diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4536 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 10.0, + "eval_steps": 500, + "global_step": 6420, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01557632398753894, + "grad_norm": 8.34332275390625, + "learning_rate": 3.1152647975077884e-06, + "loss": 0.933, + "step": 10 + }, + { + "epoch": 0.03115264797507788, + "grad_norm": 6.62861442565918, + "learning_rate": 6.230529595015577e-06, + "loss": 0.848, + "step": 20 + }, + { + "epoch": 0.04672897196261682, + "grad_norm": 3.3893256187438965, + "learning_rate": 9.345794392523365e-06, + "loss": 0.5369, + "step": 30 + }, + { + "epoch": 0.06230529595015576, + "grad_norm": 1.5483717918395996, + "learning_rate": 1.2461059190031153e-05, + "loss": 0.2689, + "step": 40 + }, + { + "epoch": 0.0778816199376947, + "grad_norm": 1.2724570035934448, + "learning_rate": 1.557632398753894e-05, + "loss": 0.1844, + "step": 50 + }, + { + "epoch": 0.09345794392523364, + "grad_norm": 1.3665391206741333, + "learning_rate": 1.869158878504673e-05, + "loss": 0.1427, + "step": 60 + }, + { + "epoch": 0.10903426791277258, + "grad_norm": 0.694703221321106, + "learning_rate": 2.1806853582554517e-05, + "loss": 0.1315, + "step": 70 + }, + { + "epoch": 0.12461059190031153, + "grad_norm": 1.038958191871643, + "learning_rate": 2.4922118380062307e-05, + "loss": 0.1092, + "step": 80 + }, + { + "epoch": 0.14018691588785046, + "grad_norm": 0.9026461839675903, + "learning_rate": 2.8037383177570094e-05, + "loss": 0.0979, + "step": 90 + }, + { + "epoch": 0.1557632398753894, + "grad_norm": 0.5277336239814758, + "learning_rate": 3.115264797507788e-05, + "loss": 0.0792, + "step": 100 + }, + { + "epoch": 0.17133956386292834, + "grad_norm": 0.7498865723609924, + "learning_rate": 3.426791277258567e-05, + "loss": 0.0664, + "step": 110 + }, + { + "epoch": 0.18691588785046728, + "grad_norm": 0.8186283707618713, + "learning_rate": 3.738317757009346e-05, + "loss": 0.0598, + "step": 120 + }, + { + "epoch": 0.20249221183800623, + "grad_norm": 0.6182209849357605, + "learning_rate": 4.049844236760125e-05, + "loss": 0.064, + "step": 130 + }, + { + "epoch": 0.21806853582554517, + "grad_norm": 0.8646546006202698, + "learning_rate": 4.3613707165109034e-05, + "loss": 0.0613, + "step": 140 + }, + { + "epoch": 0.2336448598130841, + "grad_norm": 0.535507082939148, + "learning_rate": 4.672897196261683e-05, + "loss": 0.0492, + "step": 150 + }, + { + "epoch": 0.24922118380062305, + "grad_norm": 0.6274747848510742, + "learning_rate": 4.9844236760124614e-05, + "loss": 0.0524, + "step": 160 + }, + { + "epoch": 0.26479750778816197, + "grad_norm": 0.36217236518859863, + "learning_rate": 5.2959501557632394e-05, + "loss": 0.0487, + "step": 170 + }, + { + "epoch": 0.2803738317757009, + "grad_norm": 0.5501105189323425, + "learning_rate": 5.607476635514019e-05, + "loss": 0.0442, + "step": 180 + }, + { + "epoch": 0.29595015576323985, + "grad_norm": 0.4763103723526001, + "learning_rate": 5.9190031152647974e-05, + "loss": 0.0483, + "step": 190 + }, + { + "epoch": 0.3115264797507788, + "grad_norm": 0.6724681854248047, + "learning_rate": 6.230529595015576e-05, + "loss": 0.0445, + "step": 200 + }, + { + "epoch": 0.32710280373831774, + "grad_norm": 0.6471392512321472, + "learning_rate": 6.542056074766355e-05, + "loss": 0.04, + "step": 210 + }, + { + "epoch": 0.3426791277258567, + "grad_norm": 0.40489619970321655, + "learning_rate": 6.853582554517133e-05, + "loss": 0.03, + "step": 220 + }, + { + "epoch": 0.3582554517133956, + "grad_norm": 0.5478422045707703, + "learning_rate": 7.165109034267913e-05, + "loss": 0.0386, + "step": 230 + }, + { + "epoch": 0.37383177570093457, + "grad_norm": 0.4060825705528259, + "learning_rate": 7.476635514018692e-05, + "loss": 0.0377, + "step": 240 + }, + { + "epoch": 0.3894080996884735, + "grad_norm": 0.30575844645500183, + "learning_rate": 7.788161993769471e-05, + "loss": 0.0355, + "step": 250 + }, + { + "epoch": 0.40498442367601245, + "grad_norm": 0.5236951112747192, + "learning_rate": 8.09968847352025e-05, + "loss": 0.0362, + "step": 260 + }, + { + "epoch": 0.4205607476635514, + "grad_norm": 0.45649516582489014, + "learning_rate": 8.411214953271028e-05, + "loss": 0.0351, + "step": 270 + }, + { + "epoch": 0.43613707165109034, + "grad_norm": 0.5160077214241028, + "learning_rate": 8.722741433021807e-05, + "loss": 0.0349, + "step": 280 + }, + { + "epoch": 0.4517133956386293, + "grad_norm": 0.36491018533706665, + "learning_rate": 9.034267912772585e-05, + "loss": 0.0274, + "step": 290 + }, + { + "epoch": 0.4672897196261682, + "grad_norm": 0.6136458516120911, + "learning_rate": 9.345794392523365e-05, + "loss": 0.0347, + "step": 300 + }, + { + "epoch": 0.48286604361370716, + "grad_norm": 0.283626526594162, + "learning_rate": 9.657320872274144e-05, + "loss": 0.0344, + "step": 310 + }, + { + "epoch": 0.4984423676012461, + "grad_norm": 0.3656242787837982, + "learning_rate": 9.968847352024923e-05, + "loss": 0.0331, + "step": 320 + }, + { + "epoch": 0.514018691588785, + "grad_norm": 0.33881592750549316, + "learning_rate": 9.99994627125104e-05, + "loss": 0.0312, + "step": 330 + }, + { + "epoch": 0.5295950155763239, + "grad_norm": 0.39000996947288513, + "learning_rate": 9.999760543724762e-05, + "loss": 0.0278, + "step": 340 + }, + { + "epoch": 0.5451713395638629, + "grad_norm": 0.3204681873321533, + "learning_rate": 9.999442159029957e-05, + "loss": 0.0276, + "step": 350 + }, + { + "epoch": 0.5607476635514018, + "grad_norm": 0.3364989757537842, + "learning_rate": 9.998991125614234e-05, + "loss": 0.0284, + "step": 360 + }, + { + "epoch": 0.5763239875389408, + "grad_norm": 0.3593066930770874, + "learning_rate": 9.998407455444729e-05, + "loss": 0.0272, + "step": 370 + }, + { + "epoch": 0.5919003115264797, + "grad_norm": 0.3878420293331146, + "learning_rate": 9.99769116400779e-05, + "loss": 0.027, + "step": 380 + }, + { + "epoch": 0.6074766355140186, + "grad_norm": 0.4487762153148651, + "learning_rate": 9.996842270308566e-05, + "loss": 0.0272, + "step": 390 + }, + { + "epoch": 0.6230529595015576, + "grad_norm": 0.5334095358848572, + "learning_rate": 9.9958607968705e-05, + "loss": 0.0283, + "step": 400 + }, + { + "epoch": 0.6386292834890965, + "grad_norm": 0.32835790514945984, + "learning_rate": 9.99474676973474e-05, + "loss": 0.0264, + "step": 410 + }, + { + "epoch": 0.6542056074766355, + "grad_norm": 0.3653208911418915, + "learning_rate": 9.993500218459437e-05, + "loss": 0.0241, + "step": 420 + }, + { + "epoch": 0.6697819314641744, + "grad_norm": 0.4939619302749634, + "learning_rate": 9.992121176118967e-05, + "loss": 0.0233, + "step": 430 + }, + { + "epoch": 0.6853582554517134, + "grad_norm": 0.39662104845046997, + "learning_rate": 9.990609679303056e-05, + "loss": 0.0272, + "step": 440 + }, + { + "epoch": 0.7009345794392523, + "grad_norm": 0.3198733925819397, + "learning_rate": 9.988965768115798e-05, + "loss": 0.0248, + "step": 450 + }, + { + "epoch": 0.7165109034267912, + "grad_norm": 0.4477022886276245, + "learning_rate": 9.987189486174606e-05, + "loss": 0.0307, + "step": 460 + }, + { + "epoch": 0.7320872274143302, + "grad_norm": 0.27620789408683777, + "learning_rate": 9.98528088060904e-05, + "loss": 0.0246, + "step": 470 + }, + { + "epoch": 0.7476635514018691, + "grad_norm": 0.2453584223985672, + "learning_rate": 9.983240002059571e-05, + "loss": 0.0257, + "step": 480 + }, + { + "epoch": 0.7632398753894081, + "grad_norm": 0.282216340303421, + "learning_rate": 9.981066904676224e-05, + "loss": 0.0206, + "step": 490 + }, + { + "epoch": 0.778816199376947, + "grad_norm": 0.20669862627983093, + "learning_rate": 9.97876164611715e-05, + "loss": 0.0208, + "step": 500 + }, + { + "epoch": 0.794392523364486, + "grad_norm": 0.3541461229324341, + "learning_rate": 9.976324287547092e-05, + "loss": 0.0211, + "step": 510 + }, + { + "epoch": 0.8099688473520249, + "grad_norm": 0.4043068289756775, + "learning_rate": 9.973754893635767e-05, + "loss": 0.02, + "step": 520 + }, + { + "epoch": 0.8255451713395638, + "grad_norm": 0.3331877589225769, + "learning_rate": 9.971053532556143e-05, + "loss": 0.0251, + "step": 530 + }, + { + "epoch": 0.8411214953271028, + "grad_norm": 0.4757959544658661, + "learning_rate": 9.968220275982632e-05, + "loss": 0.0216, + "step": 540 + }, + { + "epoch": 0.8566978193146417, + "grad_norm": 0.35191771388053894, + "learning_rate": 9.965255199089197e-05, + "loss": 0.0281, + "step": 550 + }, + { + "epoch": 0.8722741433021807, + "grad_norm": 0.30520376563072205, + "learning_rate": 9.962158380547346e-05, + "loss": 0.0258, + "step": 560 + }, + { + "epoch": 0.8878504672897196, + "grad_norm": 0.46462851762771606, + "learning_rate": 9.958929902524048e-05, + "loss": 0.0261, + "step": 570 + }, + { + "epoch": 0.9034267912772586, + "grad_norm": 0.26396533846855164, + "learning_rate": 9.955569850679558e-05, + "loss": 0.022, + "step": 580 + }, + { + "epoch": 0.9190031152647975, + "grad_norm": 0.2325705736875534, + "learning_rate": 9.952078314165139e-05, + "loss": 0.021, + "step": 590 + }, + { + "epoch": 0.9345794392523364, + "grad_norm": 0.45752421021461487, + "learning_rate": 9.948455385620694e-05, + "loss": 0.0189, + "step": 600 + }, + { + "epoch": 0.9501557632398754, + "grad_norm": 0.36059829592704773, + "learning_rate": 9.944701161172316e-05, + "loss": 0.0166, + "step": 610 + }, + { + "epoch": 0.9657320872274143, + "grad_norm": 0.41589999198913574, + "learning_rate": 9.940815740429734e-05, + "loss": 0.0274, + "step": 620 + }, + { + "epoch": 0.9813084112149533, + "grad_norm": 0.26806989312171936, + "learning_rate": 9.936799226483669e-05, + "loss": 0.0223, + "step": 630 + }, + { + "epoch": 0.9968847352024922, + "grad_norm": 0.7092414498329163, + "learning_rate": 9.932651725903095e-05, + "loss": 0.0231, + "step": 640 + }, + { + "epoch": 1.0124610591900312, + "grad_norm": 0.3604114055633545, + "learning_rate": 9.92837334873242e-05, + "loss": 0.0241, + "step": 650 + }, + { + "epoch": 1.02803738317757, + "grad_norm": 0.3966909646987915, + "learning_rate": 9.923964208488563e-05, + "loss": 0.0227, + "step": 660 + }, + { + "epoch": 1.043613707165109, + "grad_norm": 0.3709312975406647, + "learning_rate": 9.919424422157937e-05, + "loss": 0.0206, + "step": 670 + }, + { + "epoch": 1.0591900311526479, + "grad_norm": 0.2638329267501831, + "learning_rate": 9.914754110193349e-05, + "loss": 0.0216, + "step": 680 + }, + { + "epoch": 1.074766355140187, + "grad_norm": 0.37985852360725403, + "learning_rate": 9.909953396510807e-05, + "loss": 0.0214, + "step": 690 + }, + { + "epoch": 1.0903426791277258, + "grad_norm": 0.30944398045539856, + "learning_rate": 9.90502240848623e-05, + "loss": 0.0212, + "step": 700 + }, + { + "epoch": 1.1059190031152648, + "grad_norm": 0.1825455278158188, + "learning_rate": 9.899961276952064e-05, + "loss": 0.0186, + "step": 710 + }, + { + "epoch": 1.1214953271028036, + "grad_norm": 0.29276084899902344, + "learning_rate": 9.894770136193814e-05, + "loss": 0.0178, + "step": 720 + }, + { + "epoch": 1.1370716510903427, + "grad_norm": 0.23537325859069824, + "learning_rate": 9.889449123946486e-05, + "loss": 0.0189, + "step": 730 + }, + { + "epoch": 1.1526479750778815, + "grad_norm": 0.22725921869277954, + "learning_rate": 9.883998381390924e-05, + "loss": 0.0219, + "step": 740 + }, + { + "epoch": 1.1682242990654206, + "grad_norm": 0.3256939649581909, + "learning_rate": 9.878418053150069e-05, + "loss": 0.0219, + "step": 750 + }, + { + "epoch": 1.1838006230529594, + "grad_norm": 0.35934165120124817, + "learning_rate": 9.872708287285122e-05, + "loss": 0.0195, + "step": 760 + }, + { + "epoch": 1.1993769470404985, + "grad_norm": 0.21802407503128052, + "learning_rate": 9.86686923529161e-05, + "loss": 0.0216, + "step": 770 + }, + { + "epoch": 1.2149532710280373, + "grad_norm": 0.22510743141174316, + "learning_rate": 9.860901052095374e-05, + "loss": 0.02, + "step": 780 + }, + { + "epoch": 1.2305295950155763, + "grad_norm": 0.24169179797172546, + "learning_rate": 9.854803896048457e-05, + "loss": 0.0174, + "step": 790 + }, + { + "epoch": 1.2461059190031152, + "grad_norm": 0.19489426910877228, + "learning_rate": 9.848577928924895e-05, + "loss": 0.0185, + "step": 800 + }, + { + "epoch": 1.2616822429906542, + "grad_norm": 0.2777082324028015, + "learning_rate": 9.842223315916434e-05, + "loss": 0.0204, + "step": 810 + }, + { + "epoch": 1.277258566978193, + "grad_norm": 0.3644140362739563, + "learning_rate": 9.835740225628145e-05, + "loss": 0.0209, + "step": 820 + }, + { + "epoch": 1.2928348909657321, + "grad_norm": 0.1896691918373108, + "learning_rate": 9.829128830073944e-05, + "loss": 0.0166, + "step": 830 + }, + { + "epoch": 1.308411214953271, + "grad_norm": 0.20399421453475952, + "learning_rate": 9.822389304672034e-05, + "loss": 0.0205, + "step": 840 + }, + { + "epoch": 1.32398753894081, + "grad_norm": 0.23192964494228363, + "learning_rate": 9.815521828240251e-05, + "loss": 0.0215, + "step": 850 + }, + { + "epoch": 1.3395638629283488, + "grad_norm": 0.17685532569885254, + "learning_rate": 9.808526582991318e-05, + "loss": 0.0161, + "step": 860 + }, + { + "epoch": 1.355140186915888, + "grad_norm": 0.3119238615036011, + "learning_rate": 9.801403754528009e-05, + "loss": 0.0175, + "step": 870 + }, + { + "epoch": 1.3707165109034267, + "grad_norm": 0.29733213782310486, + "learning_rate": 9.794153531838226e-05, + "loss": 0.0186, + "step": 880 + }, + { + "epoch": 1.3862928348909658, + "grad_norm": 0.22840946912765503, + "learning_rate": 9.786776107289983e-05, + "loss": 0.019, + "step": 890 + }, + { + "epoch": 1.4018691588785046, + "grad_norm": 0.26108086109161377, + "learning_rate": 9.779271676626306e-05, + "loss": 0.0164, + "step": 900 + }, + { + "epoch": 1.4174454828660437, + "grad_norm": 0.30991408228874207, + "learning_rate": 9.771640438960036e-05, + "loss": 0.0146, + "step": 910 + }, + { + "epoch": 1.4330218068535825, + "grad_norm": 0.31594666838645935, + "learning_rate": 9.763882596768544e-05, + "loss": 0.0202, + "step": 920 + }, + { + "epoch": 1.4485981308411215, + "grad_norm": 0.16621960699558258, + "learning_rate": 9.755998355888365e-05, + "loss": 0.0172, + "step": 930 + }, + { + "epoch": 1.4641744548286604, + "grad_norm": 0.20799924433231354, + "learning_rate": 9.747987925509733e-05, + "loss": 0.0162, + "step": 940 + }, + { + "epoch": 1.4797507788161994, + "grad_norm": 0.30253657698631287, + "learning_rate": 9.739851518171026e-05, + "loss": 0.0203, + "step": 950 + }, + { + "epoch": 1.4953271028037383, + "grad_norm": 0.2623201906681061, + "learning_rate": 9.731589349753139e-05, + "loss": 0.0167, + "step": 960 + }, + { + "epoch": 1.5109034267912773, + "grad_norm": 0.19337186217308044, + "learning_rate": 9.723201639473738e-05, + "loss": 0.0191, + "step": 970 + }, + { + "epoch": 1.5264797507788161, + "grad_norm": 0.19426818192005157, + "learning_rate": 9.714688609881467e-05, + "loss": 0.0166, + "step": 980 + }, + { + "epoch": 1.542056074766355, + "grad_norm": 0.16336968541145325, + "learning_rate": 9.706050486850019e-05, + "loss": 0.021, + "step": 990 + }, + { + "epoch": 1.557632398753894, + "grad_norm": 0.30551159381866455, + "learning_rate": 9.697287499572161e-05, + "loss": 0.0213, + "step": 1000 + }, + { + "epoch": 1.573208722741433, + "grad_norm": 0.2476799190044403, + "learning_rate": 9.688399880553645e-05, + "loss": 0.0187, + "step": 1010 + }, + { + "epoch": 1.588785046728972, + "grad_norm": 0.2057810127735138, + "learning_rate": 9.679387865607035e-05, + "loss": 0.0195, + "step": 1020 + }, + { + "epoch": 1.6043613707165107, + "grad_norm": 0.2896953225135803, + "learning_rate": 9.670251693845464e-05, + "loss": 0.0212, + "step": 1030 + }, + { + "epoch": 1.6199376947040498, + "grad_norm": 0.2235778421163559, + "learning_rate": 9.660991607676276e-05, + "loss": 0.0196, + "step": 1040 + }, + { + "epoch": 1.6355140186915889, + "grad_norm": 0.24319937825202942, + "learning_rate": 9.651607852794597e-05, + "loss": 0.0164, + "step": 1050 + }, + { + "epoch": 1.6510903426791277, + "grad_norm": 0.3737725615501404, + "learning_rate": 9.642100678176824e-05, + "loss": 0.0197, + "step": 1060 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 0.2973679006099701, + "learning_rate": 9.632470336074009e-05, + "loss": 0.0179, + "step": 1070 + }, + { + "epoch": 1.6822429906542056, + "grad_norm": 0.23015610873699188, + "learning_rate": 9.622717082005175e-05, + "loss": 0.0197, + "step": 1080 + }, + { + "epoch": 1.6978193146417446, + "grad_norm": 0.3532279431819916, + "learning_rate": 9.612841174750524e-05, + "loss": 0.0176, + "step": 1090 + }, + { + "epoch": 1.7133956386292835, + "grad_norm": 0.2900121212005615, + "learning_rate": 9.602842876344589e-05, + "loss": 0.0198, + "step": 1100 + }, + { + "epoch": 1.7289719626168223, + "grad_norm": 0.3288438022136688, + "learning_rate": 9.592722452069264e-05, + "loss": 0.014, + "step": 1110 + }, + { + "epoch": 1.7445482866043613, + "grad_norm": 0.275817334651947, + "learning_rate": 9.582480170446774e-05, + "loss": 0.018, + "step": 1120 + }, + { + "epoch": 1.7601246105919004, + "grad_norm": 0.2955857515335083, + "learning_rate": 9.57211630323255e-05, + "loss": 0.0231, + "step": 1130 + }, + { + "epoch": 1.7757009345794392, + "grad_norm": 0.3142707049846649, + "learning_rate": 9.561631125408017e-05, + "loss": 0.0188, + "step": 1140 + }, + { + "epoch": 1.791277258566978, + "grad_norm": 0.2019612342119217, + "learning_rate": 9.551024915173299e-05, + "loss": 0.0177, + "step": 1150 + }, + { + "epoch": 1.8068535825545171, + "grad_norm": 0.3271605968475342, + "learning_rate": 9.540297953939836e-05, + "loss": 0.0178, + "step": 1160 + }, + { + "epoch": 1.8224299065420562, + "grad_norm": 0.2196618616580963, + "learning_rate": 9.529450526322922e-05, + "loss": 0.0172, + "step": 1170 + }, + { + "epoch": 1.838006230529595, + "grad_norm": 0.22195592522621155, + "learning_rate": 9.518482920134144e-05, + "loss": 0.013, + "step": 1180 + }, + { + "epoch": 1.8535825545171338, + "grad_norm": 0.40921470522880554, + "learning_rate": 9.507395426373756e-05, + "loss": 0.0166, + "step": 1190 + }, + { + "epoch": 1.8691588785046729, + "grad_norm": 0.30641061067581177, + "learning_rate": 9.496188339222954e-05, + "loss": 0.0153, + "step": 1200 + }, + { + "epoch": 1.884735202492212, + "grad_norm": 0.21686281263828278, + "learning_rate": 9.484861956036067e-05, + "loss": 0.0165, + "step": 1210 + }, + { + "epoch": 1.9003115264797508, + "grad_norm": 0.24141277372837067, + "learning_rate": 9.47341657733267e-05, + "loss": 0.0194, + "step": 1220 + }, + { + "epoch": 1.9158878504672896, + "grad_norm": 0.14065827429294586, + "learning_rate": 9.461852506789617e-05, + "loss": 0.0189, + "step": 1230 + }, + { + "epoch": 1.9314641744548287, + "grad_norm": 0.16774609684944153, + "learning_rate": 9.45017005123297e-05, + "loss": 0.0167, + "step": 1240 + }, + { + "epoch": 1.9470404984423677, + "grad_norm": 0.1050788089632988, + "learning_rate": 9.43836952062987e-05, + "loss": 0.0146, + "step": 1250 + }, + { + "epoch": 1.9626168224299065, + "grad_norm": 0.2471192628145218, + "learning_rate": 9.426451228080308e-05, + "loss": 0.0174, + "step": 1260 + }, + { + "epoch": 1.9781931464174454, + "grad_norm": 0.21543623507022858, + "learning_rate": 9.414415489808814e-05, + "loss": 0.0144, + "step": 1270 + }, + { + "epoch": 1.9937694704049844, + "grad_norm": 0.17441585659980774, + "learning_rate": 9.402262625156077e-05, + "loss": 0.0164, + "step": 1280 + }, + { + "epoch": 2.0093457943925235, + "grad_norm": 0.2228071242570877, + "learning_rate": 9.389992956570462e-05, + "loss": 0.0173, + "step": 1290 + }, + { + "epoch": 2.0249221183800623, + "grad_norm": 0.25380459427833557, + "learning_rate": 9.377606809599457e-05, + "loss": 0.0166, + "step": 1300 + }, + { + "epoch": 2.040498442367601, + "grad_norm": 0.2757023870944977, + "learning_rate": 9.365104512881039e-05, + "loss": 0.014, + "step": 1310 + }, + { + "epoch": 2.05607476635514, + "grad_norm": 0.19488276541233063, + "learning_rate": 9.35248639813495e-05, + "loss": 0.0148, + "step": 1320 + }, + { + "epoch": 2.0716510903426792, + "grad_norm": 0.179704949259758, + "learning_rate": 9.3397528001539e-05, + "loss": 0.0187, + "step": 1330 + }, + { + "epoch": 2.087227414330218, + "grad_norm": 0.19472235441207886, + "learning_rate": 9.326904056794679e-05, + "loss": 0.0144, + "step": 1340 + }, + { + "epoch": 2.102803738317757, + "grad_norm": 0.39669862389564514, + "learning_rate": 9.313940508969195e-05, + "loss": 0.0168, + "step": 1350 + }, + { + "epoch": 2.1183800623052957, + "grad_norm": 0.2911001741886139, + "learning_rate": 9.300862500635434e-05, + "loss": 0.0173, + "step": 1360 + }, + { + "epoch": 2.133956386292835, + "grad_norm": 0.18247844278812408, + "learning_rate": 9.287670378788322e-05, + "loss": 0.0144, + "step": 1370 + }, + { + "epoch": 2.149532710280374, + "grad_norm": 0.255506694316864, + "learning_rate": 9.27436449345053e-05, + "loss": 0.0151, + "step": 1380 + }, + { + "epoch": 2.1651090342679127, + "grad_norm": 0.32190999388694763, + "learning_rate": 9.26094519766318e-05, + "loss": 0.0177, + "step": 1390 + }, + { + "epoch": 2.1806853582554515, + "grad_norm": 0.3123270571231842, + "learning_rate": 9.247412847476477e-05, + "loss": 0.018, + "step": 1400 + }, + { + "epoch": 2.196261682242991, + "grad_norm": 0.2271157205104828, + "learning_rate": 9.233767801940274e-05, + "loss": 0.015, + "step": 1410 + }, + { + "epoch": 2.2118380062305296, + "grad_norm": 0.2909000515937805, + "learning_rate": 9.220010423094528e-05, + "loss": 0.0131, + "step": 1420 + }, + { + "epoch": 2.2274143302180685, + "grad_norm": 0.2275337427854538, + "learning_rate": 9.206141075959702e-05, + "loss": 0.0193, + "step": 1430 + }, + { + "epoch": 2.2429906542056073, + "grad_norm": 0.19818776845932007, + "learning_rate": 9.192160128527088e-05, + "loss": 0.0137, + "step": 1440 + }, + { + "epoch": 2.2585669781931466, + "grad_norm": 0.1689874231815338, + "learning_rate": 9.17806795174903e-05, + "loss": 0.0127, + "step": 1450 + }, + { + "epoch": 2.2741433021806854, + "grad_norm": 0.31005653738975525, + "learning_rate": 9.163864919529089e-05, + "loss": 0.0166, + "step": 1460 + }, + { + "epoch": 2.289719626168224, + "grad_norm": 0.20109283924102783, + "learning_rate": 9.149551408712117e-05, + "loss": 0.0141, + "step": 1470 + }, + { + "epoch": 2.305295950155763, + "grad_norm": 0.2736987769603729, + "learning_rate": 9.13512779907427e-05, + "loss": 0.0159, + "step": 1480 + }, + { + "epoch": 2.3208722741433023, + "grad_norm": 0.21294985711574554, + "learning_rate": 9.120594473312916e-05, + "loss": 0.0178, + "step": 1490 + }, + { + "epoch": 2.336448598130841, + "grad_norm": 0.2750362753868103, + "learning_rate": 9.105951817036492e-05, + "loss": 0.0129, + "step": 1500 + }, + { + "epoch": 2.35202492211838, + "grad_norm": 0.25375640392303467, + "learning_rate": 9.091200218754268e-05, + "loss": 0.0173, + "step": 1510 + }, + { + "epoch": 2.367601246105919, + "grad_norm": 0.2495325803756714, + "learning_rate": 9.076340069866042e-05, + "loss": 0.0142, + "step": 1520 + }, + { + "epoch": 2.383177570093458, + "grad_norm": 0.2479378879070282, + "learning_rate": 9.06137176465175e-05, + "loss": 0.0136, + "step": 1530 + }, + { + "epoch": 2.398753894080997, + "grad_norm": 0.2946663200855255, + "learning_rate": 9.04629570026101e-05, + "loss": 0.0144, + "step": 1540 + }, + { + "epoch": 2.4143302180685358, + "grad_norm": 0.2197083681821823, + "learning_rate": 9.03111227670258e-05, + "loss": 0.0132, + "step": 1550 + }, + { + "epoch": 2.4299065420560746, + "grad_norm": 0.16844919323921204, + "learning_rate": 9.015821896833752e-05, + "loss": 0.0165, + "step": 1560 + }, + { + "epoch": 2.445482866043614, + "grad_norm": 0.1574881672859192, + "learning_rate": 9.000424966349654e-05, + "loss": 0.0165, + "step": 1570 + }, + { + "epoch": 2.4610591900311527, + "grad_norm": 0.15327616035938263, + "learning_rate": 8.984921893772491e-05, + "loss": 0.0154, + "step": 1580 + }, + { + "epoch": 2.4766355140186915, + "grad_norm": 0.14823028445243835, + "learning_rate": 8.969313090440703e-05, + "loss": 0.015, + "step": 1590 + }, + { + "epoch": 2.4922118380062304, + "grad_norm": 0.2471739500761032, + "learning_rate": 8.953598970498058e-05, + "loss": 0.0147, + "step": 1600 + }, + { + "epoch": 2.507788161993769, + "grad_norm": 0.11376072466373444, + "learning_rate": 8.937779950882654e-05, + "loss": 0.0144, + "step": 1610 + }, + { + "epoch": 2.5233644859813085, + "grad_norm": 0.3323586583137512, + "learning_rate": 8.921856451315862e-05, + "loss": 0.014, + "step": 1620 + }, + { + "epoch": 2.5389408099688473, + "grad_norm": 0.234357088804245, + "learning_rate": 8.905828894291186e-05, + "loss": 0.0145, + "step": 1630 + }, + { + "epoch": 2.554517133956386, + "grad_norm": 0.1384000927209854, + "learning_rate": 8.889697705063063e-05, + "loss": 0.0161, + "step": 1640 + }, + { + "epoch": 2.5700934579439254, + "grad_norm": 0.2684932351112366, + "learning_rate": 8.873463311635565e-05, + "loss": 0.0164, + "step": 1650 + }, + { + "epoch": 2.5856697819314642, + "grad_norm": 0.2894107401371002, + "learning_rate": 8.857126144751056e-05, + "loss": 0.0136, + "step": 1660 + }, + { + "epoch": 2.601246105919003, + "grad_norm": 0.22467093169689178, + "learning_rate": 8.840686637878755e-05, + "loss": 0.0188, + "step": 1670 + }, + { + "epoch": 2.616822429906542, + "grad_norm": 0.27161693572998047, + "learning_rate": 8.824145227203238e-05, + "loss": 0.0151, + "step": 1680 + }, + { + "epoch": 2.6323987538940807, + "grad_norm": 0.18942400813102722, + "learning_rate": 8.807502351612865e-05, + "loss": 0.0161, + "step": 1690 + }, + { + "epoch": 2.64797507788162, + "grad_norm": 0.23328815400600433, + "learning_rate": 8.790758452688132e-05, + "loss": 0.0147, + "step": 1700 + }, + { + "epoch": 2.663551401869159, + "grad_norm": 0.20751376450061798, + "learning_rate": 8.773913974689962e-05, + "loss": 0.0145, + "step": 1710 + }, + { + "epoch": 2.6791277258566977, + "grad_norm": 0.19469401240348816, + "learning_rate": 8.75696936454791e-05, + "loss": 0.0156, + "step": 1720 + }, + { + "epoch": 2.694704049844237, + "grad_norm": 0.20457860827445984, + "learning_rate": 8.739925071848308e-05, + "loss": 0.0136, + "step": 1730 + }, + { + "epoch": 2.710280373831776, + "grad_norm": 0.24321670830249786, + "learning_rate": 8.722781548822332e-05, + "loss": 0.018, + "step": 1740 + }, + { + "epoch": 2.7258566978193146, + "grad_norm": 0.20383647084236145, + "learning_rate": 8.705539250334016e-05, + "loss": 0.0147, + "step": 1750 + }, + { + "epoch": 2.7414330218068534, + "grad_norm": 0.22205962240695953, + "learning_rate": 8.688198633868168e-05, + "loss": 0.0143, + "step": 1760 + }, + { + "epoch": 2.7570093457943923, + "grad_norm": 0.1648181974887848, + "learning_rate": 8.670760159518234e-05, + "loss": 0.0213, + "step": 1770 + }, + { + "epoch": 2.7725856697819315, + "grad_norm": 0.2582213282585144, + "learning_rate": 8.653224289974104e-05, + "loss": 0.0128, + "step": 1780 + }, + { + "epoch": 2.7881619937694704, + "grad_norm": 0.2634594142436981, + "learning_rate": 8.63559149050982e-05, + "loss": 0.0158, + "step": 1790 + }, + { + "epoch": 2.803738317757009, + "grad_norm": 0.19994807243347168, + "learning_rate": 8.617862228971234e-05, + "loss": 0.0135, + "step": 1800 + }, + { + "epoch": 2.8193146417445485, + "grad_norm": 0.22563837468624115, + "learning_rate": 8.600036975763605e-05, + "loss": 0.0136, + "step": 1810 + }, + { + "epoch": 2.8348909657320873, + "grad_norm": 0.13740035891532898, + "learning_rate": 8.582116203839108e-05, + "loss": 0.0159, + "step": 1820 + }, + { + "epoch": 2.850467289719626, + "grad_norm": 0.21851687133312225, + "learning_rate": 8.564100388684284e-05, + "loss": 0.0156, + "step": 1830 + }, + { + "epoch": 2.866043613707165, + "grad_norm": 0.21716952323913574, + "learning_rate": 8.545990008307431e-05, + "loss": 0.0103, + "step": 1840 + }, + { + "epoch": 2.881619937694704, + "grad_norm": 0.20476889610290527, + "learning_rate": 8.527785543225921e-05, + "loss": 0.016, + "step": 1850 + }, + { + "epoch": 2.897196261682243, + "grad_norm": 0.23600561916828156, + "learning_rate": 8.509487476453442e-05, + "loss": 0.0154, + "step": 1860 + }, + { + "epoch": 2.912772585669782, + "grad_norm": 0.1938115656375885, + "learning_rate": 8.49109629348719e-05, + "loss": 0.012, + "step": 1870 + }, + { + "epoch": 2.9283489096573208, + "grad_norm": 0.2717917263507843, + "learning_rate": 8.472612482294987e-05, + "loss": 0.0125, + "step": 1880 + }, + { + "epoch": 2.94392523364486, + "grad_norm": 0.18718834221363068, + "learning_rate": 8.454036533302331e-05, + "loss": 0.0146, + "step": 1890 + }, + { + "epoch": 2.959501557632399, + "grad_norm": 0.13792039453983307, + "learning_rate": 8.435368939379388e-05, + "loss": 0.0157, + "step": 1900 + }, + { + "epoch": 2.9750778816199377, + "grad_norm": 0.19632574915885925, + "learning_rate": 8.416610195827905e-05, + "loss": 0.0152, + "step": 1910 + }, + { + "epoch": 2.9906542056074765, + "grad_norm": 0.28728196024894714, + "learning_rate": 8.397760800368082e-05, + "loss": 0.0123, + "step": 1920 + }, + { + "epoch": 3.0062305295950154, + "grad_norm": 0.3165406584739685, + "learning_rate": 8.378821253125356e-05, + "loss": 0.0151, + "step": 1930 + }, + { + "epoch": 3.0218068535825546, + "grad_norm": 0.19019992649555206, + "learning_rate": 8.359792056617135e-05, + "loss": 0.0162, + "step": 1940 + }, + { + "epoch": 3.0373831775700935, + "grad_norm": 0.21490029990673065, + "learning_rate": 8.340673715739466e-05, + "loss": 0.0148, + "step": 1950 + }, + { + "epoch": 3.0529595015576323, + "grad_norm": 0.16761478781700134, + "learning_rate": 8.321466737753634e-05, + "loss": 0.0152, + "step": 1960 + }, + { + "epoch": 3.068535825545171, + "grad_norm": 0.200679212808609, + "learning_rate": 8.302171632272709e-05, + "loss": 0.0132, + "step": 1970 + }, + { + "epoch": 3.0841121495327104, + "grad_norm": 0.15712881088256836, + "learning_rate": 8.282788911248021e-05, + "loss": 0.0152, + "step": 1980 + }, + { + "epoch": 3.0996884735202492, + "grad_norm": 0.19571930170059204, + "learning_rate": 8.263319088955578e-05, + "loss": 0.0138, + "step": 1990 + }, + { + "epoch": 3.115264797507788, + "grad_norm": 0.24790585041046143, + "learning_rate": 8.243762681982419e-05, + "loss": 0.0154, + "step": 2000 + }, + { + "epoch": 3.130841121495327, + "grad_norm": 0.1957317292690277, + "learning_rate": 8.224120209212907e-05, + "loss": 0.0122, + "step": 2010 + }, + { + "epoch": 3.146417445482866, + "grad_norm": 0.24267509579658508, + "learning_rate": 8.204392191814968e-05, + "loss": 0.0139, + "step": 2020 + }, + { + "epoch": 3.161993769470405, + "grad_norm": 0.23138225078582764, + "learning_rate": 8.184579153226254e-05, + "loss": 0.0136, + "step": 2030 + }, + { + "epoch": 3.177570093457944, + "grad_norm": 0.28118881583213806, + "learning_rate": 8.164681619140264e-05, + "loss": 0.0166, + "step": 2040 + }, + { + "epoch": 3.1931464174454827, + "grad_norm": 0.2736000716686249, + "learning_rate": 8.144700117492385e-05, + "loss": 0.0161, + "step": 2050 + }, + { + "epoch": 3.208722741433022, + "grad_norm": 0.21003635227680206, + "learning_rate": 8.124635178445895e-05, + "loss": 0.0142, + "step": 2060 + }, + { + "epoch": 3.2242990654205608, + "grad_norm": 0.2584760785102844, + "learning_rate": 8.104487334377889e-05, + "loss": 0.0154, + "step": 2070 + }, + { + "epoch": 3.2398753894080996, + "grad_norm": 0.2691844403743744, + "learning_rate": 8.084257119865159e-05, + "loss": 0.0138, + "step": 2080 + }, + { + "epoch": 3.2554517133956384, + "grad_norm": 0.19839587807655334, + "learning_rate": 8.063945071670006e-05, + "loss": 0.0141, + "step": 2090 + }, + { + "epoch": 3.2710280373831777, + "grad_norm": 0.28742802143096924, + "learning_rate": 8.043551728726e-05, + "loss": 0.017, + "step": 2100 + }, + { + "epoch": 3.2866043613707165, + "grad_norm": 0.17262744903564453, + "learning_rate": 8.02307763212368e-05, + "loss": 0.011, + "step": 2110 + }, + { + "epoch": 3.3021806853582554, + "grad_norm": 0.17710132896900177, + "learning_rate": 8.002523325096201e-05, + "loss": 0.0106, + "step": 2120 + }, + { + "epoch": 3.317757009345794, + "grad_norm": 0.2129012644290924, + "learning_rate": 7.981889353004915e-05, + "loss": 0.0147, + "step": 2130 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 0.14716210961341858, + "learning_rate": 7.961176263324901e-05, + "loss": 0.0112, + "step": 2140 + }, + { + "epoch": 3.3489096573208723, + "grad_norm": 0.2212764173746109, + "learning_rate": 7.940384605630447e-05, + "loss": 0.0122, + "step": 2150 + }, + { + "epoch": 3.364485981308411, + "grad_norm": 0.17234817147254944, + "learning_rate": 7.919514931580461e-05, + "loss": 0.0112, + "step": 2160 + }, + { + "epoch": 3.38006230529595, + "grad_norm": 0.1748262494802475, + "learning_rate": 7.898567794903833e-05, + "loss": 0.013, + "step": 2170 + }, + { + "epoch": 3.3956386292834893, + "grad_norm": 0.19324713945388794, + "learning_rate": 7.877543751384749e-05, + "loss": 0.0147, + "step": 2180 + }, + { + "epoch": 3.411214953271028, + "grad_norm": 0.19106893241405487, + "learning_rate": 7.856443358847943e-05, + "loss": 0.0169, + "step": 2190 + }, + { + "epoch": 3.426791277258567, + "grad_norm": 0.20462898910045624, + "learning_rate": 7.83526717714389e-05, + "loss": 0.0126, + "step": 2200 + }, + { + "epoch": 3.4423676012461057, + "grad_norm": 0.26991477608680725, + "learning_rate": 7.81401576813396e-05, + "loss": 0.0154, + "step": 2210 + }, + { + "epoch": 3.457943925233645, + "grad_norm": 0.20646274089813232, + "learning_rate": 7.792689695675506e-05, + "loss": 0.0145, + "step": 2220 + }, + { + "epoch": 3.473520249221184, + "grad_norm": 0.1414683610200882, + "learning_rate": 7.771289525606904e-05, + "loss": 0.0141, + "step": 2230 + }, + { + "epoch": 3.4890965732087227, + "grad_norm": 0.268904447555542, + "learning_rate": 7.749815825732543e-05, + "loss": 0.0113, + "step": 2240 + }, + { + "epoch": 3.5046728971962615, + "grad_norm": 0.09890452027320862, + "learning_rate": 7.72826916580775e-05, + "loss": 0.0115, + "step": 2250 + }, + { + "epoch": 3.520249221183801, + "grad_norm": 0.22059427201747894, + "learning_rate": 7.706650117523686e-05, + "loss": 0.0138, + "step": 2260 + }, + { + "epoch": 3.5358255451713396, + "grad_norm": 0.15525758266448975, + "learning_rate": 7.684959254492167e-05, + "loss": 0.0138, + "step": 2270 + }, + { + "epoch": 3.5514018691588785, + "grad_norm": 0.17785225808620453, + "learning_rate": 7.66319715223045e-05, + "loss": 0.0117, + "step": 2280 + }, + { + "epoch": 3.5669781931464173, + "grad_norm": 0.2259814292192459, + "learning_rate": 7.641364388145963e-05, + "loss": 0.0159, + "step": 2290 + }, + { + "epoch": 3.582554517133956, + "grad_norm": 0.289288192987442, + "learning_rate": 7.619461541520978e-05, + "loss": 0.0154, + "step": 2300 + }, + { + "epoch": 3.5981308411214954, + "grad_norm": 0.1844770759344101, + "learning_rate": 7.597489193497254e-05, + "loss": 0.0123, + "step": 2310 + }, + { + "epoch": 3.6137071651090342, + "grad_norm": 0.275309681892395, + "learning_rate": 7.575447927060601e-05, + "loss": 0.014, + "step": 2320 + }, + { + "epoch": 3.629283489096573, + "grad_norm": 0.3368551433086395, + "learning_rate": 7.55333832702543e-05, + "loss": 0.0128, + "step": 2330 + }, + { + "epoch": 3.6448598130841123, + "grad_norm": 0.14441683888435364, + "learning_rate": 7.53116098001922e-05, + "loss": 0.0124, + "step": 2340 + }, + { + "epoch": 3.660436137071651, + "grad_norm": 0.16262127459049225, + "learning_rate": 7.508916474466966e-05, + "loss": 0.0111, + "step": 2350 + }, + { + "epoch": 3.67601246105919, + "grad_norm": 0.28471696376800537, + "learning_rate": 7.486605400575555e-05, + "loss": 0.0155, + "step": 2360 + }, + { + "epoch": 3.691588785046729, + "grad_norm": 0.16003911197185516, + "learning_rate": 7.464228350318118e-05, + "loss": 0.0137, + "step": 2370 + }, + { + "epoch": 3.7071651090342677, + "grad_norm": 0.13349305093288422, + "learning_rate": 7.441785917418314e-05, + "loss": 0.0122, + "step": 2380 + }, + { + "epoch": 3.722741433021807, + "grad_norm": 0.2013760805130005, + "learning_rate": 7.41927869733458e-05, + "loss": 0.0129, + "step": 2390 + }, + { + "epoch": 3.7383177570093458, + "grad_norm": 0.14687639474868774, + "learning_rate": 7.396707287244331e-05, + "loss": 0.0108, + "step": 2400 + }, + { + "epoch": 3.7538940809968846, + "grad_norm": 0.20814476907253265, + "learning_rate": 7.374072286028124e-05, + "loss": 0.0143, + "step": 2410 + }, + { + "epoch": 3.769470404984424, + "grad_norm": 0.11024034023284912, + "learning_rate": 7.351374294253751e-05, + "loss": 0.0115, + "step": 2420 + }, + { + "epoch": 3.7850467289719627, + "grad_norm": 0.25545600056648254, + "learning_rate": 7.328613914160318e-05, + "loss": 0.0172, + "step": 2430 + }, + { + "epoch": 3.8006230529595015, + "grad_norm": 0.17019154131412506, + "learning_rate": 7.305791749642263e-05, + "loss": 0.0113, + "step": 2440 + }, + { + "epoch": 3.8161993769470404, + "grad_norm": 0.17234203219413757, + "learning_rate": 7.282908406233332e-05, + "loss": 0.0165, + "step": 2450 + }, + { + "epoch": 3.831775700934579, + "grad_norm": 0.13886360824108124, + "learning_rate": 7.25996449109051e-05, + "loss": 0.0185, + "step": 2460 + }, + { + "epoch": 3.8473520249221185, + "grad_norm": 0.1880546510219574, + "learning_rate": 7.236960612977914e-05, + "loss": 0.0135, + "step": 2470 + }, + { + "epoch": 3.8629283489096573, + "grad_norm": 0.33895474672317505, + "learning_rate": 7.213897382250643e-05, + "loss": 0.0135, + "step": 2480 + }, + { + "epoch": 3.878504672897196, + "grad_norm": 0.1464867889881134, + "learning_rate": 7.190775410838578e-05, + "loss": 0.0097, + "step": 2490 + }, + { + "epoch": 3.8940809968847354, + "grad_norm": 0.25255146622657776, + "learning_rate": 7.167595312230155e-05, + "loss": 0.017, + "step": 2500 + }, + { + "epoch": 3.9096573208722742, + "grad_norm": 0.29201412200927734, + "learning_rate": 7.144357701456075e-05, + "loss": 0.014, + "step": 2510 + }, + { + "epoch": 3.925233644859813, + "grad_norm": 0.13257095217704773, + "learning_rate": 7.121063195072999e-05, + "loss": 0.0101, + "step": 2520 + }, + { + "epoch": 3.940809968847352, + "grad_norm": 0.15739849209785461, + "learning_rate": 7.097712411147173e-05, + "loss": 0.014, + "step": 2530 + }, + { + "epoch": 3.9563862928348907, + "grad_norm": 0.2477468103170395, + "learning_rate": 7.074305969238048e-05, + "loss": 0.01, + "step": 2540 + }, + { + "epoch": 3.97196261682243, + "grad_norm": 0.2315158247947693, + "learning_rate": 7.050844490381829e-05, + "loss": 0.014, + "step": 2550 + }, + { + "epoch": 3.987538940809969, + "grad_norm": 0.11539182811975479, + "learning_rate": 7.027328597074998e-05, + "loss": 0.0119, + "step": 2560 + }, + { + "epoch": 4.003115264797508, + "grad_norm": 0.21218988299369812, + "learning_rate": 7.003758913257799e-05, + "loss": 0.0157, + "step": 2570 + }, + { + "epoch": 4.018691588785047, + "grad_norm": 0.1664864420890808, + "learning_rate": 6.980136064297686e-05, + "loss": 0.0144, + "step": 2580 + }, + { + "epoch": 4.034267912772585, + "grad_norm": 0.13880150020122528, + "learning_rate": 6.956460676972726e-05, + "loss": 0.0115, + "step": 2590 + }, + { + "epoch": 4.049844236760125, + "grad_norm": 0.17875036597251892, + "learning_rate": 6.932733379454978e-05, + "loss": 0.0128, + "step": 2600 + }, + { + "epoch": 4.065420560747664, + "grad_norm": 0.19681435823440552, + "learning_rate": 6.908954801293806e-05, + "loss": 0.0105, + "step": 2610 + }, + { + "epoch": 4.080996884735202, + "grad_norm": 0.1514422744512558, + "learning_rate": 6.8851255733992e-05, + "loss": 0.0106, + "step": 2620 + }, + { + "epoch": 4.0965732087227416, + "grad_norm": 0.18962527811527252, + "learning_rate": 6.861246328025017e-05, + "loss": 0.0131, + "step": 2630 + }, + { + "epoch": 4.11214953271028, + "grad_norm": 0.23904556035995483, + "learning_rate": 6.837317698752214e-05, + "loss": 0.01, + "step": 2640 + }, + { + "epoch": 4.127725856697819, + "grad_norm": 0.30831441283226013, + "learning_rate": 6.813340320472042e-05, + "loss": 0.0138, + "step": 2650 + }, + { + "epoch": 4.1433021806853585, + "grad_norm": 0.16034604609012604, + "learning_rate": 6.789314829369188e-05, + "loss": 0.0105, + "step": 2660 + }, + { + "epoch": 4.158878504672897, + "grad_norm": 0.16733208298683167, + "learning_rate": 6.765241862904909e-05, + "loss": 0.0105, + "step": 2670 + }, + { + "epoch": 4.174454828660436, + "grad_norm": 0.2249719798564911, + "learning_rate": 6.741122059800103e-05, + "loss": 0.0141, + "step": 2680 + }, + { + "epoch": 4.190031152647975, + "grad_norm": 0.1738113909959793, + "learning_rate": 6.716956060018377e-05, + "loss": 0.0106, + "step": 2690 + }, + { + "epoch": 4.205607476635514, + "grad_norm": 0.19900630414485931, + "learning_rate": 6.692744504749062e-05, + "loss": 0.0121, + "step": 2700 + }, + { + "epoch": 4.221183800623053, + "grad_norm": 0.16744938492774963, + "learning_rate": 6.668488036390196e-05, + "loss": 0.0112, + "step": 2710 + }, + { + "epoch": 4.2367601246105915, + "grad_norm": 0.2038905769586563, + "learning_rate": 6.644187298531482e-05, + "loss": 0.0109, + "step": 2720 + }, + { + "epoch": 4.252336448598131, + "grad_norm": 0.22416190803050995, + "learning_rate": 6.619842935937214e-05, + "loss": 0.0184, + "step": 2730 + }, + { + "epoch": 4.26791277258567, + "grad_norm": 0.21082550287246704, + "learning_rate": 6.59545559452917e-05, + "loss": 0.0107, + "step": 2740 + }, + { + "epoch": 4.283489096573208, + "grad_norm": 0.1576286405324936, + "learning_rate": 6.571025921369469e-05, + "loss": 0.0098, + "step": 2750 + }, + { + "epoch": 4.299065420560748, + "grad_norm": 0.1269635260105133, + "learning_rate": 6.54655456464341e-05, + "loss": 0.0106, + "step": 2760 + }, + { + "epoch": 4.314641744548287, + "grad_norm": 0.20629334449768066, + "learning_rate": 6.522042173642262e-05, + "loss": 0.0126, + "step": 2770 + }, + { + "epoch": 4.330218068535825, + "grad_norm": 0.1645572930574417, + "learning_rate": 6.497489398746052e-05, + "loss": 0.0136, + "step": 2780 + }, + { + "epoch": 4.345794392523365, + "grad_norm": 0.1145990639925003, + "learning_rate": 6.4728968914063e-05, + "loss": 0.0103, + "step": 2790 + }, + { + "epoch": 4.361370716510903, + "grad_norm": 0.15445075929164886, + "learning_rate": 6.448265304128732e-05, + "loss": 0.0121, + "step": 2800 + }, + { + "epoch": 4.376947040498442, + "grad_norm": 0.10428357869386673, + "learning_rate": 6.42359529045597e-05, + "loss": 0.0094, + "step": 2810 + }, + { + "epoch": 4.392523364485982, + "grad_norm": 0.2353171408176422, + "learning_rate": 6.398887504950199e-05, + "loss": 0.0104, + "step": 2820 + }, + { + "epoch": 4.40809968847352, + "grad_norm": 0.16263040900230408, + "learning_rate": 6.374142603175782e-05, + "loss": 0.0114, + "step": 2830 + }, + { + "epoch": 4.423676012461059, + "grad_norm": 0.22931188344955444, + "learning_rate": 6.349361241681886e-05, + "loss": 0.0135, + "step": 2840 + }, + { + "epoch": 4.4392523364485985, + "grad_norm": 0.16527006030082703, + "learning_rate": 6.32454407798505e-05, + "loss": 0.012, + "step": 2850 + }, + { + "epoch": 4.454828660436137, + "grad_norm": 0.1626274138689041, + "learning_rate": 6.29969177055174e-05, + "loss": 0.0098, + "step": 2860 + }, + { + "epoch": 4.470404984423676, + "grad_norm": 0.19715692102909088, + "learning_rate": 6.274804978780889e-05, + "loss": 0.0142, + "step": 2870 + }, + { + "epoch": 4.485981308411215, + "grad_norm": 0.16883446276187897, + "learning_rate": 6.249884362986377e-05, + "loss": 0.0097, + "step": 2880 + }, + { + "epoch": 4.501557632398754, + "grad_norm": 0.1753227561712265, + "learning_rate": 6.224930584379542e-05, + "loss": 0.0127, + "step": 2890 + }, + { + "epoch": 4.517133956386293, + "grad_norm": 0.24264495074748993, + "learning_rate": 6.199944305051614e-05, + "loss": 0.0139, + "step": 2900 + }, + { + "epoch": 4.5327102803738315, + "grad_norm": 0.16255994141101837, + "learning_rate": 6.174926187956156e-05, + "loss": 0.0112, + "step": 2910 + }, + { + "epoch": 4.548286604361371, + "grad_norm": 0.11260908842086792, + "learning_rate": 6.149876896891473e-05, + "loss": 0.0118, + "step": 2920 + }, + { + "epoch": 4.563862928348909, + "grad_norm": 0.11248435080051422, + "learning_rate": 6.124797096483e-05, + "loss": 0.0101, + "step": 2930 + }, + { + "epoch": 4.579439252336448, + "grad_norm": 0.10404013842344284, + "learning_rate": 6.099687452165669e-05, + "loss": 0.0095, + "step": 2940 + }, + { + "epoch": 4.595015576323988, + "grad_norm": 0.19719412922859192, + "learning_rate": 6.074548630166247e-05, + "loss": 0.0154, + "step": 2950 + }, + { + "epoch": 4.610591900311526, + "grad_norm": 0.25961896777153015, + "learning_rate": 6.0493812974856703e-05, + "loss": 0.0113, + "step": 2960 + }, + { + "epoch": 4.626168224299065, + "grad_norm": 0.18753407895565033, + "learning_rate": 6.024186121881335e-05, + "loss": 0.0093, + "step": 2970 + }, + { + "epoch": 4.641744548286605, + "grad_norm": 0.15325991809368134, + "learning_rate": 5.998963771849386e-05, + "loss": 0.0122, + "step": 2980 + }, + { + "epoch": 4.657320872274143, + "grad_norm": 0.13568300008773804, + "learning_rate": 5.973714916606984e-05, + "loss": 0.0097, + "step": 2990 + }, + { + "epoch": 4.672897196261682, + "grad_norm": 0.21129286289215088, + "learning_rate": 5.9484402260745386e-05, + "loss": 0.0106, + "step": 3000 + }, + { + "epoch": 4.688473520249222, + "grad_norm": 0.16718155145645142, + "learning_rate": 5.923140370857946e-05, + "loss": 0.0092, + "step": 3010 + }, + { + "epoch": 4.70404984423676, + "grad_norm": 0.18136487901210785, + "learning_rate": 5.8978160222307796e-05, + "loss": 0.0098, + "step": 3020 + }, + { + "epoch": 4.719626168224299, + "grad_norm": 0.12381799519062042, + "learning_rate": 5.8724678521165024e-05, + "loss": 0.0089, + "step": 3030 + }, + { + "epoch": 4.735202492211838, + "grad_norm": 0.231089249253273, + "learning_rate": 5.847096533070615e-05, + "loss": 0.0113, + "step": 3040 + }, + { + "epoch": 4.750778816199377, + "grad_norm": 0.13842178881168365, + "learning_rate": 5.8217027382628244e-05, + "loss": 0.0134, + "step": 3050 + }, + { + "epoch": 4.766355140186916, + "grad_norm": 0.1382349133491516, + "learning_rate": 5.7962871414591844e-05, + "loss": 0.011, + "step": 3060 + }, + { + "epoch": 4.781931464174455, + "grad_norm": 0.20510578155517578, + "learning_rate": 5.770850417004211e-05, + "loss": 0.0124, + "step": 3070 + }, + { + "epoch": 4.797507788161994, + "grad_norm": 0.09913448244333267, + "learning_rate": 5.745393239802994e-05, + "loss": 0.0136, + "step": 3080 + }, + { + "epoch": 4.813084112149532, + "grad_norm": 0.16298551857471466, + "learning_rate": 5.719916285303293e-05, + "loss": 0.0109, + "step": 3090 + }, + { + "epoch": 4.8286604361370715, + "grad_norm": 0.14564213156700134, + "learning_rate": 5.694420229477606e-05, + "loss": 0.012, + "step": 3100 + }, + { + "epoch": 4.844236760124611, + "grad_norm": 0.16995395720005035, + "learning_rate": 5.66890574880525e-05, + "loss": 0.0085, + "step": 3110 + }, + { + "epoch": 4.859813084112149, + "grad_norm": 0.11849524825811386, + "learning_rate": 5.6433735202543945e-05, + "loss": 0.0115, + "step": 3120 + }, + { + "epoch": 4.8753894080996885, + "grad_norm": 0.22196464240550995, + "learning_rate": 5.617824221264116e-05, + "loss": 0.011, + "step": 3130 + }, + { + "epoch": 4.890965732087228, + "grad_norm": 0.15856057405471802, + "learning_rate": 5.5922585297264096e-05, + "loss": 0.011, + "step": 3140 + }, + { + "epoch": 4.906542056074766, + "grad_norm": 0.1632566601037979, + "learning_rate": 5.5666771239682116e-05, + "loss": 0.0116, + "step": 3150 + }, + { + "epoch": 4.922118380062305, + "grad_norm": 0.1795213520526886, + "learning_rate": 5.541080682733398e-05, + "loss": 0.0148, + "step": 3160 + }, + { + "epoch": 4.937694704049845, + "grad_norm": 0.1980634480714798, + "learning_rate": 5.5154698851647776e-05, + "loss": 0.0125, + "step": 3170 + }, + { + "epoch": 4.953271028037383, + "grad_norm": 0.1593889445066452, + "learning_rate": 5.489845410786075e-05, + "loss": 0.0101, + "step": 3180 + }, + { + "epoch": 4.968847352024922, + "grad_norm": 0.14490163326263428, + "learning_rate": 5.4642079394838905e-05, + "loss": 0.0085, + "step": 3190 + }, + { + "epoch": 4.984423676012461, + "grad_norm": 0.21932144463062286, + "learning_rate": 5.4385581514896724e-05, + "loss": 0.012, + "step": 3200 + }, + { + "epoch": 5.0, + "grad_norm": 0.17571520805358887, + "learning_rate": 5.4128967273616625e-05, + "loss": 0.0111, + "step": 3210 + }, + { + "epoch": 5.015576323987539, + "grad_norm": 0.2138480544090271, + "learning_rate": 5.3872243479668415e-05, + "loss": 0.011, + "step": 3220 + }, + { + "epoch": 5.031152647975078, + "grad_norm": 0.13243074715137482, + "learning_rate": 5.361541694462865e-05, + "loss": 0.0084, + "step": 3230 + }, + { + "epoch": 5.046728971962617, + "grad_norm": 0.15625926852226257, + "learning_rate": 5.335849448279984e-05, + "loss": 0.0102, + "step": 3240 + }, + { + "epoch": 5.062305295950155, + "grad_norm": 0.18019627034664154, + "learning_rate": 5.310148291102973e-05, + "loss": 0.0166, + "step": 3250 + }, + { + "epoch": 5.077881619937695, + "grad_norm": 0.2128341943025589, + "learning_rate": 5.284438904853036e-05, + "loss": 0.0129, + "step": 3260 + }, + { + "epoch": 5.093457943925234, + "grad_norm": 0.21096031367778778, + "learning_rate": 5.258721971669719e-05, + "loss": 0.0098, + "step": 3270 + }, + { + "epoch": 5.109034267912772, + "grad_norm": 0.17452724277973175, + "learning_rate": 5.23299817389281e-05, + "loss": 0.0114, + "step": 3280 + }, + { + "epoch": 5.1246105919003115, + "grad_norm": 0.0774535983800888, + "learning_rate": 5.2072681940442306e-05, + "loss": 0.0114, + "step": 3290 + }, + { + "epoch": 5.140186915887851, + "grad_norm": 0.16518208384513855, + "learning_rate": 5.181532714809929e-05, + "loss": 0.0113, + "step": 3300 + }, + { + "epoch": 5.155763239875389, + "grad_norm": 0.19055700302124023, + "learning_rate": 5.155792419021766e-05, + "loss": 0.0094, + "step": 3310 + }, + { + "epoch": 5.1713395638629285, + "grad_norm": 0.2532927691936493, + "learning_rate": 5.130047989639404e-05, + "loss": 0.0117, + "step": 3320 + }, + { + "epoch": 5.186915887850467, + "grad_norm": 0.15372776985168457, + "learning_rate": 5.104300109732178e-05, + "loss": 0.0123, + "step": 3330 + }, + { + "epoch": 5.202492211838006, + "grad_norm": 0.2493387758731842, + "learning_rate": 5.078549462460972e-05, + "loss": 0.0103, + "step": 3340 + }, + { + "epoch": 5.218068535825545, + "grad_norm": 0.3557164669036865, + "learning_rate": 5.0527967310600976e-05, + "loss": 0.018, + "step": 3350 + }, + { + "epoch": 5.233644859813084, + "grad_norm": 0.19296665489673615, + "learning_rate": 5.027042598819165e-05, + "loss": 0.0087, + "step": 3360 + }, + { + "epoch": 5.249221183800623, + "grad_norm": 0.1797320544719696, + "learning_rate": 5.001287749064952e-05, + "loss": 0.0124, + "step": 3370 + }, + { + "epoch": 5.264797507788162, + "grad_norm": 0.1910810023546219, + "learning_rate": 4.975532865143276e-05, + "loss": 0.0111, + "step": 3380 + }, + { + "epoch": 5.280373831775701, + "grad_norm": 0.18078486621379852, + "learning_rate": 4.9497786304008535e-05, + "loss": 0.0105, + "step": 3390 + }, + { + "epoch": 5.29595015576324, + "grad_norm": 0.1647537797689438, + "learning_rate": 4.924025728167186e-05, + "loss": 0.0102, + "step": 3400 + }, + { + "epoch": 5.311526479750778, + "grad_norm": 0.1424635648727417, + "learning_rate": 4.8982748417364154e-05, + "loss": 0.0129, + "step": 3410 + }, + { + "epoch": 5.327102803738318, + "grad_norm": 0.237388014793396, + "learning_rate": 4.872526654349196e-05, + "loss": 0.0081, + "step": 3420 + }, + { + "epoch": 5.342679127725857, + "grad_norm": 0.17875955998897552, + "learning_rate": 4.8467818491745756e-05, + "loss": 0.0098, + "step": 3430 + }, + { + "epoch": 5.358255451713395, + "grad_norm": 0.18949951231479645, + "learning_rate": 4.8210411092918584e-05, + "loss": 0.0143, + "step": 3440 + }, + { + "epoch": 5.373831775700935, + "grad_norm": 0.1877906322479248, + "learning_rate": 4.795305117672486e-05, + "loss": 0.0116, + "step": 3450 + }, + { + "epoch": 5.389408099688474, + "grad_norm": 0.08062563091516495, + "learning_rate": 4.7695745571619156e-05, + "loss": 0.0115, + "step": 3460 + }, + { + "epoch": 5.404984423676012, + "grad_norm": 0.2015499770641327, + "learning_rate": 4.7438501104615046e-05, + "loss": 0.0098, + "step": 3470 + }, + { + "epoch": 5.420560747663552, + "grad_norm": 0.15176242589950562, + "learning_rate": 4.7181324601103905e-05, + "loss": 0.0109, + "step": 3480 + }, + { + "epoch": 5.43613707165109, + "grad_norm": 0.11479273438453674, + "learning_rate": 4.6924222884673906e-05, + "loss": 0.0114, + "step": 3490 + }, + { + "epoch": 5.451713395638629, + "grad_norm": 0.20649981498718262, + "learning_rate": 4.666720277692889e-05, + "loss": 0.0111, + "step": 3500 + }, + { + "epoch": 5.4672897196261685, + "grad_norm": 0.1916668713092804, + "learning_rate": 4.6410271097307375e-05, + "loss": 0.0129, + "step": 3510 + }, + { + "epoch": 5.482866043613707, + "grad_norm": 0.14754751324653625, + "learning_rate": 4.6153434662901677e-05, + "loss": 0.0102, + "step": 3520 + }, + { + "epoch": 5.498442367601246, + "grad_norm": 0.1280348300933838, + "learning_rate": 4.589670028827697e-05, + "loss": 0.0103, + "step": 3530 + }, + { + "epoch": 5.5140186915887845, + "grad_norm": 0.19268469512462616, + "learning_rate": 4.5640074785290546e-05, + "loss": 0.007, + "step": 3540 + }, + { + "epoch": 5.529595015576324, + "grad_norm": 0.2448658049106598, + "learning_rate": 4.538356496291101e-05, + "loss": 0.0089, + "step": 3550 + }, + { + "epoch": 5.545171339563863, + "grad_norm": 0.15902170538902283, + "learning_rate": 4.512717762703765e-05, + "loss": 0.0103, + "step": 3560 + }, + { + "epoch": 5.5607476635514015, + "grad_norm": 0.1977582573890686, + "learning_rate": 4.487091958031984e-05, + "loss": 0.0095, + "step": 3570 + }, + { + "epoch": 5.576323987538941, + "grad_norm": 0.1744050830602646, + "learning_rate": 4.461479762197659e-05, + "loss": 0.0093, + "step": 3580 + }, + { + "epoch": 5.59190031152648, + "grad_norm": 0.17722630500793457, + "learning_rate": 4.435881854761612e-05, + "loss": 0.0095, + "step": 3590 + }, + { + "epoch": 5.607476635514018, + "grad_norm": 0.17866544425487518, + "learning_rate": 4.410298914905554e-05, + "loss": 0.0109, + "step": 3600 + }, + { + "epoch": 5.623052959501558, + "grad_norm": 0.11874900013208389, + "learning_rate": 4.384731621414066e-05, + "loss": 0.0093, + "step": 3610 + }, + { + "epoch": 5.638629283489097, + "grad_norm": 0.10844465345144272, + "learning_rate": 4.359180652656587e-05, + "loss": 0.0127, + "step": 3620 + }, + { + "epoch": 5.654205607476635, + "grad_norm": 0.0915406346321106, + "learning_rate": 4.3336466865694175e-05, + "loss": 0.0107, + "step": 3630 + }, + { + "epoch": 5.669781931464175, + "grad_norm": 0.12223953753709793, + "learning_rate": 4.308130400637733e-05, + "loss": 0.0102, + "step": 3640 + }, + { + "epoch": 5.685358255451713, + "grad_norm": 0.15521594882011414, + "learning_rate": 4.2826324718776044e-05, + "loss": 0.0131, + "step": 3650 + }, + { + "epoch": 5.700934579439252, + "grad_norm": 0.14360561966896057, + "learning_rate": 4.2571535768180387e-05, + "loss": 0.0111, + "step": 3660 + }, + { + "epoch": 5.716510903426792, + "grad_norm": 0.20258069038391113, + "learning_rate": 4.2316943914830236e-05, + "loss": 0.0097, + "step": 3670 + }, + { + "epoch": 5.73208722741433, + "grad_norm": 0.17553642392158508, + "learning_rate": 4.206255591373603e-05, + "loss": 0.0102, + "step": 3680 + }, + { + "epoch": 5.747663551401869, + "grad_norm": 0.1354387104511261, + "learning_rate": 4.1808378514499356e-05, + "loss": 0.011, + "step": 3690 + }, + { + "epoch": 5.763239875389408, + "grad_norm": 0.1975325047969818, + "learning_rate": 4.155441846113404e-05, + "loss": 0.0106, + "step": 3700 + }, + { + "epoch": 5.778816199376947, + "grad_norm": 0.14944128692150116, + "learning_rate": 4.1300682491887096e-05, + "loss": 0.0079, + "step": 3710 + }, + { + "epoch": 5.794392523364486, + "grad_norm": 0.17943407595157623, + "learning_rate": 4.104717733905999e-05, + "loss": 0.0106, + "step": 3720 + }, + { + "epoch": 5.809968847352025, + "grad_norm": 0.12425987422466278, + "learning_rate": 4.079390972882999e-05, + "loss": 0.0105, + "step": 3730 + }, + { + "epoch": 5.825545171339564, + "grad_norm": 0.19627559185028076, + "learning_rate": 4.054088638107174e-05, + "loss": 0.0127, + "step": 3740 + }, + { + "epoch": 5.841121495327103, + "grad_norm": 0.12838999927043915, + "learning_rate": 4.028811400917894e-05, + "loss": 0.0081, + "step": 3750 + }, + { + "epoch": 5.8566978193146415, + "grad_norm": 0.12284855544567108, + "learning_rate": 4.00355993198862e-05, + "loss": 0.0081, + "step": 3760 + }, + { + "epoch": 5.872274143302181, + "grad_norm": 0.09254119545221329, + "learning_rate": 3.9783349013091094e-05, + "loss": 0.0101, + "step": 3770 + }, + { + "epoch": 5.88785046728972, + "grad_norm": 0.15401200950145721, + "learning_rate": 3.95313697816765e-05, + "loss": 0.0111, + "step": 3780 + }, + { + "epoch": 5.9034267912772584, + "grad_norm": 0.13822408020496368, + "learning_rate": 3.927966831133285e-05, + "loss": 0.0085, + "step": 3790 + }, + { + "epoch": 5.919003115264798, + "grad_norm": 0.13557831943035126, + "learning_rate": 3.902825128038086e-05, + "loss": 0.0088, + "step": 3800 + }, + { + "epoch": 5.934579439252336, + "grad_norm": 0.19986526668071747, + "learning_rate": 3.877712535959429e-05, + "loss": 0.0097, + "step": 3810 + }, + { + "epoch": 5.950155763239875, + "grad_norm": 0.09881894290447235, + "learning_rate": 3.8526297212022954e-05, + "loss": 0.0107, + "step": 3820 + }, + { + "epoch": 5.965732087227415, + "grad_norm": 0.20597843825817108, + "learning_rate": 3.8275773492815935e-05, + "loss": 0.0117, + "step": 3830 + }, + { + "epoch": 5.981308411214953, + "grad_norm": 0.16056543588638306, + "learning_rate": 3.802556084904504e-05, + "loss": 0.0094, + "step": 3840 + }, + { + "epoch": 5.996884735202492, + "grad_norm": 0.1428888738155365, + "learning_rate": 3.7775665919528365e-05, + "loss": 0.0096, + "step": 3850 + }, + { + "epoch": 6.012461059190031, + "grad_norm": 0.3193315267562866, + "learning_rate": 3.752609533465421e-05, + "loss": 0.011, + "step": 3860 + }, + { + "epoch": 6.02803738317757, + "grad_norm": 0.12124066054821014, + "learning_rate": 3.72768557162051e-05, + "loss": 0.0108, + "step": 3870 + }, + { + "epoch": 6.043613707165109, + "grad_norm": 0.1567990630865097, + "learning_rate": 3.702795367718222e-05, + "loss": 0.0117, + "step": 3880 + }, + { + "epoch": 6.059190031152648, + "grad_norm": 0.14114873111248016, + "learning_rate": 3.677939582162973e-05, + "loss": 0.0079, + "step": 3890 + }, + { + "epoch": 6.074766355140187, + "grad_norm": 0.12263435125350952, + "learning_rate": 3.653118874445975e-05, + "loss": 0.009, + "step": 3900 + }, + { + "epoch": 6.090342679127726, + "grad_norm": 0.24416130781173706, + "learning_rate": 3.6283339031277314e-05, + "loss": 0.0105, + "step": 3910 + }, + { + "epoch": 6.105919003115265, + "grad_norm": 0.14247013628482819, + "learning_rate": 3.603585325820556e-05, + "loss": 0.0117, + "step": 3920 + }, + { + "epoch": 6.121495327102804, + "grad_norm": 0.14784014225006104, + "learning_rate": 3.578873799171137e-05, + "loss": 0.0102, + "step": 3930 + }, + { + "epoch": 6.137071651090342, + "grad_norm": 0.1328940987586975, + "learning_rate": 3.5541999788431044e-05, + "loss": 0.0079, + "step": 3940 + }, + { + "epoch": 6.1526479750778815, + "grad_norm": 0.12007959187030792, + "learning_rate": 3.529564519499641e-05, + "loss": 0.009, + "step": 3950 + }, + { + "epoch": 6.168224299065421, + "grad_norm": 0.16528165340423584, + "learning_rate": 3.5049680747861025e-05, + "loss": 0.0094, + "step": 3960 + }, + { + "epoch": 6.183800623052959, + "grad_norm": 0.2054515779018402, + "learning_rate": 3.4804112973126914e-05, + "loss": 0.0118, + "step": 3970 + }, + { + "epoch": 6.1993769470404985, + "grad_norm": 0.16610035300254822, + "learning_rate": 3.455894838637122e-05, + "loss": 0.0087, + "step": 3980 + }, + { + "epoch": 6.214953271028038, + "grad_norm": 0.13173583149909973, + "learning_rate": 3.4314193492473465e-05, + "loss": 0.0084, + "step": 3990 + }, + { + "epoch": 6.230529595015576, + "grad_norm": 0.19198912382125854, + "learning_rate": 3.406985478544287e-05, + "loss": 0.0128, + "step": 4000 + }, + { + "epoch": 6.246105919003115, + "grad_norm": 0.11047305911779404, + "learning_rate": 3.382593874824615e-05, + "loss": 0.0091, + "step": 4010 + }, + { + "epoch": 6.261682242990654, + "grad_norm": 0.11866667866706848, + "learning_rate": 3.3582451852635404e-05, + "loss": 0.0073, + "step": 4020 + }, + { + "epoch": 6.277258566978193, + "grad_norm": 0.526498556137085, + "learning_rate": 3.333940055897646e-05, + "loss": 0.0081, + "step": 4030 + }, + { + "epoch": 6.292834890965732, + "grad_norm": 0.23729725182056427, + "learning_rate": 3.3096791316077425e-05, + "loss": 0.0124, + "step": 4040 + }, + { + "epoch": 6.308411214953271, + "grad_norm": 0.2855716943740845, + "learning_rate": 3.285463056101764e-05, + "loss": 0.0118, + "step": 4050 + }, + { + "epoch": 6.32398753894081, + "grad_norm": 0.17290759086608887, + "learning_rate": 3.2612924718976826e-05, + "loss": 0.0091, + "step": 4060 + }, + { + "epoch": 6.339563862928349, + "grad_norm": 0.16979916393756866, + "learning_rate": 3.237168020306465e-05, + "loss": 0.0094, + "step": 4070 + }, + { + "epoch": 6.355140186915888, + "grad_norm": 0.1293216347694397, + "learning_rate": 3.2130903414150535e-05, + "loss": 0.0098, + "step": 4080 + }, + { + "epoch": 6.370716510903427, + "grad_norm": 0.171615332365036, + "learning_rate": 3.189060074069385e-05, + "loss": 0.0114, + "step": 4090 + }, + { + "epoch": 6.386292834890965, + "grad_norm": 0.17334286868572235, + "learning_rate": 3.165077855857437e-05, + "loss": 0.0093, + "step": 4100 + }, + { + "epoch": 6.401869158878505, + "grad_norm": 0.07420803606510162, + "learning_rate": 3.141144323092319e-05, + "loss": 0.008, + "step": 4110 + }, + { + "epoch": 6.417445482866044, + "grad_norm": 0.19926607608795166, + "learning_rate": 3.117260110795379e-05, + "loss": 0.009, + "step": 4120 + }, + { + "epoch": 6.433021806853582, + "grad_norm": 0.10032382607460022, + "learning_rate": 3.093425852679363e-05, + "loss": 0.0078, + "step": 4130 + }, + { + "epoch": 6.4485981308411215, + "grad_norm": 0.29928913712501526, + "learning_rate": 3.069642181131592e-05, + "loss": 0.0091, + "step": 4140 + }, + { + "epoch": 6.464174454828661, + "grad_norm": 0.09755891561508179, + "learning_rate": 3.0459097271971982e-05, + "loss": 0.0074, + "step": 4150 + }, + { + "epoch": 6.479750778816199, + "grad_norm": 0.10077785700559616, + "learning_rate": 3.0222291205623603e-05, + "loss": 0.0094, + "step": 4160 + }, + { + "epoch": 6.4953271028037385, + "grad_norm": 0.13297952711582184, + "learning_rate": 2.9986009895376192e-05, + "loss": 0.0107, + "step": 4170 + }, + { + "epoch": 6.510903426791277, + "grad_norm": 0.10207609832286835, + "learning_rate": 2.9750259610411884e-05, + "loss": 0.01, + "step": 4180 + }, + { + "epoch": 6.526479750778816, + "grad_norm": 0.2845441401004791, + "learning_rate": 2.9515046605823304e-05, + "loss": 0.0095, + "step": 4190 + }, + { + "epoch": 6.542056074766355, + "grad_norm": 0.16185392439365387, + "learning_rate": 2.9280377122447556e-05, + "loss": 0.0093, + "step": 4200 + }, + { + "epoch": 6.557632398753894, + "grad_norm": 0.1834220588207245, + "learning_rate": 2.9046257386700704e-05, + "loss": 0.0086, + "step": 4210 + }, + { + "epoch": 6.573208722741433, + "grad_norm": 0.12148604542016983, + "learning_rate": 2.8812693610412444e-05, + "loss": 0.0074, + "step": 4220 + }, + { + "epoch": 6.588785046728972, + "grad_norm": 0.10846960544586182, + "learning_rate": 2.8579691990661433e-05, + "loss": 0.0087, + "step": 4230 + }, + { + "epoch": 6.604361370716511, + "grad_norm": 0.17413875460624695, + "learning_rate": 2.834725870961079e-05, + "loss": 0.0085, + "step": 4240 + }, + { + "epoch": 6.61993769470405, + "grad_norm": 0.08588691055774689, + "learning_rate": 2.8115399934343995e-05, + "loss": 0.0085, + "step": 4250 + }, + { + "epoch": 6.635514018691588, + "grad_norm": 0.14263932406902313, + "learning_rate": 2.7884121816701463e-05, + "loss": 0.0109, + "step": 4260 + }, + { + "epoch": 6.651090342679128, + "grad_norm": 0.07699258625507355, + "learning_rate": 2.7653430493117073e-05, + "loss": 0.0107, + "step": 4270 + }, + { + "epoch": 6.666666666666667, + "grad_norm": 0.1988629847764969, + "learning_rate": 2.7423332084455544e-05, + "loss": 0.0093, + "step": 4280 + }, + { + "epoch": 6.682242990654205, + "grad_norm": 0.16722887754440308, + "learning_rate": 2.7193832695849907e-05, + "loss": 0.0104, + "step": 4290 + }, + { + "epoch": 6.697819314641745, + "grad_norm": 0.16533584892749786, + "learning_rate": 2.696493841653961e-05, + "loss": 0.01, + "step": 4300 + }, + { + "epoch": 6.713395638629283, + "grad_norm": 0.10756443440914154, + "learning_rate": 2.6736655319708925e-05, + "loss": 0.0079, + "step": 4310 + }, + { + "epoch": 6.728971962616822, + "grad_norm": 0.19490206241607666, + "learning_rate": 2.650898946232574e-05, + "loss": 0.0084, + "step": 4320 + }, + { + "epoch": 6.744548286604362, + "grad_norm": 0.15268468856811523, + "learning_rate": 2.6281946884980957e-05, + "loss": 0.0105, + "step": 4330 + }, + { + "epoch": 6.7601246105919, + "grad_norm": 0.11121109873056412, + "learning_rate": 2.60555336117282e-05, + "loss": 0.0092, + "step": 4340 + }, + { + "epoch": 6.775700934579439, + "grad_norm": 0.1262609362602234, + "learning_rate": 2.582975564992385e-05, + "loss": 0.0087, + "step": 4350 + }, + { + "epoch": 6.7912772585669785, + "grad_norm": 0.1314808428287506, + "learning_rate": 2.560461899006791e-05, + "loss": 0.0077, + "step": 4360 + }, + { + "epoch": 6.806853582554517, + "grad_norm": 0.19681373238563538, + "learning_rate": 2.5380129605644753e-05, + "loss": 0.0106, + "step": 4370 + }, + { + "epoch": 6.822429906542056, + "grad_norm": 0.17111235857009888, + "learning_rate": 2.5156293452964908e-05, + "loss": 0.0072, + "step": 4380 + }, + { + "epoch": 6.838006230529595, + "grad_norm": 0.12845902144908905, + "learning_rate": 2.4933116471006796e-05, + "loss": 0.0096, + "step": 4390 + }, + { + "epoch": 6.853582554517134, + "grad_norm": 0.19043442606925964, + "learning_rate": 2.471060458125933e-05, + "loss": 0.0092, + "step": 4400 + }, + { + "epoch": 6.869158878504673, + "grad_norm": 0.08261138945817947, + "learning_rate": 2.448876368756472e-05, + "loss": 0.0063, + "step": 4410 + }, + { + "epoch": 6.8847352024922115, + "grad_norm": 0.1296285092830658, + "learning_rate": 2.4267599675961784e-05, + "loss": 0.0079, + "step": 4420 + }, + { + "epoch": 6.900311526479751, + "grad_norm": 0.10047978162765503, + "learning_rate": 2.4047118414529886e-05, + "loss": 0.0073, + "step": 4430 + }, + { + "epoch": 6.91588785046729, + "grad_norm": 0.10485507547855377, + "learning_rate": 2.3827325753233166e-05, + "loss": 0.0087, + "step": 4440 + }, + { + "epoch": 6.931464174454828, + "grad_norm": 0.08776561915874481, + "learning_rate": 2.3608227523765292e-05, + "loss": 0.0074, + "step": 4450 + }, + { + "epoch": 6.947040498442368, + "grad_norm": 0.11480440944433212, + "learning_rate": 2.338982953939488e-05, + "loss": 0.0062, + "step": 4460 + }, + { + "epoch": 6.962616822429906, + "grad_norm": 0.2410510927438736, + "learning_rate": 2.3172137594811044e-05, + "loss": 0.0097, + "step": 4470 + }, + { + "epoch": 6.978193146417445, + "grad_norm": 0.2679021656513214, + "learning_rate": 2.2955157465969835e-05, + "loss": 0.0096, + "step": 4480 + }, + { + "epoch": 6.993769470404985, + "grad_norm": 0.12317413091659546, + "learning_rate": 2.273889490994082e-05, + "loss": 0.0093, + "step": 4490 + }, + { + "epoch": 7.009345794392523, + "grad_norm": 0.16951963305473328, + "learning_rate": 2.2523355664754498e-05, + "loss": 0.0087, + "step": 4500 + }, + { + "epoch": 7.024922118380062, + "grad_norm": 0.13611598312854767, + "learning_rate": 2.2308545449249963e-05, + "loss": 0.0083, + "step": 4510 + }, + { + "epoch": 7.040498442367602, + "grad_norm": 0.0933770090341568, + "learning_rate": 2.209446996292312e-05, + "loss": 0.008, + "step": 4520 + }, + { + "epoch": 7.05607476635514, + "grad_norm": 0.08887217938899994, + "learning_rate": 2.188113488577559e-05, + "loss": 0.008, + "step": 4530 + }, + { + "epoch": 7.071651090342679, + "grad_norm": 0.10375925153493881, + "learning_rate": 2.166854587816391e-05, + "loss": 0.0083, + "step": 4540 + }, + { + "epoch": 7.087227414330218, + "grad_norm": 0.298216849565506, + "learning_rate": 2.1456708580649375e-05, + "loss": 0.0085, + "step": 4550 + }, + { + "epoch": 7.102803738317757, + "grad_norm": 0.12560240924358368, + "learning_rate": 2.1245628613848394e-05, + "loss": 0.0081, + "step": 4560 + }, + { + "epoch": 7.118380062305296, + "grad_norm": 0.11159154027700424, + "learning_rate": 2.1035311578283296e-05, + "loss": 0.0079, + "step": 4570 + }, + { + "epoch": 7.133956386292835, + "grad_norm": 0.17967365682125092, + "learning_rate": 2.0825763054233844e-05, + "loss": 0.0079, + "step": 4580 + }, + { + "epoch": 7.149532710280374, + "grad_norm": 0.10396911203861237, + "learning_rate": 2.061698860158905e-05, + "loss": 0.0066, + "step": 4590 + }, + { + "epoch": 7.165109034267913, + "grad_norm": 0.14478343725204468, + "learning_rate": 2.0408993759699773e-05, + "loss": 0.0093, + "step": 4600 + }, + { + "epoch": 7.1806853582554515, + "grad_norm": 0.16734841465950012, + "learning_rate": 2.0201784047231675e-05, + "loss": 0.0074, + "step": 4610 + }, + { + "epoch": 7.196261682242991, + "grad_norm": 0.21679063141345978, + "learning_rate": 1.999536496201878e-05, + "loss": 0.0073, + "step": 4620 + }, + { + "epoch": 7.211838006230529, + "grad_norm": 0.07951363176107407, + "learning_rate": 1.9789741980917682e-05, + "loss": 0.0067, + "step": 4630 + }, + { + "epoch": 7.2274143302180685, + "grad_norm": 0.1287214457988739, + "learning_rate": 1.95849205596621e-05, + "loss": 0.0098, + "step": 4640 + }, + { + "epoch": 7.242990654205608, + "grad_norm": 0.17212507128715515, + "learning_rate": 1.938090613271834e-05, + "loss": 0.0078, + "step": 4650 + }, + { + "epoch": 7.258566978193146, + "grad_norm": 0.17881302535533905, + "learning_rate": 1.9177704113140815e-05, + "loss": 0.0068, + "step": 4660 + }, + { + "epoch": 7.274143302180685, + "grad_norm": 0.13449600338935852, + "learning_rate": 1.897531989242867e-05, + "loss": 0.0106, + "step": 4670 + }, + { + "epoch": 7.289719626168225, + "grad_norm": 0.12616336345672607, + "learning_rate": 1.8773758840382622e-05, + "loss": 0.0069, + "step": 4680 + }, + { + "epoch": 7.305295950155763, + "grad_norm": 0.16827356815338135, + "learning_rate": 1.8573026304962443e-05, + "loss": 0.007, + "step": 4690 + }, + { + "epoch": 7.320872274143302, + "grad_norm": 0.1590610295534134, + "learning_rate": 1.837312761214518e-05, + "loss": 0.0099, + "step": 4700 + }, + { + "epoch": 7.336448598130841, + "grad_norm": 0.094545878469944, + "learning_rate": 1.8174068065783765e-05, + "loss": 0.0064, + "step": 4710 + }, + { + "epoch": 7.35202492211838, + "grad_norm": 0.09499020129442215, + "learning_rate": 1.7975852947466278e-05, + "loss": 0.0077, + "step": 4720 + }, + { + "epoch": 7.367601246105919, + "grad_norm": 0.0822395458817482, + "learning_rate": 1.7778487516375876e-05, + "loss": 0.0062, + "step": 4730 + }, + { + "epoch": 7.383177570093458, + "grad_norm": 0.09064001590013504, + "learning_rate": 1.7581977009151157e-05, + "loss": 0.0071, + "step": 4740 + }, + { + "epoch": 7.398753894080997, + "grad_norm": 0.11729217320680618, + "learning_rate": 1.7386326639747384e-05, + "loss": 0.007, + "step": 4750 + }, + { + "epoch": 7.414330218068536, + "grad_norm": 0.1214100793004036, + "learning_rate": 1.7191541599297926e-05, + "loss": 0.0075, + "step": 4760 + }, + { + "epoch": 7.429906542056075, + "grad_norm": 0.18827670812606812, + "learning_rate": 1.6997627055976696e-05, + "loss": 0.0085, + "step": 4770 + }, + { + "epoch": 7.445482866043614, + "grad_norm": 0.15139153599739075, + "learning_rate": 1.6804588154860962e-05, + "loss": 0.0069, + "step": 4780 + }, + { + "epoch": 7.461059190031152, + "grad_norm": 0.16040366888046265, + "learning_rate": 1.661243001779479e-05, + "loss": 0.007, + "step": 4790 + }, + { + "epoch": 7.4766355140186915, + "grad_norm": 0.20151734352111816, + "learning_rate": 1.6421157743253244e-05, + "loss": 0.0102, + "step": 4800 + }, + { + "epoch": 7.492211838006231, + "grad_norm": 0.08017118275165558, + "learning_rate": 1.6230776406207066e-05, + "loss": 0.0065, + "step": 4810 + }, + { + "epoch": 7.507788161993769, + "grad_norm": 0.06299513578414917, + "learning_rate": 1.604129105798795e-05, + "loss": 0.0093, + "step": 4820 + }, + { + "epoch": 7.5233644859813085, + "grad_norm": 0.08470407128334045, + "learning_rate": 1.585270672615465e-05, + "loss": 0.008, + "step": 4830 + }, + { + "epoch": 7.538940809968848, + "grad_norm": 0.09864120930433273, + "learning_rate": 1.5665028414359485e-05, + "loss": 0.0064, + "step": 4840 + }, + { + "epoch": 7.554517133956386, + "grad_norm": 0.08766548335552216, + "learning_rate": 1.547826110221565e-05, + "loss": 0.0083, + "step": 4850 + }, + { + "epoch": 7.570093457943925, + "grad_norm": 0.1513214409351349, + "learning_rate": 1.529240974516499e-05, + "loss": 0.0085, + "step": 4860 + }, + { + "epoch": 7.585669781931464, + "grad_norm": 0.13386106491088867, + "learning_rate": 1.5107479274346641e-05, + "loss": 0.0088, + "step": 4870 + }, + { + "epoch": 7.601246105919003, + "grad_norm": 0.12557023763656616, + "learning_rate": 1.492347459646613e-05, + "loss": 0.0102, + "step": 4880 + }, + { + "epoch": 7.616822429906542, + "grad_norm": 0.1672525256872177, + "learning_rate": 1.4740400593665154e-05, + "loss": 0.0104, + "step": 4890 + }, + { + "epoch": 7.632398753894081, + "grad_norm": 0.08632241189479828, + "learning_rate": 1.4558262123392125e-05, + "loss": 0.0065, + "step": 4900 + }, + { + "epoch": 7.64797507788162, + "grad_norm": 0.10372444242238998, + "learning_rate": 1.4377064018273245e-05, + "loss": 0.0067, + "step": 4910 + }, + { + "epoch": 7.663551401869158, + "grad_norm": 0.2614145278930664, + "learning_rate": 1.4196811085984247e-05, + "loss": 0.0087, + "step": 4920 + }, + { + "epoch": 7.679127725856698, + "grad_norm": 0.1608159989118576, + "learning_rate": 1.4017508109122912e-05, + "loss": 0.0064, + "step": 4930 + }, + { + "epoch": 7.694704049844237, + "grad_norm": 0.07221081852912903, + "learning_rate": 1.3839159845082118e-05, + "loss": 0.0067, + "step": 4940 + }, + { + "epoch": 7.710280373831775, + "grad_norm": 0.1406513899564743, + "learning_rate": 1.3661771025923653e-05, + "loss": 0.007, + "step": 4950 + }, + { + "epoch": 7.725856697819315, + "grad_norm": 0.16034474968910217, + "learning_rate": 1.3485346358252587e-05, + "loss": 0.0052, + "step": 4960 + }, + { + "epoch": 7.741433021806854, + "grad_norm": 0.08381876349449158, + "learning_rate": 1.3309890523092488e-05, + "loss": 0.0076, + "step": 4970 + }, + { + "epoch": 7.757009345794392, + "grad_norm": 0.13160599768161774, + "learning_rate": 1.3135408175761176e-05, + "loss": 0.0057, + "step": 4980 + }, + { + "epoch": 7.7725856697819315, + "grad_norm": 0.1081743836402893, + "learning_rate": 1.2961903945747172e-05, + "loss": 0.0076, + "step": 4990 + }, + { + "epoch": 7.788161993769471, + "grad_norm": 0.14261646568775177, + "learning_rate": 1.2789382436586945e-05, + "loss": 0.0073, + "step": 5000 + }, + { + "epoch": 7.803738317757009, + "grad_norm": 0.10383739322423935, + "learning_rate": 1.2617848225742667e-05, + "loss": 0.0078, + "step": 5010 + }, + { + "epoch": 7.8193146417445485, + "grad_norm": 0.07915589958429337, + "learning_rate": 1.2447305864480868e-05, + "loss": 0.0087, + "step": 5020 + }, + { + "epoch": 7.834890965732087, + "grad_norm": 0.15104737877845764, + "learning_rate": 1.2277759877751599e-05, + "loss": 0.0074, + "step": 5030 + }, + { + "epoch": 7.850467289719626, + "grad_norm": 0.1244148463010788, + "learning_rate": 1.2109214764068416e-05, + "loss": 0.0085, + "step": 5040 + }, + { + "epoch": 7.866043613707165, + "grad_norm": 0.12754645943641663, + "learning_rate": 1.1941674995389001e-05, + "loss": 0.0079, + "step": 5050 + }, + { + "epoch": 7.881619937694704, + "grad_norm": 0.16898225247859955, + "learning_rate": 1.1775145016996487e-05, + "loss": 0.0102, + "step": 5060 + }, + { + "epoch": 7.897196261682243, + "grad_norm": 0.12284207344055176, + "learning_rate": 1.160962924738157e-05, + "loss": 0.006, + "step": 5070 + }, + { + "epoch": 7.9127725856697815, + "grad_norm": 0.3757985830307007, + "learning_rate": 1.144513207812526e-05, + "loss": 0.0063, + "step": 5080 + }, + { + "epoch": 7.928348909657321, + "grad_norm": 0.24468566477298737, + "learning_rate": 1.1281657873782286e-05, + "loss": 0.0094, + "step": 5090 + }, + { + "epoch": 7.94392523364486, + "grad_norm": 0.1097400113940239, + "learning_rate": 1.111921097176542e-05, + "loss": 0.0046, + "step": 5100 + }, + { + "epoch": 7.959501557632398, + "grad_norm": 0.0909595787525177, + "learning_rate": 1.0957795682230265e-05, + "loss": 0.0095, + "step": 5110 + }, + { + "epoch": 7.975077881619938, + "grad_norm": 0.14016325771808624, + "learning_rate": 1.0797416287960993e-05, + "loss": 0.0099, + "step": 5120 + }, + { + "epoch": 7.990654205607477, + "grad_norm": 0.20573468506336212, + "learning_rate": 1.0638077044256661e-05, + "loss": 0.0067, + "step": 5130 + }, + { + "epoch": 8.006230529595015, + "grad_norm": 0.1662212461233139, + "learning_rate": 1.0479782178818327e-05, + "loss": 0.0072, + "step": 5140 + }, + { + "epoch": 8.021806853582554, + "grad_norm": 0.11567887663841248, + "learning_rate": 1.032253589163687e-05, + "loss": 0.0089, + "step": 5150 + }, + { + "epoch": 8.037383177570094, + "grad_norm": 0.16256652772426605, + "learning_rate": 1.0166342354881519e-05, + "loss": 0.0077, + "step": 5160 + }, + { + "epoch": 8.052959501557632, + "grad_norm": 0.2063686102628708, + "learning_rate": 1.0011205712789223e-05, + "loss": 0.007, + "step": 5170 + }, + { + "epoch": 8.06853582554517, + "grad_norm": 0.11452540010213852, + "learning_rate": 9.857130081554666e-06, + "loss": 0.0098, + "step": 5180 + }, + { + "epoch": 8.08411214953271, + "grad_norm": 0.18076153099536896, + "learning_rate": 9.70411954922099e-06, + "loss": 0.0079, + "step": 5190 + }, + { + "epoch": 8.09968847352025, + "grad_norm": 0.11946731805801392, + "learning_rate": 9.552178175571452e-06, + "loss": 0.0132, + "step": 5200 + }, + { + "epoch": 8.115264797507788, + "grad_norm": 0.07168768346309662, + "learning_rate": 9.401309992021573e-06, + "loss": 0.0087, + "step": 5210 + }, + { + "epoch": 8.130841121495328, + "grad_norm": 0.17897409200668335, + "learning_rate": 9.251519001512282e-06, + "loss": 0.0097, + "step": 5220 + }, + { + "epoch": 8.146417445482866, + "grad_norm": 0.07458654046058655, + "learning_rate": 9.102809178403655e-06, + "loss": 0.0062, + "step": 5230 + }, + { + "epoch": 8.161993769470405, + "grad_norm": 0.11129793524742126, + "learning_rate": 8.95518446836946e-06, + "loss": 0.0073, + "step": 5240 + }, + { + "epoch": 8.177570093457945, + "grad_norm": 0.11445678025484085, + "learning_rate": 8.808648788292512e-06, + "loss": 0.0058, + "step": 5250 + }, + { + "epoch": 8.193146417445483, + "grad_norm": 0.11773504316806793, + "learning_rate": 8.663206026160658e-06, + "loss": 0.0051, + "step": 5260 + }, + { + "epoch": 8.208722741433021, + "grad_norm": 0.06362207978963852, + "learning_rate": 8.518860040963733e-06, + "loss": 0.0067, + "step": 5270 + }, + { + "epoch": 8.22429906542056, + "grad_norm": 0.07297392934560776, + "learning_rate": 8.375614662591098e-06, + "loss": 0.0047, + "step": 5280 + }, + { + "epoch": 8.2398753894081, + "grad_norm": 0.17183497548103333, + "learning_rate": 8.233473691730003e-06, + "loss": 0.0059, + "step": 5290 + }, + { + "epoch": 8.255451713395638, + "grad_norm": 0.17913080751895905, + "learning_rate": 8.092440899764848e-06, + "loss": 0.0078, + "step": 5300 + }, + { + "epoch": 8.271028037383177, + "grad_norm": 0.2050745040178299, + "learning_rate": 7.95252002867698e-06, + "loss": 0.0071, + "step": 5310 + }, + { + "epoch": 8.286604361370717, + "grad_norm": 0.08249979466199875, + "learning_rate": 7.81371479094552e-06, + "loss": 0.0077, + "step": 5320 + }, + { + "epoch": 8.302180685358255, + "grad_norm": 0.10749052464962006, + "learning_rate": 7.67602886944881e-06, + "loss": 0.0064, + "step": 5330 + }, + { + "epoch": 8.317757009345794, + "grad_norm": 0.12287521362304688, + "learning_rate": 7.539465917366706e-06, + "loss": 0.0063, + "step": 5340 + }, + { + "epoch": 8.333333333333334, + "grad_norm": 0.11348897218704224, + "learning_rate": 7.404029558083653e-06, + "loss": 0.0081, + "step": 5350 + }, + { + "epoch": 8.348909657320872, + "grad_norm": 0.13174685835838318, + "learning_rate": 7.269723385092514e-06, + "loss": 0.0049, + "step": 5360 + }, + { + "epoch": 8.36448598130841, + "grad_norm": 0.07876933366060257, + "learning_rate": 7.1365509618992964e-06, + "loss": 0.0091, + "step": 5370 + }, + { + "epoch": 8.38006230529595, + "grad_norm": 0.1581820249557495, + "learning_rate": 7.00451582192852e-06, + "loss": 0.0079, + "step": 5380 + }, + { + "epoch": 8.39563862928349, + "grad_norm": 0.18994443118572235, + "learning_rate": 6.8736214684295305e-06, + "loss": 0.0074, + "step": 5390 + }, + { + "epoch": 8.411214953271028, + "grad_norm": 0.14710044860839844, + "learning_rate": 6.743871374383526e-06, + "loss": 0.0068, + "step": 5400 + }, + { + "epoch": 8.426791277258568, + "grad_norm": 0.05785469338297844, + "learning_rate": 6.615268982411365e-06, + "loss": 0.0059, + "step": 5410 + }, + { + "epoch": 8.442367601246106, + "grad_norm": 0.06289026886224747, + "learning_rate": 6.487817704682342e-06, + "loss": 0.0068, + "step": 5420 + }, + { + "epoch": 8.457943925233645, + "grad_norm": 0.08994963765144348, + "learning_rate": 6.361520922823499e-06, + "loss": 0.0085, + "step": 5430 + }, + { + "epoch": 8.473520249221183, + "grad_norm": 0.051708612591028214, + "learning_rate": 6.2363819878300255e-06, + "loss": 0.0053, + "step": 5440 + }, + { + "epoch": 8.489096573208723, + "grad_norm": 0.08145775645971298, + "learning_rate": 6.112404219976297e-06, + "loss": 0.0064, + "step": 5450 + }, + { + "epoch": 8.504672897196262, + "grad_norm": 0.12217671424150467, + "learning_rate": 5.989590908727771e-06, + "loss": 0.008, + "step": 5460 + }, + { + "epoch": 8.5202492211838, + "grad_norm": 0.21912190318107605, + "learning_rate": 5.8679453126537335e-06, + "loss": 0.0069, + "step": 5470 + }, + { + "epoch": 8.53582554517134, + "grad_norm": 0.17678362131118774, + "learning_rate": 5.74747065934082e-06, + "loss": 0.0066, + "step": 5480 + }, + { + "epoch": 8.551401869158878, + "grad_norm": 0.1932913213968277, + "learning_rate": 5.6281701453073935e-06, + "loss": 0.0056, + "step": 5490 + }, + { + "epoch": 8.566978193146417, + "grad_norm": 0.09892082959413528, + "learning_rate": 5.510046935918739e-06, + "loss": 0.0049, + "step": 5500 + }, + { + "epoch": 8.582554517133957, + "grad_norm": 0.07008782029151917, + "learning_rate": 5.393104165303031e-06, + "loss": 0.0051, + "step": 5510 + }, + { + "epoch": 8.598130841121495, + "grad_norm": 0.11751464009284973, + "learning_rate": 5.277344936268258e-06, + "loss": 0.0057, + "step": 5520 + }, + { + "epoch": 8.613707165109034, + "grad_norm": 0.07049732655286789, + "learning_rate": 5.162772320219805e-06, + "loss": 0.006, + "step": 5530 + }, + { + "epoch": 8.629283489096574, + "grad_norm": 0.08728863298892975, + "learning_rate": 5.049389357079032e-06, + "loss": 0.0092, + "step": 5540 + }, + { + "epoch": 8.644859813084112, + "grad_norm": 0.22238792479038239, + "learning_rate": 4.93719905520259e-06, + "loss": 0.0063, + "step": 5550 + }, + { + "epoch": 8.66043613707165, + "grad_norm": 0.10668980330228806, + "learning_rate": 4.82620439130258e-06, + "loss": 0.0073, + "step": 5560 + }, + { + "epoch": 8.67601246105919, + "grad_norm": 0.17555032670497894, + "learning_rate": 4.71640831036762e-06, + "loss": 0.0069, + "step": 5570 + }, + { + "epoch": 8.69158878504673, + "grad_norm": 0.11057223379611969, + "learning_rate": 4.6078137255846584e-06, + "loss": 0.0065, + "step": 5580 + }, + { + "epoch": 8.707165109034268, + "grad_norm": 0.14288246631622314, + "learning_rate": 4.500423518261715e-06, + "loss": 0.009, + "step": 5590 + }, + { + "epoch": 8.722741433021806, + "grad_norm": 0.10396552085876465, + "learning_rate": 4.394240537751415e-06, + "loss": 0.006, + "step": 5600 + }, + { + "epoch": 8.738317757009346, + "grad_norm": 0.04296548292040825, + "learning_rate": 4.289267601375396e-06, + "loss": 0.0055, + "step": 5610 + }, + { + "epoch": 8.753894080996885, + "grad_norm": 0.11966723948717117, + "learning_rate": 4.18550749434955e-06, + "loss": 0.0076, + "step": 5620 + }, + { + "epoch": 8.769470404984423, + "grad_norm": 0.08584234863519669, + "learning_rate": 4.082962969710114e-06, + "loss": 0.009, + "step": 5630 + }, + { + "epoch": 8.785046728971963, + "grad_norm": 0.09861597418785095, + "learning_rate": 3.981636748240647e-06, + "loss": 0.0056, + "step": 5640 + }, + { + "epoch": 8.800623052959502, + "grad_norm": 0.10276372730731964, + "learning_rate": 3.881531518399851e-06, + "loss": 0.0076, + "step": 5650 + }, + { + "epoch": 8.81619937694704, + "grad_norm": 0.10706163197755814, + "learning_rate": 3.7826499362501656e-06, + "loss": 0.006, + "step": 5660 + }, + { + "epoch": 8.83177570093458, + "grad_norm": 0.09831231832504272, + "learning_rate": 3.6849946253874055e-06, + "loss": 0.0065, + "step": 5670 + }, + { + "epoch": 8.847352024922118, + "grad_norm": 0.2387414127588272, + "learning_rate": 3.588568176871049e-06, + "loss": 0.0092, + "step": 5680 + }, + { + "epoch": 8.862928348909657, + "grad_norm": 0.10069181025028229, + "learning_rate": 3.49337314915556e-06, + "loss": 0.0047, + "step": 5690 + }, + { + "epoch": 8.878504672897197, + "grad_norm": 0.054378148168325424, + "learning_rate": 3.3994120680224563e-06, + "loss": 0.0055, + "step": 5700 + }, + { + "epoch": 8.894080996884735, + "grad_norm": 0.08076715469360352, + "learning_rate": 3.3066874265133443e-06, + "loss": 0.0059, + "step": 5710 + }, + { + "epoch": 8.909657320872274, + "grad_norm": 0.093848317861557, + "learning_rate": 3.2152016848637224e-06, + "loss": 0.0047, + "step": 5720 + }, + { + "epoch": 8.925233644859812, + "grad_norm": 0.07754657417535782, + "learning_rate": 3.124957270437723e-06, + "loss": 0.0075, + "step": 5730 + }, + { + "epoch": 8.940809968847352, + "grad_norm": 0.06180989369750023, + "learning_rate": 3.0359565776637233e-06, + "loss": 0.009, + "step": 5740 + }, + { + "epoch": 8.95638629283489, + "grad_norm": 0.10458532720804214, + "learning_rate": 2.9482019679707815e-06, + "loss": 0.0067, + "step": 5750 + }, + { + "epoch": 8.97196261682243, + "grad_norm": 0.07835295796394348, + "learning_rate": 2.8616957697260303e-06, + "loss": 0.0053, + "step": 5760 + }, + { + "epoch": 8.98753894080997, + "grad_norm": 0.12571848928928375, + "learning_rate": 2.7764402781728595e-06, + "loss": 0.0062, + "step": 5770 + }, + { + "epoch": 9.003115264797508, + "grad_norm": 0.14288344979286194, + "learning_rate": 2.692437755370003e-06, + "loss": 0.0067, + "step": 5780 + }, + { + "epoch": 9.018691588785046, + "grad_norm": 0.07175132632255554, + "learning_rate": 2.609690430131584e-06, + "loss": 0.0059, + "step": 5790 + }, + { + "epoch": 9.034267912772586, + "grad_norm": 0.16830457746982574, + "learning_rate": 2.528200497967914e-06, + "loss": 0.0082, + "step": 5800 + }, + { + "epoch": 9.049844236760125, + "grad_norm": 0.10765346884727478, + "learning_rate": 2.447970121027282e-06, + "loss": 0.0055, + "step": 5810 + }, + { + "epoch": 9.065420560747663, + "grad_norm": 0.10951469093561172, + "learning_rate": 2.369001428038559e-06, + "loss": 0.008, + "step": 5820 + }, + { + "epoch": 9.080996884735203, + "grad_norm": 0.0858723595738411, + "learning_rate": 2.291296514254726e-06, + "loss": 0.005, + "step": 5830 + }, + { + "epoch": 9.096573208722742, + "grad_norm": 0.10129673779010773, + "learning_rate": 2.2148574413973e-06, + "loss": 0.0065, + "step": 5840 + }, + { + "epoch": 9.11214953271028, + "grad_norm": 0.1688499003648758, + "learning_rate": 2.1396862376015903e-06, + "loss": 0.0076, + "step": 5850 + }, + { + "epoch": 9.12772585669782, + "grad_norm": 0.0803821012377739, + "learning_rate": 2.065784897362938e-06, + "loss": 0.0069, + "step": 5860 + }, + { + "epoch": 9.143302180685358, + "grad_norm": 0.08840493857860565, + "learning_rate": 1.9931553814837635e-06, + "loss": 0.008, + "step": 5870 + }, + { + "epoch": 9.158878504672897, + "grad_norm": 0.08763741701841354, + "learning_rate": 1.921799617021519e-06, + "loss": 0.0076, + "step": 5880 + }, + { + "epoch": 9.174454828660435, + "grad_norm": 0.06915774941444397, + "learning_rate": 1.8517194972376305e-06, + "loss": 0.0074, + "step": 5890 + }, + { + "epoch": 9.190031152647975, + "grad_norm": 0.08171873539686203, + "learning_rate": 1.7829168815471863e-06, + "loss": 0.0064, + "step": 5900 + }, + { + "epoch": 9.205607476635514, + "grad_norm": 0.08857376873493195, + "learning_rate": 1.715393595469661e-06, + "loss": 0.0053, + "step": 5910 + }, + { + "epoch": 9.221183800623052, + "grad_norm": 0.19674229621887207, + "learning_rate": 1.6491514305804423e-06, + "loss": 0.0069, + "step": 5920 + }, + { + "epoch": 9.236760124610592, + "grad_norm": 0.05321294069290161, + "learning_rate": 1.584192144463298e-06, + "loss": 0.0066, + "step": 5930 + }, + { + "epoch": 9.25233644859813, + "grad_norm": 0.14856597781181335, + "learning_rate": 1.5205174606637783e-06, + "loss": 0.0064, + "step": 5940 + }, + { + "epoch": 9.26791277258567, + "grad_norm": 0.11522695422172546, + "learning_rate": 1.4581290686434378e-06, + "loss": 0.0094, + "step": 5950 + }, + { + "epoch": 9.28348909657321, + "grad_norm": 0.06241914629936218, + "learning_rate": 1.397028623735036e-06, + "loss": 0.005, + "step": 5960 + }, + { + "epoch": 9.299065420560748, + "grad_norm": 0.17963765561580658, + "learning_rate": 1.3372177470986236e-06, + "loss": 0.0058, + "step": 5970 + }, + { + "epoch": 9.314641744548286, + "grad_norm": 0.16115325689315796, + "learning_rate": 1.2786980256785042e-06, + "loss": 0.0113, + "step": 5980 + }, + { + "epoch": 9.330218068535826, + "grad_norm": 0.16287362575531006, + "learning_rate": 1.221471012161146e-06, + "loss": 0.0082, + "step": 5990 + }, + { + "epoch": 9.345794392523365, + "grad_norm": 0.07424895465373993, + "learning_rate": 1.1655382249339875e-06, + "loss": 0.0068, + "step": 6000 + }, + { + "epoch": 9.361370716510903, + "grad_norm": 0.0644395723938942, + "learning_rate": 1.1109011480451347e-06, + "loss": 0.0073, + "step": 6010 + }, + { + "epoch": 9.376947040498443, + "grad_norm": 0.10327418893575668, + "learning_rate": 1.0575612311640003e-06, + "loss": 0.0062, + "step": 6020 + }, + { + "epoch": 9.392523364485982, + "grad_norm": 0.09636327624320984, + "learning_rate": 1.0055198895428274e-06, + "loss": 0.0074, + "step": 6030 + }, + { + "epoch": 9.40809968847352, + "grad_norm": 0.06338019669055939, + "learning_rate": 9.547785039791536e-07, + "loss": 0.0057, + "step": 6040 + }, + { + "epoch": 9.423676012461058, + "grad_norm": 0.12510456144809723, + "learning_rate": 9.053384207791504e-07, + "loss": 0.0071, + "step": 6050 + }, + { + "epoch": 9.439252336448599, + "grad_norm": 0.09605526179075241, + "learning_rate": 8.572009517219426e-07, + "loss": 0.0056, + "step": 6060 + }, + { + "epoch": 9.454828660436137, + "grad_norm": 0.0817190483212471, + "learning_rate": 8.10367374024773e-07, + "loss": 0.0057, + "step": 6070 + }, + { + "epoch": 9.470404984423675, + "grad_norm": 0.08908645808696747, + "learning_rate": 7.648389303090919e-07, + "loss": 0.0099, + "step": 6080 + }, + { + "epoch": 9.485981308411215, + "grad_norm": 0.21740996837615967, + "learning_rate": 7.206168285676551e-07, + "loss": 0.0071, + "step": 6090 + }, + { + "epoch": 9.501557632398754, + "grad_norm": 0.12292931228876114, + "learning_rate": 6.777022421324108e-07, + "loss": 0.0073, + "step": 6100 + }, + { + "epoch": 9.517133956386292, + "grad_norm": 0.2714439332485199, + "learning_rate": 6.360963096434025e-07, + "loss": 0.0052, + "step": 6110 + }, + { + "epoch": 9.532710280373832, + "grad_norm": 0.08194523304700851, + "learning_rate": 5.958001350185205e-07, + "loss": 0.009, + "step": 6120 + }, + { + "epoch": 9.54828660436137, + "grad_norm": 0.10809244960546494, + "learning_rate": 5.568147874242646e-07, + "loss": 0.0076, + "step": 6130 + }, + { + "epoch": 9.56386292834891, + "grad_norm": 0.07969952374696732, + "learning_rate": 5.191413012473445e-07, + "loss": 0.0079, + "step": 6140 + }, + { + "epoch": 9.57943925233645, + "grad_norm": 0.15038545429706573, + "learning_rate": 4.827806760672183e-07, + "loss": 0.0054, + "step": 6150 + }, + { + "epoch": 9.595015576323988, + "grad_norm": 0.12301000207662582, + "learning_rate": 4.4773387662961354e-07, + "loss": 0.0068, + "step": 6160 + }, + { + "epoch": 9.610591900311526, + "grad_norm": 0.13655084371566772, + "learning_rate": 4.140018328209039e-07, + "loss": 0.0081, + "step": 6170 + }, + { + "epoch": 9.626168224299064, + "grad_norm": 0.07633527368307114, + "learning_rate": 3.8158543964342263e-07, + "loss": 0.0052, + "step": 6180 + }, + { + "epoch": 9.641744548286605, + "grad_norm": 0.1401819884777069, + "learning_rate": 3.504855571917598e-07, + "loss": 0.0056, + "step": 6190 + }, + { + "epoch": 9.657320872274143, + "grad_norm": 0.12164486199617386, + "learning_rate": 3.207030106298969e-07, + "loss": 0.0062, + "step": 6200 + }, + { + "epoch": 9.672897196261681, + "grad_norm": 0.15020953118801117, + "learning_rate": 2.922385901693414e-07, + "loss": 0.006, + "step": 6210 + }, + { + "epoch": 9.688473520249222, + "grad_norm": 0.10112406313419342, + "learning_rate": 2.650930510481375e-07, + "loss": 0.0057, + "step": 6220 + }, + { + "epoch": 9.70404984423676, + "grad_norm": 0.13086533546447754, + "learning_rate": 2.3926711351086596e-07, + "loss": 0.0075, + "step": 6230 + }, + { + "epoch": 9.719626168224298, + "grad_norm": 0.10105488449335098, + "learning_rate": 2.1476146278948117e-07, + "loss": 0.0053, + "step": 6240 + }, + { + "epoch": 9.735202492211839, + "grad_norm": 0.1257752776145935, + "learning_rate": 1.9157674908517608e-07, + "loss": 0.0078, + "step": 6250 + }, + { + "epoch": 9.750778816199377, + "grad_norm": 0.1095588207244873, + "learning_rate": 1.697135875510958e-07, + "loss": 0.0054, + "step": 6260 + }, + { + "epoch": 9.766355140186915, + "grad_norm": 0.14588817954063416, + "learning_rate": 1.4917255827605613e-07, + "loss": 0.006, + "step": 6270 + }, + { + "epoch": 9.781931464174455, + "grad_norm": 0.07080316543579102, + "learning_rate": 1.2995420626910616e-07, + "loss": 0.0074, + "step": 6280 + }, + { + "epoch": 9.797507788161994, + "grad_norm": 0.14342112839221954, + "learning_rate": 1.120590414451006e-07, + "loss": 0.0071, + "step": 6290 + }, + { + "epoch": 9.813084112149532, + "grad_norm": 0.10840759426355362, + "learning_rate": 9.548753861115533e-08, + "loss": 0.0068, + "step": 6300 + }, + { + "epoch": 9.828660436137072, + "grad_norm": 0.11529383808374405, + "learning_rate": 8.02401374540629e-08, + "loss": 0.0082, + "step": 6310 + }, + { + "epoch": 9.84423676012461, + "grad_norm": 0.1757618486881256, + "learning_rate": 6.631724252860183e-08, + "loss": 0.0073, + "step": 6320 + }, + { + "epoch": 9.85981308411215, + "grad_norm": 0.08475689589977264, + "learning_rate": 5.37192232468342e-08, + "loss": 0.0072, + "step": 6330 + }, + { + "epoch": 9.87538940809969, + "grad_norm": 0.11995267122983932, + "learning_rate": 4.244641386828008e-08, + "loss": 0.0066, + "step": 6340 + }, + { + "epoch": 9.890965732087228, + "grad_norm": 0.11214066296815872, + "learning_rate": 3.2499113491052394e-08, + "loss": 0.0085, + "step": 6350 + }, + { + "epoch": 9.906542056074766, + "grad_norm": 0.2757103741168976, + "learning_rate": 2.387758604394108e-08, + "loss": 0.0067, + "step": 6360 + }, + { + "epoch": 9.922118380062305, + "grad_norm": 0.08542575687170029, + "learning_rate": 1.6582060279379807e-08, + "loss": 0.0053, + "step": 6370 + }, + { + "epoch": 9.937694704049845, + "grad_norm": 0.08663801848888397, + "learning_rate": 1.0612729767400798e-08, + "loss": 0.0052, + "step": 6380 + }, + { + "epoch": 9.953271028037383, + "grad_norm": 0.11144966632127762, + "learning_rate": 5.969752890483404e-09, + "loss": 0.0059, + "step": 6390 + }, + { + "epoch": 9.968847352024921, + "grad_norm": 0.05011506378650665, + "learning_rate": 2.6532528393574676e-09, + "loss": 0.0048, + "step": 6400 + }, + { + "epoch": 9.984423676012462, + "grad_norm": 0.08475054800510406, + "learning_rate": 6.633176097392558e-10, + "loss": 0.0056, + "step": 6410 + }, + { + "epoch": 10.0, + "grad_norm": 0.11929858475923538, + "learning_rate": 0.0, + "loss": 0.008, + "step": 6420 + }, + { + "epoch": 10.0, + "step": 6420, + "total_flos": 6.893756797515905e+17, + "train_loss": 0.017978323890459127, + "train_runtime": 6212.5327, + "train_samples_per_second": 50.627, + "train_steps_per_second": 1.033 + } + ], + "logging_steps": 10, + "max_steps": 6420, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 10000, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 6.893756797515905e+17, + "train_batch_size": 49, + "trial_name": null, + "trial_params": null +}