": 0.5661, + "grad_norm": 5.5209174156188965, + "learning_rate": 1.2220877791222088e-05, + "loss": 0.4719, "step": 4020 }, { "epoch": 2.105496801148975, - "grad_norm": 6.230477333068848, - "learning_rate": 1.219667803321967e-05, - "loss": 0.5823, + "grad_norm": 4.070075988769531, + "learning_rate": 1.2198878011219887e-05, + "loss": 0.464, "step": 4030 }, { "epoch": 2.110719415067241, - "grad_norm": 6.373923301696777, - "learning_rate": 1.217467825321747e-05, - "loss": 0.5674, + "grad_norm": 4.487546443939209, + "learning_rate": 1.2176878231217689e-05, + "loss": 0.4228, "step": 4040 }, { "epoch": 2.1159420289855073, - "grad_norm": 4.496073246002197, - "learning_rate": 1.2152678473215269e-05, - "loss": 0.5449, + "grad_norm": 4.75532341003418, + "learning_rate": 1.2154878451215489e-05, + "loss": 0.4523, "step": 4050 }, { "epoch": 2.121164642903773, - "grad_norm": 5.086572170257568, - "learning_rate": 1.2130678693213069e-05, - "loss": 0.478, + "grad_norm": 3.34230375289917, + "learning_rate": 1.2132878671213289e-05, + "loss": 0.4985, "step": 4060 }, { "epoch": 2.1263872568220394, - "grad_norm": 3.9062867164611816, - "learning_rate": 1.2108678913210869e-05, - "loss": 0.4974, + "grad_norm": 4.725452423095703, + "learning_rate": 1.2110878891211088e-05, + "loss": 0.5518, "step": 4070 }, { "epoch": 2.1316098707403057, - "grad_norm": 5.478116989135742, - "learning_rate": 1.208667913320867e-05, - "loss": 0.6237, + "grad_norm": 4.407351970672607, + "learning_rate": 1.2088879111208888e-05, + "loss": 0.5639, "step": 4080 }, { "epoch": 2.1368324846585716, - "grad_norm": 4.788442611694336, - "learning_rate": 1.206467935320647e-05, - "loss": 0.4777, + "grad_norm": 4.79113245010376, + "learning_rate": 1.206687933120669e-05, + "loss": 0.4428, "step": 4090 }, { "epoch": 2.142055098576838, - "grad_norm": 5.6343889236450195, - "learning_rate": 1.2042679573204268e-05, - "loss": 0.5733, + "grad_norm": 5.133504390716553, + "learning_rate": 1.204487955120449e-05, + "loss": 0.5339, "step": 4100 }, { "epoch": 2.1472777124951037, - "grad_norm": 3.3089094161987305, - "learning_rate": 1.2020679793202068e-05, - "loss": 0.4911, + "grad_norm": 5.313210964202881, + "learning_rate": 1.202287977120229e-05, + "loss": 0.561, "step": 4110 }, { "epoch": 2.15250032641337, - "grad_norm": 4.506272315979004, - "learning_rate": 1.1998680013199868e-05, - "loss": 0.516, + "grad_norm": 4.001386642456055, + "learning_rate": 1.200087999120009e-05, + "loss": 0.3734, "step": 4120 }, { "epoch": 2.157722940331636, - "grad_norm": 5.48958158493042, - "learning_rate": 1.1976680233197668e-05, - "loss": 0.5535, + "grad_norm": 5.470261096954346, + "learning_rate": 1.1978880211197889e-05, + "loss": 0.5102, "step": 4130 }, { "epoch": 2.162945554249902, - "grad_norm": 4.435242652893066, - "learning_rate": 1.195468045319547e-05, - "loss": 0.5861, + "grad_norm": 5.0790791511535645, + "learning_rate": 1.195688043119569e-05, + "loss": 0.5521, "step": 4140 }, { "epoch": 2.1681681681681684, - "grad_norm": 4.332143306732178, - "learning_rate": 1.1932680673193269e-05, - "loss": 0.5009, + "grad_norm": 3.801112413406372, + "learning_rate": 1.193488065119349e-05, + "loss": 0.4709, "step": 4150 }, { "epoch": 2.173390782086434, - "grad_norm": 4.4287848472595215, - "learning_rate": 1.1910680893191069e-05, - "loss": 0.4957, + "grad_norm": 6.1106181144714355, + "learning_rate": 1.1912880871191288e-05, + "loss": 0.4338, "step": 4160 }, { "epoch": 2.1786133960047005, - "grad_norm": 5.4941229820251465, - "learning_rate": 1.1888681113188869e-05, - "loss": 0.5236, + "grad_norm": 5.111484050750732, + "learning_rate": 1.1890881091189088e-05, + "loss": 0.5373, "step": 4170 }, { "epoch": 2.1838360099229663, - "grad_norm": 3.3228816986083984, - "learning_rate": 1.1866681333186668e-05, - "loss": 0.4355, + "grad_norm": 4.768340587615967, + "learning_rate": 1.1868881311186888e-05, + "loss": 0.4295, "step": 4180 }, { "epoch": 2.1890586238412326, - "grad_norm": 6.6818952560424805, - "learning_rate": 1.184468155318447e-05, - "loss": 0.4244, + "grad_norm": 6.232722282409668, + "learning_rate": 1.1846881531184688e-05, + "loss": 0.545, "step": 4190 }, { "epoch": 2.1942812377594985, - "grad_norm": 4.551324844360352, - "learning_rate": 1.182268177318227e-05, - "loss": 0.5724, + "grad_norm": 6.43731689453125, + "learning_rate": 1.182488175118249e-05, + "loss": 0.4732, "step": 4200 }, { "epoch": 2.1995038516777647, - "grad_norm": 3.3153746128082275, - "learning_rate": 1.180068199318007e-05, - "loss": 0.4918, + "grad_norm": 3.004621744155884, + "learning_rate": 1.180288197118029e-05, + "loss": 0.4664, "step": 4210 }, { "epoch": 2.204726465596031, - "grad_norm": 4.966384410858154, - "learning_rate": 1.177868221317787e-05, - "loss": 0.4823, + "grad_norm": 4.8401031494140625, + "learning_rate": 1.1780882191178089e-05, + "loss": 0.4279, "step": 4220 }, { "epoch": 2.209949079514297, - "grad_norm": 4.28817081451416, - "learning_rate": 1.1756682433175671e-05, - "loss": 0.4717, + "grad_norm": 3.1851627826690674, + "learning_rate": 1.1758882411175889e-05, + "loss": 0.4551, "step": 4230 }, { "epoch": 2.215171693432563, - "grad_norm": 3.8175127506256104, - "learning_rate": 1.173468265317347e-05, - "loss": 0.4902, + "grad_norm": 5.632800102233887, + "learning_rate": 1.1736882631173689e-05, + "loss": 0.4816, "step": 4240 }, { "epoch": 2.220394307350829, - "grad_norm": 3.9820878505706787, - "learning_rate": 1.1712682873171269e-05, - "loss": 0.5859, + "grad_norm": 4.3907575607299805, + "learning_rate": 1.171488285117149e-05, + "loss": 0.5119, "step": 4250 }, { "epoch": 2.2256169212690953, - "grad_norm": 4.197390079498291, - "learning_rate": 1.1690683093169069e-05, - "loss": 0.5415, + "grad_norm": 3.5601589679718018, + "learning_rate": 1.169288307116929e-05, + "loss": 0.4853, "step": 4260 }, { "epoch": 2.230839535187361, - "grad_norm": 6.945038795471191, - "learning_rate": 1.1668683313166869e-05, - "loss": 0.5892, + "grad_norm": 5.264835834503174, + "learning_rate": 1.167088329116709e-05, + "loss": 0.4494, "step": 4270 }, { "epoch": 2.2360621491056274, - "grad_norm": 4.9087982177734375, - "learning_rate": 1.1646683533164668e-05, - "loss": 0.5109, + "grad_norm": 4.804422378540039, + "learning_rate": 1.164888351116489e-05, + "loss": 0.5515, "step": 4280 }, { "epoch": 2.2412847630238932, - "grad_norm": 5.819643497467041, - "learning_rate": 1.1624683753162468e-05, - "loss": 0.6031, + "grad_norm": 5.466322898864746, + "learning_rate": 1.162688373116269e-05, + "loss": 0.5372, "step": 4290 }, { "epoch": 2.2465073769421595, - "grad_norm": 5.164392948150635, - "learning_rate": 1.160268397316027e-05, - "loss": 0.5399, + "grad_norm": 8.20328426361084, + "learning_rate": 1.1604883951160488e-05, + "loss": 0.4693, "step": 4300 }, { "epoch": 2.251729990860426, - "grad_norm": 5.87556791305542, - "learning_rate": 1.158068419315807e-05, - "loss": 0.5744, + "grad_norm": 5.77635383605957, + "learning_rate": 1.158288417115829e-05, + "loss": 0.5372, "step": 4310 }, { "epoch": 2.2569526047786916, - "grad_norm": 2.9850735664367676, - "learning_rate": 1.155868441315587e-05, - "loss": 0.4386, + "grad_norm": 3.531426429748535, + "learning_rate": 1.1560884391156089e-05, + "loss": 0.5374, "step": 4320 }, { "epoch": 2.262175218696958, - "grad_norm": 4.8038530349731445, - "learning_rate": 1.153668463315367e-05, - "loss": 0.5148, + "grad_norm": 4.663374900817871, + "learning_rate": 1.1538884611153889e-05, + "loss": 0.5393, "step": 4330 }, { "epoch": 2.2673978326152238, - "grad_norm": 5.2686004638671875, - "learning_rate": 1.151468485315147e-05, - "loss": 0.5089, + "grad_norm": 4.4545722007751465, + "learning_rate": 1.1516884831151689e-05, + "loss": 0.5145, "step": 4340 }, { "epoch": 2.27262044653349, - "grad_norm": 6.421589374542236, - "learning_rate": 1.149268507314927e-05, - "loss": 0.6375, + "grad_norm": 5.173253536224365, + "learning_rate": 1.1494885051149489e-05, + "loss": 0.4239, "step": 4350 }, { "epoch": 2.2778430604517563, - "grad_norm": 5.114859104156494, - "learning_rate": 1.147068529314707e-05, - "loss": 0.4898, + "grad_norm": 4.701399803161621, + "learning_rate": 1.147288527114729e-05, + "loss": 0.5093, "step": 4360 }, { "epoch": 2.283065674370022, - "grad_norm": 3.2084782123565674, - "learning_rate": 1.144868551314487e-05, - "loss": 0.5357, + "grad_norm": 4.533555030822754, + "learning_rate": 1.145088549114509e-05, + "loss": 0.439, "step": 4370 }, { "epoch": 2.2882882882882885, - "grad_norm": 4.514894962310791, - "learning_rate": 1.142668573314267e-05, - "loss": 0.4976, + "grad_norm": 6.554306983947754, + "learning_rate": 1.142888571114289e-05, + "loss": 0.5251, "step": 4380 }, { "epoch": 2.2935109022065543, - "grad_norm": 4.541710376739502, - "learning_rate": 1.1404685953140468e-05, - "loss": 0.5235, + "grad_norm": 2.6566102504730225, + "learning_rate": 1.140688593114069e-05, + "loss": 0.4977, "step": 4390 }, { "epoch": 2.2987335161248206, - "grad_norm": 3.9480032920837402, - "learning_rate": 1.1382686173138268e-05, - "loss": 0.571, + "grad_norm": 3.8855392932891846, + "learning_rate": 1.138488615113849e-05, + "loss": 0.4881, "step": 4400 }, { "epoch": 2.3039561300430864, - "grad_norm": 5.118963718414307, - "learning_rate": 1.136068639313607e-05, - "loss": 0.5245, + "grad_norm": 6.29968786239624, + "learning_rate": 1.1362886371136291e-05, + "loss": 0.5018, "step": 4410 }, { "epoch": 2.3091787439613527, - "grad_norm": 4.763105869293213, - "learning_rate": 1.133868661313387e-05, - "loss": 0.6002, + "grad_norm": 5.406229496002197, + "learning_rate": 1.134088659113409e-05, + "loss": 0.5306, "step": 4420 }, { "epoch": 2.3144013578796185, - "grad_norm": 3.849142074584961, - "learning_rate": 1.1316686833131669e-05, - "loss": 0.6067, + "grad_norm": 3.506462574005127, + "learning_rate": 1.131888681113189e-05, + "loss": 0.462, "step": 4430 }, { "epoch": 2.319623971797885, - "grad_norm": 6.650665760040283, - "learning_rate": 1.1294687053129469e-05, - "loss": 0.5129, + "grad_norm": 4.1984758377075195, + "learning_rate": 1.1296887031129689e-05, + "loss": 0.5107, "step": 4440 }, { "epoch": 2.324846585716151, - "grad_norm": 4.083095073699951, - "learning_rate": 1.1272687273127269e-05, - "loss": 0.5995, + "grad_norm": 4.599320411682129, + "learning_rate": 1.1274887251127488e-05, + "loss": 0.5, "step": 4450 }, { "epoch": 2.330069199634417, - "grad_norm": 5.624079704284668, - "learning_rate": 1.125068749312507e-05, - "loss": 0.5652, + "grad_norm": 6.1861572265625, + "learning_rate": 1.1252887471125288e-05, + "loss": 0.5485, "step": 4460 }, { "epoch": 2.3352918135526832, - "grad_norm": 6.2935028076171875, - "learning_rate": 1.122868771312287e-05, - "loss": 0.5226, + "grad_norm": 5.439957141876221, + "learning_rate": 1.123088769112309e-05, + "loss": 0.4871, "step": 4470 }, { "epoch": 2.340514427470949, - "grad_norm": 7.2632646560668945, - "learning_rate": 1.120668793312067e-05, - "loss": 0.5898, + "grad_norm": 4.366857051849365, + "learning_rate": 1.120888791112089e-05, + "loss": 0.5262, "step": 4480 }, { "epoch": 2.3457370413892153, - "grad_norm": 5.291088104248047, - "learning_rate": 1.118468815311847e-05, - "loss": 0.5239, + "grad_norm": 4.805356979370117, + "learning_rate": 1.118688813111869e-05, + "loss": 0.5043, "step": 4490 }, { "epoch": 2.3509596553074816, - "grad_norm": 4.0247907638549805, - "learning_rate": 1.1162688373116271e-05, - "loss": 0.5263, + "grad_norm": 4.125453948974609, + "learning_rate": 1.116488835111649e-05, + "loss": 0.6307, "step": 4500 }, { "epoch": 2.3509596553074816, - "eval_loss": 0.943185031414032, - "eval_runtime": 46.3699, - "eval_samples_per_second": 36.705, - "eval_steps_per_second": 4.593, + "eval_loss": 0.9579859972000122, + "eval_runtime": 46.2028, + "eval_samples_per_second": 36.838, + "eval_steps_per_second": 4.61, "step": 4500 }, { "epoch": 2.3561822692257475, - "grad_norm": 6.087654113769531, - "learning_rate": 1.1140688593114071e-05, - "loss": 0.5452, + "grad_norm": 5.159942626953125, + "learning_rate": 1.1142888571114289e-05, + "loss": 0.5249, "step": 4510 }, { "epoch": 2.3614048831440138, - "grad_norm": 5.286149024963379, - "learning_rate": 1.1118688813111871e-05, - "loss": 0.4648, + "grad_norm": 3.7154197692871094, + "learning_rate": 1.112088879111209e-05, + "loss": 0.4912, "step": 4520 }, { "epoch": 2.3666274970622796, - "grad_norm": 5.043085098266602, - "learning_rate": 1.1096689033109669e-05, - "loss": 0.6365, + "grad_norm": 5.892611026763916, + "learning_rate": 1.109888901110989e-05, + "loss": 0.5103, "step": 4530 }, { "epoch": 2.371850110980546, - "grad_norm": 4.788007736206055, - "learning_rate": 1.1074689253107469e-05, - "loss": 0.4857, + "grad_norm": 6.050294399261475, + "learning_rate": 1.107688923110769e-05, + "loss": 0.538, "step": 4540 }, { "epoch": 2.3770727248988117, - "grad_norm": 5.524052619934082, - "learning_rate": 1.1052689473105269e-05, - "loss": 0.6689, + "grad_norm": 3.6718509197235107, + "learning_rate": 1.105488945110549e-05, + "loss": 0.4608, "step": 4550 }, { "epoch": 2.382295338817078, - "grad_norm": 4.40060567855835, - "learning_rate": 1.1030689693103069e-05, - "loss": 0.4604, + "grad_norm": 4.4985880851745605, + "learning_rate": 1.103288967110329e-05, + "loss": 0.5294, "step": 4560 }, { "epoch": 2.387517952735344, - "grad_norm": 4.678870677947998, - "learning_rate": 1.100868991310087e-05, - "loss": 0.5077, + "grad_norm": 6.062043190002441, + "learning_rate": 1.1010889891101091e-05, + "loss": 0.438, "step": 4570 }, { "epoch": 2.39274056665361, - "grad_norm": 5.824218273162842, - "learning_rate": 1.098669013309867e-05, - "loss": 0.5265, + "grad_norm": 3.617093801498413, + "learning_rate": 1.098889011109889e-05, + "loss": 0.4725, "step": 4580 }, { "epoch": 2.3979631805718764, - "grad_norm": 6.174131393432617, - "learning_rate": 1.096469035309647e-05, - "loss": 0.4931, + "grad_norm": 6.143939018249512, + "learning_rate": 1.096689033109669e-05, + "loss": 0.4644, "step": 4590 }, { "epoch": 2.4031857944901422, - "grad_norm": 5.839408874511719, - "learning_rate": 1.094269057309427e-05, - "loss": 0.53, + "grad_norm": 5.8897929191589355, + "learning_rate": 1.094489055109449e-05, + "loss": 0.5271, "step": 4600 }, { "epoch": 2.4084084084084085, - "grad_norm": 5.057196140289307, - "learning_rate": 1.0920690793092071e-05, - "loss": 0.5008, + "grad_norm": 3.4078855514526367, + "learning_rate": 1.0922890771092289e-05, + "loss": 0.4713, "step": 4610 }, { "epoch": 2.4136310223266744, - "grad_norm": 4.919235706329346, - "learning_rate": 1.0898691013089871e-05, - "loss": 0.5645, + "grad_norm": 4.900807857513428, + "learning_rate": 1.0900890991090089e-05, + "loss": 0.4928, "step": 4620 }, { "epoch": 2.4188536362449407, - "grad_norm": 4.900165557861328, - "learning_rate": 1.087669123308767e-05, - "loss": 0.5037, + "grad_norm": 4.195567607879639, + "learning_rate": 1.087889121108789e-05, + "loss": 0.4793, "step": 4630 }, { "epoch": 2.4240762501632065, - "grad_norm": 5.373619079589844, - "learning_rate": 1.085469145308547e-05, - "loss": 0.5159, + "grad_norm": 5.371895790100098, + "learning_rate": 1.085689143108569e-05, + "loss": 0.4348, "step": 4640 }, { "epoch": 2.4292988640814728, - "grad_norm": 4.071446418762207, - "learning_rate": 1.083269167308327e-05, - "loss": 0.4935, + "grad_norm": 7.091486930847168, + "learning_rate": 1.083489165108349e-05, + "loss": 0.5869, "step": 4650 }, { "epoch": 2.434521477999739, - "grad_norm": 4.382746696472168, - "learning_rate": 1.0810691893081072e-05, - "loss": 0.5618, + "grad_norm": 6.140565395355225, + "learning_rate": 1.081289187108129e-05, + "loss": 0.4909, "step": 4660 }, { "epoch": 2.439744091918005, - "grad_norm": 5.00177001953125, - "learning_rate": 1.0788692113078868e-05, - "loss": 0.5216, + "grad_norm": 3.8957531452178955, + "learning_rate": 1.079089209107909e-05, + "loss": 0.3953, "step": 4670 }, { "epoch": 2.444966705836271, - "grad_norm": 5.724264621734619, - "learning_rate": 1.076669233307667e-05, - "loss": 0.5516, + "grad_norm": 5.203509330749512, + "learning_rate": 1.0768892311076891e-05, + "loss": 0.5462, "step": 4680 }, { "epoch": 2.450189319754537, - "grad_norm": 5.690515518188477, - "learning_rate": 1.074469255307447e-05, - "loss": 0.5124, + "grad_norm": 4.643135070800781, + "learning_rate": 1.0746892531074691e-05, + "loss": 0.4987, "step": 4690 }, { "epoch": 2.4554119336728033, - "grad_norm": 5.372811794281006, - "learning_rate": 1.072269277307227e-05, - "loss": 0.514, + "grad_norm": 4.381008148193359, + "learning_rate": 1.0724892751072491e-05, + "loss": 0.5452, "step": 4700 }, { "epoch": 2.460634547591069, - "grad_norm": 3.5407040119171143, - "learning_rate": 1.070069299307007e-05, - "loss": 0.5131, + "grad_norm": 3.5113565921783447, + "learning_rate": 1.070289297107029e-05, + "loss": 0.5308, "step": 4710 }, { "epoch": 2.4658571615093354, - "grad_norm": 5.465915679931641, - "learning_rate": 1.067869321306787e-05, - "loss": 0.4962, + "grad_norm": 6.373734951019287, + "learning_rate": 1.0680893191068089e-05, + "loss": 0.4833, "step": 4720 }, { "epoch": 2.4710797754276017, - "grad_norm": 3.357609272003174, - "learning_rate": 1.065669343306567e-05, - "loss": 0.495, + "grad_norm": 6.036679744720459, + "learning_rate": 1.0658893411065889e-05, + "loss": 0.5638, "step": 4730 }, { "epoch": 2.4763023893458675, - "grad_norm": 6.003732204437256, - "learning_rate": 1.063469365306347e-05, - "loss": 0.47, + "grad_norm": 5.156855583190918, + "learning_rate": 1.063689363106369e-05, + "loss": 0.4734, "step": 4740 }, { "epoch": 2.481525003264134, - "grad_norm": 6.457789897918701, - "learning_rate": 1.061269387306127e-05, - "loss": 0.5406, + "grad_norm": 5.211863040924072, + "learning_rate": 1.061489385106149e-05, + "loss": 0.4596, "step": 4750 }, { "epoch": 2.4867476171823997, - "grad_norm": 4.927131652832031, - "learning_rate": 1.059069409305907e-05, - "loss": 0.5804, + "grad_norm": 3.838738203048706, + "learning_rate": 1.059289407105929e-05, + "loss": 0.4321, "step": 4760 }, { "epoch": 2.491970231100666, - "grad_norm": 4.805700302124023, - "learning_rate": 1.0568694313056872e-05, - "loss": 0.6091, + "grad_norm": 5.004904270172119, + "learning_rate": 1.057089429105709e-05, + "loss": 0.527, "step": 4770 }, { "epoch": 2.497192845018932, - "grad_norm": 5.979767799377441, - "learning_rate": 1.0546694533054671e-05, - "loss": 0.5219, + "grad_norm": 5.926597595214844, + "learning_rate": 1.054889451105489e-05, + "loss": 0.5087, "step": 4780 }, { "epoch": 2.502415458937198, - "grad_norm": 3.6271190643310547, - "learning_rate": 1.0524694753052471e-05, - "loss": 0.5208, + "grad_norm": 4.630094528198242, + "learning_rate": 1.0526894731052691e-05, + "loss": 0.4924, "step": 4790 }, { "epoch": 2.507638072855464, - "grad_norm": 4.7155442237854, - "learning_rate": 1.0502694973050271e-05, - "loss": 0.5011, + "grad_norm": 4.727828025817871, + "learning_rate": 1.050489495105049e-05, + "loss": 0.5099, "step": 4800 }, { "epoch": 2.51286068677373, - "grad_norm": 5.767155647277832, - "learning_rate": 1.048069519304807e-05, - "loss": 0.4557, + "grad_norm": 2.598799228668213, + "learning_rate": 1.048289517104829e-05, + "loss": 0.4881, "step": 4810 }, { "epoch": 2.5180833006919965, - "grad_norm": 3.412713050842285, - "learning_rate": 1.0458695413045869e-05, - "loss": 0.5196, + "grad_norm": 4.259073734283447, + "learning_rate": 1.046089539104609e-05, + "loss": 0.4783, "step": 4820 }, { "epoch": 2.5233059146102623, - "grad_norm": 6.184292793273926, - "learning_rate": 1.0436695633043669e-05, - "loss": 0.6146, + "grad_norm": 3.2685747146606445, + "learning_rate": 1.043889561104389e-05, + "loss": 0.4725, "step": 4830 }, { "epoch": 2.5285285285285286, - "grad_norm": 6.836652755737305, - "learning_rate": 1.041469585304147e-05, - "loss": 0.5338, + "grad_norm": 4.925142288208008, + "learning_rate": 1.0416895831041692e-05, + "loss": 0.5037, "step": 4840 }, { "epoch": 2.5337511424467944, - "grad_norm": 6.3216118812561035, - "learning_rate": 1.039269607303927e-05, - "loss": 0.5146, + "grad_norm": 5.959577560424805, + "learning_rate": 1.0394896051039492e-05, + "loss": 0.5677, "step": 4850 }, { "epoch": 2.5389737563650607, - "grad_norm": 5.718419075012207, - "learning_rate": 1.037069629303707e-05, - "loss": 0.5192, + "grad_norm": 4.998350620269775, + "learning_rate": 1.0372896271037291e-05, + "loss": 0.4071, "step": 4860 }, { "epoch": 2.544196370283327, - "grad_norm": 4.821221351623535, - "learning_rate": 1.034869651303487e-05, - "loss": 0.5044, + "grad_norm": 5.131379127502441, + "learning_rate": 1.035089649103509e-05, + "loss": 0.5444, "step": 4870 }, { "epoch": 2.549418984201593, - "grad_norm": 3.7728681564331055, - "learning_rate": 1.0326696733032671e-05, - "loss": 0.4557, + "grad_norm": 5.296921730041504, + "learning_rate": 1.032889671103289e-05, + "loss": 0.4554, "step": 4880 }, { "epoch": 2.554641598119859, - "grad_norm": 4.283219337463379, - "learning_rate": 1.0304696953030471e-05, - "loss": 0.6728, + "grad_norm": 5.137699127197266, + "learning_rate": 1.030689693103069e-05, + "loss": 0.5124, "step": 4890 }, { "epoch": 2.559864212038125, - "grad_norm": 3.8312339782714844, - "learning_rate": 1.0282697173028271e-05, - "loss": 0.4257, + "grad_norm": 5.180246829986572, + "learning_rate": 1.028489715102849e-05, + "loss": 0.5194, "step": 4900 }, { "epoch": 2.5650868259563913, - "grad_norm": 4.710253715515137, - "learning_rate": 1.0260697393026071e-05, - "loss": 0.4818, + "grad_norm": 5.066071510314941, + "learning_rate": 1.026289737102629e-05, + "loss": 0.5039, "step": 4910 }, { "epoch": 2.5703094398746575, - "grad_norm": 5.925652980804443, - "learning_rate": 1.023869761302387e-05, - "loss": 0.5167, + "grad_norm": 4.890101432800293, + "learning_rate": 1.024089759102409e-05, + "loss": 0.3757, "step": 4920 }, { "epoch": 2.5755320537929234, - "grad_norm": 4.76073694229126, - "learning_rate": 1.0216697833021672e-05, - "loss": 0.4363, + "grad_norm": 4.352961540222168, + "learning_rate": 1.021889781102189e-05, + "loss": 0.5546, "step": 4930 }, { "epoch": 2.580754667711189, - "grad_norm": 3.673877716064453, - "learning_rate": 1.0194698053019472e-05, - "loss": 0.5326, + "grad_norm": 4.833639144897461, + "learning_rate": 1.019689803101969e-05, + "loss": 0.4063, "step": 4940 }, { "epoch": 2.5859772816294555, - "grad_norm": 4.946410655975342, - "learning_rate": 1.017269827301727e-05, - "loss": 0.6238, + "grad_norm": 4.127015113830566, + "learning_rate": 1.0174898251017492e-05, + "loss": 0.5065, "step": 4950 }, { "epoch": 2.591199895547722, - "grad_norm": 4.7166290283203125, - "learning_rate": 1.015069849301507e-05, - "loss": 0.479, + "grad_norm": 4.817408561706543, + "learning_rate": 1.0152898471015291e-05, + "loss": 0.4395, "step": 4960 }, { "epoch": 2.5964225094659876, - "grad_norm": 4.268213748931885, - "learning_rate": 1.012869871301287e-05, - "loss": 0.5868, + "grad_norm": 3.6117539405822754, + "learning_rate": 1.0130898691013091e-05, + "loss": 0.5218, "step": 4970 }, { "epoch": 2.601645123384254, - "grad_norm": 5.25217866897583, - "learning_rate": 1.010669893301067e-05, - "loss": 0.5622, + "grad_norm": 3.3799283504486084, + "learning_rate": 1.0108898911010891e-05, + "loss": 0.4014, "step": 4980 }, { "epoch": 2.6068677373025197, - "grad_norm": 5.709757328033447, - "learning_rate": 1.0084699153008471e-05, - "loss": 0.4127, + "grad_norm": 5.896636962890625, + "learning_rate": 1.008689913100869e-05, + "loss": 0.5036, "step": 4990 }, { "epoch": 2.612090351220786, - "grad_norm": 5.0968732833862305, - "learning_rate": 1.0062699373006271e-05, - "loss": 0.5278, + "grad_norm": 5.3043670654296875, + "learning_rate": 1.0064899351006492e-05, + "loss": 0.4719, "step": 5000 }, { "epoch": 2.612090351220786, - "eval_loss": 0.9182845950126648, - "eval_runtime": 46.4198, - "eval_samples_per_second": 36.665, - "eval_steps_per_second": 4.589, + "eval_loss": 0.9424123764038086, + "eval_runtime": 46.1964, + "eval_samples_per_second": 36.843, + "eval_steps_per_second": 4.611, "step": 5000 }, { "epoch": 2.6173129651390523, - "grad_norm": 4.902764320373535, - "learning_rate": 1.004069959300407e-05, - "loss": 0.5596, + "grad_norm": 4.1417365074157715, + "learning_rate": 1.004289957100429e-05, + "loss": 0.5549, "step": 5010 }, { "epoch": 2.622535579057318, - "grad_norm": 5.16074800491333, - "learning_rate": 1.001869981300187e-05, - "loss": 0.496, + "grad_norm": 5.097470760345459, + "learning_rate": 1.002089979100209e-05, + "loss": 0.4905, "step": 5020 }, { "epoch": 2.6277581929755844, - "grad_norm": 5.447412490844727, - "learning_rate": 9.99670003299967e-06, - "loss": 0.5532, + "grad_norm": 4.4659318923950195, + "learning_rate": 9.998900010999892e-06, + "loss": 0.4778, "step": 5030 }, { "epoch": 2.6329808068938503, - "grad_norm": 2.698256492614746, - "learning_rate": 9.974700252997472e-06, - "loss": 0.5541, + "grad_norm": 4.431312561035156, + "learning_rate": 9.97690023099769e-06, + "loss": 0.5327, "step": 5040 }, { "epoch": 2.6382034208121166, - "grad_norm": 5.540690898895264, - "learning_rate": 9.95270047299527e-06, - "loss": 0.5367, + "grad_norm": 5.72623348236084, + "learning_rate": 9.95490045099549e-06, + "loss": 0.4634, "step": 5050 }, { "epoch": 2.6434260347303824, - "grad_norm": 3.423121929168701, - "learning_rate": 9.93070069299307e-06, - "loss": 0.5361, + "grad_norm": 6.397616863250732, + "learning_rate": 9.932900670993291e-06, + "loss": 0.5599, "step": 5060 }, { "epoch": 2.6486486486486487, - "grad_norm": 5.744271278381348, - "learning_rate": 9.908700912990871e-06, - "loss": 0.5378, + "grad_norm": 4.112700462341309, + "learning_rate": 9.910900890991091e-06, + "loss": 0.5478, "step": 5070 }, { "epoch": 2.6538712625669145, - "grad_norm": 5.6931843757629395, - "learning_rate": 9.886701132988671e-06, - "loss": 0.5504, + "grad_norm": 4.668074607849121, + "learning_rate": 9.888901110988891e-06, + "loss": 0.4968, "step": 5080 }, { "epoch": 2.659093876485181, - "grad_norm": 5.743236064910889, - "learning_rate": 9.864701352986471e-06, - "loss": 0.5685, + "grad_norm": 5.368128776550293, + "learning_rate": 9.86690133098669e-06, + "loss": 0.5542, "step": 5090 }, { "epoch": 2.664316490403447, - "grad_norm": 5.774685382843018, - "learning_rate": 9.842701572984271e-06, - "loss": 0.6709, + "grad_norm": 6.339516639709473, + "learning_rate": 9.84490155098449e-06, + "loss": 0.5513, "step": 5100 }, { "epoch": 2.669539104321713, - "grad_norm": 3.9477055072784424, - "learning_rate": 9.82070179298207e-06, - "loss": 0.4714, + "grad_norm": 4.424732685089111, + "learning_rate": 9.82290177098229e-06, + "loss": 0.4676, "step": 5110 }, { "epoch": 2.674761718239979, - "grad_norm": 4.734736442565918, - "learning_rate": 9.79870201297987e-06, - "loss": 0.4373, + "grad_norm": 3.3977339267730713, + "learning_rate": 9.80090199098009e-06, + "loss": 0.6237, "step": 5120 }, { "epoch": 2.679984332158245, - "grad_norm": 4.351528167724609, - "learning_rate": 9.77670223297767e-06, - "loss": 0.5032, + "grad_norm": 4.895110607147217, + "learning_rate": 9.77890221097789e-06, + "loss": 0.533, "step": 5130 }, { "epoch": 2.6852069460765113, - "grad_norm": 5.193398475646973, - "learning_rate": 9.75470245297547e-06, - "loss": 0.4897, + "grad_norm": 5.128396987915039, + "learning_rate": 9.756902430975692e-06, + "loss": 0.4814, "step": 5140 }, { "epoch": 2.6904295599947776, - "grad_norm": 3.739581823348999, - "learning_rate": 9.732702672973272e-06, - "loss": 0.4485, + "grad_norm": 4.816542148590088, + "learning_rate": 9.734902650973491e-06, + "loss": 0.432, "step": 5150 }, { "epoch": 2.6956521739130435, - "grad_norm": 4.9811787605285645, - "learning_rate": 9.710702892971072e-06, - "loss": 0.5092, + "grad_norm": 6.79871940612793, + "learning_rate": 9.712902870971291e-06, + "loss": 0.5283, "step": 5160 }, { "epoch": 2.7008747878313093, - "grad_norm": 5.68161678314209, - "learning_rate": 9.688703112968871e-06, - "loss": 0.5234, + "grad_norm": 4.02092170715332, + "learning_rate": 9.690903090969091e-06, + "loss": 0.5039, "step": 5170 }, { "epoch": 2.7060974017495756, - "grad_norm": 4.802379131317139, - "learning_rate": 9.666703332966671e-06, - "loss": 0.4929, + "grad_norm": 2.9405927658081055, + "learning_rate": 9.668903310966891e-06, + "loss": 0.5141, "step": 5180 }, { "epoch": 2.711320015667842, - "grad_norm": 4.281787872314453, - "learning_rate": 9.644703552964471e-06, - "loss": 0.5053, + "grad_norm": 5.391539573669434, + "learning_rate": 9.64690353096469e-06, + "loss": 0.5294, "step": 5190 }, { "epoch": 2.7165426295861077, - "grad_norm": 5.739396572113037, - "learning_rate": 9.622703772962271e-06, - "loss": 0.572, + "grad_norm": 5.314087867736816, + "learning_rate": 9.62490375096249e-06, + "loss": 0.5635, "step": 5200 }, { "epoch": 2.721765243504374, - "grad_norm": 4.055601596832275, - "learning_rate": 9.60070399296007e-06, - "loss": 0.5056, + "grad_norm": 4.858371734619141, + "learning_rate": 9.60290397096029e-06, + "loss": 0.4825, "step": 5210 }, { "epoch": 2.72698785742264, - "grad_norm": 6.050775527954102, - "learning_rate": 9.57870421295787e-06, - "loss": 0.4743, + "grad_norm": 6.224414348602295, + "learning_rate": 9.580904190958092e-06, + "loss": 0.5579, "step": 5220 }, { "epoch": 2.732210471340906, - "grad_norm": 2.8432188034057617, - "learning_rate": 9.556704432955672e-06, - "loss": 0.4505, + "grad_norm": 5.094522476196289, + "learning_rate": 9.558904410955892e-06, + "loss": 0.5563, "step": 5230 }, { "epoch": 2.7374330852591724, - "grad_norm": 5.975009918212891, - "learning_rate": 9.534704652953472e-06, - "loss": 0.5604, + "grad_norm": 4.434865474700928, + "learning_rate": 9.536904630953692e-06, + "loss": 0.5524, "step": 5240 }, { "epoch": 2.7426556991774382, - "grad_norm": 4.10011100769043, - "learning_rate": 9.512704872951272e-06, - "loss": 0.4997, + "grad_norm": 5.384510040283203, + "learning_rate": 9.514904850951491e-06, + "loss": 0.5541, "step": 5250 }, { "epoch": 2.7478783130957045, - "grad_norm": 3.3179147243499756, - "learning_rate": 9.490705092949071e-06, - "loss": 0.4576, + "grad_norm": 4.52743673324585, + "learning_rate": 9.492905070949291e-06, + "loss": 0.4929, "step": 5260 }, { "epoch": 2.7531009270139704, - "grad_norm": 5.42781925201416, - "learning_rate": 9.468705312946871e-06, - "loss": 0.5374, + "grad_norm": 4.675116062164307, + "learning_rate": 9.470905290947091e-06, + "loss": 0.5256, "step": 5270 }, { "epoch": 2.7583235409322366, - "grad_norm": 5.195150852203369, - "learning_rate": 9.446705532944671e-06, - "loss": 0.4939, + "grad_norm": 4.984778881072998, + "learning_rate": 9.44890551094489e-06, + "loss": 0.4393, "step": 5280 }, { "epoch": 2.763546154850503, - "grad_norm": 3.490081548690796, - "learning_rate": 9.424705752942471e-06, - "loss": 0.4875, + "grad_norm": 6.062976837158203, + "learning_rate": 9.42690573094269e-06, + "loss": 0.4756, "step": 5290 }, { "epoch": 2.7687687687687688, - "grad_norm": 2.548433303833008, - "learning_rate": 9.40270597294027e-06, - "loss": 0.5088, + "grad_norm": 4.108662128448486, + "learning_rate": 9.404905950940492e-06, + "loss": 0.5881, "step": 5300 }, { "epoch": 2.7739913826870346, - "grad_norm": 4.056173801422119, - "learning_rate": 9.380706192938072e-06, - "loss": 0.5062, + "grad_norm": 3.5533432960510254, + "learning_rate": 9.382906170938292e-06, + "loss": 0.4687, "step": 5310 }, { "epoch": 2.779213996605301, - "grad_norm": 3.9009056091308594, - "learning_rate": 9.358706412935872e-06, - "loss": 0.4859, + "grad_norm": 3.260845422744751, + "learning_rate": 9.36090639093609e-06, + "loss": 0.4397, "step": 5320 }, { "epoch": 2.784436610523567, - "grad_norm": 7.616847038269043, - "learning_rate": 9.336706632933672e-06, - "loss": 0.4976, + "grad_norm": 3.3019182682037354, + "learning_rate": 9.338906610933892e-06, + "loss": 0.5021, "step": 5330 }, { "epoch": 2.789659224441833, - "grad_norm": 3.8724076747894287, - "learning_rate": 9.314706852931472e-06, - "loss": 0.4965, + "grad_norm": 5.403345108032227, + "learning_rate": 9.316906830931691e-06, + "loss": 0.5703, "step": 5340 }, { "epoch": 2.7948818383600993, - "grad_norm": 6.395397663116455, - "learning_rate": 9.292707072929272e-06, - "loss": 0.5324, + "grad_norm": 5.3398003578186035, + "learning_rate": 9.294907050929491e-06, + "loss": 0.4649, "step": 5350 }, { "epoch": 2.800104452278365, - "grad_norm": 5.191283702850342, - "learning_rate": 9.270707292927071e-06, - "loss": 0.4321, + "grad_norm": 4.496288776397705, + "learning_rate": 9.272907270927291e-06, + "loss": 0.5249, "step": 5360 }, { "epoch": 2.8053270661966314, - "grad_norm": 5.025651454925537, - "learning_rate": 9.248707512924871e-06, - "loss": 0.5549, + "grad_norm": 5.002990245819092, + "learning_rate": 9.250907490925091e-06, + "loss": 0.4271, "step": 5370 }, { "epoch": 2.8105496801148977, - "grad_norm": 6.335343360900879, - "learning_rate": 9.226707732922671e-06, - "loss": 0.5857, + "grad_norm": 5.431394100189209, + "learning_rate": 9.228907710922892e-06, + "loss": 0.4901, "step": 5380 }, { "epoch": 2.8157722940331635, - "grad_norm": 4.999082565307617, - "learning_rate": 9.204707952920473e-06, - "loss": 0.4673, + "grad_norm": 3.670156955718994, + "learning_rate": 9.20690793092069e-06, + "loss": 0.4489, "step": 5390 }, { "epoch": 2.82099490795143, - "grad_norm": 5.395633697509766, - "learning_rate": 9.182708172918272e-06, - "loss": 0.5711, + "grad_norm": 4.886593818664551, + "learning_rate": 9.18490815091849e-06, + "loss": 0.5646, "step": 5400 }, { "epoch": 2.8262175218696957, - "grad_norm": 3.730288028717041, - "learning_rate": 9.16070839291607e-06, - "loss": 0.5602, + "grad_norm": 4.911600112915039, + "learning_rate": 9.162908370916292e-06, + "loss": 0.5012, "step": 5410 }, { "epoch": 2.831440135787962, - "grad_norm": 6.485812664031982, - "learning_rate": 9.138708612913872e-06, - "loss": 0.5606, + "grad_norm": 7.215076446533203, + "learning_rate": 9.140908590914092e-06, + "loss": 0.5025, "step": 5420 }, { "epoch": 2.8366627497062282, - "grad_norm": 6.046615123748779, - "learning_rate": 9.116708832911672e-06, - "loss": 0.5591, + "grad_norm": 5.0257391929626465, + "learning_rate": 9.118908810911892e-06, + "loss": 0.4007, "step": 5430 }, { "epoch": 2.841885363624494, - "grad_norm": 6.039712429046631, - "learning_rate": 9.094709052909472e-06, - "loss": 0.7427, + "grad_norm": 4.077221870422363, + "learning_rate": 9.096909030909691e-06, + "loss": 0.5079, "step": 5440 }, { "epoch": 2.84710797754276, - "grad_norm": 5.540979385375977, - "learning_rate": 9.072709272907272e-06, - "loss": 0.5996, + "grad_norm": 5.463025093078613, + "learning_rate": 9.074909250907493e-06, + "loss": 0.476, "step": 5450 }, { "epoch": 2.852330591461026, - "grad_norm": 4.887878894805908, - "learning_rate": 9.050709492905071e-06, - "loss": 0.5769, + "grad_norm": 7.295045375823975, + "learning_rate": 9.052909470905291e-06, + "loss": 0.485, "step": 5460 }, { "epoch": 2.8575532053792925, - "grad_norm": 3.2837166786193848, - "learning_rate": 9.028709712902873e-06, - "loss": 0.4668, + "grad_norm": 5.430480480194092, + "learning_rate": 9.030909690903091e-06, + "loss": 0.6185, "step": 5470 }, { "epoch": 2.8627758192975583, - "grad_norm": 6.3544440269470215, - "learning_rate": 9.006709932900671e-06, - "loss": 0.4784, + "grad_norm": 4.49359655380249, + "learning_rate": 9.00890991090089e-06, + "loss": 0.524, "step": 5480 }, { "epoch": 2.8679984332158246, - "grad_norm": 3.9592199325561523, - "learning_rate": 8.98471015289847e-06, - "loss": 0.521, + "grad_norm": 4.84905481338501, + "learning_rate": 8.986910130898692e-06, + "loss": 0.4687, "step": 5490 }, { "epoch": 2.8732210471340904, - "grad_norm": 3.39884614944458, - "learning_rate": 8.962710372896272e-06, - "loss": 0.5307, + "grad_norm": 6.2004780769348145, + "learning_rate": 8.964910350896492e-06, + "loss": 0.5525, "step": 5500 }, { "epoch": 2.8732210471340904, - "eval_loss": 0.9099580645561218, - "eval_runtime": 46.3657, - "eval_samples_per_second": 36.708, - "eval_steps_per_second": 4.594, + "eval_loss": 0.9267139434814453, + "eval_runtime": 46.3854, + "eval_samples_per_second": 36.693, + "eval_steps_per_second": 4.592, "step": 5500 } ],