{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.3877787807814457e-13, "eval_steps": 500, "global_step": 40000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 3.469446951953614e-17, "grad_norm": 2.059882164001465, "learning_rate": 1.25e-06, "loss": 10.4676, "step": 10 }, { "epoch": 6.938893903907228e-17, "grad_norm": 2.0199081897735596, "learning_rate": 2.5e-06, "loss": 10.4296, "step": 20 }, { "epoch": 1.0408340855860843e-16, "grad_norm": 2.84952449798584, "learning_rate": 3.75e-06, "loss": 10.3519, "step": 30 }, { "epoch": 1.3877787807814457e-16, "grad_norm": 2.7102675437927246, "learning_rate": 5e-06, "loss": 10.2176, "step": 40 }, { "epoch": 1.734723475976807e-16, "grad_norm": 2.003204822540283, "learning_rate": 6.25e-06, "loss": 10.0326, "step": 50 }, { "epoch": 2.0816681711721685e-16, "grad_norm": 1.6600241661071777, "learning_rate": 7.5e-06, "loss": 9.8876, "step": 60 }, { "epoch": 2.42861286636753e-16, "grad_norm": 1.5886650085449219, "learning_rate": 8.750000000000001e-06, "loss": 9.7683, "step": 70 }, { "epoch": 2.7755575615628914e-16, "grad_norm": 1.5405967235565186, "learning_rate": 1e-05, "loss": 9.6956, "step": 80 }, { "epoch": 3.122502256758253e-16, "grad_norm": 1.5582334995269775, "learning_rate": 1.1249999999999999e-05, "loss": 9.6354, "step": 90 }, { "epoch": 3.469446951953614e-16, "grad_norm": 1.5802150964736938, "learning_rate": 1.25e-05, "loss": 9.5755, "step": 100 }, { "epoch": 3.8163916471489756e-16, "grad_norm": 1.5428049564361572, "learning_rate": 1.375e-05, "loss": 9.503, "step": 110 }, { "epoch": 4.163336342344337e-16, "grad_norm": 1.3599138259887695, "learning_rate": 1.5e-05, "loss": 9.4323, "step": 120 }, { "epoch": 4.510281037539698e-16, "grad_norm": 1.5508506298065186, "learning_rate": 1.6250000000000002e-05, "loss": 9.3374, "step": 130 }, { "epoch": 4.85722573273506e-16, "grad_norm": 1.5241225957870483, "learning_rate": 1.7500000000000002e-05, "loss": 9.258, "step": 140 }, { "epoch": 5.204170427930421e-16, "grad_norm": 1.4980995655059814, "learning_rate": 1.875e-05, "loss": 9.1478, "step": 150 }, { "epoch": 5.551115123125783e-16, "grad_norm": 1.3956459760665894, "learning_rate": 2e-05, "loss": 8.995, "step": 160 }, { "epoch": 5.898059818321144e-16, "grad_norm": 1.5634582042694092, "learning_rate": 2.125e-05, "loss": 8.8657, "step": 170 }, { "epoch": 6.245004513516506e-16, "grad_norm": 1.3934783935546875, "learning_rate": 2.2499999999999998e-05, "loss": 8.7483, "step": 180 }, { "epoch": 6.591949208711867e-16, "grad_norm": 1.4147783517837524, "learning_rate": 2.375e-05, "loss": 8.6278, "step": 190 }, { "epoch": 6.938893903907228e-16, "grad_norm": 1.3870561122894287, "learning_rate": 2.5e-05, "loss": 8.5143, "step": 200 }, { "epoch": 7.28583859910259e-16, "grad_norm": 1.2880840301513672, "learning_rate": 2.625e-05, "loss": 8.4048, "step": 210 }, { "epoch": 7.632783294297951e-16, "grad_norm": 1.858474850654602, "learning_rate": 2.75e-05, "loss": 8.2866, "step": 220 }, { "epoch": 7.979727989493313e-16, "grad_norm": 1.286628007888794, "learning_rate": 2.875e-05, "loss": 8.1876, "step": 230 }, { "epoch": 8.326672684688674e-16, "grad_norm": 1.2863655090332031, "learning_rate": 3e-05, "loss": 8.069, "step": 240 }, { "epoch": 8.673617379884035e-16, "grad_norm": 1.1988719701766968, "learning_rate": 3.125e-05, "loss": 7.9953, "step": 250 }, { "epoch": 9.020562075079397e-16, "grad_norm": 1.163584589958191, "learning_rate": 3.2500000000000004e-05, "loss": 7.8655, "step": 260 }, { "epoch": 9.367506770274758e-16, "grad_norm": 1.11767578125, "learning_rate": 3.375e-05, "loss": 7.7913, "step": 270 }, { "epoch": 9.71445146547012e-16, "grad_norm": 1.6332619190216064, "learning_rate": 3.5000000000000004e-05, "loss": 7.696, "step": 280 }, { "epoch": 1.0061396160665481e-15, "grad_norm": 1.3233344554901123, "learning_rate": 3.625e-05, "loss": 7.5894, "step": 290 }, { "epoch": 1.0408340855860843e-15, "grad_norm": 1.1899006366729736, "learning_rate": 3.75e-05, "loss": 7.4906, "step": 300 }, { "epoch": 1.0755285551056204e-15, "grad_norm": 2.0172367095947266, "learning_rate": 3.875e-05, "loss": 7.4279, "step": 310 }, { "epoch": 1.1102230246251565e-15, "grad_norm": 1.7248649597167969, "learning_rate": 4e-05, "loss": 7.3457, "step": 320 }, { "epoch": 1.1449174941446927e-15, "grad_norm": 1.4723551273345947, "learning_rate": 4.125e-05, "loss": 7.3005, "step": 330 }, { "epoch": 1.1796119636642288e-15, "grad_norm": 1.8123193979263306, "learning_rate": 4.25e-05, "loss": 7.2081, "step": 340 }, { "epoch": 1.214306433183765e-15, "grad_norm": 1.6105283498764038, "learning_rate": 4.375e-05, "loss": 7.1459, "step": 350 }, { "epoch": 1.2490009027033011e-15, "grad_norm": 1.5732420682907104, "learning_rate": 4.4999999999999996e-05, "loss": 7.0739, "step": 360 }, { "epoch": 1.2836953722228372e-15, "grad_norm": 1.1241536140441895, "learning_rate": 4.625e-05, "loss": 7.0124, "step": 370 }, { "epoch": 1.3183898417423734e-15, "grad_norm": 1.6548717021942139, "learning_rate": 4.75e-05, "loss": 6.9543, "step": 380 }, { "epoch": 1.3530843112619095e-15, "grad_norm": 2.06797194480896, "learning_rate": 4.8750000000000006e-05, "loss": 6.9175, "step": 390 }, { "epoch": 1.3877787807814457e-15, "grad_norm": 1.65916907787323, "learning_rate": 5e-05, "loss": 6.8333, "step": 400 }, { "epoch": 1.4224732503009818e-15, "grad_norm": 1.4789600372314453, "learning_rate": 5.125e-05, "loss": 6.7854, "step": 410 }, { "epoch": 1.457167719820518e-15, "grad_norm": 1.7124913930892944, "learning_rate": 5.25e-05, "loss": 6.7137, "step": 420 }, { "epoch": 1.4918621893400541e-15, "grad_norm": 1.8768134117126465, "learning_rate": 5.375e-05, "loss": 6.678, "step": 430 }, { "epoch": 1.5265566588595902e-15, "grad_norm": 1.6320363283157349, "learning_rate": 5.5e-05, "loss": 6.6103, "step": 440 }, { "epoch": 1.5612511283791264e-15, "grad_norm": 1.8146885633468628, "learning_rate": 5.6250000000000005e-05, "loss": 6.5636, "step": 450 }, { "epoch": 1.5959455978986625e-15, "grad_norm": 3.034447193145752, "learning_rate": 5.75e-05, "loss": 6.4887, "step": 460 }, { "epoch": 1.6306400674181987e-15, "grad_norm": 1.5114161968231201, "learning_rate": 5.875e-05, "loss": 6.4869, "step": 470 }, { "epoch": 1.6653345369377348e-15, "grad_norm": 1.404179334640503, "learning_rate": 6e-05, "loss": 6.4174, "step": 480 }, { "epoch": 1.700029006457271e-15, "grad_norm": 1.6470967531204224, "learning_rate": 6.125e-05, "loss": 6.3717, "step": 490 }, { "epoch": 1.734723475976807e-15, "grad_norm": 1.8696486949920654, "learning_rate": 6.25e-05, "loss": 6.3303, "step": 500 }, { "epoch": 1.7694179454963432e-15, "grad_norm": 1.7590429782867432, "learning_rate": 6.375e-05, "loss": 6.2535, "step": 510 }, { "epoch": 1.8041124150158794e-15, "grad_norm": 1.7268675565719604, "learning_rate": 6.500000000000001e-05, "loss": 6.246, "step": 520 }, { "epoch": 1.8388068845354155e-15, "grad_norm": 1.4760897159576416, "learning_rate": 6.625000000000001e-05, "loss": 6.2175, "step": 530 }, { "epoch": 1.8735013540549517e-15, "grad_norm": 1.8681708574295044, "learning_rate": 6.75e-05, "loss": 6.1467, "step": 540 }, { "epoch": 1.908195823574488e-15, "grad_norm": 1.4182546138763428, "learning_rate": 6.875e-05, "loss": 6.1119, "step": 550 }, { "epoch": 1.942890293094024e-15, "grad_norm": 2.5084240436553955, "learning_rate": 7.000000000000001e-05, "loss": 6.0749, "step": 560 }, { "epoch": 1.97758476261356e-15, "grad_norm": 1.5843790769577026, "learning_rate": 7.125e-05, "loss": 6.0533, "step": 570 }, { "epoch": 2.0122792321330962e-15, "grad_norm": 2.0986595153808594, "learning_rate": 7.25e-05, "loss": 5.9555, "step": 580 }, { "epoch": 2.0469737016526324e-15, "grad_norm": 1.3557995557785034, "learning_rate": 7.375e-05, "loss": 6.0013, "step": 590 }, { "epoch": 2.0816681711721685e-15, "grad_norm": 2.205937385559082, "learning_rate": 7.5e-05, "loss": 5.9732, "step": 600 }, { "epoch": 2.1163626406917047e-15, "grad_norm": 1.8265631198883057, "learning_rate": 7.625e-05, "loss": 5.921, "step": 610 }, { "epoch": 2.1510571102112408e-15, "grad_norm": 1.903822660446167, "learning_rate": 7.75e-05, "loss": 5.9057, "step": 620 }, { "epoch": 2.185751579730777e-15, "grad_norm": 1.9713225364685059, "learning_rate": 7.875e-05, "loss": 5.8418, "step": 630 }, { "epoch": 2.220446049250313e-15, "grad_norm": 2.186734437942505, "learning_rate": 8e-05, "loss": 5.794, "step": 640 }, { "epoch": 2.2551405187698492e-15, "grad_norm": 1.5751311779022217, "learning_rate": 8.125000000000001e-05, "loss": 5.7945, "step": 650 }, { "epoch": 2.2898349882893854e-15, "grad_norm": 1.9784258604049683, "learning_rate": 8.25e-05, "loss": 5.7716, "step": 660 }, { "epoch": 2.3245294578089215e-15, "grad_norm": 2.25524640083313, "learning_rate": 8.375e-05, "loss": 5.7332, "step": 670 }, { "epoch": 2.3592239273284576e-15, "grad_norm": 1.9643704891204834, "learning_rate": 8.5e-05, "loss": 5.6788, "step": 680 }, { "epoch": 2.3939183968479938e-15, "grad_norm": 1.5461243391036987, "learning_rate": 8.625e-05, "loss": 5.7208, "step": 690 }, { "epoch": 2.42861286636753e-15, "grad_norm": 2.2134976387023926, "learning_rate": 8.75e-05, "loss": 5.666, "step": 700 }, { "epoch": 2.463307335887066e-15, "grad_norm": 2.60357666015625, "learning_rate": 8.875e-05, "loss": 5.6247, "step": 710 }, { "epoch": 2.4980018054066022e-15, "grad_norm": 1.6550157070159912, "learning_rate": 8.999999999999999e-05, "loss": 5.6663, "step": 720 }, { "epoch": 2.5326962749261384e-15, "grad_norm": 1.5992182493209839, "learning_rate": 9.125e-05, "loss": 5.613, "step": 730 }, { "epoch": 2.5673907444456745e-15, "grad_norm": 1.9118322134017944, "learning_rate": 9.25e-05, "loss": 5.5903, "step": 740 }, { "epoch": 2.6020852139652106e-15, "grad_norm": 1.9484632015228271, "learning_rate": 9.375e-05, "loss": 5.5604, "step": 750 }, { "epoch": 2.6367796834847468e-15, "grad_norm": 3.3100740909576416, "learning_rate": 9.5e-05, "loss": 5.5074, "step": 760 }, { "epoch": 2.671474153004283e-15, "grad_norm": 1.7070995569229126, "learning_rate": 9.625000000000001e-05, "loss": 5.53, "step": 770 }, { "epoch": 2.706168622523819e-15, "grad_norm": 1.4375962018966675, "learning_rate": 9.750000000000001e-05, "loss": 5.4964, "step": 780 }, { "epoch": 2.7408630920433552e-15, "grad_norm": 1.6255340576171875, "learning_rate": 9.875e-05, "loss": 5.4764, "step": 790 }, { "epoch": 2.7755575615628914e-15, "grad_norm": 1.7709786891937256, "learning_rate": 0.0001, "loss": 5.4533, "step": 800 }, { "epoch": 2.8102520310824275e-15, "grad_norm": 1.7416393756866455, "learning_rate": 0.00010125000000000001, "loss": 5.4423, "step": 810 }, { "epoch": 2.8449465006019636e-15, "grad_norm": 2.6547224521636963, "learning_rate": 0.0001025, "loss": 5.4091, "step": 820 }, { "epoch": 2.8796409701214998e-15, "grad_norm": 1.6572755575180054, "learning_rate": 0.00010375, "loss": 5.3704, "step": 830 }, { "epoch": 2.914335439641036e-15, "grad_norm": 1.5294963121414185, "learning_rate": 0.000105, "loss": 5.3569, "step": 840 }, { "epoch": 2.949029909160572e-15, "grad_norm": 1.5555777549743652, "learning_rate": 0.00010625, "loss": 5.3561, "step": 850 }, { "epoch": 2.9837243786801082e-15, "grad_norm": 1.7305047512054443, "learning_rate": 0.0001075, "loss": 5.3702, "step": 860 }, { "epoch": 3.0184188481996443e-15, "grad_norm": 1.611474871635437, "learning_rate": 0.00010875, "loss": 5.2871, "step": 870 }, { "epoch": 3.0531133177191805e-15, "grad_norm": 1.4751167297363281, "learning_rate": 0.00011, "loss": 5.2821, "step": 880 }, { "epoch": 3.0878077872387166e-15, "grad_norm": 1.9502071142196655, "learning_rate": 0.00011125000000000001, "loss": 5.2779, "step": 890 }, { "epoch": 3.1225022567582528e-15, "grad_norm": 1.7957355976104736, "learning_rate": 0.00011250000000000001, "loss": 5.2163, "step": 900 }, { "epoch": 3.157196726277789e-15, "grad_norm": 1.678847074508667, "learning_rate": 0.00011375, "loss": 5.2114, "step": 910 }, { "epoch": 3.191891195797325e-15, "grad_norm": 1.5330493450164795, "learning_rate": 0.000115, "loss": 5.249, "step": 920 }, { "epoch": 3.226585665316861e-15, "grad_norm": 1.8024704456329346, "learning_rate": 0.00011625, "loss": 5.1994, "step": 930 }, { "epoch": 3.2612801348363973e-15, "grad_norm": 1.4165303707122803, "learning_rate": 0.0001175, "loss": 5.174, "step": 940 }, { "epoch": 3.2959746043559335e-15, "grad_norm": 1.6327879428863525, "learning_rate": 0.00011875, "loss": 5.1543, "step": 950 }, { "epoch": 3.3306690738754696e-15, "grad_norm": 1.6701730489730835, "learning_rate": 0.00012, "loss": 5.1493, "step": 960 }, { "epoch": 3.3653635433950058e-15, "grad_norm": 1.427886962890625, "learning_rate": 0.00012124999999999999, "loss": 5.1112, "step": 970 }, { "epoch": 3.400058012914542e-15, "grad_norm": 1.8350648880004883, "learning_rate": 0.0001225, "loss": 5.1055, "step": 980 }, { "epoch": 3.434752482434078e-15, "grad_norm": 1.6441937685012817, "learning_rate": 0.00012375, "loss": 5.1197, "step": 990 }, { "epoch": 3.469446951953614e-15, "grad_norm": 1.337614893913269, "learning_rate": 0.000125, "loss": 5.0768, "step": 1000 }, { "epoch": 3.5041414214731503e-15, "grad_norm": 2.2115330696105957, "learning_rate": 0.00012499999555227336, "loss": 5.0724, "step": 1010 }, { "epoch": 3.5388358909926865e-15, "grad_norm": 1.6734768152236938, "learning_rate": 0.0001249999822090941, "loss": 5.0083, "step": 1020 }, { "epoch": 3.573530360512223e-15, "grad_norm": 1.8418147563934326, "learning_rate": 0.00012499995997046438, "loss": 4.9944, "step": 1030 }, { "epoch": 3.608224830031759e-15, "grad_norm": 1.372340440750122, "learning_rate": 0.0001249999288363877, "loss": 4.9754, "step": 1040 }, { "epoch": 3.642919299551295e-15, "grad_norm": 1.4691674709320068, "learning_rate": 0.00012499988880686897, "loss": 4.9704, "step": 1050 }, { "epoch": 3.677613769070831e-15, "grad_norm": 1.265261173248291, "learning_rate": 0.00012499983988191452, "loss": 4.9329, "step": 1060 }, { "epoch": 3.712308238590367e-15, "grad_norm": 1.665921926498413, "learning_rate": 0.00012499978206153208, "loss": 4.932, "step": 1070 }, { "epoch": 3.747002708109903e-15, "grad_norm": 1.2975447177886963, "learning_rate": 0.00012499971534573085, "loss": 4.8933, "step": 1080 }, { "epoch": 3.7816971776294395e-15, "grad_norm": 1.5446079969406128, "learning_rate": 0.0001249996397345213, "loss": 4.9345, "step": 1090 }, { "epoch": 3.816391647148976e-15, "grad_norm": 1.2354469299316406, "learning_rate": 0.00012499955522791544, "loss": 4.8401, "step": 1100 }, { "epoch": 3.851086116668512e-15, "grad_norm": 1.3809199333190918, "learning_rate": 0.00012499946182592664, "loss": 4.8644, "step": 1110 }, { "epoch": 3.885780586188048e-15, "grad_norm": 1.429549217224121, "learning_rate": 0.00012499935952856964, "loss": 4.8872, "step": 1120 }, { "epoch": 3.920475055707584e-15, "grad_norm": 1.8042439222335815, "learning_rate": 0.0001249992483358606, "loss": 4.8464, "step": 1130 }, { "epoch": 3.95516952522712e-15, "grad_norm": 1.9081182479858398, "learning_rate": 0.00012499912824781717, "loss": 4.8257, "step": 1140 }, { "epoch": 3.989863994746656e-15, "grad_norm": 1.1181987524032593, "learning_rate": 0.00012499899926445826, "loss": 4.8461, "step": 1150 }, { "epoch": 4.0245584642661925e-15, "grad_norm": 1.6167137622833252, "learning_rate": 0.00012499886138580434, "loss": 4.8056, "step": 1160 }, { "epoch": 4.0592529337857286e-15, "grad_norm": 1.4424612522125244, "learning_rate": 0.0001249987146118772, "loss": 4.7486, "step": 1170 }, { "epoch": 4.093947403305265e-15, "grad_norm": 1.3669044971466064, "learning_rate": 0.0001249985589427, "loss": 4.7285, "step": 1180 }, { "epoch": 4.128641872824801e-15, "grad_norm": 1.228793978691101, "learning_rate": 0.00012499839437829743, "loss": 4.7177, "step": 1190 }, { "epoch": 4.163336342344337e-15, "grad_norm": 1.4458509683609009, "learning_rate": 0.00012499822091869547, "loss": 4.7215, "step": 1200 }, { "epoch": 4.198030811863873e-15, "grad_norm": 1.2814877033233643, "learning_rate": 0.00012499803856392153, "loss": 4.713, "step": 1210 }, { "epoch": 4.232725281383409e-15, "grad_norm": 1.3616384267807007, "learning_rate": 0.0001249978473140045, "loss": 4.7023, "step": 1220 }, { "epoch": 4.2674197509029455e-15, "grad_norm": 1.2756009101867676, "learning_rate": 0.00012499764716897462, "loss": 4.628, "step": 1230 }, { "epoch": 4.3021142204224816e-15, "grad_norm": 1.4610651731491089, "learning_rate": 0.00012499743812886352, "loss": 4.6577, "step": 1240 }, { "epoch": 4.336808689942018e-15, "grad_norm": 1.10690176486969, "learning_rate": 0.00012499722019370424, "loss": 4.6738, "step": 1250 }, { "epoch": 4.371503159461554e-15, "grad_norm": 1.3380756378173828, "learning_rate": 0.00012499699336353126, "loss": 4.6506, "step": 1260 }, { "epoch": 4.40619762898109e-15, "grad_norm": 1.230385422706604, "learning_rate": 0.0001249967576383805, "loss": 4.6027, "step": 1270 }, { "epoch": 4.440892098500626e-15, "grad_norm": 1.4060776233673096, "learning_rate": 0.00012499651301828914, "loss": 4.5954, "step": 1280 }, { "epoch": 4.475586568020162e-15, "grad_norm": 1.2003498077392578, "learning_rate": 0.00012499625950329593, "loss": 4.6002, "step": 1290 }, { "epoch": 4.5102810375396984e-15, "grad_norm": 1.2301654815673828, "learning_rate": 0.00012499599709344097, "loss": 4.6161, "step": 1300 }, { "epoch": 4.5449755070592346e-15, "grad_norm": 1.3027172088623047, "learning_rate": 0.00012499572578876574, "loss": 4.5972, "step": 1310 }, { "epoch": 4.579669976578771e-15, "grad_norm": 1.301685094833374, "learning_rate": 0.00012499544558931315, "loss": 4.5269, "step": 1320 }, { "epoch": 4.614364446098307e-15, "grad_norm": 1.1482892036437988, "learning_rate": 0.0001249951564951275, "loss": 4.5355, "step": 1330 }, { "epoch": 4.649058915617843e-15, "grad_norm": 1.2399473190307617, "learning_rate": 0.00012499485850625447, "loss": 4.5606, "step": 1340 }, { "epoch": 4.683753385137379e-15, "grad_norm": 1.1606062650680542, "learning_rate": 0.00012499455162274126, "loss": 4.5053, "step": 1350 }, { "epoch": 4.718447854656915e-15, "grad_norm": 1.3789100646972656, "learning_rate": 0.0001249942358446364, "loss": 4.4779, "step": 1360 }, { "epoch": 4.7531423241764514e-15, "grad_norm": 1.2012298107147217, "learning_rate": 0.00012499391117198975, "loss": 4.4876, "step": 1370 }, { "epoch": 4.7878367936959876e-15, "grad_norm": 1.5306181907653809, "learning_rate": 0.0001249935776048527, "loss": 4.4854, "step": 1380 }, { "epoch": 4.822531263215524e-15, "grad_norm": 1.4287185668945312, "learning_rate": 0.000124993235143278, "loss": 4.4821, "step": 1390 }, { "epoch": 4.85722573273506e-15, "grad_norm": 1.1439909934997559, "learning_rate": 0.00012499288378731982, "loss": 4.4563, "step": 1400 }, { "epoch": 4.891920202254596e-15, "grad_norm": 1.1890869140625, "learning_rate": 0.0001249925235370337, "loss": 4.4593, "step": 1410 }, { "epoch": 4.926614671774132e-15, "grad_norm": 1.1952084302902222, "learning_rate": 0.00012499215439247665, "loss": 4.4619, "step": 1420 }, { "epoch": 4.961309141293668e-15, "grad_norm": 1.3953648805618286, "learning_rate": 0.00012499177635370699, "loss": 4.4402, "step": 1430 }, { "epoch": 4.9960036108132044e-15, "grad_norm": 1.1769664287567139, "learning_rate": 0.00012499138942078454, "loss": 4.4175, "step": 1440 }, { "epoch": 5.0306980803327406e-15, "grad_norm": 1.1352949142456055, "learning_rate": 0.0001249909935937705, "loss": 4.357, "step": 1450 }, { "epoch": 5.065392549852277e-15, "grad_norm": 1.3775118589401245, "learning_rate": 0.00012499058887272742, "loss": 4.3855, "step": 1460 }, { "epoch": 5.100087019371813e-15, "grad_norm": 1.3279666900634766, "learning_rate": 0.00012499017525771934, "loss": 4.3738, "step": 1470 }, { "epoch": 5.134781488891349e-15, "grad_norm": 1.2900484800338745, "learning_rate": 0.00012498975274881165, "loss": 4.3762, "step": 1480 }, { "epoch": 5.169475958410885e-15, "grad_norm": 1.2940657138824463, "learning_rate": 0.0001249893213460712, "loss": 4.3587, "step": 1490 }, { "epoch": 5.204170427930421e-15, "grad_norm": 1.458011269569397, "learning_rate": 0.00012498888104956618, "loss": 4.3502, "step": 1500 }, { "epoch": 5.2388648974499574e-15, "grad_norm": 1.1659231185913086, "learning_rate": 0.00012498843185936625, "loss": 4.3701, "step": 1510 }, { "epoch": 5.2735593669694936e-15, "grad_norm": 1.1005889177322388, "learning_rate": 0.0001249879737755424, "loss": 4.3509, "step": 1520 }, { "epoch": 5.30825383648903e-15, "grad_norm": 1.3251738548278809, "learning_rate": 0.00012498750679816711, "loss": 4.3244, "step": 1530 }, { "epoch": 5.342948306008566e-15, "grad_norm": 1.2482818365097046, "learning_rate": 0.0001249870309273142, "loss": 4.3266, "step": 1540 }, { "epoch": 5.377642775528102e-15, "grad_norm": 1.2870382070541382, "learning_rate": 0.00012498654616305895, "loss": 4.2933, "step": 1550 }, { "epoch": 5.412337245047638e-15, "grad_norm": 1.0139496326446533, "learning_rate": 0.00012498605250547802, "loss": 4.3087, "step": 1560 }, { "epoch": 5.447031714567174e-15, "grad_norm": 1.1183644533157349, "learning_rate": 0.00012498554995464948, "loss": 4.2773, "step": 1570 }, { "epoch": 5.4817261840867104e-15, "grad_norm": 1.3766967058181763, "learning_rate": 0.00012498503851065277, "loss": 4.2515, "step": 1580 }, { "epoch": 5.5164206536062466e-15, "grad_norm": 1.2576528787612915, "learning_rate": 0.0001249845181735688, "loss": 4.2491, "step": 1590 }, { "epoch": 5.551115123125783e-15, "grad_norm": 1.172000527381897, "learning_rate": 0.00012498398894347985, "loss": 4.2878, "step": 1600 }, { "epoch": 5.585809592645319e-15, "grad_norm": 1.386810064315796, "learning_rate": 0.0001249834508204696, "loss": 4.2644, "step": 1610 }, { "epoch": 5.620504062164855e-15, "grad_norm": 1.1654918193817139, "learning_rate": 0.0001249829038046232, "loss": 4.2442, "step": 1620 }, { "epoch": 5.655198531684391e-15, "grad_norm": 1.3576298952102661, "learning_rate": 0.0001249823478960271, "loss": 4.196, "step": 1630 }, { "epoch": 5.689893001203927e-15, "grad_norm": 1.1857887506484985, "learning_rate": 0.00012498178309476923, "loss": 4.1786, "step": 1640 }, { "epoch": 5.7245874707234634e-15, "grad_norm": 1.069484829902649, "learning_rate": 0.00012498120940093892, "loss": 4.2418, "step": 1650 }, { "epoch": 5.7592819402429996e-15, "grad_norm": 1.0802791118621826, "learning_rate": 0.00012498062681462685, "loss": 4.2214, "step": 1660 }, { "epoch": 5.793976409762536e-15, "grad_norm": 1.0958795547485352, "learning_rate": 0.00012498003533592523, "loss": 4.1764, "step": 1670 }, { "epoch": 5.828670879282072e-15, "grad_norm": 1.0173470973968506, "learning_rate": 0.00012497943496492754, "loss": 4.227, "step": 1680 }, { "epoch": 5.863365348801608e-15, "grad_norm": 1.2646253108978271, "learning_rate": 0.00012497882570172873, "loss": 4.1982, "step": 1690 }, { "epoch": 5.898059818321144e-15, "grad_norm": 1.1745597124099731, "learning_rate": 0.00012497820754642515, "loss": 4.1515, "step": 1700 }, { "epoch": 5.93275428784068e-15, "grad_norm": 1.0884398221969604, "learning_rate": 0.00012497758049911456, "loss": 4.1671, "step": 1710 }, { "epoch": 5.9674487573602164e-15, "grad_norm": 1.1412599086761475, "learning_rate": 0.00012497694455989615, "loss": 4.155, "step": 1720 }, { "epoch": 6.0021432268797525e-15, "grad_norm": 1.1433255672454834, "learning_rate": 0.00012497629972887044, "loss": 4.1641, "step": 1730 }, { "epoch": 6.036837696399289e-15, "grad_norm": 1.1854442358016968, "learning_rate": 0.00012497564600613944, "loss": 4.1494, "step": 1740 }, { "epoch": 6.071532165918825e-15, "grad_norm": 1.4519822597503662, "learning_rate": 0.00012497498339180652, "loss": 4.1221, "step": 1750 }, { "epoch": 6.106226635438361e-15, "grad_norm": 1.35395085811615, "learning_rate": 0.00012497431188597646, "loss": 4.1453, "step": 1760 }, { "epoch": 6.140921104957897e-15, "grad_norm": 1.1102803945541382, "learning_rate": 0.00012497363148875545, "loss": 4.0972, "step": 1770 }, { "epoch": 6.175615574477433e-15, "grad_norm": 1.2988463640213013, "learning_rate": 0.00012497294220025108, "loss": 4.0983, "step": 1780 }, { "epoch": 6.2103100439969694e-15, "grad_norm": 0.983448326587677, "learning_rate": 0.0001249722440205724, "loss": 4.0682, "step": 1790 }, { "epoch": 6.2450045135165055e-15, "grad_norm": 1.0195692777633667, "learning_rate": 0.0001249715369498298, "loss": 4.0459, "step": 1800 }, { "epoch": 6.279698983036042e-15, "grad_norm": 1.127953290939331, "learning_rate": 0.00012497082098813505, "loss": 4.0654, "step": 1810 }, { "epoch": 6.314393452555578e-15, "grad_norm": 1.1295812129974365, "learning_rate": 0.00012497009613560145, "loss": 4.0835, "step": 1820 }, { "epoch": 6.349087922075114e-15, "grad_norm": 1.0090285539627075, "learning_rate": 0.00012496936239234357, "loss": 4.0332, "step": 1830 }, { "epoch": 6.38378239159465e-15, "grad_norm": 1.004581332206726, "learning_rate": 0.0001249686197584775, "loss": 3.9935, "step": 1840 }, { "epoch": 6.418476861114186e-15, "grad_norm": 1.0376604795455933, "learning_rate": 0.0001249678682341206, "loss": 4.0361, "step": 1850 }, { "epoch": 6.453171330633722e-15, "grad_norm": 1.1699234247207642, "learning_rate": 0.0001249671078193918, "loss": 4.0171, "step": 1860 }, { "epoch": 6.4878658001532585e-15, "grad_norm": 1.3391773700714111, "learning_rate": 0.0001249663385144113, "loss": 3.9618, "step": 1870 }, { "epoch": 6.522560269672795e-15, "grad_norm": 1.1559298038482666, "learning_rate": 0.0001249655603193008, "loss": 4.0009, "step": 1880 }, { "epoch": 6.557254739192331e-15, "grad_norm": 1.1633734703063965, "learning_rate": 0.00012496477323418334, "loss": 4.0034, "step": 1890 }, { "epoch": 6.591949208711867e-15, "grad_norm": 0.9894998669624329, "learning_rate": 0.00012496397725918337, "loss": 4.002, "step": 1900 }, { "epoch": 6.626643678231403e-15, "grad_norm": 0.9801426529884338, "learning_rate": 0.0001249631723944268, "loss": 3.9568, "step": 1910 }, { "epoch": 6.661338147750939e-15, "grad_norm": 1.0301562547683716, "learning_rate": 0.0001249623586400409, "loss": 3.9791, "step": 1920 }, { "epoch": 6.696032617270475e-15, "grad_norm": 1.1770894527435303, "learning_rate": 0.00012496153599615435, "loss": 3.9298, "step": 1930 }, { "epoch": 6.7307270867900115e-15, "grad_norm": 0.9789376854896545, "learning_rate": 0.00012496070446289727, "loss": 3.9238, "step": 1940 }, { "epoch": 6.765421556309548e-15, "grad_norm": 0.9871399402618408, "learning_rate": 0.00012495986404040117, "loss": 3.9246, "step": 1950 }, { "epoch": 6.800116025829084e-15, "grad_norm": 1.2827492952346802, "learning_rate": 0.0001249590147287989, "loss": 3.8912, "step": 1960 }, { "epoch": 6.83481049534862e-15, "grad_norm": 1.0770854949951172, "learning_rate": 0.0001249581565282248, "loss": 3.8792, "step": 1970 }, { "epoch": 6.869504964868156e-15, "grad_norm": 1.0476429462432861, "learning_rate": 0.00012495728943881458, "loss": 3.8788, "step": 1980 }, { "epoch": 6.904199434387692e-15, "grad_norm": 1.1102715730667114, "learning_rate": 0.00012495641346070538, "loss": 3.8513, "step": 1990 }, { "epoch": 6.938893903907228e-15, "grad_norm": 1.0551728010177612, "learning_rate": 0.00012495552859403573, "loss": 3.8296, "step": 2000 }, { "epoch": 6.9735883734267645e-15, "grad_norm": 0.929436445236206, "learning_rate": 0.00012495463483894556, "loss": 3.8433, "step": 2010 }, { "epoch": 7.008282842946301e-15, "grad_norm": 1.1489485502243042, "learning_rate": 0.0001249537321955762, "loss": 3.8707, "step": 2020 }, { "epoch": 7.042977312465837e-15, "grad_norm": 0.9174073934555054, "learning_rate": 0.00012495282066407037, "loss": 3.8331, "step": 2030 }, { "epoch": 7.077671781985373e-15, "grad_norm": 0.9909847974777222, "learning_rate": 0.00012495190024457226, "loss": 3.7932, "step": 2040 }, { "epoch": 7.112366251504909e-15, "grad_norm": 1.0692604780197144, "learning_rate": 0.00012495097093722743, "loss": 3.8332, "step": 2050 }, { "epoch": 7.147060721024445e-15, "grad_norm": 1.086592674255371, "learning_rate": 0.00012495003274218283, "loss": 3.8201, "step": 2060 }, { "epoch": 7.181755190543981e-15, "grad_norm": 1.133994698524475, "learning_rate": 0.0001249490856595868, "loss": 3.7603, "step": 2070 }, { "epoch": 7.216449660063518e-15, "grad_norm": 1.0277695655822754, "learning_rate": 0.00012494812968958917, "loss": 3.7709, "step": 2080 }, { "epoch": 7.251144129583054e-15, "grad_norm": 1.0337709188461304, "learning_rate": 0.00012494716483234106, "loss": 3.78, "step": 2090 }, { "epoch": 7.28583859910259e-15, "grad_norm": 1.2172855138778687, "learning_rate": 0.00012494619108799508, "loss": 3.7766, "step": 2100 }, { "epoch": 7.320533068622126e-15, "grad_norm": 1.0768190622329712, "learning_rate": 0.00012494520845670522, "loss": 3.7496, "step": 2110 }, { "epoch": 7.355227538141662e-15, "grad_norm": 1.0554888248443604, "learning_rate": 0.0001249442169386269, "loss": 3.7975, "step": 2120 }, { "epoch": 7.389922007661198e-15, "grad_norm": 1.0549849271774292, "learning_rate": 0.00012494321653391688, "loss": 3.7459, "step": 2130 }, { "epoch": 7.424616477180734e-15, "grad_norm": 0.9857177734375, "learning_rate": 0.00012494220724273337, "loss": 3.7636, "step": 2140 }, { "epoch": 7.45931094670027e-15, "grad_norm": 0.8990267515182495, "learning_rate": 0.00012494118906523598, "loss": 3.7326, "step": 2150 }, { "epoch": 7.494005416219807e-15, "grad_norm": 1.0241947174072266, "learning_rate": 0.00012494016200158576, "loss": 3.7196, "step": 2160 }, { "epoch": 7.528699885739343e-15, "grad_norm": 0.9795858263969421, "learning_rate": 0.0001249391260519451, "loss": 3.7181, "step": 2170 }, { "epoch": 7.563394355258879e-15, "grad_norm": 1.0870239734649658, "learning_rate": 0.00012493808121647785, "loss": 3.705, "step": 2180 }, { "epoch": 7.598088824778415e-15, "grad_norm": 0.9662448763847351, "learning_rate": 0.00012493702749534922, "loss": 3.6994, "step": 2190 }, { "epoch": 7.632783294297951e-15, "grad_norm": 0.9734577536582947, "learning_rate": 0.0001249359648887258, "loss": 3.6622, "step": 2200 }, { "epoch": 7.667477763817487e-15, "grad_norm": 0.9365831613540649, "learning_rate": 0.00012493489339677577, "loss": 3.707, "step": 2210 }, { "epoch": 7.702172233337023e-15, "grad_norm": 0.9854034185409546, "learning_rate": 0.00012493381301966844, "loss": 3.6758, "step": 2220 }, { "epoch": 7.73686670285656e-15, "grad_norm": 1.3100858926773071, "learning_rate": 0.00012493272375757472, "loss": 3.7149, "step": 2230 }, { "epoch": 7.771561172376096e-15, "grad_norm": 1.0683401823043823, "learning_rate": 0.00012493162561066687, "loss": 3.6953, "step": 2240 }, { "epoch": 7.806255641895632e-15, "grad_norm": 1.1088753938674927, "learning_rate": 0.00012493051857911854, "loss": 3.6704, "step": 2250 }, { "epoch": 7.840950111415168e-15, "grad_norm": 0.998305082321167, "learning_rate": 0.00012492940266310481, "loss": 3.669, "step": 2260 }, { "epoch": 7.875644580934704e-15, "grad_norm": 0.9200564026832581, "learning_rate": 0.00012492827786280214, "loss": 3.6216, "step": 2270 }, { "epoch": 7.91033905045424e-15, "grad_norm": 0.9278472661972046, "learning_rate": 0.0001249271441783884, "loss": 3.6529, "step": 2280 }, { "epoch": 7.945033519973776e-15, "grad_norm": 1.0532819032669067, "learning_rate": 0.0001249260016100429, "loss": 3.6178, "step": 2290 }, { "epoch": 7.979727989493313e-15, "grad_norm": 1.0741575956344604, "learning_rate": 0.00012492485015794632, "loss": 3.5803, "step": 2300 }, { "epoch": 8.014422459012849e-15, "grad_norm": 0.983441174030304, "learning_rate": 0.00012492368982228071, "loss": 3.6084, "step": 2310 }, { "epoch": 8.049116928532385e-15, "grad_norm": 0.9482371807098389, "learning_rate": 0.00012492252060322966, "loss": 3.5885, "step": 2320 }, { "epoch": 8.083811398051921e-15, "grad_norm": 0.8566581010818481, "learning_rate": 0.00012492134250097794, "loss": 3.6332, "step": 2330 }, { "epoch": 8.118505867571457e-15, "grad_norm": 0.873976469039917, "learning_rate": 0.00012492015551571198, "loss": 3.5845, "step": 2340 }, { "epoch": 8.153200337090993e-15, "grad_norm": 1.0078606605529785, "learning_rate": 0.0001249189596476194, "loss": 3.5859, "step": 2350 }, { "epoch": 8.18789480661053e-15, "grad_norm": 0.911682665348053, "learning_rate": 0.0001249177548968894, "loss": 3.6152, "step": 2360 }, { "epoch": 8.222589276130066e-15, "grad_norm": 0.9052351117134094, "learning_rate": 0.00012491654126371246, "loss": 3.573, "step": 2370 }, { "epoch": 8.257283745649602e-15, "grad_norm": 0.890880823135376, "learning_rate": 0.00012491531874828048, "loss": 3.5395, "step": 2380 }, { "epoch": 8.291978215169138e-15, "grad_norm": 0.8542453050613403, "learning_rate": 0.00012491408735078682, "loss": 3.5854, "step": 2390 }, { "epoch": 8.326672684688674e-15, "grad_norm": 0.9755571484565735, "learning_rate": 0.0001249128470714262, "loss": 3.5526, "step": 2400 }, { "epoch": 8.36136715420821e-15, "grad_norm": 0.8511834144592285, "learning_rate": 0.0001249115979103948, "loss": 3.5334, "step": 2410 }, { "epoch": 8.396061623727746e-15, "grad_norm": 1.0363075733184814, "learning_rate": 0.0001249103398678901, "loss": 3.5602, "step": 2420 }, { "epoch": 8.430756093247282e-15, "grad_norm": 0.9961519241333008, "learning_rate": 0.00012490907294411107, "loss": 3.5203, "step": 2430 }, { "epoch": 8.465450562766819e-15, "grad_norm": 1.0298234224319458, "learning_rate": 0.00012490779713925811, "loss": 3.555, "step": 2440 }, { "epoch": 8.500145032286355e-15, "grad_norm": 0.9939607381820679, "learning_rate": 0.00012490651245353294, "loss": 3.5251, "step": 2450 }, { "epoch": 8.534839501805891e-15, "grad_norm": 0.8517451882362366, "learning_rate": 0.00012490521888713872, "loss": 3.5475, "step": 2460 }, { "epoch": 8.569533971325427e-15, "grad_norm": 0.9422284364700317, "learning_rate": 0.00012490391644028, "loss": 3.5295, "step": 2470 }, { "epoch": 8.604228440844963e-15, "grad_norm": 0.9677397012710571, "learning_rate": 0.0001249026051131628, "loss": 3.5413, "step": 2480 }, { "epoch": 8.6389229103645e-15, "grad_norm": 1.1343128681182861, "learning_rate": 0.00012490128490599443, "loss": 3.5545, "step": 2490 }, { "epoch": 8.673617379884035e-15, "grad_norm": 0.9107974171638489, "learning_rate": 0.00012489995581898373, "loss": 3.4829, "step": 2500 }, { "epoch": 8.708311849403572e-15, "grad_norm": 0.9497007131576538, "learning_rate": 0.00012489861785234085, "loss": 3.4752, "step": 2510 }, { "epoch": 8.743006318923108e-15, "grad_norm": 0.9502478837966919, "learning_rate": 0.0001248972710062774, "loss": 3.4816, "step": 2520 }, { "epoch": 8.777700788442644e-15, "grad_norm": 0.964961588382721, "learning_rate": 0.00012489591528100634, "loss": 3.5242, "step": 2530 }, { "epoch": 8.81239525796218e-15, "grad_norm": 0.9320996403694153, "learning_rate": 0.00012489455067674212, "loss": 3.4762, "step": 2540 }, { "epoch": 8.847089727481716e-15, "grad_norm": 0.9358189105987549, "learning_rate": 0.0001248931771937005, "loss": 3.4861, "step": 2550 }, { "epoch": 8.881784197001252e-15, "grad_norm": 1.1089838743209839, "learning_rate": 0.00012489179483209865, "loss": 3.4547, "step": 2560 }, { "epoch": 8.916478666520788e-15, "grad_norm": 1.0396053791046143, "learning_rate": 0.00012489040359215525, "loss": 3.4833, "step": 2570 }, { "epoch": 8.951173136040325e-15, "grad_norm": 0.9273746013641357, "learning_rate": 0.00012488900347409028, "loss": 3.503, "step": 2580 }, { "epoch": 8.98586760555986e-15, "grad_norm": 0.8459292054176331, "learning_rate": 0.00012488759447812513, "loss": 3.4606, "step": 2590 }, { "epoch": 9.020562075079397e-15, "grad_norm": 0.8773356080055237, "learning_rate": 0.0001248861766044827, "loss": 3.4342, "step": 2600 }, { "epoch": 9.055256544598933e-15, "grad_norm": 0.8420624732971191, "learning_rate": 0.00012488474985338714, "loss": 3.4566, "step": 2610 }, { "epoch": 9.089951014118469e-15, "grad_norm": 0.9201327562332153, "learning_rate": 0.0001248833142250641, "loss": 3.4269, "step": 2620 }, { "epoch": 9.124645483638005e-15, "grad_norm": 0.9463579654693604, "learning_rate": 0.0001248818697197406, "loss": 3.4823, "step": 2630 }, { "epoch": 9.159339953157541e-15, "grad_norm": 0.9335482120513916, "learning_rate": 0.00012488041633764512, "loss": 3.4596, "step": 2640 }, { "epoch": 9.194034422677078e-15, "grad_norm": 0.8705974817276001, "learning_rate": 0.00012487895407900745, "loss": 3.4176, "step": 2650 }, { "epoch": 9.228728892196614e-15, "grad_norm": 0.9120906591415405, "learning_rate": 0.00012487748294405884, "loss": 3.4254, "step": 2660 }, { "epoch": 9.26342336171615e-15, "grad_norm": 0.8400965332984924, "learning_rate": 0.000124876002933032, "loss": 3.4056, "step": 2670 }, { "epoch": 9.298117831235686e-15, "grad_norm": 0.9838013052940369, "learning_rate": 0.0001248745140461609, "loss": 3.4299, "step": 2680 }, { "epoch": 9.332812300755222e-15, "grad_norm": 1.0034348964691162, "learning_rate": 0.000124873016283681, "loss": 3.4461, "step": 2690 }, { "epoch": 9.367506770274758e-15, "grad_norm": 0.9233691096305847, "learning_rate": 0.0001248715096458292, "loss": 3.4205, "step": 2700 }, { "epoch": 9.402201239794294e-15, "grad_norm": 0.8897681832313538, "learning_rate": 0.00012486999413284378, "loss": 3.3935, "step": 2710 }, { "epoch": 9.43689570931383e-15, "grad_norm": 0.837489902973175, "learning_rate": 0.00012486846974496435, "loss": 3.4099, "step": 2720 }, { "epoch": 9.471590178833367e-15, "grad_norm": 0.9077757596969604, "learning_rate": 0.000124866936482432, "loss": 3.4279, "step": 2730 }, { "epoch": 9.506284648352903e-15, "grad_norm": 0.8528566956520081, "learning_rate": 0.0001248653943454892, "loss": 3.4041, "step": 2740 }, { "epoch": 9.540979117872439e-15, "grad_norm": 1.0278834104537964, "learning_rate": 0.00012486384333437986, "loss": 3.3991, "step": 2750 }, { "epoch": 9.575673587391975e-15, "grad_norm": 0.8195092678070068, "learning_rate": 0.0001248622834493492, "loss": 3.3596, "step": 2760 }, { "epoch": 9.610368056911511e-15, "grad_norm": 0.8521397709846497, "learning_rate": 0.00012486071469064393, "loss": 3.4261, "step": 2770 }, { "epoch": 9.645062526431047e-15, "grad_norm": 0.8101736903190613, "learning_rate": 0.00012485913705851211, "loss": 3.414, "step": 2780 }, { "epoch": 9.679756995950584e-15, "grad_norm": 0.997776210308075, "learning_rate": 0.00012485755055320332, "loss": 3.3907, "step": 2790 }, { "epoch": 9.71445146547012e-15, "grad_norm": 1.0188149213790894, "learning_rate": 0.00012485595517496833, "loss": 3.3448, "step": 2800 }, { "epoch": 9.749145934989656e-15, "grad_norm": 0.8486977219581604, "learning_rate": 0.0001248543509240595, "loss": 3.3361, "step": 2810 }, { "epoch": 9.783840404509192e-15, "grad_norm": 0.854432225227356, "learning_rate": 0.00012485273780073056, "loss": 3.3748, "step": 2820 }, { "epoch": 9.818534874028728e-15, "grad_norm": 0.864069402217865, "learning_rate": 0.00012485111580523654, "loss": 3.3994, "step": 2830 }, { "epoch": 9.853229343548264e-15, "grad_norm": 0.8119328022003174, "learning_rate": 0.00012484948493783399, "loss": 3.3534, "step": 2840 }, { "epoch": 9.8879238130678e-15, "grad_norm": 1.0056599378585815, "learning_rate": 0.0001248478451987808, "loss": 3.3437, "step": 2850 }, { "epoch": 9.922618282587337e-15, "grad_norm": 0.9675257802009583, "learning_rate": 0.00012484619658833628, "loss": 3.364, "step": 2860 }, { "epoch": 9.957312752106873e-15, "grad_norm": 0.7408711314201355, "learning_rate": 0.00012484453910676114, "loss": 3.3092, "step": 2870 }, { "epoch": 9.992007221626409e-15, "grad_norm": 0.853739321231842, "learning_rate": 0.00012484287275431753, "loss": 3.3983, "step": 2880 }, { "epoch": 1.0026701691145945e-14, "grad_norm": 0.7800089120864868, "learning_rate": 0.00012484119753126892, "loss": 3.3405, "step": 2890 }, { "epoch": 1.0061396160665481e-14, "grad_norm": 0.8963075280189514, "learning_rate": 0.00012483951343788028, "loss": 3.3917, "step": 2900 }, { "epoch": 1.0096090630185017e-14, "grad_norm": 0.8680503368377686, "learning_rate": 0.0001248378204744179, "loss": 3.3383, "step": 2910 }, { "epoch": 1.0130785099704553e-14, "grad_norm": 0.8258677124977112, "learning_rate": 0.00012483611864114953, "loss": 3.3745, "step": 2920 }, { "epoch": 1.016547956922409e-14, "grad_norm": 0.9842739701271057, "learning_rate": 0.00012483440793834428, "loss": 3.3098, "step": 2930 }, { "epoch": 1.0200174038743626e-14, "grad_norm": 0.8482029438018799, "learning_rate": 0.0001248326883662727, "loss": 3.3041, "step": 2940 }, { "epoch": 1.0234868508263162e-14, "grad_norm": 0.8042598366737366, "learning_rate": 0.0001248309599252067, "loss": 3.2598, "step": 2950 }, { "epoch": 1.0269562977782698e-14, "grad_norm": 0.87558513879776, "learning_rate": 0.00012482922261541966, "loss": 3.3353, "step": 2960 }, { "epoch": 1.0304257447302234e-14, "grad_norm": 0.8558880090713501, "learning_rate": 0.00012482747643718627, "loss": 3.3133, "step": 2970 }, { "epoch": 1.033895191682177e-14, "grad_norm": 0.9676014184951782, "learning_rate": 0.0001248257213907827, "loss": 3.3086, "step": 2980 }, { "epoch": 1.0373646386341306e-14, "grad_norm": 0.9024289846420288, "learning_rate": 0.00012482395747648653, "loss": 3.3167, "step": 2990 }, { "epoch": 1.0408340855860843e-14, "grad_norm": 0.789986789226532, "learning_rate": 0.00012482218469457663, "loss": 3.2991, "step": 3000 }, { "epoch": 1.0443035325380379e-14, "grad_norm": 0.8115243911743164, "learning_rate": 0.00012482040304533342, "loss": 3.3262, "step": 3010 }, { "epoch": 1.0477729794899915e-14, "grad_norm": 0.7944703102111816, "learning_rate": 0.00012481861252903862, "loss": 3.3281, "step": 3020 }, { "epoch": 1.0512424264419451e-14, "grad_norm": 0.7910351157188416, "learning_rate": 0.0001248168131459754, "loss": 3.2934, "step": 3030 }, { "epoch": 1.0547118733938987e-14, "grad_norm": 0.8015826940536499, "learning_rate": 0.0001248150048964283, "loss": 3.3015, "step": 3040 }, { "epoch": 1.0581813203458523e-14, "grad_norm": 0.8789092302322388, "learning_rate": 0.00012481318778068327, "loss": 3.3021, "step": 3050 }, { "epoch": 1.061650767297806e-14, "grad_norm": 0.8058632016181946, "learning_rate": 0.00012481136179902773, "loss": 3.2615, "step": 3060 }, { "epoch": 1.0651202142497596e-14, "grad_norm": 0.8875707983970642, "learning_rate": 0.00012480952695175036, "loss": 3.2707, "step": 3070 }, { "epoch": 1.0685896612017132e-14, "grad_norm": 0.8026496767997742, "learning_rate": 0.00012480768323914137, "loss": 3.3037, "step": 3080 }, { "epoch": 1.0720591081536668e-14, "grad_norm": 0.7037297487258911, "learning_rate": 0.00012480583066149233, "loss": 3.2929, "step": 3090 }, { "epoch": 1.0755285551056204e-14, "grad_norm": 0.9366029500961304, "learning_rate": 0.00012480396921909622, "loss": 3.3181, "step": 3100 }, { "epoch": 1.078998002057574e-14, "grad_norm": 0.8701136112213135, "learning_rate": 0.00012480209891224737, "loss": 3.2339, "step": 3110 }, { "epoch": 1.0824674490095276e-14, "grad_norm": 0.7917385101318359, "learning_rate": 0.00012480021974124159, "loss": 3.2912, "step": 3120 }, { "epoch": 1.0859368959614812e-14, "grad_norm": 0.7848642468452454, "learning_rate": 0.00012479833170637604, "loss": 3.2809, "step": 3130 }, { "epoch": 1.0894063429134349e-14, "grad_norm": 0.8274076581001282, "learning_rate": 0.0001247964348079493, "loss": 3.2778, "step": 3140 }, { "epoch": 1.0928757898653885e-14, "grad_norm": 0.8585158586502075, "learning_rate": 0.00012479452904626135, "loss": 3.2572, "step": 3150 }, { "epoch": 1.0963452368173421e-14, "grad_norm": 1.0393214225769043, "learning_rate": 0.00012479261442161356, "loss": 3.2655, "step": 3160 }, { "epoch": 1.0998146837692957e-14, "grad_norm": 0.8175503015518188, "learning_rate": 0.0001247906909343087, "loss": 3.2628, "step": 3170 }, { "epoch": 1.1032841307212493e-14, "grad_norm": 0.7699459791183472, "learning_rate": 0.00012478875858465099, "loss": 3.2814, "step": 3180 }, { "epoch": 1.106753577673203e-14, "grad_norm": 0.8340761661529541, "learning_rate": 0.00012478681737294601, "loss": 3.267, "step": 3190 }, { "epoch": 1.1102230246251565e-14, "grad_norm": 0.7933850288391113, "learning_rate": 0.0001247848672995007, "loss": 3.2521, "step": 3200 }, { "epoch": 1.1136924715771102e-14, "grad_norm": 0.885556161403656, "learning_rate": 0.0001247829083646235, "loss": 3.2765, "step": 3210 }, { "epoch": 1.1171619185290638e-14, "grad_norm": 0.7902571558952332, "learning_rate": 0.00012478094056862413, "loss": 3.244, "step": 3220 }, { "epoch": 1.1206313654810174e-14, "grad_norm": 0.8358456492424011, "learning_rate": 0.0001247789639118139, "loss": 3.2017, "step": 3230 }, { "epoch": 1.124100812432971e-14, "grad_norm": 0.8746649622917175, "learning_rate": 0.00012477697839450528, "loss": 3.264, "step": 3240 }, { "epoch": 1.1275702593849246e-14, "grad_norm": 0.8939067721366882, "learning_rate": 0.00012477498401701232, "loss": 3.2381, "step": 3250 }, { "epoch": 1.1310397063368782e-14, "grad_norm": 0.8101410269737244, "learning_rate": 0.0001247729807796504, "loss": 3.2599, "step": 3260 }, { "epoch": 1.1345091532888318e-14, "grad_norm": 0.8689619898796082, "learning_rate": 0.00012477096868273633, "loss": 3.2214, "step": 3270 }, { "epoch": 1.1379786002407855e-14, "grad_norm": 0.7271187901496887, "learning_rate": 0.00012476894772658827, "loss": 3.2226, "step": 3280 }, { "epoch": 1.141448047192739e-14, "grad_norm": 0.8111050128936768, "learning_rate": 0.00012476691791152587, "loss": 3.2506, "step": 3290 }, { "epoch": 1.1449174941446927e-14, "grad_norm": 0.748626708984375, "learning_rate": 0.0001247648792378701, "loss": 3.2043, "step": 3300 }, { "epoch": 1.1483869410966463e-14, "grad_norm": 0.7505301833152771, "learning_rate": 0.00012476283170594334, "loss": 3.2578, "step": 3310 }, { "epoch": 1.1518563880485999e-14, "grad_norm": 0.8929470181465149, "learning_rate": 0.00012476077531606942, "loss": 3.2256, "step": 3320 }, { "epoch": 1.1553258350005535e-14, "grad_norm": 0.8270478248596191, "learning_rate": 0.00012475871006857353, "loss": 3.1908, "step": 3330 }, { "epoch": 1.1587952819525071e-14, "grad_norm": 0.865400493144989, "learning_rate": 0.00012475663596378227, "loss": 3.1757, "step": 3340 }, { "epoch": 1.1622647289044608e-14, "grad_norm": 0.7414234280586243, "learning_rate": 0.00012475455300202363, "loss": 3.199, "step": 3350 }, { "epoch": 1.1657341758564144e-14, "grad_norm": 0.8095191121101379, "learning_rate": 0.00012475246118362701, "loss": 3.2173, "step": 3360 }, { "epoch": 1.169203622808368e-14, "grad_norm": 0.725797176361084, "learning_rate": 0.00012475036050892325, "loss": 3.2307, "step": 3370 }, { "epoch": 1.1726730697603216e-14, "grad_norm": 0.8418781757354736, "learning_rate": 0.00012474825097824453, "loss": 3.2124, "step": 3380 }, { "epoch": 1.1761425167122752e-14, "grad_norm": 0.7218002080917358, "learning_rate": 0.00012474613259192443, "loss": 3.1903, "step": 3390 }, { "epoch": 1.1796119636642288e-14, "grad_norm": 0.8026348948478699, "learning_rate": 0.00012474400535029801, "loss": 3.245, "step": 3400 }, { "epoch": 1.1830814106161824e-14, "grad_norm": 0.8022415637969971, "learning_rate": 0.00012474186925370162, "loss": 3.2177, "step": 3410 }, { "epoch": 1.186550857568136e-14, "grad_norm": 0.8232307434082031, "learning_rate": 0.0001247397243024731, "loss": 3.1853, "step": 3420 }, { "epoch": 1.1900203045200897e-14, "grad_norm": 0.8155747056007385, "learning_rate": 0.00012473757049695164, "loss": 3.199, "step": 3430 }, { "epoch": 1.1934897514720433e-14, "grad_norm": 0.8763492107391357, "learning_rate": 0.00012473540783747785, "loss": 3.2141, "step": 3440 }, { "epoch": 1.1969591984239969e-14, "grad_norm": 0.8439213037490845, "learning_rate": 0.00012473323632439376, "loss": 3.1905, "step": 3450 }, { "epoch": 1.2004286453759505e-14, "grad_norm": 0.7893644571304321, "learning_rate": 0.00012473105595804272, "loss": 3.1821, "step": 3460 }, { "epoch": 1.2038980923279041e-14, "grad_norm": 0.7765392661094666, "learning_rate": 0.00012472886673876957, "loss": 3.171, "step": 3470 }, { "epoch": 1.2073675392798577e-14, "grad_norm": 0.7502172589302063, "learning_rate": 0.00012472666866692052, "loss": 3.2033, "step": 3480 }, { "epoch": 1.2108369862318114e-14, "grad_norm": 0.9560279250144958, "learning_rate": 0.00012472446174284317, "loss": 3.2092, "step": 3490 }, { "epoch": 1.214306433183765e-14, "grad_norm": 0.7791681289672852, "learning_rate": 0.00012472224596688655, "loss": 3.1672, "step": 3500 }, { "epoch": 1.2177758801357186e-14, "grad_norm": 0.8664537668228149, "learning_rate": 0.00012472002133940102, "loss": 3.178, "step": 3510 }, { "epoch": 1.2212453270876722e-14, "grad_norm": 0.7568835616111755, "learning_rate": 0.00012471778786073842, "loss": 3.1745, "step": 3520 }, { "epoch": 1.2247147740396258e-14, "grad_norm": 0.8368754386901855, "learning_rate": 0.00012471554553125194, "loss": 3.1694, "step": 3530 }, { "epoch": 1.2281842209915794e-14, "grad_norm": 0.7846680283546448, "learning_rate": 0.00012471329435129617, "loss": 3.1584, "step": 3540 }, { "epoch": 1.231653667943533e-14, "grad_norm": 0.8405663371086121, "learning_rate": 0.00012471103432122715, "loss": 3.2079, "step": 3550 }, { "epoch": 1.2351231148954867e-14, "grad_norm": 0.7632923722267151, "learning_rate": 0.00012470876544140228, "loss": 3.1319, "step": 3560 }, { "epoch": 1.2385925618474403e-14, "grad_norm": 0.7066273093223572, "learning_rate": 0.00012470648771218033, "loss": 3.1714, "step": 3570 }, { "epoch": 1.2420620087993939e-14, "grad_norm": 0.8684081435203552, "learning_rate": 0.00012470420113392153, "loss": 3.1661, "step": 3580 }, { "epoch": 1.2455314557513475e-14, "grad_norm": 0.855876088142395, "learning_rate": 0.00012470190570698747, "loss": 3.172, "step": 3590 }, { "epoch": 1.2490009027033011e-14, "grad_norm": 0.9231800436973572, "learning_rate": 0.0001246996014317412, "loss": 3.1419, "step": 3600 }, { "epoch": 1.2524703496552547e-14, "grad_norm": 0.8291738629341125, "learning_rate": 0.00012469728830854704, "loss": 3.1532, "step": 3610 }, { "epoch": 1.2559397966072083e-14, "grad_norm": 0.7272961735725403, "learning_rate": 0.00012469496633777085, "loss": 3.1593, "step": 3620 }, { "epoch": 1.259409243559162e-14, "grad_norm": 0.766878604888916, "learning_rate": 0.00012469263551977982, "loss": 3.1636, "step": 3630 }, { "epoch": 1.2628786905111156e-14, "grad_norm": 0.7307482957839966, "learning_rate": 0.00012469029585494251, "loss": 3.1927, "step": 3640 }, { "epoch": 1.2663481374630692e-14, "grad_norm": 0.8179423809051514, "learning_rate": 0.00012468794734362898, "loss": 3.1511, "step": 3650 }, { "epoch": 1.2698175844150228e-14, "grad_norm": 0.7904072999954224, "learning_rate": 0.00012468558998621055, "loss": 3.1535, "step": 3660 }, { "epoch": 1.2732870313669764e-14, "grad_norm": 0.7376141548156738, "learning_rate": 0.0001246832237830601, "loss": 3.1587, "step": 3670 }, { "epoch": 1.27675647831893e-14, "grad_norm": 0.8373693227767944, "learning_rate": 0.00012468084873455177, "loss": 3.125, "step": 3680 }, { "epoch": 1.2802259252708836e-14, "grad_norm": 0.8021637201309204, "learning_rate": 0.00012467846484106118, "loss": 3.1097, "step": 3690 }, { "epoch": 1.2836953722228372e-14, "grad_norm": 0.6624990701675415, "learning_rate": 0.0001246760721029653, "loss": 3.1559, "step": 3700 }, { "epoch": 1.2871648191747909e-14, "grad_norm": 0.7306764125823975, "learning_rate": 0.00012467367052064254, "loss": 3.1394, "step": 3710 }, { "epoch": 1.2906342661267445e-14, "grad_norm": 0.8226941227912903, "learning_rate": 0.00012467126009447268, "loss": 3.1413, "step": 3720 }, { "epoch": 1.2941037130786981e-14, "grad_norm": 0.745802104473114, "learning_rate": 0.0001246688408248369, "loss": 3.1527, "step": 3730 }, { "epoch": 1.2975731600306517e-14, "grad_norm": 0.7246356010437012, "learning_rate": 0.0001246664127121178, "loss": 3.1477, "step": 3740 }, { "epoch": 1.3010426069826053e-14, "grad_norm": 0.8900009393692017, "learning_rate": 0.00012466397575669935, "loss": 3.1083, "step": 3750 }, { "epoch": 1.304512053934559e-14, "grad_norm": 0.7691717743873596, "learning_rate": 0.00012466152995896696, "loss": 3.1085, "step": 3760 }, { "epoch": 1.3079815008865125e-14, "grad_norm": 0.8326911926269531, "learning_rate": 0.0001246590753193074, "loss": 3.1119, "step": 3770 }, { "epoch": 1.3114509478384662e-14, "grad_norm": 0.8034533858299255, "learning_rate": 0.00012465661183810886, "loss": 3.1335, "step": 3780 }, { "epoch": 1.3149203947904198e-14, "grad_norm": 0.735568642616272, "learning_rate": 0.00012465413951576087, "loss": 3.0916, "step": 3790 }, { "epoch": 1.3183898417423734e-14, "grad_norm": 0.7771469950675964, "learning_rate": 0.00012465165835265447, "loss": 3.1634, "step": 3800 }, { "epoch": 1.321859288694327e-14, "grad_norm": 0.796273410320282, "learning_rate": 0.00012464916834918199, "loss": 3.1279, "step": 3810 }, { "epoch": 1.3253287356462806e-14, "grad_norm": 0.7363583445549011, "learning_rate": 0.00012464666950573724, "loss": 3.1375, "step": 3820 }, { "epoch": 1.3287981825982342e-14, "grad_norm": 0.7906473875045776, "learning_rate": 0.00012464416182271534, "loss": 3.1187, "step": 3830 }, { "epoch": 1.3322676295501878e-14, "grad_norm": 0.7574383616447449, "learning_rate": 0.0001246416453005129, "loss": 3.1362, "step": 3840 }, { "epoch": 1.3357370765021415e-14, "grad_norm": 0.8083314895629883, "learning_rate": 0.00012463911993952788, "loss": 3.13, "step": 3850 }, { "epoch": 1.339206523454095e-14, "grad_norm": 0.7779651284217834, "learning_rate": 0.00012463658574015963, "loss": 3.0876, "step": 3860 }, { "epoch": 1.3426759704060487e-14, "grad_norm": 0.7964084148406982, "learning_rate": 0.00012463404270280894, "loss": 3.0905, "step": 3870 }, { "epoch": 1.3461454173580023e-14, "grad_norm": 0.7373225092887878, "learning_rate": 0.00012463149082787793, "loss": 3.1033, "step": 3880 }, { "epoch": 1.3496148643099559e-14, "grad_norm": 0.7947068810462952, "learning_rate": 0.00012462893011577018, "loss": 3.1148, "step": 3890 }, { "epoch": 1.3530843112619095e-14, "grad_norm": 0.745068371295929, "learning_rate": 0.00012462636056689063, "loss": 3.0971, "step": 3900 }, { "epoch": 1.3565537582138631e-14, "grad_norm": 0.7376861572265625, "learning_rate": 0.00012462378218164568, "loss": 3.045, "step": 3910 }, { "epoch": 1.3600232051658168e-14, "grad_norm": 0.7279322743415833, "learning_rate": 0.000124621194960443, "loss": 3.0916, "step": 3920 }, { "epoch": 1.3634926521177704e-14, "grad_norm": 0.7057489156723022, "learning_rate": 0.0001246185989036918, "loss": 3.0979, "step": 3930 }, { "epoch": 1.366962099069724e-14, "grad_norm": 0.7181816101074219, "learning_rate": 0.00012461599401180262, "loss": 3.1033, "step": 3940 }, { "epoch": 1.3704315460216776e-14, "grad_norm": 0.7674596905708313, "learning_rate": 0.00012461338028518735, "loss": 3.1109, "step": 3950 }, { "epoch": 1.3739009929736312e-14, "grad_norm": 0.703161895275116, "learning_rate": 0.00012461075772425938, "loss": 3.1337, "step": 3960 }, { "epoch": 1.3773704399255848e-14, "grad_norm": 0.7776753902435303, "learning_rate": 0.00012460812632943344, "loss": 3.099, "step": 3970 }, { "epoch": 1.3808398868775384e-14, "grad_norm": 0.7839871048927307, "learning_rate": 0.00012460548610112565, "loss": 3.1496, "step": 3980 }, { "epoch": 1.384309333829492e-14, "grad_norm": 0.7940945029258728, "learning_rate": 0.00012460283703975353, "loss": 3.1124, "step": 3990 }, { "epoch": 1.3877787807814457e-14, "grad_norm": 0.664508581161499, "learning_rate": 0.000124600179145736, "loss": 3.0913, "step": 4000 }, { "epoch": 1.3912482277333993e-14, "grad_norm": 0.7119874954223633, "learning_rate": 0.00012459751241949342, "loss": 3.0839, "step": 4010 }, { "epoch": 1.3947176746853529e-14, "grad_norm": 0.7712161540985107, "learning_rate": 0.0001245948368614475, "loss": 3.0912, "step": 4020 }, { "epoch": 1.3981871216373065e-14, "grad_norm": 0.7478469014167786, "learning_rate": 0.0001245921524720213, "loss": 3.0979, "step": 4030 }, { "epoch": 1.4016565685892601e-14, "grad_norm": 0.7218197584152222, "learning_rate": 0.00012458945925163944, "loss": 3.0667, "step": 4040 }, { "epoch": 1.4051260155412137e-14, "grad_norm": 0.7821571230888367, "learning_rate": 0.00012458675720072773, "loss": 3.0999, "step": 4050 }, { "epoch": 1.4085954624931674e-14, "grad_norm": 0.7311422824859619, "learning_rate": 0.00012458404631971354, "loss": 3.081, "step": 4060 }, { "epoch": 1.412064909445121e-14, "grad_norm": 0.8397068381309509, "learning_rate": 0.00012458132660902554, "loss": 3.0924, "step": 4070 }, { "epoch": 1.4155343563970746e-14, "grad_norm": 0.7732449173927307, "learning_rate": 0.00012457859806909381, "loss": 3.0772, "step": 4080 }, { "epoch": 1.4190038033490282e-14, "grad_norm": 0.7348335981369019, "learning_rate": 0.0001245758607003499, "loss": 3.0793, "step": 4090 }, { "epoch": 1.4224732503009818e-14, "grad_norm": 0.7658295035362244, "learning_rate": 0.00012457311450322664, "loss": 3.1329, "step": 4100 }, { "epoch": 1.4259426972529354e-14, "grad_norm": 0.6935727000236511, "learning_rate": 0.00012457035947815838, "loss": 3.05, "step": 4110 }, { "epoch": 1.429412144204889e-14, "grad_norm": 0.7814844846725464, "learning_rate": 0.00012456759562558077, "loss": 3.1054, "step": 4120 }, { "epoch": 1.4328815911568427e-14, "grad_norm": 0.6381024122238159, "learning_rate": 0.0001245648229459309, "loss": 3.0311, "step": 4130 }, { "epoch": 1.4363510381087963e-14, "grad_norm": 0.6988679766654968, "learning_rate": 0.00012456204143964722, "loss": 3.0814, "step": 4140 }, { "epoch": 1.43982048506075e-14, "grad_norm": 0.819586455821991, "learning_rate": 0.00012455925110716961, "loss": 3.1005, "step": 4150 }, { "epoch": 1.4432899320127035e-14, "grad_norm": 0.7517282962799072, "learning_rate": 0.00012455645194893938, "loss": 3.0747, "step": 4160 }, { "epoch": 1.446759378964657e-14, "grad_norm": 0.7151929140090942, "learning_rate": 0.00012455364396539916, "loss": 3.0484, "step": 4170 }, { "epoch": 1.4502288259166107e-14, "grad_norm": 0.6947036385536194, "learning_rate": 0.00012455082715699298, "loss": 3.0632, "step": 4180 }, { "epoch": 1.4536982728685643e-14, "grad_norm": 0.7482947111129761, "learning_rate": 0.00012454800152416633, "loss": 3.0686, "step": 4190 }, { "epoch": 1.457167719820518e-14, "grad_norm": 0.7070441842079163, "learning_rate": 0.00012454516706736607, "loss": 3.0484, "step": 4200 }, { "epoch": 1.4606371667724716e-14, "grad_norm": 0.7110937833786011, "learning_rate": 0.0001245423237870404, "loss": 3.0285, "step": 4210 }, { "epoch": 1.4641066137244252e-14, "grad_norm": 0.7551254630088806, "learning_rate": 0.000124539471683639, "loss": 3.0369, "step": 4220 }, { "epoch": 1.4675760606763788e-14, "grad_norm": 0.707839846611023, "learning_rate": 0.00012453661075761288, "loss": 3.0446, "step": 4230 }, { "epoch": 1.4710455076283324e-14, "grad_norm": 0.7582014203071594, "learning_rate": 0.0001245337410094145, "loss": 3.0662, "step": 4240 }, { "epoch": 1.474514954580286e-14, "grad_norm": 0.7325688004493713, "learning_rate": 0.00012453086243949767, "loss": 3.0507, "step": 4250 }, { "epoch": 1.4779844015322396e-14, "grad_norm": 0.8526148200035095, "learning_rate": 0.0001245279750483176, "loss": 3.0473, "step": 4260 }, { "epoch": 1.4814538484841933e-14, "grad_norm": 0.7357615828514099, "learning_rate": 0.00012452507883633091, "loss": 3.0856, "step": 4270 }, { "epoch": 1.484923295436147e-14, "grad_norm": 0.7400020360946655, "learning_rate": 0.00012452217380399562, "loss": 3.0653, "step": 4280 }, { "epoch": 1.4883927423881005e-14, "grad_norm": 0.7040058374404907, "learning_rate": 0.00012451925995177115, "loss": 3.0479, "step": 4290 }, { "epoch": 1.491862189340054e-14, "grad_norm": 0.6828770041465759, "learning_rate": 0.00012451633728011828, "loss": 3.0164, "step": 4300 }, { "epoch": 1.4953316362920077e-14, "grad_norm": 0.6935179829597473, "learning_rate": 0.00012451340578949918, "loss": 3.0753, "step": 4310 }, { "epoch": 1.4988010832439613e-14, "grad_norm": 0.7052716016769409, "learning_rate": 0.0001245104654803775, "loss": 3.0632, "step": 4320 }, { "epoch": 1.502270530195915e-14, "grad_norm": 0.7265757918357849, "learning_rate": 0.00012450751635321818, "loss": 3.0561, "step": 4330 }, { "epoch": 1.5057399771478686e-14, "grad_norm": 0.7397698760032654, "learning_rate": 0.00012450455840848763, "loss": 3.0288, "step": 4340 }, { "epoch": 1.5092094240998222e-14, "grad_norm": 0.7340617775917053, "learning_rate": 0.00012450159164665358, "loss": 3.0381, "step": 4350 }, { "epoch": 1.5126788710517758e-14, "grad_norm": 0.7598298788070679, "learning_rate": 0.00012449861606818525, "loss": 3.0216, "step": 4360 }, { "epoch": 1.5161483180037294e-14, "grad_norm": 0.8035849928855896, "learning_rate": 0.00012449563167355317, "loss": 3.0564, "step": 4370 }, { "epoch": 1.519617764955683e-14, "grad_norm": 0.6707798838615417, "learning_rate": 0.00012449263846322928, "loss": 3.0268, "step": 4380 }, { "epoch": 1.5230872119076366e-14, "grad_norm": 0.7244170308113098, "learning_rate": 0.00012448963643768698, "loss": 3.0537, "step": 4390 }, { "epoch": 1.5265566588595902e-14, "grad_norm": 0.6986050009727478, "learning_rate": 0.00012448662559740096, "loss": 2.9723, "step": 4400 }, { "epoch": 1.530026105811544e-14, "grad_norm": 0.7201911807060242, "learning_rate": 0.0001244836059428474, "loss": 3.0466, "step": 4410 }, { "epoch": 1.5334955527634975e-14, "grad_norm": 0.761934220790863, "learning_rate": 0.00012448057747450384, "loss": 3.0419, "step": 4420 }, { "epoch": 1.536964999715451e-14, "grad_norm": 0.6656420230865479, "learning_rate": 0.00012447754019284915, "loss": 3.025, "step": 4430 }, { "epoch": 1.5404344466674047e-14, "grad_norm": 0.7581021785736084, "learning_rate": 0.0001244744940983637, "loss": 3.0394, "step": 4440 }, { "epoch": 1.5439038936193583e-14, "grad_norm": 0.6712950468063354, "learning_rate": 0.0001244714391915292, "loss": 3.0675, "step": 4450 }, { "epoch": 1.547373340571312e-14, "grad_norm": 0.6931546330451965, "learning_rate": 0.00012446837547282872, "loss": 3.0258, "step": 4460 }, { "epoch": 1.5508427875232655e-14, "grad_norm": 0.703609049320221, "learning_rate": 0.00012446530294274682, "loss": 3.0065, "step": 4470 }, { "epoch": 1.554312234475219e-14, "grad_norm": 0.7280946969985962, "learning_rate": 0.00012446222160176934, "loss": 3.0539, "step": 4480 }, { "epoch": 1.5577816814271728e-14, "grad_norm": 0.6789142489433289, "learning_rate": 0.0001244591314503836, "loss": 3.011, "step": 4490 }, { "epoch": 1.5612511283791264e-14, "grad_norm": 0.718810498714447, "learning_rate": 0.00012445603248907824, "loss": 3.0051, "step": 4500 }, { "epoch": 1.56472057533108e-14, "grad_norm": 0.7286238074302673, "learning_rate": 0.00012445292471834337, "loss": 3.0238, "step": 4510 }, { "epoch": 1.5681900222830336e-14, "grad_norm": 0.6888225078582764, "learning_rate": 0.00012444980813867046, "loss": 3.0543, "step": 4520 }, { "epoch": 1.5716594692349872e-14, "grad_norm": 0.6972569823265076, "learning_rate": 0.00012444668275055236, "loss": 3.0073, "step": 4530 }, { "epoch": 1.5751289161869408e-14, "grad_norm": 0.7302638292312622, "learning_rate": 0.00012444354855448332, "loss": 3.0285, "step": 4540 }, { "epoch": 1.5785983631388945e-14, "grad_norm": 0.6544486880302429, "learning_rate": 0.000124440405550959, "loss": 3.0129, "step": 4550 }, { "epoch": 1.582067810090848e-14, "grad_norm": 0.6626681089401245, "learning_rate": 0.0001244372537404764, "loss": 3.0166, "step": 4560 }, { "epoch": 1.5855372570428017e-14, "grad_norm": 0.6533512473106384, "learning_rate": 0.000124434093123534, "loss": 3.0039, "step": 4570 }, { "epoch": 1.5890067039947553e-14, "grad_norm": 0.7138797044754028, "learning_rate": 0.00012443092370063163, "loss": 2.9873, "step": 4580 }, { "epoch": 1.592476150946709e-14, "grad_norm": 0.7039511799812317, "learning_rate": 0.00012442774547227043, "loss": 3.0062, "step": 4590 }, { "epoch": 1.5959455978986625e-14, "grad_norm": 0.6465252041816711, "learning_rate": 0.0001244245584389531, "loss": 3.0021, "step": 4600 }, { "epoch": 1.599415044850616e-14, "grad_norm": 0.67662513256073, "learning_rate": 0.00012442136260118358, "loss": 3.0011, "step": 4610 }, { "epoch": 1.6028844918025698e-14, "grad_norm": 0.7998048663139343, "learning_rate": 0.0001244181579594673, "loss": 3.0431, "step": 4620 }, { "epoch": 1.6063539387545234e-14, "grad_norm": 0.8624919056892395, "learning_rate": 0.00012441494451431105, "loss": 3.0185, "step": 4630 }, { "epoch": 1.609823385706477e-14, "grad_norm": 0.664972186088562, "learning_rate": 0.00012441172226622296, "loss": 3.0267, "step": 4640 }, { "epoch": 1.6132928326584306e-14, "grad_norm": 0.7613922953605652, "learning_rate": 0.00012440849121571264, "loss": 2.9809, "step": 4650 }, { "epoch": 1.6167622796103842e-14, "grad_norm": 0.7426005005836487, "learning_rate": 0.00012440525136329106, "loss": 3.0183, "step": 4660 }, { "epoch": 1.6202317265623378e-14, "grad_norm": 0.7002527117729187, "learning_rate": 0.00012440200270947055, "loss": 3.0095, "step": 4670 }, { "epoch": 1.6237011735142914e-14, "grad_norm": 0.7750053405761719, "learning_rate": 0.00012439874525476487, "loss": 2.9579, "step": 4680 }, { "epoch": 1.627170620466245e-14, "grad_norm": 0.7063431143760681, "learning_rate": 0.00012439547899968913, "loss": 3.0042, "step": 4690 }, { "epoch": 1.6306400674181987e-14, "grad_norm": 0.6393726468086243, "learning_rate": 0.00012439220394475992, "loss": 2.964, "step": 4700 }, { "epoch": 1.6341095143701523e-14, "grad_norm": 0.7400582432746887, "learning_rate": 0.00012438892009049513, "loss": 2.9746, "step": 4710 }, { "epoch": 1.637578961322106e-14, "grad_norm": 0.6828471422195435, "learning_rate": 0.00012438562743741406, "loss": 2.9645, "step": 4720 }, { "epoch": 1.6410484082740595e-14, "grad_norm": 0.6524949669837952, "learning_rate": 0.00012438232598603743, "loss": 2.9727, "step": 4730 }, { "epoch": 1.644517855226013e-14, "grad_norm": 0.7743234038352966, "learning_rate": 0.00012437901573688733, "loss": 3.0058, "step": 4740 }, { "epoch": 1.6479873021779667e-14, "grad_norm": 0.7012312412261963, "learning_rate": 0.00012437569669048723, "loss": 3.0187, "step": 4750 }, { "epoch": 1.6514567491299204e-14, "grad_norm": 0.7543424367904663, "learning_rate": 0.00012437236884736206, "loss": 3.0004, "step": 4760 }, { "epoch": 1.654926196081874e-14, "grad_norm": 0.8476557731628418, "learning_rate": 0.00012436903220803804, "loss": 2.9777, "step": 4770 }, { "epoch": 1.6583956430338276e-14, "grad_norm": 0.7226364016532898, "learning_rate": 0.00012436568677304282, "loss": 2.9946, "step": 4780 }, { "epoch": 1.6618650899857812e-14, "grad_norm": 0.6563786268234253, "learning_rate": 0.00012436233254290552, "loss": 3.0027, "step": 4790 }, { "epoch": 1.6653345369377348e-14, "grad_norm": 0.6584700345993042, "learning_rate": 0.00012435896951815654, "loss": 2.9749, "step": 4800 }, { "epoch": 1.6688039838896884e-14, "grad_norm": 0.7400755882263184, "learning_rate": 0.0001243555976993277, "loss": 2.9758, "step": 4810 }, { "epoch": 1.672273430841642e-14, "grad_norm": 0.7412940263748169, "learning_rate": 0.00012435221708695225, "loss": 2.9689, "step": 4820 }, { "epoch": 1.6757428777935957e-14, "grad_norm": 0.6534263491630554, "learning_rate": 0.0001243488276815648, "loss": 2.9621, "step": 4830 }, { "epoch": 1.6792123247455493e-14, "grad_norm": 0.7333821058273315, "learning_rate": 0.00012434542948370133, "loss": 3.0075, "step": 4840 }, { "epoch": 1.682681771697503e-14, "grad_norm": 0.7161802649497986, "learning_rate": 0.0001243420224938993, "loss": 2.9997, "step": 4850 }, { "epoch": 1.6861512186494565e-14, "grad_norm": 0.7017641067504883, "learning_rate": 0.0001243386067126974, "loss": 3.0028, "step": 4860 }, { "epoch": 1.68962066560141e-14, "grad_norm": 0.8049578666687012, "learning_rate": 0.00012433518214063587, "loss": 2.9586, "step": 4870 }, { "epoch": 1.6930901125533637e-14, "grad_norm": 0.7451969385147095, "learning_rate": 0.00012433174877825626, "loss": 3.0144, "step": 4880 }, { "epoch": 1.6965595595053173e-14, "grad_norm": 0.6921694874763489, "learning_rate": 0.00012432830662610156, "loss": 3.0061, "step": 4890 }, { "epoch": 1.700029006457271e-14, "grad_norm": 0.8296787738800049, "learning_rate": 0.00012432485568471607, "loss": 2.9784, "step": 4900 }, { "epoch": 1.7034984534092246e-14, "grad_norm": 0.6721057295799255, "learning_rate": 0.00012432139595464555, "loss": 2.979, "step": 4910 }, { "epoch": 1.7069679003611782e-14, "grad_norm": 0.6850361227989197, "learning_rate": 0.00012431792743643712, "loss": 2.9515, "step": 4920 }, { "epoch": 1.7104373473131318e-14, "grad_norm": 0.8247771263122559, "learning_rate": 0.0001243144501306393, "loss": 2.9764, "step": 4930 }, { "epoch": 1.7139067942650854e-14, "grad_norm": 0.6937596797943115, "learning_rate": 0.000124310964037802, "loss": 2.9705, "step": 4940 }, { "epoch": 1.717376241217039e-14, "grad_norm": 0.6781917214393616, "learning_rate": 0.00012430746915847651, "loss": 2.9389, "step": 4950 }, { "epoch": 1.7208456881689926e-14, "grad_norm": 0.6863271594047546, "learning_rate": 0.0001243039654932155, "loss": 2.9409, "step": 4960 }, { "epoch": 1.7243151351209463e-14, "grad_norm": 0.6420871019363403, "learning_rate": 0.00012430045304257306, "loss": 2.9371, "step": 4970 }, { "epoch": 1.7277845820729e-14, "grad_norm": 0.6796463131904602, "learning_rate": 0.00012429693180710465, "loss": 2.9199, "step": 4980 }, { "epoch": 1.7312540290248535e-14, "grad_norm": 0.6276524066925049, "learning_rate": 0.00012429340178736714, "loss": 2.9677, "step": 4990 }, { "epoch": 1.734723475976807e-14, "grad_norm": 0.6770766973495483, "learning_rate": 0.00012428986298391876, "loss": 2.9736, "step": 5000 }, { "epoch": 1.7381929229287607e-14, "grad_norm": 0.6607031226158142, "learning_rate": 0.00012428631539731914, "loss": 2.9594, "step": 5010 }, { "epoch": 1.7416623698807143e-14, "grad_norm": 0.6835026741027832, "learning_rate": 0.00012428275902812928, "loss": 2.9282, "step": 5020 }, { "epoch": 1.745131816832668e-14, "grad_norm": 0.6466560363769531, "learning_rate": 0.00012427919387691163, "loss": 2.9579, "step": 5030 }, { "epoch": 1.7486012637846216e-14, "grad_norm": 0.7365676760673523, "learning_rate": 0.00012427561994422996, "loss": 2.9674, "step": 5040 }, { "epoch": 1.7520707107365752e-14, "grad_norm": 0.7608414888381958, "learning_rate": 0.00012427203723064945, "loss": 2.9798, "step": 5050 }, { "epoch": 1.7555401576885288e-14, "grad_norm": 0.6575977206230164, "learning_rate": 0.0001242684457367367, "loss": 2.9766, "step": 5060 }, { "epoch": 1.7590096046404824e-14, "grad_norm": 0.6910486817359924, "learning_rate": 0.00012426484546305966, "loss": 2.9616, "step": 5070 }, { "epoch": 1.762479051592436e-14, "grad_norm": 0.6427014470100403, "learning_rate": 0.0001242612364101877, "loss": 2.9542, "step": 5080 }, { "epoch": 1.7659484985443896e-14, "grad_norm": 0.609898567199707, "learning_rate": 0.00012425761857869154, "loss": 2.9328, "step": 5090 }, { "epoch": 1.7694179454963432e-14, "grad_norm": 0.7632775902748108, "learning_rate": 0.00012425399196914328, "loss": 2.9154, "step": 5100 }, { "epoch": 1.772887392448297e-14, "grad_norm": 0.6610209345817566, "learning_rate": 0.00012425035658211652, "loss": 2.9137, "step": 5110 }, { "epoch": 1.7763568394002505e-14, "grad_norm": 0.6405383348464966, "learning_rate": 0.0001242467124181861, "loss": 2.9273, "step": 5120 }, { "epoch": 1.779826286352204e-14, "grad_norm": 0.6587648391723633, "learning_rate": 0.00012424305947792832, "loss": 2.9234, "step": 5130 }, { "epoch": 1.7832957333041577e-14, "grad_norm": 0.6330050826072693, "learning_rate": 0.0001242393977619209, "loss": 2.9744, "step": 5140 }, { "epoch": 1.7867651802561113e-14, "grad_norm": 0.687660276889801, "learning_rate": 0.00012423572727074283, "loss": 2.9257, "step": 5150 }, { "epoch": 1.790234627208065e-14, "grad_norm": 0.7093472480773926, "learning_rate": 0.00012423204800497467, "loss": 2.9181, "step": 5160 }, { "epoch": 1.7937040741600185e-14, "grad_norm": 0.6153919696807861, "learning_rate": 0.00012422835996519818, "loss": 2.9188, "step": 5170 }, { "epoch": 1.797173521111972e-14, "grad_norm": 0.6825246810913086, "learning_rate": 0.00012422466315199662, "loss": 2.952, "step": 5180 }, { "epoch": 1.8006429680639258e-14, "grad_norm": 0.7290875911712646, "learning_rate": 0.00012422095756595462, "loss": 2.9403, "step": 5190 }, { "epoch": 1.8041124150158794e-14, "grad_norm": 0.6530689597129822, "learning_rate": 0.00012421724320765818, "loss": 2.9441, "step": 5200 }, { "epoch": 1.807581861967833e-14, "grad_norm": 0.6551821231842041, "learning_rate": 0.0001242135200776947, "loss": 2.9665, "step": 5210 }, { "epoch": 1.8110513089197866e-14, "grad_norm": 0.632440984249115, "learning_rate": 0.00012420978817665292, "loss": 2.9421, "step": 5220 }, { "epoch": 1.8145207558717402e-14, "grad_norm": 0.6050541400909424, "learning_rate": 0.0001242060475051231, "loss": 2.9677, "step": 5230 }, { "epoch": 1.8179902028236938e-14, "grad_norm": 0.5892105102539062, "learning_rate": 0.00012420229806369668, "loss": 2.9478, "step": 5240 }, { "epoch": 1.8214596497756474e-14, "grad_norm": 0.7979709506034851, "learning_rate": 0.00012419853985296667, "loss": 2.8984, "step": 5250 }, { "epoch": 1.824929096727601e-14, "grad_norm": 0.6521009802818298, "learning_rate": 0.0001241947728735274, "loss": 2.9431, "step": 5260 }, { "epoch": 1.8283985436795547e-14, "grad_norm": 0.6817916035652161, "learning_rate": 0.00012419099712597455, "loss": 2.9223, "step": 5270 }, { "epoch": 1.8318679906315083e-14, "grad_norm": 0.6648424863815308, "learning_rate": 0.00012418721261090528, "loss": 2.9049, "step": 5280 }, { "epoch": 1.835337437583462e-14, "grad_norm": 1.6042283773422241, "learning_rate": 0.000124183419328918, "loss": 2.9293, "step": 5290 }, { "epoch": 1.8388068845354155e-14, "grad_norm": 0.6467262506484985, "learning_rate": 0.00012417961728061267, "loss": 2.944, "step": 5300 }, { "epoch": 1.842276331487369e-14, "grad_norm": 0.6570130586624146, "learning_rate": 0.0001241758064665905, "loss": 2.9254, "step": 5310 }, { "epoch": 1.8457457784393227e-14, "grad_norm": 0.6388654708862305, "learning_rate": 0.00012417198688745412, "loss": 2.9584, "step": 5320 }, { "epoch": 1.8492152253912764e-14, "grad_norm": 0.6613774299621582, "learning_rate": 0.0001241681585438076, "loss": 2.952, "step": 5330 }, { "epoch": 1.85268467234323e-14, "grad_norm": 0.6348766088485718, "learning_rate": 0.00012416432143625637, "loss": 2.9303, "step": 5340 }, { "epoch": 1.8561541192951836e-14, "grad_norm": 0.6680907011032104, "learning_rate": 0.00012416047556540718, "loss": 2.9291, "step": 5350 }, { "epoch": 1.8596235662471372e-14, "grad_norm": 0.6389169692993164, "learning_rate": 0.00012415662093186828, "loss": 2.9317, "step": 5360 }, { "epoch": 1.8630930131990908e-14, "grad_norm": 0.6902768611907959, "learning_rate": 0.0001241527575362492, "loss": 2.9196, "step": 5370 }, { "epoch": 1.8665624601510444e-14, "grad_norm": 0.8409585952758789, "learning_rate": 0.00012414888537916097, "loss": 2.8977, "step": 5380 }, { "epoch": 1.870031907102998e-14, "grad_norm": 0.6743999123573303, "learning_rate": 0.00012414500446121586, "loss": 2.9191, "step": 5390 }, { "epoch": 1.8735013540549517e-14, "grad_norm": 0.6337765455245972, "learning_rate": 0.00012414111478302765, "loss": 2.9361, "step": 5400 }, { "epoch": 1.8769708010069053e-14, "grad_norm": 0.6448245048522949, "learning_rate": 0.00012413721634521143, "loss": 2.9253, "step": 5410 }, { "epoch": 1.880440247958859e-14, "grad_norm": 0.6430071592330933, "learning_rate": 0.00012413330914838374, "loss": 2.9431, "step": 5420 }, { "epoch": 1.8839096949108125e-14, "grad_norm": 0.6534712314605713, "learning_rate": 0.00012412939319316244, "loss": 2.9187, "step": 5430 }, { "epoch": 1.887379141862766e-14, "grad_norm": 0.6882981657981873, "learning_rate": 0.00012412546848016682, "loss": 2.8997, "step": 5440 }, { "epoch": 1.8908485888147197e-14, "grad_norm": 0.6201768517494202, "learning_rate": 0.00012412153501001753, "loss": 2.9373, "step": 5450 }, { "epoch": 1.8943180357666733e-14, "grad_norm": 0.6958590149879456, "learning_rate": 0.00012411759278333664, "loss": 2.9264, "step": 5460 }, { "epoch": 1.897787482718627e-14, "grad_norm": 0.6530311107635498, "learning_rate": 0.00012411364180074753, "loss": 2.8944, "step": 5470 }, { "epoch": 1.9012569296705806e-14, "grad_norm": 0.682150661945343, "learning_rate": 0.00012410968206287508, "loss": 2.9771, "step": 5480 }, { "epoch": 1.9047263766225342e-14, "grad_norm": 0.6742517352104187, "learning_rate": 0.00012410571357034542, "loss": 2.9456, "step": 5490 }, { "epoch": 1.9081958235744878e-14, "grad_norm": 0.7005282044410706, "learning_rate": 0.00012410173632378615, "loss": 2.9166, "step": 5500 }, { "epoch": 1.9116652705264414e-14, "grad_norm": 0.602691113948822, "learning_rate": 0.0001240977503238263, "loss": 2.9055, "step": 5510 }, { "epoch": 1.915134717478395e-14, "grad_norm": 0.6978890895843506, "learning_rate": 0.00012409375557109612, "loss": 2.9218, "step": 5520 }, { "epoch": 1.9186041644303486e-14, "grad_norm": 0.6568784713745117, "learning_rate": 0.00012408975206622743, "loss": 2.8922, "step": 5530 }, { "epoch": 1.9220736113823023e-14, "grad_norm": 0.623920202255249, "learning_rate": 0.00012408573980985331, "loss": 2.9262, "step": 5540 }, { "epoch": 1.925543058334256e-14, "grad_norm": 0.6240105032920837, "learning_rate": 0.0001240817188026083, "loss": 2.8942, "step": 5550 }, { "epoch": 1.9290125052862095e-14, "grad_norm": 0.7104952931404114, "learning_rate": 0.00012407768904512822, "loss": 2.8989, "step": 5560 }, { "epoch": 1.932481952238163e-14, "grad_norm": 0.6404186487197876, "learning_rate": 0.0001240736505380504, "loss": 2.9236, "step": 5570 }, { "epoch": 1.9359513991901167e-14, "grad_norm": 0.6760048270225525, "learning_rate": 0.00012406960328201346, "loss": 2.9343, "step": 5580 }, { "epoch": 1.9394208461420703e-14, "grad_norm": 0.6658358573913574, "learning_rate": 0.0001240655472776575, "loss": 2.9258, "step": 5590 }, { "epoch": 1.942890293094024e-14, "grad_norm": 0.6508139967918396, "learning_rate": 0.00012406148252562386, "loss": 2.8869, "step": 5600 }, { "epoch": 1.9463597400459776e-14, "grad_norm": 0.6359962224960327, "learning_rate": 0.0001240574090265554, "loss": 2.888, "step": 5610 }, { "epoch": 1.9498291869979312e-14, "grad_norm": 0.621383011341095, "learning_rate": 0.00012405332678109627, "loss": 2.927, "step": 5620 }, { "epoch": 1.9532986339498848e-14, "grad_norm": 0.6612720489501953, "learning_rate": 0.00012404923578989208, "loss": 2.8693, "step": 5630 }, { "epoch": 1.9567680809018384e-14, "grad_norm": 0.6199637651443481, "learning_rate": 0.00012404513605358976, "loss": 2.8743, "step": 5640 }, { "epoch": 1.960237527853792e-14, "grad_norm": 0.6133468151092529, "learning_rate": 0.00012404102757283765, "loss": 2.893, "step": 5650 }, { "epoch": 1.9637069748057456e-14, "grad_norm": 0.6827776432037354, "learning_rate": 0.0001240369103482855, "loss": 2.9087, "step": 5660 }, { "epoch": 1.9671764217576992e-14, "grad_norm": 0.631712019443512, "learning_rate": 0.00012403278438058438, "loss": 2.8942, "step": 5670 }, { "epoch": 1.970645868709653e-14, "grad_norm": 0.7147265672683716, "learning_rate": 0.00012402864967038678, "loss": 2.9353, "step": 5680 }, { "epoch": 1.9741153156616065e-14, "grad_norm": 0.715764045715332, "learning_rate": 0.0001240245062183466, "loss": 2.8852, "step": 5690 }, { "epoch": 1.97758476261356e-14, "grad_norm": 0.6232671141624451, "learning_rate": 0.00012402035402511906, "loss": 2.8825, "step": 5700 }, { "epoch": 1.9810542095655137e-14, "grad_norm": 0.6820979118347168, "learning_rate": 0.00012401619309136079, "loss": 2.8815, "step": 5710 }, { "epoch": 1.9845236565174673e-14, "grad_norm": 0.6694291234016418, "learning_rate": 0.0001240120234177298, "loss": 2.8676, "step": 5720 }, { "epoch": 1.987993103469421e-14, "grad_norm": 0.6339734792709351, "learning_rate": 0.00012400784500488554, "loss": 2.905, "step": 5730 }, { "epoch": 1.9914625504213745e-14, "grad_norm": 0.7280707359313965, "learning_rate": 0.00012400365785348875, "loss": 2.9286, "step": 5740 }, { "epoch": 1.9949319973733282e-14, "grad_norm": 0.6926658749580383, "learning_rate": 0.00012399946196420158, "loss": 2.8762, "step": 5750 }, { "epoch": 1.9984014443252818e-14, "grad_norm": 0.6961377859115601, "learning_rate": 0.0001239952573376876, "loss": 2.8813, "step": 5760 }, { "epoch": 2.0018708912772354e-14, "grad_norm": 0.6775835752487183, "learning_rate": 0.0001239910439746117, "loss": 2.9005, "step": 5770 }, { "epoch": 2.005340338229189e-14, "grad_norm": 0.6083140969276428, "learning_rate": 0.00012398682187564027, "loss": 2.9169, "step": 5780 }, { "epoch": 2.0088097851811426e-14, "grad_norm": 0.6381298303604126, "learning_rate": 0.0001239825910414409, "loss": 2.9203, "step": 5790 }, { "epoch": 2.0122792321330962e-14, "grad_norm": 0.6996100544929504, "learning_rate": 0.00012397835147268273, "loss": 2.8936, "step": 5800 }, { "epoch": 2.01574867908505e-14, "grad_norm": 0.6299130916595459, "learning_rate": 0.00012397410317003618, "loss": 2.8778, "step": 5810 }, { "epoch": 2.0192181260370035e-14, "grad_norm": 0.7721242904663086, "learning_rate": 0.00012396984613417307, "loss": 2.9, "step": 5820 }, { "epoch": 2.022687572988957e-14, "grad_norm": 0.6079384684562683, "learning_rate": 0.00012396558036576662, "loss": 2.9107, "step": 5830 }, { "epoch": 2.0261570199409107e-14, "grad_norm": 0.608013927936554, "learning_rate": 0.00012396130586549145, "loss": 2.9224, "step": 5840 }, { "epoch": 2.0296264668928643e-14, "grad_norm": 0.6412060856819153, "learning_rate": 0.0001239570226340235, "loss": 2.8421, "step": 5850 }, { "epoch": 2.033095913844818e-14, "grad_norm": 0.6592738628387451, "learning_rate": 0.00012395273067204016, "loss": 2.8933, "step": 5860 }, { "epoch": 2.0365653607967715e-14, "grad_norm": 0.645621657371521, "learning_rate": 0.00012394842998022018, "loss": 2.8887, "step": 5870 }, { "epoch": 2.040034807748725e-14, "grad_norm": 0.6985535621643066, "learning_rate": 0.00012394412055924362, "loss": 2.9227, "step": 5880 }, { "epoch": 2.0435042547006788e-14, "grad_norm": 0.629172146320343, "learning_rate": 0.00012393980240979203, "loss": 2.8777, "step": 5890 }, { "epoch": 2.0469737016526324e-14, "grad_norm": 0.6524924635887146, "learning_rate": 0.00012393547553254825, "loss": 2.8934, "step": 5900 }, { "epoch": 2.050443148604586e-14, "grad_norm": 0.6359326243400574, "learning_rate": 0.00012393113992819656, "loss": 2.8785, "step": 5910 }, { "epoch": 2.0539125955565396e-14, "grad_norm": 0.6454670429229736, "learning_rate": 0.00012392679559742258, "loss": 2.8971, "step": 5920 }, { "epoch": 2.0573820425084932e-14, "grad_norm": 0.6657472848892212, "learning_rate": 0.00012392244254091334, "loss": 2.8438, "step": 5930 }, { "epoch": 2.0608514894604468e-14, "grad_norm": 0.654632031917572, "learning_rate": 0.00012391808075935725, "loss": 2.8725, "step": 5940 }, { "epoch": 2.0643209364124004e-14, "grad_norm": 0.6664455533027649, "learning_rate": 0.00012391371025344408, "loss": 2.8699, "step": 5950 }, { "epoch": 2.067790383364354e-14, "grad_norm": 0.5696113109588623, "learning_rate": 0.00012390933102386496, "loss": 2.9134, "step": 5960 }, { "epoch": 2.0712598303163077e-14, "grad_norm": 0.6414527893066406, "learning_rate": 0.0001239049430713125, "loss": 2.8996, "step": 5970 }, { "epoch": 2.0747292772682613e-14, "grad_norm": 0.6280456185340881, "learning_rate": 0.00012390054639648053, "loss": 2.8818, "step": 5980 }, { "epoch": 2.078198724220215e-14, "grad_norm": 0.6286094188690186, "learning_rate": 0.0001238961410000644, "loss": 2.8834, "step": 5990 }, { "epoch": 2.0816681711721685e-14, "grad_norm": 0.6175819039344788, "learning_rate": 0.0001238917268827608, "loss": 2.8226, "step": 6000 }, { "epoch": 2.085137618124122e-14, "grad_norm": 0.6722970604896545, "learning_rate": 0.0001238873040452677, "loss": 2.85, "step": 6010 }, { "epoch": 2.0886070650760757e-14, "grad_norm": 0.6301952600479126, "learning_rate": 0.00012388287248828465, "loss": 2.8329, "step": 6020 }, { "epoch": 2.0920765120280294e-14, "grad_norm": 0.6365863084793091, "learning_rate": 0.00012387843221251237, "loss": 2.8686, "step": 6030 }, { "epoch": 2.095545958979983e-14, "grad_norm": 0.6493092775344849, "learning_rate": 0.00012387398321865307, "loss": 2.8552, "step": 6040 }, { "epoch": 2.0990154059319366e-14, "grad_norm": 0.609856903553009, "learning_rate": 0.00012386952550741038, "loss": 2.8543, "step": 6050 }, { "epoch": 2.1024848528838902e-14, "grad_norm": 0.6454996466636658, "learning_rate": 0.00012386505907948918, "loss": 2.8776, "step": 6060 }, { "epoch": 2.1059542998358438e-14, "grad_norm": 0.6441075205802917, "learning_rate": 0.0001238605839355958, "loss": 2.8504, "step": 6070 }, { "epoch": 2.1094237467877974e-14, "grad_norm": 0.6317734122276306, "learning_rate": 0.00012385610007643798, "loss": 2.8823, "step": 6080 }, { "epoch": 2.112893193739751e-14, "grad_norm": 0.6229076981544495, "learning_rate": 0.0001238516075027248, "loss": 2.8654, "step": 6090 }, { "epoch": 2.1163626406917047e-14, "grad_norm": 0.6474233269691467, "learning_rate": 0.00012384710621516672, "loss": 2.8643, "step": 6100 }, { "epoch": 2.1198320876436583e-14, "grad_norm": 0.6765882968902588, "learning_rate": 0.00012384259621447553, "loss": 2.8319, "step": 6110 }, { "epoch": 2.123301534595612e-14, "grad_norm": 0.6636559367179871, "learning_rate": 0.00012383807750136451, "loss": 2.8558, "step": 6120 }, { "epoch": 2.1267709815475655e-14, "grad_norm": 0.6156606078147888, "learning_rate": 0.00012383355007654825, "loss": 2.8715, "step": 6130 }, { "epoch": 2.130240428499519e-14, "grad_norm": 0.8312148451805115, "learning_rate": 0.00012382901394074268, "loss": 2.8484, "step": 6140 }, { "epoch": 2.1337098754514727e-14, "grad_norm": 0.6334800124168396, "learning_rate": 0.0001238244690946652, "loss": 2.8587, "step": 6150 }, { "epoch": 2.1371793224034263e-14, "grad_norm": 0.6986563801765442, "learning_rate": 0.0001238199155390345, "loss": 2.8159, "step": 6160 }, { "epoch": 2.14064876935538e-14, "grad_norm": 0.6897568106651306, "learning_rate": 0.00012381535327457072, "loss": 2.8479, "step": 6170 }, { "epoch": 2.1441182163073336e-14, "grad_norm": 0.5966699719429016, "learning_rate": 0.00012381078230199534, "loss": 2.8351, "step": 6180 }, { "epoch": 2.1475876632592872e-14, "grad_norm": 0.5847378373146057, "learning_rate": 0.00012380620262203117, "loss": 2.8494, "step": 6190 }, { "epoch": 2.1510571102112408e-14, "grad_norm": 0.6262260675430298, "learning_rate": 0.00012380161423540252, "loss": 2.8346, "step": 6200 }, { "epoch": 2.1545265571631944e-14, "grad_norm": 0.634401261806488, "learning_rate": 0.00012379701714283493, "loss": 2.8715, "step": 6210 }, { "epoch": 2.157996004115148e-14, "grad_norm": 0.6387273073196411, "learning_rate": 0.00012379241134505544, "loss": 2.8755, "step": 6220 }, { "epoch": 2.1614654510671016e-14, "grad_norm": 0.5762618184089661, "learning_rate": 0.0001237877968427924, "loss": 2.8404, "step": 6230 }, { "epoch": 2.1649348980190553e-14, "grad_norm": 0.7550005316734314, "learning_rate": 0.00012378317363677556, "loss": 2.8225, "step": 6240 }, { "epoch": 2.168404344971009e-14, "grad_norm": 0.6284468770027161, "learning_rate": 0.00012377854172773605, "loss": 2.8472, "step": 6250 }, { "epoch": 2.1718737919229625e-14, "grad_norm": 0.7554931640625, "learning_rate": 0.00012377390111640636, "loss": 2.8413, "step": 6260 }, { "epoch": 2.175343238874916e-14, "grad_norm": 0.5965588092803955, "learning_rate": 0.00012376925180352034, "loss": 2.8602, "step": 6270 }, { "epoch": 2.1788126858268697e-14, "grad_norm": 0.6671257615089417, "learning_rate": 0.00012376459378981325, "loss": 2.8254, "step": 6280 }, { "epoch": 2.1822821327788233e-14, "grad_norm": 0.6184029579162598, "learning_rate": 0.00012375992707602172, "loss": 2.9014, "step": 6290 }, { "epoch": 2.185751579730777e-14, "grad_norm": 0.6221917867660522, "learning_rate": 0.00012375525166288378, "loss": 2.8729, "step": 6300 }, { "epoch": 2.1892210266827306e-14, "grad_norm": 0.6048524975776672, "learning_rate": 0.00012375056755113876, "loss": 2.8307, "step": 6310 }, { "epoch": 2.1926904736346842e-14, "grad_norm": 0.6449043154716492, "learning_rate": 0.00012374587474152742, "loss": 2.8353, "step": 6320 }, { "epoch": 2.1961599205866378e-14, "grad_norm": 0.6057182550430298, "learning_rate": 0.0001237411732347919, "loss": 2.8443, "step": 6330 }, { "epoch": 2.1996293675385914e-14, "grad_norm": 0.6340845823287964, "learning_rate": 0.00012373646303167571, "loss": 2.8116, "step": 6340 }, { "epoch": 2.203098814490545e-14, "grad_norm": 0.6390292048454285, "learning_rate": 0.0001237317441329237, "loss": 2.8703, "step": 6350 }, { "epoch": 2.2065682614424986e-14, "grad_norm": 0.6227693557739258, "learning_rate": 0.00012372701653928217, "loss": 2.8352, "step": 6360 }, { "epoch": 2.2100377083944522e-14, "grad_norm": 0.6227917075157166, "learning_rate": 0.00012372228025149873, "loss": 2.8472, "step": 6370 }, { "epoch": 2.213507155346406e-14, "grad_norm": 0.6174140572547913, "learning_rate": 0.00012371753527032234, "loss": 2.7987, "step": 6380 }, { "epoch": 2.2169766022983595e-14, "grad_norm": 0.6746034026145935, "learning_rate": 0.00012371278159650342, "loss": 2.8404, "step": 6390 }, { "epoch": 2.220446049250313e-14, "grad_norm": 0.6226774454116821, "learning_rate": 0.0001237080192307937, "loss": 2.8227, "step": 6400 }, { "epoch": 2.2239154962022667e-14, "grad_norm": 0.6395253539085388, "learning_rate": 0.00012370324817394634, "loss": 2.8609, "step": 6410 }, { "epoch": 2.2273849431542203e-14, "grad_norm": 0.6412621140480042, "learning_rate": 0.00012369846842671584, "loss": 2.8587, "step": 6420 }, { "epoch": 2.230854390106174e-14, "grad_norm": 0.607075035572052, "learning_rate": 0.00012369367998985804, "loss": 2.8602, "step": 6430 }, { "epoch": 2.2343238370581275e-14, "grad_norm": 0.572821855545044, "learning_rate": 0.00012368888286413016, "loss": 2.8654, "step": 6440 }, { "epoch": 2.237793284010081e-14, "grad_norm": 0.6066297888755798, "learning_rate": 0.00012368407705029094, "loss": 2.8549, "step": 6450 }, { "epoch": 2.2412627309620348e-14, "grad_norm": 0.6861979961395264, "learning_rate": 0.0001236792625491003, "loss": 2.8098, "step": 6460 }, { "epoch": 2.2447321779139884e-14, "grad_norm": 0.648659884929657, "learning_rate": 0.0001236744393613196, "loss": 2.8064, "step": 6470 }, { "epoch": 2.248201624865942e-14, "grad_norm": 0.6738684773445129, "learning_rate": 0.0001236696074877116, "loss": 2.8447, "step": 6480 }, { "epoch": 2.2516710718178956e-14, "grad_norm": 0.657389223575592, "learning_rate": 0.00012366476692904043, "loss": 2.7898, "step": 6490 }, { "epoch": 2.2551405187698492e-14, "grad_norm": 0.6917227506637573, "learning_rate": 0.0001236599176860716, "loss": 2.835, "step": 6500 }, { "epoch": 2.2586099657218028e-14, "grad_norm": 0.6928380727767944, "learning_rate": 0.00012365505975957196, "loss": 2.8476, "step": 6510 }, { "epoch": 2.2620794126737565e-14, "grad_norm": 0.6112115979194641, "learning_rate": 0.00012365019315030974, "loss": 2.869, "step": 6520 }, { "epoch": 2.26554885962571e-14, "grad_norm": 0.6323057413101196, "learning_rate": 0.00012364531785905453, "loss": 2.7834, "step": 6530 }, { "epoch": 2.2690183065776637e-14, "grad_norm": 0.6016417145729065, "learning_rate": 0.00012364043388657738, "loss": 2.8257, "step": 6540 }, { "epoch": 2.2724877535296173e-14, "grad_norm": 0.6313493847846985, "learning_rate": 0.0001236355412336506, "loss": 2.8249, "step": 6550 }, { "epoch": 2.275957200481571e-14, "grad_norm": 0.7013261318206787, "learning_rate": 0.00012363063990104793, "loss": 2.8047, "step": 6560 }, { "epoch": 2.2794266474335245e-14, "grad_norm": 0.7066786885261536, "learning_rate": 0.00012362572988954448, "loss": 2.8711, "step": 6570 }, { "epoch": 2.282896094385478e-14, "grad_norm": 0.6478776335716248, "learning_rate": 0.00012362081119991669, "loss": 2.8112, "step": 6580 }, { "epoch": 2.2863655413374318e-14, "grad_norm": 0.6609697341918945, "learning_rate": 0.00012361588383294245, "loss": 2.8192, "step": 6590 }, { "epoch": 2.2898349882893854e-14, "grad_norm": 0.639938235282898, "learning_rate": 0.00012361094778940102, "loss": 2.8362, "step": 6600 }, { "epoch": 2.293304435241339e-14, "grad_norm": 0.6834721565246582, "learning_rate": 0.00012360600307007288, "loss": 2.8514, "step": 6610 }, { "epoch": 2.2967738821932926e-14, "grad_norm": 0.5977281332015991, "learning_rate": 0.0001236010496757401, "loss": 2.8234, "step": 6620 }, { "epoch": 2.3002433291452462e-14, "grad_norm": 0.6234945058822632, "learning_rate": 0.00012359608760718595, "loss": 2.8238, "step": 6630 }, { "epoch": 2.3037127760971998e-14, "grad_norm": 0.6152732372283936, "learning_rate": 0.00012359111686519516, "loss": 2.8078, "step": 6640 }, { "epoch": 2.3071822230491534e-14, "grad_norm": 0.607958197593689, "learning_rate": 0.00012358613745055383, "loss": 2.8881, "step": 6650 }, { "epoch": 2.310651670001107e-14, "grad_norm": 0.6179840564727783, "learning_rate": 0.00012358114936404942, "loss": 2.7897, "step": 6660 }, { "epoch": 2.3141211169530607e-14, "grad_norm": 0.6033633947372437, "learning_rate": 0.00012357615260647067, "loss": 2.8085, "step": 6670 }, { "epoch": 2.3175905639050143e-14, "grad_norm": 0.6591785550117493, "learning_rate": 0.00012357114717860787, "loss": 2.8311, "step": 6680 }, { "epoch": 2.321060010856968e-14, "grad_norm": 0.6293900012969971, "learning_rate": 0.00012356613308125254, "loss": 2.8145, "step": 6690 }, { "epoch": 2.3245294578089215e-14, "grad_norm": 0.7538554668426514, "learning_rate": 0.00012356111031519763, "loss": 2.7805, "step": 6700 }, { "epoch": 2.327998904760875e-14, "grad_norm": 0.6200160384178162, "learning_rate": 0.00012355607888123743, "loss": 2.7536, "step": 6710 }, { "epoch": 2.3314683517128287e-14, "grad_norm": 0.6362175941467285, "learning_rate": 0.00012355103878016762, "loss": 2.7995, "step": 6720 }, { "epoch": 2.3349377986647823e-14, "grad_norm": 0.6065363883972168, "learning_rate": 0.00012354599001278526, "loss": 2.8308, "step": 6730 }, { "epoch": 2.338407245616736e-14, "grad_norm": 0.6287219524383545, "learning_rate": 0.0001235409325798888, "loss": 2.8395, "step": 6740 }, { "epoch": 2.3418766925686896e-14, "grad_norm": 0.5881543159484863, "learning_rate": 0.00012353586648227797, "loss": 2.8078, "step": 6750 }, { "epoch": 2.3453461395206432e-14, "grad_norm": 0.615274965763092, "learning_rate": 0.00012353079172075396, "loss": 2.8399, "step": 6760 }, { "epoch": 2.3488155864725968e-14, "grad_norm": 0.6396788954734802, "learning_rate": 0.00012352570829611927, "loss": 2.8324, "step": 6770 }, { "epoch": 2.3522850334245504e-14, "grad_norm": 0.6074019074440002, "learning_rate": 0.00012352061620917788, "loss": 2.8336, "step": 6780 }, { "epoch": 2.355754480376504e-14, "grad_norm": 0.6960377097129822, "learning_rate": 0.00012351551546073495, "loss": 2.8094, "step": 6790 }, { "epoch": 2.3592239273284576e-14, "grad_norm": 0.6103847622871399, "learning_rate": 0.0001235104060515972, "loss": 2.8158, "step": 6800 }, { "epoch": 2.3626933742804113e-14, "grad_norm": 0.5879714488983154, "learning_rate": 0.00012350528798257262, "loss": 2.8173, "step": 6810 }, { "epoch": 2.366162821232365e-14, "grad_norm": 0.6215558052062988, "learning_rate": 0.00012350016125447057, "loss": 2.8328, "step": 6820 }, { "epoch": 2.3696322681843185e-14, "grad_norm": 0.5923135876655579, "learning_rate": 0.00012349502586810182, "loss": 2.8362, "step": 6830 }, { "epoch": 2.373101715136272e-14, "grad_norm": 0.6567990183830261, "learning_rate": 0.00012348988182427848, "loss": 2.8209, "step": 6840 }, { "epoch": 2.3765711620882257e-14, "grad_norm": 0.6301283240318298, "learning_rate": 0.00012348472912381402, "loss": 2.8083, "step": 6850 }, { "epoch": 2.3800406090401793e-14, "grad_norm": 0.5850421190261841, "learning_rate": 0.00012347956776752333, "loss": 2.785, "step": 6860 }, { "epoch": 2.383510055992133e-14, "grad_norm": 0.6071661710739136, "learning_rate": 0.0001234743977562226, "loss": 2.8054, "step": 6870 }, { "epoch": 2.3869795029440866e-14, "grad_norm": 0.6008660793304443, "learning_rate": 0.0001234692190907294, "loss": 2.8299, "step": 6880 }, { "epoch": 2.3904489498960402e-14, "grad_norm": 0.6322605013847351, "learning_rate": 0.00012346403177186278, "loss": 2.8022, "step": 6890 }, { "epoch": 2.3939183968479938e-14, "grad_norm": 0.6217210292816162, "learning_rate": 0.000123458835800443, "loss": 2.803, "step": 6900 }, { "epoch": 2.3973878437999474e-14, "grad_norm": 0.5809029936790466, "learning_rate": 0.0001234536311772918, "loss": 2.7823, "step": 6910 }, { "epoch": 2.400857290751901e-14, "grad_norm": 0.6093421578407288, "learning_rate": 0.0001234484179032322, "loss": 2.7999, "step": 6920 }, { "epoch": 2.4043267377038546e-14, "grad_norm": 0.5995078086853027, "learning_rate": 0.00012344319597908863, "loss": 2.8394, "step": 6930 }, { "epoch": 2.4077961846558082e-14, "grad_norm": 0.603911280632019, "learning_rate": 0.00012343796540568693, "loss": 2.789, "step": 6940 }, { "epoch": 2.411265631607762e-14, "grad_norm": 0.6865026354789734, "learning_rate": 0.00012343272618385427, "loss": 2.8198, "step": 6950 }, { "epoch": 2.4147350785597155e-14, "grad_norm": 0.6227596402168274, "learning_rate": 0.00012342747831441917, "loss": 2.8016, "step": 6960 }, { "epoch": 2.418204525511669e-14, "grad_norm": 0.5985812544822693, "learning_rate": 0.00012342222179821156, "loss": 2.7629, "step": 6970 }, { "epoch": 2.4216739724636227e-14, "grad_norm": 0.6254674792289734, "learning_rate": 0.00012341695663606268, "loss": 2.788, "step": 6980 }, { "epoch": 2.4251434194155763e-14, "grad_norm": 0.6176968216896057, "learning_rate": 0.00012341168282880516, "loss": 2.8045, "step": 6990 }, { "epoch": 2.42861286636753e-14, "grad_norm": 0.6041901111602783, "learning_rate": 0.00012340640037727306, "loss": 2.793, "step": 7000 }, { "epoch": 2.4320823133194835e-14, "grad_norm": 0.617764413356781, "learning_rate": 0.00012340110928230172, "loss": 2.7948, "step": 7010 }, { "epoch": 2.4355517602714372e-14, "grad_norm": 0.6121236681938171, "learning_rate": 0.0001233958095447279, "loss": 2.7938, "step": 7020 }, { "epoch": 2.4390212072233908e-14, "grad_norm": 0.6262118816375732, "learning_rate": 0.0001233905011653897, "loss": 2.8051, "step": 7030 }, { "epoch": 2.4424906541753444e-14, "grad_norm": 0.616005003452301, "learning_rate": 0.00012338518414512655, "loss": 2.7638, "step": 7040 }, { "epoch": 2.445960101127298e-14, "grad_norm": 0.6222718358039856, "learning_rate": 0.00012337985848477934, "loss": 2.8271, "step": 7050 }, { "epoch": 2.4494295480792516e-14, "grad_norm": 0.5830873847007751, "learning_rate": 0.0001233745241851903, "loss": 2.8136, "step": 7060 }, { "epoch": 2.4528989950312052e-14, "grad_norm": 0.594871461391449, "learning_rate": 0.00012336918124720295, "loss": 2.8073, "step": 7070 }, { "epoch": 2.456368441983159e-14, "grad_norm": 0.6114310622215271, "learning_rate": 0.00012336382967166224, "loss": 2.7951, "step": 7080 }, { "epoch": 2.4598378889351125e-14, "grad_norm": 0.6184341311454773, "learning_rate": 0.00012335846945941452, "loss": 2.8123, "step": 7090 }, { "epoch": 2.463307335887066e-14, "grad_norm": 0.652316153049469, "learning_rate": 0.00012335310061130738, "loss": 2.7882, "step": 7100 }, { "epoch": 2.4667767828390197e-14, "grad_norm": 0.6358462572097778, "learning_rate": 0.00012334772312818995, "loss": 2.8237, "step": 7110 }, { "epoch": 2.4702462297909733e-14, "grad_norm": 0.6402984261512756, "learning_rate": 0.00012334233701091256, "loss": 2.8005, "step": 7120 }, { "epoch": 2.473715676742927e-14, "grad_norm": 0.5787113904953003, "learning_rate": 0.00012333694226032701, "loss": 2.8035, "step": 7130 }, { "epoch": 2.4771851236948805e-14, "grad_norm": 0.6592028141021729, "learning_rate": 0.00012333153887728645, "loss": 2.7505, "step": 7140 }, { "epoch": 2.480654570646834e-14, "grad_norm": 0.6272876262664795, "learning_rate": 0.00012332612686264535, "loss": 2.7635, "step": 7150 }, { "epoch": 2.4841240175987878e-14, "grad_norm": 0.6601934432983398, "learning_rate": 0.00012332070621725956, "loss": 2.7861, "step": 7160 }, { "epoch": 2.4875934645507414e-14, "grad_norm": 0.6227889060974121, "learning_rate": 0.00012331527694198636, "loss": 2.7583, "step": 7170 }, { "epoch": 2.491062911502695e-14, "grad_norm": 0.5864116549491882, "learning_rate": 0.00012330983903768427, "loss": 2.8562, "step": 7180 }, { "epoch": 2.4945323584546486e-14, "grad_norm": 0.6195365190505981, "learning_rate": 0.00012330439250521333, "loss": 2.7984, "step": 7190 }, { "epoch": 2.4980018054066022e-14, "grad_norm": 0.6609063744544983, "learning_rate": 0.00012329893734543482, "loss": 2.7764, "step": 7200 }, { "epoch": 2.5014712523585558e-14, "grad_norm": 0.6414263844490051, "learning_rate": 0.0001232934735592114, "loss": 2.7534, "step": 7210 }, { "epoch": 2.5049406993105094e-14, "grad_norm": 0.6003432869911194, "learning_rate": 0.00012328800114740716, "loss": 2.7802, "step": 7220 }, { "epoch": 2.508410146262463e-14, "grad_norm": 0.6355634331703186, "learning_rate": 0.00012328252011088752, "loss": 2.8132, "step": 7230 }, { "epoch": 2.5118795932144167e-14, "grad_norm": 0.5887851119041443, "learning_rate": 0.00012327703045051923, "loss": 2.8034, "step": 7240 }, { "epoch": 2.5153490401663703e-14, "grad_norm": 0.6077080368995667, "learning_rate": 0.00012327153216717048, "loss": 2.8108, "step": 7250 }, { "epoch": 2.518818487118324e-14, "grad_norm": 0.6021937131881714, "learning_rate": 0.00012326602526171071, "loss": 2.779, "step": 7260 }, { "epoch": 2.5222879340702775e-14, "grad_norm": 0.5979954600334167, "learning_rate": 0.00012326050973501083, "loss": 2.793, "step": 7270 }, { "epoch": 2.525757381022231e-14, "grad_norm": 0.6141809821128845, "learning_rate": 0.00012325498558794307, "loss": 2.7632, "step": 7280 }, { "epoch": 2.5292268279741847e-14, "grad_norm": 0.6196129322052002, "learning_rate": 0.00012324945282138098, "loss": 2.7959, "step": 7290 }, { "epoch": 2.5326962749261384e-14, "grad_norm": 0.6411095857620239, "learning_rate": 0.0001232439114361996, "loss": 2.7636, "step": 7300 }, { "epoch": 2.536165721878092e-14, "grad_norm": 0.5794517397880554, "learning_rate": 0.00012323836143327522, "loss": 2.7771, "step": 7310 }, { "epoch": 2.5396351688300456e-14, "grad_norm": 0.6242613196372986, "learning_rate": 0.0001232328028134855, "loss": 2.8133, "step": 7320 }, { "epoch": 2.5431046157819992e-14, "grad_norm": 0.5768901705741882, "learning_rate": 0.00012322723557770953, "loss": 2.7639, "step": 7330 }, { "epoch": 2.5465740627339528e-14, "grad_norm": 0.5999317765235901, "learning_rate": 0.00012322165972682766, "loss": 2.789, "step": 7340 }, { "epoch": 2.5500435096859064e-14, "grad_norm": 0.5961325168609619, "learning_rate": 0.00012321607526172173, "loss": 2.7277, "step": 7350 }, { "epoch": 2.55351295663786e-14, "grad_norm": 0.57700514793396, "learning_rate": 0.00012321048218327482, "loss": 2.781, "step": 7360 }, { "epoch": 2.5569824035898137e-14, "grad_norm": 0.5936629772186279, "learning_rate": 0.00012320488049237147, "loss": 2.7722, "step": 7370 }, { "epoch": 2.5604518505417673e-14, "grad_norm": 0.5948240160942078, "learning_rate": 0.00012319927018989753, "loss": 2.7529, "step": 7380 }, { "epoch": 2.563921297493721e-14, "grad_norm": 0.623176097869873, "learning_rate": 0.00012319365127674018, "loss": 2.7821, "step": 7390 }, { "epoch": 2.5673907444456745e-14, "grad_norm": 0.5807058811187744, "learning_rate": 0.00012318802375378807, "loss": 2.7979, "step": 7400 }, { "epoch": 2.570860191397628e-14, "grad_norm": 0.6008669137954712, "learning_rate": 0.00012318238762193108, "loss": 2.7768, "step": 7410 }, { "epoch": 2.5743296383495817e-14, "grad_norm": 0.6285163760185242, "learning_rate": 0.00012317674288206057, "loss": 2.7606, "step": 7420 }, { "epoch": 2.5777990853015353e-14, "grad_norm": 0.5936919450759888, "learning_rate": 0.00012317108953506915, "loss": 2.8103, "step": 7430 }, { "epoch": 2.581268532253489e-14, "grad_norm": 0.5742027163505554, "learning_rate": 0.0001231654275818509, "loss": 2.8136, "step": 7440 }, { "epoch": 2.5847379792054426e-14, "grad_norm": 0.59840327501297, "learning_rate": 0.0001231597570233012, "loss": 2.7795, "step": 7450 }, { "epoch": 2.5882074261573962e-14, "grad_norm": 0.6117955446243286, "learning_rate": 0.00012315407786031677, "loss": 2.7421, "step": 7460 }, { "epoch": 2.5916768731093498e-14, "grad_norm": 0.5565841794013977, "learning_rate": 0.00012314839009379576, "loss": 2.7423, "step": 7470 }, { "epoch": 2.5951463200613034e-14, "grad_norm": 0.5690497159957886, "learning_rate": 0.0001231426937246376, "loss": 2.7622, "step": 7480 }, { "epoch": 2.598615767013257e-14, "grad_norm": 0.5705859065055847, "learning_rate": 0.00012313698875374314, "loss": 2.7874, "step": 7490 }, { "epoch": 2.6020852139652106e-14, "grad_norm": 0.6228348016738892, "learning_rate": 0.0001231312751820146, "loss": 2.7713, "step": 7500 }, { "epoch": 2.6055546609171643e-14, "grad_norm": 0.5669484734535217, "learning_rate": 0.0001231255530103555, "loss": 2.7938, "step": 7510 }, { "epoch": 2.609024107869118e-14, "grad_norm": 0.5787023901939392, "learning_rate": 0.00012311982223967076, "loss": 2.7835, "step": 7520 }, { "epoch": 2.6124935548210715e-14, "grad_norm": 0.634845495223999, "learning_rate": 0.00012311408287086662, "loss": 2.7788, "step": 7530 }, { "epoch": 2.615963001773025e-14, "grad_norm": 0.5812243223190308, "learning_rate": 0.0001231083349048508, "loss": 2.7889, "step": 7540 }, { "epoch": 2.6194324487249787e-14, "grad_norm": 0.6303025484085083, "learning_rate": 0.00012310257834253216, "loss": 2.7835, "step": 7550 }, { "epoch": 2.6229018956769323e-14, "grad_norm": 0.5807857513427734, "learning_rate": 0.00012309681318482119, "loss": 2.7609, "step": 7560 }, { "epoch": 2.626371342628886e-14, "grad_norm": 0.6607471108436584, "learning_rate": 0.0001230910394326295, "loss": 2.7468, "step": 7570 }, { "epoch": 2.6298407895808396e-14, "grad_norm": 0.615468442440033, "learning_rate": 0.0001230852570868702, "loss": 2.7761, "step": 7580 }, { "epoch": 2.6333102365327932e-14, "grad_norm": 0.5996100306510925, "learning_rate": 0.00012307946614845773, "loss": 2.8171, "step": 7590 }, { "epoch": 2.6367796834847468e-14, "grad_norm": 0.6120663285255432, "learning_rate": 0.00012307366661830786, "loss": 2.7656, "step": 7600 }, { "epoch": 2.6402491304367004e-14, "grad_norm": 0.6031818389892578, "learning_rate": 0.0001230678584973377, "loss": 2.7838, "step": 7610 }, { "epoch": 2.643718577388654e-14, "grad_norm": 0.5723419189453125, "learning_rate": 0.00012306204178646584, "loss": 2.7639, "step": 7620 }, { "epoch": 2.6471880243406076e-14, "grad_norm": 0.6247963309288025, "learning_rate": 0.00012305621648661205, "loss": 2.765, "step": 7630 }, { "epoch": 2.6506574712925612e-14, "grad_norm": 0.6081961989402771, "learning_rate": 0.00012305038259869763, "loss": 2.7715, "step": 7640 }, { "epoch": 2.654126918244515e-14, "grad_norm": 0.5922369360923767, "learning_rate": 0.00012304454012364508, "loss": 2.7828, "step": 7650 }, { "epoch": 2.6575963651964685e-14, "grad_norm": 0.5739519000053406, "learning_rate": 0.00012303868906237843, "loss": 2.7408, "step": 7660 }, { "epoch": 2.661065812148422e-14, "grad_norm": 0.5887702107429504, "learning_rate": 0.0001230328294158229, "loss": 2.761, "step": 7670 }, { "epoch": 2.6645352591003757e-14, "grad_norm": 0.5994238257408142, "learning_rate": 0.00012302696118490514, "loss": 2.7418, "step": 7680 }, { "epoch": 2.6680047060523293e-14, "grad_norm": 0.6180620789527893, "learning_rate": 0.00012302108437055323, "loss": 2.7516, "step": 7690 }, { "epoch": 2.671474153004283e-14, "grad_norm": 0.6270855069160461, "learning_rate": 0.0001230151989736965, "loss": 2.7644, "step": 7700 }, { "epoch": 2.6749435999562365e-14, "grad_norm": 0.5789709687232971, "learning_rate": 0.00012300930499526563, "loss": 2.7593, "step": 7710 }, { "epoch": 2.67841304690819e-14, "grad_norm": 0.6262583136558533, "learning_rate": 0.00012300340243619277, "loss": 2.7774, "step": 7720 }, { "epoch": 2.6818824938601438e-14, "grad_norm": 0.6528642773628235, "learning_rate": 0.00012299749129741132, "loss": 2.7506, "step": 7730 }, { "epoch": 2.6853519408120974e-14, "grad_norm": 0.6030797958374023, "learning_rate": 0.0001229915715798561, "loss": 2.7717, "step": 7740 }, { "epoch": 2.688821387764051e-14, "grad_norm": 0.6062162518501282, "learning_rate": 0.00012298564328446323, "loss": 2.7474, "step": 7750 }, { "epoch": 2.6922908347160046e-14, "grad_norm": 0.6079098582267761, "learning_rate": 0.00012297970641217027, "loss": 2.769, "step": 7760 }, { "epoch": 2.6957602816679582e-14, "grad_norm": 0.5754967331886292, "learning_rate": 0.00012297376096391603, "loss": 2.7884, "step": 7770 }, { "epoch": 2.6992297286199118e-14, "grad_norm": 0.6070712208747864, "learning_rate": 0.00012296780694064075, "loss": 2.7576, "step": 7780 }, { "epoch": 2.7026991755718655e-14, "grad_norm": 0.5947773456573486, "learning_rate": 0.00012296184434328603, "loss": 2.7884, "step": 7790 }, { "epoch": 2.706168622523819e-14, "grad_norm": 0.6258195638656616, "learning_rate": 0.00012295587317279476, "loss": 2.7783, "step": 7800 }, { "epoch": 2.7096380694757727e-14, "grad_norm": 0.577340841293335, "learning_rate": 0.00012294989343011127, "loss": 2.7706, "step": 7810 }, { "epoch": 2.7131075164277263e-14, "grad_norm": 0.5984086394309998, "learning_rate": 0.0001229439051161812, "loss": 2.7765, "step": 7820 }, { "epoch": 2.71657696337968e-14, "grad_norm": 0.6325775980949402, "learning_rate": 0.00012293790823195153, "loss": 2.7736, "step": 7830 }, { "epoch": 2.7200464103316335e-14, "grad_norm": 0.6026847958564758, "learning_rate": 0.00012293190277837062, "loss": 2.7395, "step": 7840 }, { "epoch": 2.723515857283587e-14, "grad_norm": 0.6213665008544922, "learning_rate": 0.0001229258887563882, "loss": 2.7589, "step": 7850 }, { "epoch": 2.7269853042355408e-14, "grad_norm": 0.5980018973350525, "learning_rate": 0.0001229198661669553, "loss": 2.7864, "step": 7860 }, { "epoch": 2.7304547511874944e-14, "grad_norm": 0.5779904723167419, "learning_rate": 0.0001229138350110244, "loss": 2.7416, "step": 7870 }, { "epoch": 2.733924198139448e-14, "grad_norm": 0.6063337326049805, "learning_rate": 0.00012290779528954923, "loss": 2.7357, "step": 7880 }, { "epoch": 2.7373936450914016e-14, "grad_norm": 0.5802647471427917, "learning_rate": 0.0001229017470034849, "loss": 2.7885, "step": 7890 }, { "epoch": 2.7408630920433552e-14, "grad_norm": 0.6271678805351257, "learning_rate": 0.00012289569015378794, "loss": 2.7623, "step": 7900 }, { "epoch": 2.7443325389953088e-14, "grad_norm": 0.5970304608345032, "learning_rate": 0.00012288962474141615, "loss": 2.7606, "step": 7910 }, { "epoch": 2.7478019859472624e-14, "grad_norm": 0.6127659678459167, "learning_rate": 0.0001228835507673288, "loss": 2.7001, "step": 7920 }, { "epoch": 2.751271432899216e-14, "grad_norm": 0.558969259262085, "learning_rate": 0.00012287746823248634, "loss": 2.7885, "step": 7930 }, { "epoch": 2.7547408798511697e-14, "grad_norm": 0.5748438239097595, "learning_rate": 0.00012287137713785072, "loss": 2.7743, "step": 7940 }, { "epoch": 2.7582103268031233e-14, "grad_norm": 0.6025915741920471, "learning_rate": 0.00012286527748438516, "loss": 2.7956, "step": 7950 }, { "epoch": 2.761679773755077e-14, "grad_norm": 0.6111184358596802, "learning_rate": 0.00012285916927305432, "loss": 2.7556, "step": 7960 }, { "epoch": 2.7651492207070305e-14, "grad_norm": 0.5614121556282043, "learning_rate": 0.00012285305250482414, "loss": 2.7773, "step": 7970 }, { "epoch": 2.768618667658984e-14, "grad_norm": 0.6533632278442383, "learning_rate": 0.00012284692718066191, "loss": 2.7013, "step": 7980 }, { "epoch": 2.7720881146109377e-14, "grad_norm": 0.6663220524787903, "learning_rate": 0.00012284079330153634, "loss": 2.7684, "step": 7990 }, { "epoch": 2.7755575615628914e-14, "grad_norm": 0.6494764685630798, "learning_rate": 0.0001228346508684174, "loss": 2.7334, "step": 8000 }, { "epoch": 2.779027008514845e-14, "grad_norm": 0.624011218547821, "learning_rate": 0.0001228284998822765, "loss": 2.7389, "step": 8010 }, { "epoch": 2.7824964554667986e-14, "grad_norm": 0.5492384433746338, "learning_rate": 0.00012282234034408635, "loss": 2.77, "step": 8020 }, { "epoch": 2.7859659024187522e-14, "grad_norm": 0.6274101138114929, "learning_rate": 0.00012281617225482104, "loss": 2.7577, "step": 8030 }, { "epoch": 2.7894353493707058e-14, "grad_norm": 0.6603718400001526, "learning_rate": 0.00012280999561545597, "loss": 2.733, "step": 8040 }, { "epoch": 2.7929047963226594e-14, "grad_norm": 0.6220136880874634, "learning_rate": 0.000122803810426968, "loss": 2.7278, "step": 8050 }, { "epoch": 2.796374243274613e-14, "grad_norm": 0.6459081768989563, "learning_rate": 0.00012279761669033517, "loss": 2.7604, "step": 8060 }, { "epoch": 2.7998436902265667e-14, "grad_norm": 0.5705736875534058, "learning_rate": 0.000122791414406537, "loss": 2.7384, "step": 8070 }, { "epoch": 2.8033131371785203e-14, "grad_norm": 0.5645270943641663, "learning_rate": 0.00012278520357655433, "loss": 2.7478, "step": 8080 }, { "epoch": 2.806782584130474e-14, "grad_norm": 0.6190901398658752, "learning_rate": 0.0001227789842013694, "loss": 2.7671, "step": 8090 }, { "epoch": 2.8102520310824275e-14, "grad_norm": 0.5790607929229736, "learning_rate": 0.00012277275628196565, "loss": 2.7454, "step": 8100 }, { "epoch": 2.813721478034381e-14, "grad_norm": 0.5672294497489929, "learning_rate": 0.00012276651981932805, "loss": 2.7206, "step": 8110 }, { "epoch": 2.8171909249863347e-14, "grad_norm": 0.5600792765617371, "learning_rate": 0.00012276027481444282, "loss": 2.7372, "step": 8120 }, { "epoch": 2.8206603719382883e-14, "grad_norm": 0.5849307775497437, "learning_rate": 0.00012275402126829752, "loss": 2.7532, "step": 8130 }, { "epoch": 2.824129818890242e-14, "grad_norm": 0.5883999466896057, "learning_rate": 0.00012274775918188116, "loss": 2.7245, "step": 8140 }, { "epoch": 2.8275992658421956e-14, "grad_norm": 0.6170220375061035, "learning_rate": 0.000122741488556184, "loss": 2.6967, "step": 8150 }, { "epoch": 2.8310687127941492e-14, "grad_norm": 0.6291216611862183, "learning_rate": 0.00012273520939219768, "loss": 2.7956, "step": 8160 }, { "epoch": 2.8345381597461028e-14, "grad_norm": 0.5851698517799377, "learning_rate": 0.0001227289216909152, "loss": 2.7443, "step": 8170 }, { "epoch": 2.8380076066980564e-14, "grad_norm": 0.5850293636322021, "learning_rate": 0.0001227226254533309, "loss": 2.7149, "step": 8180 }, { "epoch": 2.84147705365001e-14, "grad_norm": 0.6075373291969299, "learning_rate": 0.0001227163206804405, "loss": 2.7778, "step": 8190 }, { "epoch": 2.8449465006019636e-14, "grad_norm": 0.6128525733947754, "learning_rate": 0.00012271000737324104, "loss": 2.685, "step": 8200 }, { "epoch": 2.848415947553917e-14, "grad_norm": 0.6673964858055115, "learning_rate": 0.00012270368553273086, "loss": 2.7608, "step": 8210 }, { "epoch": 2.851885394505871e-14, "grad_norm": 0.5610538721084595, "learning_rate": 0.0001226973551599098, "loss": 2.7181, "step": 8220 }, { "epoch": 2.8553548414578245e-14, "grad_norm": 0.605940043926239, "learning_rate": 0.00012269101625577886, "loss": 2.7726, "step": 8230 }, { "epoch": 2.858824288409778e-14, "grad_norm": 0.6409590244293213, "learning_rate": 0.00012268466882134056, "loss": 2.7471, "step": 8240 }, { "epoch": 2.862293735361732e-14, "grad_norm": 0.6503705382347107, "learning_rate": 0.00012267831285759864, "loss": 2.7417, "step": 8250 }, { "epoch": 2.8657631823136853e-14, "grad_norm": 0.5680670738220215, "learning_rate": 0.00012267194836555826, "loss": 2.732, "step": 8260 }, { "epoch": 2.869232629265639e-14, "grad_norm": 0.6083877086639404, "learning_rate": 0.00012266557534622592, "loss": 2.7409, "step": 8270 }, { "epoch": 2.8727020762175925e-14, "grad_norm": 0.5614429712295532, "learning_rate": 0.00012265919380060946, "loss": 2.7775, "step": 8280 }, { "epoch": 2.876171523169546e-14, "grad_norm": 0.617866039276123, "learning_rate": 0.00012265280372971804, "loss": 2.7278, "step": 8290 }, { "epoch": 2.8796409701215e-14, "grad_norm": 0.6277989745140076, "learning_rate": 0.0001226464051345622, "loss": 2.7529, "step": 8300 }, { "epoch": 2.8831104170734534e-14, "grad_norm": 0.5595750212669373, "learning_rate": 0.00012263999801615384, "loss": 2.7334, "step": 8310 }, { "epoch": 2.886579864025407e-14, "grad_norm": 0.5465537905693054, "learning_rate": 0.00012263358237550617, "loss": 2.7358, "step": 8320 }, { "epoch": 2.8900493109773606e-14, "grad_norm": 0.6243159174919128, "learning_rate": 0.0001226271582136338, "loss": 2.7752, "step": 8330 }, { "epoch": 2.893518757929314e-14, "grad_norm": 0.5803197622299194, "learning_rate": 0.0001226207255315526, "loss": 2.7699, "step": 8340 }, { "epoch": 2.896988204881268e-14, "grad_norm": 0.5915257930755615, "learning_rate": 0.00012261428433027992, "loss": 2.7449, "step": 8350 }, { "epoch": 2.9004576518332215e-14, "grad_norm": 0.5839261412620544, "learning_rate": 0.00012260783461083433, "loss": 2.7751, "step": 8360 }, { "epoch": 2.903927098785175e-14, "grad_norm": 0.7227705717086792, "learning_rate": 0.0001226013763742358, "loss": 2.7591, "step": 8370 }, { "epoch": 2.907396545737129e-14, "grad_norm": 0.5660240650177002, "learning_rate": 0.00012259490962150565, "loss": 2.7397, "step": 8380 }, { "epoch": 2.910865992689082e-14, "grad_norm": 0.6363061666488647, "learning_rate": 0.00012258843435366656, "loss": 2.7448, "step": 8390 }, { "epoch": 2.914335439641036e-14, "grad_norm": 0.6057271957397461, "learning_rate": 0.0001225819505717425, "loss": 2.7149, "step": 8400 }, { "epoch": 2.9178048865929895e-14, "grad_norm": 0.5713937878608704, "learning_rate": 0.00012257545827675886, "loss": 2.7466, "step": 8410 }, { "epoch": 2.921274333544943e-14, "grad_norm": 0.5718899965286255, "learning_rate": 0.00012256895746974234, "loss": 2.7149, "step": 8420 }, { "epoch": 2.924743780496897e-14, "grad_norm": 0.5998473167419434, "learning_rate": 0.00012256244815172092, "loss": 2.6777, "step": 8430 }, { "epoch": 2.9282132274488504e-14, "grad_norm": 0.5892776846885681, "learning_rate": 0.00012255593032372409, "loss": 2.7479, "step": 8440 }, { "epoch": 2.931682674400804e-14, "grad_norm": 0.5821269154548645, "learning_rate": 0.00012254940398678254, "loss": 2.734, "step": 8450 }, { "epoch": 2.9351521213527576e-14, "grad_norm": 0.5835372805595398, "learning_rate": 0.00012254286914192836, "loss": 2.7269, "step": 8460 }, { "epoch": 2.938621568304711e-14, "grad_norm": 0.5851445198059082, "learning_rate": 0.00012253632579019497, "loss": 2.7514, "step": 8470 }, { "epoch": 2.942091015256665e-14, "grad_norm": 0.6556898355484009, "learning_rate": 0.00012252977393261715, "loss": 2.741, "step": 8480 }, { "epoch": 2.9455604622086184e-14, "grad_norm": 0.590623676776886, "learning_rate": 0.000122523213570231, "loss": 2.6861, "step": 8490 }, { "epoch": 2.949029909160572e-14, "grad_norm": 0.5724770426750183, "learning_rate": 0.00012251664470407404, "loss": 2.7654, "step": 8500 }, { "epoch": 2.9524993561125257e-14, "grad_norm": 0.5692627429962158, "learning_rate": 0.00012251006733518503, "loss": 2.7257, "step": 8510 }, { "epoch": 2.955968803064479e-14, "grad_norm": 0.5844013690948486, "learning_rate": 0.00012250348146460415, "loss": 2.7177, "step": 8520 }, { "epoch": 2.959438250016433e-14, "grad_norm": 0.566704273223877, "learning_rate": 0.0001224968870933729, "loss": 2.7525, "step": 8530 }, { "epoch": 2.9629076969683865e-14, "grad_norm": 0.585543692111969, "learning_rate": 0.0001224902842225341, "loss": 2.7383, "step": 8540 }, { "epoch": 2.96637714392034e-14, "grad_norm": 0.6217796802520752, "learning_rate": 0.00012248367285313193, "loss": 2.7183, "step": 8550 }, { "epoch": 2.969846590872294e-14, "grad_norm": 0.6890034675598145, "learning_rate": 0.00012247705298621198, "loss": 2.722, "step": 8560 }, { "epoch": 2.9733160378242474e-14, "grad_norm": 0.6077756285667419, "learning_rate": 0.00012247042462282107, "loss": 2.7477, "step": 8570 }, { "epoch": 2.976785484776201e-14, "grad_norm": 0.6220186352729797, "learning_rate": 0.00012246378776400744, "loss": 2.6992, "step": 8580 }, { "epoch": 2.9802549317281546e-14, "grad_norm": 0.5710828304290771, "learning_rate": 0.00012245714241082063, "loss": 2.7624, "step": 8590 }, { "epoch": 2.983724378680108e-14, "grad_norm": 0.567408561706543, "learning_rate": 0.00012245048856431156, "loss": 2.7329, "step": 8600 }, { "epoch": 2.987193825632062e-14, "grad_norm": 0.6110029220581055, "learning_rate": 0.0001224438262255325, "loss": 2.6758, "step": 8610 }, { "epoch": 2.9906632725840154e-14, "grad_norm": 0.6251906752586365, "learning_rate": 0.00012243715539553702, "loss": 2.7018, "step": 8620 }, { "epoch": 2.994132719535969e-14, "grad_norm": 0.5519136190414429, "learning_rate": 0.00012243047607538005, "loss": 2.7108, "step": 8630 }, { "epoch": 2.9976021664879227e-14, "grad_norm": 0.5632822513580322, "learning_rate": 0.00012242378826611786, "loss": 2.7139, "step": 8640 }, { "epoch": 3.001071613439876e-14, "grad_norm": 0.6092585921287537, "learning_rate": 0.0001224170919688081, "loss": 2.725, "step": 8650 }, { "epoch": 3.00454106039183e-14, "grad_norm": 0.6034282445907593, "learning_rate": 0.0001224103871845097, "loss": 2.7626, "step": 8660 }, { "epoch": 3.0080105073437835e-14, "grad_norm": 0.5606814026832581, "learning_rate": 0.000122403673914283, "loss": 2.6922, "step": 8670 }, { "epoch": 3.011479954295737e-14, "grad_norm": 0.5960850715637207, "learning_rate": 0.0001223969521591896, "loss": 2.6765, "step": 8680 }, { "epoch": 3.014949401247691e-14, "grad_norm": 0.5511080622673035, "learning_rate": 0.00012239022192029253, "loss": 2.7584, "step": 8690 }, { "epoch": 3.0184188481996443e-14, "grad_norm": 0.6101716756820679, "learning_rate": 0.0001223834831986561, "loss": 2.7236, "step": 8700 }, { "epoch": 3.021888295151598e-14, "grad_norm": 0.624607264995575, "learning_rate": 0.00012237673599534596, "loss": 2.7553, "step": 8710 }, { "epoch": 3.0253577421035516e-14, "grad_norm": 0.5755387544631958, "learning_rate": 0.00012236998031142917, "loss": 2.7136, "step": 8720 }, { "epoch": 3.028827189055505e-14, "grad_norm": 0.5803837180137634, "learning_rate": 0.00012236321614797407, "loss": 2.6449, "step": 8730 }, { "epoch": 3.032296636007459e-14, "grad_norm": 0.5518720746040344, "learning_rate": 0.00012235644350605033, "loss": 2.7298, "step": 8740 }, { "epoch": 3.0357660829594124e-14, "grad_norm": 0.5799957513809204, "learning_rate": 0.00012234966238672897, "loss": 2.7526, "step": 8750 }, { "epoch": 3.039235529911366e-14, "grad_norm": 0.708888828754425, "learning_rate": 0.0001223428727910824, "loss": 2.744, "step": 8760 }, { "epoch": 3.0427049768633196e-14, "grad_norm": 0.5876485109329224, "learning_rate": 0.00012233607472018437, "loss": 2.7345, "step": 8770 }, { "epoch": 3.046174423815273e-14, "grad_norm": 0.6189063787460327, "learning_rate": 0.00012232926817510987, "loss": 2.7278, "step": 8780 }, { "epoch": 3.049643870767227e-14, "grad_norm": 0.5861189961433411, "learning_rate": 0.0001223224531569353, "loss": 2.6509, "step": 8790 }, { "epoch": 3.0531133177191805e-14, "grad_norm": 0.5623735189437866, "learning_rate": 0.00012231562966673847, "loss": 2.7257, "step": 8800 }, { "epoch": 3.056582764671134e-14, "grad_norm": 1.2481086254119873, "learning_rate": 0.00012230879770559834, "loss": 2.74, "step": 8810 }, { "epoch": 3.060052211623088e-14, "grad_norm": 0.6280727982521057, "learning_rate": 0.00012230195727459543, "loss": 2.7404, "step": 8820 }, { "epoch": 3.0635216585750413e-14, "grad_norm": 0.5677340030670166, "learning_rate": 0.00012229510837481145, "loss": 2.7261, "step": 8830 }, { "epoch": 3.066991105526995e-14, "grad_norm": 0.5785607099533081, "learning_rate": 0.0001222882510073295, "loss": 2.7085, "step": 8840 }, { "epoch": 3.0704605524789486e-14, "grad_norm": 0.5807548761367798, "learning_rate": 0.00012228138517323398, "loss": 2.7537, "step": 8850 }, { "epoch": 3.073929999430902e-14, "grad_norm": 0.5745427012443542, "learning_rate": 0.00012227451087361074, "loss": 2.7004, "step": 8860 }, { "epoch": 3.077399446382856e-14, "grad_norm": 0.571537435054779, "learning_rate": 0.00012226762810954684, "loss": 2.7365, "step": 8870 }, { "epoch": 3.0808688933348094e-14, "grad_norm": 0.5421806573867798, "learning_rate": 0.00012226073688213074, "loss": 2.7508, "step": 8880 }, { "epoch": 3.084338340286763e-14, "grad_norm": 0.5521705746650696, "learning_rate": 0.0001222538371924522, "loss": 2.6762, "step": 8890 }, { "epoch": 3.0878077872387166e-14, "grad_norm": 0.5346919894218445, "learning_rate": 0.0001222469290416024, "loss": 2.7107, "step": 8900 }, { "epoch": 3.09127723419067e-14, "grad_norm": 0.6001712083816528, "learning_rate": 0.00012224001243067375, "loss": 2.6955, "step": 8910 }, { "epoch": 3.094746681142624e-14, "grad_norm": 0.5604910254478455, "learning_rate": 0.00012223308736076009, "loss": 2.7299, "step": 8920 }, { "epoch": 3.0982161280945775e-14, "grad_norm": 0.5772704482078552, "learning_rate": 0.00012222615383295656, "loss": 2.7112, "step": 8930 }, { "epoch": 3.101685575046531e-14, "grad_norm": 0.5923368334770203, "learning_rate": 0.00012221921184835962, "loss": 2.6758, "step": 8940 }, { "epoch": 3.105155021998485e-14, "grad_norm": 0.5484918355941772, "learning_rate": 0.0001222122614080671, "loss": 2.6498, "step": 8950 }, { "epoch": 3.108624468950438e-14, "grad_norm": 0.5630719065666199, "learning_rate": 0.00012220530251317812, "loss": 2.6971, "step": 8960 }, { "epoch": 3.112093915902392e-14, "grad_norm": 0.5648552775382996, "learning_rate": 0.00012219833516479323, "loss": 2.7187, "step": 8970 }, { "epoch": 3.1155633628543455e-14, "grad_norm": 0.5888478755950928, "learning_rate": 0.0001221913593640142, "loss": 2.7029, "step": 8980 }, { "epoch": 3.119032809806299e-14, "grad_norm": 0.5593474507331848, "learning_rate": 0.00012218437511194423, "loss": 2.6742, "step": 8990 }, { "epoch": 3.122502256758253e-14, "grad_norm": 0.5743793845176697, "learning_rate": 0.0001221773824096878, "loss": 2.7001, "step": 9000 }, { "epoch": 3.1259717037102064e-14, "grad_norm": 0.6176651120185852, "learning_rate": 0.00012217038125835071, "loss": 2.7313, "step": 9010 }, { "epoch": 3.12944115066216e-14, "grad_norm": 0.6372074484825134, "learning_rate": 0.0001221633716590402, "loss": 2.7505, "step": 9020 }, { "epoch": 3.1329105976141136e-14, "grad_norm": 0.6081959009170532, "learning_rate": 0.00012215635361286474, "loss": 2.7486, "step": 9030 }, { "epoch": 3.136380044566067e-14, "grad_norm": 0.5636699199676514, "learning_rate": 0.00012214932712093418, "loss": 2.7276, "step": 9040 }, { "epoch": 3.139849491518021e-14, "grad_norm": 0.6293176412582397, "learning_rate": 0.0001221422921843597, "loss": 2.7396, "step": 9050 }, { "epoch": 3.1433189384699745e-14, "grad_norm": 0.5791905522346497, "learning_rate": 0.00012213524880425383, "loss": 2.7555, "step": 9060 }, { "epoch": 3.146788385421928e-14, "grad_norm": 0.5599613189697266, "learning_rate": 0.0001221281969817304, "loss": 2.7389, "step": 9070 }, { "epoch": 3.1502578323738817e-14, "grad_norm": 0.6040120720863342, "learning_rate": 0.0001221211367179046, "loss": 2.6862, "step": 9080 }, { "epoch": 3.153727279325835e-14, "grad_norm": 0.6232734322547913, "learning_rate": 0.00012211406801389293, "loss": 2.6781, "step": 9090 }, { "epoch": 3.157196726277789e-14, "grad_norm": 0.5845705270767212, "learning_rate": 0.0001221069908708133, "loss": 2.7119, "step": 9100 }, { "epoch": 3.1606661732297425e-14, "grad_norm": 0.6415331959724426, "learning_rate": 0.0001220999052897848, "loss": 2.7123, "step": 9110 }, { "epoch": 3.164135620181696e-14, "grad_norm": 0.6314709186553955, "learning_rate": 0.00012209281127192808, "loss": 2.6999, "step": 9120 }, { "epoch": 3.16760506713365e-14, "grad_norm": 0.5718731880187988, "learning_rate": 0.00012208570881836492, "loss": 2.7078, "step": 9130 }, { "epoch": 3.1710745140856034e-14, "grad_norm": 0.621759831905365, "learning_rate": 0.00012207859793021851, "loss": 2.7132, "step": 9140 }, { "epoch": 3.174543961037557e-14, "grad_norm": 0.5511876940727234, "learning_rate": 0.00012207147860861343, "loss": 2.728, "step": 9150 }, { "epoch": 3.1780134079895106e-14, "grad_norm": 0.5818421840667725, "learning_rate": 0.00012206435085467546, "loss": 2.7007, "step": 9160 }, { "epoch": 3.181482854941464e-14, "grad_norm": 0.6205571889877319, "learning_rate": 0.00012205721466953188, "loss": 2.6789, "step": 9170 }, { "epoch": 3.184952301893418e-14, "grad_norm": 0.5689705014228821, "learning_rate": 0.00012205007005431113, "loss": 2.7167, "step": 9180 }, { "epoch": 3.1884217488453714e-14, "grad_norm": 0.5966915488243103, "learning_rate": 0.00012204291701014312, "loss": 2.7275, "step": 9190 }, { "epoch": 3.191891195797325e-14, "grad_norm": 0.623271644115448, "learning_rate": 0.00012203575553815905, "loss": 2.6639, "step": 9200 }, { "epoch": 3.1953606427492787e-14, "grad_norm": 0.5837920308113098, "learning_rate": 0.00012202858563949141, "loss": 2.7065, "step": 9210 }, { "epoch": 3.198830089701232e-14, "grad_norm": 0.5521780252456665, "learning_rate": 0.0001220214073152741, "loss": 2.6737, "step": 9220 }, { "epoch": 3.202299536653186e-14, "grad_norm": 0.5871712565422058, "learning_rate": 0.00012201422056664228, "loss": 2.6877, "step": 9230 }, { "epoch": 3.2057689836051395e-14, "grad_norm": 0.5816060900688171, "learning_rate": 0.00012200702539473246, "loss": 2.7111, "step": 9240 }, { "epoch": 3.209238430557093e-14, "grad_norm": 0.634639322757721, "learning_rate": 0.00012199982180068253, "loss": 2.6752, "step": 9250 }, { "epoch": 3.212707877509047e-14, "grad_norm": 0.549468994140625, "learning_rate": 0.00012199260978563165, "loss": 2.6834, "step": 9260 }, { "epoch": 3.2161773244610004e-14, "grad_norm": 0.6105870008468628, "learning_rate": 0.00012198538935072034, "loss": 2.6939, "step": 9270 }, { "epoch": 3.219646771412954e-14, "grad_norm": 0.6179273724555969, "learning_rate": 0.00012197816049709049, "loss": 2.6807, "step": 9280 }, { "epoch": 3.2231162183649076e-14, "grad_norm": 0.5950314998626709, "learning_rate": 0.00012197092322588523, "loss": 2.693, "step": 9290 }, { "epoch": 3.226585665316861e-14, "grad_norm": 0.5621364712715149, "learning_rate": 0.00012196367753824908, "loss": 2.7024, "step": 9300 }, { "epoch": 3.230055112268815e-14, "grad_norm": 0.5971581339836121, "learning_rate": 0.0001219564234353279, "loss": 2.6937, "step": 9310 }, { "epoch": 3.2335245592207684e-14, "grad_norm": 0.5951065421104431, "learning_rate": 0.00012194916091826886, "loss": 2.6743, "step": 9320 }, { "epoch": 3.236994006172722e-14, "grad_norm": 0.5457784533500671, "learning_rate": 0.00012194188998822047, "loss": 2.6907, "step": 9330 }, { "epoch": 3.2404634531246757e-14, "grad_norm": 0.5363090634346008, "learning_rate": 0.00012193461064633255, "loss": 2.682, "step": 9340 }, { "epoch": 3.243932900076629e-14, "grad_norm": 0.577541708946228, "learning_rate": 0.00012192732289375626, "loss": 2.7062, "step": 9350 }, { "epoch": 3.247402347028583e-14, "grad_norm": 0.5671541690826416, "learning_rate": 0.00012192002673164413, "loss": 2.6875, "step": 9360 }, { "epoch": 3.2508717939805365e-14, "grad_norm": 0.5511887073516846, "learning_rate": 0.00012191272216114995, "loss": 2.7233, "step": 9370 }, { "epoch": 3.25434124093249e-14, "grad_norm": 0.5917375683784485, "learning_rate": 0.0001219054091834289, "loss": 2.7159, "step": 9380 }, { "epoch": 3.257810687884444e-14, "grad_norm": 0.6373002529144287, "learning_rate": 0.00012189808779963744, "loss": 2.7144, "step": 9390 }, { "epoch": 3.2612801348363973e-14, "grad_norm": 0.5870112776756287, "learning_rate": 0.00012189075801093341, "loss": 2.7207, "step": 9400 }, { "epoch": 3.264749581788351e-14, "grad_norm": 0.6011103391647339, "learning_rate": 0.00012188341981847592, "loss": 2.7116, "step": 9410 }, { "epoch": 3.2682190287403046e-14, "grad_norm": 0.5740630030632019, "learning_rate": 0.00012187607322342546, "loss": 2.6676, "step": 9420 }, { "epoch": 3.271688475692258e-14, "grad_norm": 0.5608379244804382, "learning_rate": 0.00012186871822694385, "loss": 2.713, "step": 9430 }, { "epoch": 3.275157922644212e-14, "grad_norm": 0.5525686740875244, "learning_rate": 0.0001218613548301942, "loss": 2.682, "step": 9440 }, { "epoch": 3.2786273695961654e-14, "grad_norm": 0.5498374104499817, "learning_rate": 0.00012185398303434098, "loss": 2.6981, "step": 9450 }, { "epoch": 3.282096816548119e-14, "grad_norm": 0.5598671436309814, "learning_rate": 0.00012184660284054994, "loss": 2.6716, "step": 9460 }, { "epoch": 3.2855662635000726e-14, "grad_norm": 0.5362412929534912, "learning_rate": 0.00012183921424998823, "loss": 2.6444, "step": 9470 }, { "epoch": 3.289035710452026e-14, "grad_norm": 0.5831873416900635, "learning_rate": 0.00012183181726382428, "loss": 2.6865, "step": 9480 }, { "epoch": 3.29250515740398e-14, "grad_norm": 0.5916926860809326, "learning_rate": 0.00012182441188322785, "loss": 2.685, "step": 9490 }, { "epoch": 3.2959746043559335e-14, "grad_norm": 0.5916888117790222, "learning_rate": 0.00012181699810937007, "loss": 2.668, "step": 9500 }, { "epoch": 3.299444051307887e-14, "grad_norm": 0.6125104427337646, "learning_rate": 0.00012180957594342334, "loss": 2.6832, "step": 9510 }, { "epoch": 3.302913498259841e-14, "grad_norm": 0.5710282921791077, "learning_rate": 0.0001218021453865614, "loss": 2.6799, "step": 9520 }, { "epoch": 3.3063829452117943e-14, "grad_norm": 0.6019191145896912, "learning_rate": 0.00012179470643995935, "loss": 2.7126, "step": 9530 }, { "epoch": 3.309852392163748e-14, "grad_norm": 0.5669811367988586, "learning_rate": 0.00012178725910479359, "loss": 2.6851, "step": 9540 }, { "epoch": 3.3133218391157016e-14, "grad_norm": 0.6142165660858154, "learning_rate": 0.00012177980338224186, "loss": 2.701, "step": 9550 }, { "epoch": 3.316791286067655e-14, "grad_norm": 0.5776997208595276, "learning_rate": 0.0001217723392734832, "loss": 2.6216, "step": 9560 }, { "epoch": 3.320260733019609e-14, "grad_norm": 0.619737446308136, "learning_rate": 0.00012176486677969799, "loss": 2.6508, "step": 9570 }, { "epoch": 3.3237301799715624e-14, "grad_norm": 0.5746952295303345, "learning_rate": 0.00012175738590206796, "loss": 2.6668, "step": 9580 }, { "epoch": 3.327199626923516e-14, "grad_norm": 0.6124783754348755, "learning_rate": 0.00012174989664177614, "loss": 2.718, "step": 9590 }, { "epoch": 3.3306690738754696e-14, "grad_norm": 0.5837702751159668, "learning_rate": 0.00012174239900000692, "loss": 2.7008, "step": 9600 }, { "epoch": 3.334138520827423e-14, "grad_norm": 0.5223130583763123, "learning_rate": 0.00012173489297794594, "loss": 2.6723, "step": 9610 }, { "epoch": 3.337607967779377e-14, "grad_norm": 0.5533768534660339, "learning_rate": 0.00012172737857678022, "loss": 2.6501, "step": 9620 }, { "epoch": 3.3410774147313305e-14, "grad_norm": 0.570774495601654, "learning_rate": 0.00012171985579769812, "loss": 2.7115, "step": 9630 }, { "epoch": 3.344546861683284e-14, "grad_norm": 0.5782768130302429, "learning_rate": 0.00012171232464188929, "loss": 2.7095, "step": 9640 }, { "epoch": 3.348016308635238e-14, "grad_norm": 0.5455970168113708, "learning_rate": 0.00012170478511054472, "loss": 2.6928, "step": 9650 }, { "epoch": 3.351485755587191e-14, "grad_norm": 0.5745911598205566, "learning_rate": 0.00012169723720485675, "loss": 2.6883, "step": 9660 }, { "epoch": 3.354955202539145e-14, "grad_norm": 0.5536209940910339, "learning_rate": 0.00012168968092601897, "loss": 2.7121, "step": 9670 }, { "epoch": 3.3584246494910985e-14, "grad_norm": 0.6238003373146057, "learning_rate": 0.00012168211627522635, "loss": 2.6781, "step": 9680 }, { "epoch": 3.361894096443052e-14, "grad_norm": 0.6139904260635376, "learning_rate": 0.00012167454325367519, "loss": 2.7177, "step": 9690 }, { "epoch": 3.365363543395006e-14, "grad_norm": 0.5982169508934021, "learning_rate": 0.00012166696186256309, "loss": 2.6694, "step": 9700 }, { "epoch": 3.3688329903469594e-14, "grad_norm": 0.5608279705047607, "learning_rate": 0.00012165937210308899, "loss": 2.6571, "step": 9710 }, { "epoch": 3.372302437298913e-14, "grad_norm": 0.5683742165565491, "learning_rate": 0.00012165177397645312, "loss": 2.6828, "step": 9720 }, { "epoch": 3.3757718842508666e-14, "grad_norm": 0.5868356823921204, "learning_rate": 0.00012164416748385711, "loss": 2.6591, "step": 9730 }, { "epoch": 3.37924133120282e-14, "grad_norm": 0.5818955302238464, "learning_rate": 0.0001216365526265038, "loss": 2.695, "step": 9740 }, { "epoch": 3.382710778154774e-14, "grad_norm": 0.5848982334136963, "learning_rate": 0.00012162892940559745, "loss": 2.7156, "step": 9750 }, { "epoch": 3.3861802251067274e-14, "grad_norm": 0.5479873418807983, "learning_rate": 0.00012162129782234362, "loss": 2.6976, "step": 9760 }, { "epoch": 3.389649672058681e-14, "grad_norm": 0.6345406770706177, "learning_rate": 0.00012161365787794913, "loss": 2.6746, "step": 9770 }, { "epoch": 3.3931191190106347e-14, "grad_norm": 0.5861382484436035, "learning_rate": 0.00012160600957362222, "loss": 2.7152, "step": 9780 }, { "epoch": 3.396588565962588e-14, "grad_norm": 0.5547277331352234, "learning_rate": 0.00012159835291057236, "loss": 2.673, "step": 9790 }, { "epoch": 3.400058012914542e-14, "grad_norm": 0.5698844194412231, "learning_rate": 0.00012159068789001044, "loss": 2.7139, "step": 9800 }, { "epoch": 3.4035274598664955e-14, "grad_norm": 0.8091624975204468, "learning_rate": 0.00012158301451314855, "loss": 2.6552, "step": 9810 }, { "epoch": 3.406996906818449e-14, "grad_norm": 0.5570766925811768, "learning_rate": 0.00012157533278120023, "loss": 2.6672, "step": 9820 }, { "epoch": 3.410466353770403e-14, "grad_norm": 0.5472370386123657, "learning_rate": 0.00012156764269538026, "loss": 2.6835, "step": 9830 }, { "epoch": 3.4139358007223564e-14, "grad_norm": 0.6653771996498108, "learning_rate": 0.00012155994425690474, "loss": 2.6816, "step": 9840 }, { "epoch": 3.41740524767431e-14, "grad_norm": 0.5860264301300049, "learning_rate": 0.00012155223746699113, "loss": 2.6924, "step": 9850 }, { "epoch": 3.4208746946262636e-14, "grad_norm": 0.6670499444007874, "learning_rate": 0.00012154452232685818, "loss": 2.7091, "step": 9860 }, { "epoch": 3.424344141578217e-14, "grad_norm": 0.6064344644546509, "learning_rate": 0.000121536798837726, "loss": 2.6724, "step": 9870 }, { "epoch": 3.427813588530171e-14, "grad_norm": 0.6080465912818909, "learning_rate": 0.00012152906700081597, "loss": 2.6526, "step": 9880 }, { "epoch": 3.4312830354821244e-14, "grad_norm": 0.5855628848075867, "learning_rate": 0.00012152132681735083, "loss": 2.7032, "step": 9890 }, { "epoch": 3.434752482434078e-14, "grad_norm": 0.6088993549346924, "learning_rate": 0.00012151357828855459, "loss": 2.7014, "step": 9900 }, { "epoch": 3.4382219293860317e-14, "grad_norm": 0.5548629760742188, "learning_rate": 0.00012150582141565264, "loss": 2.6475, "step": 9910 }, { "epoch": 3.441691376337985e-14, "grad_norm": 0.566682755947113, "learning_rate": 0.00012149805619987167, "loss": 2.6986, "step": 9920 }, { "epoch": 3.445160823289939e-14, "grad_norm": 0.578998863697052, "learning_rate": 0.00012149028264243966, "loss": 2.7094, "step": 9930 }, { "epoch": 3.4486302702418925e-14, "grad_norm": 0.5747069716453552, "learning_rate": 0.00012148250074458596, "loss": 2.6475, "step": 9940 }, { "epoch": 3.452099717193846e-14, "grad_norm": 0.5545374155044556, "learning_rate": 0.00012147471050754117, "loss": 2.6687, "step": 9950 }, { "epoch": 3.4555691641458e-14, "grad_norm": 0.5327395796775818, "learning_rate": 0.00012146691193253729, "loss": 2.6715, "step": 9960 }, { "epoch": 3.4590386110977533e-14, "grad_norm": 0.5546936988830566, "learning_rate": 0.00012145910502080757, "loss": 2.6394, "step": 9970 }, { "epoch": 3.462508058049707e-14, "grad_norm": 0.5563433766365051, "learning_rate": 0.0001214512897735866, "loss": 2.6754, "step": 9980 }, { "epoch": 3.4659775050016606e-14, "grad_norm": 0.540101945400238, "learning_rate": 0.00012144346619211032, "loss": 2.6905, "step": 9990 }, { "epoch": 3.469446951953614e-14, "grad_norm": 0.5790048241615295, "learning_rate": 0.00012143563427761595, "loss": 2.7047, "step": 10000 }, { "epoch": 3.472916398905568e-14, "grad_norm": 0.5508939027786255, "learning_rate": 0.00012142779403134204, "loss": 2.6852, "step": 10010 }, { "epoch": 3.4763858458575214e-14, "grad_norm": 0.5846458673477173, "learning_rate": 0.00012141994545452845, "loss": 2.6838, "step": 10020 }, { "epoch": 3.479855292809475e-14, "grad_norm": 0.5380244255065918, "learning_rate": 0.00012141208854841637, "loss": 2.6712, "step": 10030 }, { "epoch": 3.4833247397614286e-14, "grad_norm": 0.5601175427436829, "learning_rate": 0.00012140422331424831, "loss": 2.6569, "step": 10040 }, { "epoch": 3.486794186713382e-14, "grad_norm": 0.5798640847206116, "learning_rate": 0.0001213963497532681, "loss": 2.6627, "step": 10050 }, { "epoch": 3.490263633665336e-14, "grad_norm": 0.5623951554298401, "learning_rate": 0.00012138846786672085, "loss": 2.6673, "step": 10060 }, { "epoch": 3.4937330806172895e-14, "grad_norm": 0.5707474946975708, "learning_rate": 0.00012138057765585302, "loss": 2.7051, "step": 10070 }, { "epoch": 3.497202527569243e-14, "grad_norm": 0.53314608335495, "learning_rate": 0.00012137267912191237, "loss": 2.6422, "step": 10080 }, { "epoch": 3.500671974521197e-14, "grad_norm": 0.5453944802284241, "learning_rate": 0.00012136477226614802, "loss": 2.6818, "step": 10090 }, { "epoch": 3.5041414214731503e-14, "grad_norm": 0.5814880728721619, "learning_rate": 0.00012135685708981033, "loss": 2.685, "step": 10100 }, { "epoch": 3.507610868425104e-14, "grad_norm": 0.56483393907547, "learning_rate": 0.00012134893359415103, "loss": 2.6788, "step": 10110 }, { "epoch": 3.5110803153770576e-14, "grad_norm": 0.5587108731269836, "learning_rate": 0.00012134100178042318, "loss": 2.6658, "step": 10120 }, { "epoch": 3.514549762329011e-14, "grad_norm": 0.5657208561897278, "learning_rate": 0.00012133306164988107, "loss": 2.7003, "step": 10130 }, { "epoch": 3.518019209280965e-14, "grad_norm": 0.6640677452087402, "learning_rate": 0.00012132511320378042, "loss": 2.6484, "step": 10140 }, { "epoch": 3.5214886562329184e-14, "grad_norm": 0.5597197413444519, "learning_rate": 0.00012131715644337818, "loss": 2.6834, "step": 10150 }, { "epoch": 3.524958103184872e-14, "grad_norm": 0.5415320992469788, "learning_rate": 0.00012130919136993265, "loss": 2.6685, "step": 10160 }, { "epoch": 3.5284275501368256e-14, "grad_norm": 0.5572232007980347, "learning_rate": 0.00012130121798470343, "loss": 2.7016, "step": 10170 }, { "epoch": 3.531896997088779e-14, "grad_norm": 0.5812506675720215, "learning_rate": 0.00012129323628895148, "loss": 2.7074, "step": 10180 }, { "epoch": 3.535366444040733e-14, "grad_norm": 0.5825610756874084, "learning_rate": 0.00012128524628393899, "loss": 2.6977, "step": 10190 }, { "epoch": 3.5388358909926865e-14, "grad_norm": 0.5504910349845886, "learning_rate": 0.00012127724797092951, "loss": 2.6921, "step": 10200 }, { "epoch": 3.54230533794464e-14, "grad_norm": 0.5425430536270142, "learning_rate": 0.00012126924135118794, "loss": 2.6729, "step": 10210 }, { "epoch": 3.545774784896594e-14, "grad_norm": 0.5812125205993652, "learning_rate": 0.00012126122642598044, "loss": 2.6814, "step": 10220 }, { "epoch": 3.549244231848547e-14, "grad_norm": 0.5435065627098083, "learning_rate": 0.00012125320319657447, "loss": 2.6638, "step": 10230 }, { "epoch": 3.552713678800501e-14, "grad_norm": 0.5627261400222778, "learning_rate": 0.0001212451716642389, "loss": 2.6627, "step": 10240 }, { "epoch": 3.5561831257524545e-14, "grad_norm": 0.6183850765228271, "learning_rate": 0.0001212371318302438, "loss": 2.6387, "step": 10250 }, { "epoch": 3.559652572704408e-14, "grad_norm": 0.566496729850769, "learning_rate": 0.00012122908369586063, "loss": 2.67, "step": 10260 }, { "epoch": 3.563122019656362e-14, "grad_norm": 0.5774244070053101, "learning_rate": 0.00012122102726236208, "loss": 2.685, "step": 10270 }, { "epoch": 3.5665914666083154e-14, "grad_norm": 0.5807313919067383, "learning_rate": 0.00012121296253102228, "loss": 2.6629, "step": 10280 }, { "epoch": 3.570060913560269e-14, "grad_norm": 0.5669015645980835, "learning_rate": 0.00012120488950311654, "loss": 2.6798, "step": 10290 }, { "epoch": 3.5735303605122226e-14, "grad_norm": 0.5510312914848328, "learning_rate": 0.00012119680817992155, "loss": 2.6609, "step": 10300 }, { "epoch": 3.576999807464176e-14, "grad_norm": 0.5884196758270264, "learning_rate": 0.00012118871856271533, "loss": 2.6313, "step": 10310 }, { "epoch": 3.58046925441613e-14, "grad_norm": 0.5801659822463989, "learning_rate": 0.00012118062065277717, "loss": 2.6356, "step": 10320 }, { "epoch": 3.5839387013680835e-14, "grad_norm": 0.5544509887695312, "learning_rate": 0.00012117251445138767, "loss": 2.6455, "step": 10330 }, { "epoch": 3.587408148320037e-14, "grad_norm": 0.5923222303390503, "learning_rate": 0.00012116439995982876, "loss": 2.6844, "step": 10340 }, { "epoch": 3.590877595271991e-14, "grad_norm": 0.6032731533050537, "learning_rate": 0.0001211562771793837, "loss": 2.6457, "step": 10350 }, { "epoch": 3.594347042223944e-14, "grad_norm": 0.6142696738243103, "learning_rate": 0.00012114814611133699, "loss": 2.6504, "step": 10360 }, { "epoch": 3.597816489175898e-14, "grad_norm": 0.567385196685791, "learning_rate": 0.00012114000675697455, "loss": 2.6404, "step": 10370 }, { "epoch": 3.6012859361278515e-14, "grad_norm": 0.5451346635818481, "learning_rate": 0.0001211318591175835, "loss": 2.6651, "step": 10380 }, { "epoch": 3.604755383079805e-14, "grad_norm": 0.5365930795669556, "learning_rate": 0.00012112370319445233, "loss": 2.6605, "step": 10390 }, { "epoch": 3.608224830031759e-14, "grad_norm": 0.534575879573822, "learning_rate": 0.00012111553898887088, "loss": 2.6609, "step": 10400 }, { "epoch": 3.6116942769837124e-14, "grad_norm": 0.5483484268188477, "learning_rate": 0.00012110736650213015, "loss": 2.6727, "step": 10410 }, { "epoch": 3.615163723935666e-14, "grad_norm": 0.5970900058746338, "learning_rate": 0.00012109918573552262, "loss": 2.6446, "step": 10420 }, { "epoch": 3.6186331708876196e-14, "grad_norm": 0.5492020845413208, "learning_rate": 0.000121090996690342, "loss": 2.6588, "step": 10430 }, { "epoch": 3.622102617839573e-14, "grad_norm": 0.5779070854187012, "learning_rate": 0.0001210827993678833, "loss": 2.6353, "step": 10440 }, { "epoch": 3.625572064791527e-14, "grad_norm": 0.5656790137290955, "learning_rate": 0.00012107459376944287, "loss": 2.6334, "step": 10450 }, { "epoch": 3.6290415117434804e-14, "grad_norm": 0.5465889573097229, "learning_rate": 0.00012106637989631834, "loss": 2.6689, "step": 10460 }, { "epoch": 3.632510958695434e-14, "grad_norm": 0.5334776639938354, "learning_rate": 0.00012105815774980869, "loss": 2.6558, "step": 10470 }, { "epoch": 3.6359804056473877e-14, "grad_norm": 0.5591927170753479, "learning_rate": 0.00012104992733121415, "loss": 2.6656, "step": 10480 }, { "epoch": 3.639449852599341e-14, "grad_norm": 0.6258677840232849, "learning_rate": 0.0001210416886418363, "loss": 2.6835, "step": 10490 }, { "epoch": 3.642919299551295e-14, "grad_norm": 0.6224485039710999, "learning_rate": 0.00012103344168297801, "loss": 2.7034, "step": 10500 }, { "epoch": 3.6463887465032485e-14, "grad_norm": 0.638816237449646, "learning_rate": 0.0001210251864559435, "loss": 2.6559, "step": 10510 }, { "epoch": 3.649858193455202e-14, "grad_norm": 0.5851678848266602, "learning_rate": 0.00012101692296203824, "loss": 2.663, "step": 10520 }, { "epoch": 3.653327640407156e-14, "grad_norm": 0.5923005938529968, "learning_rate": 0.00012100865120256903, "loss": 2.6559, "step": 10530 }, { "epoch": 3.6567970873591094e-14, "grad_norm": 0.5698672533035278, "learning_rate": 0.00012100037117884397, "loss": 2.6667, "step": 10540 }, { "epoch": 3.660266534311063e-14, "grad_norm": 0.5528374910354614, "learning_rate": 0.00012099208289217248, "loss": 2.6437, "step": 10550 }, { "epoch": 3.6637359812630166e-14, "grad_norm": 0.5654798150062561, "learning_rate": 0.00012098378634386528, "loss": 2.6279, "step": 10560 }, { "epoch": 3.66720542821497e-14, "grad_norm": 0.5696104168891907, "learning_rate": 0.00012097548153523442, "loss": 2.6067, "step": 10570 }, { "epoch": 3.670674875166924e-14, "grad_norm": 0.5288019776344299, "learning_rate": 0.00012096716846759318, "loss": 2.6252, "step": 10580 }, { "epoch": 3.6741443221188774e-14, "grad_norm": 0.5784745812416077, "learning_rate": 0.00012095884714225628, "loss": 2.6419, "step": 10590 }, { "epoch": 3.677613769070831e-14, "grad_norm": 0.6024763584136963, "learning_rate": 0.00012095051756053959, "loss": 2.6758, "step": 10600 }, { "epoch": 3.6810832160227847e-14, "grad_norm": 0.5749506950378418, "learning_rate": 0.00012094217972376042, "loss": 2.6881, "step": 10610 }, { "epoch": 3.684552662974738e-14, "grad_norm": 0.5717706084251404, "learning_rate": 0.00012093383363323727, "loss": 2.6221, "step": 10620 }, { "epoch": 3.688022109926692e-14, "grad_norm": 0.6194654703140259, "learning_rate": 0.00012092547929029004, "loss": 2.6723, "step": 10630 }, { "epoch": 3.6914915568786455e-14, "grad_norm": 0.5832949280738831, "learning_rate": 0.0001209171166962399, "loss": 2.6898, "step": 10640 }, { "epoch": 3.694961003830599e-14, "grad_norm": 0.6204744577407837, "learning_rate": 0.00012090874585240931, "loss": 2.6726, "step": 10650 }, { "epoch": 3.698430450782553e-14, "grad_norm": 0.5442847609519958, "learning_rate": 0.00012090036676012204, "loss": 2.6619, "step": 10660 }, { "epoch": 3.7018998977345063e-14, "grad_norm": 0.548698365688324, "learning_rate": 0.0001208919794207032, "loss": 2.6577, "step": 10670 }, { "epoch": 3.70536934468646e-14, "grad_norm": 0.6984880566596985, "learning_rate": 0.00012088358383547914, "loss": 2.6856, "step": 10680 }, { "epoch": 3.7088387916384136e-14, "grad_norm": 0.6136687994003296, "learning_rate": 0.00012087518000577756, "loss": 2.6888, "step": 10690 }, { "epoch": 3.712308238590367e-14, "grad_norm": 0.5529924035072327, "learning_rate": 0.00012086676793292747, "loss": 2.6579, "step": 10700 }, { "epoch": 3.715777685542321e-14, "grad_norm": 0.5829877853393555, "learning_rate": 0.00012085834761825914, "loss": 2.6954, "step": 10710 }, { "epoch": 3.7192471324942744e-14, "grad_norm": 0.5561708211898804, "learning_rate": 0.00012084991906310421, "loss": 2.6789, "step": 10720 }, { "epoch": 3.722716579446228e-14, "grad_norm": 0.5761757493019104, "learning_rate": 0.00012084148226879554, "loss": 2.6846, "step": 10730 }, { "epoch": 3.7261860263981816e-14, "grad_norm": 0.5632902383804321, "learning_rate": 0.00012083303723666738, "loss": 2.678, "step": 10740 }, { "epoch": 3.729655473350135e-14, "grad_norm": 0.591071367263794, "learning_rate": 0.00012082458396805518, "loss": 2.6991, "step": 10750 }, { "epoch": 3.733124920302089e-14, "grad_norm": 0.5634849071502686, "learning_rate": 0.00012081612246429581, "loss": 2.6427, "step": 10760 }, { "epoch": 3.7365943672540425e-14, "grad_norm": 0.5397931337356567, "learning_rate": 0.00012080765272672735, "loss": 2.6581, "step": 10770 }, { "epoch": 3.740063814205996e-14, "grad_norm": 0.5409976243972778, "learning_rate": 0.00012079917475668923, "loss": 2.6917, "step": 10780 }, { "epoch": 3.74353326115795e-14, "grad_norm": 0.5667703151702881, "learning_rate": 0.00012079068855552218, "loss": 2.6765, "step": 10790 }, { "epoch": 3.7470027081099033e-14, "grad_norm": 0.5640769600868225, "learning_rate": 0.00012078219412456818, "loss": 2.6227, "step": 10800 }, { "epoch": 3.750472155061857e-14, "grad_norm": 0.5736445188522339, "learning_rate": 0.0001207736914651706, "loss": 2.6176, "step": 10810 }, { "epoch": 3.7539416020138106e-14, "grad_norm": 0.5572129487991333, "learning_rate": 0.00012076518057867402, "loss": 2.6109, "step": 10820 }, { "epoch": 3.757411048965764e-14, "grad_norm": 0.5357043743133545, "learning_rate": 0.0001207566614664244, "loss": 2.6382, "step": 10830 }, { "epoch": 3.760880495917718e-14, "grad_norm": 0.566803514957428, "learning_rate": 0.00012074813412976891, "loss": 2.6547, "step": 10840 }, { "epoch": 3.7643499428696714e-14, "grad_norm": 0.5812789797782898, "learning_rate": 0.00012073959857005613, "loss": 2.6585, "step": 10850 }, { "epoch": 3.767819389821625e-14, "grad_norm": 0.5764651298522949, "learning_rate": 0.00012073105478863587, "loss": 2.6741, "step": 10860 }, { "epoch": 3.7712888367735786e-14, "grad_norm": 0.5509382486343384, "learning_rate": 0.00012072250278685925, "loss": 2.684, "step": 10870 }, { "epoch": 3.774758283725532e-14, "grad_norm": 0.5900574326515198, "learning_rate": 0.00012071394256607868, "loss": 2.686, "step": 10880 }, { "epoch": 3.778227730677486e-14, "grad_norm": 0.6026166677474976, "learning_rate": 0.00012070537412764793, "loss": 2.6643, "step": 10890 }, { "epoch": 3.7816971776294395e-14, "grad_norm": 0.5752912163734436, "learning_rate": 0.00012069679747292198, "loss": 2.6261, "step": 10900 }, { "epoch": 3.785166624581393e-14, "grad_norm": 0.5603379011154175, "learning_rate": 0.00012068821260325719, "loss": 2.6572, "step": 10910 }, { "epoch": 3.788636071533347e-14, "grad_norm": 0.5572377443313599, "learning_rate": 0.00012067961952001114, "loss": 2.6697, "step": 10920 }, { "epoch": 3.7921055184853e-14, "grad_norm": 0.5342600345611572, "learning_rate": 0.00012067101822454276, "loss": 2.634, "step": 10930 }, { "epoch": 3.795574965437254e-14, "grad_norm": 0.5681015253067017, "learning_rate": 0.00012066240871821234, "loss": 2.6557, "step": 10940 }, { "epoch": 3.7990444123892075e-14, "grad_norm": 0.5537115931510925, "learning_rate": 0.0001206537910023813, "loss": 2.6697, "step": 10950 }, { "epoch": 3.802513859341161e-14, "grad_norm": 0.5433657765388489, "learning_rate": 0.00012064516507841252, "loss": 2.638, "step": 10960 }, { "epoch": 3.805983306293115e-14, "grad_norm": 0.6463860273361206, "learning_rate": 0.00012063653094767011, "loss": 2.6424, "step": 10970 }, { "epoch": 3.8094527532450684e-14, "grad_norm": 0.5781089663505554, "learning_rate": 0.00012062788861151946, "loss": 2.6617, "step": 10980 }, { "epoch": 3.812922200197022e-14, "grad_norm": 0.5787239074707031, "learning_rate": 0.0001206192380713273, "loss": 2.6778, "step": 10990 }, { "epoch": 3.8163916471489756e-14, "grad_norm": 0.5605751872062683, "learning_rate": 0.00012061057932846161, "loss": 2.6747, "step": 11000 }, { "epoch": 3.819861094100929e-14, "grad_norm": 0.5613300204277039, "learning_rate": 0.00012060191238429175, "loss": 2.6408, "step": 11010 }, { "epoch": 3.823330541052883e-14, "grad_norm": 0.5740814208984375, "learning_rate": 0.00012059323724018828, "loss": 2.6296, "step": 11020 }, { "epoch": 3.8267999880048365e-14, "grad_norm": 0.5383687019348145, "learning_rate": 0.00012058455389752313, "loss": 2.6292, "step": 11030 }, { "epoch": 3.83026943495679e-14, "grad_norm": 0.5756822228431702, "learning_rate": 0.00012057586235766945, "loss": 2.6406, "step": 11040 }, { "epoch": 3.8337388819087437e-14, "grad_norm": 0.521607518196106, "learning_rate": 0.00012056716262200175, "loss": 2.6188, "step": 11050 }, { "epoch": 3.837208328860697e-14, "grad_norm": 0.567572832107544, "learning_rate": 0.00012055845469189585, "loss": 2.6317, "step": 11060 }, { "epoch": 3.840677775812651e-14, "grad_norm": 0.5961544513702393, "learning_rate": 0.00012054973856872879, "loss": 2.6609, "step": 11070 }, { "epoch": 3.8441472227646045e-14, "grad_norm": 0.5538514852523804, "learning_rate": 0.00012054101425387897, "loss": 2.6738, "step": 11080 }, { "epoch": 3.847616669716558e-14, "grad_norm": 0.5784700512886047, "learning_rate": 0.00012053228174872609, "loss": 2.6545, "step": 11090 }, { "epoch": 3.851086116668512e-14, "grad_norm": 0.574701189994812, "learning_rate": 0.00012052354105465107, "loss": 2.6297, "step": 11100 }, { "epoch": 3.8545555636204654e-14, "grad_norm": 0.6799556016921997, "learning_rate": 0.00012051479217303624, "loss": 2.6019, "step": 11110 }, { "epoch": 3.858025010572419e-14, "grad_norm": 0.5675702691078186, "learning_rate": 0.00012050603510526509, "loss": 2.6273, "step": 11120 }, { "epoch": 3.8614944575243726e-14, "grad_norm": 0.5607612729072571, "learning_rate": 0.00012049726985272252, "loss": 2.6344, "step": 11130 }, { "epoch": 3.864963904476326e-14, "grad_norm": 0.5474817156791687, "learning_rate": 0.00012048849641679466, "loss": 2.6449, "step": 11140 }, { "epoch": 3.86843335142828e-14, "grad_norm": 0.6409746408462524, "learning_rate": 0.00012047971479886896, "loss": 2.6313, "step": 11150 }, { "epoch": 3.8719027983802334e-14, "grad_norm": 0.5643471479415894, "learning_rate": 0.00012047092500033416, "loss": 2.6686, "step": 11160 }, { "epoch": 3.875372245332187e-14, "grad_norm": 0.5952200889587402, "learning_rate": 0.00012046212702258028, "loss": 2.6282, "step": 11170 }, { "epoch": 3.8788416922841407e-14, "grad_norm": 0.6142899990081787, "learning_rate": 0.00012045332086699865, "loss": 2.6347, "step": 11180 }, { "epoch": 3.882311139236094e-14, "grad_norm": 0.5656185150146484, "learning_rate": 0.00012044450653498191, "loss": 2.6513, "step": 11190 }, { "epoch": 3.885780586188048e-14, "grad_norm": 0.5427243113517761, "learning_rate": 0.00012043568402792392, "loss": 2.6338, "step": 11200 }, { "epoch": 3.8892500331400015e-14, "grad_norm": 0.5593475699424744, "learning_rate": 0.00012042685334721996, "loss": 2.6539, "step": 11210 }, { "epoch": 3.892719480091955e-14, "grad_norm": 0.5298470854759216, "learning_rate": 0.00012041801449426645, "loss": 2.6334, "step": 11220 }, { "epoch": 3.896188927043909e-14, "grad_norm": 0.5626364946365356, "learning_rate": 0.0001204091674704612, "loss": 2.6507, "step": 11230 }, { "epoch": 3.8996583739958623e-14, "grad_norm": 0.5656890869140625, "learning_rate": 0.00012040031227720331, "loss": 2.6504, "step": 11240 }, { "epoch": 3.903127820947816e-14, "grad_norm": 0.5670046210289001, "learning_rate": 0.00012039144891589315, "loss": 2.6238, "step": 11250 }, { "epoch": 3.9065972678997696e-14, "grad_norm": 0.5892581939697266, "learning_rate": 0.00012038257738793237, "loss": 2.6296, "step": 11260 }, { "epoch": 3.910066714851723e-14, "grad_norm": 0.5419957041740417, "learning_rate": 0.00012037369769472393, "loss": 2.6491, "step": 11270 }, { "epoch": 3.913536161803677e-14, "grad_norm": 0.5589016079902649, "learning_rate": 0.0001203648098376721, "loss": 2.6619, "step": 11280 }, { "epoch": 3.9170056087556304e-14, "grad_norm": 0.5550463795661926, "learning_rate": 0.00012035591381818236, "loss": 2.6162, "step": 11290 }, { "epoch": 3.920475055707584e-14, "grad_norm": 0.5637468099594116, "learning_rate": 0.0001203470096376616, "loss": 2.6315, "step": 11300 }, { "epoch": 3.9239445026595376e-14, "grad_norm": 0.5676747560501099, "learning_rate": 0.00012033809729751793, "loss": 2.6341, "step": 11310 }, { "epoch": 3.927413949611491e-14, "grad_norm": 0.552956223487854, "learning_rate": 0.00012032917679916073, "loss": 2.6101, "step": 11320 }, { "epoch": 3.930883396563445e-14, "grad_norm": 0.5755835771560669, "learning_rate": 0.00012032024814400072, "loss": 2.6117, "step": 11330 }, { "epoch": 3.9343528435153985e-14, "grad_norm": 0.6126993894577026, "learning_rate": 0.00012031131133344989, "loss": 2.6754, "step": 11340 }, { "epoch": 3.937822290467352e-14, "grad_norm": 0.5606957674026489, "learning_rate": 0.00012030236636892149, "loss": 2.6349, "step": 11350 }, { "epoch": 3.941291737419306e-14, "grad_norm": 0.5757265090942383, "learning_rate": 0.00012029341325183012, "loss": 2.6379, "step": 11360 }, { "epoch": 3.9447611843712593e-14, "grad_norm": 0.5416944622993469, "learning_rate": 0.00012028445198359166, "loss": 2.6605, "step": 11370 }, { "epoch": 3.948230631323213e-14, "grad_norm": 0.5534752607345581, "learning_rate": 0.00012027548256562321, "loss": 2.6393, "step": 11380 }, { "epoch": 3.9517000782751666e-14, "grad_norm": 0.5805953741073608, "learning_rate": 0.00012026650499934324, "loss": 2.612, "step": 11390 }, { "epoch": 3.95516952522712e-14, "grad_norm": 0.5517542362213135, "learning_rate": 0.00012025751928617145, "loss": 2.6845, "step": 11400 }, { "epoch": 3.958638972179074e-14, "grad_norm": 0.569674551486969, "learning_rate": 0.00012024852542752886, "loss": 2.6308, "step": 11410 }, { "epoch": 3.9621084191310274e-14, "grad_norm": 0.5921044945716858, "learning_rate": 0.00012023952342483777, "loss": 2.6378, "step": 11420 }, { "epoch": 3.965577866082981e-14, "grad_norm": 0.5645716190338135, "learning_rate": 0.0001202305132795218, "loss": 2.6291, "step": 11430 }, { "epoch": 3.9690473130349346e-14, "grad_norm": 0.5561150312423706, "learning_rate": 0.00012022149499300579, "loss": 2.6196, "step": 11440 }, { "epoch": 3.972516759986888e-14, "grad_norm": 0.5569630861282349, "learning_rate": 0.00012021246856671591, "loss": 2.6105, "step": 11450 }, { "epoch": 3.975986206938842e-14, "grad_norm": 0.5313461422920227, "learning_rate": 0.00012020343400207964, "loss": 2.6398, "step": 11460 }, { "epoch": 3.9794556538907955e-14, "grad_norm": 0.5543445944786072, "learning_rate": 0.00012019439130052567, "loss": 2.596, "step": 11470 }, { "epoch": 3.982925100842749e-14, "grad_norm": 0.5432611107826233, "learning_rate": 0.00012018534046348405, "loss": 2.6541, "step": 11480 }, { "epoch": 3.986394547794703e-14, "grad_norm": 0.5745410919189453, "learning_rate": 0.00012017628149238613, "loss": 2.6318, "step": 11490 }, { "epoch": 3.9898639947466563e-14, "grad_norm": 0.5659303665161133, "learning_rate": 0.00012016721438866444, "loss": 2.6463, "step": 11500 }, { "epoch": 3.99333344169861e-14, "grad_norm": 0.5563286542892456, "learning_rate": 0.00012015813915375293, "loss": 2.6534, "step": 11510 }, { "epoch": 3.9968028886505635e-14, "grad_norm": 0.5822678804397583, "learning_rate": 0.00012014905578908669, "loss": 2.6352, "step": 11520 }, { "epoch": 4.000272335602517e-14, "grad_norm": 0.5812376737594604, "learning_rate": 0.00012013996429610226, "loss": 2.6419, "step": 11530 }, { "epoch": 4.003741782554471e-14, "grad_norm": 0.5659502148628235, "learning_rate": 0.00012013086467623734, "loss": 2.6373, "step": 11540 }, { "epoch": 4.0072112295064244e-14, "grad_norm": 0.5606178045272827, "learning_rate": 0.00012012175693093094, "loss": 2.6268, "step": 11550 }, { "epoch": 4.010680676458378e-14, "grad_norm": 0.5537297129631042, "learning_rate": 0.00012011264106162341, "loss": 2.6662, "step": 11560 }, { "epoch": 4.0141501234103316e-14, "grad_norm": 0.5817919373512268, "learning_rate": 0.00012010351706975634, "loss": 2.6328, "step": 11570 }, { "epoch": 4.017619570362285e-14, "grad_norm": 0.5590681433677673, "learning_rate": 0.00012009438495677256, "loss": 2.6098, "step": 11580 }, { "epoch": 4.021089017314239e-14, "grad_norm": 0.5767475366592407, "learning_rate": 0.00012008524472411631, "loss": 2.6625, "step": 11590 }, { "epoch": 4.0245584642661925e-14, "grad_norm": 0.5483654141426086, "learning_rate": 0.00012007609637323298, "loss": 2.636, "step": 11600 }, { "epoch": 4.028027911218146e-14, "grad_norm": 0.5649356842041016, "learning_rate": 0.00012006693990556936, "loss": 2.6185, "step": 11610 }, { "epoch": 4.0314973581701e-14, "grad_norm": 0.5686025619506836, "learning_rate": 0.0001200577753225734, "loss": 2.6063, "step": 11620 }, { "epoch": 4.034966805122053e-14, "grad_norm": 0.5777961015701294, "learning_rate": 0.00012004860262569446, "loss": 2.6026, "step": 11630 }, { "epoch": 4.038436252074007e-14, "grad_norm": 0.5522383451461792, "learning_rate": 0.00012003942181638307, "loss": 2.6193, "step": 11640 }, { "epoch": 4.0419056990259605e-14, "grad_norm": 0.552580714225769, "learning_rate": 0.00012003023289609114, "loss": 2.6112, "step": 11650 }, { "epoch": 4.045375145977914e-14, "grad_norm": 0.5552770495414734, "learning_rate": 0.0001200210358662718, "loss": 2.6362, "step": 11660 }, { "epoch": 4.048844592929868e-14, "grad_norm": 0.5760284066200256, "learning_rate": 0.00012001183072837946, "loss": 2.662, "step": 11670 }, { "epoch": 4.0523140398818214e-14, "grad_norm": 0.5394372940063477, "learning_rate": 0.00012000261748386987, "loss": 2.6156, "step": 11680 }, { "epoch": 4.055783486833775e-14, "grad_norm": 0.5755141973495483, "learning_rate": 0.00011999339613420002, "loss": 2.5856, "step": 11690 }, { "epoch": 4.0592529337857286e-14, "grad_norm": 0.6035390496253967, "learning_rate": 0.00011998416668082817, "loss": 2.6192, "step": 11700 }, { "epoch": 4.062722380737682e-14, "grad_norm": 0.6035121083259583, "learning_rate": 0.00011997492912521389, "loss": 2.6647, "step": 11710 }, { "epoch": 4.066191827689636e-14, "grad_norm": 0.5590195059776306, "learning_rate": 0.00011996568346881801, "loss": 2.6132, "step": 11720 }, { "epoch": 4.0696612746415894e-14, "grad_norm": 0.5206135511398315, "learning_rate": 0.00011995642971310265, "loss": 2.6212, "step": 11730 }, { "epoch": 4.073130721593543e-14, "grad_norm": 0.6124399304389954, "learning_rate": 0.00011994716785953123, "loss": 2.6565, "step": 11740 }, { "epoch": 4.0766001685454967e-14, "grad_norm": 0.6187917590141296, "learning_rate": 0.00011993789790956842, "loss": 2.6357, "step": 11750 }, { "epoch": 4.08006961549745e-14, "grad_norm": 0.5645016431808472, "learning_rate": 0.00011992861986468018, "loss": 2.6157, "step": 11760 }, { "epoch": 4.083539062449404e-14, "grad_norm": 0.5608179569244385, "learning_rate": 0.00011991933372633376, "loss": 2.6462, "step": 11770 }, { "epoch": 4.0870085094013575e-14, "grad_norm": 0.5390154123306274, "learning_rate": 0.00011991003949599767, "loss": 2.6041, "step": 11780 }, { "epoch": 4.090477956353311e-14, "grad_norm": 0.5423350930213928, "learning_rate": 0.00011990073717514174, "loss": 2.6391, "step": 11790 }, { "epoch": 4.093947403305265e-14, "grad_norm": 0.5699427723884583, "learning_rate": 0.00011989142676523702, "loss": 2.616, "step": 11800 }, { "epoch": 4.0974168502572184e-14, "grad_norm": 0.5459266901016235, "learning_rate": 0.0001198821082677559, "loss": 2.6072, "step": 11810 }, { "epoch": 4.100886297209172e-14, "grad_norm": 0.5847983360290527, "learning_rate": 0.000119872781684172, "loss": 2.6666, "step": 11820 }, { "epoch": 4.1043557441611256e-14, "grad_norm": 0.5482750535011292, "learning_rate": 0.00011986344701596025, "loss": 2.639, "step": 11830 }, { "epoch": 4.107825191113079e-14, "grad_norm": 0.6347272396087646, "learning_rate": 0.00011985410426459684, "loss": 2.6168, "step": 11840 }, { "epoch": 4.111294638065033e-14, "grad_norm": 0.5724297761917114, "learning_rate": 0.00011984475343155923, "loss": 2.6301, "step": 11850 }, { "epoch": 4.1147640850169864e-14, "grad_norm": 0.550179660320282, "learning_rate": 0.0001198353945183262, "loss": 2.6237, "step": 11860 }, { "epoch": 4.11823353196894e-14, "grad_norm": 0.5592813491821289, "learning_rate": 0.00011982602752637778, "loss": 2.6613, "step": 11870 }, { "epoch": 4.1217029789208937e-14, "grad_norm": 0.519370436668396, "learning_rate": 0.00011981665245719527, "loss": 2.6162, "step": 11880 }, { "epoch": 4.125172425872847e-14, "grad_norm": 0.559050977230072, "learning_rate": 0.00011980726931226125, "loss": 2.6136, "step": 11890 }, { "epoch": 4.128641872824801e-14, "grad_norm": 0.5757958889007568, "learning_rate": 0.0001197978780930596, "loss": 2.627, "step": 11900 }, { "epoch": 4.1321113197767545e-14, "grad_norm": 0.5696915984153748, "learning_rate": 0.00011978847880107546, "loss": 2.6221, "step": 11910 }, { "epoch": 4.135580766728708e-14, "grad_norm": 0.5624293684959412, "learning_rate": 0.00011977907143779522, "loss": 2.5679, "step": 11920 }, { "epoch": 4.139050213680662e-14, "grad_norm": 0.5883429646492004, "learning_rate": 0.00011976965600470661, "loss": 2.6036, "step": 11930 }, { "epoch": 4.1425196606326153e-14, "grad_norm": 0.5801324248313904, "learning_rate": 0.00011976023250329857, "loss": 2.5854, "step": 11940 }, { "epoch": 4.145989107584569e-14, "grad_norm": 0.5817421078681946, "learning_rate": 0.00011975080093506138, "loss": 2.6583, "step": 11950 }, { "epoch": 4.1494585545365226e-14, "grad_norm": 0.5694712996482849, "learning_rate": 0.00011974136130148652, "loss": 2.6231, "step": 11960 }, { "epoch": 4.152928001488476e-14, "grad_norm": 0.5700250864028931, "learning_rate": 0.00011973191360406684, "loss": 2.6191, "step": 11970 }, { "epoch": 4.15639744844043e-14, "grad_norm": 0.5368398427963257, "learning_rate": 0.00011972245784429633, "loss": 2.6465, "step": 11980 }, { "epoch": 4.1598668953923834e-14, "grad_norm": 0.5441164970397949, "learning_rate": 0.00011971299402367042, "loss": 2.6489, "step": 11990 }, { "epoch": 4.163336342344337e-14, "grad_norm": 0.532524049282074, "learning_rate": 0.00011970352214368569, "loss": 2.6427, "step": 12000 }, { "epoch": 4.1668057892962906e-14, "grad_norm": 0.5443849563598633, "learning_rate": 0.00011969404220584004, "loss": 2.6287, "step": 12010 }, { "epoch": 4.170275236248244e-14, "grad_norm": 0.5504503846168518, "learning_rate": 0.00011968455421163263, "loss": 2.6254, "step": 12020 }, { "epoch": 4.173744683200198e-14, "grad_norm": 0.5832147002220154, "learning_rate": 0.00011967505816256394, "loss": 2.6028, "step": 12030 }, { "epoch": 4.1772141301521515e-14, "grad_norm": 0.5957428216934204, "learning_rate": 0.00011966555406013568, "loss": 2.6695, "step": 12040 }, { "epoch": 4.180683577104105e-14, "grad_norm": 0.5506970882415771, "learning_rate": 0.0001196560419058508, "loss": 2.6256, "step": 12050 }, { "epoch": 4.184153024056059e-14, "grad_norm": 0.5359552502632141, "learning_rate": 0.00011964652170121361, "loss": 2.6167, "step": 12060 }, { "epoch": 4.1876224710080123e-14, "grad_norm": 0.5298483371734619, "learning_rate": 0.00011963699344772962, "loss": 2.6094, "step": 12070 }, { "epoch": 4.191091917959966e-14, "grad_norm": 0.5812767744064331, "learning_rate": 0.00011962745714690567, "loss": 2.6655, "step": 12080 }, { "epoch": 4.1945613649119196e-14, "grad_norm": 0.6042730212211609, "learning_rate": 0.00011961791280024982, "loss": 2.6294, "step": 12090 }, { "epoch": 4.198030811863873e-14, "grad_norm": 0.5572322607040405, "learning_rate": 0.00011960836040927142, "loss": 2.6183, "step": 12100 }, { "epoch": 4.201500258815827e-14, "grad_norm": 0.5696256756782532, "learning_rate": 0.00011959879997548115, "loss": 2.6191, "step": 12110 }, { "epoch": 4.2049697057677804e-14, "grad_norm": 0.5843503475189209, "learning_rate": 0.00011958923150039085, "loss": 2.6641, "step": 12120 }, { "epoch": 4.208439152719734e-14, "grad_norm": 0.652480959892273, "learning_rate": 0.0001195796549855137, "loss": 2.5923, "step": 12130 }, { "epoch": 4.2119085996716876e-14, "grad_norm": 0.5661613941192627, "learning_rate": 0.00011957007043236419, "loss": 2.5911, "step": 12140 }, { "epoch": 4.215378046623641e-14, "grad_norm": 0.5775526165962219, "learning_rate": 0.000119560477842458, "loss": 2.6109, "step": 12150 }, { "epoch": 4.218847493575595e-14, "grad_norm": 0.537965714931488, "learning_rate": 0.00011955087721731211, "loss": 2.6116, "step": 12160 }, { "epoch": 4.2223169405275485e-14, "grad_norm": 0.548711359500885, "learning_rate": 0.00011954126855844477, "loss": 2.557, "step": 12170 }, { "epoch": 4.225786387479502e-14, "grad_norm": 0.5780961513519287, "learning_rate": 0.00011953165186737555, "loss": 2.6098, "step": 12180 }, { "epoch": 4.229255834431456e-14, "grad_norm": 0.5589568018913269, "learning_rate": 0.0001195220271456252, "loss": 2.5827, "step": 12190 }, { "epoch": 4.232725281383409e-14, "grad_norm": 0.5608492493629456, "learning_rate": 0.00011951239439471581, "loss": 2.6289, "step": 12200 }, { "epoch": 4.236194728335363e-14, "grad_norm": 0.5216711163520813, "learning_rate": 0.0001195027536161707, "loss": 2.6074, "step": 12210 }, { "epoch": 4.2396641752873165e-14, "grad_norm": 0.5567456483840942, "learning_rate": 0.00011949310481151452, "loss": 2.6545, "step": 12220 }, { "epoch": 4.24313362223927e-14, "grad_norm": 0.517013669013977, "learning_rate": 0.00011948344798227307, "loss": 2.6, "step": 12230 }, { "epoch": 4.246603069191224e-14, "grad_norm": 0.5287975072860718, "learning_rate": 0.00011947378312997359, "loss": 2.607, "step": 12240 }, { "epoch": 4.2500725161431774e-14, "grad_norm": 0.5762810111045837, "learning_rate": 0.0001194641102561444, "loss": 2.6185, "step": 12250 }, { "epoch": 4.253541963095131e-14, "grad_norm": 0.5864116549491882, "learning_rate": 0.00011945442936231523, "loss": 2.6285, "step": 12260 }, { "epoch": 4.2570114100470846e-14, "grad_norm": 0.5565986037254333, "learning_rate": 0.00011944474045001703, "loss": 2.6188, "step": 12270 }, { "epoch": 4.260480856999038e-14, "grad_norm": 0.5509280562400818, "learning_rate": 0.00011943504352078203, "loss": 2.5953, "step": 12280 }, { "epoch": 4.263950303950992e-14, "grad_norm": 0.5864250659942627, "learning_rate": 0.00011942533857614367, "loss": 2.622, "step": 12290 }, { "epoch": 4.2674197509029455e-14, "grad_norm": 0.5425901412963867, "learning_rate": 0.00011941562561763674, "loss": 2.5956, "step": 12300 }, { "epoch": 4.270889197854899e-14, "grad_norm": 0.5524466037750244, "learning_rate": 0.00011940590464679724, "loss": 2.625, "step": 12310 }, { "epoch": 4.2743586448068527e-14, "grad_norm": 0.5633988380432129, "learning_rate": 0.00011939617566516248, "loss": 2.5978, "step": 12320 }, { "epoch": 4.277828091758806e-14, "grad_norm": 0.5909106731414795, "learning_rate": 0.000119386438674271, "loss": 2.6177, "step": 12330 }, { "epoch": 4.28129753871076e-14, "grad_norm": 0.5547037124633789, "learning_rate": 0.00011937669367566265, "loss": 2.5994, "step": 12340 }, { "epoch": 4.2847669856627135e-14, "grad_norm": 0.561391294002533, "learning_rate": 0.00011936694067087845, "loss": 2.6145, "step": 12350 }, { "epoch": 4.288236432614667e-14, "grad_norm": 0.5450502038002014, "learning_rate": 0.00011935717966146081, "loss": 2.5715, "step": 12360 }, { "epoch": 4.291705879566621e-14, "grad_norm": 0.543514609336853, "learning_rate": 0.00011934741064895335, "loss": 2.5882, "step": 12370 }, { "epoch": 4.2951753265185744e-14, "grad_norm": 0.5429597496986389, "learning_rate": 0.00011933763363490095, "loss": 2.6238, "step": 12380 }, { "epoch": 4.298644773470528e-14, "grad_norm": 0.5505301356315613, "learning_rate": 0.00011932784862084972, "loss": 2.641, "step": 12390 }, { "epoch": 4.3021142204224816e-14, "grad_norm": 0.5507093071937561, "learning_rate": 0.00011931805560834714, "loss": 2.6174, "step": 12400 }, { "epoch": 4.305583667374435e-14, "grad_norm": 0.5339269638061523, "learning_rate": 0.00011930825459894183, "loss": 2.6097, "step": 12410 }, { "epoch": 4.309053114326389e-14, "grad_norm": 0.5710604190826416, "learning_rate": 0.00011929844559418378, "loss": 2.5583, "step": 12420 }, { "epoch": 4.3125225612783424e-14, "grad_norm": 0.6010050177574158, "learning_rate": 0.00011928862859562419, "loss": 2.6186, "step": 12430 }, { "epoch": 4.315992008230296e-14, "grad_norm": 0.5539467334747314, "learning_rate": 0.00011927880360481551, "loss": 2.6245, "step": 12440 }, { "epoch": 4.3194614551822497e-14, "grad_norm": 0.5998532772064209, "learning_rate": 0.00011926897062331149, "loss": 2.6388, "step": 12450 }, { "epoch": 4.322930902134203e-14, "grad_norm": 0.5505980253219604, "learning_rate": 0.00011925912965266716, "loss": 2.6163, "step": 12460 }, { "epoch": 4.326400349086157e-14, "grad_norm": 0.5457847714424133, "learning_rate": 0.00011924928069443874, "loss": 2.6089, "step": 12470 }, { "epoch": 4.3298697960381105e-14, "grad_norm": 0.5573354363441467, "learning_rate": 0.0001192394237501838, "loss": 2.635, "step": 12480 }, { "epoch": 4.333339242990064e-14, "grad_norm": 0.5457021594047546, "learning_rate": 0.00011922955882146109, "loss": 2.6321, "step": 12490 }, { "epoch": 4.336808689942018e-14, "grad_norm": 0.540162205696106, "learning_rate": 0.0001192196859098307, "loss": 2.6193, "step": 12500 }, { "epoch": 4.3402781368939714e-14, "grad_norm": 0.5336670279502869, "learning_rate": 0.00011920980501685391, "loss": 2.5897, "step": 12510 }, { "epoch": 4.343747583845925e-14, "grad_norm": 0.5586407780647278, "learning_rate": 0.00011919991614409332, "loss": 2.616, "step": 12520 }, { "epoch": 4.3472170307978786e-14, "grad_norm": 0.5536615252494812, "learning_rate": 0.00011919001929311279, "loss": 2.5713, "step": 12530 }, { "epoch": 4.350686477749832e-14, "grad_norm": 0.5569251179695129, "learning_rate": 0.00011918011446547737, "loss": 2.6192, "step": 12540 }, { "epoch": 4.354155924701786e-14, "grad_norm": 0.5395358204841614, "learning_rate": 0.00011917020166275347, "loss": 2.5802, "step": 12550 }, { "epoch": 4.3576253716537394e-14, "grad_norm": 0.5648619532585144, "learning_rate": 0.0001191602808865087, "loss": 2.6661, "step": 12560 }, { "epoch": 4.361094818605693e-14, "grad_norm": 0.5851330757141113, "learning_rate": 0.00011915035213831195, "loss": 2.618, "step": 12570 }, { "epoch": 4.3645642655576466e-14, "grad_norm": 0.5539594292640686, "learning_rate": 0.00011914041541973335, "loss": 2.6173, "step": 12580 }, { "epoch": 4.3680337125096e-14, "grad_norm": 0.6369144916534424, "learning_rate": 0.0001191304707323443, "loss": 2.5756, "step": 12590 }, { "epoch": 4.371503159461554e-14, "grad_norm": 0.6140754222869873, "learning_rate": 0.0001191205180777175, "loss": 2.6504, "step": 12600 }, { "epoch": 4.3749726064135075e-14, "grad_norm": 0.5697541832923889, "learning_rate": 0.00011911055745742683, "loss": 2.5948, "step": 12610 }, { "epoch": 4.378442053365461e-14, "grad_norm": 0.5734003186225891, "learning_rate": 0.00011910058887304754, "loss": 2.6252, "step": 12620 }, { "epoch": 4.381911500317415e-14, "grad_norm": 0.6034184694290161, "learning_rate": 0.00011909061232615603, "loss": 2.6389, "step": 12630 }, { "epoch": 4.3853809472693683e-14, "grad_norm": 0.571704089641571, "learning_rate": 0.00011908062781833002, "loss": 2.586, "step": 12640 }, { "epoch": 4.388850394221322e-14, "grad_norm": 0.5650452375411987, "learning_rate": 0.00011907063535114845, "loss": 2.6163, "step": 12650 }, { "epoch": 4.3923198411732756e-14, "grad_norm": 0.54019695520401, "learning_rate": 0.00011906063492619158, "loss": 2.5803, "step": 12660 }, { "epoch": 4.395789288125229e-14, "grad_norm": 0.5577172040939331, "learning_rate": 0.00011905062654504087, "loss": 2.5951, "step": 12670 }, { "epoch": 4.399258735077183e-14, "grad_norm": 0.5828005075454712, "learning_rate": 0.00011904061020927906, "loss": 2.6312, "step": 12680 }, { "epoch": 4.4027281820291364e-14, "grad_norm": 0.5627454519271851, "learning_rate": 0.00011903058592049014, "loss": 2.564, "step": 12690 }, { "epoch": 4.40619762898109e-14, "grad_norm": 0.577043890953064, "learning_rate": 0.00011902055368025938, "loss": 2.5946, "step": 12700 }, { "epoch": 4.4096670759330436e-14, "grad_norm": 0.5481210350990295, "learning_rate": 0.00011901051349017329, "loss": 2.6147, "step": 12710 }, { "epoch": 4.413136522884997e-14, "grad_norm": 0.5662303566932678, "learning_rate": 0.00011900046535181967, "loss": 2.6169, "step": 12720 }, { "epoch": 4.416605969836951e-14, "grad_norm": 0.5580433011054993, "learning_rate": 0.00011899040926678746, "loss": 2.6043, "step": 12730 }, { "epoch": 4.4200754167889045e-14, "grad_norm": 0.5725385546684265, "learning_rate": 0.00011898034523666703, "loss": 2.623, "step": 12740 }, { "epoch": 4.423544863740858e-14, "grad_norm": 0.5644475817680359, "learning_rate": 0.00011897027326304988, "loss": 2.5529, "step": 12750 }, { "epoch": 4.427014310692812e-14, "grad_norm": 0.5381366014480591, "learning_rate": 0.00011896019334752881, "loss": 2.6118, "step": 12760 }, { "epoch": 4.4304837576447653e-14, "grad_norm": 0.5362083911895752, "learning_rate": 0.00011895010549169786, "loss": 2.6119, "step": 12770 }, { "epoch": 4.433953204596719e-14, "grad_norm": 0.5617156028747559, "learning_rate": 0.00011894000969715238, "loss": 2.5808, "step": 12780 }, { "epoch": 4.4374226515486725e-14, "grad_norm": 0.5939880609512329, "learning_rate": 0.00011892990596548888, "loss": 2.5978, "step": 12790 }, { "epoch": 4.440892098500626e-14, "grad_norm": 0.5492604970932007, "learning_rate": 0.00011891979429830521, "loss": 2.5827, "step": 12800 }, { "epoch": 4.44436154545258e-14, "grad_norm": 0.5652876496315002, "learning_rate": 0.00011890967469720045, "loss": 2.601, "step": 12810 }, { "epoch": 4.4478309924045334e-14, "grad_norm": 0.5466254353523254, "learning_rate": 0.00011889954716377489, "loss": 2.6531, "step": 12820 }, { "epoch": 4.451300439356487e-14, "grad_norm": 0.5494062900543213, "learning_rate": 0.00011888941169963015, "loss": 2.5886, "step": 12830 }, { "epoch": 4.4547698863084406e-14, "grad_norm": 0.5212080478668213, "learning_rate": 0.00011887926830636906, "loss": 2.6026, "step": 12840 }, { "epoch": 4.458239333260394e-14, "grad_norm": 0.5500509142875671, "learning_rate": 0.00011886911698559569, "loss": 2.6111, "step": 12850 }, { "epoch": 4.461708780212348e-14, "grad_norm": 0.5546611547470093, "learning_rate": 0.0001188589577389154, "loss": 2.6162, "step": 12860 }, { "epoch": 4.4651782271643015e-14, "grad_norm": 0.5458112955093384, "learning_rate": 0.00011884879056793479, "loss": 2.5945, "step": 12870 }, { "epoch": 4.468647674116255e-14, "grad_norm": 0.5375398397445679, "learning_rate": 0.00011883861547426171, "loss": 2.6107, "step": 12880 }, { "epoch": 4.472117121068209e-14, "grad_norm": 0.5326858162879944, "learning_rate": 0.00011882843245950524, "loss": 2.5827, "step": 12890 }, { "epoch": 4.475586568020162e-14, "grad_norm": 0.5281192064285278, "learning_rate": 0.00011881824152527577, "loss": 2.5609, "step": 12900 }, { "epoch": 4.479056014972116e-14, "grad_norm": 0.6037759780883789, "learning_rate": 0.00011880804267318488, "loss": 2.629, "step": 12910 }, { "epoch": 4.4825254619240695e-14, "grad_norm": 7.128318786621094, "learning_rate": 0.00011879783590484547, "loss": 2.6185, "step": 12920 }, { "epoch": 4.485994908876023e-14, "grad_norm": 0.5508823990821838, "learning_rate": 0.00011878762122187162, "loss": 2.6054, "step": 12930 }, { "epoch": 4.489464355827977e-14, "grad_norm": 0.5807201266288757, "learning_rate": 0.00011877739862587868, "loss": 2.5923, "step": 12940 }, { "epoch": 4.4929338027799304e-14, "grad_norm": 0.544528067111969, "learning_rate": 0.0001187671681184833, "loss": 2.6288, "step": 12950 }, { "epoch": 4.496403249731884e-14, "grad_norm": 0.5562353730201721, "learning_rate": 0.00011875692970130331, "loss": 2.5664, "step": 12960 }, { "epoch": 4.4998726966838376e-14, "grad_norm": 0.5448423027992249, "learning_rate": 0.00011874668337595786, "loss": 2.5972, "step": 12970 }, { "epoch": 4.503342143635791e-14, "grad_norm": 0.5516115427017212, "learning_rate": 0.00011873642914406734, "loss": 2.5769, "step": 12980 }, { "epoch": 4.506811590587745e-14, "grad_norm": 0.5339480638504028, "learning_rate": 0.00011872616700725332, "loss": 2.5917, "step": 12990 }, { "epoch": 4.5102810375396984e-14, "grad_norm": 0.5338950753211975, "learning_rate": 0.00011871589696713867, "loss": 2.6279, "step": 13000 }, { "epoch": 4.513750484491652e-14, "grad_norm": 0.588991641998291, "learning_rate": 0.00011870561902534752, "loss": 2.5885, "step": 13010 }, { "epoch": 4.5172199314436057e-14, "grad_norm": 0.5608024597167969, "learning_rate": 0.00011869533318350527, "loss": 2.6228, "step": 13020 }, { "epoch": 4.520689378395559e-14, "grad_norm": 0.6109429001808167, "learning_rate": 0.00011868503944323849, "loss": 2.5784, "step": 13030 }, { "epoch": 4.524158825347513e-14, "grad_norm": 0.5742112994194031, "learning_rate": 0.00011867473780617508, "loss": 2.6436, "step": 13040 }, { "epoch": 4.5276282722994665e-14, "grad_norm": 0.5615503191947937, "learning_rate": 0.00011866442827394414, "loss": 2.6037, "step": 13050 }, { "epoch": 4.53109771925142e-14, "grad_norm": 0.5737802982330322, "learning_rate": 0.00011865411084817602, "loss": 2.5768, "step": 13060 }, { "epoch": 4.534567166203374e-14, "grad_norm": 0.5946771502494812, "learning_rate": 0.00011864378553050235, "loss": 2.6304, "step": 13070 }, { "epoch": 4.5380366131553274e-14, "grad_norm": 0.5678924918174744, "learning_rate": 0.000118633452322556, "loss": 2.5834, "step": 13080 }, { "epoch": 4.541506060107281e-14, "grad_norm": 0.5528861284255981, "learning_rate": 0.00011862311122597105, "loss": 2.5845, "step": 13090 }, { "epoch": 4.5449755070592346e-14, "grad_norm": 0.5377615690231323, "learning_rate": 0.00011861276224238288, "loss": 2.5869, "step": 13100 }, { "epoch": 4.548444954011188e-14, "grad_norm": 0.5299820899963379, "learning_rate": 0.00011860240537342808, "loss": 2.5748, "step": 13110 }, { "epoch": 4.551914400963142e-14, "grad_norm": 0.5899865627288818, "learning_rate": 0.00011859204062074451, "loss": 2.5884, "step": 13120 }, { "epoch": 4.5553838479150954e-14, "grad_norm": 0.5695735216140747, "learning_rate": 0.00011858166798597123, "loss": 2.5891, "step": 13130 }, { "epoch": 4.558853294867049e-14, "grad_norm": 0.5352872610092163, "learning_rate": 0.00011857128747074865, "loss": 2.5974, "step": 13140 }, { "epoch": 4.5623227418190027e-14, "grad_norm": 0.6546672582626343, "learning_rate": 0.0001185608990767183, "loss": 2.5602, "step": 13150 }, { "epoch": 4.565792188770956e-14, "grad_norm": 0.5655081272125244, "learning_rate": 0.00011855050280552303, "loss": 2.5978, "step": 13160 }, { "epoch": 4.56926163572291e-14, "grad_norm": 0.5425516963005066, "learning_rate": 0.00011854009865880695, "loss": 2.5905, "step": 13170 }, { "epoch": 4.5727310826748635e-14, "grad_norm": 0.5369641780853271, "learning_rate": 0.00011852968663821534, "loss": 2.593, "step": 13180 }, { "epoch": 4.576200529626817e-14, "grad_norm": 0.5432778000831604, "learning_rate": 0.0001185192667453948, "loss": 2.5813, "step": 13190 }, { "epoch": 4.579669976578771e-14, "grad_norm": 0.5387287139892578, "learning_rate": 0.00011850883898199316, "loss": 2.6126, "step": 13200 }, { "epoch": 4.5831394235307243e-14, "grad_norm": 0.5455820560455322, "learning_rate": 0.00011849840334965945, "loss": 2.5877, "step": 13210 }, { "epoch": 4.586608870482678e-14, "grad_norm": 0.5439821481704712, "learning_rate": 0.00011848795985004399, "loss": 2.6032, "step": 13220 }, { "epoch": 4.5900783174346316e-14, "grad_norm": 0.5427988767623901, "learning_rate": 0.0001184775084847983, "loss": 2.5937, "step": 13230 }, { "epoch": 4.593547764386585e-14, "grad_norm": 0.5350154042243958, "learning_rate": 0.00011846704925557523, "loss": 2.635, "step": 13240 }, { "epoch": 4.597017211338539e-14, "grad_norm": 0.5307956337928772, "learning_rate": 0.00011845658216402878, "loss": 2.5764, "step": 13250 }, { "epoch": 4.6004866582904924e-14, "grad_norm": 0.529161274433136, "learning_rate": 0.00011844610721181424, "loss": 2.5741, "step": 13260 }, { "epoch": 4.603956105242446e-14, "grad_norm": 0.5679211616516113, "learning_rate": 0.00011843562440058813, "loss": 2.5507, "step": 13270 }, { "epoch": 4.6074255521943996e-14, "grad_norm": 0.5356968641281128, "learning_rate": 0.00011842513373200822, "loss": 2.6071, "step": 13280 }, { "epoch": 4.610894999146353e-14, "grad_norm": 0.5480650663375854, "learning_rate": 0.00011841463520773352, "loss": 2.5772, "step": 13290 }, { "epoch": 4.614364446098307e-14, "grad_norm": 0.583526611328125, "learning_rate": 0.00011840412882942426, "loss": 2.5841, "step": 13300 }, { "epoch": 4.6178338930502605e-14, "grad_norm": 0.5327820777893066, "learning_rate": 0.00011839361459874198, "loss": 2.6138, "step": 13310 }, { "epoch": 4.621303340002214e-14, "grad_norm": 0.5968527793884277, "learning_rate": 0.00011838309251734937, "loss": 2.5793, "step": 13320 }, { "epoch": 4.624772786954168e-14, "grad_norm": 0.5413755178451538, "learning_rate": 0.00011837256258691043, "loss": 2.6116, "step": 13330 }, { "epoch": 4.6282422339061213e-14, "grad_norm": 0.6006259322166443, "learning_rate": 0.00011836202480909036, "loss": 2.5877, "step": 13340 }, { "epoch": 4.631711680858075e-14, "grad_norm": 0.5978817939758301, "learning_rate": 0.00011835147918555562, "loss": 2.5663, "step": 13350 }, { "epoch": 4.6351811278100286e-14, "grad_norm": 0.6145845055580139, "learning_rate": 0.00011834092571797394, "loss": 2.5729, "step": 13360 }, { "epoch": 4.638650574761982e-14, "grad_norm": 0.5176492929458618, "learning_rate": 0.00011833036440801424, "loss": 2.5916, "step": 13370 }, { "epoch": 4.642120021713936e-14, "grad_norm": 0.5443931818008423, "learning_rate": 0.00011831979525734667, "loss": 2.5971, "step": 13380 }, { "epoch": 4.6455894686658894e-14, "grad_norm": 0.5576615929603577, "learning_rate": 0.00011830921826764272, "loss": 2.599, "step": 13390 }, { "epoch": 4.649058915617843e-14, "grad_norm": 0.5775448083877563, "learning_rate": 0.000118298633440575, "loss": 2.5907, "step": 13400 }, { "epoch": 4.6525283625697966e-14, "grad_norm": 0.5758131146430969, "learning_rate": 0.0001182880407778174, "loss": 2.6018, "step": 13410 }, { "epoch": 4.65599780952175e-14, "grad_norm": 0.5840325355529785, "learning_rate": 0.0001182774402810451, "loss": 2.5884, "step": 13420 }, { "epoch": 4.659467256473704e-14, "grad_norm": 0.534463107585907, "learning_rate": 0.00011826683195193442, "loss": 2.5944, "step": 13430 }, { "epoch": 4.6629367034256575e-14, "grad_norm": 0.5451862215995789, "learning_rate": 0.00011825621579216304, "loss": 2.5948, "step": 13440 }, { "epoch": 4.666406150377611e-14, "grad_norm": 0.5377612709999084, "learning_rate": 0.0001182455918034098, "loss": 2.6219, "step": 13450 }, { "epoch": 4.669875597329565e-14, "grad_norm": 0.561937689781189, "learning_rate": 0.00011823495998735474, "loss": 2.577, "step": 13460 }, { "epoch": 4.673345044281518e-14, "grad_norm": 0.5340191721916199, "learning_rate": 0.00011822432034567925, "loss": 2.586, "step": 13470 }, { "epoch": 4.676814491233472e-14, "grad_norm": 0.638986349105835, "learning_rate": 0.00011821367288006588, "loss": 2.602, "step": 13480 }, { "epoch": 4.6802839381854255e-14, "grad_norm": 0.5439431667327881, "learning_rate": 0.0001182030175921984, "loss": 2.5799, "step": 13490 }, { "epoch": 4.683753385137379e-14, "grad_norm": 0.5417417883872986, "learning_rate": 0.00011819235448376189, "loss": 2.5643, "step": 13500 }, { "epoch": 4.687222832089333e-14, "grad_norm": 0.69687420129776, "learning_rate": 0.00011818168355644264, "loss": 2.5929, "step": 13510 }, { "epoch": 4.6906922790412864e-14, "grad_norm": 0.5478177070617676, "learning_rate": 0.00011817100481192814, "loss": 2.5932, "step": 13520 }, { "epoch": 4.69416172599324e-14, "grad_norm": 0.548803985118866, "learning_rate": 0.00011816031825190715, "loss": 2.6089, "step": 13530 }, { "epoch": 4.6976311729451936e-14, "grad_norm": 0.5715242028236389, "learning_rate": 0.00011814962387806966, "loss": 2.5912, "step": 13540 }, { "epoch": 4.701100619897147e-14, "grad_norm": 0.556463897228241, "learning_rate": 0.00011813892169210687, "loss": 2.5685, "step": 13550 }, { "epoch": 4.704570066849101e-14, "grad_norm": 0.5359580516815186, "learning_rate": 0.00011812821169571127, "loss": 2.5752, "step": 13560 }, { "epoch": 4.7080395138010545e-14, "grad_norm": 0.5664352178573608, "learning_rate": 0.00011811749389057654, "loss": 2.5985, "step": 13570 }, { "epoch": 4.711508960753008e-14, "grad_norm": 0.5962624549865723, "learning_rate": 0.0001181067682783976, "loss": 2.5578, "step": 13580 }, { "epoch": 4.7149784077049617e-14, "grad_norm": 0.5728591680526733, "learning_rate": 0.00011809603486087064, "loss": 2.5837, "step": 13590 }, { "epoch": 4.718447854656915e-14, "grad_norm": 0.5329049229621887, "learning_rate": 0.00011808529363969304, "loss": 2.5836, "step": 13600 }, { "epoch": 4.721917301608869e-14, "grad_norm": 0.5493348240852356, "learning_rate": 0.00011807454461656342, "loss": 2.5726, "step": 13610 }, { "epoch": 4.7253867485608225e-14, "grad_norm": 0.5395370721817017, "learning_rate": 0.00011806378779318165, "loss": 2.5868, "step": 13620 }, { "epoch": 4.728856195512776e-14, "grad_norm": 0.5301405787467957, "learning_rate": 0.00011805302317124884, "loss": 2.6122, "step": 13630 }, { "epoch": 4.73232564246473e-14, "grad_norm": 0.5987261533737183, "learning_rate": 0.00011804225075246733, "loss": 2.5393, "step": 13640 }, { "epoch": 4.7357950894166834e-14, "grad_norm": 0.5598045587539673, "learning_rate": 0.00011803147053854068, "loss": 2.5951, "step": 13650 }, { "epoch": 4.739264536368637e-14, "grad_norm": 0.5527815222740173, "learning_rate": 0.00011802068253117367, "loss": 2.5853, "step": 13660 }, { "epoch": 4.7427339833205906e-14, "grad_norm": 0.5564942359924316, "learning_rate": 0.00011800988673207233, "loss": 2.5652, "step": 13670 }, { "epoch": 4.746203430272544e-14, "grad_norm": 0.5785066485404968, "learning_rate": 0.00011799908314294395, "loss": 2.5686, "step": 13680 }, { "epoch": 4.749672877224498e-14, "grad_norm": 0.5673598051071167, "learning_rate": 0.00011798827176549701, "loss": 2.5513, "step": 13690 }, { "epoch": 4.7531423241764514e-14, "grad_norm": 0.5645158290863037, "learning_rate": 0.00011797745260144124, "loss": 2.574, "step": 13700 }, { "epoch": 4.756611771128405e-14, "grad_norm": 0.5313735604286194, "learning_rate": 0.00011796662565248758, "loss": 2.6084, "step": 13710 }, { "epoch": 4.7600812180803587e-14, "grad_norm": 0.5619309544563293, "learning_rate": 0.00011795579092034825, "loss": 2.5916, "step": 13720 }, { "epoch": 4.763550665032312e-14, "grad_norm": 0.5441076755523682, "learning_rate": 0.00011794494840673666, "loss": 2.6132, "step": 13730 }, { "epoch": 4.767020111984266e-14, "grad_norm": 0.5343903303146362, "learning_rate": 0.00011793409811336743, "loss": 2.5842, "step": 13740 }, { "epoch": 4.7704895589362195e-14, "grad_norm": 0.5376540422439575, "learning_rate": 0.00011792324004195646, "loss": 2.546, "step": 13750 }, { "epoch": 4.773959005888173e-14, "grad_norm": 0.5330073833465576, "learning_rate": 0.00011791237419422089, "loss": 2.5954, "step": 13760 }, { "epoch": 4.777428452840127e-14, "grad_norm": 0.5527759790420532, "learning_rate": 0.00011790150057187901, "loss": 2.5527, "step": 13770 }, { "epoch": 4.7808978997920804e-14, "grad_norm": 0.5579999089241028, "learning_rate": 0.00011789061917665042, "loss": 2.5968, "step": 13780 }, { "epoch": 4.784367346744034e-14, "grad_norm": 0.5511219501495361, "learning_rate": 0.00011787973001025591, "loss": 2.5306, "step": 13790 }, { "epoch": 4.7878367936959876e-14, "grad_norm": 0.5246829986572266, "learning_rate": 0.00011786883307441752, "loss": 2.5802, "step": 13800 }, { "epoch": 4.791306240647941e-14, "grad_norm": 0.5993335843086243, "learning_rate": 0.0001178579283708585, "loss": 2.5981, "step": 13810 }, { "epoch": 4.794775687599895e-14, "grad_norm": 0.5508777499198914, "learning_rate": 0.00011784701590130332, "loss": 2.579, "step": 13820 }, { "epoch": 4.7982451345518484e-14, "grad_norm": 0.6985280513763428, "learning_rate": 0.0001178360956674777, "loss": 2.6202, "step": 13830 }, { "epoch": 4.801714581503802e-14, "grad_norm": 0.5840509533882141, "learning_rate": 0.00011782516767110861, "loss": 2.5305, "step": 13840 }, { "epoch": 4.8051840284557557e-14, "grad_norm": 0.5208919644355774, "learning_rate": 0.00011781423191392418, "loss": 2.564, "step": 13850 }, { "epoch": 4.808653475407709e-14, "grad_norm": 0.534591794013977, "learning_rate": 0.0001178032883976538, "loss": 2.5566, "step": 13860 }, { "epoch": 4.812122922359663e-14, "grad_norm": 0.542080283164978, "learning_rate": 0.00011779233712402813, "loss": 2.5937, "step": 13870 }, { "epoch": 4.8155923693116165e-14, "grad_norm": 0.5175138115882874, "learning_rate": 0.00011778137809477901, "loss": 2.5593, "step": 13880 }, { "epoch": 4.81906181626357e-14, "grad_norm": 0.5634547472000122, "learning_rate": 0.00011777041131163953, "loss": 2.5676, "step": 13890 }, { "epoch": 4.822531263215524e-14, "grad_norm": 0.5322561860084534, "learning_rate": 0.00011775943677634394, "loss": 2.5812, "step": 13900 }, { "epoch": 4.8260007101674773e-14, "grad_norm": 0.5139391422271729, "learning_rate": 0.0001177484544906278, "loss": 2.6231, "step": 13910 }, { "epoch": 4.829470157119431e-14, "grad_norm": 0.6136922836303711, "learning_rate": 0.00011773746445622789, "loss": 2.5467, "step": 13920 }, { "epoch": 4.8329396040713846e-14, "grad_norm": 0.578454852104187, "learning_rate": 0.00011772646667488213, "loss": 2.5885, "step": 13930 }, { "epoch": 4.836409051023338e-14, "grad_norm": 0.5696459412574768, "learning_rate": 0.00011771546114832979, "loss": 2.5831, "step": 13940 }, { "epoch": 4.839878497975292e-14, "grad_norm": 0.5399787425994873, "learning_rate": 0.00011770444787831123, "loss": 2.543, "step": 13950 }, { "epoch": 4.8433479449272454e-14, "grad_norm": 0.5378800630569458, "learning_rate": 0.00011769342686656816, "loss": 2.565, "step": 13960 }, { "epoch": 4.846817391879199e-14, "grad_norm": 0.5741323828697205, "learning_rate": 0.00011768239811484344, "loss": 2.5902, "step": 13970 }, { "epoch": 4.8502868388311526e-14, "grad_norm": 0.5610677003860474, "learning_rate": 0.00011767136162488118, "loss": 2.6045, "step": 13980 }, { "epoch": 4.853756285783106e-14, "grad_norm": 0.5359668135643005, "learning_rate": 0.00011766031739842668, "loss": 2.5982, "step": 13990 }, { "epoch": 4.85722573273506e-14, "grad_norm": 0.5685226917266846, "learning_rate": 0.00011764926543722652, "loss": 2.5556, "step": 14000 }, { "epoch": 4.8606951796870135e-14, "grad_norm": 0.5761728882789612, "learning_rate": 0.00011763820574302844, "loss": 2.5927, "step": 14010 }, { "epoch": 4.864164626638967e-14, "grad_norm": 0.5258231163024902, "learning_rate": 0.00011762713831758147, "loss": 2.5459, "step": 14020 }, { "epoch": 4.867634073590921e-14, "grad_norm": 0.5527777075767517, "learning_rate": 0.00011761606316263581, "loss": 2.5838, "step": 14030 }, { "epoch": 4.8711035205428743e-14, "grad_norm": 0.574434757232666, "learning_rate": 0.0001176049802799429, "loss": 2.5515, "step": 14040 }, { "epoch": 4.874572967494828e-14, "grad_norm": 0.553231954574585, "learning_rate": 0.00011759388967125543, "loss": 2.574, "step": 14050 }, { "epoch": 4.8780424144467815e-14, "grad_norm": 0.5908355116844177, "learning_rate": 0.00011758279133832724, "loss": 2.5559, "step": 14060 }, { "epoch": 4.881511861398735e-14, "grad_norm": 0.5476428866386414, "learning_rate": 0.00011757168528291344, "loss": 2.5719, "step": 14070 }, { "epoch": 4.884981308350689e-14, "grad_norm": 0.5303199291229248, "learning_rate": 0.00011756057150677038, "loss": 2.5657, "step": 14080 }, { "epoch": 4.8884507553026424e-14, "grad_norm": 0.5466170310974121, "learning_rate": 0.00011754945001165561, "loss": 2.5216, "step": 14090 }, { "epoch": 4.891920202254596e-14, "grad_norm": 0.5349481105804443, "learning_rate": 0.00011753832079932788, "loss": 2.5556, "step": 14100 }, { "epoch": 4.8953896492065496e-14, "grad_norm": 0.571894645690918, "learning_rate": 0.0001175271838715472, "loss": 2.5648, "step": 14110 }, { "epoch": 4.898859096158503e-14, "grad_norm": 0.5681653022766113, "learning_rate": 0.00011751603923007476, "loss": 2.5712, "step": 14120 }, { "epoch": 4.902328543110457e-14, "grad_norm": 0.5367128849029541, "learning_rate": 0.000117504886876673, "loss": 2.5526, "step": 14130 }, { "epoch": 4.9057979900624105e-14, "grad_norm": 0.5409322381019592, "learning_rate": 0.00011749372681310556, "loss": 2.5663, "step": 14140 }, { "epoch": 4.909267437014364e-14, "grad_norm": 0.5490410923957825, "learning_rate": 0.00011748255904113731, "loss": 2.5615, "step": 14150 }, { "epoch": 4.912736883966318e-14, "grad_norm": 0.5343555808067322, "learning_rate": 0.00011747138356253436, "loss": 2.5461, "step": 14160 }, { "epoch": 4.916206330918271e-14, "grad_norm": 0.5514886379241943, "learning_rate": 0.00011746020037906397, "loss": 2.5306, "step": 14170 }, { "epoch": 4.919675777870225e-14, "grad_norm": 0.5801485180854797, "learning_rate": 0.0001174490094924947, "loss": 2.5959, "step": 14180 }, { "epoch": 4.9231452248221785e-14, "grad_norm": 0.5687825679779053, "learning_rate": 0.00011743781090459629, "loss": 2.5915, "step": 14190 }, { "epoch": 4.926614671774132e-14, "grad_norm": 0.5580301284790039, "learning_rate": 0.00011742660461713969, "loss": 2.5466, "step": 14200 }, { "epoch": 4.930084118726086e-14, "grad_norm": 0.5375584363937378, "learning_rate": 0.0001174153906318971, "loss": 2.5648, "step": 14210 }, { "epoch": 4.9335535656780394e-14, "grad_norm": 0.5548380613327026, "learning_rate": 0.00011740416895064188, "loss": 2.5733, "step": 14220 }, { "epoch": 4.937023012629993e-14, "grad_norm": 0.5488592982292175, "learning_rate": 0.00011739293957514864, "loss": 2.582, "step": 14230 }, { "epoch": 4.9404924595819466e-14, "grad_norm": 0.5376275181770325, "learning_rate": 0.00011738170250719324, "loss": 2.561, "step": 14240 }, { "epoch": 4.9439619065339e-14, "grad_norm": 0.5361543297767639, "learning_rate": 0.00011737045774855272, "loss": 2.5991, "step": 14250 }, { "epoch": 4.947431353485854e-14, "grad_norm": 0.5696021318435669, "learning_rate": 0.00011735920530100533, "loss": 2.5449, "step": 14260 }, { "epoch": 4.9509008004378074e-14, "grad_norm": 0.5537202954292297, "learning_rate": 0.00011734794516633056, "loss": 2.5568, "step": 14270 }, { "epoch": 4.954370247389761e-14, "grad_norm": 0.5594831705093384, "learning_rate": 0.00011733667734630908, "loss": 2.5309, "step": 14280 }, { "epoch": 4.9578396943417147e-14, "grad_norm": 0.5532917380332947, "learning_rate": 0.00011732540184272284, "loss": 2.5769, "step": 14290 }, { "epoch": 4.961309141293668e-14, "grad_norm": 0.5452396869659424, "learning_rate": 0.0001173141186573549, "loss": 2.5617, "step": 14300 }, { "epoch": 4.964778588245622e-14, "grad_norm": 0.5379423499107361, "learning_rate": 0.00011730282779198965, "loss": 2.6028, "step": 14310 }, { "epoch": 4.9682480351975755e-14, "grad_norm": 0.5519096255302429, "learning_rate": 0.00011729152924841266, "loss": 2.5415, "step": 14320 }, { "epoch": 4.971717482149529e-14, "grad_norm": 0.5651238560676575, "learning_rate": 0.00011728022302841064, "loss": 2.5434, "step": 14330 }, { "epoch": 4.975186929101483e-14, "grad_norm": 0.5512981414794922, "learning_rate": 0.00011726890913377162, "loss": 2.5633, "step": 14340 }, { "epoch": 4.9786563760534364e-14, "grad_norm": 0.51771479845047, "learning_rate": 0.00011725758756628474, "loss": 2.5391, "step": 14350 }, { "epoch": 4.98212582300539e-14, "grad_norm": 0.5614038109779358, "learning_rate": 0.00011724625832774047, "loss": 2.5874, "step": 14360 }, { "epoch": 4.9855952699573436e-14, "grad_norm": 0.5777345299720764, "learning_rate": 0.0001172349214199304, "loss": 2.5368, "step": 14370 }, { "epoch": 4.989064716909297e-14, "grad_norm": 0.5544854998588562, "learning_rate": 0.00011722357684464736, "loss": 2.6046, "step": 14380 }, { "epoch": 4.992534163861251e-14, "grad_norm": 0.529388427734375, "learning_rate": 0.00011721222460368541, "loss": 2.5534, "step": 14390 }, { "epoch": 4.9960036108132044e-14, "grad_norm": 0.5564414858818054, "learning_rate": 0.00011720086469883982, "loss": 2.54, "step": 14400 }, { "epoch": 4.999473057765158e-14, "grad_norm": 0.552510678768158, "learning_rate": 0.00011718949713190702, "loss": 2.5574, "step": 14410 }, { "epoch": 5.0029425047171117e-14, "grad_norm": 0.5445631742477417, "learning_rate": 0.00011717812190468475, "loss": 2.5478, "step": 14420 }, { "epoch": 5.006411951669065e-14, "grad_norm": 0.5174502730369568, "learning_rate": 0.00011716673901897183, "loss": 2.5599, "step": 14430 }, { "epoch": 5.009881398621019e-14, "grad_norm": 0.5761746764183044, "learning_rate": 0.00011715534847656843, "loss": 2.5561, "step": 14440 }, { "epoch": 5.0133508455729725e-14, "grad_norm": 0.5453391075134277, "learning_rate": 0.00011714395027927585, "loss": 2.5609, "step": 14450 }, { "epoch": 5.016820292524926e-14, "grad_norm": 0.5521064400672913, "learning_rate": 0.00011713254442889661, "loss": 2.5583, "step": 14460 }, { "epoch": 5.02028973947688e-14, "grad_norm": 0.5426429510116577, "learning_rate": 0.00011712113092723445, "loss": 2.5237, "step": 14470 }, { "epoch": 5.0237591864288333e-14, "grad_norm": 0.5781782269477844, "learning_rate": 0.0001171097097760943, "loss": 2.5883, "step": 14480 }, { "epoch": 5.027228633380787e-14, "grad_norm": 0.5599090456962585, "learning_rate": 0.00011709828097728234, "loss": 2.5778, "step": 14490 }, { "epoch": 5.0306980803327406e-14, "grad_norm": 0.5537056922912598, "learning_rate": 0.00011708684453260594, "loss": 2.5399, "step": 14500 }, { "epoch": 5.034167527284694e-14, "grad_norm": 0.5456365346908569, "learning_rate": 0.00011707540044387366, "loss": 2.5783, "step": 14510 }, { "epoch": 5.037636974236648e-14, "grad_norm": 0.54795902967453, "learning_rate": 0.00011706394871289529, "loss": 2.5676, "step": 14520 }, { "epoch": 5.0411064211886014e-14, "grad_norm": 0.5519582629203796, "learning_rate": 0.0001170524893414818, "loss": 2.5733, "step": 14530 }, { "epoch": 5.044575868140555e-14, "grad_norm": 0.5288481116294861, "learning_rate": 0.00011704102233144543, "loss": 2.5937, "step": 14540 }, { "epoch": 5.0480453150925086e-14, "grad_norm": 0.5995100736618042, "learning_rate": 0.00011702954768459956, "loss": 2.5212, "step": 14550 }, { "epoch": 5.051514762044462e-14, "grad_norm": 0.5527558326721191, "learning_rate": 0.00011701806540275883, "loss": 2.5848, "step": 14560 }, { "epoch": 5.054984208996416e-14, "grad_norm": 0.555608868598938, "learning_rate": 0.00011700657548773903, "loss": 2.5726, "step": 14570 }, { "epoch": 5.0584536559483695e-14, "grad_norm": 0.541709303855896, "learning_rate": 0.00011699507794135723, "loss": 2.5798, "step": 14580 }, { "epoch": 5.061923102900323e-14, "grad_norm": 0.5199000239372253, "learning_rate": 0.00011698357276543163, "loss": 2.5751, "step": 14590 }, { "epoch": 5.065392549852277e-14, "grad_norm": 0.5290905833244324, "learning_rate": 0.00011697205996178172, "loss": 2.5737, "step": 14600 }, { "epoch": 5.0688619968042303e-14, "grad_norm": 0.5537115335464478, "learning_rate": 0.00011696053953222812, "loss": 2.5435, "step": 14610 }, { "epoch": 5.072331443756184e-14, "grad_norm": 0.5295954346656799, "learning_rate": 0.00011694901147859267, "loss": 2.5742, "step": 14620 }, { "epoch": 5.0758008907081376e-14, "grad_norm": 0.5597634315490723, "learning_rate": 0.00011693747580269848, "loss": 2.559, "step": 14630 }, { "epoch": 5.079270337660091e-14, "grad_norm": 0.5364637970924377, "learning_rate": 0.00011692593250636978, "loss": 2.544, "step": 14640 }, { "epoch": 5.082739784612045e-14, "grad_norm": 0.6161866784095764, "learning_rate": 0.00011691438159143208, "loss": 2.5901, "step": 14650 }, { "epoch": 5.0862092315639984e-14, "grad_norm": 0.534414529800415, "learning_rate": 0.00011690282305971199, "loss": 2.5509, "step": 14660 }, { "epoch": 5.089678678515952e-14, "grad_norm": 0.5473677515983582, "learning_rate": 0.00011689125691303745, "loss": 2.5549, "step": 14670 }, { "epoch": 5.0931481254679056e-14, "grad_norm": 0.5224444270133972, "learning_rate": 0.00011687968315323756, "loss": 2.6162, "step": 14680 }, { "epoch": 5.096617572419859e-14, "grad_norm": 0.5379287004470825, "learning_rate": 0.00011686810178214255, "loss": 2.5655, "step": 14690 }, { "epoch": 5.100087019371813e-14, "grad_norm": 0.5700736045837402, "learning_rate": 0.00011685651280158396, "loss": 2.5749, "step": 14700 }, { "epoch": 5.1035564663237665e-14, "grad_norm": 0.559236466884613, "learning_rate": 0.00011684491621339448, "loss": 2.5421, "step": 14710 }, { "epoch": 5.10702591327572e-14, "grad_norm": 0.5382214188575745, "learning_rate": 0.00011683331201940799, "loss": 2.5546, "step": 14720 }, { "epoch": 5.110495360227674e-14, "grad_norm": 0.5729016661643982, "learning_rate": 0.00011682170022145962, "loss": 2.6251, "step": 14730 }, { "epoch": 5.113964807179627e-14, "grad_norm": 0.5396726727485657, "learning_rate": 0.00011681008082138566, "loss": 2.5794, "step": 14740 }, { "epoch": 5.117434254131581e-14, "grad_norm": 0.5531055331230164, "learning_rate": 0.00011679845382102361, "loss": 2.5475, "step": 14750 }, { "epoch": 5.1209037010835345e-14, "grad_norm": 0.5643865466117859, "learning_rate": 0.0001167868192222122, "loss": 2.5754, "step": 14760 }, { "epoch": 5.124373148035488e-14, "grad_norm": 0.5243997573852539, "learning_rate": 0.00011677517702679136, "loss": 2.5327, "step": 14770 }, { "epoch": 5.127842594987442e-14, "grad_norm": 0.5776858925819397, "learning_rate": 0.00011676352723660215, "loss": 2.5375, "step": 14780 }, { "epoch": 5.1313120419393954e-14, "grad_norm": 0.5442910194396973, "learning_rate": 0.00011675186985348691, "loss": 2.5411, "step": 14790 }, { "epoch": 5.134781488891349e-14, "grad_norm": 0.5380579233169556, "learning_rate": 0.00011674020487928918, "loss": 2.5711, "step": 14800 }, { "epoch": 5.1382509358433026e-14, "grad_norm": 0.5720306634902954, "learning_rate": 0.00011672853231585363, "loss": 2.5418, "step": 14810 }, { "epoch": 5.141720382795256e-14, "grad_norm": 0.6422322988510132, "learning_rate": 0.00011671685216502619, "loss": 2.5427, "step": 14820 }, { "epoch": 5.14518982974721e-14, "grad_norm": 0.542958676815033, "learning_rate": 0.00011670516442865399, "loss": 2.5616, "step": 14830 }, { "epoch": 5.1486592766991635e-14, "grad_norm": 0.5656629204750061, "learning_rate": 0.00011669346910858534, "loss": 2.5805, "step": 14840 }, { "epoch": 5.152128723651117e-14, "grad_norm": 0.5813406109809875, "learning_rate": 0.00011668176620666974, "loss": 2.5699, "step": 14850 }, { "epoch": 5.155598170603071e-14, "grad_norm": 0.5416033864021301, "learning_rate": 0.00011667005572475792, "loss": 2.5598, "step": 14860 }, { "epoch": 5.159067617555024e-14, "grad_norm": 0.5441697835922241, "learning_rate": 0.00011665833766470179, "loss": 2.549, "step": 14870 }, { "epoch": 5.162537064506978e-14, "grad_norm": 0.5306109189987183, "learning_rate": 0.00011664661202835446, "loss": 2.5739, "step": 14880 }, { "epoch": 5.1660065114589315e-14, "grad_norm": 0.5772628784179688, "learning_rate": 0.00011663487881757022, "loss": 2.5631, "step": 14890 }, { "epoch": 5.169475958410885e-14, "grad_norm": 0.5332619547843933, "learning_rate": 0.0001166231380342046, "loss": 2.569, "step": 14900 }, { "epoch": 5.172945405362839e-14, "grad_norm": 0.5505924820899963, "learning_rate": 0.00011661138968011428, "loss": 2.5692, "step": 14910 }, { "epoch": 5.1764148523147924e-14, "grad_norm": 0.5826533436775208, "learning_rate": 0.0001165996337571572, "loss": 2.5218, "step": 14920 }, { "epoch": 5.179884299266746e-14, "grad_norm": 0.5563478469848633, "learning_rate": 0.00011658787026719243, "loss": 2.5487, "step": 14930 }, { "epoch": 5.1833537462186996e-14, "grad_norm": 0.5934200882911682, "learning_rate": 0.00011657609921208025, "loss": 2.5908, "step": 14940 }, { "epoch": 5.186823193170653e-14, "grad_norm": 0.5359223484992981, "learning_rate": 0.00011656432059368218, "loss": 2.5475, "step": 14950 }, { "epoch": 5.190292640122607e-14, "grad_norm": 0.5529601573944092, "learning_rate": 0.0001165525344138609, "loss": 2.5442, "step": 14960 }, { "epoch": 5.1937620870745604e-14, "grad_norm": 0.5458753108978271, "learning_rate": 0.00011654074067448028, "loss": 2.5695, "step": 14970 }, { "epoch": 5.197231534026514e-14, "grad_norm": 0.5586483478546143, "learning_rate": 0.00011652893937740542, "loss": 2.5641, "step": 14980 }, { "epoch": 5.2007009809784677e-14, "grad_norm": 0.5747271776199341, "learning_rate": 0.00011651713052450258, "loss": 2.5444, "step": 14990 }, { "epoch": 5.204170427930421e-14, "grad_norm": 0.5256319046020508, "learning_rate": 0.00011650531411763923, "loss": 2.535, "step": 15000 }, { "epoch": 5.207639874882375e-14, "grad_norm": 0.5340853333473206, "learning_rate": 0.00011649349015868402, "loss": 2.5646, "step": 15010 }, { "epoch": 5.2111093218343285e-14, "grad_norm": 0.564560055732727, "learning_rate": 0.00011648165864950683, "loss": 2.5612, "step": 15020 }, { "epoch": 5.214578768786282e-14, "grad_norm": 0.5368573665618896, "learning_rate": 0.0001164698195919787, "loss": 2.5419, "step": 15030 }, { "epoch": 5.218048215738236e-14, "grad_norm": 0.5839037895202637, "learning_rate": 0.00011645797298797188, "loss": 2.5679, "step": 15040 }, { "epoch": 5.2215176626901894e-14, "grad_norm": 0.560958206653595, "learning_rate": 0.0001164461188393598, "loss": 2.5764, "step": 15050 }, { "epoch": 5.224987109642143e-14, "grad_norm": 0.60573810338974, "learning_rate": 0.0001164342571480171, "loss": 2.5646, "step": 15060 }, { "epoch": 5.2284565565940966e-14, "grad_norm": 0.5247768759727478, "learning_rate": 0.0001164223879158196, "loss": 2.5099, "step": 15070 }, { "epoch": 5.23192600354605e-14, "grad_norm": 0.5521934628486633, "learning_rate": 0.00011641051114464432, "loss": 2.5404, "step": 15080 }, { "epoch": 5.235395450498004e-14, "grad_norm": 0.5505820512771606, "learning_rate": 0.00011639862683636946, "loss": 2.5889, "step": 15090 }, { "epoch": 5.2388648974499574e-14, "grad_norm": 0.633589506149292, "learning_rate": 0.00011638673499287444, "loss": 2.5611, "step": 15100 }, { "epoch": 5.242334344401911e-14, "grad_norm": 0.5748184323310852, "learning_rate": 0.00011637483561603982, "loss": 2.5672, "step": 15110 }, { "epoch": 5.2458037913538647e-14, "grad_norm": 0.5151926279067993, "learning_rate": 0.00011636292870774742, "loss": 2.534, "step": 15120 }, { "epoch": 5.249273238305818e-14, "grad_norm": 0.5578928589820862, "learning_rate": 0.0001163510142698802, "loss": 2.5807, "step": 15130 }, { "epoch": 5.252742685257772e-14, "grad_norm": 0.5663720965385437, "learning_rate": 0.00011633909230432232, "loss": 2.5768, "step": 15140 }, { "epoch": 5.2562121322097255e-14, "grad_norm": 0.5544734001159668, "learning_rate": 0.00011632716281295916, "loss": 2.573, "step": 15150 }, { "epoch": 5.259681579161679e-14, "grad_norm": 0.6001941561698914, "learning_rate": 0.00011631522579767721, "loss": 2.5854, "step": 15160 }, { "epoch": 5.263151026113633e-14, "grad_norm": 0.5568806529045105, "learning_rate": 0.00011630328126036428, "loss": 2.5779, "step": 15170 }, { "epoch": 5.2666204730655863e-14, "grad_norm": 0.5305020213127136, "learning_rate": 0.00011629132920290925, "loss": 2.5698, "step": 15180 }, { "epoch": 5.27008992001754e-14, "grad_norm": 0.5353072881698608, "learning_rate": 0.00011627936962720225, "loss": 2.5752, "step": 15190 }, { "epoch": 5.2735593669694936e-14, "grad_norm": 0.5480389595031738, "learning_rate": 0.00011626740253513457, "loss": 2.5855, "step": 15200 }, { "epoch": 5.277028813921447e-14, "grad_norm": 0.5681470632553101, "learning_rate": 0.00011625542792859869, "loss": 2.5584, "step": 15210 }, { "epoch": 5.280498260873401e-14, "grad_norm": 0.5390207767486572, "learning_rate": 0.00011624344580948836, "loss": 2.5091, "step": 15220 }, { "epoch": 5.2839677078253544e-14, "grad_norm": 0.5354315638542175, "learning_rate": 0.00011623145617969838, "loss": 2.5519, "step": 15230 }, { "epoch": 5.287437154777308e-14, "grad_norm": 0.5431820154190063, "learning_rate": 0.00011621945904112481, "loss": 2.532, "step": 15240 }, { "epoch": 5.2909066017292616e-14, "grad_norm": 0.5721626281738281, "learning_rate": 0.00011620745439566494, "loss": 2.5974, "step": 15250 }, { "epoch": 5.294376048681215e-14, "grad_norm": 0.5566370487213135, "learning_rate": 0.00011619544224521715, "loss": 2.5447, "step": 15260 }, { "epoch": 5.297845495633169e-14, "grad_norm": 0.5757797956466675, "learning_rate": 0.00011618342259168109, "loss": 2.5275, "step": 15270 }, { "epoch": 5.3013149425851225e-14, "grad_norm": 0.5613824725151062, "learning_rate": 0.00011617139543695756, "loss": 2.5937, "step": 15280 }, { "epoch": 5.304784389537076e-14, "grad_norm": 0.7483511567115784, "learning_rate": 0.00011615936078294856, "loss": 2.5561, "step": 15290 }, { "epoch": 5.30825383648903e-14, "grad_norm": 0.5827199816703796, "learning_rate": 0.00011614731863155725, "loss": 2.5274, "step": 15300 }, { "epoch": 5.3117232834409833e-14, "grad_norm": 0.5997164845466614, "learning_rate": 0.00011613526898468797, "loss": 2.511, "step": 15310 }, { "epoch": 5.315192730392937e-14, "grad_norm": 0.5375935435295105, "learning_rate": 0.00011612321184424633, "loss": 2.5415, "step": 15320 }, { "epoch": 5.3186621773448906e-14, "grad_norm": 0.5964475870132446, "learning_rate": 0.00011611114721213903, "loss": 2.5503, "step": 15330 }, { "epoch": 5.322131624296844e-14, "grad_norm": 0.5353571176528931, "learning_rate": 0.00011609907509027397, "loss": 2.5208, "step": 15340 }, { "epoch": 5.325601071248798e-14, "grad_norm": 0.5662490129470825, "learning_rate": 0.00011608699548056027, "loss": 2.5626, "step": 15350 }, { "epoch": 5.3290705182007514e-14, "grad_norm": 0.5450071692466736, "learning_rate": 0.0001160749083849082, "loss": 2.5313, "step": 15360 }, { "epoch": 5.332539965152705e-14, "grad_norm": 0.5414485931396484, "learning_rate": 0.00011606281380522926, "loss": 2.5088, "step": 15370 }, { "epoch": 5.3360094121046586e-14, "grad_norm": 0.5370718836784363, "learning_rate": 0.00011605071174343608, "loss": 2.5475, "step": 15380 }, { "epoch": 5.339478859056612e-14, "grad_norm": 0.5550580024719238, "learning_rate": 0.00011603860220144252, "loss": 2.5699, "step": 15390 }, { "epoch": 5.342948306008566e-14, "grad_norm": 0.561282753944397, "learning_rate": 0.00011602648518116357, "loss": 2.5314, "step": 15400 }, { "epoch": 5.3464177529605195e-14, "grad_norm": 0.5301901698112488, "learning_rate": 0.00011601436068451548, "loss": 2.5718, "step": 15410 }, { "epoch": 5.349887199912473e-14, "grad_norm": 0.5355451703071594, "learning_rate": 0.00011600222871341555, "loss": 2.5906, "step": 15420 }, { "epoch": 5.353356646864427e-14, "grad_norm": 0.5401766300201416, "learning_rate": 0.00011599008926978242, "loss": 2.5616, "step": 15430 }, { "epoch": 5.35682609381638e-14, "grad_norm": 0.5213943719863892, "learning_rate": 0.00011597794235553583, "loss": 2.5396, "step": 15440 }, { "epoch": 5.360295540768334e-14, "grad_norm": 0.526264488697052, "learning_rate": 0.00011596578797259668, "loss": 2.5383, "step": 15450 }, { "epoch": 5.3637649877202875e-14, "grad_norm": 0.5283395051956177, "learning_rate": 0.00011595362612288712, "loss": 2.5265, "step": 15460 }, { "epoch": 5.367234434672241e-14, "grad_norm": 0.5648586750030518, "learning_rate": 0.0001159414568083304, "loss": 2.5457, "step": 15470 }, { "epoch": 5.370703881624195e-14, "grad_norm": 0.5128897428512573, "learning_rate": 0.00011592928003085104, "loss": 2.5603, "step": 15480 }, { "epoch": 5.3741733285761484e-14, "grad_norm": 0.5494193434715271, "learning_rate": 0.00011591709579237464, "loss": 2.5267, "step": 15490 }, { "epoch": 5.377642775528102e-14, "grad_norm": 0.5784398317337036, "learning_rate": 0.00011590490409482805, "loss": 2.596, "step": 15500 }, { "epoch": 5.3811122224800556e-14, "grad_norm": 0.5181556940078735, "learning_rate": 0.00011589270494013931, "loss": 2.5193, "step": 15510 }, { "epoch": 5.384581669432009e-14, "grad_norm": 0.5153165459632874, "learning_rate": 0.00011588049833023758, "loss": 2.5407, "step": 15520 }, { "epoch": 5.388051116383963e-14, "grad_norm": 0.5639307498931885, "learning_rate": 0.00011586828426705324, "loss": 2.5517, "step": 15530 }, { "epoch": 5.3915205633359164e-14, "grad_norm": 0.5338780283927917, "learning_rate": 0.00011585606275251785, "loss": 2.5474, "step": 15540 }, { "epoch": 5.39499001028787e-14, "grad_norm": 0.5461327433586121, "learning_rate": 0.00011584383378856412, "loss": 2.5341, "step": 15550 }, { "epoch": 5.3984594572398237e-14, "grad_norm": 0.5791953206062317, "learning_rate": 0.00011583159737712598, "loss": 2.5263, "step": 15560 }, { "epoch": 5.401928904191777e-14, "grad_norm": 0.526917040348053, "learning_rate": 0.00011581935352013848, "loss": 2.5414, "step": 15570 }, { "epoch": 5.405398351143731e-14, "grad_norm": 0.5548356175422668, "learning_rate": 0.00011580710221953789, "loss": 2.4873, "step": 15580 }, { "epoch": 5.4088677980956845e-14, "grad_norm": 0.5661717057228088, "learning_rate": 0.00011579484347726167, "loss": 2.5244, "step": 15590 }, { "epoch": 5.412337245047638e-14, "grad_norm": 0.5400513410568237, "learning_rate": 0.00011578257729524841, "loss": 2.5741, "step": 15600 }, { "epoch": 5.415806691999592e-14, "grad_norm": 0.5538696050643921, "learning_rate": 0.0001157703036754379, "loss": 2.54, "step": 15610 }, { "epoch": 5.4192761389515454e-14, "grad_norm": 0.5118833184242249, "learning_rate": 0.00011575802261977114, "loss": 2.5599, "step": 15620 }, { "epoch": 5.422745585903499e-14, "grad_norm": 0.5391678810119629, "learning_rate": 0.00011574573413019024, "loss": 2.5592, "step": 15630 }, { "epoch": 5.4262150328554526e-14, "grad_norm": 0.5180425047874451, "learning_rate": 0.00011573343820863854, "loss": 2.5187, "step": 15640 }, { "epoch": 5.429684479807406e-14, "grad_norm": 0.5640853047370911, "learning_rate": 0.00011572113485706051, "loss": 2.5694, "step": 15650 }, { "epoch": 5.43315392675936e-14, "grad_norm": 0.561102032661438, "learning_rate": 0.00011570882407740183, "loss": 2.5506, "step": 15660 }, { "epoch": 5.4366233737113134e-14, "grad_norm": 0.537375271320343, "learning_rate": 0.00011569650587160934, "loss": 2.5504, "step": 15670 }, { "epoch": 5.440092820663267e-14, "grad_norm": 0.5470637679100037, "learning_rate": 0.0001156841802416311, "loss": 2.5554, "step": 15680 }, { "epoch": 5.4435622676152207e-14, "grad_norm": 0.5525790452957153, "learning_rate": 0.00011567184718941623, "loss": 2.535, "step": 15690 }, { "epoch": 5.447031714567174e-14, "grad_norm": 0.5441362857818604, "learning_rate": 0.00011565950671691513, "loss": 2.5386, "step": 15700 }, { "epoch": 5.450501161519128e-14, "grad_norm": 0.5305981636047363, "learning_rate": 0.00011564715882607938, "loss": 2.56, "step": 15710 }, { "epoch": 5.4539706084710815e-14, "grad_norm": 0.5535471439361572, "learning_rate": 0.00011563480351886163, "loss": 2.5289, "step": 15720 }, { "epoch": 5.457440055423035e-14, "grad_norm": 0.5360457897186279, "learning_rate": 0.00011562244079721578, "loss": 2.558, "step": 15730 }, { "epoch": 5.460909502374989e-14, "grad_norm": 0.5244848728179932, "learning_rate": 0.00011561007066309693, "loss": 2.5691, "step": 15740 }, { "epoch": 5.4643789493269423e-14, "grad_norm": 0.5497969388961792, "learning_rate": 0.00011559769311846126, "loss": 2.5413, "step": 15750 }, { "epoch": 5.467848396278896e-14, "grad_norm": 0.5356274247169495, "learning_rate": 0.00011558530816526619, "loss": 2.5195, "step": 15760 }, { "epoch": 5.4713178432308496e-14, "grad_norm": 0.5813896059989929, "learning_rate": 0.0001155729158054703, "loss": 2.5175, "step": 15770 }, { "epoch": 5.474787290182803e-14, "grad_norm": 0.5635690689086914, "learning_rate": 0.00011556051604103332, "loss": 2.5365, "step": 15780 }, { "epoch": 5.478256737134757e-14, "grad_norm": 0.5746093988418579, "learning_rate": 0.00011554810887391619, "loss": 2.5744, "step": 15790 }, { "epoch": 5.4817261840867104e-14, "grad_norm": 0.5539220571517944, "learning_rate": 0.00011553569430608098, "loss": 2.5299, "step": 15800 }, { "epoch": 5.485195631038664e-14, "grad_norm": 0.5210923552513123, "learning_rate": 0.00011552327233949094, "loss": 2.5347, "step": 15810 }, { "epoch": 5.4886650779906176e-14, "grad_norm": 0.5245906710624695, "learning_rate": 0.0001155108429761105, "loss": 2.561, "step": 15820 }, { "epoch": 5.492134524942571e-14, "grad_norm": 0.5548833608627319, "learning_rate": 0.00011549840621790529, "loss": 2.5899, "step": 15830 }, { "epoch": 5.495603971894525e-14, "grad_norm": 0.5443552732467651, "learning_rate": 0.00011548596206684203, "loss": 2.5646, "step": 15840 }, { "epoch": 5.4990734188464785e-14, "grad_norm": 0.5679439902305603, "learning_rate": 0.00011547351052488868, "loss": 2.5261, "step": 15850 }, { "epoch": 5.502542865798432e-14, "grad_norm": 0.5438655614852905, "learning_rate": 0.00011546105159401435, "loss": 2.5585, "step": 15860 }, { "epoch": 5.506012312750386e-14, "grad_norm": 0.5614205002784729, "learning_rate": 0.00011544858527618929, "loss": 2.5791, "step": 15870 }, { "epoch": 5.5094817597023393e-14, "grad_norm": 0.5507932901382446, "learning_rate": 0.00011543611157338494, "loss": 2.5506, "step": 15880 }, { "epoch": 5.512951206654293e-14, "grad_norm": 0.5429747700691223, "learning_rate": 0.00011542363048757394, "loss": 2.5515, "step": 15890 }, { "epoch": 5.5164206536062466e-14, "grad_norm": 0.5417023301124573, "learning_rate": 0.00011541114202073004, "loss": 2.5555, "step": 15900 }, { "epoch": 5.5198901005582e-14, "grad_norm": 0.5488260984420776, "learning_rate": 0.0001153986461748282, "loss": 2.5388, "step": 15910 }, { "epoch": 5.523359547510154e-14, "grad_norm": 0.5407751202583313, "learning_rate": 0.00011538614295184452, "loss": 2.5508, "step": 15920 }, { "epoch": 5.5268289944621074e-14, "grad_norm": 0.5396977066993713, "learning_rate": 0.00011537363235375628, "loss": 2.572, "step": 15930 }, { "epoch": 5.530298441414061e-14, "grad_norm": 0.5705907344818115, "learning_rate": 0.00011536111438254191, "loss": 2.531, "step": 15940 }, { "epoch": 5.5337678883660146e-14, "grad_norm": 0.5599509477615356, "learning_rate": 0.00011534858904018106, "loss": 2.5207, "step": 15950 }, { "epoch": 5.537237335317968e-14, "grad_norm": 0.5833021998405457, "learning_rate": 0.00011533605632865444, "loss": 2.5705, "step": 15960 }, { "epoch": 5.540706782269922e-14, "grad_norm": 0.558687686920166, "learning_rate": 0.00011532351624994406, "loss": 2.5395, "step": 15970 }, { "epoch": 5.5441762292218755e-14, "grad_norm": 0.5186316967010498, "learning_rate": 0.00011531096880603295, "loss": 2.5658, "step": 15980 }, { "epoch": 5.547645676173829e-14, "grad_norm": 0.5533387064933777, "learning_rate": 0.00011529841399890545, "loss": 2.5361, "step": 15990 }, { "epoch": 5.551115123125783e-14, "grad_norm": 0.5393924713134766, "learning_rate": 0.00011528585183054698, "loss": 2.5256, "step": 16000 }, { "epoch": 5.5545845700777363e-14, "grad_norm": 0.5295600295066833, "learning_rate": 0.00011527328230294409, "loss": 2.5423, "step": 16010 }, { "epoch": 5.55805401702969e-14, "grad_norm": 0.5582622289657593, "learning_rate": 0.00011526070541808461, "loss": 2.5019, "step": 16020 }, { "epoch": 5.5615234639816435e-14, "grad_norm": 0.5129712224006653, "learning_rate": 0.00011524812117795744, "loss": 2.5223, "step": 16030 }, { "epoch": 5.564992910933597e-14, "grad_norm": 0.5804534554481506, "learning_rate": 0.00011523552958455263, "loss": 2.522, "step": 16040 }, { "epoch": 5.568462357885551e-14, "grad_norm": 0.5604347586631775, "learning_rate": 0.0001152229306398615, "loss": 2.5417, "step": 16050 }, { "epoch": 5.5719318048375044e-14, "grad_norm": 0.5249770283699036, "learning_rate": 0.00011521032434587641, "loss": 2.567, "step": 16060 }, { "epoch": 5.575401251789458e-14, "grad_norm": 0.5818070769309998, "learning_rate": 0.00011519771070459098, "loss": 2.5578, "step": 16070 }, { "epoch": 5.5788706987414116e-14, "grad_norm": 0.5565850734710693, "learning_rate": 0.0001151850897179999, "loss": 2.5519, "step": 16080 }, { "epoch": 5.582340145693365e-14, "grad_norm": 0.5391707420349121, "learning_rate": 0.00011517246138809911, "loss": 2.5285, "step": 16090 }, { "epoch": 5.585809592645319e-14, "grad_norm": 0.5336205959320068, "learning_rate": 0.00011515982571688567, "loss": 2.5242, "step": 16100 }, { "epoch": 5.5892790395972725e-14, "grad_norm": 0.5376940965652466, "learning_rate": 0.00011514718270635776, "loss": 2.4809, "step": 16110 }, { "epoch": 5.592748486549226e-14, "grad_norm": 0.5394768714904785, "learning_rate": 0.00011513453235851484, "loss": 2.5463, "step": 16120 }, { "epoch": 5.59621793350118e-14, "grad_norm": 0.5739537477493286, "learning_rate": 0.00011512187467535737, "loss": 2.5743, "step": 16130 }, { "epoch": 5.599687380453133e-14, "grad_norm": 0.5189250111579895, "learning_rate": 0.0001151092096588871, "loss": 2.5429, "step": 16140 }, { "epoch": 5.603156827405087e-14, "grad_norm": 0.55143141746521, "learning_rate": 0.0001150965373111069, "loss": 2.5461, "step": 16150 }, { "epoch": 5.6066262743570405e-14, "grad_norm": 0.5344709753990173, "learning_rate": 0.00011508385763402076, "loss": 2.5582, "step": 16160 }, { "epoch": 5.610095721308994e-14, "grad_norm": 0.5284020304679871, "learning_rate": 0.0001150711706296339, "loss": 2.5755, "step": 16170 }, { "epoch": 5.613565168260948e-14, "grad_norm": 0.5606048703193665, "learning_rate": 0.00011505847629995263, "loss": 2.5297, "step": 16180 }, { "epoch": 5.6170346152129014e-14, "grad_norm": 0.5551612377166748, "learning_rate": 0.00011504577464698446, "loss": 2.5226, "step": 16190 }, { "epoch": 5.620504062164855e-14, "grad_norm": 0.5223903059959412, "learning_rate": 0.00011503306567273803, "loss": 2.51, "step": 16200 }, { "epoch": 5.6239735091168086e-14, "grad_norm": 0.5345178246498108, "learning_rate": 0.00011502034937922319, "loss": 2.5583, "step": 16210 }, { "epoch": 5.627442956068762e-14, "grad_norm": 0.5461723208427429, "learning_rate": 0.0001150076257684509, "loss": 2.5245, "step": 16220 }, { "epoch": 5.630912403020716e-14, "grad_norm": 0.5746565461158752, "learning_rate": 0.00011499489484243328, "loss": 2.5743, "step": 16230 }, { "epoch": 5.6343818499726694e-14, "grad_norm": 0.5508471131324768, "learning_rate": 0.00011498215660318359, "loss": 2.5536, "step": 16240 }, { "epoch": 5.637851296924623e-14, "grad_norm": 0.5428469181060791, "learning_rate": 0.00011496941105271632, "loss": 2.5752, "step": 16250 }, { "epoch": 5.6413207438765767e-14, "grad_norm": 0.5340019464492798, "learning_rate": 0.00011495665819304706, "loss": 2.5198, "step": 16260 }, { "epoch": 5.64479019082853e-14, "grad_norm": 0.5428141951560974, "learning_rate": 0.00011494389802619255, "loss": 2.5328, "step": 16270 }, { "epoch": 5.648259637780484e-14, "grad_norm": 0.5191581845283508, "learning_rate": 0.00011493113055417072, "loss": 2.5366, "step": 16280 }, { "epoch": 5.6517290847324375e-14, "grad_norm": 0.5232183933258057, "learning_rate": 0.00011491835577900062, "loss": 2.5627, "step": 16290 }, { "epoch": 5.655198531684391e-14, "grad_norm": 0.5586358308792114, "learning_rate": 0.00011490557370270246, "loss": 2.5373, "step": 16300 }, { "epoch": 5.658667978636345e-14, "grad_norm": 0.5445131063461304, "learning_rate": 0.00011489278432729762, "loss": 2.5356, "step": 16310 }, { "epoch": 5.6621374255882984e-14, "grad_norm": 0.5339245796203613, "learning_rate": 0.00011487998765480865, "loss": 2.5042, "step": 16320 }, { "epoch": 5.665606872540252e-14, "grad_norm": 0.5061741471290588, "learning_rate": 0.00011486718368725923, "loss": 2.5612, "step": 16330 }, { "epoch": 5.6690763194922056e-14, "grad_norm": 0.5400462746620178, "learning_rate": 0.00011485437242667418, "loss": 2.578, "step": 16340 }, { "epoch": 5.672545766444159e-14, "grad_norm": 0.5363628268241882, "learning_rate": 0.00011484155387507952, "loss": 2.5731, "step": 16350 }, { "epoch": 5.676015213396113e-14, "grad_norm": 0.5839107036590576, "learning_rate": 0.00011482872803450233, "loss": 2.5499, "step": 16360 }, { "epoch": 5.6794846603480664e-14, "grad_norm": 0.5320642590522766, "learning_rate": 0.000114815894906971, "loss": 2.5768, "step": 16370 }, { "epoch": 5.68295410730002e-14, "grad_norm": 0.5700929164886475, "learning_rate": 0.00011480305449451489, "loss": 2.5606, "step": 16380 }, { "epoch": 5.686423554251974e-14, "grad_norm": 0.5623654127120972, "learning_rate": 0.00011479020679916466, "loss": 2.5606, "step": 16390 }, { "epoch": 5.689893001203927e-14, "grad_norm": 2.301480770111084, "learning_rate": 0.00011477735182295204, "loss": 2.5724, "step": 16400 }, { "epoch": 5.693362448155881e-14, "grad_norm": 0.5709985494613647, "learning_rate": 0.00011476448956790992, "loss": 2.5235, "step": 16410 }, { "epoch": 5.696831895107834e-14, "grad_norm": 0.5600557327270508, "learning_rate": 0.0001147516200360724, "loss": 2.5096, "step": 16420 }, { "epoch": 5.700301342059788e-14, "grad_norm": 0.5572566986083984, "learning_rate": 0.00011473874322947464, "loss": 2.5356, "step": 16430 }, { "epoch": 5.703770789011742e-14, "grad_norm": 0.5243510007858276, "learning_rate": 0.00011472585915015301, "loss": 2.5426, "step": 16440 }, { "epoch": 5.707240235963695e-14, "grad_norm": 0.5255234837532043, "learning_rate": 0.000114712967800145, "loss": 2.5526, "step": 16450 }, { "epoch": 5.710709682915649e-14, "grad_norm": 0.5184243321418762, "learning_rate": 0.00011470006918148931, "loss": 2.5491, "step": 16460 }, { "epoch": 5.714179129867603e-14, "grad_norm": 0.5674170851707458, "learning_rate": 0.00011468716329622571, "loss": 2.5463, "step": 16470 }, { "epoch": 5.717648576819556e-14, "grad_norm": 0.5391799211502075, "learning_rate": 0.00011467425014639517, "loss": 2.5578, "step": 16480 }, { "epoch": 5.72111802377151e-14, "grad_norm": 0.5484810471534729, "learning_rate": 0.00011466132973403979, "loss": 2.5363, "step": 16490 }, { "epoch": 5.724587470723463e-14, "grad_norm": 0.5319463014602661, "learning_rate": 0.0001146484020612028, "loss": 2.5791, "step": 16500 }, { "epoch": 5.728056917675417e-14, "grad_norm": 0.5631074905395508, "learning_rate": 0.00011463546712992864, "loss": 2.5268, "step": 16510 }, { "epoch": 5.731526364627371e-14, "grad_norm": 0.536239743232727, "learning_rate": 0.00011462252494226284, "loss": 2.5146, "step": 16520 }, { "epoch": 5.734995811579324e-14, "grad_norm": 0.5943508744239807, "learning_rate": 0.00011460957550025208, "loss": 2.5338, "step": 16530 }, { "epoch": 5.738465258531278e-14, "grad_norm": 0.9517995715141296, "learning_rate": 0.00011459661880594423, "loss": 2.5253, "step": 16540 }, { "epoch": 5.741934705483231e-14, "grad_norm": 0.5457494854927063, "learning_rate": 0.00011458365486138826, "loss": 2.5465, "step": 16550 }, { "epoch": 5.745404152435185e-14, "grad_norm": 0.5370190143585205, "learning_rate": 0.00011457068366863431, "loss": 2.5668, "step": 16560 }, { "epoch": 5.748873599387139e-14, "grad_norm": 0.5427057147026062, "learning_rate": 0.00011455770522973367, "loss": 2.5296, "step": 16570 }, { "epoch": 5.752343046339092e-14, "grad_norm": 0.5405125021934509, "learning_rate": 0.00011454471954673877, "loss": 2.5268, "step": 16580 }, { "epoch": 5.755812493291046e-14, "grad_norm": 0.536435067653656, "learning_rate": 0.00011453172662170315, "loss": 2.5161, "step": 16590 }, { "epoch": 5.759281940243e-14, "grad_norm": 0.6013194918632507, "learning_rate": 0.00011451872645668158, "loss": 2.553, "step": 16600 }, { "epoch": 5.762751387194953e-14, "grad_norm": 0.5528194904327393, "learning_rate": 0.0001145057190537299, "loss": 2.5017, "step": 16610 }, { "epoch": 5.766220834146907e-14, "grad_norm": 0.5527042746543884, "learning_rate": 0.00011449270441490511, "loss": 2.5197, "step": 16620 }, { "epoch": 5.76969028109886e-14, "grad_norm": 0.5475438237190247, "learning_rate": 0.00011447968254226537, "loss": 2.5419, "step": 16630 }, { "epoch": 5.773159728050814e-14, "grad_norm": 0.5729413628578186, "learning_rate": 0.00011446665343787, "loss": 2.5447, "step": 16640 }, { "epoch": 5.776629175002768e-14, "grad_norm": 0.5400980710983276, "learning_rate": 0.0001144536171037794, "loss": 2.5259, "step": 16650 }, { "epoch": 5.780098621954721e-14, "grad_norm": 0.5440073609352112, "learning_rate": 0.00011444057354205518, "loss": 2.5601, "step": 16660 }, { "epoch": 5.783568068906675e-14, "grad_norm": 0.55428147315979, "learning_rate": 0.00011442752275476006, "loss": 2.5213, "step": 16670 }, { "epoch": 5.787037515858628e-14, "grad_norm": 0.5320708751678467, "learning_rate": 0.00011441446474395792, "loss": 2.5242, "step": 16680 }, { "epoch": 5.790506962810582e-14, "grad_norm": 0.5731686353683472, "learning_rate": 0.00011440139951171375, "loss": 2.5009, "step": 16690 }, { "epoch": 5.793976409762536e-14, "grad_norm": 0.5450450778007507, "learning_rate": 0.00011438832706009373, "loss": 2.546, "step": 16700 }, { "epoch": 5.797445856714489e-14, "grad_norm": 0.5333134531974792, "learning_rate": 0.00011437524739116515, "loss": 2.539, "step": 16710 }, { "epoch": 5.800915303666443e-14, "grad_norm": 0.5719906687736511, "learning_rate": 0.00011436216050699642, "loss": 2.5675, "step": 16720 }, { "epoch": 5.804384750618397e-14, "grad_norm": 0.540851891040802, "learning_rate": 0.00011434906640965714, "loss": 2.5061, "step": 16730 }, { "epoch": 5.80785419757035e-14, "grad_norm": 0.5760574340820312, "learning_rate": 0.00011433596510121805, "loss": 2.5128, "step": 16740 }, { "epoch": 5.811323644522304e-14, "grad_norm": 0.5905861258506775, "learning_rate": 0.00011432285658375096, "loss": 2.5423, "step": 16750 }, { "epoch": 5.814793091474257e-14, "grad_norm": 0.5620747208595276, "learning_rate": 0.00011430974085932891, "loss": 2.5051, "step": 16760 }, { "epoch": 5.818262538426211e-14, "grad_norm": 0.5377799272537231, "learning_rate": 0.000114296617930026, "loss": 2.5312, "step": 16770 }, { "epoch": 5.821731985378165e-14, "grad_norm": 0.5396204590797424, "learning_rate": 0.00011428348779791755, "loss": 2.5528, "step": 16780 }, { "epoch": 5.825201432330118e-14, "grad_norm": 0.5298066139221191, "learning_rate": 0.00011427035046507997, "loss": 2.5454, "step": 16790 }, { "epoch": 5.828670879282072e-14, "grad_norm": 0.5812476873397827, "learning_rate": 0.00011425720593359079, "loss": 2.5768, "step": 16800 }, { "epoch": 5.832140326234025e-14, "grad_norm": 0.5467714071273804, "learning_rate": 0.00011424405420552872, "loss": 2.4836, "step": 16810 }, { "epoch": 5.835609773185979e-14, "grad_norm": 0.5447736382484436, "learning_rate": 0.00011423089528297362, "loss": 2.518, "step": 16820 }, { "epoch": 5.839079220137933e-14, "grad_norm": 0.5243205428123474, "learning_rate": 0.0001142177291680064, "loss": 2.5163, "step": 16830 }, { "epoch": 5.842548667089886e-14, "grad_norm": 0.5419864654541016, "learning_rate": 0.00011420455586270921, "loss": 2.5105, "step": 16840 }, { "epoch": 5.84601811404184e-14, "grad_norm": 0.5588043928146362, "learning_rate": 0.0001141913753691653, "loss": 2.5288, "step": 16850 }, { "epoch": 5.849487560993794e-14, "grad_norm": 5.650465965270996, "learning_rate": 0.00011417818768945902, "loss": 2.545, "step": 16860 }, { "epoch": 5.852957007945747e-14, "grad_norm": 0.5843492746353149, "learning_rate": 0.0001141649928256759, "loss": 2.524, "step": 16870 }, { "epoch": 5.856426454897701e-14, "grad_norm": 0.5527926683425903, "learning_rate": 0.00011415179077990263, "loss": 2.534, "step": 16880 }, { "epoch": 5.859895901849654e-14, "grad_norm": 0.5339957475662231, "learning_rate": 0.00011413858155422693, "loss": 2.5362, "step": 16890 }, { "epoch": 5.863365348801608e-14, "grad_norm": 0.5480198860168457, "learning_rate": 0.0001141253651507378, "loss": 2.5337, "step": 16900 }, { "epoch": 5.866834795753562e-14, "grad_norm": 0.5504103899002075, "learning_rate": 0.00011411214157152525, "loss": 2.5107, "step": 16910 }, { "epoch": 5.870304242705515e-14, "grad_norm": 0.6335641741752625, "learning_rate": 0.0001140989108186805, "loss": 2.5014, "step": 16920 }, { "epoch": 5.873773689657469e-14, "grad_norm": 0.5518343448638916, "learning_rate": 0.00011408567289429584, "loss": 2.4951, "step": 16930 }, { "epoch": 5.877243136609422e-14, "grad_norm": 0.5349966287612915, "learning_rate": 0.00011407242780046481, "loss": 2.5334, "step": 16940 }, { "epoch": 5.880712583561376e-14, "grad_norm": 0.5435717701911926, "learning_rate": 0.00011405917553928192, "loss": 2.5459, "step": 16950 }, { "epoch": 5.88418203051333e-14, "grad_norm": 0.5830144286155701, "learning_rate": 0.00011404591611284296, "loss": 2.5706, "step": 16960 }, { "epoch": 5.887651477465283e-14, "grad_norm": 0.5589125156402588, "learning_rate": 0.00011403264952324478, "loss": 2.5151, "step": 16970 }, { "epoch": 5.891120924417237e-14, "grad_norm": 0.5386729836463928, "learning_rate": 0.00011401937577258538, "loss": 2.5103, "step": 16980 }, { "epoch": 5.89459037136919e-14, "grad_norm": 0.559823215007782, "learning_rate": 0.00011400609486296387, "loss": 2.5388, "step": 16990 }, { "epoch": 5.898059818321144e-14, "grad_norm": 0.5390809774398804, "learning_rate": 0.00011399280679648052, "loss": 2.4727, "step": 17000 }, { "epoch": 5.901529265273098e-14, "grad_norm": 0.5837151408195496, "learning_rate": 0.00011397951157523674, "loss": 2.5512, "step": 17010 }, { "epoch": 5.904998712225051e-14, "grad_norm": 0.5374026298522949, "learning_rate": 0.00011396620920133503, "loss": 2.5178, "step": 17020 }, { "epoch": 5.908468159177005e-14, "grad_norm": 0.53143709897995, "learning_rate": 0.00011395289967687908, "loss": 2.5141, "step": 17030 }, { "epoch": 5.911937606128959e-14, "grad_norm": 0.5505455732345581, "learning_rate": 0.00011393958300397362, "loss": 2.5131, "step": 17040 }, { "epoch": 5.915407053080912e-14, "grad_norm": 0.5129123330116272, "learning_rate": 0.00011392625918472465, "loss": 2.4988, "step": 17050 }, { "epoch": 5.918876500032866e-14, "grad_norm": 0.6044517755508423, "learning_rate": 0.00011391292822123912, "loss": 2.5225, "step": 17060 }, { "epoch": 5.92234594698482e-14, "grad_norm": 0.5559740662574768, "learning_rate": 0.0001138995901156253, "loss": 2.5086, "step": 17070 }, { "epoch": 5.925815393936773e-14, "grad_norm": 0.5918977856636047, "learning_rate": 0.00011388624486999242, "loss": 2.5064, "step": 17080 }, { "epoch": 5.929284840888727e-14, "grad_norm": 0.5470765829086304, "learning_rate": 0.00011387289248645096, "loss": 2.5249, "step": 17090 }, { "epoch": 5.93275428784068e-14, "grad_norm": 0.5320781469345093, "learning_rate": 0.00011385953296711251, "loss": 2.5406, "step": 17100 }, { "epoch": 5.936223734792634e-14, "grad_norm": 0.567104697227478, "learning_rate": 0.00011384616631408968, "loss": 2.5415, "step": 17110 }, { "epoch": 5.939693181744587e-14, "grad_norm": 0.533332109451294, "learning_rate": 0.00011383279252949634, "loss": 2.5024, "step": 17120 }, { "epoch": 5.943162628696541e-14, "grad_norm": 0.5323771238327026, "learning_rate": 0.00011381941161544746, "loss": 2.4895, "step": 17130 }, { "epoch": 5.946632075648495e-14, "grad_norm": 0.5428128838539124, "learning_rate": 0.00011380602357405909, "loss": 2.5012, "step": 17140 }, { "epoch": 5.950101522600448e-14, "grad_norm": 0.508678138256073, "learning_rate": 0.00011379262840744842, "loss": 2.4823, "step": 17150 }, { "epoch": 5.953570969552402e-14, "grad_norm": 0.5860413908958435, "learning_rate": 0.00011377922611773381, "loss": 2.5129, "step": 17160 }, { "epoch": 5.957040416504356e-14, "grad_norm": 0.5267683267593384, "learning_rate": 0.00011376581670703469, "loss": 2.5282, "step": 17170 }, { "epoch": 5.960509863456309e-14, "grad_norm": 0.5486993193626404, "learning_rate": 0.00011375240017747166, "loss": 2.5106, "step": 17180 }, { "epoch": 5.963979310408263e-14, "grad_norm": 0.5839786529541016, "learning_rate": 0.00011373897653116644, "loss": 2.5121, "step": 17190 }, { "epoch": 5.967448757360216e-14, "grad_norm": 0.5442468523979187, "learning_rate": 0.00011372554577024185, "loss": 2.5103, "step": 17200 }, { "epoch": 5.97091820431217e-14, "grad_norm": 0.5417844653129578, "learning_rate": 0.00011371210789682183, "loss": 2.4847, "step": 17210 }, { "epoch": 5.974387651264124e-14, "grad_norm": 0.5374609231948853, "learning_rate": 0.0001136986629130315, "loss": 2.5259, "step": 17220 }, { "epoch": 5.977857098216077e-14, "grad_norm": 0.5553344488143921, "learning_rate": 0.00011368521082099705, "loss": 2.5113, "step": 17230 }, { "epoch": 5.981326545168031e-14, "grad_norm": 0.5479012131690979, "learning_rate": 0.0001136717516228458, "loss": 2.5023, "step": 17240 }, { "epoch": 5.984795992119984e-14, "grad_norm": 0.5643675923347473, "learning_rate": 0.00011365828532070623, "loss": 2.5259, "step": 17250 }, { "epoch": 5.988265439071938e-14, "grad_norm": 0.546603262424469, "learning_rate": 0.00011364481191670791, "loss": 2.5177, "step": 17260 }, { "epoch": 5.991734886023892e-14, "grad_norm": 0.5368894338607788, "learning_rate": 0.00011363133141298153, "loss": 2.511, "step": 17270 }, { "epoch": 5.995204332975845e-14, "grad_norm": 0.5368651747703552, "learning_rate": 0.00011361784381165895, "loss": 2.5252, "step": 17280 }, { "epoch": 5.998673779927799e-14, "grad_norm": 0.5610401630401611, "learning_rate": 0.00011360434911487307, "loss": 2.5119, "step": 17290 }, { "epoch": 6.002143226879753e-14, "grad_norm": 0.5276376605033875, "learning_rate": 0.000113590847324758, "loss": 2.5195, "step": 17300 }, { "epoch": 6.005612673831706e-14, "grad_norm": 0.5725240111351013, "learning_rate": 0.00011357733844344891, "loss": 2.5343, "step": 17310 }, { "epoch": 6.00908212078366e-14, "grad_norm": 0.5880289673805237, "learning_rate": 0.00011356382247308212, "loss": 2.5473, "step": 17320 }, { "epoch": 6.012551567735613e-14, "grad_norm": 0.5489155054092407, "learning_rate": 0.00011355029941579509, "loss": 2.497, "step": 17330 }, { "epoch": 6.016021014687567e-14, "grad_norm": 0.563105583190918, "learning_rate": 0.00011353676927372635, "loss": 2.5044, "step": 17340 }, { "epoch": 6.01949046163952e-14, "grad_norm": 0.5597757697105408, "learning_rate": 0.00011352323204901556, "loss": 2.5263, "step": 17350 }, { "epoch": 6.022959908591474e-14, "grad_norm": 0.5185726284980774, "learning_rate": 0.00011350968774380353, "loss": 2.5391, "step": 17360 }, { "epoch": 6.026429355543428e-14, "grad_norm": 0.5423191785812378, "learning_rate": 0.00011349613636023217, "loss": 2.5487, "step": 17370 }, { "epoch": 6.029898802495381e-14, "grad_norm": 0.5404453277587891, "learning_rate": 0.00011348257790044456, "loss": 2.49, "step": 17380 }, { "epoch": 6.033368249447335e-14, "grad_norm": 0.5483809113502502, "learning_rate": 0.00011346901236658479, "loss": 2.5348, "step": 17390 }, { "epoch": 6.036837696399289e-14, "grad_norm": 0.5183385014533997, "learning_rate": 0.00011345543976079813, "loss": 2.5484, "step": 17400 }, { "epoch": 6.040307143351242e-14, "grad_norm": 0.5443412661552429, "learning_rate": 0.00011344186008523105, "loss": 2.5244, "step": 17410 }, { "epoch": 6.043776590303196e-14, "grad_norm": 0.5335543751716614, "learning_rate": 0.00011342827334203099, "loss": 2.5388, "step": 17420 }, { "epoch": 6.04724603725515e-14, "grad_norm": 0.5167184472084045, "learning_rate": 0.00011341467953334658, "loss": 2.5548, "step": 17430 }, { "epoch": 6.050715484207103e-14, "grad_norm": 0.5353938937187195, "learning_rate": 0.0001134010786613276, "loss": 2.5183, "step": 17440 }, { "epoch": 6.054184931159057e-14, "grad_norm": 0.5373006463050842, "learning_rate": 0.00011338747072812487, "loss": 2.5705, "step": 17450 }, { "epoch": 6.05765437811101e-14, "grad_norm": 0.6030579805374146, "learning_rate": 0.00011337385573589039, "loss": 2.5462, "step": 17460 }, { "epoch": 6.061123825062964e-14, "grad_norm": 0.5316969156265259, "learning_rate": 0.00011336023368677726, "loss": 2.5149, "step": 17470 }, { "epoch": 6.064593272014918e-14, "grad_norm": 0.5319227576255798, "learning_rate": 0.00011334660458293966, "loss": 2.4864, "step": 17480 }, { "epoch": 6.068062718966871e-14, "grad_norm": 0.5413691401481628, "learning_rate": 0.00011333296842653294, "loss": 2.5373, "step": 17490 }, { "epoch": 6.071532165918825e-14, "grad_norm": 0.5249924659729004, "learning_rate": 0.00011331932521971355, "loss": 2.5503, "step": 17500 }, { "epoch": 6.075001612870778e-14, "grad_norm": 0.5698750019073486, "learning_rate": 0.00011330567496463903, "loss": 2.5118, "step": 17510 }, { "epoch": 6.078471059822732e-14, "grad_norm": 0.5553235411643982, "learning_rate": 0.00011329201766346806, "loss": 2.5287, "step": 17520 }, { "epoch": 6.081940506774686e-14, "grad_norm": 0.5540741086006165, "learning_rate": 0.00011327835331836039, "loss": 2.498, "step": 17530 }, { "epoch": 6.085409953726639e-14, "grad_norm": 0.5794776082038879, "learning_rate": 0.00011326468193147698, "loss": 2.5211, "step": 17540 }, { "epoch": 6.088879400678593e-14, "grad_norm": 0.540596067905426, "learning_rate": 0.00011325100350497977, "loss": 2.5562, "step": 17550 }, { "epoch": 6.092348847630547e-14, "grad_norm": 0.564656138420105, "learning_rate": 0.00011323731804103193, "loss": 2.5065, "step": 17560 }, { "epoch": 6.0958182945825e-14, "grad_norm": 0.5726520419120789, "learning_rate": 0.00011322362554179773, "loss": 2.5408, "step": 17570 }, { "epoch": 6.099287741534454e-14, "grad_norm": 0.5542612671852112, "learning_rate": 0.00011320992600944244, "loss": 2.536, "step": 17580 }, { "epoch": 6.102757188486407e-14, "grad_norm": 0.5465985536575317, "learning_rate": 0.00011319621944613259, "loss": 2.5224, "step": 17590 }, { "epoch": 6.106226635438361e-14, "grad_norm": 0.5468568801879883, "learning_rate": 0.0001131825058540357, "loss": 2.5303, "step": 17600 }, { "epoch": 6.109696082390315e-14, "grad_norm": 0.5542080402374268, "learning_rate": 0.00011316878523532052, "loss": 2.5355, "step": 17610 }, { "epoch": 6.113165529342268e-14, "grad_norm": 0.532297670841217, "learning_rate": 0.00011315505759215682, "loss": 2.4946, "step": 17620 }, { "epoch": 6.116634976294222e-14, "grad_norm": 0.5781600475311279, "learning_rate": 0.00011314132292671547, "loss": 2.5356, "step": 17630 }, { "epoch": 6.120104423246175e-14, "grad_norm": 0.5541283488273621, "learning_rate": 0.00011312758124116852, "loss": 2.5431, "step": 17640 }, { "epoch": 6.123573870198129e-14, "grad_norm": 0.5636920928955078, "learning_rate": 0.00011311383253768912, "loss": 2.4938, "step": 17650 }, { "epoch": 6.127043317150083e-14, "grad_norm": 0.5401487946510315, "learning_rate": 0.00011310007681845151, "loss": 2.5026, "step": 17660 }, { "epoch": 6.130512764102036e-14, "grad_norm": 0.54365074634552, "learning_rate": 0.00011308631408563099, "loss": 2.5201, "step": 17670 }, { "epoch": 6.13398221105399e-14, "grad_norm": 0.5401392579078674, "learning_rate": 0.00011307254434140404, "loss": 2.5119, "step": 17680 }, { "epoch": 6.137451658005944e-14, "grad_norm": 0.5396437048912048, "learning_rate": 0.00011305876758794827, "loss": 2.5154, "step": 17690 }, { "epoch": 6.140921104957897e-14, "grad_norm": 0.5574985146522522, "learning_rate": 0.0001130449838274423, "loss": 2.5144, "step": 17700 }, { "epoch": 6.144390551909851e-14, "grad_norm": 0.5133096575737, "learning_rate": 0.00011303119306206592, "loss": 2.4963, "step": 17710 }, { "epoch": 6.147859998861804e-14, "grad_norm": 0.5571476817131042, "learning_rate": 0.00011301739529400004, "loss": 2.5289, "step": 17720 }, { "epoch": 6.151329445813758e-14, "grad_norm": 0.5599967241287231, "learning_rate": 0.00011300359052542666, "loss": 2.5226, "step": 17730 }, { "epoch": 6.154798892765712e-14, "grad_norm": 0.539832592010498, "learning_rate": 0.00011298977875852887, "loss": 2.5338, "step": 17740 }, { "epoch": 6.158268339717665e-14, "grad_norm": 0.514394223690033, "learning_rate": 0.00011297595999549089, "loss": 2.5055, "step": 17750 }, { "epoch": 6.161737786669619e-14, "grad_norm": 0.5469964742660522, "learning_rate": 0.00011296213423849802, "loss": 2.5398, "step": 17760 }, { "epoch": 6.165207233621572e-14, "grad_norm": 0.5725789666175842, "learning_rate": 0.00011294830148973672, "loss": 2.4884, "step": 17770 }, { "epoch": 6.168676680573526e-14, "grad_norm": 0.5525121092796326, "learning_rate": 0.00011293446175139449, "loss": 2.5567, "step": 17780 }, { "epoch": 6.17214612752548e-14, "grad_norm": 0.5252744555473328, "learning_rate": 0.00011292061502565998, "loss": 2.5209, "step": 17790 }, { "epoch": 6.175615574477433e-14, "grad_norm": 0.5808729529380798, "learning_rate": 0.00011290676131472293, "loss": 2.5174, "step": 17800 }, { "epoch": 6.179085021429387e-14, "grad_norm": 0.53389573097229, "learning_rate": 0.00011289290062077417, "loss": 2.5011, "step": 17810 }, { "epoch": 6.18255446838134e-14, "grad_norm": 0.5689623355865479, "learning_rate": 0.00011287903294600568, "loss": 2.5187, "step": 17820 }, { "epoch": 6.186023915333294e-14, "grad_norm": 0.5292587280273438, "learning_rate": 0.00011286515829261049, "loss": 2.5191, "step": 17830 }, { "epoch": 6.189493362285248e-14, "grad_norm": 0.5460833311080933, "learning_rate": 0.00011285127666278277, "loss": 2.4941, "step": 17840 }, { "epoch": 6.192962809237201e-14, "grad_norm": 0.5159164071083069, "learning_rate": 0.00011283738805871773, "loss": 2.5197, "step": 17850 }, { "epoch": 6.196432256189155e-14, "grad_norm": 0.5569590330123901, "learning_rate": 0.00011282349248261182, "loss": 2.5003, "step": 17860 }, { "epoch": 6.199901703141109e-14, "grad_norm": 0.5681961178779602, "learning_rate": 0.00011280958993666245, "loss": 2.5009, "step": 17870 }, { "epoch": 6.203371150093062e-14, "grad_norm": 0.5337896943092346, "learning_rate": 0.00011279568042306818, "loss": 2.5434, "step": 17880 }, { "epoch": 6.206840597045016e-14, "grad_norm": 0.5115234851837158, "learning_rate": 0.00011278176394402873, "loss": 2.5202, "step": 17890 }, { "epoch": 6.21031004399697e-14, "grad_norm": 0.552435576915741, "learning_rate": 0.00011276784050174482, "loss": 2.4979, "step": 17900 }, { "epoch": 6.213779490948923e-14, "grad_norm": 0.5590205788612366, "learning_rate": 0.00011275391009841836, "loss": 2.4454, "step": 17910 }, { "epoch": 6.217248937900877e-14, "grad_norm": 0.5707004070281982, "learning_rate": 0.0001127399727362523, "loss": 2.5139, "step": 17920 }, { "epoch": 6.22071838485283e-14, "grad_norm": 0.5299373865127563, "learning_rate": 0.00011272602841745069, "loss": 2.4943, "step": 17930 }, { "epoch": 6.224187831804784e-14, "grad_norm": 0.6077837347984314, "learning_rate": 0.00011271207714421878, "loss": 2.5047, "step": 17940 }, { "epoch": 6.227657278756737e-14, "grad_norm": 0.5461963415145874, "learning_rate": 0.00011269811891876276, "loss": 2.5165, "step": 17950 }, { "epoch": 6.231126725708691e-14, "grad_norm": 0.5792241096496582, "learning_rate": 0.00011268415374329007, "loss": 2.5071, "step": 17960 }, { "epoch": 6.234596172660645e-14, "grad_norm": 0.5340027809143066, "learning_rate": 0.00011267018162000913, "loss": 2.4953, "step": 17970 }, { "epoch": 6.238065619612598e-14, "grad_norm": 0.5708136558532715, "learning_rate": 0.00011265620255112954, "loss": 2.4815, "step": 17980 }, { "epoch": 6.241535066564552e-14, "grad_norm": 0.5353647470474243, "learning_rate": 0.00011264221653886198, "loss": 2.4812, "step": 17990 }, { "epoch": 6.245004513516506e-14, "grad_norm": 0.5278773903846741, "learning_rate": 0.00011262822358541818, "loss": 2.4995, "step": 18000 }, { "epoch": 6.248473960468459e-14, "grad_norm": 0.5246478319168091, "learning_rate": 0.00011261422369301104, "loss": 2.5117, "step": 18010 }, { "epoch": 6.251943407420413e-14, "grad_norm": 0.553086519241333, "learning_rate": 0.00011260021686385451, "loss": 2.5382, "step": 18020 }, { "epoch": 6.255412854372366e-14, "grad_norm": 0.5432382822036743, "learning_rate": 0.00011258620310016365, "loss": 2.5312, "step": 18030 }, { "epoch": 6.25888230132432e-14, "grad_norm": 0.5756500959396362, "learning_rate": 0.0001125721824041546, "loss": 2.5297, "step": 18040 }, { "epoch": 6.262351748276274e-14, "grad_norm": 0.562449038028717, "learning_rate": 0.00011255815477804464, "loss": 2.505, "step": 18050 }, { "epoch": 6.265821195228227e-14, "grad_norm": 0.5464043021202087, "learning_rate": 0.0001125441202240521, "loss": 2.5196, "step": 18060 }, { "epoch": 6.269290642180181e-14, "grad_norm": 0.5284660458564758, "learning_rate": 0.00011253007874439643, "loss": 2.4985, "step": 18070 }, { "epoch": 6.272760089132134e-14, "grad_norm": 0.6191226243972778, "learning_rate": 0.00011251603034129815, "loss": 2.488, "step": 18080 }, { "epoch": 6.276229536084088e-14, "grad_norm": 0.5304848551750183, "learning_rate": 0.00011250197501697894, "loss": 2.4844, "step": 18090 }, { "epoch": 6.279698983036042e-14, "grad_norm": 0.5190476775169373, "learning_rate": 0.0001124879127736615, "loss": 2.5052, "step": 18100 }, { "epoch": 6.283168429987995e-14, "grad_norm": 0.5681871771812439, "learning_rate": 0.00011247384361356962, "loss": 2.513, "step": 18110 }, { "epoch": 6.286637876939949e-14, "grad_norm": 0.5361435413360596, "learning_rate": 0.00011245976753892826, "loss": 2.5194, "step": 18120 }, { "epoch": 6.290107323891903e-14, "grad_norm": 0.5663570761680603, "learning_rate": 0.00011244568455196345, "loss": 2.4737, "step": 18130 }, { "epoch": 6.293576770843856e-14, "grad_norm": 0.5340004563331604, "learning_rate": 0.00011243159465490225, "loss": 2.4973, "step": 18140 }, { "epoch": 6.29704621779581e-14, "grad_norm": 0.5454726815223694, "learning_rate": 0.00011241749784997285, "loss": 2.4724, "step": 18150 }, { "epoch": 6.300515664747763e-14, "grad_norm": 0.5248898267745972, "learning_rate": 0.00011240339413940457, "loss": 2.5047, "step": 18160 }, { "epoch": 6.303985111699717e-14, "grad_norm": 0.5378693342208862, "learning_rate": 0.00011238928352542778, "loss": 2.5095, "step": 18170 }, { "epoch": 6.30745455865167e-14, "grad_norm": 0.5470671057701111, "learning_rate": 0.00011237516601027395, "loss": 2.4683, "step": 18180 }, { "epoch": 6.310924005603624e-14, "grad_norm": 0.5215895175933838, "learning_rate": 0.00011236104159617565, "loss": 2.5024, "step": 18190 }, { "epoch": 6.314393452555578e-14, "grad_norm": 0.5249385833740234, "learning_rate": 0.00011234691028536652, "loss": 2.4743, "step": 18200 }, { "epoch": 6.317862899507531e-14, "grad_norm": 0.5559395551681519, "learning_rate": 0.00011233277208008131, "loss": 2.5033, "step": 18210 }, { "epoch": 6.321332346459485e-14, "grad_norm": 0.5396894812583923, "learning_rate": 0.00011231862698255587, "loss": 2.5192, "step": 18220 }, { "epoch": 6.324801793411439e-14, "grad_norm": 0.5971764922142029, "learning_rate": 0.00011230447499502709, "loss": 2.4925, "step": 18230 }, { "epoch": 6.328271240363392e-14, "grad_norm": 0.5469611883163452, "learning_rate": 0.00011229031611973304, "loss": 2.5099, "step": 18240 }, { "epoch": 6.331740687315346e-14, "grad_norm": 0.5433498024940491, "learning_rate": 0.00011227615035891277, "loss": 2.5113, "step": 18250 }, { "epoch": 6.3352101342673e-14, "grad_norm": 0.5283875465393066, "learning_rate": 0.00011226197771480652, "loss": 2.5381, "step": 18260 }, { "epoch": 6.338679581219253e-14, "grad_norm": 0.5410704016685486, "learning_rate": 0.00011224779818965554, "loss": 2.4975, "step": 18270 }, { "epoch": 6.342149028171207e-14, "grad_norm": 0.5834761261940002, "learning_rate": 0.0001122336117857022, "loss": 2.5452, "step": 18280 }, { "epoch": 6.34561847512316e-14, "grad_norm": 0.5384276509284973, "learning_rate": 0.00011221941850518997, "loss": 2.515, "step": 18290 }, { "epoch": 6.349087922075114e-14, "grad_norm": 0.561103343963623, "learning_rate": 0.00011220521835036337, "loss": 2.5136, "step": 18300 }, { "epoch": 6.352557369027068e-14, "grad_norm": 0.5413253307342529, "learning_rate": 0.00011219101132346808, "loss": 2.494, "step": 18310 }, { "epoch": 6.356026815979021e-14, "grad_norm": 0.5530669689178467, "learning_rate": 0.00011217679742675079, "loss": 2.4885, "step": 18320 }, { "epoch": 6.359496262930975e-14, "grad_norm": 0.5412546992301941, "learning_rate": 0.00011216257666245928, "loss": 2.5359, "step": 18330 }, { "epoch": 6.362965709882928e-14, "grad_norm": 0.6071393489837646, "learning_rate": 0.0001121483490328425, "loss": 2.4688, "step": 18340 }, { "epoch": 6.366435156834882e-14, "grad_norm": 0.5455372333526611, "learning_rate": 0.00011213411454015039, "loss": 2.4935, "step": 18350 }, { "epoch": 6.369904603786836e-14, "grad_norm": 0.524079442024231, "learning_rate": 0.00011211987318663402, "loss": 2.506, "step": 18360 }, { "epoch": 6.373374050738789e-14, "grad_norm": 0.5481063723564148, "learning_rate": 0.00011210562497454552, "loss": 2.5005, "step": 18370 }, { "epoch": 6.376843497690743e-14, "grad_norm": 0.525812566280365, "learning_rate": 0.00011209136990613814, "loss": 2.4656, "step": 18380 }, { "epoch": 6.380312944642696e-14, "grad_norm": 0.6106922626495361, "learning_rate": 0.0001120771079836662, "loss": 2.5113, "step": 18390 }, { "epoch": 6.38378239159465e-14, "grad_norm": 0.5507948994636536, "learning_rate": 0.0001120628392093851, "loss": 2.4898, "step": 18400 }, { "epoch": 6.387251838546604e-14, "grad_norm": 0.5459674596786499, "learning_rate": 0.00011204856358555132, "loss": 2.5273, "step": 18410 }, { "epoch": 6.390721285498557e-14, "grad_norm": 0.5210742354393005, "learning_rate": 0.00011203428111442244, "loss": 2.4894, "step": 18420 }, { "epoch": 6.394190732450511e-14, "grad_norm": 0.5709819197654724, "learning_rate": 0.00011201999179825706, "loss": 2.5141, "step": 18430 }, { "epoch": 6.397660179402465e-14, "grad_norm": 0.5245206356048584, "learning_rate": 0.00011200569563931497, "loss": 2.4794, "step": 18440 }, { "epoch": 6.401129626354418e-14, "grad_norm": 0.5363589525222778, "learning_rate": 0.00011199139263985694, "loss": 2.501, "step": 18450 }, { "epoch": 6.404599073306372e-14, "grad_norm": 0.5181222558021545, "learning_rate": 0.00011197708280214492, "loss": 2.5071, "step": 18460 }, { "epoch": 6.408068520258325e-14, "grad_norm": 0.5505267977714539, "learning_rate": 0.00011196276612844184, "loss": 2.5083, "step": 18470 }, { "epoch": 6.411537967210279e-14, "grad_norm": 0.5368124842643738, "learning_rate": 0.00011194844262101176, "loss": 2.499, "step": 18480 }, { "epoch": 6.415007414162233e-14, "grad_norm": 0.530672550201416, "learning_rate": 0.00011193411228211986, "loss": 2.5083, "step": 18490 }, { "epoch": 6.418476861114186e-14, "grad_norm": 0.5555778741836548, "learning_rate": 0.00011191977511403232, "loss": 2.4964, "step": 18500 }, { "epoch": 6.42194630806614e-14, "grad_norm": 0.5375620126724243, "learning_rate": 0.00011190543111901646, "loss": 2.4884, "step": 18510 }, { "epoch": 6.425415755018093e-14, "grad_norm": 0.5615679025650024, "learning_rate": 0.00011189108029934065, "loss": 2.5241, "step": 18520 }, { "epoch": 6.428885201970047e-14, "grad_norm": 0.5278034806251526, "learning_rate": 0.00011187672265727435, "loss": 2.468, "step": 18530 }, { "epoch": 6.432354648922001e-14, "grad_norm": 0.5734915733337402, "learning_rate": 0.00011186235819508811, "loss": 2.5092, "step": 18540 }, { "epoch": 6.435824095873954e-14, "grad_norm": 0.5346309542655945, "learning_rate": 0.00011184798691505352, "loss": 2.5124, "step": 18550 }, { "epoch": 6.439293542825908e-14, "grad_norm": 0.5273226499557495, "learning_rate": 0.0001118336088194433, "loss": 2.5091, "step": 18560 }, { "epoch": 6.442762989777862e-14, "grad_norm": 0.5459911823272705, "learning_rate": 0.0001118192239105312, "loss": 2.5094, "step": 18570 }, { "epoch": 6.446232436729815e-14, "grad_norm": 0.577604353427887, "learning_rate": 0.00011180483219059209, "loss": 2.5061, "step": 18580 }, { "epoch": 6.449701883681769e-14, "grad_norm": 0.5916627645492554, "learning_rate": 0.00011179043366190189, "loss": 2.5393, "step": 18590 }, { "epoch": 6.453171330633722e-14, "grad_norm": 0.5683515071868896, "learning_rate": 0.00011177602832673761, "loss": 2.5055, "step": 18600 }, { "epoch": 6.456640777585676e-14, "grad_norm": 0.5666828155517578, "learning_rate": 0.0001117616161873773, "loss": 2.4908, "step": 18610 }, { "epoch": 6.46011022453763e-14, "grad_norm": 0.5650627613067627, "learning_rate": 0.00011174719724610014, "loss": 2.512, "step": 18620 }, { "epoch": 6.463579671489583e-14, "grad_norm": 0.5429067015647888, "learning_rate": 0.00011173277150518638, "loss": 2.4947, "step": 18630 }, { "epoch": 6.467049118441537e-14, "grad_norm": 0.5361812114715576, "learning_rate": 0.00011171833896691728, "loss": 2.48, "step": 18640 }, { "epoch": 6.47051856539349e-14, "grad_norm": 0.5597516894340515, "learning_rate": 0.00011170389963357525, "loss": 2.5101, "step": 18650 }, { "epoch": 6.473988012345444e-14, "grad_norm": 0.5501545667648315, "learning_rate": 0.00011168945350744375, "loss": 2.5133, "step": 18660 }, { "epoch": 6.477457459297398e-14, "grad_norm": 1.2668966054916382, "learning_rate": 0.00011167500059080729, "loss": 2.5371, "step": 18670 }, { "epoch": 6.480926906249351e-14, "grad_norm": 0.5531436800956726, "learning_rate": 0.00011166054088595149, "loss": 2.5183, "step": 18680 }, { "epoch": 6.484396353201305e-14, "grad_norm": 0.5494400262832642, "learning_rate": 0.00011164607439516304, "loss": 2.4815, "step": 18690 }, { "epoch": 6.487865800153259e-14, "grad_norm": 0.5631104707717896, "learning_rate": 0.00011163160112072967, "loss": 2.4948, "step": 18700 }, { "epoch": 6.491335247105212e-14, "grad_norm": 0.5864558219909668, "learning_rate": 0.0001116171210649402, "loss": 2.4795, "step": 18710 }, { "epoch": 6.494804694057166e-14, "grad_norm": 0.5446715354919434, "learning_rate": 0.00011160263423008454, "loss": 2.5041, "step": 18720 }, { "epoch": 6.49827414100912e-14, "grad_norm": 0.5411696434020996, "learning_rate": 0.00011158814061845367, "loss": 2.4932, "step": 18730 }, { "epoch": 6.501743587961073e-14, "grad_norm": 0.5354647636413574, "learning_rate": 0.00011157364023233961, "loss": 2.5019, "step": 18740 }, { "epoch": 6.505213034913027e-14, "grad_norm": 0.5475752353668213, "learning_rate": 0.00011155913307403547, "loss": 2.4701, "step": 18750 }, { "epoch": 6.50868248186498e-14, "grad_norm": 0.5245879292488098, "learning_rate": 0.00011154461914583543, "loss": 2.5036, "step": 18760 }, { "epoch": 6.512151928816934e-14, "grad_norm": 0.5319998860359192, "learning_rate": 0.00011153009845003476, "loss": 2.466, "step": 18770 }, { "epoch": 6.515621375768887e-14, "grad_norm": 0.5333579182624817, "learning_rate": 0.00011151557098892978, "loss": 2.4575, "step": 18780 }, { "epoch": 6.519090822720841e-14, "grad_norm": 0.5491250157356262, "learning_rate": 0.00011150103676481786, "loss": 2.4974, "step": 18790 }, { "epoch": 6.522560269672795e-14, "grad_norm": 0.5404905676841736, "learning_rate": 0.00011148649577999748, "loss": 2.5054, "step": 18800 }, { "epoch": 6.526029716624748e-14, "grad_norm": 0.5411607027053833, "learning_rate": 0.00011147194803676819, "loss": 2.5062, "step": 18810 }, { "epoch": 6.529499163576702e-14, "grad_norm": 0.5467842817306519, "learning_rate": 0.00011145739353743055, "loss": 2.5221, "step": 18820 }, { "epoch": 6.532968610528656e-14, "grad_norm": 0.5387471914291382, "learning_rate": 0.00011144283228428626, "loss": 2.5111, "step": 18830 }, { "epoch": 6.536438057480609e-14, "grad_norm": 0.5706613659858704, "learning_rate": 0.00011142826427963805, "loss": 2.4713, "step": 18840 }, { "epoch": 6.539907504432563e-14, "grad_norm": 0.5445410013198853, "learning_rate": 0.00011141368952578972, "loss": 2.4864, "step": 18850 }, { "epoch": 6.543376951384516e-14, "grad_norm": 0.5372044444084167, "learning_rate": 0.00011139910802504612, "loss": 2.5043, "step": 18860 }, { "epoch": 6.54684639833647e-14, "grad_norm": 0.5836686491966248, "learning_rate": 0.00011138451977971325, "loss": 2.5023, "step": 18870 }, { "epoch": 6.550315845288424e-14, "grad_norm": 0.554682195186615, "learning_rate": 0.00011136992479209805, "loss": 2.4954, "step": 18880 }, { "epoch": 6.553785292240377e-14, "grad_norm": 0.544811487197876, "learning_rate": 0.00011135532306450862, "loss": 2.4749, "step": 18890 }, { "epoch": 6.557254739192331e-14, "grad_norm": 0.5202591419219971, "learning_rate": 0.00011134071459925413, "loss": 2.4601, "step": 18900 }, { "epoch": 6.560724186144284e-14, "grad_norm": 0.5454455018043518, "learning_rate": 0.00011132609939864471, "loss": 2.4302, "step": 18910 }, { "epoch": 6.564193633096238e-14, "grad_norm": 0.5810677409172058, "learning_rate": 0.00011131147746499169, "loss": 2.5297, "step": 18920 }, { "epoch": 6.567663080048192e-14, "grad_norm": 0.5487270951271057, "learning_rate": 0.00011129684880060738, "loss": 2.5142, "step": 18930 }, { "epoch": 6.571132527000145e-14, "grad_norm": 0.5296702980995178, "learning_rate": 0.00011128221340780516, "loss": 2.5453, "step": 18940 }, { "epoch": 6.574601973952099e-14, "grad_norm": 0.5315213799476624, "learning_rate": 0.00011126757128889953, "loss": 2.5238, "step": 18950 }, { "epoch": 6.578071420904053e-14, "grad_norm": 0.5195649266242981, "learning_rate": 0.00011125292244620598, "loss": 2.4793, "step": 18960 }, { "epoch": 6.581540867856006e-14, "grad_norm": 0.5336682200431824, "learning_rate": 0.0001112382668820411, "loss": 2.5088, "step": 18970 }, { "epoch": 6.58501031480796e-14, "grad_norm": 0.5446407198905945, "learning_rate": 0.0001112236045987226, "loss": 2.4833, "step": 18980 }, { "epoch": 6.588479761759913e-14, "grad_norm": 0.5301308631896973, "learning_rate": 0.0001112089355985691, "loss": 2.4778, "step": 18990 }, { "epoch": 6.591949208711867e-14, "grad_norm": 0.5434473752975464, "learning_rate": 0.00011119425988390044, "loss": 2.469, "step": 19000 }, { "epoch": 6.59541865566382e-14, "grad_norm": 0.5530282855033875, "learning_rate": 0.00011117957745703746, "loss": 2.4906, "step": 19010 }, { "epoch": 6.598888102615774e-14, "grad_norm": 0.5448588728904724, "learning_rate": 0.00011116488832030202, "loss": 2.4819, "step": 19020 }, { "epoch": 6.602357549567728e-14, "grad_norm": 0.5461463332176208, "learning_rate": 0.00011115019247601712, "loss": 2.4754, "step": 19030 }, { "epoch": 6.605826996519681e-14, "grad_norm": 0.5540622472763062, "learning_rate": 0.00011113548992650674, "loss": 2.4854, "step": 19040 }, { "epoch": 6.609296443471635e-14, "grad_norm": 0.5357370972633362, "learning_rate": 0.00011112078067409599, "loss": 2.514, "step": 19050 }, { "epoch": 6.612765890423589e-14, "grad_norm": 0.5229300856590271, "learning_rate": 0.00011110606472111103, "loss": 2.4871, "step": 19060 }, { "epoch": 6.616235337375542e-14, "grad_norm": 0.5163809061050415, "learning_rate": 0.000111091342069879, "loss": 2.5189, "step": 19070 }, { "epoch": 6.619704784327496e-14, "grad_norm": 0.5304562449455261, "learning_rate": 0.00011107661272272825, "loss": 2.4957, "step": 19080 }, { "epoch": 6.62317423127945e-14, "grad_norm": 0.5474101305007935, "learning_rate": 0.000111061876681988, "loss": 2.4763, "step": 19090 }, { "epoch": 6.626643678231403e-14, "grad_norm": 0.5419023633003235, "learning_rate": 0.0001110471339499887, "loss": 2.453, "step": 19100 }, { "epoch": 6.630113125183357e-14, "grad_norm": 0.557370126247406, "learning_rate": 0.00011103238452906175, "loss": 2.4729, "step": 19110 }, { "epoch": 6.63358257213531e-14, "grad_norm": 0.5596783757209778, "learning_rate": 0.00011101762842153966, "loss": 2.477, "step": 19120 }, { "epoch": 6.637052019087264e-14, "grad_norm": 0.5540986061096191, "learning_rate": 0.00011100286562975599, "loss": 2.4734, "step": 19130 }, { "epoch": 6.640521466039218e-14, "grad_norm": 0.575932502746582, "learning_rate": 0.00011098809615604533, "loss": 2.4749, "step": 19140 }, { "epoch": 6.643990912991171e-14, "grad_norm": 0.5654997825622559, "learning_rate": 0.00011097332000274333, "loss": 2.5014, "step": 19150 }, { "epoch": 6.647460359943125e-14, "grad_norm": 0.5501792430877686, "learning_rate": 0.00011095853717218675, "loss": 2.5001, "step": 19160 }, { "epoch": 6.650929806895078e-14, "grad_norm": 0.564298152923584, "learning_rate": 0.00011094374766671334, "loss": 2.4911, "step": 19170 }, { "epoch": 6.654399253847032e-14, "grad_norm": 0.5425021052360535, "learning_rate": 0.00011092895148866195, "loss": 2.5014, "step": 19180 }, { "epoch": 6.657868700798986e-14, "grad_norm": 0.5221104025840759, "learning_rate": 0.00011091414864037249, "loss": 2.4677, "step": 19190 }, { "epoch": 6.661338147750939e-14, "grad_norm": 0.5310280323028564, "learning_rate": 0.00011089933912418583, "loss": 2.5021, "step": 19200 }, { "epoch": 6.664807594702893e-14, "grad_norm": 0.5758768320083618, "learning_rate": 0.00011088452294244403, "loss": 2.4468, "step": 19210 }, { "epoch": 6.668277041654846e-14, "grad_norm": 0.5560582876205444, "learning_rate": 0.00011086970009749011, "loss": 2.489, "step": 19220 }, { "epoch": 6.6717464886068e-14, "grad_norm": 0.5486220121383667, "learning_rate": 0.00011085487059166821, "loss": 2.5032, "step": 19230 }, { "epoch": 6.675215935558754e-14, "grad_norm": 0.5555339455604553, "learning_rate": 0.00011084003442732346, "loss": 2.4786, "step": 19240 }, { "epoch": 6.678685382510707e-14, "grad_norm": 0.5304591059684753, "learning_rate": 0.00011082519160680206, "loss": 2.4809, "step": 19250 }, { "epoch": 6.682154829462661e-14, "grad_norm": 0.5313590168952942, "learning_rate": 0.00011081034213245132, "loss": 2.4738, "step": 19260 }, { "epoch": 6.685624276414615e-14, "grad_norm": 0.5461806058883667, "learning_rate": 0.00011079548600661951, "loss": 2.4773, "step": 19270 }, { "epoch": 6.689093723366568e-14, "grad_norm": 0.5252965092658997, "learning_rate": 0.00011078062323165603, "loss": 2.451, "step": 19280 }, { "epoch": 6.692563170318522e-14, "grad_norm": 0.5292608141899109, "learning_rate": 0.00011076575380991129, "loss": 2.4549, "step": 19290 }, { "epoch": 6.696032617270475e-14, "grad_norm": 0.5467344522476196, "learning_rate": 0.00011075087774373674, "loss": 2.4968, "step": 19300 }, { "epoch": 6.699502064222429e-14, "grad_norm": 0.5536628365516663, "learning_rate": 0.00011073599503548495, "loss": 2.4316, "step": 19310 }, { "epoch": 6.702971511174383e-14, "grad_norm": 0.5083438158035278, "learning_rate": 0.00011072110568750946, "loss": 2.4758, "step": 19320 }, { "epoch": 6.706440958126336e-14, "grad_norm": 0.5538945198059082, "learning_rate": 0.00011070620970216487, "loss": 2.4839, "step": 19330 }, { "epoch": 6.70991040507829e-14, "grad_norm": 0.5428655743598938, "learning_rate": 0.0001106913070818069, "loss": 2.5022, "step": 19340 }, { "epoch": 6.713379852030243e-14, "grad_norm": 0.5342002511024475, "learning_rate": 0.00011067639782879223, "loss": 2.4694, "step": 19350 }, { "epoch": 6.716849298982197e-14, "grad_norm": 0.5520106554031372, "learning_rate": 0.00011066148194547863, "loss": 2.5356, "step": 19360 }, { "epoch": 6.720318745934151e-14, "grad_norm": 0.5650825500488281, "learning_rate": 0.00011064655943422499, "loss": 2.4987, "step": 19370 }, { "epoch": 6.723788192886104e-14, "grad_norm": 0.5288682579994202, "learning_rate": 0.00011063163029739108, "loss": 2.5146, "step": 19380 }, { "epoch": 6.727257639838058e-14, "grad_norm": 0.5569751858711243, "learning_rate": 0.00011061669453733789, "loss": 2.4942, "step": 19390 }, { "epoch": 6.730727086790012e-14, "grad_norm": 0.5461465716362, "learning_rate": 0.00011060175215642733, "loss": 2.5302, "step": 19400 }, { "epoch": 6.734196533741965e-14, "grad_norm": 0.5334799885749817, "learning_rate": 0.00011058680315702243, "loss": 2.482, "step": 19410 }, { "epoch": 6.737665980693919e-14, "grad_norm": 0.5749849677085876, "learning_rate": 0.00011057184754148725, "loss": 2.4968, "step": 19420 }, { "epoch": 6.741135427645872e-14, "grad_norm": 0.5719369649887085, "learning_rate": 0.00011055688531218689, "loss": 2.4929, "step": 19430 }, { "epoch": 6.744604874597826e-14, "grad_norm": 0.5526955723762512, "learning_rate": 0.00011054191647148748, "loss": 2.4793, "step": 19440 }, { "epoch": 6.74807432154978e-14, "grad_norm": 0.5557788014411926, "learning_rate": 0.00011052694102175622, "loss": 2.4939, "step": 19450 }, { "epoch": 6.751543768501733e-14, "grad_norm": 0.5536144375801086, "learning_rate": 0.00011051195896536137, "loss": 2.4756, "step": 19460 }, { "epoch": 6.755013215453687e-14, "grad_norm": 0.5338996052742004, "learning_rate": 0.00011049697030467222, "loss": 2.4657, "step": 19470 }, { "epoch": 6.75848266240564e-14, "grad_norm": 0.5396523475646973, "learning_rate": 0.00011048197504205904, "loss": 2.5086, "step": 19480 }, { "epoch": 6.761952109357594e-14, "grad_norm": 0.5463858246803284, "learning_rate": 0.00011046697317989325, "loss": 2.5004, "step": 19490 }, { "epoch": 6.765421556309548e-14, "grad_norm": 0.5425831079483032, "learning_rate": 0.00011045196472054729, "loss": 2.506, "step": 19500 }, { "epoch": 6.768891003261501e-14, "grad_norm": 0.5444839000701904, "learning_rate": 0.00011043694966639453, "loss": 2.4879, "step": 19510 }, { "epoch": 6.772360450213455e-14, "grad_norm": 0.5455523729324341, "learning_rate": 0.00011042192801980955, "loss": 2.4886, "step": 19520 }, { "epoch": 6.775829897165409e-14, "grad_norm": 0.5731580853462219, "learning_rate": 0.00011040689978316787, "loss": 2.4538, "step": 19530 }, { "epoch": 6.779299344117362e-14, "grad_norm": 0.5895419716835022, "learning_rate": 0.00011039186495884606, "loss": 2.4938, "step": 19540 }, { "epoch": 6.782768791069316e-14, "grad_norm": 0.5637843608856201, "learning_rate": 0.00011037682354922177, "loss": 2.561, "step": 19550 }, { "epoch": 6.786238238021269e-14, "grad_norm": 0.5516568422317505, "learning_rate": 0.00011036177555667366, "loss": 2.4983, "step": 19560 }, { "epoch": 6.789707684973223e-14, "grad_norm": 0.5442430973052979, "learning_rate": 0.00011034672098358145, "loss": 2.4769, "step": 19570 }, { "epoch": 6.793177131925177e-14, "grad_norm": 0.5120355486869812, "learning_rate": 0.00011033165983232587, "loss": 2.4709, "step": 19580 }, { "epoch": 6.79664657887713e-14, "grad_norm": 0.5566269755363464, "learning_rate": 0.00011031659210528874, "loss": 2.4626, "step": 19590 }, { "epoch": 6.800116025829084e-14, "grad_norm": 0.5734180212020874, "learning_rate": 0.00011030151780485287, "loss": 2.4687, "step": 19600 }, { "epoch": 6.803585472781037e-14, "grad_norm": 0.5828995704650879, "learning_rate": 0.00011028643693340213, "loss": 2.4787, "step": 19610 }, { "epoch": 6.807054919732991e-14, "grad_norm": 0.5549116730690002, "learning_rate": 0.00011027134949332145, "loss": 2.4627, "step": 19620 }, { "epoch": 6.810524366684945e-14, "grad_norm": 0.5761750340461731, "learning_rate": 0.00011025625548699675, "loss": 2.4717, "step": 19630 }, { "epoch": 6.813993813636898e-14, "grad_norm": 0.5413475632667542, "learning_rate": 0.00011024115491681506, "loss": 2.4988, "step": 19640 }, { "epoch": 6.817463260588852e-14, "grad_norm": 0.5273314118385315, "learning_rate": 0.00011022604778516435, "loss": 2.5182, "step": 19650 }, { "epoch": 6.820932707540805e-14, "grad_norm": 0.536911129951477, "learning_rate": 0.00011021093409443372, "loss": 2.5071, "step": 19660 }, { "epoch": 6.824402154492759e-14, "grad_norm": 0.5291702747344971, "learning_rate": 0.00011019581384701327, "loss": 2.5102, "step": 19670 }, { "epoch": 6.827871601444713e-14, "grad_norm": 0.5709646940231323, "learning_rate": 0.00011018068704529412, "loss": 2.5013, "step": 19680 }, { "epoch": 6.831341048396666e-14, "grad_norm": 0.5637699961662292, "learning_rate": 0.00011016555369166844, "loss": 2.4709, "step": 19690 }, { "epoch": 6.83481049534862e-14, "grad_norm": 0.5360698103904724, "learning_rate": 0.00011015041378852948, "loss": 2.4377, "step": 19700 }, { "epoch": 6.838279942300574e-14, "grad_norm": 0.5258808135986328, "learning_rate": 0.00011013526733827144, "loss": 2.4885, "step": 19710 }, { "epoch": 6.841749389252527e-14, "grad_norm": 0.5394079089164734, "learning_rate": 0.0001101201143432896, "loss": 2.4851, "step": 19720 }, { "epoch": 6.845218836204481e-14, "grad_norm": 0.5656520128250122, "learning_rate": 0.00011010495480598032, "loss": 2.4889, "step": 19730 }, { "epoch": 6.848688283156434e-14, "grad_norm": 0.5421004295349121, "learning_rate": 0.00011008978872874091, "loss": 2.4876, "step": 19740 }, { "epoch": 6.852157730108388e-14, "grad_norm": 0.5397576093673706, "learning_rate": 0.00011007461611396978, "loss": 2.4793, "step": 19750 }, { "epoch": 6.855627177060342e-14, "grad_norm": 0.5464798808097839, "learning_rate": 0.00011005943696406633, "loss": 2.4922, "step": 19760 }, { "epoch": 6.859096624012295e-14, "grad_norm": 0.5836300253868103, "learning_rate": 0.000110044251281431, "loss": 2.4636, "step": 19770 }, { "epoch": 6.862566070964249e-14, "grad_norm": 0.5341587066650391, "learning_rate": 0.00011002905906846532, "loss": 2.4767, "step": 19780 }, { "epoch": 6.866035517916202e-14, "grad_norm": 0.5632769465446472, "learning_rate": 0.00011001386032757176, "loss": 2.4592, "step": 19790 }, { "epoch": 6.869504964868156e-14, "grad_norm": 0.5714156031608582, "learning_rate": 0.00010999865506115392, "loss": 2.4672, "step": 19800 }, { "epoch": 6.87297441182011e-14, "grad_norm": 0.5642797350883484, "learning_rate": 0.00010998344327161632, "loss": 2.4799, "step": 19810 }, { "epoch": 6.876443858772063e-14, "grad_norm": 0.5402487516403198, "learning_rate": 0.00010996822496136462, "loss": 2.5318, "step": 19820 }, { "epoch": 6.879913305724017e-14, "grad_norm": 0.5305373668670654, "learning_rate": 0.00010995300013280546, "loss": 2.4785, "step": 19830 }, { "epoch": 6.88338275267597e-14, "grad_norm": 0.5176288485527039, "learning_rate": 0.0001099377687883465, "loss": 2.4586, "step": 19840 }, { "epoch": 6.886852199627924e-14, "grad_norm": 0.5459149479866028, "learning_rate": 0.00010992253093039644, "loss": 2.4392, "step": 19850 }, { "epoch": 6.890321646579878e-14, "grad_norm": 0.546907365322113, "learning_rate": 0.00010990728656136504, "loss": 2.4827, "step": 19860 }, { "epoch": 6.893791093531831e-14, "grad_norm": 0.5168582797050476, "learning_rate": 0.00010989203568366305, "loss": 2.4812, "step": 19870 }, { "epoch": 6.897260540483785e-14, "grad_norm": 0.5224096179008484, "learning_rate": 0.00010987677829970228, "loss": 2.4753, "step": 19880 }, { "epoch": 6.900729987435739e-14, "grad_norm": 0.5758994221687317, "learning_rate": 0.00010986151441189555, "loss": 2.4946, "step": 19890 }, { "epoch": 6.904199434387692e-14, "grad_norm": 0.5573430061340332, "learning_rate": 0.0001098462440226567, "loss": 2.4887, "step": 19900 }, { "epoch": 6.907668881339646e-14, "grad_norm": 0.5479156970977783, "learning_rate": 0.00010983096713440061, "loss": 2.4824, "step": 19910 }, { "epoch": 6.9111383282916e-14, "grad_norm": 0.5739791393280029, "learning_rate": 0.00010981568374954318, "loss": 2.4947, "step": 19920 }, { "epoch": 6.914607775243553e-14, "grad_norm": 0.5480680465698242, "learning_rate": 0.00010980039387050137, "loss": 2.4661, "step": 19930 }, { "epoch": 6.918077222195507e-14, "grad_norm": 0.567875862121582, "learning_rate": 0.00010978509749969317, "loss": 2.5152, "step": 19940 }, { "epoch": 6.92154666914746e-14, "grad_norm": 0.5395058989524841, "learning_rate": 0.00010976979463953748, "loss": 2.469, "step": 19950 }, { "epoch": 6.925016116099414e-14, "grad_norm": 0.5594972968101501, "learning_rate": 0.0001097544852924544, "loss": 2.4874, "step": 19960 }, { "epoch": 6.928485563051368e-14, "grad_norm": 0.5452079176902771, "learning_rate": 0.00010973916946086492, "loss": 2.488, "step": 19970 }, { "epoch": 6.931955010003321e-14, "grad_norm": 0.5320824384689331, "learning_rate": 0.00010972384714719116, "loss": 2.5008, "step": 19980 }, { "epoch": 6.935424456955275e-14, "grad_norm": 0.5120947360992432, "learning_rate": 0.00010970851835385616, "loss": 2.5163, "step": 19990 }, { "epoch": 6.938893903907228e-14, "grad_norm": 0.53548663854599, "learning_rate": 0.00010969318308328404, "loss": 2.4664, "step": 20000 }, { "epoch": 6.942363350859182e-14, "grad_norm": 0.546826183795929, "learning_rate": 0.00010967784133789998, "loss": 2.5116, "step": 20010 }, { "epoch": 6.945832797811136e-14, "grad_norm": 0.557449996471405, "learning_rate": 0.0001096624931201301, "loss": 2.4862, "step": 20020 }, { "epoch": 6.949302244763089e-14, "grad_norm": 0.5504398345947266, "learning_rate": 0.00010964713843240164, "loss": 2.4936, "step": 20030 }, { "epoch": 6.952771691715043e-14, "grad_norm": 0.5999268889427185, "learning_rate": 0.00010963177727714275, "loss": 2.488, "step": 20040 }, { "epoch": 6.956241138666996e-14, "grad_norm": 0.5488281846046448, "learning_rate": 0.00010961640965678273, "loss": 2.4918, "step": 20050 }, { "epoch": 6.95971058561895e-14, "grad_norm": 0.5361968874931335, "learning_rate": 0.00010960103557375177, "loss": 2.4716, "step": 20060 }, { "epoch": 6.963180032570904e-14, "grad_norm": 0.6802313327789307, "learning_rate": 0.0001095856550304812, "loss": 2.5051, "step": 20070 }, { "epoch": 6.966649479522857e-14, "grad_norm": 0.5496341586112976, "learning_rate": 0.0001095702680294033, "loss": 2.4779, "step": 20080 }, { "epoch": 6.970118926474811e-14, "grad_norm": 0.5313570499420166, "learning_rate": 0.00010955487457295139, "loss": 2.4812, "step": 20090 }, { "epoch": 6.973588373426765e-14, "grad_norm": 0.547789990901947, "learning_rate": 0.00010953947466355982, "loss": 2.5083, "step": 20100 }, { "epoch": 6.977057820378718e-14, "grad_norm": 0.5429452061653137, "learning_rate": 0.00010952406830366396, "loss": 2.4392, "step": 20110 }, { "epoch": 6.980527267330672e-14, "grad_norm": 0.558617353439331, "learning_rate": 0.00010950865549570017, "loss": 2.4996, "step": 20120 }, { "epoch": 6.983996714282625e-14, "grad_norm": 0.5643292665481567, "learning_rate": 0.00010949323624210586, "loss": 2.4741, "step": 20130 }, { "epoch": 6.987466161234579e-14, "grad_norm": 0.5443257093429565, "learning_rate": 0.00010947781054531948, "loss": 2.4936, "step": 20140 }, { "epoch": 6.990935608186533e-14, "grad_norm": 0.5446560978889465, "learning_rate": 0.00010946237840778044, "loss": 2.497, "step": 20150 }, { "epoch": 6.994405055138486e-14, "grad_norm": 0.5450677275657654, "learning_rate": 0.0001094469398319292, "loss": 2.5303, "step": 20160 }, { "epoch": 6.99787450209044e-14, "grad_norm": 0.5499958395957947, "learning_rate": 0.00010943149482020726, "loss": 2.4798, "step": 20170 }, { "epoch": 7.001343949042393e-14, "grad_norm": 0.5409227013587952, "learning_rate": 0.0001094160433750571, "loss": 2.5096, "step": 20180 }, { "epoch": 7.004813395994347e-14, "grad_norm": 0.5505097508430481, "learning_rate": 0.00010940058549892223, "loss": 2.4544, "step": 20190 }, { "epoch": 7.008282842946301e-14, "grad_norm": 0.5452449321746826, "learning_rate": 0.0001093851211942472, "loss": 2.4832, "step": 20200 }, { "epoch": 7.011752289898254e-14, "grad_norm": 0.5642987489700317, "learning_rate": 0.00010936965046347752, "loss": 2.4709, "step": 20210 }, { "epoch": 7.015221736850208e-14, "grad_norm": 0.5106102824211121, "learning_rate": 0.0001093541733090598, "loss": 2.4651, "step": 20220 }, { "epoch": 7.018691183802162e-14, "grad_norm": 0.5501532554626465, "learning_rate": 0.00010933868973344157, "loss": 2.4642, "step": 20230 }, { "epoch": 7.022160630754115e-14, "grad_norm": 0.5147389769554138, "learning_rate": 0.00010932319973907147, "loss": 2.4591, "step": 20240 }, { "epoch": 7.025630077706069e-14, "grad_norm": 0.5887203216552734, "learning_rate": 0.00010930770332839908, "loss": 2.462, "step": 20250 }, { "epoch": 7.029099524658022e-14, "grad_norm": 0.5659809112548828, "learning_rate": 0.00010929220050387501, "loss": 2.5018, "step": 20260 }, { "epoch": 7.032568971609976e-14, "grad_norm": 0.5285807847976685, "learning_rate": 0.00010927669126795096, "loss": 2.4689, "step": 20270 }, { "epoch": 7.03603841856193e-14, "grad_norm": 0.5422607660293579, "learning_rate": 0.00010926117562307951, "loss": 2.4791, "step": 20280 }, { "epoch": 7.039507865513883e-14, "grad_norm": 0.5254136323928833, "learning_rate": 0.00010924565357171438, "loss": 2.4989, "step": 20290 }, { "epoch": 7.042977312465837e-14, "grad_norm": 0.5661361217498779, "learning_rate": 0.00010923012511631023, "loss": 2.4603, "step": 20300 }, { "epoch": 7.04644675941779e-14, "grad_norm": 0.563640832901001, "learning_rate": 0.00010921459025932272, "loss": 2.4854, "step": 20310 }, { "epoch": 7.049916206369744e-14, "grad_norm": 0.5855013728141785, "learning_rate": 0.00010919904900320859, "loss": 2.4424, "step": 20320 }, { "epoch": 7.053385653321698e-14, "grad_norm": 0.5631839036941528, "learning_rate": 0.00010918350135042557, "loss": 2.4503, "step": 20330 }, { "epoch": 7.056855100273651e-14, "grad_norm": 0.5787818431854248, "learning_rate": 0.00010916794730343233, "loss": 2.494, "step": 20340 }, { "epoch": 7.060324547225605e-14, "grad_norm": 0.5661528706550598, "learning_rate": 0.00010915238686468867, "loss": 2.4434, "step": 20350 }, { "epoch": 7.063793994177558e-14, "grad_norm": 0.53556889295578, "learning_rate": 0.00010913682003665531, "loss": 2.4807, "step": 20360 }, { "epoch": 7.067263441129512e-14, "grad_norm": 0.5245079398155212, "learning_rate": 0.00010912124682179402, "loss": 2.4819, "step": 20370 }, { "epoch": 7.070732888081466e-14, "grad_norm": 0.5670454502105713, "learning_rate": 0.00010910566722256755, "loss": 2.5003, "step": 20380 }, { "epoch": 7.074202335033419e-14, "grad_norm": 0.5716837644577026, "learning_rate": 0.0001090900812414397, "loss": 2.4842, "step": 20390 }, { "epoch": 7.077671781985373e-14, "grad_norm": 0.5194280743598938, "learning_rate": 0.00010907448888087524, "loss": 2.457, "step": 20400 }, { "epoch": 7.081141228937327e-14, "grad_norm": 0.5540274977684021, "learning_rate": 0.00010905889014334, "loss": 2.4766, "step": 20410 }, { "epoch": 7.08461067588928e-14, "grad_norm": 0.5792196989059448, "learning_rate": 0.00010904328503130075, "loss": 2.4756, "step": 20420 }, { "epoch": 7.088080122841234e-14, "grad_norm": 0.5349225997924805, "learning_rate": 0.00010902767354722532, "loss": 2.4893, "step": 20430 }, { "epoch": 7.091549569793187e-14, "grad_norm": 0.5710325241088867, "learning_rate": 0.00010901205569358255, "loss": 2.4845, "step": 20440 }, { "epoch": 7.095019016745141e-14, "grad_norm": 0.5320618748664856, "learning_rate": 0.00010899643147284223, "loss": 2.5036, "step": 20450 }, { "epoch": 7.098488463697095e-14, "grad_norm": 0.5902583599090576, "learning_rate": 0.00010898080088747521, "loss": 2.4874, "step": 20460 }, { "epoch": 7.101957910649048e-14, "grad_norm": 0.5565060377120972, "learning_rate": 0.00010896516393995335, "loss": 2.4673, "step": 20470 }, { "epoch": 7.105427357601002e-14, "grad_norm": 0.5409265160560608, "learning_rate": 0.0001089495206327495, "loss": 2.4619, "step": 20480 }, { "epoch": 7.108896804552955e-14, "grad_norm": 0.5623031258583069, "learning_rate": 0.00010893387096833749, "loss": 2.5107, "step": 20490 }, { "epoch": 7.112366251504909e-14, "grad_norm": 0.5879287123680115, "learning_rate": 0.00010891821494919218, "loss": 2.4345, "step": 20500 }, { "epoch": 7.115835698456863e-14, "grad_norm": 0.603547990322113, "learning_rate": 0.00010890255257778946, "loss": 2.4859, "step": 20510 }, { "epoch": 7.119305145408816e-14, "grad_norm": 0.5388978123664856, "learning_rate": 0.00010888688385660618, "loss": 2.4804, "step": 20520 }, { "epoch": 7.12277459236077e-14, "grad_norm": 0.5459191203117371, "learning_rate": 0.00010887120878812022, "loss": 2.5145, "step": 20530 }, { "epoch": 7.126244039312724e-14, "grad_norm": 0.5158421397209167, "learning_rate": 0.00010885552737481045, "loss": 2.47, "step": 20540 }, { "epoch": 7.129713486264677e-14, "grad_norm": 0.5249080061912537, "learning_rate": 0.00010883983961915676, "loss": 2.476, "step": 20550 }, { "epoch": 7.133182933216631e-14, "grad_norm": 0.5484512448310852, "learning_rate": 0.00010882414552364003, "loss": 2.4701, "step": 20560 }, { "epoch": 7.136652380168584e-14, "grad_norm": 0.5599285364151001, "learning_rate": 0.00010880844509074215, "loss": 2.4748, "step": 20570 }, { "epoch": 7.140121827120538e-14, "grad_norm": 0.5244728326797485, "learning_rate": 0.00010879273832294599, "loss": 2.4794, "step": 20580 }, { "epoch": 7.143591274072492e-14, "grad_norm": 0.5397453904151917, "learning_rate": 0.00010877702522273549, "loss": 2.4564, "step": 20590 }, { "epoch": 7.147060721024445e-14, "grad_norm": 0.5273917317390442, "learning_rate": 0.00010876130579259548, "loss": 2.5308, "step": 20600 }, { "epoch": 7.150530167976399e-14, "grad_norm": 0.5538013577461243, "learning_rate": 0.00010874558003501187, "loss": 2.5147, "step": 20610 }, { "epoch": 7.153999614928352e-14, "grad_norm": 0.5621039867401123, "learning_rate": 0.00010872984795247158, "loss": 2.4647, "step": 20620 }, { "epoch": 7.157469061880306e-14, "grad_norm": 0.5375024676322937, "learning_rate": 0.00010871410954746247, "loss": 2.4581, "step": 20630 }, { "epoch": 7.16093850883226e-14, "grad_norm": 0.5447106957435608, "learning_rate": 0.00010869836482247345, "loss": 2.5021, "step": 20640 }, { "epoch": 7.164407955784213e-14, "grad_norm": 0.5415632724761963, "learning_rate": 0.00010868261377999443, "loss": 2.4638, "step": 20650 }, { "epoch": 7.167877402736167e-14, "grad_norm": 0.5276161432266235, "learning_rate": 0.00010866685642251627, "loss": 2.4978, "step": 20660 }, { "epoch": 7.17134684968812e-14, "grad_norm": 0.5449094176292419, "learning_rate": 0.00010865109275253088, "loss": 2.4937, "step": 20670 }, { "epoch": 7.174816296640074e-14, "grad_norm": 0.5871606469154358, "learning_rate": 0.00010863532277253113, "loss": 2.4642, "step": 20680 }, { "epoch": 7.178285743592028e-14, "grad_norm": 0.5447830557823181, "learning_rate": 0.00010861954648501093, "loss": 2.493, "step": 20690 }, { "epoch": 7.181755190543981e-14, "grad_norm": 0.5734698176383972, "learning_rate": 0.00010860376389246516, "loss": 2.4721, "step": 20700 }, { "epoch": 7.185224637495935e-14, "grad_norm": 0.5447969436645508, "learning_rate": 0.00010858797499738968, "loss": 2.5107, "step": 20710 }, { "epoch": 7.188694084447889e-14, "grad_norm": 0.5451892614364624, "learning_rate": 0.00010857217980228138, "loss": 2.5066, "step": 20720 }, { "epoch": 7.192163531399842e-14, "grad_norm": 0.5331855416297913, "learning_rate": 0.00010855637830963813, "loss": 2.4808, "step": 20730 }, { "epoch": 7.195632978351796e-14, "grad_norm": 0.574570894241333, "learning_rate": 0.00010854057052195882, "loss": 2.5035, "step": 20740 }, { "epoch": 7.19910242530375e-14, "grad_norm": 0.5657070279121399, "learning_rate": 0.0001085247564417433, "loss": 2.4648, "step": 20750 }, { "epoch": 7.202571872255703e-14, "grad_norm": 0.5390982031822205, "learning_rate": 0.00010850893607149244, "loss": 2.4821, "step": 20760 }, { "epoch": 7.206041319207657e-14, "grad_norm": 0.5670458078384399, "learning_rate": 0.00010849310941370808, "loss": 2.4853, "step": 20770 }, { "epoch": 7.20951076615961e-14, "grad_norm": 0.5551409721374512, "learning_rate": 0.00010847727647089307, "loss": 2.4959, "step": 20780 }, { "epoch": 7.212980213111564e-14, "grad_norm": 0.5481020212173462, "learning_rate": 0.00010846143724555125, "loss": 2.4622, "step": 20790 }, { "epoch": 7.216449660063518e-14, "grad_norm": 0.5483428239822388, "learning_rate": 0.00010844559174018749, "loss": 2.5006, "step": 20800 }, { "epoch": 7.219919107015471e-14, "grad_norm": 0.5714631676673889, "learning_rate": 0.0001084297399573076, "loss": 2.4867, "step": 20810 }, { "epoch": 7.223388553967425e-14, "grad_norm": 0.5375357866287231, "learning_rate": 0.0001084138818994184, "loss": 2.4707, "step": 20820 }, { "epoch": 7.226858000919378e-14, "grad_norm": 0.5199787020683289, "learning_rate": 0.00010839801756902768, "loss": 2.4935, "step": 20830 }, { "epoch": 7.230327447871332e-14, "grad_norm": 0.5619285702705383, "learning_rate": 0.0001083821469686443, "loss": 2.4928, "step": 20840 }, { "epoch": 7.233796894823286e-14, "grad_norm": 0.5524587035179138, "learning_rate": 0.00010836627010077805, "loss": 2.4907, "step": 20850 }, { "epoch": 7.237266341775239e-14, "grad_norm": 0.5499525666236877, "learning_rate": 0.00010835038696793966, "loss": 2.4805, "step": 20860 }, { "epoch": 7.240735788727193e-14, "grad_norm": 0.5761500597000122, "learning_rate": 0.00010833449757264096, "loss": 2.4699, "step": 20870 }, { "epoch": 7.244205235679146e-14, "grad_norm": 0.5719471573829651, "learning_rate": 0.00010831860191739474, "loss": 2.4693, "step": 20880 }, { "epoch": 7.2476746826311e-14, "grad_norm": 0.5548362731933594, "learning_rate": 0.00010830270000471472, "loss": 2.473, "step": 20890 }, { "epoch": 7.251144129583054e-14, "grad_norm": 0.525255560874939, "learning_rate": 0.00010828679183711567, "loss": 2.4697, "step": 20900 }, { "epoch": 7.254613576535007e-14, "grad_norm": 0.6528435349464417, "learning_rate": 0.00010827087741711334, "loss": 2.4471, "step": 20910 }, { "epoch": 7.258083023486961e-14, "grad_norm": 0.5734004378318787, "learning_rate": 0.00010825495674722442, "loss": 2.4868, "step": 20920 }, { "epoch": 7.261552470438914e-14, "grad_norm": 0.5519812703132629, "learning_rate": 0.00010823902982996665, "loss": 2.4944, "step": 20930 }, { "epoch": 7.265021917390868e-14, "grad_norm": 0.5498120188713074, "learning_rate": 0.00010822309666785875, "loss": 2.4735, "step": 20940 }, { "epoch": 7.268491364342822e-14, "grad_norm": 0.5658668279647827, "learning_rate": 0.0001082071572634204, "loss": 2.4691, "step": 20950 }, { "epoch": 7.271960811294775e-14, "grad_norm": 0.5333907008171082, "learning_rate": 0.00010819121161917228, "loss": 2.4822, "step": 20960 }, { "epoch": 7.275430258246729e-14, "grad_norm": 0.5568821430206299, "learning_rate": 0.00010817525973763604, "loss": 2.5228, "step": 20970 }, { "epoch": 7.278899705198683e-14, "grad_norm": 0.5381320714950562, "learning_rate": 0.00010815930162133437, "loss": 2.4583, "step": 20980 }, { "epoch": 7.282369152150636e-14, "grad_norm": 0.532949686050415, "learning_rate": 0.00010814333727279087, "loss": 2.4483, "step": 20990 }, { "epoch": 7.28583859910259e-14, "grad_norm": 0.5358375906944275, "learning_rate": 0.00010812736669453017, "loss": 2.5063, "step": 21000 }, { "epoch": 7.289308046054543e-14, "grad_norm": 0.5486124753952026, "learning_rate": 0.00010811138988907791, "loss": 2.4454, "step": 21010 }, { "epoch": 7.292777493006497e-14, "grad_norm": 0.5352567434310913, "learning_rate": 0.00010809540685896065, "loss": 2.4321, "step": 21020 }, { "epoch": 7.296246939958451e-14, "grad_norm": 0.556169867515564, "learning_rate": 0.000108079417606706, "loss": 2.4628, "step": 21030 }, { "epoch": 7.299716386910404e-14, "grad_norm": 0.5550949573516846, "learning_rate": 0.00010806342213484249, "loss": 2.4639, "step": 21040 }, { "epoch": 7.303185833862358e-14, "grad_norm": 0.5550413727760315, "learning_rate": 0.0001080474204458997, "loss": 2.472, "step": 21050 }, { "epoch": 7.306655280814311e-14, "grad_norm": 0.5689463019371033, "learning_rate": 0.00010803141254240812, "loss": 2.4776, "step": 21060 }, { "epoch": 7.310124727766265e-14, "grad_norm": 0.5414127707481384, "learning_rate": 0.00010801539842689929, "loss": 2.4857, "step": 21070 }, { "epoch": 7.313594174718219e-14, "grad_norm": 0.5424641370773315, "learning_rate": 0.0001079993781019057, "loss": 2.4802, "step": 21080 }, { "epoch": 7.317063621670172e-14, "grad_norm": 0.5456169843673706, "learning_rate": 0.00010798335156996082, "loss": 2.4586, "step": 21090 }, { "epoch": 7.320533068622126e-14, "grad_norm": 0.5868222117424011, "learning_rate": 0.00010796731883359913, "loss": 2.4687, "step": 21100 }, { "epoch": 7.32400251557408e-14, "grad_norm": 0.5341931581497192, "learning_rate": 0.00010795127989535603, "loss": 2.4652, "step": 21110 }, { "epoch": 7.327471962526033e-14, "grad_norm": 0.5531726479530334, "learning_rate": 0.00010793523475776797, "loss": 2.5092, "step": 21120 }, { "epoch": 7.330941409477987e-14, "grad_norm": 0.5502501726150513, "learning_rate": 0.00010791918342337234, "loss": 2.5051, "step": 21130 }, { "epoch": 7.33441085642994e-14, "grad_norm": 0.5467191338539124, "learning_rate": 0.00010790312589470753, "loss": 2.4491, "step": 21140 }, { "epoch": 7.337880303381894e-14, "grad_norm": 0.5939664840698242, "learning_rate": 0.00010788706217431288, "loss": 2.5232, "step": 21150 }, { "epoch": 7.341349750333848e-14, "grad_norm": 0.5390699505805969, "learning_rate": 0.00010787099226472875, "loss": 2.4887, "step": 21160 }, { "epoch": 7.344819197285801e-14, "grad_norm": 0.5577099323272705, "learning_rate": 0.00010785491616849644, "loss": 2.4692, "step": 21170 }, { "epoch": 7.348288644237755e-14, "grad_norm": 0.5406773090362549, "learning_rate": 0.00010783883388815827, "loss": 2.4989, "step": 21180 }, { "epoch": 7.351758091189708e-14, "grad_norm": 0.5508929491043091, "learning_rate": 0.0001078227454262575, "loss": 2.4983, "step": 21190 }, { "epoch": 7.355227538141662e-14, "grad_norm": 0.5449211597442627, "learning_rate": 0.00010780665078533837, "loss": 2.4556, "step": 21200 }, { "epoch": 7.358696985093616e-14, "grad_norm": 0.529397189617157, "learning_rate": 0.0001077905499679461, "loss": 2.4688, "step": 21210 }, { "epoch": 7.362166432045569e-14, "grad_norm": 0.5529760122299194, "learning_rate": 0.00010777444297662695, "loss": 2.4824, "step": 21220 }, { "epoch": 7.365635878997523e-14, "grad_norm": 0.5533760786056519, "learning_rate": 0.00010775832981392805, "loss": 2.4906, "step": 21230 }, { "epoch": 7.369105325949477e-14, "grad_norm": 0.5240657925605774, "learning_rate": 0.00010774221048239758, "loss": 2.4402, "step": 21240 }, { "epoch": 7.37257477290143e-14, "grad_norm": 0.5496078133583069, "learning_rate": 0.00010772608498458468, "loss": 2.4575, "step": 21250 }, { "epoch": 7.376044219853384e-14, "grad_norm": 0.6144286394119263, "learning_rate": 0.00010770995332303943, "loss": 2.4989, "step": 21260 }, { "epoch": 7.379513666805337e-14, "grad_norm": 0.5672423839569092, "learning_rate": 0.00010769381550031295, "loss": 2.4606, "step": 21270 }, { "epoch": 7.382983113757291e-14, "grad_norm": 0.5673855543136597, "learning_rate": 0.00010767767151895727, "loss": 2.4791, "step": 21280 }, { "epoch": 7.386452560709245e-14, "grad_norm": 0.5319676995277405, "learning_rate": 0.00010766152138152542, "loss": 2.4986, "step": 21290 }, { "epoch": 7.389922007661198e-14, "grad_norm": 0.565649688243866, "learning_rate": 0.00010764536509057144, "loss": 2.472, "step": 21300 }, { "epoch": 7.393391454613152e-14, "grad_norm": 0.5360819101333618, "learning_rate": 0.00010762920264865027, "loss": 2.4704, "step": 21310 }, { "epoch": 7.396860901565105e-14, "grad_norm": 0.5423641800880432, "learning_rate": 0.00010761303405831788, "loss": 2.4695, "step": 21320 }, { "epoch": 7.400330348517059e-14, "grad_norm": 0.540692925453186, "learning_rate": 0.00010759685932213118, "loss": 2.459, "step": 21330 }, { "epoch": 7.403799795469013e-14, "grad_norm": 3.4901013374328613, "learning_rate": 0.00010758067844264809, "loss": 2.4805, "step": 21340 }, { "epoch": 7.407269242420966e-14, "grad_norm": 0.5963630080223083, "learning_rate": 0.00010756449142242744, "loss": 2.4711, "step": 21350 }, { "epoch": 7.41073868937292e-14, "grad_norm": 0.5437384247779846, "learning_rate": 0.00010754829826402911, "loss": 2.5057, "step": 21360 }, { "epoch": 7.414208136324874e-14, "grad_norm": 0.5660656690597534, "learning_rate": 0.00010753209897001388, "loss": 2.5066, "step": 21370 }, { "epoch": 7.417677583276827e-14, "grad_norm": 0.5427932739257812, "learning_rate": 0.00010751589354294352, "loss": 2.4741, "step": 21380 }, { "epoch": 7.421147030228781e-14, "grad_norm": 0.5723586678504944, "learning_rate": 0.00010749968198538082, "loss": 2.4606, "step": 21390 }, { "epoch": 7.424616477180734e-14, "grad_norm": 0.5819128751754761, "learning_rate": 0.00010748346429988946, "loss": 2.4697, "step": 21400 }, { "epoch": 7.428085924132688e-14, "grad_norm": 0.531648576259613, "learning_rate": 0.00010746724048903416, "loss": 2.4578, "step": 21410 }, { "epoch": 7.431555371084642e-14, "grad_norm": 0.5543997287750244, "learning_rate": 0.00010745101055538053, "loss": 2.4327, "step": 21420 }, { "epoch": 7.435024818036595e-14, "grad_norm": 0.5367705225944519, "learning_rate": 0.00010743477450149525, "loss": 2.4554, "step": 21430 }, { "epoch": 7.438494264988549e-14, "grad_norm": 0.5267646312713623, "learning_rate": 0.00010741853232994585, "loss": 2.4607, "step": 21440 }, { "epoch": 7.441963711940502e-14, "grad_norm": 0.528314471244812, "learning_rate": 0.00010740228404330095, "loss": 2.4793, "step": 21450 }, { "epoch": 7.445433158892456e-14, "grad_norm": 0.542198121547699, "learning_rate": 0.00010738602964413004, "loss": 2.4684, "step": 21460 }, { "epoch": 7.44890260584441e-14, "grad_norm": 0.5223922729492188, "learning_rate": 0.0001073697691350036, "loss": 2.4436, "step": 21470 }, { "epoch": 7.452372052796363e-14, "grad_norm": 0.534054160118103, "learning_rate": 0.00010735350251849314, "loss": 2.4754, "step": 21480 }, { "epoch": 7.455841499748317e-14, "grad_norm": 0.5450066328048706, "learning_rate": 0.00010733722979717106, "loss": 2.4619, "step": 21490 }, { "epoch": 7.45931094670027e-14, "grad_norm": 0.5549519658088684, "learning_rate": 0.00010732095097361075, "loss": 2.4769, "step": 21500 }, { "epoch": 7.462780393652224e-14, "grad_norm": 0.5721248388290405, "learning_rate": 0.00010730466605038656, "loss": 2.453, "step": 21510 }, { "epoch": 7.466249840604178e-14, "grad_norm": 0.5575054883956909, "learning_rate": 0.0001072883750300738, "loss": 2.4524, "step": 21520 }, { "epoch": 7.469719287556131e-14, "grad_norm": 0.552089512348175, "learning_rate": 0.00010727207791524877, "loss": 2.4456, "step": 21530 }, { "epoch": 7.473188734508085e-14, "grad_norm": 0.5402194261550903, "learning_rate": 0.00010725577470848874, "loss": 2.4968, "step": 21540 }, { "epoch": 7.476658181460039e-14, "grad_norm": 0.5235426425933838, "learning_rate": 0.0001072394654123719, "loss": 2.4621, "step": 21550 }, { "epoch": 7.480127628411992e-14, "grad_norm": 0.5170088410377502, "learning_rate": 0.0001072231500294774, "loss": 2.4672, "step": 21560 }, { "epoch": 7.483597075363946e-14, "grad_norm": 0.5508016347885132, "learning_rate": 0.00010720682856238541, "loss": 2.4789, "step": 21570 }, { "epoch": 7.4870665223159e-14, "grad_norm": 0.5521009564399719, "learning_rate": 0.00010719050101367703, "loss": 2.477, "step": 21580 }, { "epoch": 7.490535969267853e-14, "grad_norm": 0.5368630290031433, "learning_rate": 0.00010717416738593431, "loss": 2.4119, "step": 21590 }, { "epoch": 7.494005416219807e-14, "grad_norm": 0.5298675298690796, "learning_rate": 0.00010715782768174029, "loss": 2.4486, "step": 21600 }, { "epoch": 7.49747486317176e-14, "grad_norm": 0.5234464406967163, "learning_rate": 0.00010714148190367892, "loss": 2.4751, "step": 21610 }, { "epoch": 7.500944310123714e-14, "grad_norm": 0.549571692943573, "learning_rate": 0.00010712513005433517, "loss": 2.4749, "step": 21620 }, { "epoch": 7.504413757075667e-14, "grad_norm": 0.5316060781478882, "learning_rate": 0.00010710877213629493, "loss": 2.465, "step": 21630 }, { "epoch": 7.507883204027621e-14, "grad_norm": 0.5356999635696411, "learning_rate": 0.00010709240815214508, "loss": 2.4943, "step": 21640 }, { "epoch": 7.511352650979575e-14, "grad_norm": 0.5633673667907715, "learning_rate": 0.00010707603810447342, "loss": 2.4214, "step": 21650 }, { "epoch": 7.514822097931528e-14, "grad_norm": 0.5260785222053528, "learning_rate": 0.00010705966199586875, "loss": 2.4523, "step": 21660 }, { "epoch": 7.518291544883482e-14, "grad_norm": 0.553510844707489, "learning_rate": 0.00010704327982892083, "loss": 2.4563, "step": 21670 }, { "epoch": 7.521760991835436e-14, "grad_norm": 0.5295028686523438, "learning_rate": 0.00010702689160622031, "loss": 2.4744, "step": 21680 }, { "epoch": 7.525230438787389e-14, "grad_norm": 0.5942913889884949, "learning_rate": 0.00010701049733035888, "loss": 2.4364, "step": 21690 }, { "epoch": 7.528699885739343e-14, "grad_norm": 0.5400057435035706, "learning_rate": 0.00010699409700392915, "loss": 2.4673, "step": 21700 }, { "epoch": 7.532169332691296e-14, "grad_norm": 0.5246851444244385, "learning_rate": 0.00010697769062952468, "loss": 2.4394, "step": 21710 }, { "epoch": 7.53563877964325e-14, "grad_norm": 0.5686829090118408, "learning_rate": 0.00010696127820974002, "loss": 2.4464, "step": 21720 }, { "epoch": 7.539108226595204e-14, "grad_norm": 0.5592916011810303, "learning_rate": 0.00010694485974717062, "loss": 2.5107, "step": 21730 }, { "epoch": 7.542577673547157e-14, "grad_norm": 0.5354910492897034, "learning_rate": 0.00010692843524441293, "loss": 2.4806, "step": 21740 }, { "epoch": 7.546047120499111e-14, "grad_norm": 0.633556604385376, "learning_rate": 0.00010691200470406435, "loss": 2.5071, "step": 21750 }, { "epoch": 7.549516567451064e-14, "grad_norm": 0.5572131872177124, "learning_rate": 0.00010689556812872324, "loss": 2.4582, "step": 21760 }, { "epoch": 7.552986014403018e-14, "grad_norm": 0.5620610117912292, "learning_rate": 0.00010687912552098888, "loss": 2.4664, "step": 21770 }, { "epoch": 7.556455461354972e-14, "grad_norm": 0.554878830909729, "learning_rate": 0.00010686267688346155, "loss": 2.4659, "step": 21780 }, { "epoch": 7.559924908306925e-14, "grad_norm": 0.5567675232887268, "learning_rate": 0.00010684622221874243, "loss": 2.4846, "step": 21790 }, { "epoch": 7.563394355258879e-14, "grad_norm": 0.5306344032287598, "learning_rate": 0.0001068297615294337, "loss": 2.426, "step": 21800 }, { "epoch": 7.566863802210833e-14, "grad_norm": 0.534853458404541, "learning_rate": 0.0001068132948181385, "loss": 2.4723, "step": 21810 }, { "epoch": 7.570333249162786e-14, "grad_norm": 0.5342264771461487, "learning_rate": 0.00010679682208746086, "loss": 2.4804, "step": 21820 }, { "epoch": 7.57380269611474e-14, "grad_norm": 0.5241295099258423, "learning_rate": 0.00010678034334000582, "loss": 2.4185, "step": 21830 }, { "epoch": 7.577272143066693e-14, "grad_norm": 0.539743959903717, "learning_rate": 0.00010676385857837933, "loss": 2.468, "step": 21840 }, { "epoch": 7.580741590018647e-14, "grad_norm": 0.5519183278083801, "learning_rate": 0.00010674736780518832, "loss": 2.4333, "step": 21850 }, { "epoch": 7.5842110369706e-14, "grad_norm": 0.5577542781829834, "learning_rate": 0.00010673087102304072, "loss": 2.4591, "step": 21860 }, { "epoch": 7.587680483922554e-14, "grad_norm": 0.5541655421257019, "learning_rate": 0.0001067143682345453, "loss": 2.4747, "step": 21870 }, { "epoch": 7.591149930874508e-14, "grad_norm": 0.5553311705589294, "learning_rate": 0.00010669785944231183, "loss": 2.4751, "step": 21880 }, { "epoch": 7.594619377826461e-14, "grad_norm": 0.5441707968711853, "learning_rate": 0.00010668134464895105, "loss": 2.4812, "step": 21890 }, { "epoch": 7.598088824778415e-14, "grad_norm": 0.5616073608398438, "learning_rate": 0.00010666482385707461, "loss": 2.4415, "step": 21900 }, { "epoch": 7.601558271730369e-14, "grad_norm": 0.5500084757804871, "learning_rate": 0.00010664829706929519, "loss": 2.4647, "step": 21910 }, { "epoch": 7.605027718682322e-14, "grad_norm": 0.5322548747062683, "learning_rate": 0.0001066317642882263, "loss": 2.4314, "step": 21920 }, { "epoch": 7.608497165634276e-14, "grad_norm": 0.5296065211296082, "learning_rate": 0.00010661522551648248, "loss": 2.4486, "step": 21930 }, { "epoch": 7.61196661258623e-14, "grad_norm": 0.5536172389984131, "learning_rate": 0.00010659868075667918, "loss": 2.484, "step": 21940 }, { "epoch": 7.615436059538183e-14, "grad_norm": 0.5488125681877136, "learning_rate": 0.00010658213001143282, "loss": 2.4955, "step": 21950 }, { "epoch": 7.618905506490137e-14, "grad_norm": 0.5635329484939575, "learning_rate": 0.00010656557328336076, "loss": 2.4864, "step": 21960 }, { "epoch": 7.62237495344209e-14, "grad_norm": 0.5638188123703003, "learning_rate": 0.00010654901057508132, "loss": 2.4325, "step": 21970 }, { "epoch": 7.625844400394044e-14, "grad_norm": 0.5351075530052185, "learning_rate": 0.00010653244188921373, "loss": 2.4559, "step": 21980 }, { "epoch": 7.629313847345998e-14, "grad_norm": 0.5462030172348022, "learning_rate": 0.0001065158672283782, "loss": 2.4503, "step": 21990 }, { "epoch": 7.632783294297951e-14, "grad_norm": 0.5850192904472351, "learning_rate": 0.00010649928659519583, "loss": 2.4613, "step": 22000 }, { "epoch": 7.636252741249905e-14, "grad_norm": 0.5417519211769104, "learning_rate": 0.00010648269999228874, "loss": 2.4718, "step": 22010 }, { "epoch": 7.639722188201858e-14, "grad_norm": 0.5947379469871521, "learning_rate": 0.00010646610742227995, "loss": 2.5133, "step": 22020 }, { "epoch": 7.643191635153812e-14, "grad_norm": 0.5540027618408203, "learning_rate": 0.00010644950888779345, "loss": 2.4637, "step": 22030 }, { "epoch": 7.646661082105766e-14, "grad_norm": 0.5651462078094482, "learning_rate": 0.00010643290439145413, "loss": 2.4321, "step": 22040 }, { "epoch": 7.650130529057719e-14, "grad_norm": 0.5574852824211121, "learning_rate": 0.00010641629393588786, "loss": 2.4575, "step": 22050 }, { "epoch": 7.653599976009673e-14, "grad_norm": 0.5367576479911804, "learning_rate": 0.00010639967752372142, "loss": 2.4354, "step": 22060 }, { "epoch": 7.657069422961627e-14, "grad_norm": 0.5538796186447144, "learning_rate": 0.00010638305515758259, "loss": 2.4661, "step": 22070 }, { "epoch": 7.66053886991358e-14, "grad_norm": 0.5897275805473328, "learning_rate": 0.00010636642684010001, "loss": 2.4498, "step": 22080 }, { "epoch": 7.664008316865534e-14, "grad_norm": 0.5515284538269043, "learning_rate": 0.00010634979257390335, "loss": 2.4646, "step": 22090 }, { "epoch": 7.667477763817487e-14, "grad_norm": 0.5694953203201294, "learning_rate": 0.00010633315236162318, "loss": 2.475, "step": 22100 }, { "epoch": 7.670947210769441e-14, "grad_norm": 0.5848621129989624, "learning_rate": 0.00010631650620589094, "loss": 2.4369, "step": 22110 }, { "epoch": 7.674416657721395e-14, "grad_norm": 0.5386332273483276, "learning_rate": 0.00010629985410933916, "loss": 2.4016, "step": 22120 }, { "epoch": 7.677886104673348e-14, "grad_norm": 0.5463725924491882, "learning_rate": 0.00010628319607460115, "loss": 2.4878, "step": 22130 }, { "epoch": 7.681355551625302e-14, "grad_norm": 0.5319032073020935, "learning_rate": 0.00010626653210431127, "loss": 2.4641, "step": 22140 }, { "epoch": 7.684824998577255e-14, "grad_norm": 0.5187978148460388, "learning_rate": 0.00010624986220110479, "loss": 2.4683, "step": 22150 }, { "epoch": 7.688294445529209e-14, "grad_norm": 0.5393623113632202, "learning_rate": 0.0001062331863676179, "loss": 2.5109, "step": 22160 }, { "epoch": 7.691763892481163e-14, "grad_norm": 0.6076535582542419, "learning_rate": 0.00010621650460648774, "loss": 2.4859, "step": 22170 }, { "epoch": 7.695233339433116e-14, "grad_norm": 0.5526183247566223, "learning_rate": 0.0001061998169203524, "loss": 2.4245, "step": 22180 }, { "epoch": 7.69870278638507e-14, "grad_norm": 0.5289157032966614, "learning_rate": 0.00010618312331185088, "loss": 2.4335, "step": 22190 }, { "epoch": 7.702172233337023e-14, "grad_norm": 0.559384286403656, "learning_rate": 0.00010616642378362314, "loss": 2.4649, "step": 22200 }, { "epoch": 7.705641680288977e-14, "grad_norm": 0.5375635027885437, "learning_rate": 0.00010614971833831006, "loss": 2.4651, "step": 22210 }, { "epoch": 7.709111127240931e-14, "grad_norm": 0.5795628428459167, "learning_rate": 0.00010613300697855346, "loss": 2.4674, "step": 22220 }, { "epoch": 7.712580574192884e-14, "grad_norm": 0.5666950345039368, "learning_rate": 0.0001061162897069961, "loss": 2.468, "step": 22230 }, { "epoch": 7.716050021144838e-14, "grad_norm": 0.533785879611969, "learning_rate": 0.00010609956652628169, "loss": 2.4499, "step": 22240 }, { "epoch": 7.719519468096792e-14, "grad_norm": 0.5402054190635681, "learning_rate": 0.00010608283743905482, "loss": 2.4155, "step": 22250 }, { "epoch": 7.722988915048745e-14, "grad_norm": 0.5492618680000305, "learning_rate": 0.00010606610244796108, "loss": 2.4416, "step": 22260 }, { "epoch": 7.726458362000699e-14, "grad_norm": 0.5439861416816711, "learning_rate": 0.00010604936155564694, "loss": 2.4702, "step": 22270 }, { "epoch": 7.729927808952652e-14, "grad_norm": 0.5594990253448486, "learning_rate": 0.0001060326147647599, "loss": 2.4599, "step": 22280 }, { "epoch": 7.733397255904606e-14, "grad_norm": 0.5380552411079407, "learning_rate": 0.0001060158620779482, "loss": 2.4338, "step": 22290 }, { "epoch": 7.73686670285656e-14, "grad_norm": 0.5592819452285767, "learning_rate": 0.00010599910349786123, "loss": 2.4326, "step": 22300 }, { "epoch": 7.740336149808513e-14, "grad_norm": 0.5650115013122559, "learning_rate": 0.0001059823390271492, "loss": 2.474, "step": 22310 }, { "epoch": 7.743805596760467e-14, "grad_norm": 0.5557757616043091, "learning_rate": 0.00010596556866846321, "loss": 2.4898, "step": 22320 }, { "epoch": 7.74727504371242e-14, "grad_norm": 0.5244106650352478, "learning_rate": 0.00010594879242445542, "loss": 2.4416, "step": 22330 }, { "epoch": 7.750744490664374e-14, "grad_norm": 0.5351771116256714, "learning_rate": 0.00010593201029777881, "loss": 2.4448, "step": 22340 }, { "epoch": 7.754213937616328e-14, "grad_norm": 0.5692909359931946, "learning_rate": 0.00010591522229108734, "loss": 2.4611, "step": 22350 }, { "epoch": 7.757683384568281e-14, "grad_norm": 0.5262425541877747, "learning_rate": 0.00010589842840703589, "loss": 2.4647, "step": 22360 }, { "epoch": 7.761152831520235e-14, "grad_norm": 0.5579456686973572, "learning_rate": 0.00010588162864828024, "loss": 2.4236, "step": 22370 }, { "epoch": 7.764622278472189e-14, "grad_norm": 0.5645949840545654, "learning_rate": 0.00010586482301747716, "loss": 2.4613, "step": 22380 }, { "epoch": 7.768091725424142e-14, "grad_norm": 0.5212802886962891, "learning_rate": 0.00010584801151728431, "loss": 2.4248, "step": 22390 }, { "epoch": 7.771561172376096e-14, "grad_norm": 0.5517788529396057, "learning_rate": 0.00010583119415036028, "loss": 2.4461, "step": 22400 }, { "epoch": 7.77503061932805e-14, "grad_norm": 0.526538610458374, "learning_rate": 0.0001058143709193646, "loss": 2.4782, "step": 22410 }, { "epoch": 7.778500066280003e-14, "grad_norm": 0.5653492212295532, "learning_rate": 0.00010579754182695771, "loss": 2.4697, "step": 22420 }, { "epoch": 7.781969513231957e-14, "grad_norm": 0.5217351913452148, "learning_rate": 0.00010578070687580098, "loss": 2.4791, "step": 22430 }, { "epoch": 7.78543896018391e-14, "grad_norm": 0.5704909563064575, "learning_rate": 0.00010576386606855671, "loss": 2.4798, "step": 22440 }, { "epoch": 7.788908407135864e-14, "grad_norm": 0.5320776104927063, "learning_rate": 0.00010574701940788817, "loss": 2.4543, "step": 22450 }, { "epoch": 7.792377854087817e-14, "grad_norm": 0.5399250388145447, "learning_rate": 0.00010573016689645947, "loss": 2.4063, "step": 22460 }, { "epoch": 7.795847301039771e-14, "grad_norm": 0.5420365929603577, "learning_rate": 0.00010571330853693569, "loss": 2.4313, "step": 22470 }, { "epoch": 7.799316747991725e-14, "grad_norm": 0.5538212060928345, "learning_rate": 0.00010569644433198285, "loss": 2.4503, "step": 22480 }, { "epoch": 7.802786194943678e-14, "grad_norm": 0.5553978681564331, "learning_rate": 0.00010567957428426787, "loss": 2.4762, "step": 22490 }, { "epoch": 7.806255641895632e-14, "grad_norm": 0.5194708704948425, "learning_rate": 0.00010566269839645863, "loss": 2.4411, "step": 22500 }, { "epoch": 7.809725088847586e-14, "grad_norm": 0.5431689620018005, "learning_rate": 0.00010564581667122387, "loss": 2.4552, "step": 22510 }, { "epoch": 7.813194535799539e-14, "grad_norm": 0.5276854038238525, "learning_rate": 0.00010562892911123332, "loss": 2.499, "step": 22520 }, { "epoch": 7.816663982751493e-14, "grad_norm": 0.5555587410926819, "learning_rate": 0.00010561203571915758, "loss": 2.4835, "step": 22530 }, { "epoch": 7.820133429703446e-14, "grad_norm": 0.55650794506073, "learning_rate": 0.0001055951364976682, "loss": 2.4479, "step": 22540 }, { "epoch": 7.8236028766554e-14, "grad_norm": 0.5517092347145081, "learning_rate": 0.00010557823144943764, "loss": 2.4473, "step": 22550 }, { "epoch": 7.827072323607354e-14, "grad_norm": 0.5301164388656616, "learning_rate": 0.0001055613205771393, "loss": 2.4386, "step": 22560 }, { "epoch": 7.830541770559307e-14, "grad_norm": 0.5448908805847168, "learning_rate": 0.0001055444038834475, "loss": 2.4642, "step": 22570 }, { "epoch": 7.834011217511261e-14, "grad_norm": 0.5587882995605469, "learning_rate": 0.00010552748137103743, "loss": 2.4634, "step": 22580 }, { "epoch": 7.837480664463214e-14, "grad_norm": 0.5381949543952942, "learning_rate": 0.00010551055304258526, "loss": 2.4447, "step": 22590 }, { "epoch": 7.840950111415168e-14, "grad_norm": 0.5667337775230408, "learning_rate": 0.00010549361890076808, "loss": 2.4281, "step": 22600 }, { "epoch": 7.844419558367122e-14, "grad_norm": 0.5692315697669983, "learning_rate": 0.00010547667894826388, "loss": 2.4804, "step": 22610 }, { "epoch": 7.847889005319075e-14, "grad_norm": 0.538310706615448, "learning_rate": 0.00010545973318775153, "loss": 2.4722, "step": 22620 }, { "epoch": 7.851358452271029e-14, "grad_norm": 0.5276389122009277, "learning_rate": 0.00010544278162191087, "loss": 2.4685, "step": 22630 }, { "epoch": 7.854827899222983e-14, "grad_norm": 0.583629846572876, "learning_rate": 0.00010542582425342265, "loss": 2.429, "step": 22640 }, { "epoch": 7.858297346174936e-14, "grad_norm": 0.5244770050048828, "learning_rate": 0.00010540886108496855, "loss": 2.4449, "step": 22650 }, { "epoch": 7.86176679312689e-14, "grad_norm": 0.6011843681335449, "learning_rate": 0.00010539189211923112, "loss": 2.4603, "step": 22660 }, { "epoch": 7.865236240078843e-14, "grad_norm": 0.5309341549873352, "learning_rate": 0.00010537491735889387, "loss": 2.5017, "step": 22670 }, { "epoch": 7.868705687030797e-14, "grad_norm": 0.5551658868789673, "learning_rate": 0.0001053579368066412, "loss": 2.4598, "step": 22680 }, { "epoch": 7.87217513398275e-14, "grad_norm": 0.5819404125213623, "learning_rate": 0.00010534095046515846, "loss": 2.481, "step": 22690 }, { "epoch": 7.875644580934704e-14, "grad_norm": 0.5131975412368774, "learning_rate": 0.00010532395833713187, "loss": 2.4845, "step": 22700 }, { "epoch": 7.879114027886658e-14, "grad_norm": 0.5470442771911621, "learning_rate": 0.00010530696042524859, "loss": 2.4716, "step": 22710 }, { "epoch": 7.882583474838611e-14, "grad_norm": 0.625395655632019, "learning_rate": 0.00010528995673219672, "loss": 2.4639, "step": 22720 }, { "epoch": 7.886052921790565e-14, "grad_norm": 0.5847963690757751, "learning_rate": 0.00010527294726066523, "loss": 2.4505, "step": 22730 }, { "epoch": 7.889522368742519e-14, "grad_norm": 0.5468015074729919, "learning_rate": 0.00010525593201334402, "loss": 2.4722, "step": 22740 }, { "epoch": 7.892991815694472e-14, "grad_norm": 0.5571111440658569, "learning_rate": 0.0001052389109929239, "loss": 2.4959, "step": 22750 }, { "epoch": 7.896461262646426e-14, "grad_norm": 0.5975949764251709, "learning_rate": 0.0001052218842020966, "loss": 2.4535, "step": 22760 }, { "epoch": 7.89993070959838e-14, "grad_norm": 0.5716298222541809, "learning_rate": 0.0001052048516435548, "loss": 2.4594, "step": 22770 }, { "epoch": 7.903400156550333e-14, "grad_norm": 0.5460346341133118, "learning_rate": 0.000105187813319992, "loss": 2.492, "step": 22780 }, { "epoch": 7.906869603502287e-14, "grad_norm": 0.5538521409034729, "learning_rate": 0.00010517076923410268, "loss": 2.4793, "step": 22790 }, { "epoch": 7.91033905045424e-14, "grad_norm": 0.5596922039985657, "learning_rate": 0.00010515371938858221, "loss": 2.4312, "step": 22800 }, { "epoch": 7.913808497406194e-14, "grad_norm": 0.5711422562599182, "learning_rate": 0.0001051366637861269, "loss": 2.4362, "step": 22810 }, { "epoch": 7.917277944358148e-14, "grad_norm": 0.5200814008712769, "learning_rate": 0.00010511960242943391, "loss": 2.4696, "step": 22820 }, { "epoch": 7.920747391310101e-14, "grad_norm": 0.5570937395095825, "learning_rate": 0.00010510253532120139, "loss": 2.4677, "step": 22830 }, { "epoch": 7.924216838262055e-14, "grad_norm": 0.5624849796295166, "learning_rate": 0.00010508546246412833, "loss": 2.4654, "step": 22840 }, { "epoch": 7.927686285214008e-14, "grad_norm": 0.5659021139144897, "learning_rate": 0.00010506838386091464, "loss": 2.4474, "step": 22850 }, { "epoch": 7.931155732165962e-14, "grad_norm": 0.5335783362388611, "learning_rate": 0.00010505129951426118, "loss": 2.4452, "step": 22860 }, { "epoch": 7.934625179117916e-14, "grad_norm": 0.5764058232307434, "learning_rate": 0.0001050342094268697, "loss": 2.4605, "step": 22870 }, { "epoch": 7.938094626069869e-14, "grad_norm": 0.5634549260139465, "learning_rate": 0.00010501711360144282, "loss": 2.4448, "step": 22880 }, { "epoch": 7.941564073021823e-14, "grad_norm": 0.5680103898048401, "learning_rate": 0.00010500001204068413, "loss": 2.445, "step": 22890 }, { "epoch": 7.945033519973776e-14, "grad_norm": 0.5491111278533936, "learning_rate": 0.00010498290474729806, "loss": 2.4849, "step": 22900 }, { "epoch": 7.94850296692573e-14, "grad_norm": 0.5625021457672119, "learning_rate": 0.00010496579172399, "loss": 2.4647, "step": 22910 }, { "epoch": 7.951972413877684e-14, "grad_norm": 0.5532258152961731, "learning_rate": 0.00010494867297346623, "loss": 2.4607, "step": 22920 }, { "epoch": 7.955441860829637e-14, "grad_norm": 0.5981163382530212, "learning_rate": 0.00010493154849843394, "loss": 2.467, "step": 22930 }, { "epoch": 7.958911307781591e-14, "grad_norm": 0.533108651638031, "learning_rate": 0.0001049144183016012, "loss": 2.43, "step": 22940 }, { "epoch": 7.962380754733545e-14, "grad_norm": 0.5516650080680847, "learning_rate": 0.00010489728238567701, "loss": 2.4522, "step": 22950 }, { "epoch": 7.965850201685498e-14, "grad_norm": 0.5617630481719971, "learning_rate": 0.00010488014075337126, "loss": 2.4602, "step": 22960 }, { "epoch": 7.969319648637452e-14, "grad_norm": 0.5608962774276733, "learning_rate": 0.00010486299340739478, "loss": 2.4625, "step": 22970 }, { "epoch": 7.972789095589405e-14, "grad_norm": 0.5526724457740784, "learning_rate": 0.00010484584035045925, "loss": 2.4388, "step": 22980 }, { "epoch": 7.976258542541359e-14, "grad_norm": 0.5507545471191406, "learning_rate": 0.00010482868158527729, "loss": 2.4793, "step": 22990 }, { "epoch": 7.979727989493313e-14, "grad_norm": 0.5289723873138428, "learning_rate": 0.00010481151711456238, "loss": 2.4517, "step": 23000 }, { "epoch": 7.983197436445266e-14, "grad_norm": 0.5477055907249451, "learning_rate": 0.00010479434694102898, "loss": 2.4323, "step": 23010 }, { "epoch": 7.98666688339722e-14, "grad_norm": 0.5647627711296082, "learning_rate": 0.00010477717106739236, "loss": 2.4398, "step": 23020 }, { "epoch": 7.990136330349173e-14, "grad_norm": 0.7289478778839111, "learning_rate": 0.00010475998949636878, "loss": 2.4805, "step": 23030 }, { "epoch": 7.993605777301127e-14, "grad_norm": 0.5549210906028748, "learning_rate": 0.00010474280223067536, "loss": 2.4887, "step": 23040 }, { "epoch": 7.997075224253081e-14, "grad_norm": 0.5224712491035461, "learning_rate": 0.00010472560927303004, "loss": 2.4164, "step": 23050 }, { "epoch": 8.000544671205034e-14, "grad_norm": 0.5357049107551575, "learning_rate": 0.00010470841062615184, "loss": 2.49, "step": 23060 }, { "epoch": 8.004014118156988e-14, "grad_norm": 0.5791611671447754, "learning_rate": 0.0001046912062927605, "loss": 2.459, "step": 23070 }, { "epoch": 8.007483565108942e-14, "grad_norm": 0.5495725274085999, "learning_rate": 0.00010467399627557679, "loss": 2.4321, "step": 23080 }, { "epoch": 8.010953012060895e-14, "grad_norm": 0.5869030356407166, "learning_rate": 0.0001046567805773223, "loss": 2.4344, "step": 23090 }, { "epoch": 8.014422459012849e-14, "grad_norm": 0.5392690896987915, "learning_rate": 0.00010463955920071953, "loss": 2.4533, "step": 23100 }, { "epoch": 8.017891905964802e-14, "grad_norm": 0.5719928741455078, "learning_rate": 0.00010462233214849194, "loss": 2.4657, "step": 23110 }, { "epoch": 8.021361352916756e-14, "grad_norm": 0.5312402844429016, "learning_rate": 0.0001046050994233638, "loss": 2.4556, "step": 23120 }, { "epoch": 8.02483079986871e-14, "grad_norm": 0.5430520176887512, "learning_rate": 0.00010458786102806033, "loss": 2.4819, "step": 23130 }, { "epoch": 8.028300246820663e-14, "grad_norm": 0.5363131761550903, "learning_rate": 0.00010457061696530763, "loss": 2.4615, "step": 23140 }, { "epoch": 8.031769693772617e-14, "grad_norm": 0.5439170002937317, "learning_rate": 0.00010455336723783271, "loss": 2.4501, "step": 23150 }, { "epoch": 8.03523914072457e-14, "grad_norm": 0.5470542311668396, "learning_rate": 0.00010453611184836345, "loss": 2.4961, "step": 23160 }, { "epoch": 8.038708587676524e-14, "grad_norm": 0.5382347702980042, "learning_rate": 0.00010451885079962867, "loss": 2.4126, "step": 23170 }, { "epoch": 8.042178034628478e-14, "grad_norm": 0.552560031414032, "learning_rate": 0.000104501584094358, "loss": 2.473, "step": 23180 }, { "epoch": 8.045647481580431e-14, "grad_norm": 0.5201101303100586, "learning_rate": 0.00010448431173528209, "loss": 2.4321, "step": 23190 }, { "epoch": 8.049116928532385e-14, "grad_norm": 0.5371541976928711, "learning_rate": 0.00010446703372513236, "loss": 2.4582, "step": 23200 }, { "epoch": 8.052586375484339e-14, "grad_norm": 0.5683630108833313, "learning_rate": 0.0001044497500666412, "loss": 2.428, "step": 23210 }, { "epoch": 8.056055822436292e-14, "grad_norm": 0.5544760227203369, "learning_rate": 0.0001044324607625419, "loss": 2.478, "step": 23220 }, { "epoch": 8.059525269388246e-14, "grad_norm": 0.5647435784339905, "learning_rate": 0.00010441516581556856, "loss": 2.4628, "step": 23230 }, { "epoch": 8.0629947163402e-14, "grad_norm": 0.5507259368896484, "learning_rate": 0.00010439786522845626, "loss": 2.4736, "step": 23240 }, { "epoch": 8.066464163292153e-14, "grad_norm": 0.5315088033676147, "learning_rate": 0.00010438055900394094, "loss": 2.4584, "step": 23250 }, { "epoch": 8.069933610244107e-14, "grad_norm": 0.5618688464164734, "learning_rate": 0.00010436324714475944, "loss": 2.4663, "step": 23260 }, { "epoch": 8.07340305719606e-14, "grad_norm": 0.5647534132003784, "learning_rate": 0.00010434592965364945, "loss": 2.4938, "step": 23270 }, { "epoch": 8.076872504148014e-14, "grad_norm": 0.5279672145843506, "learning_rate": 0.00010432860653334957, "loss": 2.4174, "step": 23280 }, { "epoch": 8.080341951099967e-14, "grad_norm": 0.5421703457832336, "learning_rate": 0.00010431127778659936, "loss": 2.4638, "step": 23290 }, { "epoch": 8.083811398051921e-14, "grad_norm": 0.5501317381858826, "learning_rate": 0.00010429394341613917, "loss": 2.4528, "step": 23300 }, { "epoch": 8.087280845003875e-14, "grad_norm": 0.5144875645637512, "learning_rate": 0.00010427660342471031, "loss": 2.4455, "step": 23310 }, { "epoch": 8.090750291955828e-14, "grad_norm": 0.5711395740509033, "learning_rate": 0.00010425925781505493, "loss": 2.4329, "step": 23320 }, { "epoch": 8.094219738907782e-14, "grad_norm": 0.5334223508834839, "learning_rate": 0.00010424190658991606, "loss": 2.4419, "step": 23330 }, { "epoch": 8.097689185859736e-14, "grad_norm": 0.5241640210151672, "learning_rate": 0.0001042245497520377, "loss": 2.4851, "step": 23340 }, { "epoch": 8.101158632811689e-14, "grad_norm": 0.5488755702972412, "learning_rate": 0.00010420718730416468, "loss": 2.4367, "step": 23350 }, { "epoch": 8.104628079763643e-14, "grad_norm": 0.5792297124862671, "learning_rate": 0.00010418981924904268, "loss": 2.4562, "step": 23360 }, { "epoch": 8.108097526715596e-14, "grad_norm": 0.5641176700592041, "learning_rate": 0.00010417244558941835, "loss": 2.4779, "step": 23370 }, { "epoch": 8.11156697366755e-14, "grad_norm": 0.5785850882530212, "learning_rate": 0.00010415506632803917, "loss": 2.4482, "step": 23380 }, { "epoch": 8.115036420619504e-14, "grad_norm": 0.5201356410980225, "learning_rate": 0.0001041376814676535, "loss": 2.4373, "step": 23390 }, { "epoch": 8.118505867571457e-14, "grad_norm": 0.5150402784347534, "learning_rate": 0.00010412029101101064, "loss": 2.4188, "step": 23400 }, { "epoch": 8.121975314523411e-14, "grad_norm": 0.5293119549751282, "learning_rate": 0.00010410289496086072, "loss": 2.4721, "step": 23410 }, { "epoch": 8.125444761475364e-14, "grad_norm": 0.5274626612663269, "learning_rate": 0.0001040854933199548, "loss": 2.4334, "step": 23420 }, { "epoch": 8.128914208427318e-14, "grad_norm": 0.540248453617096, "learning_rate": 0.00010406808609104475, "loss": 2.4475, "step": 23430 }, { "epoch": 8.132383655379272e-14, "grad_norm": 0.5396832227706909, "learning_rate": 0.00010405067327688343, "loss": 2.4276, "step": 23440 }, { "epoch": 8.135853102331225e-14, "grad_norm": 0.541689932346344, "learning_rate": 0.0001040332548802245, "loss": 2.4438, "step": 23450 }, { "epoch": 8.139322549283179e-14, "grad_norm": 0.5421687960624695, "learning_rate": 0.00010401583090382253, "loss": 2.4486, "step": 23460 }, { "epoch": 8.142791996235133e-14, "grad_norm": 0.5435052514076233, "learning_rate": 0.00010399840135043297, "loss": 2.4597, "step": 23470 }, { "epoch": 8.146261443187086e-14, "grad_norm": 0.5398709177970886, "learning_rate": 0.00010398096622281215, "loss": 2.4608, "step": 23480 }, { "epoch": 8.14973089013904e-14, "grad_norm": 0.5517069101333618, "learning_rate": 0.0001039635255237173, "loss": 2.4839, "step": 23490 }, { "epoch": 8.153200337090993e-14, "grad_norm": 0.6210787296295166, "learning_rate": 0.00010394607925590651, "loss": 2.4465, "step": 23500 }, { "epoch": 8.156669784042947e-14, "grad_norm": 0.6217390298843384, "learning_rate": 0.00010392862742213874, "loss": 2.4347, "step": 23510 }, { "epoch": 8.1601392309949e-14, "grad_norm": 0.5792965888977051, "learning_rate": 0.00010391117002517387, "loss": 2.435, "step": 23520 }, { "epoch": 8.163608677946854e-14, "grad_norm": 0.558043360710144, "learning_rate": 0.00010389370706777265, "loss": 2.4255, "step": 23530 }, { "epoch": 8.167078124898808e-14, "grad_norm": 0.5260147452354431, "learning_rate": 0.00010387623855269665, "loss": 2.4756, "step": 23540 }, { "epoch": 8.170547571850761e-14, "grad_norm": 0.5640802383422852, "learning_rate": 0.0001038587644827084, "loss": 2.4458, "step": 23550 }, { "epoch": 8.174017018802715e-14, "grad_norm": 0.5505959391593933, "learning_rate": 0.00010384128486057126, "loss": 2.4616, "step": 23560 }, { "epoch": 8.177486465754669e-14, "grad_norm": 0.569680392742157, "learning_rate": 0.00010382379968904949, "loss": 2.4074, "step": 23570 }, { "epoch": 8.180955912706622e-14, "grad_norm": 0.5425609350204468, "learning_rate": 0.00010380630897090823, "loss": 2.45, "step": 23580 }, { "epoch": 8.184425359658576e-14, "grad_norm": 0.5553896427154541, "learning_rate": 0.00010378881270891346, "loss": 2.4873, "step": 23590 }, { "epoch": 8.18789480661053e-14, "grad_norm": 0.5645985007286072, "learning_rate": 0.00010377131090583208, "loss": 2.4574, "step": 23600 }, { "epoch": 8.191364253562483e-14, "grad_norm": 0.5942535400390625, "learning_rate": 0.00010375380356443184, "loss": 2.4166, "step": 23610 }, { "epoch": 8.194833700514437e-14, "grad_norm": 0.5579211711883545, "learning_rate": 0.00010373629068748141, "loss": 2.4552, "step": 23620 }, { "epoch": 8.19830314746639e-14, "grad_norm": 0.5492250919342041, "learning_rate": 0.00010371877227775024, "loss": 2.4289, "step": 23630 }, { "epoch": 8.201772594418344e-14, "grad_norm": 0.5312198400497437, "learning_rate": 0.00010370124833800876, "loss": 2.446, "step": 23640 }, { "epoch": 8.205242041370298e-14, "grad_norm": 0.5353894829750061, "learning_rate": 0.00010368371887102822, "loss": 2.4475, "step": 23650 }, { "epoch": 8.208711488322251e-14, "grad_norm": 0.5400974154472351, "learning_rate": 0.00010366618387958074, "loss": 2.4384, "step": 23660 }, { "epoch": 8.212180935274205e-14, "grad_norm": 0.5500911474227905, "learning_rate": 0.00010364864336643934, "loss": 2.4447, "step": 23670 }, { "epoch": 8.215650382226158e-14, "grad_norm": 0.5417644381523132, "learning_rate": 0.00010363109733437791, "loss": 2.4703, "step": 23680 }, { "epoch": 8.219119829178112e-14, "grad_norm": 0.5828315615653992, "learning_rate": 0.0001036135457861712, "loss": 2.4443, "step": 23690 }, { "epoch": 8.222589276130066e-14, "grad_norm": 0.5461673736572266, "learning_rate": 0.0001035959887245948, "loss": 2.4533, "step": 23700 }, { "epoch": 8.226058723082019e-14, "grad_norm": 0.5689979195594788, "learning_rate": 0.00010357842615242527, "loss": 2.4526, "step": 23710 }, { "epoch": 8.229528170033973e-14, "grad_norm": 0.5473804473876953, "learning_rate": 0.00010356085807243996, "loss": 2.4351, "step": 23720 }, { "epoch": 8.232997616985926e-14, "grad_norm": 0.5606071949005127, "learning_rate": 0.00010354328448741707, "loss": 2.4441, "step": 23730 }, { "epoch": 8.23646706393788e-14, "grad_norm": 0.5717425346374512, "learning_rate": 0.00010352570540013576, "loss": 2.4759, "step": 23740 }, { "epoch": 8.239936510889834e-14, "grad_norm": 0.5562634468078613, "learning_rate": 0.00010350812081337597, "loss": 2.4315, "step": 23750 }, { "epoch": 8.243405957841787e-14, "grad_norm": 0.5611185431480408, "learning_rate": 0.00010349053072991858, "loss": 2.4563, "step": 23760 }, { "epoch": 8.246875404793741e-14, "grad_norm": 0.5853745937347412, "learning_rate": 0.00010347293515254533, "loss": 2.4235, "step": 23770 }, { "epoch": 8.250344851745695e-14, "grad_norm": 0.5411030054092407, "learning_rate": 0.00010345533408403877, "loss": 2.4679, "step": 23780 }, { "epoch": 8.253814298697648e-14, "grad_norm": 0.5427550077438354, "learning_rate": 0.00010343772752718236, "loss": 2.4717, "step": 23790 }, { "epoch": 8.257283745649602e-14, "grad_norm": 0.5358166694641113, "learning_rate": 0.00010342011548476045, "loss": 2.446, "step": 23800 }, { "epoch": 8.260753192601555e-14, "grad_norm": 0.5842272639274597, "learning_rate": 0.00010340249795955821, "loss": 2.4447, "step": 23810 }, { "epoch": 8.264222639553509e-14, "grad_norm": 0.568020224571228, "learning_rate": 0.0001033848749543617, "loss": 2.3976, "step": 23820 }, { "epoch": 8.267692086505463e-14, "grad_norm": 0.5496253371238708, "learning_rate": 0.00010336724647195786, "loss": 2.4477, "step": 23830 }, { "epoch": 8.271161533457416e-14, "grad_norm": 0.556081235408783, "learning_rate": 0.00010334961251513448, "loss": 2.4648, "step": 23840 }, { "epoch": 8.27463098040937e-14, "grad_norm": 0.585643470287323, "learning_rate": 0.00010333197308668023, "loss": 2.438, "step": 23850 }, { "epoch": 8.278100427361323e-14, "grad_norm": 0.6747328639030457, "learning_rate": 0.0001033143281893846, "loss": 2.4043, "step": 23860 }, { "epoch": 8.281569874313277e-14, "grad_norm": 0.5339018702507019, "learning_rate": 0.000103296677826038, "loss": 2.4342, "step": 23870 }, { "epoch": 8.285039321265231e-14, "grad_norm": 0.5384336113929749, "learning_rate": 0.0001032790219994317, "loss": 2.4397, "step": 23880 }, { "epoch": 8.288508768217184e-14, "grad_norm": 0.5520952939987183, "learning_rate": 0.00010326136071235778, "loss": 2.4782, "step": 23890 }, { "epoch": 8.291978215169138e-14, "grad_norm": 0.5595542192459106, "learning_rate": 0.00010324369396760925, "loss": 2.4399, "step": 23900 }, { "epoch": 8.295447662121092e-14, "grad_norm": 0.5649265050888062, "learning_rate": 0.00010322602176797993, "loss": 2.4578, "step": 23910 }, { "epoch": 8.298917109073045e-14, "grad_norm": 0.5226211547851562, "learning_rate": 0.00010320834411626455, "loss": 2.4345, "step": 23920 }, { "epoch": 8.302386556024999e-14, "grad_norm": 0.5851272344589233, "learning_rate": 0.00010319066101525866, "loss": 2.4703, "step": 23930 }, { "epoch": 8.305856002976952e-14, "grad_norm": 0.5590537190437317, "learning_rate": 0.00010317297246775871, "loss": 2.4353, "step": 23940 }, { "epoch": 8.309325449928906e-14, "grad_norm": 0.519929051399231, "learning_rate": 0.00010315527847656196, "loss": 2.4134, "step": 23950 }, { "epoch": 8.31279489688086e-14, "grad_norm": 0.5434836745262146, "learning_rate": 0.0001031375790444666, "loss": 2.4082, "step": 23960 }, { "epoch": 8.316264343832813e-14, "grad_norm": 0.5530901551246643, "learning_rate": 0.00010311987417427163, "loss": 2.4178, "step": 23970 }, { "epoch": 8.319733790784767e-14, "grad_norm": 0.5496677756309509, "learning_rate": 0.00010310216386877688, "loss": 2.4435, "step": 23980 }, { "epoch": 8.32320323773672e-14, "grad_norm": 0.6129196286201477, "learning_rate": 0.00010308444813078316, "loss": 2.427, "step": 23990 }, { "epoch": 8.326672684688674e-14, "grad_norm": 0.5833174586296082, "learning_rate": 0.000103066726963092, "loss": 2.442, "step": 24000 }, { "epoch": 8.330142131640628e-14, "grad_norm": 0.547748327255249, "learning_rate": 0.00010304900036850588, "loss": 2.4207, "step": 24010 }, { "epoch": 8.333611578592581e-14, "grad_norm": 0.5436789989471436, "learning_rate": 0.0001030312683498281, "loss": 2.4559, "step": 24020 }, { "epoch": 8.337081025544535e-14, "grad_norm": 0.540674090385437, "learning_rate": 0.00010301353090986282, "loss": 2.4796, "step": 24030 }, { "epoch": 8.340550472496489e-14, "grad_norm": 0.550979495048523, "learning_rate": 0.00010299578805141506, "loss": 2.4173, "step": 24040 }, { "epoch": 8.344019919448442e-14, "grad_norm": 0.5416810512542725, "learning_rate": 0.00010297803977729073, "loss": 2.4582, "step": 24050 }, { "epoch": 8.347489366400396e-14, "grad_norm": 0.5426979660987854, "learning_rate": 0.00010296028609029652, "loss": 2.4573, "step": 24060 }, { "epoch": 8.350958813352349e-14, "grad_norm": 0.5547951459884644, "learning_rate": 0.00010294252699324008, "loss": 2.4107, "step": 24070 }, { "epoch": 8.354428260304303e-14, "grad_norm": 0.5451095700263977, "learning_rate": 0.0001029247624889298, "loss": 2.4181, "step": 24080 }, { "epoch": 8.357897707256257e-14, "grad_norm": 0.5537813305854797, "learning_rate": 0.00010290699258017504, "loss": 2.4807, "step": 24090 }, { "epoch": 8.36136715420821e-14, "grad_norm": 0.5572213530540466, "learning_rate": 0.00010288921726978591, "loss": 2.4407, "step": 24100 }, { "epoch": 8.364836601160164e-14, "grad_norm": 0.5413472652435303, "learning_rate": 0.00010287143656057344, "loss": 2.4769, "step": 24110 }, { "epoch": 8.368306048112117e-14, "grad_norm": 0.5270729660987854, "learning_rate": 0.00010285365045534951, "loss": 2.4237, "step": 24120 }, { "epoch": 8.371775495064071e-14, "grad_norm": 0.5397841334342957, "learning_rate": 0.00010283585895692683, "loss": 2.4293, "step": 24130 }, { "epoch": 8.375244942016025e-14, "grad_norm": 0.5565266609191895, "learning_rate": 0.00010281806206811893, "loss": 2.4468, "step": 24140 }, { "epoch": 8.378714388967978e-14, "grad_norm": 0.5353559851646423, "learning_rate": 0.0001028002597917403, "loss": 2.4471, "step": 24150 }, { "epoch": 8.382183835919932e-14, "grad_norm": 0.5436949729919434, "learning_rate": 0.0001027824521306062, "loss": 2.4208, "step": 24160 }, { "epoch": 8.385653282871885e-14, "grad_norm": 0.5640153884887695, "learning_rate": 0.00010276463908753274, "loss": 2.4413, "step": 24170 }, { "epoch": 8.389122729823839e-14, "grad_norm": 0.5579166412353516, "learning_rate": 0.0001027468206653369, "loss": 2.4448, "step": 24180 }, { "epoch": 8.392592176775793e-14, "grad_norm": 0.5268977284431458, "learning_rate": 0.00010272899686683652, "loss": 2.422, "step": 24190 }, { "epoch": 8.396061623727746e-14, "grad_norm": 0.5350064635276794, "learning_rate": 0.0001027111676948503, "loss": 2.4515, "step": 24200 }, { "epoch": 8.3995310706797e-14, "grad_norm": 0.5973467826843262, "learning_rate": 0.00010269333315219772, "loss": 2.4555, "step": 24210 }, { "epoch": 8.403000517631654e-14, "grad_norm": 0.5740598440170288, "learning_rate": 0.00010267549324169921, "loss": 2.4465, "step": 24220 }, { "epoch": 8.406469964583607e-14, "grad_norm": 0.5835289359092712, "learning_rate": 0.00010265764796617598, "loss": 2.4501, "step": 24230 }, { "epoch": 8.409939411535561e-14, "grad_norm": 0.5345718860626221, "learning_rate": 0.00010263979732845008, "loss": 2.4052, "step": 24240 }, { "epoch": 8.413408858487514e-14, "grad_norm": 0.528450608253479, "learning_rate": 0.00010262194133134448, "loss": 2.4341, "step": 24250 }, { "epoch": 8.416878305439468e-14, "grad_norm": 0.5462702512741089, "learning_rate": 0.0001026040799776829, "loss": 2.4485, "step": 24260 }, { "epoch": 8.420347752391422e-14, "grad_norm": 0.554424524307251, "learning_rate": 0.00010258621327029003, "loss": 2.439, "step": 24270 }, { "epoch": 8.423817199343375e-14, "grad_norm": 0.5529745817184448, "learning_rate": 0.00010256834121199128, "loss": 2.4391, "step": 24280 }, { "epoch": 8.427286646295329e-14, "grad_norm": 0.546566903591156, "learning_rate": 0.00010255046380561297, "loss": 2.4299, "step": 24290 }, { "epoch": 8.430756093247282e-14, "grad_norm": 0.5375308394432068, "learning_rate": 0.00010253258105398226, "loss": 2.4428, "step": 24300 }, { "epoch": 8.434225540199236e-14, "grad_norm": 0.5460305213928223, "learning_rate": 0.00010251469295992715, "loss": 2.4481, "step": 24310 }, { "epoch": 8.43769498715119e-14, "grad_norm": 0.5621047616004944, "learning_rate": 0.00010249679952627654, "loss": 2.442, "step": 24320 }, { "epoch": 8.441164434103143e-14, "grad_norm": 0.5566433668136597, "learning_rate": 0.00010247890075586005, "loss": 2.446, "step": 24330 }, { "epoch": 8.444633881055097e-14, "grad_norm": 0.6320019960403442, "learning_rate": 0.00010246099665150824, "loss": 2.4132, "step": 24340 }, { "epoch": 8.44810332800705e-14, "grad_norm": 0.5224804878234863, "learning_rate": 0.00010244308721605252, "loss": 2.407, "step": 24350 }, { "epoch": 8.451572774959004e-14, "grad_norm": 0.5632887482643127, "learning_rate": 0.00010242517245232507, "loss": 2.4232, "step": 24360 }, { "epoch": 8.455042221910958e-14, "grad_norm": 0.5807772874832153, "learning_rate": 0.00010240725236315897, "loss": 2.3992, "step": 24370 }, { "epoch": 8.458511668862911e-14, "grad_norm": 0.5366011261940002, "learning_rate": 0.00010238932695138813, "loss": 2.4764, "step": 24380 }, { "epoch": 8.461981115814865e-14, "grad_norm": 0.5551806092262268, "learning_rate": 0.00010237139621984732, "loss": 2.4427, "step": 24390 }, { "epoch": 8.465450562766819e-14, "grad_norm": 0.5381510853767395, "learning_rate": 0.0001023534601713721, "loss": 2.4461, "step": 24400 }, { "epoch": 8.468920009718772e-14, "grad_norm": 0.5696653723716736, "learning_rate": 0.00010233551880879892, "loss": 2.4476, "step": 24410 }, { "epoch": 8.472389456670726e-14, "grad_norm": 0.5415769219398499, "learning_rate": 0.00010231757213496506, "loss": 2.4443, "step": 24420 }, { "epoch": 8.47585890362268e-14, "grad_norm": 0.5730569958686829, "learning_rate": 0.00010229962015270861, "loss": 2.4635, "step": 24430 }, { "epoch": 8.479328350574633e-14, "grad_norm": 0.5357567071914673, "learning_rate": 0.00010228166286486852, "loss": 2.4116, "step": 24440 }, { "epoch": 8.482797797526587e-14, "grad_norm": 0.5372189879417419, "learning_rate": 0.00010226370027428458, "loss": 2.4587, "step": 24450 }, { "epoch": 8.48626724447854e-14, "grad_norm": 0.5638733506202698, "learning_rate": 0.00010224573238379746, "loss": 2.4583, "step": 24460 }, { "epoch": 8.489736691430494e-14, "grad_norm": 0.5191668272018433, "learning_rate": 0.00010222775919624859, "loss": 2.424, "step": 24470 }, { "epoch": 8.493206138382448e-14, "grad_norm": 0.5942763090133667, "learning_rate": 0.00010220978071448027, "loss": 2.4164, "step": 24480 }, { "epoch": 8.496675585334401e-14, "grad_norm": 0.6608238816261292, "learning_rate": 0.00010219179694133565, "loss": 2.4406, "step": 24490 }, { "epoch": 8.500145032286355e-14, "grad_norm": 0.5707746744155884, "learning_rate": 0.00010217380787965872, "loss": 2.4721, "step": 24500 }, { "epoch": 8.503614479238308e-14, "grad_norm": 0.5460811257362366, "learning_rate": 0.00010215581353229428, "loss": 2.4335, "step": 24510 }, { "epoch": 8.507083926190262e-14, "grad_norm": 0.5624075531959534, "learning_rate": 0.000102137813902088, "loss": 2.4291, "step": 24520 }, { "epoch": 8.510553373142216e-14, "grad_norm": 0.6187969446182251, "learning_rate": 0.00010211980899188636, "loss": 2.4369, "step": 24530 }, { "epoch": 8.514022820094169e-14, "grad_norm": 0.5811499357223511, "learning_rate": 0.00010210179880453665, "loss": 2.4196, "step": 24540 }, { "epoch": 8.517492267046123e-14, "grad_norm": 0.5846900343894958, "learning_rate": 0.00010208378334288709, "loss": 2.4249, "step": 24550 }, { "epoch": 8.520961713998076e-14, "grad_norm": 0.5898990035057068, "learning_rate": 0.00010206576260978662, "loss": 2.457, "step": 24560 }, { "epoch": 8.52443116095003e-14, "grad_norm": 0.5401241779327393, "learning_rate": 0.00010204773660808507, "loss": 2.4254, "step": 24570 }, { "epoch": 8.527900607901984e-14, "grad_norm": 0.5377113223075867, "learning_rate": 0.0001020297053406331, "loss": 2.4285, "step": 24580 }, { "epoch": 8.531370054853937e-14, "grad_norm": 0.5494766235351562, "learning_rate": 0.00010201166881028222, "loss": 2.4377, "step": 24590 }, { "epoch": 8.534839501805891e-14, "grad_norm": 0.537217915058136, "learning_rate": 0.00010199362701988474, "loss": 2.4199, "step": 24600 }, { "epoch": 8.538308948757845e-14, "grad_norm": 0.5634428262710571, "learning_rate": 0.00010197557997229378, "loss": 2.455, "step": 24610 }, { "epoch": 8.541778395709798e-14, "grad_norm": 0.5590505599975586, "learning_rate": 0.00010195752767036338, "loss": 2.4335, "step": 24620 }, { "epoch": 8.545247842661752e-14, "grad_norm": 0.5488409996032715, "learning_rate": 0.00010193947011694831, "loss": 2.4252, "step": 24630 }, { "epoch": 8.548717289613705e-14, "grad_norm": 0.5998058319091797, "learning_rate": 0.00010192140731490426, "loss": 2.4428, "step": 24640 }, { "epoch": 8.552186736565659e-14, "grad_norm": 0.5596379637718201, "learning_rate": 0.0001019033392670877, "loss": 2.4765, "step": 24650 }, { "epoch": 8.555656183517613e-14, "grad_norm": 0.5710581541061401, "learning_rate": 0.0001018852659763559, "loss": 2.4163, "step": 24660 }, { "epoch": 8.559125630469566e-14, "grad_norm": 0.5784903764724731, "learning_rate": 0.000101867187445567, "loss": 2.4416, "step": 24670 }, { "epoch": 8.56259507742152e-14, "grad_norm": 0.5416696071624756, "learning_rate": 0.00010184910367758, "loss": 2.4366, "step": 24680 }, { "epoch": 8.566064524373473e-14, "grad_norm": 0.5605747699737549, "learning_rate": 0.00010183101467525466, "loss": 2.4487, "step": 24690 }, { "epoch": 8.569533971325427e-14, "grad_norm": 0.5907326340675354, "learning_rate": 0.00010181292044145162, "loss": 2.3744, "step": 24700 }, { "epoch": 8.573003418277381e-14, "grad_norm": 0.5764821767807007, "learning_rate": 0.00010179482097903232, "loss": 2.4426, "step": 24710 }, { "epoch": 8.576472865229334e-14, "grad_norm": 0.5645548701286316, "learning_rate": 0.00010177671629085902, "loss": 2.4185, "step": 24720 }, { "epoch": 8.579942312181288e-14, "grad_norm": 0.5438414216041565, "learning_rate": 0.00010175860637979486, "loss": 2.4371, "step": 24730 }, { "epoch": 8.583411759133242e-14, "grad_norm": 0.5295098423957825, "learning_rate": 0.00010174049124870371, "loss": 2.4521, "step": 24740 }, { "epoch": 8.586881206085195e-14, "grad_norm": 0.5535319447517395, "learning_rate": 0.0001017223709004504, "loss": 2.4741, "step": 24750 }, { "epoch": 8.590350653037149e-14, "grad_norm": 0.5624488592147827, "learning_rate": 0.00010170424533790041, "loss": 2.4271, "step": 24760 }, { "epoch": 8.593820099989102e-14, "grad_norm": 0.5461381673812866, "learning_rate": 0.00010168611456392023, "loss": 2.3982, "step": 24770 }, { "epoch": 8.597289546941056e-14, "grad_norm": 0.5630932450294495, "learning_rate": 0.00010166797858137702, "loss": 2.4132, "step": 24780 }, { "epoch": 8.60075899389301e-14, "grad_norm": 0.5556327700614929, "learning_rate": 0.00010164983739313887, "loss": 2.4328, "step": 24790 }, { "epoch": 8.604228440844963e-14, "grad_norm": 0.5386098623275757, "learning_rate": 0.00010163169100207464, "loss": 2.4432, "step": 24800 }, { "epoch": 8.607697887796917e-14, "grad_norm": 0.5948113203048706, "learning_rate": 0.00010161353941105405, "loss": 2.4475, "step": 24810 }, { "epoch": 8.61116733474887e-14, "grad_norm": 0.5606451630592346, "learning_rate": 0.00010159538262294756, "loss": 2.4357, "step": 24820 }, { "epoch": 8.614636781700824e-14, "grad_norm": 0.5348959565162659, "learning_rate": 0.00010157722064062652, "loss": 2.4394, "step": 24830 }, { "epoch": 8.618106228652778e-14, "grad_norm": 0.5467357635498047, "learning_rate": 0.00010155905346696316, "loss": 2.4408, "step": 24840 }, { "epoch": 8.621575675604731e-14, "grad_norm": 0.5643408894538879, "learning_rate": 0.00010154088110483041, "loss": 2.439, "step": 24850 }, { "epoch": 8.625045122556685e-14, "grad_norm": 0.5665498375892639, "learning_rate": 0.00010152270355710207, "loss": 2.453, "step": 24860 }, { "epoch": 8.628514569508638e-14, "grad_norm": 0.5647155046463013, "learning_rate": 0.00010150452082665276, "loss": 2.4247, "step": 24870 }, { "epoch": 8.631984016460592e-14, "grad_norm": 0.5456105470657349, "learning_rate": 0.00010148633291635793, "loss": 2.3948, "step": 24880 }, { "epoch": 8.635453463412546e-14, "grad_norm": 0.5434112548828125, "learning_rate": 0.00010146813982909387, "loss": 2.4456, "step": 24890 }, { "epoch": 8.638922910364499e-14, "grad_norm": 0.5479350090026855, "learning_rate": 0.00010144994156773763, "loss": 2.4442, "step": 24900 }, { "epoch": 8.642392357316453e-14, "grad_norm": 0.5361886620521545, "learning_rate": 0.00010143173813516712, "loss": 2.4468, "step": 24910 }, { "epoch": 8.645861804268407e-14, "grad_norm": 0.5714845061302185, "learning_rate": 0.00010141352953426104, "loss": 2.3641, "step": 24920 }, { "epoch": 8.64933125122036e-14, "grad_norm": 0.5839753150939941, "learning_rate": 0.00010139531576789894, "loss": 2.415, "step": 24930 }, { "epoch": 8.652800698172314e-14, "grad_norm": 0.5667299032211304, "learning_rate": 0.00010137709683896117, "loss": 2.4217, "step": 24940 }, { "epoch": 8.656270145124267e-14, "grad_norm": 0.5767189264297485, "learning_rate": 0.00010135887275032888, "loss": 2.4283, "step": 24950 }, { "epoch": 8.659739592076221e-14, "grad_norm": 0.5576906800270081, "learning_rate": 0.00010134064350488407, "loss": 2.4343, "step": 24960 }, { "epoch": 8.663209039028175e-14, "grad_norm": 0.5479781627655029, "learning_rate": 0.00010132240910550951, "loss": 2.469, "step": 24970 }, { "epoch": 8.666678485980128e-14, "grad_norm": 0.5552141666412354, "learning_rate": 0.00010130416955508886, "loss": 2.4516, "step": 24980 }, { "epoch": 8.670147932932082e-14, "grad_norm": 0.5666444897651672, "learning_rate": 0.00010128592485650651, "loss": 2.4391, "step": 24990 }, { "epoch": 8.673617379884035e-14, "grad_norm": 0.5287792682647705, "learning_rate": 0.00010126767501264772, "loss": 2.4863, "step": 25000 }, { "epoch": 8.677086826835989e-14, "grad_norm": 0.6180870532989502, "learning_rate": 0.00010124942002639854, "loss": 2.4257, "step": 25010 }, { "epoch": 8.680556273787943e-14, "grad_norm": 0.550003707408905, "learning_rate": 0.00010123115990064586, "loss": 2.4289, "step": 25020 }, { "epoch": 8.684025720739896e-14, "grad_norm": 0.5443030595779419, "learning_rate": 0.00010121289463827733, "loss": 2.4508, "step": 25030 }, { "epoch": 8.68749516769185e-14, "grad_norm": 0.5467247366905212, "learning_rate": 0.00010119462424218142, "loss": 2.414, "step": 25040 }, { "epoch": 8.690964614643804e-14, "grad_norm": 0.5823155641555786, "learning_rate": 0.0001011763487152475, "loss": 2.4293, "step": 25050 }, { "epoch": 8.694434061595757e-14, "grad_norm": 0.5574851036071777, "learning_rate": 0.00010115806806036566, "loss": 2.4554, "step": 25060 }, { "epoch": 8.697903508547711e-14, "grad_norm": 0.532813549041748, "learning_rate": 0.00010113978228042683, "loss": 2.4149, "step": 25070 }, { "epoch": 8.701372955499664e-14, "grad_norm": 0.5546402335166931, "learning_rate": 0.00010112149137832274, "loss": 2.4482, "step": 25080 }, { "epoch": 8.704842402451618e-14, "grad_norm": 0.5316426157951355, "learning_rate": 0.00010110319535694594, "loss": 2.4307, "step": 25090 }, { "epoch": 8.708311849403572e-14, "grad_norm": 0.536320686340332, "learning_rate": 0.0001010848942191898, "loss": 2.4359, "step": 25100 }, { "epoch": 8.711781296355525e-14, "grad_norm": 0.5448563694953918, "learning_rate": 0.00010106658796794848, "loss": 2.4705, "step": 25110 }, { "epoch": 8.715250743307479e-14, "grad_norm": 0.5408859252929688, "learning_rate": 0.00010104827660611694, "loss": 2.4444, "step": 25120 }, { "epoch": 8.718720190259432e-14, "grad_norm": 0.5918940305709839, "learning_rate": 0.000101029960136591, "loss": 2.416, "step": 25130 }, { "epoch": 8.722189637211386e-14, "grad_norm": 0.5435788035392761, "learning_rate": 0.00010101163856226723, "loss": 2.46, "step": 25140 }, { "epoch": 8.72565908416334e-14, "grad_norm": 0.54753577709198, "learning_rate": 0.00010099331188604302, "loss": 2.409, "step": 25150 }, { "epoch": 8.729128531115293e-14, "grad_norm": 0.5487062335014343, "learning_rate": 0.0001009749801108166, "loss": 2.4622, "step": 25160 }, { "epoch": 8.732597978067247e-14, "grad_norm": 0.5764562487602234, "learning_rate": 0.00010095664323948696, "loss": 2.4309, "step": 25170 }, { "epoch": 8.7360674250192e-14, "grad_norm": 0.5641457438468933, "learning_rate": 0.00010093830127495393, "loss": 2.4372, "step": 25180 }, { "epoch": 8.739536871971154e-14, "grad_norm": 0.5365095734596252, "learning_rate": 0.00010091995422011814, "loss": 2.4154, "step": 25190 }, { "epoch": 8.743006318923108e-14, "grad_norm": 0.5307635068893433, "learning_rate": 0.000100901602077881, "loss": 2.4348, "step": 25200 }, { "epoch": 8.746475765875061e-14, "grad_norm": 0.5511671304702759, "learning_rate": 0.00010088324485114476, "loss": 2.4521, "step": 25210 }, { "epoch": 8.749945212827015e-14, "grad_norm": 0.5285272002220154, "learning_rate": 0.00010086488254281247, "loss": 2.3807, "step": 25220 }, { "epoch": 8.753414659778969e-14, "grad_norm": 0.565654456615448, "learning_rate": 0.0001008465151557879, "loss": 2.4381, "step": 25230 }, { "epoch": 8.756884106730922e-14, "grad_norm": 0.5532052516937256, "learning_rate": 0.0001008281426929758, "loss": 2.4057, "step": 25240 }, { "epoch": 8.760353553682876e-14, "grad_norm": 0.5474978089332581, "learning_rate": 0.00010080976515728155, "loss": 2.4395, "step": 25250 }, { "epoch": 8.76382300063483e-14, "grad_norm": 0.5433254241943359, "learning_rate": 0.00010079138255161139, "loss": 2.4276, "step": 25260 }, { "epoch": 8.767292447586783e-14, "grad_norm": 0.5327993631362915, "learning_rate": 0.00010077299487887243, "loss": 2.4151, "step": 25270 }, { "epoch": 8.770761894538737e-14, "grad_norm": 0.5664665102958679, "learning_rate": 0.00010075460214197246, "loss": 2.4641, "step": 25280 }, { "epoch": 8.77423134149069e-14, "grad_norm": 0.5729371905326843, "learning_rate": 0.00010073620434382016, "loss": 2.4483, "step": 25290 }, { "epoch": 8.777700788442644e-14, "grad_norm": 0.5510352849960327, "learning_rate": 0.000100717801487325, "loss": 2.4316, "step": 25300 }, { "epoch": 8.781170235394598e-14, "grad_norm": 0.5792786478996277, "learning_rate": 0.0001006993935753972, "loss": 2.417, "step": 25310 }, { "epoch": 8.784639682346551e-14, "grad_norm": 0.5281094908714294, "learning_rate": 0.00010068098061094782, "loss": 2.4347, "step": 25320 }, { "epoch": 8.788109129298505e-14, "grad_norm": 0.5550370812416077, "learning_rate": 0.00010066256259688873, "loss": 2.4688, "step": 25330 }, { "epoch": 8.791578576250458e-14, "grad_norm": 0.5606416463851929, "learning_rate": 0.00010064413953613259, "loss": 2.4407, "step": 25340 }, { "epoch": 8.795048023202412e-14, "grad_norm": 0.5392405986785889, "learning_rate": 0.0001006257114315928, "loss": 2.4184, "step": 25350 }, { "epoch": 8.798517470154366e-14, "grad_norm": 0.5448417663574219, "learning_rate": 0.00010060727828618364, "loss": 2.4461, "step": 25360 }, { "epoch": 8.801986917106319e-14, "grad_norm": 0.5320744514465332, "learning_rate": 0.00010058884010282018, "loss": 2.4398, "step": 25370 }, { "epoch": 8.805456364058273e-14, "grad_norm": 0.5451936721801758, "learning_rate": 0.00010057039688441819, "loss": 2.4288, "step": 25380 }, { "epoch": 8.808925811010226e-14, "grad_norm": 0.5606884956359863, "learning_rate": 0.00010055194863389437, "loss": 2.4164, "step": 25390 }, { "epoch": 8.81239525796218e-14, "grad_norm": 0.5340784192085266, "learning_rate": 0.0001005334953541661, "loss": 2.4315, "step": 25400 }, { "epoch": 8.815864704914134e-14, "grad_norm": 0.5753525495529175, "learning_rate": 0.00010051503704815164, "loss": 2.411, "step": 25410 }, { "epoch": 8.819334151866087e-14, "grad_norm": 0.5375596284866333, "learning_rate": 0.00010049657371877002, "loss": 2.4086, "step": 25420 }, { "epoch": 8.822803598818041e-14, "grad_norm": 0.5719355940818787, "learning_rate": 0.00010047810536894104, "loss": 2.4479, "step": 25430 }, { "epoch": 8.826273045769994e-14, "grad_norm": 0.5312706232070923, "learning_rate": 0.00010045963200158531, "loss": 2.4249, "step": 25440 }, { "epoch": 8.829742492721948e-14, "grad_norm": 0.5433868765830994, "learning_rate": 0.00010044115361962423, "loss": 2.4444, "step": 25450 }, { "epoch": 8.833211939673902e-14, "grad_norm": 0.5680206418037415, "learning_rate": 0.00010042267022598002, "loss": 2.4685, "step": 25460 }, { "epoch": 8.836681386625855e-14, "grad_norm": 0.5691958069801331, "learning_rate": 0.00010040418182357564, "loss": 2.4382, "step": 25470 }, { "epoch": 8.840150833577809e-14, "grad_norm": 0.5626378655433655, "learning_rate": 0.0001003856884153349, "loss": 2.4163, "step": 25480 }, { "epoch": 8.843620280529763e-14, "grad_norm": 0.6258940696716309, "learning_rate": 0.00010036719000418235, "loss": 2.4299, "step": 25490 }, { "epoch": 8.847089727481716e-14, "grad_norm": 0.5369779467582703, "learning_rate": 0.00010034868659304337, "loss": 2.4407, "step": 25500 }, { "epoch": 8.85055917443367e-14, "grad_norm": 0.5413534045219421, "learning_rate": 0.0001003301781848441, "loss": 2.4182, "step": 25510 }, { "epoch": 8.854028621385623e-14, "grad_norm": 0.5495430827140808, "learning_rate": 0.00010031166478251149, "loss": 2.4254, "step": 25520 }, { "epoch": 8.857498068337577e-14, "grad_norm": 0.5528014898300171, "learning_rate": 0.00010029314638897329, "loss": 2.4499, "step": 25530 }, { "epoch": 8.860967515289531e-14, "grad_norm": 0.5822210311889648, "learning_rate": 0.00010027462300715801, "loss": 2.4143, "step": 25540 }, { "epoch": 8.864436962241484e-14, "grad_norm": 0.6025453209877014, "learning_rate": 0.00010025609463999496, "loss": 2.4614, "step": 25550 }, { "epoch": 8.867906409193438e-14, "grad_norm": 0.5334014296531677, "learning_rate": 0.00010023756129041426, "loss": 2.428, "step": 25560 }, { "epoch": 8.871375856145391e-14, "grad_norm": 0.5700635313987732, "learning_rate": 0.00010021902296134678, "loss": 2.4212, "step": 25570 }, { "epoch": 8.874845303097345e-14, "grad_norm": 0.5358031392097473, "learning_rate": 0.00010020047965572422, "loss": 2.4291, "step": 25580 }, { "epoch": 8.878314750049299e-14, "grad_norm": 0.5319196581840515, "learning_rate": 0.00010018193137647901, "loss": 2.4302, "step": 25590 }, { "epoch": 8.881784197001252e-14, "grad_norm": 0.5512889623641968, "learning_rate": 0.00010016337812654444, "loss": 2.4103, "step": 25600 }, { "epoch": 8.885253643953206e-14, "grad_norm": 0.5555306077003479, "learning_rate": 0.00010014481990885452, "loss": 2.4399, "step": 25610 }, { "epoch": 8.88872309090516e-14, "grad_norm": 0.5362983345985413, "learning_rate": 0.00010012625672634408, "loss": 2.4206, "step": 25620 }, { "epoch": 8.892192537857113e-14, "grad_norm": 0.5617801547050476, "learning_rate": 0.00010010768858194873, "loss": 2.4503, "step": 25630 }, { "epoch": 8.895661984809067e-14, "grad_norm": 0.5402596592903137, "learning_rate": 0.00010008911547860487, "loss": 2.4399, "step": 25640 }, { "epoch": 8.89913143176102e-14, "grad_norm": 0.5710116028785706, "learning_rate": 0.00010007053741924964, "loss": 2.4484, "step": 25650 }, { "epoch": 8.902600878712974e-14, "grad_norm": 0.5573415160179138, "learning_rate": 0.00010005195440682104, "loss": 2.4137, "step": 25660 }, { "epoch": 8.906070325664928e-14, "grad_norm": 0.5415785908699036, "learning_rate": 0.0001000333664442578, "loss": 2.4382, "step": 25670 }, { "epoch": 8.909539772616881e-14, "grad_norm": 0.5322752594947815, "learning_rate": 0.00010001477353449943, "loss": 2.4494, "step": 25680 }, { "epoch": 8.913009219568835e-14, "grad_norm": 0.5611672401428223, "learning_rate": 9.999617568048627e-05, "loss": 2.3897, "step": 25690 }, { "epoch": 8.916478666520788e-14, "grad_norm": 0.5721877813339233, "learning_rate": 9.99775728851594e-05, "loss": 2.4193, "step": 25700 }, { "epoch": 8.919948113472742e-14, "grad_norm": 0.5798177719116211, "learning_rate": 9.995896515146069e-05, "loss": 2.4172, "step": 25710 }, { "epoch": 8.923417560424696e-14, "grad_norm": 0.5488571524620056, "learning_rate": 9.994035248233277e-05, "loss": 2.4065, "step": 25720 }, { "epoch": 8.926887007376649e-14, "grad_norm": 0.5291709899902344, "learning_rate": 9.99217348807191e-05, "loss": 2.4064, "step": 25730 }, { "epoch": 8.930356454328603e-14, "grad_norm": 0.5631994009017944, "learning_rate": 9.990311234956389e-05, "loss": 2.4092, "step": 25740 }, { "epoch": 8.933825901280557e-14, "grad_norm": 0.5543108582496643, "learning_rate": 9.988448489181214e-05, "loss": 2.4244, "step": 25750 }, { "epoch": 8.93729534823251e-14, "grad_norm": 0.5328806042671204, "learning_rate": 9.986585251040959e-05, "loss": 2.4462, "step": 25760 }, { "epoch": 8.940764795184464e-14, "grad_norm": 0.5457202792167664, "learning_rate": 9.984721520830283e-05, "loss": 2.4395, "step": 25770 }, { "epoch": 8.944234242136417e-14, "grad_norm": 0.5546552538871765, "learning_rate": 9.982857298843919e-05, "loss": 2.4294, "step": 25780 }, { "epoch": 8.947703689088371e-14, "grad_norm": 0.553859531879425, "learning_rate": 9.980992585376672e-05, "loss": 2.4385, "step": 25790 }, { "epoch": 8.951173136040325e-14, "grad_norm": 0.5526059865951538, "learning_rate": 9.979127380723435e-05, "loss": 2.425, "step": 25800 }, { "epoch": 8.954642582992278e-14, "grad_norm": 0.5286966562271118, "learning_rate": 9.977261685179174e-05, "loss": 2.4405, "step": 25810 }, { "epoch": 8.958112029944232e-14, "grad_norm": 0.5638823509216309, "learning_rate": 9.975395499038932e-05, "loss": 2.3937, "step": 25820 }, { "epoch": 8.961581476896185e-14, "grad_norm": 0.5429010391235352, "learning_rate": 9.973528822597831e-05, "loss": 2.4462, "step": 25830 }, { "epoch": 8.965050923848139e-14, "grad_norm": 0.5960492491722107, "learning_rate": 9.971661656151069e-05, "loss": 2.3762, "step": 25840 }, { "epoch": 8.968520370800093e-14, "grad_norm": 0.5770930647850037, "learning_rate": 9.969793999993919e-05, "loss": 2.4526, "step": 25850 }, { "epoch": 8.971989817752046e-14, "grad_norm": 0.5456598401069641, "learning_rate": 9.96792585442174e-05, "loss": 2.4497, "step": 25860 }, { "epoch": 8.975459264704e-14, "grad_norm": 0.5373523235321045, "learning_rate": 9.966057219729962e-05, "loss": 2.4532, "step": 25870 }, { "epoch": 8.978928711655954e-14, "grad_norm": 0.5505486130714417, "learning_rate": 9.964188096214093e-05, "loss": 2.4302, "step": 25880 }, { "epoch": 8.982398158607907e-14, "grad_norm": 0.5573275685310364, "learning_rate": 9.962318484169716e-05, "loss": 2.4467, "step": 25890 }, { "epoch": 8.985867605559861e-14, "grad_norm": 0.5524411797523499, "learning_rate": 9.9604483838925e-05, "loss": 2.4553, "step": 25900 }, { "epoch": 8.989337052511814e-14, "grad_norm": 0.5976356267929077, "learning_rate": 9.95857779567818e-05, "loss": 2.4399, "step": 25910 }, { "epoch": 8.992806499463768e-14, "grad_norm": 0.5468243360519409, "learning_rate": 9.956706719822576e-05, "loss": 2.4335, "step": 25920 }, { "epoch": 8.996275946415722e-14, "grad_norm": 0.5408995747566223, "learning_rate": 9.954835156621583e-05, "loss": 2.4045, "step": 25930 }, { "epoch": 8.999745393367675e-14, "grad_norm": 0.5467814207077026, "learning_rate": 9.952963106371168e-05, "loss": 2.4415, "step": 25940 }, { "epoch": 9.003214840319629e-14, "grad_norm": 0.5591344833374023, "learning_rate": 9.951090569367384e-05, "loss": 2.4186, "step": 25950 }, { "epoch": 9.006684287271582e-14, "grad_norm": 0.5511786937713623, "learning_rate": 9.949217545906357e-05, "loss": 2.4102, "step": 25960 }, { "epoch": 9.010153734223536e-14, "grad_norm": 0.5627973079681396, "learning_rate": 9.947344036284289e-05, "loss": 2.4147, "step": 25970 }, { "epoch": 9.01362318117549e-14, "grad_norm": 0.552569568157196, "learning_rate": 9.945470040797455e-05, "loss": 2.405, "step": 25980 }, { "epoch": 9.017092628127443e-14, "grad_norm": 0.5722510814666748, "learning_rate": 9.943595559742218e-05, "loss": 2.3787, "step": 25990 }, { "epoch": 9.020562075079397e-14, "grad_norm": 0.5481232404708862, "learning_rate": 9.941720593415005e-05, "loss": 2.4107, "step": 26000 }, { "epoch": 9.02403152203135e-14, "grad_norm": 0.5609031915664673, "learning_rate": 9.93984514211233e-05, "loss": 2.4337, "step": 26010 }, { "epoch": 9.027500968983304e-14, "grad_norm": 0.5653215050697327, "learning_rate": 9.937969206130778e-05, "loss": 2.4414, "step": 26020 }, { "epoch": 9.030970415935258e-14, "grad_norm": 0.5843251347541809, "learning_rate": 9.936092785767014e-05, "loss": 2.3947, "step": 26030 }, { "epoch": 9.034439862887211e-14, "grad_norm": 0.5652306079864502, "learning_rate": 9.934215881317776e-05, "loss": 2.423, "step": 26040 }, { "epoch": 9.037909309839165e-14, "grad_norm": 0.56053626537323, "learning_rate": 9.93233849307988e-05, "loss": 2.4234, "step": 26050 }, { "epoch": 9.041378756791119e-14, "grad_norm": 0.5990728139877319, "learning_rate": 9.930460621350216e-05, "loss": 2.4302, "step": 26060 }, { "epoch": 9.044848203743072e-14, "grad_norm": 0.580033004283905, "learning_rate": 9.92858226642576e-05, "loss": 2.444, "step": 26070 }, { "epoch": 9.048317650695026e-14, "grad_norm": 0.526780903339386, "learning_rate": 9.926703428603551e-05, "loss": 2.4422, "step": 26080 }, { "epoch": 9.05178709764698e-14, "grad_norm": 0.5455052256584167, "learning_rate": 9.924824108180714e-05, "loss": 2.4303, "step": 26090 }, { "epoch": 9.055256544598933e-14, "grad_norm": 0.5437169075012207, "learning_rate": 9.92294430545445e-05, "loss": 2.4055, "step": 26100 }, { "epoch": 9.058725991550887e-14, "grad_norm": 0.5815073847770691, "learning_rate": 9.92106402072203e-05, "loss": 2.3876, "step": 26110 }, { "epoch": 9.06219543850284e-14, "grad_norm": 0.548892080783844, "learning_rate": 9.919183254280806e-05, "loss": 2.4073, "step": 26120 }, { "epoch": 9.065664885454794e-14, "grad_norm": 0.5694777965545654, "learning_rate": 9.917302006428204e-05, "loss": 2.4051, "step": 26130 }, { "epoch": 9.069134332406747e-14, "grad_norm": 0.5596986413002014, "learning_rate": 9.915420277461728e-05, "loss": 2.3861, "step": 26140 }, { "epoch": 9.072603779358701e-14, "grad_norm": 0.574597179889679, "learning_rate": 9.913538067678958e-05, "loss": 2.4328, "step": 26150 }, { "epoch": 9.076073226310655e-14, "grad_norm": 0.575454592704773, "learning_rate": 9.911655377377548e-05, "loss": 2.4048, "step": 26160 }, { "epoch": 9.079542673262608e-14, "grad_norm": 0.5565473437309265, "learning_rate": 9.909772206855232e-05, "loss": 2.4217, "step": 26170 }, { "epoch": 9.083012120214562e-14, "grad_norm": 0.5750633478164673, "learning_rate": 9.907888556409813e-05, "loss": 2.3933, "step": 26180 }, { "epoch": 9.086481567166516e-14, "grad_norm": 0.537359893321991, "learning_rate": 9.906004426339176e-05, "loss": 2.4218, "step": 26190 }, { "epoch": 9.089951014118469e-14, "grad_norm": 0.5552623867988586, "learning_rate": 9.90411981694128e-05, "loss": 2.4193, "step": 26200 }, { "epoch": 9.093420461070423e-14, "grad_norm": 0.5522831082344055, "learning_rate": 9.902234728514159e-05, "loss": 2.433, "step": 26210 }, { "epoch": 9.096889908022376e-14, "grad_norm": 0.5997814536094666, "learning_rate": 9.900349161355924e-05, "loss": 2.4235, "step": 26220 }, { "epoch": 9.10035935497433e-14, "grad_norm": 0.5702798366546631, "learning_rate": 9.898463115764763e-05, "loss": 2.4249, "step": 26230 }, { "epoch": 9.103828801926284e-14, "grad_norm": 0.5662654042243958, "learning_rate": 9.896576592038937e-05, "loss": 2.4384, "step": 26240 }, { "epoch": 9.107298248878237e-14, "grad_norm": 0.5421016812324524, "learning_rate": 9.894689590476782e-05, "loss": 2.4454, "step": 26250 }, { "epoch": 9.110767695830191e-14, "grad_norm": 0.5481651425361633, "learning_rate": 9.892802111376713e-05, "loss": 2.4131, "step": 26260 }, { "epoch": 9.114237142782144e-14, "grad_norm": 0.5166508555412292, "learning_rate": 9.890914155037214e-05, "loss": 2.41, "step": 26270 }, { "epoch": 9.117706589734098e-14, "grad_norm": 0.5758945345878601, "learning_rate": 9.889025721756856e-05, "loss": 2.4105, "step": 26280 }, { "epoch": 9.121176036686052e-14, "grad_norm": 0.5371818542480469, "learning_rate": 9.887136811834274e-05, "loss": 2.3896, "step": 26290 }, { "epoch": 9.124645483638005e-14, "grad_norm": 0.5899543166160583, "learning_rate": 9.885247425568184e-05, "loss": 2.4063, "step": 26300 }, { "epoch": 9.128114930589959e-14, "grad_norm": 0.5994194149971008, "learning_rate": 9.883357563257376e-05, "loss": 2.4378, "step": 26310 }, { "epoch": 9.131584377541913e-14, "grad_norm": 0.5312535166740417, "learning_rate": 9.881467225200715e-05, "loss": 2.4059, "step": 26320 }, { "epoch": 9.135053824493866e-14, "grad_norm": 0.543514609336853, "learning_rate": 9.879576411697144e-05, "loss": 2.4348, "step": 26330 }, { "epoch": 9.13852327144582e-14, "grad_norm": 0.5471076369285583, "learning_rate": 9.877685123045675e-05, "loss": 2.4503, "step": 26340 }, { "epoch": 9.141992718397773e-14, "grad_norm": 0.5650702118873596, "learning_rate": 9.875793359545401e-05, "loss": 2.4267, "step": 26350 }, { "epoch": 9.145462165349727e-14, "grad_norm": 0.5305619835853577, "learning_rate": 9.873901121495489e-05, "loss": 2.4101, "step": 26360 }, { "epoch": 9.14893161230168e-14, "grad_norm": 0.5477628111839294, "learning_rate": 9.872008409195178e-05, "loss": 2.4509, "step": 26370 }, { "epoch": 9.152401059253634e-14, "grad_norm": 0.562695324420929, "learning_rate": 9.870115222943788e-05, "loss": 2.42, "step": 26380 }, { "epoch": 9.155870506205588e-14, "grad_norm": 0.5851125717163086, "learning_rate": 9.868221563040706e-05, "loss": 2.4014, "step": 26390 }, { "epoch": 9.159339953157541e-14, "grad_norm": 0.5274739861488342, "learning_rate": 9.8663274297854e-05, "loss": 2.3877, "step": 26400 }, { "epoch": 9.162809400109495e-14, "grad_norm": 0.5716590285301208, "learning_rate": 9.86443282347741e-05, "loss": 2.4618, "step": 26410 }, { "epoch": 9.166278847061449e-14, "grad_norm": 0.5635870099067688, "learning_rate": 9.862537744416354e-05, "loss": 2.3993, "step": 26420 }, { "epoch": 9.169748294013402e-14, "grad_norm": 0.5675215721130371, "learning_rate": 9.860642192901919e-05, "loss": 2.4786, "step": 26430 }, { "epoch": 9.173217740965356e-14, "grad_norm": 0.5450778603553772, "learning_rate": 9.858746169233872e-05, "loss": 2.4081, "step": 26440 }, { "epoch": 9.17668718791731e-14, "grad_norm": 0.5730286836624146, "learning_rate": 9.856849673712053e-05, "loss": 2.393, "step": 26450 }, { "epoch": 9.180156634869263e-14, "grad_norm": 2.4852356910705566, "learning_rate": 9.854952706636376e-05, "loss": 2.4379, "step": 26460 }, { "epoch": 9.183626081821217e-14, "grad_norm": 0.5593804717063904, "learning_rate": 9.85305526830683e-05, "loss": 2.4226, "step": 26470 }, { "epoch": 9.18709552877317e-14, "grad_norm": 0.554023802280426, "learning_rate": 9.851157359023479e-05, "loss": 2.4204, "step": 26480 }, { "epoch": 9.190564975725124e-14, "grad_norm": 0.536842942237854, "learning_rate": 9.84925897908646e-05, "loss": 2.4545, "step": 26490 }, { "epoch": 9.194034422677078e-14, "grad_norm": 0.5502216815948486, "learning_rate": 9.847360128795984e-05, "loss": 2.451, "step": 26500 }, { "epoch": 9.197503869629031e-14, "grad_norm": 0.56390780210495, "learning_rate": 9.845460808452343e-05, "loss": 2.448, "step": 26510 }, { "epoch": 9.200973316580985e-14, "grad_norm": 0.5438216924667358, "learning_rate": 9.843561018355894e-05, "loss": 2.4011, "step": 26520 }, { "epoch": 9.204442763532938e-14, "grad_norm": 0.5475978851318359, "learning_rate": 9.841660758807073e-05, "loss": 2.4379, "step": 26530 }, { "epoch": 9.207912210484892e-14, "grad_norm": 0.5789387226104736, "learning_rate": 9.839760030106392e-05, "loss": 2.4147, "step": 26540 }, { "epoch": 9.211381657436846e-14, "grad_norm": 0.5657266974449158, "learning_rate": 9.837858832554431e-05, "loss": 2.4095, "step": 26550 }, { "epoch": 9.214851104388799e-14, "grad_norm": 0.5809825658798218, "learning_rate": 9.835957166451851e-05, "loss": 2.4145, "step": 26560 }, { "epoch": 9.218320551340753e-14, "grad_norm": 0.5581790208816528, "learning_rate": 9.834055032099382e-05, "loss": 2.4164, "step": 26570 }, { "epoch": 9.221789998292707e-14, "grad_norm": 0.5697193145751953, "learning_rate": 9.832152429797835e-05, "loss": 2.4191, "step": 26580 }, { "epoch": 9.22525944524466e-14, "grad_norm": 0.5555958151817322, "learning_rate": 9.830249359848082e-05, "loss": 2.4203, "step": 26590 }, { "epoch": 9.228728892196614e-14, "grad_norm": 0.561434268951416, "learning_rate": 9.828345822551083e-05, "loss": 2.4439, "step": 26600 }, { "epoch": 9.232198339148567e-14, "grad_norm": 0.5576274394989014, "learning_rate": 9.826441818207867e-05, "loss": 2.4132, "step": 26610 }, { "epoch": 9.235667786100521e-14, "grad_norm": 0.5503467917442322, "learning_rate": 9.824537347119531e-05, "loss": 2.43, "step": 26620 }, { "epoch": 9.239137233052475e-14, "grad_norm": 0.5712782144546509, "learning_rate": 9.822632409587255e-05, "loss": 2.419, "step": 26630 }, { "epoch": 9.242606680004428e-14, "grad_norm": 0.5381444692611694, "learning_rate": 9.820727005912285e-05, "loss": 2.3996, "step": 26640 }, { "epoch": 9.246076126956382e-14, "grad_norm": 0.5467784404754639, "learning_rate": 9.81882113639595e-05, "loss": 2.3802, "step": 26650 }, { "epoch": 9.249545573908335e-14, "grad_norm": 0.5598121881484985, "learning_rate": 9.816914801339638e-05, "loss": 2.4042, "step": 26660 }, { "epoch": 9.253015020860289e-14, "grad_norm": 0.5486695170402527, "learning_rate": 9.815008001044827e-05, "loss": 2.4267, "step": 26670 }, { "epoch": 9.256484467812243e-14, "grad_norm": 0.5757712125778198, "learning_rate": 9.813100735813057e-05, "loss": 2.4037, "step": 26680 }, { "epoch": 9.259953914764196e-14, "grad_norm": 0.5378996133804321, "learning_rate": 9.811193005945948e-05, "loss": 2.4653, "step": 26690 }, { "epoch": 9.26342336171615e-14, "grad_norm": 0.5453724265098572, "learning_rate": 9.809284811745189e-05, "loss": 2.4154, "step": 26700 }, { "epoch": 9.266892808668103e-14, "grad_norm": 0.5468491911888123, "learning_rate": 9.807376153512548e-05, "loss": 2.3972, "step": 26710 }, { "epoch": 9.270362255620057e-14, "grad_norm": 0.5406248569488525, "learning_rate": 9.805467031549859e-05, "loss": 2.364, "step": 26720 }, { "epoch": 9.273831702572011e-14, "grad_norm": 0.6004711389541626, "learning_rate": 9.803557446159035e-05, "loss": 2.4309, "step": 26730 }, { "epoch": 9.277301149523964e-14, "grad_norm": 0.5582687854766846, "learning_rate": 9.801647397642059e-05, "loss": 2.4335, "step": 26740 }, { "epoch": 9.280770596475918e-14, "grad_norm": 0.5522333383560181, "learning_rate": 9.799736886300988e-05, "loss": 2.4442, "step": 26750 }, { "epoch": 9.284240043427872e-14, "grad_norm": 0.5462465882301331, "learning_rate": 9.797825912437959e-05, "loss": 2.4179, "step": 26760 }, { "epoch": 9.287709490379825e-14, "grad_norm": 0.8642601370811462, "learning_rate": 9.79591447635517e-05, "loss": 2.3903, "step": 26770 }, { "epoch": 9.291178937331779e-14, "grad_norm": 0.5527167916297913, "learning_rate": 9.7940025783549e-05, "loss": 2.4143, "step": 26780 }, { "epoch": 9.294648384283732e-14, "grad_norm": 0.5519347190856934, "learning_rate": 9.792090218739499e-05, "loss": 2.4551, "step": 26790 }, { "epoch": 9.298117831235686e-14, "grad_norm": 0.5739725232124329, "learning_rate": 9.790177397811391e-05, "loss": 2.4404, "step": 26800 }, { "epoch": 9.30158727818764e-14, "grad_norm": 0.560754120349884, "learning_rate": 9.78826411587307e-05, "loss": 2.4402, "step": 26810 }, { "epoch": 9.305056725139593e-14, "grad_norm": 0.5507110953330994, "learning_rate": 9.786350373227107e-05, "loss": 2.4192, "step": 26820 }, { "epoch": 9.308526172091547e-14, "grad_norm": 0.5792728662490845, "learning_rate": 9.784436170176144e-05, "loss": 2.4055, "step": 26830 }, { "epoch": 9.3119956190435e-14, "grad_norm": 0.5400843620300293, "learning_rate": 9.782521507022895e-05, "loss": 2.4588, "step": 26840 }, { "epoch": 9.315465065995454e-14, "grad_norm": 0.5813441872596741, "learning_rate": 9.780606384070147e-05, "loss": 2.4468, "step": 26850 }, { "epoch": 9.318934512947408e-14, "grad_norm": 0.5420752167701721, "learning_rate": 9.778690801620765e-05, "loss": 2.3912, "step": 26860 }, { "epoch": 9.322403959899361e-14, "grad_norm": 0.5687354207038879, "learning_rate": 9.776774759977675e-05, "loss": 2.4144, "step": 26870 }, { "epoch": 9.325873406851315e-14, "grad_norm": 0.5424913167953491, "learning_rate": 9.774858259443885e-05, "loss": 2.4504, "step": 26880 }, { "epoch": 9.329342853803269e-14, "grad_norm": 0.5424386262893677, "learning_rate": 9.772941300322474e-05, "loss": 2.4266, "step": 26890 }, { "epoch": 9.332812300755222e-14, "grad_norm": 0.5524683594703674, "learning_rate": 9.771023882916593e-05, "loss": 2.4377, "step": 26900 }, { "epoch": 9.336281747707176e-14, "grad_norm": 0.5146583318710327, "learning_rate": 9.76910600752946e-05, "loss": 2.3925, "step": 26910 }, { "epoch": 9.33975119465913e-14, "grad_norm": 0.5464593768119812, "learning_rate": 9.767187674464378e-05, "loss": 2.4195, "step": 26920 }, { "epoch": 9.343220641611083e-14, "grad_norm": 0.6008455157279968, "learning_rate": 9.765268884024711e-05, "loss": 2.4505, "step": 26930 }, { "epoch": 9.346690088563037e-14, "grad_norm": 0.5709128379821777, "learning_rate": 9.7633496365139e-05, "loss": 2.3872, "step": 26940 }, { "epoch": 9.35015953551499e-14, "grad_norm": 0.5712220072746277, "learning_rate": 9.761429932235457e-05, "loss": 2.4096, "step": 26950 }, { "epoch": 9.353628982466944e-14, "grad_norm": 0.5559030771255493, "learning_rate": 9.759509771492966e-05, "loss": 2.3997, "step": 26960 }, { "epoch": 9.357098429418897e-14, "grad_norm": 0.5614777207374573, "learning_rate": 9.757589154590085e-05, "loss": 2.4471, "step": 26970 }, { "epoch": 9.360567876370851e-14, "grad_norm": 0.5458818078041077, "learning_rate": 9.755668081830542e-05, "loss": 2.4503, "step": 26980 }, { "epoch": 9.364037323322805e-14, "grad_norm": 0.551488995552063, "learning_rate": 9.753746553518139e-05, "loss": 2.4372, "step": 26990 }, { "epoch": 9.367506770274758e-14, "grad_norm": 0.5464067459106445, "learning_rate": 9.75182456995675e-05, "loss": 2.3949, "step": 27000 }, { "epoch": 9.370976217226712e-14, "grad_norm": 0.554745614528656, "learning_rate": 9.749902131450318e-05, "loss": 2.4028, "step": 27010 }, { "epoch": 9.374445664178666e-14, "grad_norm": 0.5856623649597168, "learning_rate": 9.747979238302864e-05, "loss": 2.4008, "step": 27020 }, { "epoch": 9.377915111130619e-14, "grad_norm": 0.5273082852363586, "learning_rate": 9.746055890818471e-05, "loss": 2.3654, "step": 27030 }, { "epoch": 9.381384558082573e-14, "grad_norm": 0.5459267497062683, "learning_rate": 9.744132089301305e-05, "loss": 2.4005, "step": 27040 }, { "epoch": 9.384854005034526e-14, "grad_norm": 0.5448989868164062, "learning_rate": 9.742207834055594e-05, "loss": 2.4032, "step": 27050 }, { "epoch": 9.38832345198648e-14, "grad_norm": 0.5623103380203247, "learning_rate": 9.740283125385648e-05, "loss": 2.4269, "step": 27060 }, { "epoch": 9.391792898938434e-14, "grad_norm": 0.5596795082092285, "learning_rate": 9.738357963595838e-05, "loss": 2.413, "step": 27070 }, { "epoch": 9.395262345890387e-14, "grad_norm": 0.5437533259391785, "learning_rate": 9.736432348990617e-05, "loss": 2.39, "step": 27080 }, { "epoch": 9.398731792842341e-14, "grad_norm": 0.5453222393989563, "learning_rate": 9.734506281874498e-05, "loss": 2.4315, "step": 27090 }, { "epoch": 9.402201239794294e-14, "grad_norm": 0.59007328748703, "learning_rate": 9.732579762552079e-05, "loss": 2.4352, "step": 27100 }, { "epoch": 9.405670686746248e-14, "grad_norm": 0.5492329597473145, "learning_rate": 9.730652791328017e-05, "loss": 2.3914, "step": 27110 }, { "epoch": 9.409140133698202e-14, "grad_norm": 0.5709962248802185, "learning_rate": 9.728725368507045e-05, "loss": 2.4044, "step": 27120 }, { "epoch": 9.412609580650155e-14, "grad_norm": 0.57121741771698, "learning_rate": 9.726797494393975e-05, "loss": 2.4039, "step": 27130 }, { "epoch": 9.416079027602109e-14, "grad_norm": 0.5629534721374512, "learning_rate": 9.724869169293677e-05, "loss": 2.4114, "step": 27140 }, { "epoch": 9.419548474554063e-14, "grad_norm": 0.5348795652389526, "learning_rate": 9.722940393511104e-05, "loss": 2.4032, "step": 27150 }, { "epoch": 9.423017921506016e-14, "grad_norm": 0.5820119380950928, "learning_rate": 9.721011167351271e-05, "loss": 2.4257, "step": 27160 }, { "epoch": 9.42648736845797e-14, "grad_norm": 0.5381873846054077, "learning_rate": 9.719081491119272e-05, "loss": 2.3826, "step": 27170 }, { "epoch": 9.429956815409923e-14, "grad_norm": 0.569635272026062, "learning_rate": 9.717151365120266e-05, "loss": 2.4018, "step": 27180 }, { "epoch": 9.433426262361877e-14, "grad_norm": 0.5293648838996887, "learning_rate": 9.715220789659489e-05, "loss": 2.3975, "step": 27190 }, { "epoch": 9.43689570931383e-14, "grad_norm": 0.5518786311149597, "learning_rate": 9.713289765042241e-05, "loss": 2.4034, "step": 27200 }, { "epoch": 9.440365156265784e-14, "grad_norm": 0.5506157875061035, "learning_rate": 9.7113582915739e-05, "loss": 2.3818, "step": 27210 }, { "epoch": 9.443834603217738e-14, "grad_norm": 0.5663739442825317, "learning_rate": 9.709426369559909e-05, "loss": 2.4116, "step": 27220 }, { "epoch": 9.447304050169691e-14, "grad_norm": 0.5841239094734192, "learning_rate": 9.707493999305788e-05, "loss": 2.4547, "step": 27230 }, { "epoch": 9.450773497121645e-14, "grad_norm": 0.5459904670715332, "learning_rate": 9.705561181117122e-05, "loss": 2.4035, "step": 27240 }, { "epoch": 9.454242944073599e-14, "grad_norm": 0.565638542175293, "learning_rate": 9.703627915299573e-05, "loss": 2.3735, "step": 27250 }, { "epoch": 9.457712391025552e-14, "grad_norm": 0.5659382939338684, "learning_rate": 9.701694202158866e-05, "loss": 2.3937, "step": 27260 }, { "epoch": 9.461181837977506e-14, "grad_norm": 0.5879226922988892, "learning_rate": 9.699760042000804e-05, "loss": 2.4235, "step": 27270 }, { "epoch": 9.46465128492946e-14, "grad_norm": 0.6044114232063293, "learning_rate": 9.697825435131257e-05, "loss": 2.4233, "step": 27280 }, { "epoch": 9.468120731881413e-14, "grad_norm": 0.5647789835929871, "learning_rate": 9.695890381856168e-05, "loss": 2.3815, "step": 27290 }, { "epoch": 9.471590178833367e-14, "grad_norm": 0.5406466126441956, "learning_rate": 9.693954882481546e-05, "loss": 2.4337, "step": 27300 }, { "epoch": 9.47505962578532e-14, "grad_norm": 0.5539856553077698, "learning_rate": 9.692018937313475e-05, "loss": 2.4508, "step": 27310 }, { "epoch": 9.478529072737274e-14, "grad_norm": 0.5306506156921387, "learning_rate": 9.690082546658107e-05, "loss": 2.402, "step": 27320 }, { "epoch": 9.481998519689228e-14, "grad_norm": 0.56385338306427, "learning_rate": 9.688145710821669e-05, "loss": 2.405, "step": 27330 }, { "epoch": 9.485467966641181e-14, "grad_norm": 0.5651365518569946, "learning_rate": 9.686208430110452e-05, "loss": 2.3925, "step": 27340 }, { "epoch": 9.488937413593135e-14, "grad_norm": 0.5459543466567993, "learning_rate": 9.684270704830821e-05, "loss": 2.4103, "step": 27350 }, { "epoch": 9.492406860545088e-14, "grad_norm": 0.5890876054763794, "learning_rate": 9.682332535289209e-05, "loss": 2.4272, "step": 27360 }, { "epoch": 9.495876307497042e-14, "grad_norm": 0.6010164022445679, "learning_rate": 9.680393921792124e-05, "loss": 2.4148, "step": 27370 }, { "epoch": 9.499345754448996e-14, "grad_norm": 0.5780273079872131, "learning_rate": 9.678454864646138e-05, "loss": 2.3868, "step": 27380 }, { "epoch": 9.502815201400949e-14, "grad_norm": 0.5567091703414917, "learning_rate": 9.676515364157898e-05, "loss": 2.4064, "step": 27390 }, { "epoch": 9.506284648352903e-14, "grad_norm": 0.5643218159675598, "learning_rate": 9.674575420634119e-05, "loss": 2.3859, "step": 27400 }, { "epoch": 9.509754095304856e-14, "grad_norm": 0.5350591540336609, "learning_rate": 9.672635034381586e-05, "loss": 2.3955, "step": 27410 }, { "epoch": 9.51322354225681e-14, "grad_norm": 0.5532533526420593, "learning_rate": 9.670694205707154e-05, "loss": 2.4074, "step": 27420 }, { "epoch": 9.516692989208764e-14, "grad_norm": 0.5565969944000244, "learning_rate": 9.66875293491775e-05, "loss": 2.4409, "step": 27430 }, { "epoch": 9.520162436160717e-14, "grad_norm": 0.5779430866241455, "learning_rate": 9.666811222320366e-05, "loss": 2.4251, "step": 27440 }, { "epoch": 9.523631883112671e-14, "grad_norm": 0.5502991676330566, "learning_rate": 9.664869068222071e-05, "loss": 2.4426, "step": 27450 }, { "epoch": 9.527101330064625e-14, "grad_norm": 0.5424234867095947, "learning_rate": 9.662926472929998e-05, "loss": 2.4258, "step": 27460 }, { "epoch": 9.530570777016578e-14, "grad_norm": 0.5370607376098633, "learning_rate": 9.660983436751351e-05, "loss": 2.3863, "step": 27470 }, { "epoch": 9.534040223968532e-14, "grad_norm": 0.5594062209129333, "learning_rate": 9.659039959993406e-05, "loss": 2.388, "step": 27480 }, { "epoch": 9.537509670920485e-14, "grad_norm": 0.5611739754676819, "learning_rate": 9.657096042963507e-05, "loss": 2.4163, "step": 27490 }, { "epoch": 9.540979117872439e-14, "grad_norm": 0.551555335521698, "learning_rate": 9.655151685969066e-05, "loss": 2.3781, "step": 27500 }, { "epoch": 9.544448564824393e-14, "grad_norm": 0.5829760432243347, "learning_rate": 9.653206889317569e-05, "loss": 2.412, "step": 27510 }, { "epoch": 9.547918011776346e-14, "grad_norm": 0.5583577156066895, "learning_rate": 9.651261653316567e-05, "loss": 2.3997, "step": 27520 }, { "epoch": 9.5513874587283e-14, "grad_norm": 0.5419352650642395, "learning_rate": 9.649315978273683e-05, "loss": 2.4371, "step": 27530 }, { "epoch": 9.554856905680253e-14, "grad_norm": 0.6126682162284851, "learning_rate": 9.647369864496607e-05, "loss": 2.3825, "step": 27540 }, { "epoch": 9.558326352632207e-14, "grad_norm": 0.5610232353210449, "learning_rate": 9.645423312293103e-05, "loss": 2.4028, "step": 27550 }, { "epoch": 9.561795799584161e-14, "grad_norm": 0.5484853982925415, "learning_rate": 9.643476321971001e-05, "loss": 2.38, "step": 27560 }, { "epoch": 9.565265246536114e-14, "grad_norm": 0.5607242584228516, "learning_rate": 9.641528893838199e-05, "loss": 2.3854, "step": 27570 }, { "epoch": 9.568734693488068e-14, "grad_norm": 0.56036376953125, "learning_rate": 9.639581028202666e-05, "loss": 2.4534, "step": 27580 }, { "epoch": 9.572204140440022e-14, "grad_norm": 0.5518860220909119, "learning_rate": 9.637632725372443e-05, "loss": 2.3911, "step": 27590 }, { "epoch": 9.575673587391975e-14, "grad_norm": 0.5895405411720276, "learning_rate": 9.635683985655633e-05, "loss": 2.4219, "step": 27600 }, { "epoch": 9.579143034343929e-14, "grad_norm": 0.5651900768280029, "learning_rate": 9.633734809360418e-05, "loss": 2.4083, "step": 27610 }, { "epoch": 9.582612481295882e-14, "grad_norm": 0.5499807000160217, "learning_rate": 9.631785196795035e-05, "loss": 2.4153, "step": 27620 }, { "epoch": 9.586081928247836e-14, "grad_norm": 0.5550734996795654, "learning_rate": 9.629835148267808e-05, "loss": 2.4034, "step": 27630 }, { "epoch": 9.58955137519979e-14, "grad_norm": 0.591543436050415, "learning_rate": 9.627884664087116e-05, "loss": 2.3837, "step": 27640 }, { "epoch": 9.593020822151743e-14, "grad_norm": 0.5524836778640747, "learning_rate": 9.62593374456141e-05, "loss": 2.4031, "step": 27650 }, { "epoch": 9.596490269103697e-14, "grad_norm": 0.5499709248542786, "learning_rate": 9.62398238999921e-05, "loss": 2.4581, "step": 27660 }, { "epoch": 9.59995971605565e-14, "grad_norm": 0.5499160289764404, "learning_rate": 9.622030600709114e-05, "loss": 2.3987, "step": 27670 }, { "epoch": 9.603429163007604e-14, "grad_norm": 0.5332122445106506, "learning_rate": 9.620078376999772e-05, "loss": 2.3805, "step": 27680 }, { "epoch": 9.606898609959558e-14, "grad_norm": 0.5701301693916321, "learning_rate": 9.618125719179914e-05, "loss": 2.4325, "step": 27690 }, { "epoch": 9.610368056911511e-14, "grad_norm": 0.5394874215126038, "learning_rate": 9.616172627558336e-05, "loss": 2.3942, "step": 27700 }, { "epoch": 9.613837503863465e-14, "grad_norm": 0.5703315138816833, "learning_rate": 9.614219102443905e-05, "loss": 2.3875, "step": 27710 }, { "epoch": 9.617306950815419e-14, "grad_norm": 0.6102027297019958, "learning_rate": 9.612265144145552e-05, "loss": 2.3464, "step": 27720 }, { "epoch": 9.620776397767372e-14, "grad_norm": 0.5472743511199951, "learning_rate": 9.61031075297228e-05, "loss": 2.3686, "step": 27730 }, { "epoch": 9.624245844719326e-14, "grad_norm": 0.5419854521751404, "learning_rate": 9.608355929233156e-05, "loss": 2.4134, "step": 27740 }, { "epoch": 9.62771529167128e-14, "grad_norm": 0.540151059627533, "learning_rate": 9.606400673237322e-05, "loss": 2.4124, "step": 27750 }, { "epoch": 9.631184738623233e-14, "grad_norm": 0.5519115328788757, "learning_rate": 9.604444985293983e-05, "loss": 2.3928, "step": 27760 }, { "epoch": 9.634654185575187e-14, "grad_norm": 0.5527346730232239, "learning_rate": 9.602488865712417e-05, "loss": 2.4438, "step": 27770 }, { "epoch": 9.63812363252714e-14, "grad_norm": 0.5474745035171509, "learning_rate": 9.600532314801961e-05, "loss": 2.4045, "step": 27780 }, { "epoch": 9.641593079479094e-14, "grad_norm": 0.5573525428771973, "learning_rate": 9.598575332872034e-05, "loss": 2.4273, "step": 27790 }, { "epoch": 9.645062526431047e-14, "grad_norm": 0.5534637570381165, "learning_rate": 9.596617920232112e-05, "loss": 2.4016, "step": 27800 }, { "epoch": 9.648531973383001e-14, "grad_norm": 0.5629870891571045, "learning_rate": 9.594660077191742e-05, "loss": 2.4158, "step": 27810 }, { "epoch": 9.652001420334955e-14, "grad_norm": 0.5376394987106323, "learning_rate": 9.592701804060545e-05, "loss": 2.3944, "step": 27820 }, { "epoch": 9.655470867286908e-14, "grad_norm": 0.5492544770240784, "learning_rate": 9.590743101148201e-05, "loss": 2.4215, "step": 27830 }, { "epoch": 9.658940314238862e-14, "grad_norm": 0.5567702054977417, "learning_rate": 9.588783968764461e-05, "loss": 2.4425, "step": 27840 }, { "epoch": 9.662409761190816e-14, "grad_norm": 0.5613316893577576, "learning_rate": 9.586824407219148e-05, "loss": 2.3965, "step": 27850 }, { "epoch": 9.665879208142769e-14, "grad_norm": 0.564526379108429, "learning_rate": 9.584864416822147e-05, "loss": 2.3961, "step": 27860 }, { "epoch": 9.669348655094723e-14, "grad_norm": 0.5353239178657532, "learning_rate": 9.582903997883419e-05, "loss": 2.4028, "step": 27870 }, { "epoch": 9.672818102046676e-14, "grad_norm": 0.5676625967025757, "learning_rate": 9.58094315071298e-05, "loss": 2.3992, "step": 27880 }, { "epoch": 9.67628754899863e-14, "grad_norm": 0.5550457239151001, "learning_rate": 9.578981875620926e-05, "loss": 2.4121, "step": 27890 }, { "epoch": 9.679756995950584e-14, "grad_norm": 0.5685123801231384, "learning_rate": 9.577020172917415e-05, "loss": 2.4094, "step": 27900 }, { "epoch": 9.683226442902537e-14, "grad_norm": 0.5427864789962769, "learning_rate": 9.575058042912673e-05, "loss": 2.4048, "step": 27910 }, { "epoch": 9.686695889854491e-14, "grad_norm": 0.5445307493209839, "learning_rate": 9.573095485916992e-05, "loss": 2.3888, "step": 27920 }, { "epoch": 9.690165336806444e-14, "grad_norm": 0.5381543040275574, "learning_rate": 9.571132502240735e-05, "loss": 2.4006, "step": 27930 }, { "epoch": 9.693634783758398e-14, "grad_norm": 0.5578057169914246, "learning_rate": 9.569169092194333e-05, "loss": 2.4264, "step": 27940 }, { "epoch": 9.697104230710352e-14, "grad_norm": 0.5368276238441467, "learning_rate": 9.56720525608828e-05, "loss": 2.3962, "step": 27950 }, { "epoch": 9.700573677662305e-14, "grad_norm": 0.5415818691253662, "learning_rate": 9.565240994233141e-05, "loss": 2.4122, "step": 27960 }, { "epoch": 9.704043124614259e-14, "grad_norm": 0.5504457354545593, "learning_rate": 9.563276306939544e-05, "loss": 2.4214, "step": 27970 }, { "epoch": 9.707512571566213e-14, "grad_norm": 0.5857498049736023, "learning_rate": 9.561311194518192e-05, "loss": 2.3815, "step": 27980 }, { "epoch": 9.710982018518166e-14, "grad_norm": 0.5661400556564331, "learning_rate": 9.559345657279848e-05, "loss": 2.3638, "step": 27990 }, { "epoch": 9.71445146547012e-14, "grad_norm": 0.5488722920417786, "learning_rate": 9.557379695535348e-05, "loss": 2.4064, "step": 28000 }, { "epoch": 9.717920912422073e-14, "grad_norm": 0.5410741567611694, "learning_rate": 9.555413309595585e-05, "loss": 2.3716, "step": 28010 }, { "epoch": 9.721390359374027e-14, "grad_norm": 0.5701033473014832, "learning_rate": 9.553446499771531e-05, "loss": 2.4288, "step": 28020 }, { "epoch": 9.72485980632598e-14, "grad_norm": 0.5200414061546326, "learning_rate": 9.551479266374221e-05, "loss": 2.3921, "step": 28030 }, { "epoch": 9.728329253277934e-14, "grad_norm": 0.5512361526489258, "learning_rate": 9.549511609714754e-05, "loss": 2.3987, "step": 28040 }, { "epoch": 9.731798700229888e-14, "grad_norm": 0.5444222092628479, "learning_rate": 9.547543530104298e-05, "loss": 2.3805, "step": 28050 }, { "epoch": 9.735268147181841e-14, "grad_norm": 0.5777656435966492, "learning_rate": 9.545575027854089e-05, "loss": 2.4195, "step": 28060 }, { "epoch": 9.738737594133795e-14, "grad_norm": 0.5626847743988037, "learning_rate": 9.543606103275426e-05, "loss": 2.4122, "step": 28070 }, { "epoch": 9.742207041085749e-14, "grad_norm": 0.5577059984207153, "learning_rate": 9.54163675667968e-05, "loss": 2.42, "step": 28080 }, { "epoch": 9.745676488037702e-14, "grad_norm": 0.5626766085624695, "learning_rate": 9.539666988378286e-05, "loss": 2.4062, "step": 28090 }, { "epoch": 9.749145934989656e-14, "grad_norm": 0.5556621551513672, "learning_rate": 9.537696798682744e-05, "loss": 2.3945, "step": 28100 }, { "epoch": 9.75261538194161e-14, "grad_norm": 0.5763980746269226, "learning_rate": 9.535726187904627e-05, "loss": 2.4187, "step": 28110 }, { "epoch": 9.756084828893563e-14, "grad_norm": 0.5768687725067139, "learning_rate": 9.533755156355564e-05, "loss": 2.4092, "step": 28120 }, { "epoch": 9.759554275845517e-14, "grad_norm": 0.5758286714553833, "learning_rate": 9.53178370434726e-05, "loss": 2.4101, "step": 28130 }, { "epoch": 9.76302372279747e-14, "grad_norm": 0.5573415756225586, "learning_rate": 9.529811832191486e-05, "loss": 2.3954, "step": 28140 }, { "epoch": 9.766493169749424e-14, "grad_norm": 0.5603094100952148, "learning_rate": 9.527839540200071e-05, "loss": 2.3959, "step": 28150 }, { "epoch": 9.769962616701378e-14, "grad_norm": 0.5600277185440063, "learning_rate": 9.525866828684919e-05, "loss": 2.3545, "step": 28160 }, { "epoch": 9.773432063653331e-14, "grad_norm": 0.5677960515022278, "learning_rate": 9.523893697957996e-05, "loss": 2.3846, "step": 28170 }, { "epoch": 9.776901510605285e-14, "grad_norm": 0.5624752044677734, "learning_rate": 9.521920148331336e-05, "loss": 2.4091, "step": 28180 }, { "epoch": 9.780370957557238e-14, "grad_norm": 0.5523982644081116, "learning_rate": 9.519946180117042e-05, "loss": 2.4132, "step": 28190 }, { "epoch": 9.783840404509192e-14, "grad_norm": 0.5584945678710938, "learning_rate": 9.517971793627275e-05, "loss": 2.4311, "step": 28200 }, { "epoch": 9.787309851461146e-14, "grad_norm": 0.5563260316848755, "learning_rate": 9.515996989174273e-05, "loss": 2.3821, "step": 28210 }, { "epoch": 9.790779298413099e-14, "grad_norm": 0.5778075456619263, "learning_rate": 9.514021767070327e-05, "loss": 2.3873, "step": 28220 }, { "epoch": 9.794248745365053e-14, "grad_norm": 0.5528417229652405, "learning_rate": 9.512046127627808e-05, "loss": 2.3612, "step": 28230 }, { "epoch": 9.797718192317006e-14, "grad_norm": 0.5679506063461304, "learning_rate": 9.51007007115914e-05, "loss": 2.4257, "step": 28240 }, { "epoch": 9.80118763926896e-14, "grad_norm": 0.5572946667671204, "learning_rate": 9.508093597976828e-05, "loss": 2.4188, "step": 28250 }, { "epoch": 9.804657086220914e-14, "grad_norm": 0.5832510590553284, "learning_rate": 9.506116708393427e-05, "loss": 2.4294, "step": 28260 }, { "epoch": 9.808126533172867e-14, "grad_norm": 0.5832209587097168, "learning_rate": 9.504139402721568e-05, "loss": 2.3996, "step": 28270 }, { "epoch": 9.811595980124821e-14, "grad_norm": 0.5659646987915039, "learning_rate": 9.502161681273943e-05, "loss": 2.4031, "step": 28280 }, { "epoch": 9.815065427076775e-14, "grad_norm": 0.5641638040542603, "learning_rate": 9.500183544363315e-05, "loss": 2.3865, "step": 28290 }, { "epoch": 9.818534874028728e-14, "grad_norm": 0.5763866901397705, "learning_rate": 9.498204992302506e-05, "loss": 2.3965, "step": 28300 }, { "epoch": 9.822004320980682e-14, "grad_norm": 0.5393321514129639, "learning_rate": 9.496226025404407e-05, "loss": 2.3965, "step": 28310 }, { "epoch": 9.825473767932635e-14, "grad_norm": 0.5649250745773315, "learning_rate": 9.494246643981978e-05, "loss": 2.3913, "step": 28320 }, { "epoch": 9.828943214884589e-14, "grad_norm": 0.5638706684112549, "learning_rate": 9.492266848348237e-05, "loss": 2.4157, "step": 28330 }, { "epoch": 9.832412661836543e-14, "grad_norm": 0.5630431771278381, "learning_rate": 9.490286638816273e-05, "loss": 2.407, "step": 28340 }, { "epoch": 9.835882108788496e-14, "grad_norm": 0.5328143835067749, "learning_rate": 9.488306015699242e-05, "loss": 2.4086, "step": 28350 }, { "epoch": 9.83935155574045e-14, "grad_norm": 0.5738922357559204, "learning_rate": 9.486324979310358e-05, "loss": 2.3429, "step": 28360 }, { "epoch": 9.842821002692403e-14, "grad_norm": 0.5720822215080261, "learning_rate": 9.484343529962909e-05, "loss": 2.3841, "step": 28370 }, { "epoch": 9.846290449644357e-14, "grad_norm": 0.5386765003204346, "learning_rate": 9.482361667970242e-05, "loss": 2.403, "step": 28380 }, { "epoch": 9.849759896596311e-14, "grad_norm": 0.6300882697105408, "learning_rate": 9.48037939364577e-05, "loss": 2.3926, "step": 28390 }, { "epoch": 9.853229343548264e-14, "grad_norm": 0.5528148412704468, "learning_rate": 9.478396707302975e-05, "loss": 2.3798, "step": 28400 }, { "epoch": 9.856698790500218e-14, "grad_norm": 0.5707433819770813, "learning_rate": 9.476413609255399e-05, "loss": 2.4132, "step": 28410 }, { "epoch": 9.860168237452172e-14, "grad_norm": 0.5395046472549438, "learning_rate": 9.474430099816656e-05, "loss": 2.3986, "step": 28420 }, { "epoch": 9.863637684404125e-14, "grad_norm": 0.5545224547386169, "learning_rate": 9.47244617930042e-05, "loss": 2.3842, "step": 28430 }, { "epoch": 9.867107131356079e-14, "grad_norm": 0.5698997974395752, "learning_rate": 9.470461848020429e-05, "loss": 2.4304, "step": 28440 }, { "epoch": 9.870576578308032e-14, "grad_norm": 0.5785558223724365, "learning_rate": 9.468477106290487e-05, "loss": 2.4188, "step": 28450 }, { "epoch": 9.874046025259986e-14, "grad_norm": 0.569245457649231, "learning_rate": 9.466491954424465e-05, "loss": 2.4161, "step": 28460 }, { "epoch": 9.87751547221194e-14, "grad_norm": 0.5549311637878418, "learning_rate": 9.4645063927363e-05, "loss": 2.3875, "step": 28470 }, { "epoch": 9.880984919163893e-14, "grad_norm": 0.5345077514648438, "learning_rate": 9.462520421539988e-05, "loss": 2.358, "step": 28480 }, { "epoch": 9.884454366115847e-14, "grad_norm": 0.5773128271102905, "learning_rate": 9.460534041149595e-05, "loss": 2.4009, "step": 28490 }, { "epoch": 9.8879238130678e-14, "grad_norm": 0.5553207993507385, "learning_rate": 9.45854725187925e-05, "loss": 2.3978, "step": 28500 }, { "epoch": 9.891393260019754e-14, "grad_norm": 0.57541424036026, "learning_rate": 9.456560054043148e-05, "loss": 2.3855, "step": 28510 }, { "epoch": 9.894862706971708e-14, "grad_norm": 0.574540913105011, "learning_rate": 9.454572447955545e-05, "loss": 2.4189, "step": 28520 }, { "epoch": 9.898332153923661e-14, "grad_norm": 0.5762327909469604, "learning_rate": 9.452584433930764e-05, "loss": 2.4013, "step": 28530 }, { "epoch": 9.901801600875615e-14, "grad_norm": 0.5913914442062378, "learning_rate": 9.450596012283192e-05, "loss": 2.3983, "step": 28540 }, { "epoch": 9.905271047827569e-14, "grad_norm": 0.5372111201286316, "learning_rate": 9.448607183327283e-05, "loss": 2.4312, "step": 28550 }, { "epoch": 9.908740494779522e-14, "grad_norm": 0.6104349493980408, "learning_rate": 9.446617947377551e-05, "loss": 2.4233, "step": 28560 }, { "epoch": 9.912209941731476e-14, "grad_norm": 0.5715567469596863, "learning_rate": 9.444628304748577e-05, "loss": 2.3863, "step": 28570 }, { "epoch": 9.915679388683429e-14, "grad_norm": 0.5544934272766113, "learning_rate": 9.442638255755007e-05, "loss": 2.3573, "step": 28580 }, { "epoch": 9.919148835635383e-14, "grad_norm": 0.5642582178115845, "learning_rate": 9.440647800711549e-05, "loss": 2.4277, "step": 28590 }, { "epoch": 9.922618282587337e-14, "grad_norm": 0.5696749091148376, "learning_rate": 9.438656939932977e-05, "loss": 2.4248, "step": 28600 }, { "epoch": 9.92608772953929e-14, "grad_norm": 0.5808570384979248, "learning_rate": 9.436665673734126e-05, "loss": 2.3788, "step": 28610 }, { "epoch": 9.929557176491244e-14, "grad_norm": 0.5817629098892212, "learning_rate": 9.434674002429901e-05, "loss": 2.4142, "step": 28620 }, { "epoch": 9.933026623443197e-14, "grad_norm": 0.5507323741912842, "learning_rate": 9.432681926335267e-05, "loss": 2.4057, "step": 28630 }, { "epoch": 9.936496070395151e-14, "grad_norm": 0.6064805388450623, "learning_rate": 9.430689445765252e-05, "loss": 2.4107, "step": 28640 }, { "epoch": 9.939965517347105e-14, "grad_norm": 0.5592120885848999, "learning_rate": 9.42869656103495e-05, "loss": 2.3776, "step": 28650 }, { "epoch": 9.943434964299058e-14, "grad_norm": 0.5458888411521912, "learning_rate": 9.426703272459521e-05, "loss": 2.4222, "step": 28660 }, { "epoch": 9.946904411251012e-14, "grad_norm": 0.5480506420135498, "learning_rate": 9.424709580354185e-05, "loss": 2.3933, "step": 28670 }, { "epoch": 9.950373858202965e-14, "grad_norm": 0.5453349351882935, "learning_rate": 9.422715485034226e-05, "loss": 2.3831, "step": 28680 }, { "epoch": 9.953843305154919e-14, "grad_norm": 0.5444241762161255, "learning_rate": 9.420720986814993e-05, "loss": 2.384, "step": 28690 }, { "epoch": 9.957312752106873e-14, "grad_norm": 0.5525915622711182, "learning_rate": 9.418726086011901e-05, "loss": 2.384, "step": 28700 }, { "epoch": 9.960782199058826e-14, "grad_norm": 0.5406032800674438, "learning_rate": 9.416730782940423e-05, "loss": 2.3985, "step": 28710 }, { "epoch": 9.96425164601078e-14, "grad_norm": 0.5486472845077515, "learning_rate": 9.414735077916104e-05, "loss": 2.3748, "step": 28720 }, { "epoch": 9.967721092962734e-14, "grad_norm": 0.5650331974029541, "learning_rate": 9.412738971254543e-05, "loss": 2.4232, "step": 28730 }, { "epoch": 9.971190539914687e-14, "grad_norm": 0.5965089201927185, "learning_rate": 9.410742463271408e-05, "loss": 2.3794, "step": 28740 }, { "epoch": 9.974659986866641e-14, "grad_norm": 0.5770296454429626, "learning_rate": 9.408745554282431e-05, "loss": 2.4222, "step": 28750 }, { "epoch": 9.978129433818594e-14, "grad_norm": 0.5308224558830261, "learning_rate": 9.406748244603407e-05, "loss": 2.3658, "step": 28760 }, { "epoch": 9.981598880770548e-14, "grad_norm": 0.5691573023796082, "learning_rate": 9.40475053455019e-05, "loss": 2.3979, "step": 28770 }, { "epoch": 9.985068327722502e-14, "grad_norm": 0.5578698515892029, "learning_rate": 9.402752424438701e-05, "loss": 2.4184, "step": 28780 }, { "epoch": 9.988537774674455e-14, "grad_norm": 0.5497139692306519, "learning_rate": 9.400753914584927e-05, "loss": 2.4108, "step": 28790 }, { "epoch": 9.992007221626409e-14, "grad_norm": 0.5545321702957153, "learning_rate": 9.398755005304911e-05, "loss": 2.402, "step": 28800 }, { "epoch": 9.995476668578362e-14, "grad_norm": 0.5551892518997192, "learning_rate": 9.396755696914769e-05, "loss": 2.4362, "step": 28810 }, { "epoch": 9.998946115530316e-14, "grad_norm": 0.5900185704231262, "learning_rate": 9.394755989730667e-05, "loss": 2.4104, "step": 28820 }, { "epoch": 1.000241556248227e-13, "grad_norm": 0.5423593521118164, "learning_rate": 9.392755884068848e-05, "loss": 2.4108, "step": 28830 }, { "epoch": 1.0005885009434223e-13, "grad_norm": 0.5859991312026978, "learning_rate": 9.390755380245608e-05, "loss": 2.4111, "step": 28840 }, { "epoch": 1.0009354456386177e-13, "grad_norm": 0.5879217982292175, "learning_rate": 9.388754478577312e-05, "loss": 2.3874, "step": 28850 }, { "epoch": 1.001282390333813e-13, "grad_norm": 0.5526617169380188, "learning_rate": 9.386753179380382e-05, "loss": 2.4018, "step": 28860 }, { "epoch": 1.0016293350290084e-13, "grad_norm": 0.5806336402893066, "learning_rate": 9.384751482971308e-05, "loss": 2.4117, "step": 28870 }, { "epoch": 1.0019762797242038e-13, "grad_norm": 0.5867785215377808, "learning_rate": 9.382749389666643e-05, "loss": 2.4146, "step": 28880 }, { "epoch": 1.0023232244193991e-13, "grad_norm": 0.5451642870903015, "learning_rate": 9.380746899782997e-05, "loss": 2.3483, "step": 28890 }, { "epoch": 1.0026701691145945e-13, "grad_norm": 0.5841920375823975, "learning_rate": 9.378744013637048e-05, "loss": 2.414, "step": 28900 }, { "epoch": 1.0030171138097899e-13, "grad_norm": 0.5587546825408936, "learning_rate": 9.376740731545537e-05, "loss": 2.4037, "step": 28910 }, { "epoch": 1.0033640585049852e-13, "grad_norm": 0.5576076507568359, "learning_rate": 9.374737053825265e-05, "loss": 2.4117, "step": 28920 }, { "epoch": 1.0037110032001806e-13, "grad_norm": 0.5532352328300476, "learning_rate": 9.372732980793093e-05, "loss": 2.3951, "step": 28930 }, { "epoch": 1.004057947895376e-13, "grad_norm": 0.5423511862754822, "learning_rate": 9.370728512765954e-05, "loss": 2.3836, "step": 28940 }, { "epoch": 1.0044048925905713e-13, "grad_norm": 0.554423451423645, "learning_rate": 9.368723650060832e-05, "loss": 2.4208, "step": 28950 }, { "epoch": 1.0047518372857667e-13, "grad_norm": 0.5415956974029541, "learning_rate": 9.36671839299478e-05, "loss": 2.4268, "step": 28960 }, { "epoch": 1.005098781980962e-13, "grad_norm": 0.5579209923744202, "learning_rate": 9.364712741884916e-05, "loss": 2.4196, "step": 28970 }, { "epoch": 1.0054457266761574e-13, "grad_norm": 0.5689019560813904, "learning_rate": 9.362706697048411e-05, "loss": 2.4173, "step": 28980 }, { "epoch": 1.0057926713713528e-13, "grad_norm": 0.5354799628257751, "learning_rate": 9.360700258802506e-05, "loss": 2.3536, "step": 28990 }, { "epoch": 1.0061396160665481e-13, "grad_norm": 0.5798614621162415, "learning_rate": 9.358693427464504e-05, "loss": 2.3976, "step": 29000 }, { "epoch": 1.0064865607617435e-13, "grad_norm": 0.5720332860946655, "learning_rate": 9.356686203351767e-05, "loss": 2.3872, "step": 29010 }, { "epoch": 1.0068335054569388e-13, "grad_norm": 0.5827178359031677, "learning_rate": 9.354678586781718e-05, "loss": 2.3985, "step": 29020 }, { "epoch": 1.0071804501521342e-13, "grad_norm": 0.5759417414665222, "learning_rate": 9.352670578071844e-05, "loss": 2.3852, "step": 29030 }, { "epoch": 1.0075273948473296e-13, "grad_norm": 0.558983564376831, "learning_rate": 9.350662177539698e-05, "loss": 2.4148, "step": 29040 }, { "epoch": 1.0078743395425249e-13, "grad_norm": 0.5506790280342102, "learning_rate": 9.348653385502889e-05, "loss": 2.3757, "step": 29050 }, { "epoch": 1.0082212842377203e-13, "grad_norm": 0.5406849384307861, "learning_rate": 9.346644202279089e-05, "loss": 2.3894, "step": 29060 }, { "epoch": 1.0085682289329156e-13, "grad_norm": 0.5463433861732483, "learning_rate": 9.344634628186036e-05, "loss": 2.4133, "step": 29070 }, { "epoch": 1.008915173628111e-13, "grad_norm": 0.5827387571334839, "learning_rate": 9.342624663541527e-05, "loss": 2.3882, "step": 29080 }, { "epoch": 1.0092621183233064e-13, "grad_norm": 0.5399664044380188, "learning_rate": 9.340614308663415e-05, "loss": 2.3833, "step": 29090 }, { "epoch": 1.0096090630185017e-13, "grad_norm": 0.5709380507469177, "learning_rate": 9.338603563869628e-05, "loss": 2.4238, "step": 29100 }, { "epoch": 1.0099560077136971e-13, "grad_norm": 0.5788446068763733, "learning_rate": 9.336592429478143e-05, "loss": 2.3957, "step": 29110 }, { "epoch": 1.0103029524088925e-13, "grad_norm": 0.5376695990562439, "learning_rate": 9.334580905807002e-05, "loss": 2.3656, "step": 29120 }, { "epoch": 1.0106498971040878e-13, "grad_norm": 0.5549370646476746, "learning_rate": 9.332568993174316e-05, "loss": 2.4083, "step": 29130 }, { "epoch": 1.0109968417992832e-13, "grad_norm": 0.571911633014679, "learning_rate": 9.330556691898248e-05, "loss": 2.387, "step": 29140 }, { "epoch": 1.0113437864944785e-13, "grad_norm": 0.5625308156013489, "learning_rate": 9.328544002297026e-05, "loss": 2.3876, "step": 29150 }, { "epoch": 1.0116907311896739e-13, "grad_norm": 0.562080442905426, "learning_rate": 9.326530924688942e-05, "loss": 2.368, "step": 29160 }, { "epoch": 1.0120376758848693e-13, "grad_norm": 0.5594699382781982, "learning_rate": 9.324517459392344e-05, "loss": 2.42, "step": 29170 }, { "epoch": 1.0123846205800646e-13, "grad_norm": 0.5677003264427185, "learning_rate": 9.322503606725644e-05, "loss": 2.4103, "step": 29180 }, { "epoch": 1.01273156527526e-13, "grad_norm": 0.5445783138275146, "learning_rate": 9.32048936700732e-05, "loss": 2.3697, "step": 29190 }, { "epoch": 1.0130785099704553e-13, "grad_norm": 0.5820077061653137, "learning_rate": 9.318474740555901e-05, "loss": 2.4018, "step": 29200 }, { "epoch": 1.0134254546656507e-13, "grad_norm": 0.5473839044570923, "learning_rate": 9.316459727689986e-05, "loss": 2.3929, "step": 29210 }, { "epoch": 1.0137723993608461e-13, "grad_norm": 0.5748271346092224, "learning_rate": 9.314444328728232e-05, "loss": 2.3786, "step": 29220 }, { "epoch": 1.0141193440560414e-13, "grad_norm": 0.5713251829147339, "learning_rate": 9.312428543989355e-05, "loss": 2.3917, "step": 29230 }, { "epoch": 1.0144662887512368e-13, "grad_norm": 0.5581631660461426, "learning_rate": 9.310412373792138e-05, "loss": 2.3981, "step": 29240 }, { "epoch": 1.0148132334464322e-13, "grad_norm": 0.5478533506393433, "learning_rate": 9.308395818455415e-05, "loss": 2.4178, "step": 29250 }, { "epoch": 1.0151601781416275e-13, "grad_norm": 0.5650857090950012, "learning_rate": 9.306378878298091e-05, "loss": 2.4073, "step": 29260 }, { "epoch": 1.0155071228368229e-13, "grad_norm": 0.5457553863525391, "learning_rate": 9.304361553639129e-05, "loss": 2.3728, "step": 29270 }, { "epoch": 1.0158540675320182e-13, "grad_norm": 0.5793463587760925, "learning_rate": 9.302343844797548e-05, "loss": 2.4012, "step": 29280 }, { "epoch": 1.0162010122272136e-13, "grad_norm": 0.577930212020874, "learning_rate": 9.300325752092431e-05, "loss": 2.4167, "step": 29290 }, { "epoch": 1.016547956922409e-13, "grad_norm": 0.5758457183837891, "learning_rate": 9.298307275842923e-05, "loss": 2.4044, "step": 29300 }, { "epoch": 1.0168949016176043e-13, "grad_norm": 0.5446056723594666, "learning_rate": 9.296288416368233e-05, "loss": 2.399, "step": 29310 }, { "epoch": 1.0172418463127997e-13, "grad_norm": 0.5814151167869568, "learning_rate": 9.294269173987623e-05, "loss": 2.4027, "step": 29320 }, { "epoch": 1.017588791007995e-13, "grad_norm": 0.549554169178009, "learning_rate": 9.292249549020415e-05, "loss": 2.3931, "step": 29330 }, { "epoch": 1.0179357357031904e-13, "grad_norm": 0.5749816298484802, "learning_rate": 9.290229541786001e-05, "loss": 2.4315, "step": 29340 }, { "epoch": 1.0182826803983858e-13, "grad_norm": 0.5720140337944031, "learning_rate": 9.288209152603821e-05, "loss": 2.43, "step": 29350 }, { "epoch": 1.0186296250935811e-13, "grad_norm": 0.5744329690933228, "learning_rate": 9.286188381793392e-05, "loss": 2.4053, "step": 29360 }, { "epoch": 1.0189765697887765e-13, "grad_norm": 0.5344810485839844, "learning_rate": 9.284167229674273e-05, "loss": 2.4221, "step": 29370 }, { "epoch": 1.0193235144839718e-13, "grad_norm": 0.5508120656013489, "learning_rate": 9.282145696566095e-05, "loss": 2.4188, "step": 29380 }, { "epoch": 1.0196704591791672e-13, "grad_norm": 0.5479004383087158, "learning_rate": 9.280123782788547e-05, "loss": 2.3935, "step": 29390 }, { "epoch": 1.0200174038743626e-13, "grad_norm": 0.5525877475738525, "learning_rate": 9.278101488661378e-05, "loss": 2.4061, "step": 29400 }, { "epoch": 1.0203643485695579e-13, "grad_norm": 0.5416966676712036, "learning_rate": 9.276078814504391e-05, "loss": 2.3864, "step": 29410 }, { "epoch": 1.0207112932647533e-13, "grad_norm": 0.5506890416145325, "learning_rate": 9.27405576063746e-05, "loss": 2.3973, "step": 29420 }, { "epoch": 1.0210582379599487e-13, "grad_norm": 0.5925576090812683, "learning_rate": 9.272032327380512e-05, "loss": 2.3737, "step": 29430 }, { "epoch": 1.021405182655144e-13, "grad_norm": 0.5581991672515869, "learning_rate": 9.270008515053534e-05, "loss": 2.3704, "step": 29440 }, { "epoch": 1.0217521273503394e-13, "grad_norm": 0.5635791420936584, "learning_rate": 9.267984323976576e-05, "loss": 2.3695, "step": 29450 }, { "epoch": 1.0220990720455347e-13, "grad_norm": 0.6458520293235779, "learning_rate": 9.265959754469745e-05, "loss": 2.4056, "step": 29460 }, { "epoch": 1.0224460167407301e-13, "grad_norm": 0.5750038623809814, "learning_rate": 9.263934806853212e-05, "loss": 2.4239, "step": 29470 }, { "epoch": 1.0227929614359255e-13, "grad_norm": 0.5710302591323853, "learning_rate": 9.261909481447204e-05, "loss": 2.3846, "step": 29480 }, { "epoch": 1.0231399061311208e-13, "grad_norm": 0.5550037622451782, "learning_rate": 9.259883778572008e-05, "loss": 2.4154, "step": 29490 }, { "epoch": 1.0234868508263162e-13, "grad_norm": 0.5662981271743774, "learning_rate": 9.257857698547972e-05, "loss": 2.3537, "step": 29500 }, { "epoch": 1.0238337955215115e-13, "grad_norm": 0.5683168768882751, "learning_rate": 9.255831241695501e-05, "loss": 2.4336, "step": 29510 }, { "epoch": 1.0241807402167069e-13, "grad_norm": 0.5538005232810974, "learning_rate": 9.253804408335062e-05, "loss": 2.3945, "step": 29520 }, { "epoch": 1.0245276849119023e-13, "grad_norm": 0.578133761882782, "learning_rate": 9.251777198787186e-05, "loss": 2.3712, "step": 29530 }, { "epoch": 1.0248746296070976e-13, "grad_norm": 0.5558757185935974, "learning_rate": 9.249749613372455e-05, "loss": 2.4145, "step": 29540 }, { "epoch": 1.025221574302293e-13, "grad_norm": 0.55946946144104, "learning_rate": 9.247721652411514e-05, "loss": 2.3769, "step": 29550 }, { "epoch": 1.0255685189974884e-13, "grad_norm": 0.5633155703544617, "learning_rate": 9.245693316225069e-05, "loss": 2.4143, "step": 29560 }, { "epoch": 1.0259154636926837e-13, "grad_norm": 0.5657464265823364, "learning_rate": 9.24366460513388e-05, "loss": 2.3648, "step": 29570 }, { "epoch": 1.0262624083878791e-13, "grad_norm": 0.5602383017539978, "learning_rate": 9.241635519458778e-05, "loss": 2.4297, "step": 29580 }, { "epoch": 1.0266093530830744e-13, "grad_norm": 0.5544652342796326, "learning_rate": 9.239606059520637e-05, "loss": 2.3609, "step": 29590 }, { "epoch": 1.0269562977782698e-13, "grad_norm": 0.5952119827270508, "learning_rate": 9.237576225640403e-05, "loss": 2.4089, "step": 29600 }, { "epoch": 1.0273032424734652e-13, "grad_norm": 0.5785110592842102, "learning_rate": 9.235546018139076e-05, "loss": 2.4179, "step": 29610 }, { "epoch": 1.0276501871686605e-13, "grad_norm": 0.5607949495315552, "learning_rate": 9.233515437337716e-05, "loss": 2.3704, "step": 29620 }, { "epoch": 1.0279971318638559e-13, "grad_norm": 0.5886299014091492, "learning_rate": 9.231484483557445e-05, "loss": 2.3645, "step": 29630 }, { "epoch": 1.0283440765590512e-13, "grad_norm": 0.5758508443832397, "learning_rate": 9.229453157119435e-05, "loss": 2.3745, "step": 29640 }, { "epoch": 1.0286910212542466e-13, "grad_norm": 0.5766371488571167, "learning_rate": 9.227421458344925e-05, "loss": 2.3632, "step": 29650 }, { "epoch": 1.029037965949442e-13, "grad_norm": 0.5849335193634033, "learning_rate": 9.225389387555212e-05, "loss": 2.3965, "step": 29660 }, { "epoch": 1.0293849106446373e-13, "grad_norm": 0.5503614544868469, "learning_rate": 9.223356945071649e-05, "loss": 2.4237, "step": 29670 }, { "epoch": 1.0297318553398327e-13, "grad_norm": 0.5847958326339722, "learning_rate": 9.22132413121565e-05, "loss": 2.3904, "step": 29680 }, { "epoch": 1.030078800035028e-13, "grad_norm": 0.5498755574226379, "learning_rate": 9.21929094630869e-05, "loss": 2.3451, "step": 29690 }, { "epoch": 1.0304257447302234e-13, "grad_norm": 0.6042669415473938, "learning_rate": 9.217257390672292e-05, "loss": 2.4128, "step": 29700 }, { "epoch": 1.0307726894254188e-13, "grad_norm": 0.5753196477890015, "learning_rate": 9.215223464628054e-05, "loss": 2.4055, "step": 29710 }, { "epoch": 1.0311196341206141e-13, "grad_norm": 0.5601356029510498, "learning_rate": 9.213189168497617e-05, "loss": 2.4056, "step": 29720 }, { "epoch": 1.0314665788158095e-13, "grad_norm": 0.5570131540298462, "learning_rate": 9.211154502602691e-05, "loss": 2.3972, "step": 29730 }, { "epoch": 1.0318135235110049e-13, "grad_norm": 0.5761812925338745, "learning_rate": 9.209119467265039e-05, "loss": 2.3577, "step": 29740 }, { "epoch": 1.0321604682062002e-13, "grad_norm": 0.5352731347084045, "learning_rate": 9.207084062806488e-05, "loss": 2.3541, "step": 29750 }, { "epoch": 1.0325074129013956e-13, "grad_norm": 0.5503727197647095, "learning_rate": 9.205048289548914e-05, "loss": 2.4398, "step": 29760 }, { "epoch": 1.032854357596591e-13, "grad_norm": 0.5439123511314392, "learning_rate": 9.203012147814263e-05, "loss": 2.4013, "step": 29770 }, { "epoch": 1.0332013022917863e-13, "grad_norm": 0.5780762434005737, "learning_rate": 9.200975637924531e-05, "loss": 2.4188, "step": 29780 }, { "epoch": 1.0335482469869817e-13, "grad_norm": 0.5554466843605042, "learning_rate": 9.19893876020177e-05, "loss": 2.4084, "step": 29790 }, { "epoch": 1.033895191682177e-13, "grad_norm": 0.5664693713188171, "learning_rate": 9.196901514968103e-05, "loss": 2.3427, "step": 29800 }, { "epoch": 1.0342421363773724e-13, "grad_norm": 0.5432242751121521, "learning_rate": 9.194863902545697e-05, "loss": 2.3762, "step": 29810 }, { "epoch": 1.0345890810725678e-13, "grad_norm": 0.5690397024154663, "learning_rate": 9.192825923256786e-05, "loss": 2.4012, "step": 29820 }, { "epoch": 1.0349360257677631e-13, "grad_norm": 0.5770685076713562, "learning_rate": 9.190787577423654e-05, "loss": 2.4224, "step": 29830 }, { "epoch": 1.0352829704629585e-13, "grad_norm": 0.5518808364868164, "learning_rate": 9.188748865368651e-05, "loss": 2.4023, "step": 29840 }, { "epoch": 1.0356299151581538e-13, "grad_norm": 0.5239438414573669, "learning_rate": 9.186709787414184e-05, "loss": 2.3788, "step": 29850 }, { "epoch": 1.0359768598533492e-13, "grad_norm": 0.5606805086135864, "learning_rate": 9.184670343882711e-05, "loss": 2.4338, "step": 29860 }, { "epoch": 1.0363238045485446e-13, "grad_norm": 0.6165766716003418, "learning_rate": 9.182630535096759e-05, "loss": 2.3868, "step": 29870 }, { "epoch": 1.0366707492437399e-13, "grad_norm": 0.5466289520263672, "learning_rate": 9.180590361378897e-05, "loss": 2.397, "step": 29880 }, { "epoch": 1.0370176939389353e-13, "grad_norm": 0.560247004032135, "learning_rate": 9.178549823051769e-05, "loss": 2.4386, "step": 29890 }, { "epoch": 1.0373646386341306e-13, "grad_norm": 0.5621964335441589, "learning_rate": 9.176508920438066e-05, "loss": 2.3633, "step": 29900 }, { "epoch": 1.037711583329326e-13, "grad_norm": 0.560436487197876, "learning_rate": 9.174467653860536e-05, "loss": 2.3959, "step": 29910 }, { "epoch": 1.0380585280245214e-13, "grad_norm": 0.5504152774810791, "learning_rate": 9.172426023641994e-05, "loss": 2.367, "step": 29920 }, { "epoch": 1.0384054727197167e-13, "grad_norm": 0.5977057218551636, "learning_rate": 9.170384030105302e-05, "loss": 2.4077, "step": 29930 }, { "epoch": 1.0387524174149121e-13, "grad_norm": 0.5540210008621216, "learning_rate": 9.168341673573382e-05, "loss": 2.3763, "step": 29940 }, { "epoch": 1.0390993621101074e-13, "grad_norm": 0.6540987491607666, "learning_rate": 9.16629895436922e-05, "loss": 2.4009, "step": 29950 }, { "epoch": 1.0394463068053028e-13, "grad_norm": 0.5653108358383179, "learning_rate": 9.164255872815854e-05, "loss": 2.3106, "step": 29960 }, { "epoch": 1.0397932515004982e-13, "grad_norm": 0.597344160079956, "learning_rate": 9.162212429236373e-05, "loss": 2.3877, "step": 29970 }, { "epoch": 1.0401401961956935e-13, "grad_norm": 0.5790861248970032, "learning_rate": 9.160168623953938e-05, "loss": 2.3756, "step": 29980 }, { "epoch": 1.0404871408908889e-13, "grad_norm": 0.5391475558280945, "learning_rate": 9.158124457291754e-05, "loss": 2.3804, "step": 29990 }, { "epoch": 1.0408340855860843e-13, "grad_norm": 0.56046462059021, "learning_rate": 9.156079929573092e-05, "loss": 2.3703, "step": 30000 }, { "epoch": 1.0411810302812796e-13, "grad_norm": 0.5473029017448425, "learning_rate": 9.154035041121272e-05, "loss": 2.3786, "step": 30010 }, { "epoch": 1.041527974976475e-13, "grad_norm": 0.5749661922454834, "learning_rate": 9.151989792259681e-05, "loss": 2.3872, "step": 30020 }, { "epoch": 1.0418749196716703e-13, "grad_norm": 0.5789147615432739, "learning_rate": 9.149944183311755e-05, "loss": 2.4254, "step": 30030 }, { "epoch": 1.0422218643668657e-13, "grad_norm": 0.5828678607940674, "learning_rate": 9.147898214600987e-05, "loss": 2.4269, "step": 30040 }, { "epoch": 1.0425688090620611e-13, "grad_norm": 0.5553803443908691, "learning_rate": 9.145851886450933e-05, "loss": 2.4198, "step": 30050 }, { "epoch": 1.0429157537572564e-13, "grad_norm": 0.5718862414360046, "learning_rate": 9.143805199185201e-05, "loss": 2.4304, "step": 30060 }, { "epoch": 1.0432626984524518e-13, "grad_norm": 0.5367109775543213, "learning_rate": 9.141758153127456e-05, "loss": 2.4023, "step": 30070 }, { "epoch": 1.0436096431476471e-13, "grad_norm": 0.5580459833145142, "learning_rate": 9.139710748601422e-05, "loss": 2.4146, "step": 30080 }, { "epoch": 1.0439565878428425e-13, "grad_norm": 0.5788630843162537, "learning_rate": 9.137662985930878e-05, "loss": 2.4179, "step": 30090 }, { "epoch": 1.0443035325380379e-13, "grad_norm": 0.5589025020599365, "learning_rate": 9.13561486543966e-05, "loss": 2.3444, "step": 30100 }, { "epoch": 1.0446504772332332e-13, "grad_norm": 0.5654733180999756, "learning_rate": 9.13356638745166e-05, "loss": 2.3857, "step": 30110 }, { "epoch": 1.0449974219284286e-13, "grad_norm": 0.5528470277786255, "learning_rate": 9.131517552290829e-05, "loss": 2.4101, "step": 30120 }, { "epoch": 1.045344366623624e-13, "grad_norm": 0.5677087903022766, "learning_rate": 9.129468360281171e-05, "loss": 2.3898, "step": 30130 }, { "epoch": 1.0456913113188193e-13, "grad_norm": 0.5377663969993591, "learning_rate": 9.127418811746748e-05, "loss": 2.3796, "step": 30140 }, { "epoch": 1.0460382560140147e-13, "grad_norm": 0.5572667121887207, "learning_rate": 9.125368907011678e-05, "loss": 2.4081, "step": 30150 }, { "epoch": 1.04638520070921e-13, "grad_norm": 0.5618417263031006, "learning_rate": 9.123318646400134e-05, "loss": 2.4152, "step": 30160 }, { "epoch": 1.0467321454044054e-13, "grad_norm": 0.5472591519355774, "learning_rate": 9.121268030236355e-05, "loss": 2.4024, "step": 30170 }, { "epoch": 1.0470790900996008e-13, "grad_norm": 0.5519917011260986, "learning_rate": 9.119217058844622e-05, "loss": 2.362, "step": 30180 }, { "epoch": 1.0474260347947961e-13, "grad_norm": 0.5361835956573486, "learning_rate": 9.117165732549278e-05, "loss": 2.4155, "step": 30190 }, { "epoch": 1.0477729794899915e-13, "grad_norm": 0.5684929490089417, "learning_rate": 9.115114051674724e-05, "loss": 2.3902, "step": 30200 }, { "epoch": 1.0481199241851868e-13, "grad_norm": 0.5910242199897766, "learning_rate": 9.113062016545416e-05, "loss": 2.3572, "step": 30210 }, { "epoch": 1.0484668688803822e-13, "grad_norm": 0.5678502321243286, "learning_rate": 9.111009627485866e-05, "loss": 2.395, "step": 30220 }, { "epoch": 1.0488138135755776e-13, "grad_norm": 0.5730435848236084, "learning_rate": 9.108956884820641e-05, "loss": 2.3669, "step": 30230 }, { "epoch": 1.0491607582707729e-13, "grad_norm": 0.5820460915565491, "learning_rate": 9.106903788874366e-05, "loss": 2.3816, "step": 30240 }, { "epoch": 1.0495077029659683e-13, "grad_norm": 0.5483666658401489, "learning_rate": 9.104850339971716e-05, "loss": 2.3812, "step": 30250 }, { "epoch": 1.0498546476611637e-13, "grad_norm": 0.5715810656547546, "learning_rate": 9.102796538437431e-05, "loss": 2.3802, "step": 30260 }, { "epoch": 1.050201592356359e-13, "grad_norm": 0.5733579397201538, "learning_rate": 9.100742384596301e-05, "loss": 2.3602, "step": 30270 }, { "epoch": 1.0505485370515544e-13, "grad_norm": 0.5614033341407776, "learning_rate": 9.098687878773172e-05, "loss": 2.3926, "step": 30280 }, { "epoch": 1.0508954817467497e-13, "grad_norm": 0.5603087544441223, "learning_rate": 9.096633021292947e-05, "loss": 2.3782, "step": 30290 }, { "epoch": 1.0512424264419451e-13, "grad_norm": 0.5952281355857849, "learning_rate": 9.094577812480585e-05, "loss": 2.3801, "step": 30300 }, { "epoch": 1.0515893711371405e-13, "grad_norm": 0.549046516418457, "learning_rate": 9.092522252661094e-05, "loss": 2.3653, "step": 30310 }, { "epoch": 1.0519363158323358e-13, "grad_norm": 0.5824404954910278, "learning_rate": 9.090466342159549e-05, "loss": 2.4012, "step": 30320 }, { "epoch": 1.0522832605275312e-13, "grad_norm": 0.5618991255760193, "learning_rate": 9.088410081301074e-05, "loss": 2.3627, "step": 30330 }, { "epoch": 1.0526302052227265e-13, "grad_norm": 0.5685769319534302, "learning_rate": 9.086353470410848e-05, "loss": 2.3657, "step": 30340 }, { "epoch": 1.0529771499179219e-13, "grad_norm": 0.5369185209274292, "learning_rate": 9.084296509814105e-05, "loss": 2.3963, "step": 30350 }, { "epoch": 1.0533240946131173e-13, "grad_norm": 0.6003386974334717, "learning_rate": 9.082239199836136e-05, "loss": 2.3795, "step": 30360 }, { "epoch": 1.0536710393083126e-13, "grad_norm": 0.5660193562507629, "learning_rate": 9.080181540802291e-05, "loss": 2.3944, "step": 30370 }, { "epoch": 1.054017984003508e-13, "grad_norm": 0.6177456378936768, "learning_rate": 9.078123533037964e-05, "loss": 2.3856, "step": 30380 }, { "epoch": 1.0543649286987034e-13, "grad_norm": 0.5715065002441406, "learning_rate": 9.076065176868616e-05, "loss": 2.4128, "step": 30390 }, { "epoch": 1.0547118733938987e-13, "grad_norm": 0.5749316215515137, "learning_rate": 9.074006472619759e-05, "loss": 2.4077, "step": 30400 }, { "epoch": 1.0550588180890941e-13, "grad_norm": 0.5545327067375183, "learning_rate": 9.071947420616955e-05, "loss": 2.4101, "step": 30410 }, { "epoch": 1.0554057627842894e-13, "grad_norm": 0.5628526210784912, "learning_rate": 9.069888021185828e-05, "loss": 2.4102, "step": 30420 }, { "epoch": 1.0557527074794848e-13, "grad_norm": 0.5903543829917908, "learning_rate": 9.067828274652055e-05, "loss": 2.3514, "step": 30430 }, { "epoch": 1.0560996521746802e-13, "grad_norm": 0.5783190131187439, "learning_rate": 9.065768181341368e-05, "loss": 2.3922, "step": 30440 }, { "epoch": 1.0564465968698755e-13, "grad_norm": 0.5590732097625732, "learning_rate": 9.063707741579549e-05, "loss": 2.4003, "step": 30450 }, { "epoch": 1.0567935415650709e-13, "grad_norm": 0.5554191470146179, "learning_rate": 9.06164695569244e-05, "loss": 2.3905, "step": 30460 }, { "epoch": 1.0571404862602662e-13, "grad_norm": 0.551842987537384, "learning_rate": 9.05958582400594e-05, "loss": 2.361, "step": 30470 }, { "epoch": 1.0574874309554616e-13, "grad_norm": 0.5651971101760864, "learning_rate": 9.057524346845998e-05, "loss": 2.3454, "step": 30480 }, { "epoch": 1.057834375650657e-13, "grad_norm": 0.5786429643630981, "learning_rate": 9.055462524538616e-05, "loss": 2.3788, "step": 30490 }, { "epoch": 1.0581813203458523e-13, "grad_norm": 0.554775595664978, "learning_rate": 9.053400357409856e-05, "loss": 2.3534, "step": 30500 }, { "epoch": 1.0585282650410477e-13, "grad_norm": 0.5686996579170227, "learning_rate": 9.051337845785832e-05, "loss": 2.3761, "step": 30510 }, { "epoch": 1.058875209736243e-13, "grad_norm": 0.5658218860626221, "learning_rate": 9.04927498999271e-05, "loss": 2.3912, "step": 30520 }, { "epoch": 1.0592221544314384e-13, "grad_norm": 0.5502528548240662, "learning_rate": 9.047211790356716e-05, "loss": 2.3616, "step": 30530 }, { "epoch": 1.0595690991266338e-13, "grad_norm": 0.5522202849388123, "learning_rate": 9.045148247204126e-05, "loss": 2.4039, "step": 30540 }, { "epoch": 1.0599160438218291e-13, "grad_norm": 0.5972748398780823, "learning_rate": 9.043084360861268e-05, "loss": 2.3748, "step": 30550 }, { "epoch": 1.0602629885170245e-13, "grad_norm": 0.5639322400093079, "learning_rate": 9.041020131654535e-05, "loss": 2.4158, "step": 30560 }, { "epoch": 1.0606099332122199e-13, "grad_norm": 0.5393741726875305, "learning_rate": 9.038955559910364e-05, "loss": 2.3468, "step": 30570 }, { "epoch": 1.0609568779074152e-13, "grad_norm": 0.5779573321342468, "learning_rate": 9.036890645955245e-05, "loss": 2.3243, "step": 30580 }, { "epoch": 1.0613038226026106e-13, "grad_norm": 0.5636646151542664, "learning_rate": 9.034825390115734e-05, "loss": 2.3593, "step": 30590 }, { "epoch": 1.061650767297806e-13, "grad_norm": 0.5610026717185974, "learning_rate": 9.032759792718428e-05, "loss": 2.4044, "step": 30600 }, { "epoch": 1.0619977119930013e-13, "grad_norm": 0.6352822184562683, "learning_rate": 9.030693854089985e-05, "loss": 2.4155, "step": 30610 }, { "epoch": 1.0623446566881967e-13, "grad_norm": 0.572272539138794, "learning_rate": 9.028627574557113e-05, "loss": 2.3655, "step": 30620 }, { "epoch": 1.062691601383392e-13, "grad_norm": 0.5713704228401184, "learning_rate": 9.026560954446582e-05, "loss": 2.3706, "step": 30630 }, { "epoch": 1.0630385460785874e-13, "grad_norm": 0.5494360327720642, "learning_rate": 9.024493994085204e-05, "loss": 2.3863, "step": 30640 }, { "epoch": 1.0633854907737827e-13, "grad_norm": 0.6130688786506653, "learning_rate": 9.022426693799856e-05, "loss": 2.3962, "step": 30650 }, { "epoch": 1.0637324354689781e-13, "grad_norm": 0.5286880135536194, "learning_rate": 9.020359053917461e-05, "loss": 2.4067, "step": 30660 }, { "epoch": 1.0640793801641735e-13, "grad_norm": 0.5531871914863586, "learning_rate": 9.018291074765e-05, "loss": 2.3847, "step": 30670 }, { "epoch": 1.0644263248593688e-13, "grad_norm": 0.6084424257278442, "learning_rate": 9.016222756669502e-05, "loss": 2.4067, "step": 30680 }, { "epoch": 1.0647732695545642e-13, "grad_norm": 0.5705227255821228, "learning_rate": 9.014154099958057e-05, "loss": 2.4086, "step": 30690 }, { "epoch": 1.0651202142497596e-13, "grad_norm": 0.574177622795105, "learning_rate": 9.012085104957808e-05, "loss": 2.397, "step": 30700 }, { "epoch": 1.0654671589449549e-13, "grad_norm": 0.5892225503921509, "learning_rate": 9.010015771995946e-05, "loss": 2.3479, "step": 30710 }, { "epoch": 1.0658141036401503e-13, "grad_norm": 0.6010050177574158, "learning_rate": 9.007946101399715e-05, "loss": 2.3953, "step": 30720 }, { "epoch": 1.0661610483353456e-13, "grad_norm": 0.5574138164520264, "learning_rate": 9.00587609349642e-05, "loss": 2.3707, "step": 30730 }, { "epoch": 1.066507993030541e-13, "grad_norm": 0.5598821043968201, "learning_rate": 9.003805748613412e-05, "loss": 2.3735, "step": 30740 }, { "epoch": 1.0668549377257364e-13, "grad_norm": 0.5784913897514343, "learning_rate": 9.001735067078103e-05, "loss": 2.3821, "step": 30750 }, { "epoch": 1.0672018824209317e-13, "grad_norm": 0.5816372632980347, "learning_rate": 8.99966404921795e-05, "loss": 2.3993, "step": 30760 }, { "epoch": 1.0675488271161271e-13, "grad_norm": 0.5827204585075378, "learning_rate": 8.997592695360466e-05, "loss": 2.4094, "step": 30770 }, { "epoch": 1.0678957718113224e-13, "grad_norm": 0.5828802585601807, "learning_rate": 8.995521005833215e-05, "loss": 2.384, "step": 30780 }, { "epoch": 1.0682427165065178e-13, "grad_norm": 0.595705509185791, "learning_rate": 8.993448980963824e-05, "loss": 2.4002, "step": 30790 }, { "epoch": 1.0685896612017132e-13, "grad_norm": 0.5614132285118103, "learning_rate": 8.99137662107996e-05, "loss": 2.4011, "step": 30800 }, { "epoch": 1.0689366058969085e-13, "grad_norm": 0.5729092359542847, "learning_rate": 8.989303926509355e-05, "loss": 2.4026, "step": 30810 }, { "epoch": 1.0692835505921039e-13, "grad_norm": 0.5416492223739624, "learning_rate": 8.987230897579784e-05, "loss": 2.3924, "step": 30820 }, { "epoch": 1.0696304952872993e-13, "grad_norm": 0.5666707158088684, "learning_rate": 8.985157534619077e-05, "loss": 2.3236, "step": 30830 }, { "epoch": 1.0699774399824946e-13, "grad_norm": 0.5519704818725586, "learning_rate": 8.983083837955119e-05, "loss": 2.4038, "step": 30840 }, { "epoch": 1.07032438467769e-13, "grad_norm": 0.5536150336265564, "learning_rate": 8.98100980791585e-05, "loss": 2.3729, "step": 30850 }, { "epoch": 1.0706713293728853e-13, "grad_norm": 0.5778900384902954, "learning_rate": 8.978935444829259e-05, "loss": 2.3874, "step": 30860 }, { "epoch": 1.0710182740680807e-13, "grad_norm": 0.5734397172927856, "learning_rate": 8.976860749023386e-05, "loss": 2.3947, "step": 30870 }, { "epoch": 1.071365218763276e-13, "grad_norm": 0.5771304368972778, "learning_rate": 8.97478572082633e-05, "loss": 2.3894, "step": 30880 }, { "epoch": 1.0717121634584714e-13, "grad_norm": 0.5638551712036133, "learning_rate": 8.972710360566238e-05, "loss": 2.3848, "step": 30890 }, { "epoch": 1.0720591081536668e-13, "grad_norm": 0.593562126159668, "learning_rate": 8.97063466857131e-05, "loss": 2.3817, "step": 30900 }, { "epoch": 1.0724060528488621e-13, "grad_norm": 0.5515971183776855, "learning_rate": 8.968558645169797e-05, "loss": 2.4155, "step": 30910 }, { "epoch": 1.0727529975440575e-13, "grad_norm": 0.5685548186302185, "learning_rate": 8.966482290690007e-05, "loss": 2.3844, "step": 30920 }, { "epoch": 1.0730999422392529e-13, "grad_norm": 0.5785879492759705, "learning_rate": 8.964405605460295e-05, "loss": 2.3962, "step": 30930 }, { "epoch": 1.0734468869344482e-13, "grad_norm": 0.5598316788673401, "learning_rate": 8.962328589809072e-05, "loss": 2.4112, "step": 30940 }, { "epoch": 1.0737938316296436e-13, "grad_norm": 0.5606583952903748, "learning_rate": 8.960251244064801e-05, "loss": 2.3566, "step": 30950 }, { "epoch": 1.074140776324839e-13, "grad_norm": 0.5630648136138916, "learning_rate": 8.958173568555997e-05, "loss": 2.3652, "step": 30960 }, { "epoch": 1.0744877210200343e-13, "grad_norm": 0.5578071475028992, "learning_rate": 8.956095563611224e-05, "loss": 2.3967, "step": 30970 }, { "epoch": 1.0748346657152297e-13, "grad_norm": 0.5791213512420654, "learning_rate": 8.954017229559102e-05, "loss": 2.3451, "step": 30980 }, { "epoch": 1.075181610410425e-13, "grad_norm": 0.5782485008239746, "learning_rate": 8.951938566728302e-05, "loss": 2.3835, "step": 30990 }, { "epoch": 1.0755285551056204e-13, "grad_norm": 0.5713821053504944, "learning_rate": 8.949859575447544e-05, "loss": 2.3751, "step": 31000 }, { "epoch": 1.0758754998008158e-13, "grad_norm": 0.5358672142028809, "learning_rate": 8.947780256045606e-05, "loss": 2.4123, "step": 31010 }, { "epoch": 1.0762224444960111e-13, "grad_norm": 0.5981467962265015, "learning_rate": 8.945700608851315e-05, "loss": 2.4035, "step": 31020 }, { "epoch": 1.0765693891912065e-13, "grad_norm": 0.5974823832511902, "learning_rate": 8.943620634193544e-05, "loss": 2.4032, "step": 31030 }, { "epoch": 1.0769163338864018e-13, "grad_norm": 0.5578307509422302, "learning_rate": 8.94154033240123e-05, "loss": 2.3719, "step": 31040 }, { "epoch": 1.0772632785815972e-13, "grad_norm": 0.5836302638053894, "learning_rate": 8.939459703803349e-05, "loss": 2.3768, "step": 31050 }, { "epoch": 1.0776102232767926e-13, "grad_norm": 0.5644739866256714, "learning_rate": 8.937378748728938e-05, "loss": 2.3578, "step": 31060 }, { "epoch": 1.0779571679719879e-13, "grad_norm": 0.5683475732803345, "learning_rate": 8.935297467507079e-05, "loss": 2.3852, "step": 31070 }, { "epoch": 1.0783041126671833e-13, "grad_norm": 0.5434136390686035, "learning_rate": 8.933215860466912e-05, "loss": 2.3951, "step": 31080 }, { "epoch": 1.0786510573623787e-13, "grad_norm": 0.5574511885643005, "learning_rate": 8.931133927937625e-05, "loss": 2.3847, "step": 31090 }, { "epoch": 1.078998002057574e-13, "grad_norm": 0.568943977355957, "learning_rate": 8.929051670248454e-05, "loss": 2.3821, "step": 31100 }, { "epoch": 1.0793449467527694e-13, "grad_norm": 0.5730789303779602, "learning_rate": 8.926969087728694e-05, "loss": 2.3917, "step": 31110 }, { "epoch": 1.0796918914479647e-13, "grad_norm": 0.5377691388130188, "learning_rate": 8.924886180707688e-05, "loss": 2.4006, "step": 31120 }, { "epoch": 1.0800388361431601e-13, "grad_norm": 0.5443070530891418, "learning_rate": 8.922802949514826e-05, "loss": 2.3766, "step": 31130 }, { "epoch": 1.0803857808383555e-13, "grad_norm": 0.5462044477462769, "learning_rate": 8.920719394479557e-05, "loss": 2.4193, "step": 31140 }, { "epoch": 1.0807327255335508e-13, "grad_norm": 0.6074146628379822, "learning_rate": 8.918635515931374e-05, "loss": 2.3409, "step": 31150 }, { "epoch": 1.0810796702287462e-13, "grad_norm": 0.5731334686279297, "learning_rate": 8.916551314199824e-05, "loss": 2.3913, "step": 31160 }, { "epoch": 1.0814266149239415e-13, "grad_norm": 0.5602709054946899, "learning_rate": 8.91446678961451e-05, "loss": 2.4027, "step": 31170 }, { "epoch": 1.0817735596191369e-13, "grad_norm": 0.5641303658485413, "learning_rate": 8.91238194250508e-05, "loss": 2.4085, "step": 31180 }, { "epoch": 1.0821205043143323e-13, "grad_norm": 0.5646292567253113, "learning_rate": 8.910296773201232e-05, "loss": 2.414, "step": 31190 }, { "epoch": 1.0824674490095276e-13, "grad_norm": 0.5732901096343994, "learning_rate": 8.908211282032719e-05, "loss": 2.3992, "step": 31200 }, { "epoch": 1.082814393704723e-13, "grad_norm": 0.6132940053939819, "learning_rate": 8.906125469329344e-05, "loss": 2.3983, "step": 31210 }, { "epoch": 1.0831613383999183e-13, "grad_norm": 0.5527704954147339, "learning_rate": 8.904039335420961e-05, "loss": 2.4009, "step": 31220 }, { "epoch": 1.0835082830951137e-13, "grad_norm": 0.5895368456840515, "learning_rate": 8.901952880637472e-05, "loss": 2.3881, "step": 31230 }, { "epoch": 1.0838552277903091e-13, "grad_norm": 0.5594503879547119, "learning_rate": 8.899866105308834e-05, "loss": 2.3825, "step": 31240 }, { "epoch": 1.0842021724855044e-13, "grad_norm": 0.5545564889907837, "learning_rate": 8.897779009765053e-05, "loss": 2.3751, "step": 31250 }, { "epoch": 1.0845491171806998e-13, "grad_norm": 0.5741826891899109, "learning_rate": 8.89569159433618e-05, "loss": 2.3597, "step": 31260 }, { "epoch": 1.0848960618758952e-13, "grad_norm": 0.5649962425231934, "learning_rate": 8.893603859352328e-05, "loss": 2.3805, "step": 31270 }, { "epoch": 1.0852430065710905e-13, "grad_norm": 0.567814826965332, "learning_rate": 8.891515805143652e-05, "loss": 2.3963, "step": 31280 }, { "epoch": 1.0855899512662859e-13, "grad_norm": 0.5952401161193848, "learning_rate": 8.889427432040358e-05, "loss": 2.3816, "step": 31290 }, { "epoch": 1.0859368959614812e-13, "grad_norm": 0.5489510893821716, "learning_rate": 8.887338740372707e-05, "loss": 2.3992, "step": 31300 }, { "epoch": 1.0862838406566766e-13, "grad_norm": 0.5577434301376343, "learning_rate": 8.885249730471008e-05, "loss": 2.3915, "step": 31310 }, { "epoch": 1.086630785351872e-13, "grad_norm": 0.5940614342689514, "learning_rate": 8.883160402665616e-05, "loss": 2.3481, "step": 31320 }, { "epoch": 1.0869777300470673e-13, "grad_norm": 0.5575900673866272, "learning_rate": 8.881070757286945e-05, "loss": 2.3556, "step": 31330 }, { "epoch": 1.0873246747422627e-13, "grad_norm": 0.5578234195709229, "learning_rate": 8.878980794665452e-05, "loss": 2.372, "step": 31340 }, { "epoch": 1.087671619437458e-13, "grad_norm": 0.5650972723960876, "learning_rate": 8.876890515131644e-05, "loss": 2.3711, "step": 31350 }, { "epoch": 1.0880185641326534e-13, "grad_norm": 0.555234432220459, "learning_rate": 8.874799919016085e-05, "loss": 2.3857, "step": 31360 }, { "epoch": 1.0883655088278488e-13, "grad_norm": 0.5718128681182861, "learning_rate": 8.872709006649385e-05, "loss": 2.3657, "step": 31370 }, { "epoch": 1.0887124535230441e-13, "grad_norm": 0.5537779927253723, "learning_rate": 8.870617778362199e-05, "loss": 2.3542, "step": 31380 }, { "epoch": 1.0890593982182395e-13, "grad_norm": 0.5621500611305237, "learning_rate": 8.868526234485242e-05, "loss": 2.379, "step": 31390 }, { "epoch": 1.0894063429134349e-13, "grad_norm": 0.5707624554634094, "learning_rate": 8.866434375349269e-05, "loss": 2.3762, "step": 31400 }, { "epoch": 1.0897532876086302e-13, "grad_norm": 0.5938093662261963, "learning_rate": 8.864342201285096e-05, "loss": 2.403, "step": 31410 }, { "epoch": 1.0901002323038256e-13, "grad_norm": 0.5831393599510193, "learning_rate": 8.862249712623575e-05, "loss": 2.3705, "step": 31420 }, { "epoch": 1.090447176999021e-13, "grad_norm": 0.5766123533248901, "learning_rate": 8.860156909695617e-05, "loss": 2.3763, "step": 31430 }, { "epoch": 1.0907941216942163e-13, "grad_norm": 0.5701380372047424, "learning_rate": 8.858063792832184e-05, "loss": 2.3741, "step": 31440 }, { "epoch": 1.0911410663894117e-13, "grad_norm": 0.599669337272644, "learning_rate": 8.855970362364282e-05, "loss": 2.3617, "step": 31450 }, { "epoch": 1.091488011084607e-13, "grad_norm": 0.5691094398498535, "learning_rate": 8.853876618622973e-05, "loss": 2.3595, "step": 31460 }, { "epoch": 1.0918349557798024e-13, "grad_norm": 0.5390432476997375, "learning_rate": 8.851782561939358e-05, "loss": 2.3881, "step": 31470 }, { "epoch": 1.0921819004749977e-13, "grad_norm": 0.5649941563606262, "learning_rate": 8.849688192644596e-05, "loss": 2.3681, "step": 31480 }, { "epoch": 1.0925288451701931e-13, "grad_norm": 0.5799488425254822, "learning_rate": 8.847593511069895e-05, "loss": 2.3888, "step": 31490 }, { "epoch": 1.0928757898653885e-13, "grad_norm": 0.551538348197937, "learning_rate": 8.845498517546512e-05, "loss": 2.3946, "step": 31500 }, { "epoch": 1.0932227345605838e-13, "grad_norm": 0.5630422234535217, "learning_rate": 8.84340321240575e-05, "loss": 2.4305, "step": 31510 }, { "epoch": 1.0935696792557792e-13, "grad_norm": 0.5818496942520142, "learning_rate": 8.841307595978964e-05, "loss": 2.3741, "step": 31520 }, { "epoch": 1.0939166239509746e-13, "grad_norm": 0.6101105213165283, "learning_rate": 8.839211668597557e-05, "loss": 2.3891, "step": 31530 }, { "epoch": 1.0942635686461699e-13, "grad_norm": 0.5517274737358093, "learning_rate": 8.837115430592983e-05, "loss": 2.3744, "step": 31540 }, { "epoch": 1.0946105133413653e-13, "grad_norm": 0.5768172144889832, "learning_rate": 8.835018882296744e-05, "loss": 2.3872, "step": 31550 }, { "epoch": 1.0949574580365606e-13, "grad_norm": 0.5702220797538757, "learning_rate": 8.832922024040392e-05, "loss": 2.3779, "step": 31560 }, { "epoch": 1.095304402731756e-13, "grad_norm": 0.5445351004600525, "learning_rate": 8.830824856155524e-05, "loss": 2.3351, "step": 31570 }, { "epoch": 1.0956513474269514e-13, "grad_norm": 0.5919806957244873, "learning_rate": 8.828727378973791e-05, "loss": 2.3435, "step": 31580 }, { "epoch": 1.0959982921221467e-13, "grad_norm": 0.6309936046600342, "learning_rate": 8.826629592826891e-05, "loss": 2.3538, "step": 31590 }, { "epoch": 1.0963452368173421e-13, "grad_norm": 0.5615207552909851, "learning_rate": 8.824531498046571e-05, "loss": 2.3653, "step": 31600 }, { "epoch": 1.0966921815125374e-13, "grad_norm": 0.596936047077179, "learning_rate": 8.822433094964627e-05, "loss": 2.368, "step": 31610 }, { "epoch": 1.0970391262077328e-13, "grad_norm": 0.5547690987586975, "learning_rate": 8.820334383912901e-05, "loss": 2.4169, "step": 31620 }, { "epoch": 1.0973860709029282e-13, "grad_norm": 0.6111728549003601, "learning_rate": 8.818235365223288e-05, "loss": 2.3573, "step": 31630 }, { "epoch": 1.0977330155981235e-13, "grad_norm": 0.5731184482574463, "learning_rate": 8.81613603922773e-05, "loss": 2.3947, "step": 31640 }, { "epoch": 1.0980799602933189e-13, "grad_norm": 0.6165626049041748, "learning_rate": 8.814036406258217e-05, "loss": 2.4145, "step": 31650 }, { "epoch": 1.0984269049885143e-13, "grad_norm": 0.5838724374771118, "learning_rate": 8.811936466646787e-05, "loss": 2.3859, "step": 31660 }, { "epoch": 1.0987738496837096e-13, "grad_norm": 0.5787611603736877, "learning_rate": 8.80983622072553e-05, "loss": 2.3965, "step": 31670 }, { "epoch": 1.099120794378905e-13, "grad_norm": 0.5470767617225647, "learning_rate": 8.807735668826577e-05, "loss": 2.3763, "step": 31680 }, { "epoch": 1.0994677390741003e-13, "grad_norm": 0.5903882384300232, "learning_rate": 8.805634811282117e-05, "loss": 2.3973, "step": 31690 }, { "epoch": 1.0998146837692957e-13, "grad_norm": 0.5805817246437073, "learning_rate": 8.80353364842438e-05, "loss": 2.4102, "step": 31700 }, { "epoch": 1.100161628464491e-13, "grad_norm": 0.5695959329605103, "learning_rate": 8.801432180585647e-05, "loss": 2.3986, "step": 31710 }, { "epoch": 1.1005085731596864e-13, "grad_norm": 0.5719563364982605, "learning_rate": 8.799330408098249e-05, "loss": 2.4126, "step": 31720 }, { "epoch": 1.1008555178548818e-13, "grad_norm": 0.5945616960525513, "learning_rate": 8.797228331294562e-05, "loss": 2.4022, "step": 31730 }, { "epoch": 1.1012024625500771e-13, "grad_norm": 0.590654730796814, "learning_rate": 8.795125950507009e-05, "loss": 2.4028, "step": 31740 }, { "epoch": 1.1015494072452725e-13, "grad_norm": 0.5769381523132324, "learning_rate": 8.793023266068065e-05, "loss": 2.3931, "step": 31750 }, { "epoch": 1.1018963519404679e-13, "grad_norm": 0.5770150423049927, "learning_rate": 8.790920278310254e-05, "loss": 2.4394, "step": 31760 }, { "epoch": 1.1022432966356632e-13, "grad_norm": 0.5763756632804871, "learning_rate": 8.788816987566143e-05, "loss": 2.3884, "step": 31770 }, { "epoch": 1.1025902413308586e-13, "grad_norm": 0.5372884273529053, "learning_rate": 8.786713394168349e-05, "loss": 2.3753, "step": 31780 }, { "epoch": 1.102937186026054e-13, "grad_norm": 0.5712601542472839, "learning_rate": 8.784609498449539e-05, "loss": 2.414, "step": 31790 }, { "epoch": 1.1032841307212493e-13, "grad_norm": 0.5619747638702393, "learning_rate": 8.782505300742424e-05, "loss": 2.4083, "step": 31800 }, { "epoch": 1.1036310754164447e-13, "grad_norm": 0.5649058818817139, "learning_rate": 8.780400801379765e-05, "loss": 2.3731, "step": 31810 }, { "epoch": 1.10397802011164e-13, "grad_norm": 0.5856074094772339, "learning_rate": 8.77829600069437e-05, "loss": 2.3451, "step": 31820 }, { "epoch": 1.1043249648068354e-13, "grad_norm": 0.5645027756690979, "learning_rate": 8.776190899019098e-05, "loss": 2.3465, "step": 31830 }, { "epoch": 1.1046719095020308e-13, "grad_norm": 0.6040491461753845, "learning_rate": 8.774085496686851e-05, "loss": 2.3454, "step": 31840 }, { "epoch": 1.1050188541972261e-13, "grad_norm": 0.5919221043586731, "learning_rate": 8.77197979403058e-05, "loss": 2.3897, "step": 31850 }, { "epoch": 1.1053657988924215e-13, "grad_norm": 0.5706390142440796, "learning_rate": 8.769873791383283e-05, "loss": 2.3742, "step": 31860 }, { "epoch": 1.1057127435876168e-13, "grad_norm": 0.5722496509552002, "learning_rate": 8.767767489078008e-05, "loss": 2.4062, "step": 31870 }, { "epoch": 1.1060596882828122e-13, "grad_norm": 0.5583400726318359, "learning_rate": 8.765660887447848e-05, "loss": 2.3883, "step": 31880 }, { "epoch": 1.1064066329780076e-13, "grad_norm": 0.5742788314819336, "learning_rate": 8.763553986825943e-05, "loss": 2.4304, "step": 31890 }, { "epoch": 1.1067535776732029e-13, "grad_norm": 0.5684919953346252, "learning_rate": 8.761446787545482e-05, "loss": 2.382, "step": 31900 }, { "epoch": 1.1071005223683983e-13, "grad_norm": 0.5926767587661743, "learning_rate": 8.759339289939698e-05, "loss": 2.3866, "step": 31910 }, { "epoch": 1.1074474670635936e-13, "grad_norm": 0.5687896609306335, "learning_rate": 8.757231494341879e-05, "loss": 2.3857, "step": 31920 }, { "epoch": 1.107794411758789e-13, "grad_norm": 0.5668625831604004, "learning_rate": 8.755123401085352e-05, "loss": 2.373, "step": 31930 }, { "epoch": 1.1081413564539844e-13, "grad_norm": 0.5736902952194214, "learning_rate": 8.753015010503491e-05, "loss": 2.3967, "step": 31940 }, { "epoch": 1.1084883011491797e-13, "grad_norm": 0.5493249297142029, "learning_rate": 8.750906322929724e-05, "loss": 2.3805, "step": 31950 }, { "epoch": 1.1088352458443751e-13, "grad_norm": 0.5499359369277954, "learning_rate": 8.748797338697517e-05, "loss": 2.4001, "step": 31960 }, { "epoch": 1.1091821905395705e-13, "grad_norm": 0.5552942752838135, "learning_rate": 8.746688058140394e-05, "loss": 2.3829, "step": 31970 }, { "epoch": 1.1095291352347658e-13, "grad_norm": 0.5958216190338135, "learning_rate": 8.744578481591912e-05, "loss": 2.3676, "step": 31980 }, { "epoch": 1.1098760799299612e-13, "grad_norm": 0.5843913555145264, "learning_rate": 8.74246860938569e-05, "loss": 2.386, "step": 31990 }, { "epoch": 1.1102230246251565e-13, "grad_norm": 0.5784711837768555, "learning_rate": 8.740358441855382e-05, "loss": 2.4122, "step": 32000 }, { "epoch": 1.1105699693203519e-13, "grad_norm": 0.5630524754524231, "learning_rate": 8.738247979334694e-05, "loss": 2.385, "step": 32010 }, { "epoch": 1.1109169140155473e-13, "grad_norm": 0.5928635597229004, "learning_rate": 8.736137222157376e-05, "loss": 2.387, "step": 32020 }, { "epoch": 1.1112638587107426e-13, "grad_norm": 0.6069881916046143, "learning_rate": 8.734026170657228e-05, "loss": 2.3804, "step": 32030 }, { "epoch": 1.111610803405938e-13, "grad_norm": 0.5936369299888611, "learning_rate": 8.731914825168093e-05, "loss": 2.3539, "step": 32040 }, { "epoch": 1.1119577481011333e-13, "grad_norm": 0.5649336576461792, "learning_rate": 8.729803186023861e-05, "loss": 2.3715, "step": 32050 }, { "epoch": 1.1123046927963287e-13, "grad_norm": 0.554183840751648, "learning_rate": 8.727691253558474e-05, "loss": 2.3794, "step": 32060 }, { "epoch": 1.1126516374915241e-13, "grad_norm": 0.5575531125068665, "learning_rate": 8.725579028105912e-05, "loss": 2.3488, "step": 32070 }, { "epoch": 1.1129985821867194e-13, "grad_norm": 0.56158047914505, "learning_rate": 8.723466510000209e-05, "loss": 2.3818, "step": 32080 }, { "epoch": 1.1133455268819148e-13, "grad_norm": 0.5541004538536072, "learning_rate": 8.721353699575436e-05, "loss": 2.3997, "step": 32090 }, { "epoch": 1.1136924715771102e-13, "grad_norm": 0.5747383236885071, "learning_rate": 8.719240597165721e-05, "loss": 2.4094, "step": 32100 }, { "epoch": 1.1140394162723055e-13, "grad_norm": 0.5577704906463623, "learning_rate": 8.717127203105232e-05, "loss": 2.3602, "step": 32110 }, { "epoch": 1.1143863609675009e-13, "grad_norm": 0.549311637878418, "learning_rate": 8.715013517728181e-05, "loss": 2.3508, "step": 32120 }, { "epoch": 1.1147333056626962e-13, "grad_norm": 0.5414339900016785, "learning_rate": 8.712899541368831e-05, "loss": 2.3625, "step": 32130 }, { "epoch": 1.1150802503578916e-13, "grad_norm": 0.5690591931343079, "learning_rate": 8.710785274361489e-05, "loss": 2.3549, "step": 32140 }, { "epoch": 1.115427195053087e-13, "grad_norm": 0.5879653692245483, "learning_rate": 8.708670717040509e-05, "loss": 2.3908, "step": 32150 }, { "epoch": 1.1157741397482823e-13, "grad_norm": 0.5840421319007874, "learning_rate": 8.70655586974029e-05, "loss": 2.3702, "step": 32160 }, { "epoch": 1.1161210844434777e-13, "grad_norm": 0.5805444717407227, "learning_rate": 8.704440732795275e-05, "loss": 2.4009, "step": 32170 }, { "epoch": 1.116468029138673e-13, "grad_norm": 0.5801488161087036, "learning_rate": 8.702325306539958e-05, "loss": 2.3605, "step": 32180 }, { "epoch": 1.1168149738338684e-13, "grad_norm": 0.5880564451217651, "learning_rate": 8.700209591308873e-05, "loss": 2.3549, "step": 32190 }, { "epoch": 1.1171619185290638e-13, "grad_norm": 0.5652386546134949, "learning_rate": 8.698093587436602e-05, "loss": 2.3677, "step": 32200 }, { "epoch": 1.1175088632242591e-13, "grad_norm": 0.6293382048606873, "learning_rate": 8.695977295257774e-05, "loss": 2.411, "step": 32210 }, { "epoch": 1.1178558079194545e-13, "grad_norm": 0.7981392741203308, "learning_rate": 8.69386071510706e-05, "loss": 2.3732, "step": 32220 }, { "epoch": 1.1182027526146499e-13, "grad_norm": 0.5790445804595947, "learning_rate": 8.691743847319182e-05, "loss": 2.3659, "step": 32230 }, { "epoch": 1.1185496973098452e-13, "grad_norm": 0.5723902583122253, "learning_rate": 8.689626692228904e-05, "loss": 2.3897, "step": 32240 }, { "epoch": 1.1188966420050406e-13, "grad_norm": 0.5546213388442993, "learning_rate": 8.687509250171034e-05, "loss": 2.3776, "step": 32250 }, { "epoch": 1.119243586700236e-13, "grad_norm": 0.6647869944572449, "learning_rate": 8.68539152148043e-05, "loss": 2.3728, "step": 32260 }, { "epoch": 1.1195905313954313e-13, "grad_norm": 0.608788251876831, "learning_rate": 8.683273506491989e-05, "loss": 2.3798, "step": 32270 }, { "epoch": 1.1199374760906267e-13, "grad_norm": 0.5694875121116638, "learning_rate": 8.681155205540659e-05, "loss": 2.3442, "step": 32280 }, { "epoch": 1.120284420785822e-13, "grad_norm": 0.5698866844177246, "learning_rate": 8.67903661896143e-05, "loss": 2.4246, "step": 32290 }, { "epoch": 1.1206313654810174e-13, "grad_norm": 0.6154861450195312, "learning_rate": 8.67691774708934e-05, "loss": 2.3541, "step": 32300 }, { "epoch": 1.1209783101762127e-13, "grad_norm": 0.5652937889099121, "learning_rate": 8.674798590259469e-05, "loss": 2.3672, "step": 32310 }, { "epoch": 1.1213252548714081e-13, "grad_norm": 0.5743907690048218, "learning_rate": 8.672679148806944e-05, "loss": 2.3894, "step": 32320 }, { "epoch": 1.1216721995666035e-13, "grad_norm": 0.5601689219474792, "learning_rate": 8.670559423066935e-05, "loss": 2.3321, "step": 32330 }, { "epoch": 1.1220191442617988e-13, "grad_norm": 0.5481255650520325, "learning_rate": 8.668439413374661e-05, "loss": 2.3629, "step": 32340 }, { "epoch": 1.1223660889569942e-13, "grad_norm": 0.5766456127166748, "learning_rate": 8.666319120065381e-05, "loss": 2.4033, "step": 32350 }, { "epoch": 1.1227130336521896e-13, "grad_norm": 0.5813055038452148, "learning_rate": 8.664198543474404e-05, "loss": 2.3795, "step": 32360 }, { "epoch": 1.1230599783473849e-13, "grad_norm": 0.5409284830093384, "learning_rate": 8.662077683937075e-05, "loss": 2.3735, "step": 32370 }, { "epoch": 1.1234069230425803e-13, "grad_norm": 0.5686102509498596, "learning_rate": 8.659956541788798e-05, "loss": 2.4167, "step": 32380 }, { "epoch": 1.1237538677377756e-13, "grad_norm": 0.6061505675315857, "learning_rate": 8.657835117365005e-05, "loss": 2.3646, "step": 32390 }, { "epoch": 1.124100812432971e-13, "grad_norm": 0.5626291036605835, "learning_rate": 8.655713411001186e-05, "loss": 2.3771, "step": 32400 }, { "epoch": 1.1244477571281664e-13, "grad_norm": 0.570082426071167, "learning_rate": 8.653591423032871e-05, "loss": 2.3735, "step": 32410 }, { "epoch": 1.1247947018233617e-13, "grad_norm": 0.562591552734375, "learning_rate": 8.651469153795633e-05, "loss": 2.4073, "step": 32420 }, { "epoch": 1.1251416465185571e-13, "grad_norm": 0.5694138407707214, "learning_rate": 8.649346603625089e-05, "loss": 2.3652, "step": 32430 }, { "epoch": 1.1254885912137524e-13, "grad_norm": 0.573952317237854, "learning_rate": 8.647223772856902e-05, "loss": 2.375, "step": 32440 }, { "epoch": 1.1258355359089478e-13, "grad_norm": 0.5744211673736572, "learning_rate": 8.645100661826783e-05, "loss": 2.3483, "step": 32450 }, { "epoch": 1.1261824806041432e-13, "grad_norm": 0.5952160954475403, "learning_rate": 8.64297727087048e-05, "loss": 2.3537, "step": 32460 }, { "epoch": 1.1265294252993385e-13, "grad_norm": 0.5928564667701721, "learning_rate": 8.640853600323792e-05, "loss": 2.3718, "step": 32470 }, { "epoch": 1.1268763699945339e-13, "grad_norm": 0.5432682037353516, "learning_rate": 8.638729650522556e-05, "loss": 2.378, "step": 32480 }, { "epoch": 1.1272233146897292e-13, "grad_norm": 0.5678886771202087, "learning_rate": 8.636605421802659e-05, "loss": 2.3977, "step": 32490 }, { "epoch": 1.1275702593849246e-13, "grad_norm": 0.5574220418930054, "learning_rate": 8.634480914500028e-05, "loss": 2.3523, "step": 32500 }, { "epoch": 1.12791720408012e-13, "grad_norm": 0.5736576914787292, "learning_rate": 8.632356128950638e-05, "loss": 2.3836, "step": 32510 }, { "epoch": 1.1282641487753153e-13, "grad_norm": 0.5750691890716553, "learning_rate": 8.6302310654905e-05, "loss": 2.3887, "step": 32520 }, { "epoch": 1.1286110934705107e-13, "grad_norm": 0.5645492672920227, "learning_rate": 8.628105724455681e-05, "loss": 2.3882, "step": 32530 }, { "epoch": 1.128958038165706e-13, "grad_norm": 0.5457736849784851, "learning_rate": 8.625980106182282e-05, "loss": 2.391, "step": 32540 }, { "epoch": 1.1293049828609014e-13, "grad_norm": 0.5672537088394165, "learning_rate": 8.62385421100645e-05, "loss": 2.381, "step": 32550 }, { "epoch": 1.1296519275560968e-13, "grad_norm": 0.5523789525032043, "learning_rate": 8.62172803926438e-05, "loss": 2.3571, "step": 32560 }, { "epoch": 1.1299988722512921e-13, "grad_norm": 0.5570961236953735, "learning_rate": 8.619601591292308e-05, "loss": 2.3898, "step": 32570 }, { "epoch": 1.1303458169464875e-13, "grad_norm": 0.5669542551040649, "learning_rate": 8.617474867426511e-05, "loss": 2.4106, "step": 32580 }, { "epoch": 1.1306927616416829e-13, "grad_norm": 0.5681347250938416, "learning_rate": 8.615347868003312e-05, "loss": 2.3626, "step": 32590 }, { "epoch": 1.1310397063368782e-13, "grad_norm": 0.5616282820701599, "learning_rate": 8.61322059335908e-05, "loss": 2.3801, "step": 32600 }, { "epoch": 1.1313866510320736e-13, "grad_norm": 0.5664324760437012, "learning_rate": 8.611093043830224e-05, "loss": 2.3804, "step": 32610 }, { "epoch": 1.131733595727269e-13, "grad_norm": 0.5464318990707397, "learning_rate": 8.608965219753195e-05, "loss": 2.3803, "step": 32620 }, { "epoch": 1.1320805404224643e-13, "grad_norm": 0.5551280379295349, "learning_rate": 8.606837121464495e-05, "loss": 2.3466, "step": 32630 }, { "epoch": 1.1324274851176597e-13, "grad_norm": 0.5611832737922668, "learning_rate": 8.604708749300661e-05, "loss": 2.3467, "step": 32640 }, { "epoch": 1.132774429812855e-13, "grad_norm": 0.594584047794342, "learning_rate": 8.602580103598279e-05, "loss": 2.3832, "step": 32650 }, { "epoch": 1.1331213745080504e-13, "grad_norm": 0.5888410210609436, "learning_rate": 8.600451184693974e-05, "loss": 2.358, "step": 32660 }, { "epoch": 1.1334683192032458e-13, "grad_norm": 0.593937337398529, "learning_rate": 8.598321992924418e-05, "loss": 2.3667, "step": 32670 }, { "epoch": 1.1338152638984411e-13, "grad_norm": 0.554601788520813, "learning_rate": 8.59619252862632e-05, "loss": 2.3895, "step": 32680 }, { "epoch": 1.1341622085936365e-13, "grad_norm": 0.5715033411979675, "learning_rate": 8.594062792136442e-05, "loss": 2.3862, "step": 32690 }, { "epoch": 1.1345091532888318e-13, "grad_norm": 0.553842306137085, "learning_rate": 8.59193278379158e-05, "loss": 2.3329, "step": 32700 }, { "epoch": 1.1348560979840272e-13, "grad_norm": 0.5777182579040527, "learning_rate": 8.589802503928579e-05, "loss": 2.3505, "step": 32710 }, { "epoch": 1.1352030426792226e-13, "grad_norm": 0.5679306983947754, "learning_rate": 8.587671952884322e-05, "loss": 2.4019, "step": 32720 }, { "epoch": 1.1355499873744179e-13, "grad_norm": 0.5512667894363403, "learning_rate": 8.58554113099574e-05, "loss": 2.3861, "step": 32730 }, { "epoch": 1.1358969320696133e-13, "grad_norm": 0.5635586977005005, "learning_rate": 8.583410038599802e-05, "loss": 2.3446, "step": 32740 }, { "epoch": 1.1362438767648086e-13, "grad_norm": 0.608222246170044, "learning_rate": 8.581278676033522e-05, "loss": 2.395, "step": 32750 }, { "epoch": 1.136590821460004e-13, "grad_norm": 0.5511956810951233, "learning_rate": 8.579147043633959e-05, "loss": 2.384, "step": 32760 }, { "epoch": 1.1369377661551994e-13, "grad_norm": 0.5502623915672302, "learning_rate": 8.577015141738209e-05, "loss": 2.3761, "step": 32770 }, { "epoch": 1.1372847108503947e-13, "grad_norm": 0.571134626865387, "learning_rate": 8.574882970683413e-05, "loss": 2.4155, "step": 32780 }, { "epoch": 1.13763165554559e-13, "grad_norm": 0.5424155592918396, "learning_rate": 8.572750530806762e-05, "loss": 2.3287, "step": 32790 }, { "epoch": 1.1379786002407855e-13, "grad_norm": 0.6013025045394897, "learning_rate": 8.570617822445478e-05, "loss": 2.4056, "step": 32800 }, { "epoch": 1.1383255449359808e-13, "grad_norm": 0.5979651212692261, "learning_rate": 8.568484845936829e-05, "loss": 2.3563, "step": 32810 }, { "epoch": 1.1386724896311762e-13, "grad_norm": 0.6004801988601685, "learning_rate": 8.56635160161813e-05, "loss": 2.3615, "step": 32820 }, { "epoch": 1.1390194343263715e-13, "grad_norm": 0.572044849395752, "learning_rate": 8.564218089826736e-05, "loss": 2.3782, "step": 32830 }, { "epoch": 1.139366379021567e-13, "grad_norm": 0.5616601705551147, "learning_rate": 8.562084310900042e-05, "loss": 2.3996, "step": 32840 }, { "epoch": 1.1397133237167623e-13, "grad_norm": 0.5803918838500977, "learning_rate": 8.559950265175485e-05, "loss": 2.3886, "step": 32850 }, { "epoch": 1.1400602684119576e-13, "grad_norm": 0.6064916253089905, "learning_rate": 8.55781595299055e-05, "loss": 2.3767, "step": 32860 }, { "epoch": 1.140407213107153e-13, "grad_norm": 0.5482872724533081, "learning_rate": 8.555681374682754e-05, "loss": 2.3694, "step": 32870 }, { "epoch": 1.1407541578023483e-13, "grad_norm": 0.5691523551940918, "learning_rate": 8.553546530589669e-05, "loss": 2.3861, "step": 32880 }, { "epoch": 1.1411011024975437e-13, "grad_norm": 0.5939934253692627, "learning_rate": 8.551411421048899e-05, "loss": 2.4058, "step": 32890 }, { "epoch": 1.141448047192739e-13, "grad_norm": 0.5736396908760071, "learning_rate": 8.549276046398093e-05, "loss": 2.3439, "step": 32900 }, { "epoch": 1.1417949918879344e-13, "grad_norm": 0.5856161713600159, "learning_rate": 8.547140406974943e-05, "loss": 2.3766, "step": 32910 }, { "epoch": 1.1421419365831298e-13, "grad_norm": 0.5442352294921875, "learning_rate": 8.54500450311718e-05, "loss": 2.3731, "step": 32920 }, { "epoch": 1.1424888812783252e-13, "grad_norm": 0.5523111820220947, "learning_rate": 8.542868335162581e-05, "loss": 2.3418, "step": 32930 }, { "epoch": 1.1428358259735205e-13, "grad_norm": 0.5798482298851013, "learning_rate": 8.540731903448963e-05, "loss": 2.3403, "step": 32940 }, { "epoch": 1.143182770668716e-13, "grad_norm": 0.5478524565696716, "learning_rate": 8.538595208314181e-05, "loss": 2.3222, "step": 32950 }, { "epoch": 1.1435297153639112e-13, "grad_norm": 0.5681385397911072, "learning_rate": 8.53645825009614e-05, "loss": 2.4055, "step": 32960 }, { "epoch": 1.1438766600591066e-13, "grad_norm": 0.5750191807746887, "learning_rate": 8.534321029132777e-05, "loss": 2.4006, "step": 32970 }, { "epoch": 1.144223604754302e-13, "grad_norm": 0.5710131525993347, "learning_rate": 8.532183545762077e-05, "loss": 2.3724, "step": 32980 }, { "epoch": 1.1445705494494973e-13, "grad_norm": 0.5715236067771912, "learning_rate": 8.530045800322065e-05, "loss": 2.3867, "step": 32990 }, { "epoch": 1.1449174941446927e-13, "grad_norm": 0.5970969796180725, "learning_rate": 8.527907793150806e-05, "loss": 2.3862, "step": 33000 }, { "epoch": 1.145264438839888e-13, "grad_norm": 0.551402747631073, "learning_rate": 8.525769524586408e-05, "loss": 2.4107, "step": 33010 }, { "epoch": 1.1456113835350834e-13, "grad_norm": 0.5816943645477295, "learning_rate": 8.523630994967019e-05, "loss": 2.3902, "step": 33020 }, { "epoch": 1.1459583282302788e-13, "grad_norm": 0.5522927045822144, "learning_rate": 8.52149220463083e-05, "loss": 2.3462, "step": 33030 }, { "epoch": 1.1463052729254741e-13, "grad_norm": 0.5651223063468933, "learning_rate": 8.51935315391607e-05, "loss": 2.3858, "step": 33040 }, { "epoch": 1.1466522176206695e-13, "grad_norm": 0.6016945838928223, "learning_rate": 8.517213843161017e-05, "loss": 2.3952, "step": 33050 }, { "epoch": 1.1469991623158649e-13, "grad_norm": 0.5603168606758118, "learning_rate": 8.515074272703979e-05, "loss": 2.383, "step": 33060 }, { "epoch": 1.1473461070110602e-13, "grad_norm": 0.6064496636390686, "learning_rate": 8.512934442883313e-05, "loss": 2.3663, "step": 33070 }, { "epoch": 1.1476930517062556e-13, "grad_norm": 0.5606302618980408, "learning_rate": 8.510794354037414e-05, "loss": 2.3796, "step": 33080 }, { "epoch": 1.148039996401451e-13, "grad_norm": 0.5702493786811829, "learning_rate": 8.508654006504719e-05, "loss": 2.3908, "step": 33090 }, { "epoch": 1.1483869410966463e-13, "grad_norm": 0.5777590870857239, "learning_rate": 8.506513400623706e-05, "loss": 2.3897, "step": 33100 }, { "epoch": 1.1487338857918417e-13, "grad_norm": 0.5714629292488098, "learning_rate": 8.504372536732892e-05, "loss": 2.3817, "step": 33110 }, { "epoch": 1.149080830487037e-13, "grad_norm": 0.591336727142334, "learning_rate": 8.502231415170839e-05, "loss": 2.3827, "step": 33120 }, { "epoch": 1.1494277751822324e-13, "grad_norm": 0.6828224062919617, "learning_rate": 8.50009003627614e-05, "loss": 2.3842, "step": 33130 }, { "epoch": 1.1497747198774277e-13, "grad_norm": 0.5781036615371704, "learning_rate": 8.497948400387447e-05, "loss": 2.3846, "step": 33140 }, { "epoch": 1.150121664572623e-13, "grad_norm": 0.5569795966148376, "learning_rate": 8.495806507843431e-05, "loss": 2.3438, "step": 33150 }, { "epoch": 1.1504686092678185e-13, "grad_norm": 0.5922937989234924, "learning_rate": 8.493664358982819e-05, "loss": 2.3495, "step": 33160 }, { "epoch": 1.1508155539630138e-13, "grad_norm": 0.5859269499778748, "learning_rate": 8.491521954144371e-05, "loss": 2.3695, "step": 33170 }, { "epoch": 1.1511624986582092e-13, "grad_norm": 0.5760020613670349, "learning_rate": 8.489379293666894e-05, "loss": 2.3248, "step": 33180 }, { "epoch": 1.1515094433534045e-13, "grad_norm": 0.5429837703704834, "learning_rate": 8.487236377889223e-05, "loss": 2.3752, "step": 33190 }, { "epoch": 1.1518563880486e-13, "grad_norm": 0.5607858300209045, "learning_rate": 8.485093207150253e-05, "loss": 2.3535, "step": 33200 }, { "epoch": 1.1522033327437953e-13, "grad_norm": 0.633112370967865, "learning_rate": 8.4829497817889e-05, "loss": 2.3384, "step": 33210 }, { "epoch": 1.1525502774389906e-13, "grad_norm": 0.59754878282547, "learning_rate": 8.48080610214413e-05, "loss": 2.381, "step": 33220 }, { "epoch": 1.152897222134186e-13, "grad_norm": 0.5716890692710876, "learning_rate": 8.478662168554949e-05, "loss": 2.3739, "step": 33230 }, { "epoch": 1.1532441668293814e-13, "grad_norm": 0.5771068930625916, "learning_rate": 8.476517981360397e-05, "loss": 2.3853, "step": 33240 }, { "epoch": 1.1535911115245767e-13, "grad_norm": 0.578281819820404, "learning_rate": 8.474373540899563e-05, "loss": 2.347, "step": 33250 }, { "epoch": 1.153938056219772e-13, "grad_norm": 0.5567849278450012, "learning_rate": 8.472228847511575e-05, "loss": 2.3935, "step": 33260 }, { "epoch": 1.1542850009149674e-13, "grad_norm": 0.5511301755905151, "learning_rate": 8.47008390153559e-05, "loss": 2.3696, "step": 33270 }, { "epoch": 1.1546319456101628e-13, "grad_norm": 0.5675747990608215, "learning_rate": 8.467938703310818e-05, "loss": 2.3979, "step": 33280 }, { "epoch": 1.1549788903053582e-13, "grad_norm": 0.5724592208862305, "learning_rate": 8.465793253176501e-05, "loss": 2.339, "step": 33290 }, { "epoch": 1.1553258350005535e-13, "grad_norm": 0.5579516291618347, "learning_rate": 8.463647551471926e-05, "loss": 2.3748, "step": 33300 }, { "epoch": 1.155672779695749e-13, "grad_norm": 0.5569655895233154, "learning_rate": 8.461501598536417e-05, "loss": 2.3581, "step": 33310 }, { "epoch": 1.1560197243909442e-13, "grad_norm": 0.5420556664466858, "learning_rate": 8.459355394709334e-05, "loss": 2.4042, "step": 33320 }, { "epoch": 1.1563666690861396e-13, "grad_norm": 0.5649698972702026, "learning_rate": 8.457208940330086e-05, "loss": 2.3443, "step": 33330 }, { "epoch": 1.156713613781335e-13, "grad_norm": 0.5542807579040527, "learning_rate": 8.455062235738111e-05, "loss": 2.3682, "step": 33340 }, { "epoch": 1.1570605584765303e-13, "grad_norm": 0.5550895929336548, "learning_rate": 8.452915281272896e-05, "loss": 2.3624, "step": 33350 }, { "epoch": 1.1574075031717257e-13, "grad_norm": 0.5680784583091736, "learning_rate": 8.450768077273964e-05, "loss": 2.3676, "step": 33360 }, { "epoch": 1.157754447866921e-13, "grad_norm": 0.5524219274520874, "learning_rate": 8.448620624080874e-05, "loss": 2.3362, "step": 33370 }, { "epoch": 1.1581013925621164e-13, "grad_norm": 0.5754779577255249, "learning_rate": 8.446472922033227e-05, "loss": 2.3422, "step": 33380 }, { "epoch": 1.1584483372573118e-13, "grad_norm": 0.5706854462623596, "learning_rate": 8.444324971470665e-05, "loss": 2.3914, "step": 33390 }, { "epoch": 1.1587952819525071e-13, "grad_norm": 0.5845461487770081, "learning_rate": 8.442176772732867e-05, "loss": 2.3917, "step": 33400 }, { "epoch": 1.1591422266477025e-13, "grad_norm": 0.5565882921218872, "learning_rate": 8.440028326159554e-05, "loss": 2.3623, "step": 33410 }, { "epoch": 1.1594891713428979e-13, "grad_norm": 0.5736425518989563, "learning_rate": 8.437879632090484e-05, "loss": 2.3874, "step": 33420 }, { "epoch": 1.1598361160380932e-13, "grad_norm": 0.551762580871582, "learning_rate": 8.435730690865452e-05, "loss": 2.3382, "step": 33430 }, { "epoch": 1.1601830607332886e-13, "grad_norm": 0.5769217610359192, "learning_rate": 8.433581502824296e-05, "loss": 2.3565, "step": 33440 }, { "epoch": 1.160530005428484e-13, "grad_norm": 0.5778464674949646, "learning_rate": 8.43143206830689e-05, "loss": 2.3878, "step": 33450 }, { "epoch": 1.1608769501236793e-13, "grad_norm": 0.5760593414306641, "learning_rate": 8.429282387653153e-05, "loss": 2.3805, "step": 33460 }, { "epoch": 1.1612238948188747e-13, "grad_norm": 0.6021638512611389, "learning_rate": 8.427132461203035e-05, "loss": 2.3893, "step": 33470 }, { "epoch": 1.16157083951407e-13, "grad_norm": 0.6395330429077148, "learning_rate": 8.424982289296529e-05, "loss": 2.3502, "step": 33480 }, { "epoch": 1.1619177842092654e-13, "grad_norm": 0.599888801574707, "learning_rate": 8.422831872273667e-05, "loss": 2.4097, "step": 33490 }, { "epoch": 1.1622647289044608e-13, "grad_norm": 0.6075834035873413, "learning_rate": 8.420681210474516e-05, "loss": 2.3824, "step": 33500 }, { "epoch": 1.162611673599656e-13, "grad_norm": 0.5773367881774902, "learning_rate": 8.41853030423919e-05, "loss": 2.378, "step": 33510 }, { "epoch": 1.1629586182948515e-13, "grad_norm": 0.5782725214958191, "learning_rate": 8.416379153907831e-05, "loss": 2.3523, "step": 33520 }, { "epoch": 1.1633055629900468e-13, "grad_norm": 0.56966632604599, "learning_rate": 8.41422775982063e-05, "loss": 2.3617, "step": 33530 }, { "epoch": 1.1636525076852422e-13, "grad_norm": 0.5540698766708374, "learning_rate": 8.412076122317805e-05, "loss": 2.3598, "step": 33540 }, { "epoch": 1.1639994523804376e-13, "grad_norm": 0.6346203088760376, "learning_rate": 8.409924241739625e-05, "loss": 2.3645, "step": 33550 }, { "epoch": 1.164346397075633e-13, "grad_norm": 0.6021291017532349, "learning_rate": 8.407772118426388e-05, "loss": 2.3568, "step": 33560 }, { "epoch": 1.1646933417708283e-13, "grad_norm": 0.586383044719696, "learning_rate": 8.405619752718433e-05, "loss": 2.3971, "step": 33570 }, { "epoch": 1.1650402864660236e-13, "grad_norm": 0.6154686808586121, "learning_rate": 8.403467144956141e-05, "loss": 2.4014, "step": 33580 }, { "epoch": 1.165387231161219e-13, "grad_norm": 0.5670958161354065, "learning_rate": 8.401314295479927e-05, "loss": 2.4008, "step": 33590 }, { "epoch": 1.1657341758564144e-13, "grad_norm": 0.55140221118927, "learning_rate": 8.399161204630248e-05, "loss": 2.3289, "step": 33600 }, { "epoch": 1.1660811205516097e-13, "grad_norm": 0.5859328508377075, "learning_rate": 8.39700787274759e-05, "loss": 2.3777, "step": 33610 }, { "epoch": 1.166428065246805e-13, "grad_norm": 0.5615634918212891, "learning_rate": 8.394854300172493e-05, "loss": 2.3779, "step": 33620 }, { "epoch": 1.1667750099420005e-13, "grad_norm": 0.6059913635253906, "learning_rate": 8.39270048724552e-05, "loss": 2.3889, "step": 33630 }, { "epoch": 1.1671219546371958e-13, "grad_norm": 0.578209638595581, "learning_rate": 8.390546434307277e-05, "loss": 2.4085, "step": 33640 }, { "epoch": 1.1674688993323912e-13, "grad_norm": 0.5496302247047424, "learning_rate": 8.388392141698413e-05, "loss": 2.3412, "step": 33650 }, { "epoch": 1.1678158440275865e-13, "grad_norm": 0.5771005153656006, "learning_rate": 8.38623760975961e-05, "loss": 2.3842, "step": 33660 }, { "epoch": 1.168162788722782e-13, "grad_norm": 0.576881468296051, "learning_rate": 8.384082838831588e-05, "loss": 2.328, "step": 33670 }, { "epoch": 1.1685097334179773e-13, "grad_norm": 0.5979596376419067, "learning_rate": 8.381927829255104e-05, "loss": 2.3809, "step": 33680 }, { "epoch": 1.1688566781131726e-13, "grad_norm": 0.6309725046157837, "learning_rate": 8.379772581370956e-05, "loss": 2.3882, "step": 33690 }, { "epoch": 1.169203622808368e-13, "grad_norm": 0.6975188851356506, "learning_rate": 8.377617095519979e-05, "loss": 2.3427, "step": 33700 }, { "epoch": 1.1695505675035633e-13, "grad_norm": 0.575898289680481, "learning_rate": 8.375461372043042e-05, "loss": 2.3421, "step": 33710 }, { "epoch": 1.1698975121987587e-13, "grad_norm": 0.5766704678535461, "learning_rate": 8.373305411281053e-05, "loss": 2.3865, "step": 33720 }, { "epoch": 1.170244456893954e-13, "grad_norm": 0.5645007491111755, "learning_rate": 8.371149213574962e-05, "loss": 2.3685, "step": 33730 }, { "epoch": 1.1705914015891494e-13, "grad_norm": 0.576490581035614, "learning_rate": 8.368992779265753e-05, "loss": 2.3716, "step": 33740 }, { "epoch": 1.1709383462843448e-13, "grad_norm": 0.5603734850883484, "learning_rate": 8.366836108694444e-05, "loss": 2.3879, "step": 33750 }, { "epoch": 1.1712852909795402e-13, "grad_norm": 0.6010884642601013, "learning_rate": 8.364679202202098e-05, "loss": 2.3267, "step": 33760 }, { "epoch": 1.1716322356747355e-13, "grad_norm": 0.5808992385864258, "learning_rate": 8.362522060129808e-05, "loss": 2.3655, "step": 33770 }, { "epoch": 1.1719791803699309e-13, "grad_norm": 0.6051324009895325, "learning_rate": 8.360364682818711e-05, "loss": 2.3689, "step": 33780 }, { "epoch": 1.1723261250651262e-13, "grad_norm": 0.5823137760162354, "learning_rate": 8.358207070609975e-05, "loss": 2.3466, "step": 33790 }, { "epoch": 1.1726730697603216e-13, "grad_norm": 0.5732608437538147, "learning_rate": 8.356049223844807e-05, "loss": 2.3658, "step": 33800 }, { "epoch": 1.173020014455517e-13, "grad_norm": 0.5626540780067444, "learning_rate": 8.353891142864455e-05, "loss": 2.3629, "step": 33810 }, { "epoch": 1.1733669591507123e-13, "grad_norm": 0.5779193043708801, "learning_rate": 8.351732828010196e-05, "loss": 2.3643, "step": 33820 }, { "epoch": 1.1737139038459077e-13, "grad_norm": 0.5539374947547913, "learning_rate": 8.349574279623355e-05, "loss": 2.3877, "step": 33830 }, { "epoch": 1.174060848541103e-13, "grad_norm": 0.5644084811210632, "learning_rate": 8.347415498045284e-05, "loss": 2.3856, "step": 33840 }, { "epoch": 1.1744077932362984e-13, "grad_norm": 0.5845109820365906, "learning_rate": 8.345256483617378e-05, "loss": 2.376, "step": 33850 }, { "epoch": 1.1747547379314938e-13, "grad_norm": 0.7960535883903503, "learning_rate": 8.343097236681064e-05, "loss": 2.3695, "step": 33860 }, { "epoch": 1.175101682626689e-13, "grad_norm": 0.5489283204078674, "learning_rate": 8.340937757577808e-05, "loss": 2.3813, "step": 33870 }, { "epoch": 1.1754486273218845e-13, "grad_norm": 0.5604497790336609, "learning_rate": 8.338778046649116e-05, "loss": 2.3783, "step": 33880 }, { "epoch": 1.1757955720170798e-13, "grad_norm": 0.5994604825973511, "learning_rate": 8.336618104236525e-05, "loss": 2.3703, "step": 33890 }, { "epoch": 1.1761425167122752e-13, "grad_norm": 0.5676986575126648, "learning_rate": 8.334457930681616e-05, "loss": 2.3587, "step": 33900 }, { "epoch": 1.1764894614074706e-13, "grad_norm": 0.552615761756897, "learning_rate": 8.332297526325993e-05, "loss": 2.3413, "step": 33910 }, { "epoch": 1.176836406102666e-13, "grad_norm": 0.6061236262321472, "learning_rate": 8.330136891511317e-05, "loss": 2.3517, "step": 33920 }, { "epoch": 1.1771833507978613e-13, "grad_norm": 0.6267610192298889, "learning_rate": 8.327976026579262e-05, "loss": 2.4058, "step": 33930 }, { "epoch": 1.1775302954930567e-13, "grad_norm": 0.5613904595375061, "learning_rate": 8.325814931871559e-05, "loss": 2.3757, "step": 33940 }, { "epoch": 1.177877240188252e-13, "grad_norm": 0.5866862535476685, "learning_rate": 8.323653607729962e-05, "loss": 2.355, "step": 33950 }, { "epoch": 1.1782241848834474e-13, "grad_norm": 0.6127223968505859, "learning_rate": 8.321492054496268e-05, "loss": 2.3781, "step": 33960 }, { "epoch": 1.1785711295786427e-13, "grad_norm": 0.5824593901634216, "learning_rate": 8.319330272512306e-05, "loss": 2.3526, "step": 33970 }, { "epoch": 1.178918074273838e-13, "grad_norm": 0.5537236332893372, "learning_rate": 8.317168262119943e-05, "loss": 2.3693, "step": 33980 }, { "epoch": 1.1792650189690335e-13, "grad_norm": 0.5712925791740417, "learning_rate": 8.315006023661085e-05, "loss": 2.3392, "step": 33990 }, { "epoch": 1.1796119636642288e-13, "grad_norm": 0.6083471179008484, "learning_rate": 8.31284355747767e-05, "loss": 2.3673, "step": 34000 }, { "epoch": 1.1799589083594242e-13, "grad_norm": 0.5400474667549133, "learning_rate": 8.310680863911674e-05, "loss": 2.3902, "step": 34010 }, { "epoch": 1.1803058530546195e-13, "grad_norm": 0.5917190909385681, "learning_rate": 8.308517943305108e-05, "loss": 2.3732, "step": 34020 }, { "epoch": 1.180652797749815e-13, "grad_norm": 0.5750717520713806, "learning_rate": 8.306354796000018e-05, "loss": 2.3621, "step": 34030 }, { "epoch": 1.1809997424450103e-13, "grad_norm": 0.5749757885932922, "learning_rate": 8.304191422338488e-05, "loss": 2.3735, "step": 34040 }, { "epoch": 1.1813466871402056e-13, "grad_norm": 0.6153334379196167, "learning_rate": 8.302027822662638e-05, "loss": 2.3696, "step": 34050 }, { "epoch": 1.181693631835401e-13, "grad_norm": 0.5806971192359924, "learning_rate": 8.299863997314622e-05, "loss": 2.368, "step": 34060 }, { "epoch": 1.1820405765305964e-13, "grad_norm": 0.5910497307777405, "learning_rate": 8.297699946636629e-05, "loss": 2.3745, "step": 34070 }, { "epoch": 1.1823875212257917e-13, "grad_norm": 0.5928359031677246, "learning_rate": 8.295535670970888e-05, "loss": 2.3765, "step": 34080 }, { "epoch": 1.182734465920987e-13, "grad_norm": 0.5602556467056274, "learning_rate": 8.293371170659656e-05, "loss": 2.3649, "step": 34090 }, { "epoch": 1.1830814106161824e-13, "grad_norm": 0.5528492331504822, "learning_rate": 8.291206446045235e-05, "loss": 2.3766, "step": 34100 }, { "epoch": 1.1834283553113778e-13, "grad_norm": 0.5817140936851501, "learning_rate": 8.289041497469956e-05, "loss": 2.3488, "step": 34110 }, { "epoch": 1.1837753000065732e-13, "grad_norm": 0.5800295472145081, "learning_rate": 8.286876325276184e-05, "loss": 2.3498, "step": 34120 }, { "epoch": 1.1841222447017685e-13, "grad_norm": 0.5797029733657837, "learning_rate": 8.284710929806326e-05, "loss": 2.3413, "step": 34130 }, { "epoch": 1.184469189396964e-13, "grad_norm": 0.575778067111969, "learning_rate": 8.282545311402817e-05, "loss": 2.3797, "step": 34140 }, { "epoch": 1.1848161340921592e-13, "grad_norm": 0.55781489610672, "learning_rate": 8.280379470408135e-05, "loss": 2.3596, "step": 34150 }, { "epoch": 1.1851630787873546e-13, "grad_norm": 0.5694294571876526, "learning_rate": 8.278213407164789e-05, "loss": 2.3569, "step": 34160 }, { "epoch": 1.18551002348255e-13, "grad_norm": 0.541737973690033, "learning_rate": 8.27604712201532e-05, "loss": 2.3316, "step": 34170 }, { "epoch": 1.1858569681777453e-13, "grad_norm": 0.566902756690979, "learning_rate": 8.27388061530231e-05, "loss": 2.3852, "step": 34180 }, { "epoch": 1.1862039128729407e-13, "grad_norm": 0.6062514781951904, "learning_rate": 8.271713887368371e-05, "loss": 2.3641, "step": 34190 }, { "epoch": 1.186550857568136e-13, "grad_norm": 0.5714603066444397, "learning_rate": 8.269546938556154e-05, "loss": 2.3499, "step": 34200 }, { "epoch": 1.1868978022633314e-13, "grad_norm": 0.5853250622749329, "learning_rate": 8.267379769208341e-05, "loss": 2.3922, "step": 34210 }, { "epoch": 1.1872447469585268e-13, "grad_norm": 0.6015211939811707, "learning_rate": 8.265212379667656e-05, "loss": 2.3426, "step": 34220 }, { "epoch": 1.1875916916537221e-13, "grad_norm": 0.5822768211364746, "learning_rate": 8.26304477027685e-05, "loss": 2.3517, "step": 34230 }, { "epoch": 1.1879386363489175e-13, "grad_norm": 0.5567986965179443, "learning_rate": 8.260876941378711e-05, "loss": 2.4099, "step": 34240 }, { "epoch": 1.1882855810441129e-13, "grad_norm": 0.5540833473205566, "learning_rate": 8.258708893316064e-05, "loss": 2.3755, "step": 34250 }, { "epoch": 1.1886325257393082e-13, "grad_norm": 0.6087158918380737, "learning_rate": 8.256540626431766e-05, "loss": 2.3911, "step": 34260 }, { "epoch": 1.1889794704345036e-13, "grad_norm": 0.5747584700584412, "learning_rate": 8.254372141068713e-05, "loss": 2.3167, "step": 34270 }, { "epoch": 1.189326415129699e-13, "grad_norm": 0.585925281047821, "learning_rate": 8.252203437569826e-05, "loss": 2.3495, "step": 34280 }, { "epoch": 1.1896733598248943e-13, "grad_norm": 0.5716273188591003, "learning_rate": 8.250034516278071e-05, "loss": 2.3446, "step": 34290 }, { "epoch": 1.1900203045200897e-13, "grad_norm": 0.5717090964317322, "learning_rate": 8.247865377536444e-05, "loss": 2.3774, "step": 34300 }, { "epoch": 1.190367249215285e-13, "grad_norm": 0.5768341422080994, "learning_rate": 8.245696021687976e-05, "loss": 2.3439, "step": 34310 }, { "epoch": 1.1907141939104804e-13, "grad_norm": 0.5790709257125854, "learning_rate": 8.24352644907573e-05, "loss": 2.369, "step": 34320 }, { "epoch": 1.1910611386056758e-13, "grad_norm": 0.611169159412384, "learning_rate": 8.241356660042809e-05, "loss": 2.365, "step": 34330 }, { "epoch": 1.191408083300871e-13, "grad_norm": 0.5986297130584717, "learning_rate": 8.239186654932342e-05, "loss": 2.3696, "step": 34340 }, { "epoch": 1.1917550279960665e-13, "grad_norm": 0.5805976986885071, "learning_rate": 8.237016434087497e-05, "loss": 2.373, "step": 34350 }, { "epoch": 1.1921019726912618e-13, "grad_norm": 0.5702353119850159, "learning_rate": 8.23484599785148e-05, "loss": 2.3619, "step": 34360 }, { "epoch": 1.1924489173864572e-13, "grad_norm": 0.5712612867355347, "learning_rate": 8.232675346567523e-05, "loss": 2.3616, "step": 34370 }, { "epoch": 1.1927958620816526e-13, "grad_norm": 0.5749824047088623, "learning_rate": 8.230504480578897e-05, "loss": 2.3525, "step": 34380 }, { "epoch": 1.193142806776848e-13, "grad_norm": 0.5778239965438843, "learning_rate": 8.228333400228905e-05, "loss": 2.3586, "step": 34390 }, { "epoch": 1.1934897514720433e-13, "grad_norm": 0.6126687526702881, "learning_rate": 8.226162105860888e-05, "loss": 2.3713, "step": 34400 }, { "epoch": 1.1938366961672386e-13, "grad_norm": 0.6402298808097839, "learning_rate": 8.223990597818212e-05, "loss": 2.3851, "step": 34410 }, { "epoch": 1.194183640862434e-13, "grad_norm": 0.5663459300994873, "learning_rate": 8.221818876444286e-05, "loss": 2.3476, "step": 34420 }, { "epoch": 1.1945305855576294e-13, "grad_norm": 0.5530163645744324, "learning_rate": 8.219646942082548e-05, "loss": 2.3895, "step": 34430 }, { "epoch": 1.1948775302528247e-13, "grad_norm": 0.5769765973091125, "learning_rate": 8.21747479507647e-05, "loss": 2.351, "step": 34440 }, { "epoch": 1.19522447494802e-13, "grad_norm": 0.5690736174583435, "learning_rate": 8.215302435769561e-05, "loss": 2.3898, "step": 34450 }, { "epoch": 1.1955714196432154e-13, "grad_norm": 0.5935394763946533, "learning_rate": 8.213129864505357e-05, "loss": 2.3775, "step": 34460 }, { "epoch": 1.1959183643384108e-13, "grad_norm": 0.5782422423362732, "learning_rate": 8.210957081627436e-05, "loss": 2.3762, "step": 34470 }, { "epoch": 1.1962653090336062e-13, "grad_norm": 0.5842005014419556, "learning_rate": 8.208784087479402e-05, "loss": 2.3355, "step": 34480 }, { "epoch": 1.1966122537288015e-13, "grad_norm": 0.588543176651001, "learning_rate": 8.206610882404896e-05, "loss": 2.3788, "step": 34490 }, { "epoch": 1.196959198423997e-13, "grad_norm": 0.5613757371902466, "learning_rate": 8.204437466747591e-05, "loss": 2.3686, "step": 34500 }, { "epoch": 1.1973061431191923e-13, "grad_norm": 0.5687458515167236, "learning_rate": 8.202263840851194e-05, "loss": 2.3483, "step": 34510 }, { "epoch": 1.1976530878143876e-13, "grad_norm": 0.587069571018219, "learning_rate": 8.200090005059445e-05, "loss": 2.3595, "step": 34520 }, { "epoch": 1.198000032509583e-13, "grad_norm": 0.5799306035041809, "learning_rate": 8.197915959716119e-05, "loss": 2.3682, "step": 34530 }, { "epoch": 1.1983469772047783e-13, "grad_norm": 0.5965436697006226, "learning_rate": 8.195741705165023e-05, "loss": 2.3788, "step": 34540 }, { "epoch": 1.1986939218999737e-13, "grad_norm": 0.6027328372001648, "learning_rate": 8.193567241749996e-05, "loss": 2.3565, "step": 34550 }, { "epoch": 1.199040866595169e-13, "grad_norm": 0.5935864448547363, "learning_rate": 8.191392569814909e-05, "loss": 2.3824, "step": 34560 }, { "epoch": 1.1993878112903644e-13, "grad_norm": 0.5817765593528748, "learning_rate": 8.189217689703667e-05, "loss": 2.3336, "step": 34570 }, { "epoch": 1.1997347559855598e-13, "grad_norm": 0.6071882843971252, "learning_rate": 8.187042601760211e-05, "loss": 2.3583, "step": 34580 }, { "epoch": 1.2000817006807551e-13, "grad_norm": 1.3745992183685303, "learning_rate": 8.184867306328514e-05, "loss": 2.3651, "step": 34590 }, { "epoch": 1.2004286453759505e-13, "grad_norm": 0.5856475234031677, "learning_rate": 8.182691803752573e-05, "loss": 2.3402, "step": 34600 }, { "epoch": 1.2007755900711459e-13, "grad_norm": 0.6026624441146851, "learning_rate": 8.180516094376433e-05, "loss": 2.3854, "step": 34610 }, { "epoch": 1.2011225347663412e-13, "grad_norm": 0.6134294271469116, "learning_rate": 8.178340178544161e-05, "loss": 2.3379, "step": 34620 }, { "epoch": 1.2014694794615366e-13, "grad_norm": 0.556504487991333, "learning_rate": 8.176164056599859e-05, "loss": 2.3595, "step": 34630 }, { "epoch": 1.201816424156732e-13, "grad_norm": 0.5768483877182007, "learning_rate": 8.173987728887662e-05, "loss": 2.3835, "step": 34640 }, { "epoch": 1.2021633688519273e-13, "grad_norm": 0.5828016400337219, "learning_rate": 8.171811195751737e-05, "loss": 2.3389, "step": 34650 }, { "epoch": 1.2025103135471227e-13, "grad_norm": 0.6071190237998962, "learning_rate": 8.169634457536284e-05, "loss": 2.3654, "step": 34660 }, { "epoch": 1.202857258242318e-13, "grad_norm": 0.5876551270484924, "learning_rate": 8.167457514585536e-05, "loss": 2.3549, "step": 34670 }, { "epoch": 1.2032042029375134e-13, "grad_norm": 0.5992572903633118, "learning_rate": 8.165280367243756e-05, "loss": 2.3552, "step": 34680 }, { "epoch": 1.2035511476327088e-13, "grad_norm": 0.5618051886558533, "learning_rate": 8.163103015855246e-05, "loss": 2.3551, "step": 34690 }, { "epoch": 1.203898092327904e-13, "grad_norm": 0.594649612903595, "learning_rate": 8.16092546076433e-05, "loss": 2.3485, "step": 34700 }, { "epoch": 1.2042450370230995e-13, "grad_norm": 0.5763967037200928, "learning_rate": 8.158747702315374e-05, "loss": 2.3297, "step": 34710 }, { "epoch": 1.2045919817182948e-13, "grad_norm": 0.5908707976341248, "learning_rate": 8.156569740852768e-05, "loss": 2.3774, "step": 34720 }, { "epoch": 1.2049389264134902e-13, "grad_norm": 0.5635332465171814, "learning_rate": 8.154391576720938e-05, "loss": 2.3724, "step": 34730 }, { "epoch": 1.2052858711086856e-13, "grad_norm": 0.5645604133605957, "learning_rate": 8.152213210264345e-05, "loss": 2.3878, "step": 34740 }, { "epoch": 1.205632815803881e-13, "grad_norm": 0.565773606300354, "learning_rate": 8.150034641827475e-05, "loss": 2.3561, "step": 34750 }, { "epoch": 1.2059797604990763e-13, "grad_norm": 0.5561349391937256, "learning_rate": 8.147855871754854e-05, "loss": 2.3522, "step": 34760 }, { "epoch": 1.2063267051942717e-13, "grad_norm": 0.5639633536338806, "learning_rate": 8.145676900391033e-05, "loss": 2.3539, "step": 34770 }, { "epoch": 1.206673649889467e-13, "grad_norm": 0.59849613904953, "learning_rate": 8.143497728080598e-05, "loss": 2.3342, "step": 34780 }, { "epoch": 1.2070205945846624e-13, "grad_norm": 0.604261040687561, "learning_rate": 8.141318355168167e-05, "loss": 2.3816, "step": 34790 }, { "epoch": 1.2073675392798577e-13, "grad_norm": 0.59687739610672, "learning_rate": 8.13913878199839e-05, "loss": 2.3294, "step": 34800 }, { "epoch": 1.207714483975053e-13, "grad_norm": 0.567751944065094, "learning_rate": 8.136959008915945e-05, "loss": 2.3391, "step": 34810 }, { "epoch": 1.2080614286702485e-13, "grad_norm": 0.5895633697509766, "learning_rate": 8.134779036265547e-05, "loss": 2.3938, "step": 34820 }, { "epoch": 1.2084083733654438e-13, "grad_norm": 0.6308859586715698, "learning_rate": 8.132598864391935e-05, "loss": 2.3587, "step": 34830 }, { "epoch": 1.2087553180606392e-13, "grad_norm": 0.5663706660270691, "learning_rate": 8.130418493639891e-05, "loss": 2.3909, "step": 34840 }, { "epoch": 1.2091022627558345e-13, "grad_norm": 0.5781893134117126, "learning_rate": 8.12823792435422e-05, "loss": 2.3242, "step": 34850 }, { "epoch": 1.20944920745103e-13, "grad_norm": 0.5742360949516296, "learning_rate": 8.126057156879757e-05, "loss": 2.3455, "step": 34860 }, { "epoch": 1.2097961521462253e-13, "grad_norm": 0.5816594958305359, "learning_rate": 8.123876191561374e-05, "loss": 2.3499, "step": 34870 }, { "epoch": 1.2101430968414206e-13, "grad_norm": 0.7265118956565857, "learning_rate": 8.121695028743974e-05, "loss": 2.3515, "step": 34880 }, { "epoch": 1.210490041536616e-13, "grad_norm": 0.548849880695343, "learning_rate": 8.119513668772484e-05, "loss": 2.3636, "step": 34890 }, { "epoch": 1.2108369862318114e-13, "grad_norm": 0.5906134843826294, "learning_rate": 8.11733211199187e-05, "loss": 2.3332, "step": 34900 }, { "epoch": 1.2111839309270067e-13, "grad_norm": 0.5523233413696289, "learning_rate": 8.115150358747128e-05, "loss": 2.379, "step": 34910 }, { "epoch": 1.211530875622202e-13, "grad_norm": 0.5992013812065125, "learning_rate": 8.112968409383283e-05, "loss": 2.3723, "step": 34920 }, { "epoch": 1.2118778203173974e-13, "grad_norm": 0.6342531442642212, "learning_rate": 8.110786264245388e-05, "loss": 2.3516, "step": 34930 }, { "epoch": 1.2122247650125928e-13, "grad_norm": 0.5616797804832458, "learning_rate": 8.108603923678535e-05, "loss": 2.3849, "step": 34940 }, { "epoch": 1.2125717097077882e-13, "grad_norm": 0.6024415493011475, "learning_rate": 8.10642138802784e-05, "loss": 2.3355, "step": 34950 }, { "epoch": 1.2129186544029835e-13, "grad_norm": 0.597852349281311, "learning_rate": 8.104238657638452e-05, "loss": 2.3494, "step": 34960 }, { "epoch": 1.213265599098179e-13, "grad_norm": 0.5954911708831787, "learning_rate": 8.102055732855552e-05, "loss": 2.3486, "step": 34970 }, { "epoch": 1.2136125437933742e-13, "grad_norm": 0.5743204951286316, "learning_rate": 8.099872614024351e-05, "loss": 2.3599, "step": 34980 }, { "epoch": 1.2139594884885696e-13, "grad_norm": 0.594490110874176, "learning_rate": 8.097689301490088e-05, "loss": 2.3822, "step": 34990 }, { "epoch": 1.214306433183765e-13, "grad_norm": 0.5598217248916626, "learning_rate": 8.095505795598039e-05, "loss": 2.3862, "step": 35000 }, { "epoch": 1.2146533778789603e-13, "grad_norm": 0.559272289276123, "learning_rate": 8.093322096693504e-05, "loss": 2.3245, "step": 35010 }, { "epoch": 1.2150003225741557e-13, "grad_norm": 0.5880220532417297, "learning_rate": 8.091138205121817e-05, "loss": 2.3917, "step": 35020 }, { "epoch": 1.215347267269351e-13, "grad_norm": 0.577689528465271, "learning_rate": 8.088954121228341e-05, "loss": 2.3176, "step": 35030 }, { "epoch": 1.2156942119645464e-13, "grad_norm": 0.573743999004364, "learning_rate": 8.086769845358472e-05, "loss": 2.3777, "step": 35040 }, { "epoch": 1.2160411566597418e-13, "grad_norm": 0.5684082508087158, "learning_rate": 8.08458537785763e-05, "loss": 2.3265, "step": 35050 }, { "epoch": 1.2163881013549371e-13, "grad_norm": 0.5936105251312256, "learning_rate": 8.082400719071276e-05, "loss": 2.3724, "step": 35060 }, { "epoch": 1.2167350460501325e-13, "grad_norm": 0.5767680406570435, "learning_rate": 8.080215869344889e-05, "loss": 2.3575, "step": 35070 }, { "epoch": 1.2170819907453279e-13, "grad_norm": 0.5876318216323853, "learning_rate": 8.078030829023988e-05, "loss": 2.3289, "step": 35080 }, { "epoch": 1.2174289354405232e-13, "grad_norm": 0.5668376684188843, "learning_rate": 8.075845598454117e-05, "loss": 2.3706, "step": 35090 }, { "epoch": 1.2177758801357186e-13, "grad_norm": 0.5826125144958496, "learning_rate": 8.07366017798085e-05, "loss": 2.3065, "step": 35100 }, { "epoch": 1.218122824830914e-13, "grad_norm": 0.5817365646362305, "learning_rate": 8.071474567949797e-05, "loss": 2.3296, "step": 35110 }, { "epoch": 1.2184697695261093e-13, "grad_norm": 0.5778868198394775, "learning_rate": 8.069288768706588e-05, "loss": 2.358, "step": 35120 }, { "epoch": 1.2188167142213047e-13, "grad_norm": 0.5673644542694092, "learning_rate": 8.067102780596893e-05, "loss": 2.3234, "step": 35130 }, { "epoch": 1.2191636589165e-13, "grad_norm": 0.5773962140083313, "learning_rate": 8.064916603966403e-05, "loss": 2.3692, "step": 35140 }, { "epoch": 1.2195106036116954e-13, "grad_norm": 0.5784286856651306, "learning_rate": 8.062730239160846e-05, "loss": 2.35, "step": 35150 }, { "epoch": 1.2198575483068907e-13, "grad_norm": 0.606734573841095, "learning_rate": 8.060543686525974e-05, "loss": 2.3489, "step": 35160 }, { "epoch": 1.220204493002086e-13, "grad_norm": 0.800177812576294, "learning_rate": 8.058356946407576e-05, "loss": 2.3558, "step": 35170 }, { "epoch": 1.2205514376972815e-13, "grad_norm": 0.5841243267059326, "learning_rate": 8.056170019151461e-05, "loss": 2.2933, "step": 35180 }, { "epoch": 1.2208983823924768e-13, "grad_norm": 0.5720636248588562, "learning_rate": 8.053982905103477e-05, "loss": 2.3546, "step": 35190 }, { "epoch": 1.2212453270876722e-13, "grad_norm": 0.5583967566490173, "learning_rate": 8.051795604609494e-05, "loss": 2.3956, "step": 35200 }, { "epoch": 1.2215922717828676e-13, "grad_norm": 0.5851749181747437, "learning_rate": 8.049608118015415e-05, "loss": 2.346, "step": 35210 }, { "epoch": 1.221939216478063e-13, "grad_norm": 0.5715795159339905, "learning_rate": 8.047420445667176e-05, "loss": 2.341, "step": 35220 }, { "epoch": 1.2222861611732583e-13, "grad_norm": 0.5693667531013489, "learning_rate": 8.045232587910732e-05, "loss": 2.3703, "step": 35230 }, { "epoch": 1.2226331058684536e-13, "grad_norm": 0.5648534893989563, "learning_rate": 8.04304454509208e-05, "loss": 2.3548, "step": 35240 }, { "epoch": 1.222980050563649e-13, "grad_norm": 0.5821748375892639, "learning_rate": 8.04085631755724e-05, "loss": 2.3482, "step": 35250 }, { "epoch": 1.2233269952588444e-13, "grad_norm": 0.5929968953132629, "learning_rate": 8.038667905652256e-05, "loss": 2.3949, "step": 35260 }, { "epoch": 1.2236739399540397e-13, "grad_norm": 0.5686498880386353, "learning_rate": 8.03647930972321e-05, "loss": 2.3248, "step": 35270 }, { "epoch": 1.224020884649235e-13, "grad_norm": 0.6429967284202576, "learning_rate": 8.03429053011621e-05, "loss": 2.3626, "step": 35280 }, { "epoch": 1.2243678293444304e-13, "grad_norm": 0.5796322226524353, "learning_rate": 8.032101567177393e-05, "loss": 2.3196, "step": 35290 }, { "epoch": 1.2247147740396258e-13, "grad_norm": 0.5674224495887756, "learning_rate": 8.029912421252919e-05, "loss": 2.3521, "step": 35300 }, { "epoch": 1.2250617187348212e-13, "grad_norm": 0.592134952545166, "learning_rate": 8.027723092688992e-05, "loss": 2.3526, "step": 35310 }, { "epoch": 1.2254086634300165e-13, "grad_norm": 0.5829426646232605, "learning_rate": 8.025533581831827e-05, "loss": 2.3726, "step": 35320 }, { "epoch": 1.225755608125212e-13, "grad_norm": 0.5911497473716736, "learning_rate": 8.023343889027683e-05, "loss": 2.3598, "step": 35330 }, { "epoch": 1.2261025528204073e-13, "grad_norm": 0.596555769443512, "learning_rate": 8.021154014622837e-05, "loss": 2.33, "step": 35340 }, { "epoch": 1.2264494975156026e-13, "grad_norm": 0.5660712718963623, "learning_rate": 8.0189639589636e-05, "loss": 2.3392, "step": 35350 }, { "epoch": 1.226796442210798e-13, "grad_norm": 0.5983642935752869, "learning_rate": 8.01677372239631e-05, "loss": 2.3478, "step": 35360 }, { "epoch": 1.2271433869059933e-13, "grad_norm": 0.5766017436981201, "learning_rate": 8.014583305267332e-05, "loss": 2.3429, "step": 35370 }, { "epoch": 1.2274903316011887e-13, "grad_norm": 0.574011504650116, "learning_rate": 8.012392707923068e-05, "loss": 2.3619, "step": 35380 }, { "epoch": 1.227837276296384e-13, "grad_norm": 0.59775310754776, "learning_rate": 8.010201930709935e-05, "loss": 2.3577, "step": 35390 }, { "epoch": 1.2281842209915794e-13, "grad_norm": 0.5729162096977234, "learning_rate": 8.00801097397439e-05, "loss": 2.3773, "step": 35400 }, { "epoch": 1.2285311656867748e-13, "grad_norm": 0.5867936611175537, "learning_rate": 8.005819838062912e-05, "loss": 2.3287, "step": 35410 }, { "epoch": 1.2288781103819701e-13, "grad_norm": 0.5693145394325256, "learning_rate": 8.003628523322009e-05, "loss": 2.3312, "step": 35420 }, { "epoch": 1.2292250550771655e-13, "grad_norm": 0.5809285640716553, "learning_rate": 8.001437030098223e-05, "loss": 2.3462, "step": 35430 }, { "epoch": 1.2295719997723609e-13, "grad_norm": 0.5765895247459412, "learning_rate": 7.999245358738116e-05, "loss": 2.341, "step": 35440 }, { "epoch": 1.2299189444675562e-13, "grad_norm": 0.5980464220046997, "learning_rate": 7.997053509588284e-05, "loss": 2.3709, "step": 35450 }, { "epoch": 1.2302658891627516e-13, "grad_norm": 0.5670158267021179, "learning_rate": 7.994861482995347e-05, "loss": 2.3324, "step": 35460 }, { "epoch": 1.230612833857947e-13, "grad_norm": 0.5877791047096252, "learning_rate": 7.992669279305957e-05, "loss": 2.3539, "step": 35470 }, { "epoch": 1.2309597785531423e-13, "grad_norm": 0.56960129737854, "learning_rate": 7.99047689886679e-05, "loss": 2.3024, "step": 35480 }, { "epoch": 1.2313067232483377e-13, "grad_norm": 0.5770565867424011, "learning_rate": 7.988284342024554e-05, "loss": 2.3593, "step": 35490 }, { "epoch": 1.231653667943533e-13, "grad_norm": 0.6013855338096619, "learning_rate": 7.986091609125984e-05, "loss": 2.4021, "step": 35500 }, { "epoch": 1.2320006126387284e-13, "grad_norm": 0.592596709728241, "learning_rate": 7.98389870051784e-05, "loss": 2.3359, "step": 35510 }, { "epoch": 1.2323475573339238e-13, "grad_norm": 0.588988721370697, "learning_rate": 7.981705616546912e-05, "loss": 2.3515, "step": 35520 }, { "epoch": 1.232694502029119e-13, "grad_norm": 0.6024110317230225, "learning_rate": 7.979512357560016e-05, "loss": 2.3771, "step": 35530 }, { "epoch": 1.2330414467243145e-13, "grad_norm": 0.6065179705619812, "learning_rate": 7.977318923903999e-05, "loss": 2.3382, "step": 35540 }, { "epoch": 1.2333883914195098e-13, "grad_norm": 0.5780998468399048, "learning_rate": 7.975125315925734e-05, "loss": 2.3819, "step": 35550 }, { "epoch": 1.2337353361147052e-13, "grad_norm": 0.5689642429351807, "learning_rate": 7.972931533972118e-05, "loss": 2.337, "step": 35560 }, { "epoch": 1.2340822808099006e-13, "grad_norm": 0.5726886987686157, "learning_rate": 7.970737578390084e-05, "loss": 2.3494, "step": 35570 }, { "epoch": 1.234429225505096e-13, "grad_norm": 0.5687869191169739, "learning_rate": 7.968543449526581e-05, "loss": 2.3698, "step": 35580 }, { "epoch": 1.2347761702002913e-13, "grad_norm": 0.5629302263259888, "learning_rate": 7.966349147728597e-05, "loss": 2.3534, "step": 35590 }, { "epoch": 1.2351231148954867e-13, "grad_norm": 0.5848135948181152, "learning_rate": 7.964154673343137e-05, "loss": 2.3854, "step": 35600 }, { "epoch": 1.235470059590682e-13, "grad_norm": 0.5817884206771851, "learning_rate": 7.961960026717245e-05, "loss": 2.3618, "step": 35610 }, { "epoch": 1.2358170042858774e-13, "grad_norm": 0.6143367886543274, "learning_rate": 7.959765208197979e-05, "loss": 2.3498, "step": 35620 }, { "epoch": 1.2361639489810727e-13, "grad_norm": 0.5631608963012695, "learning_rate": 7.957570218132433e-05, "loss": 2.3579, "step": 35630 }, { "epoch": 1.236510893676268e-13, "grad_norm": 0.5747044086456299, "learning_rate": 7.955375056867727e-05, "loss": 2.3356, "step": 35640 }, { "epoch": 1.2368578383714635e-13, "grad_norm": 0.5855399966239929, "learning_rate": 7.953179724751004e-05, "loss": 2.3771, "step": 35650 }, { "epoch": 1.2372047830666588e-13, "grad_norm": 0.5650411248207092, "learning_rate": 7.95098422212944e-05, "loss": 2.35, "step": 35660 }, { "epoch": 1.2375517277618542e-13, "grad_norm": 0.5962468981742859, "learning_rate": 7.948788549350233e-05, "loss": 2.3598, "step": 35670 }, { "epoch": 1.2378986724570495e-13, "grad_norm": 0.5714625716209412, "learning_rate": 7.946592706760611e-05, "loss": 2.3405, "step": 35680 }, { "epoch": 1.238245617152245e-13, "grad_norm": 0.5778440833091736, "learning_rate": 7.944396694707824e-05, "loss": 2.4114, "step": 35690 }, { "epoch": 1.2385925618474403e-13, "grad_norm": 0.6102004647254944, "learning_rate": 7.942200513539158e-05, "loss": 2.3529, "step": 35700 }, { "epoch": 1.2389395065426356e-13, "grad_norm": 0.5653351545333862, "learning_rate": 7.940004163601914e-05, "loss": 2.3755, "step": 35710 }, { "epoch": 1.239286451237831e-13, "grad_norm": 0.5995391607284546, "learning_rate": 7.937807645243433e-05, "loss": 2.3349, "step": 35720 }, { "epoch": 1.2396333959330263e-13, "grad_norm": 0.5708414316177368, "learning_rate": 7.93561095881107e-05, "loss": 2.3725, "step": 35730 }, { "epoch": 1.2399803406282217e-13, "grad_norm": 0.5843466520309448, "learning_rate": 7.933414104652213e-05, "loss": 2.3815, "step": 35740 }, { "epoch": 1.240327285323417e-13, "grad_norm": 0.5754653215408325, "learning_rate": 7.931217083114277e-05, "loss": 2.3681, "step": 35750 }, { "epoch": 1.2406742300186124e-13, "grad_norm": 0.5622299313545227, "learning_rate": 7.929019894544703e-05, "loss": 2.3592, "step": 35760 }, { "epoch": 1.2410211747138078e-13, "grad_norm": 0.5932627320289612, "learning_rate": 7.926822539290952e-05, "loss": 2.3569, "step": 35770 }, { "epoch": 1.2413681194090032e-13, "grad_norm": 0.5721073746681213, "learning_rate": 7.924625017700523e-05, "loss": 2.3515, "step": 35780 }, { "epoch": 1.2417150641041985e-13, "grad_norm": 0.5737369656562805, "learning_rate": 7.922427330120931e-05, "loss": 2.3386, "step": 35790 }, { "epoch": 1.242062008799394e-13, "grad_norm": 0.6121986508369446, "learning_rate": 7.920229476899723e-05, "loss": 2.3104, "step": 35800 }, { "epoch": 1.2424089534945892e-13, "grad_norm": 0.6325771808624268, "learning_rate": 7.918031458384474e-05, "loss": 2.3663, "step": 35810 }, { "epoch": 1.2427558981897846e-13, "grad_norm": 0.5984382033348083, "learning_rate": 7.915833274922777e-05, "loss": 2.3557, "step": 35820 }, { "epoch": 1.24310284288498e-13, "grad_norm": 0.5731608271598816, "learning_rate": 7.913634926862255e-05, "loss": 2.3351, "step": 35830 }, { "epoch": 1.2434497875801753e-13, "grad_norm": 0.5778874754905701, "learning_rate": 7.911436414550563e-05, "loss": 2.3454, "step": 35840 }, { "epoch": 1.2437967322753707e-13, "grad_norm": 0.6411936283111572, "learning_rate": 7.909237738335371e-05, "loss": 2.3847, "step": 35850 }, { "epoch": 1.244143676970566e-13, "grad_norm": 0.6047610640525818, "learning_rate": 7.907038898564385e-05, "loss": 2.3469, "step": 35860 }, { "epoch": 1.2444906216657614e-13, "grad_norm": 0.5471968054771423, "learning_rate": 7.904839895585329e-05, "loss": 2.373, "step": 35870 }, { "epoch": 1.2448375663609568e-13, "grad_norm": 0.6058011651039124, "learning_rate": 7.90264072974596e-05, "loss": 2.3403, "step": 35880 }, { "epoch": 1.2451845110561521e-13, "grad_norm": 0.5931527018547058, "learning_rate": 7.900441401394057e-05, "loss": 2.3567, "step": 35890 }, { "epoch": 1.2455314557513475e-13, "grad_norm": 0.5798162221908569, "learning_rate": 7.898241910877421e-05, "loss": 2.3474, "step": 35900 }, { "epoch": 1.2458784004465429e-13, "grad_norm": 0.5788148641586304, "learning_rate": 7.896042258543885e-05, "loss": 2.3676, "step": 35910 }, { "epoch": 1.2462253451417382e-13, "grad_norm": 0.5901244878768921, "learning_rate": 7.893842444741306e-05, "loss": 2.3657, "step": 35920 }, { "epoch": 1.2465722898369336e-13, "grad_norm": 0.5889849066734314, "learning_rate": 7.891642469817566e-05, "loss": 2.3413, "step": 35930 }, { "epoch": 1.246919234532129e-13, "grad_norm": 0.5655329823493958, "learning_rate": 7.889442334120569e-05, "loss": 2.3545, "step": 35940 }, { "epoch": 1.2472661792273243e-13, "grad_norm": 0.5835154056549072, "learning_rate": 7.887242037998248e-05, "loss": 2.3505, "step": 35950 }, { "epoch": 1.2476131239225197e-13, "grad_norm": 0.5707507729530334, "learning_rate": 7.885041581798564e-05, "loss": 2.3569, "step": 35960 }, { "epoch": 1.247960068617715e-13, "grad_norm": 0.5652814507484436, "learning_rate": 7.882840965869497e-05, "loss": 2.347, "step": 35970 }, { "epoch": 1.2483070133129104e-13, "grad_norm": 0.571217954158783, "learning_rate": 7.880640190559059e-05, "loss": 2.3728, "step": 35980 }, { "epoch": 1.2486539580081057e-13, "grad_norm": 0.5616277456283569, "learning_rate": 7.878439256215281e-05, "loss": 2.3496, "step": 35990 }, { "epoch": 1.249000902703301e-13, "grad_norm": 0.5896549820899963, "learning_rate": 7.876238163186222e-05, "loss": 2.3201, "step": 36000 }, { "epoch": 1.2493478473984965e-13, "grad_norm": 0.5612303018569946, "learning_rate": 7.874036911819964e-05, "loss": 2.3728, "step": 36010 }, { "epoch": 1.2496947920936918e-13, "grad_norm": 0.5657585263252258, "learning_rate": 7.871835502464621e-05, "loss": 2.3231, "step": 36020 }, { "epoch": 1.2500417367888872e-13, "grad_norm": 0.5768492221832275, "learning_rate": 7.869633935468322e-05, "loss": 2.3798, "step": 36030 }, { "epoch": 1.2503886814840826e-13, "grad_norm": 0.5913141369819641, "learning_rate": 7.867432211179229e-05, "loss": 2.3618, "step": 36040 }, { "epoch": 1.250735626179278e-13, "grad_norm": 0.5787979364395142, "learning_rate": 7.865230329945524e-05, "loss": 2.335, "step": 36050 }, { "epoch": 1.2510825708744733e-13, "grad_norm": 0.6030681133270264, "learning_rate": 7.863028292115417e-05, "loss": 2.3738, "step": 36060 }, { "epoch": 1.2514295155696686e-13, "grad_norm": 0.5645657181739807, "learning_rate": 7.86082609803714e-05, "loss": 2.3626, "step": 36070 }, { "epoch": 1.251776460264864e-13, "grad_norm": 0.5854490399360657, "learning_rate": 7.858623748058954e-05, "loss": 2.3333, "step": 36080 }, { "epoch": 1.2521234049600594e-13, "grad_norm": 0.5724776387214661, "learning_rate": 7.856421242529138e-05, "loss": 2.332, "step": 36090 }, { "epoch": 1.2524703496552547e-13, "grad_norm": 0.5738252997398376, "learning_rate": 7.854218581795999e-05, "loss": 2.3614, "step": 36100 }, { "epoch": 1.25281729435045e-13, "grad_norm": 0.6003842949867249, "learning_rate": 7.85201576620787e-05, "loss": 2.3915, "step": 36110 }, { "epoch": 1.2531642390456454e-13, "grad_norm": 0.6365310549736023, "learning_rate": 7.84981279611311e-05, "loss": 2.3526, "step": 36120 }, { "epoch": 1.2535111837408408e-13, "grad_norm": 0.5887490510940552, "learning_rate": 7.847609671860097e-05, "loss": 2.3266, "step": 36130 }, { "epoch": 1.2538581284360362e-13, "grad_norm": 0.5743741989135742, "learning_rate": 7.845406393797237e-05, "loss": 2.3602, "step": 36140 }, { "epoch": 1.2542050731312315e-13, "grad_norm": 0.6046597957611084, "learning_rate": 7.84320296227296e-05, "loss": 2.3701, "step": 36150 }, { "epoch": 1.254552017826427e-13, "grad_norm": 0.5983707308769226, "learning_rate": 7.840999377635717e-05, "loss": 2.3812, "step": 36160 }, { "epoch": 1.2548989625216223e-13, "grad_norm": 0.6231013536453247, "learning_rate": 7.838795640233989e-05, "loss": 2.398, "step": 36170 }, { "epoch": 1.2552459072168176e-13, "grad_norm": 0.5954403877258301, "learning_rate": 7.836591750416275e-05, "loss": 2.3515, "step": 36180 }, { "epoch": 1.255592851912013e-13, "grad_norm": 0.5906981825828552, "learning_rate": 7.834387708531106e-05, "loss": 2.3864, "step": 36190 }, { "epoch": 1.2559397966072083e-13, "grad_norm": 0.5987058281898499, "learning_rate": 7.832183514927027e-05, "loss": 2.3878, "step": 36200 }, { "epoch": 1.2562867413024037e-13, "grad_norm": 0.5910376906394958, "learning_rate": 7.829979169952616e-05, "loss": 2.358, "step": 36210 }, { "epoch": 1.256633685997599e-13, "grad_norm": 0.5966753363609314, "learning_rate": 7.827774673956467e-05, "loss": 2.3422, "step": 36220 }, { "epoch": 1.2569806306927944e-13, "grad_norm": 0.5908222794532776, "learning_rate": 7.825570027287208e-05, "loss": 2.3355, "step": 36230 }, { "epoch": 1.2573275753879898e-13, "grad_norm": 0.5892260670661926, "learning_rate": 7.823365230293482e-05, "loss": 2.3478, "step": 36240 }, { "epoch": 1.2576745200831851e-13, "grad_norm": 0.5660959482192993, "learning_rate": 7.821160283323957e-05, "loss": 2.3365, "step": 36250 }, { "epoch": 1.2580214647783805e-13, "grad_norm": 0.5698431730270386, "learning_rate": 7.818955186727327e-05, "loss": 2.3841, "step": 36260 }, { "epoch": 1.2583684094735759e-13, "grad_norm": 0.6009196639060974, "learning_rate": 7.81674994085231e-05, "loss": 2.3548, "step": 36270 }, { "epoch": 1.2587153541687712e-13, "grad_norm": 0.5771963000297546, "learning_rate": 7.814544546047648e-05, "loss": 2.3361, "step": 36280 }, { "epoch": 1.2590622988639666e-13, "grad_norm": 0.5903156399726868, "learning_rate": 7.812339002662102e-05, "loss": 2.3515, "step": 36290 }, { "epoch": 1.259409243559162e-13, "grad_norm": 0.5504400134086609, "learning_rate": 7.810133311044464e-05, "loss": 2.3985, "step": 36300 }, { "epoch": 1.2597561882543573e-13, "grad_norm": 0.5811771154403687, "learning_rate": 7.807927471543541e-05, "loss": 2.323, "step": 36310 }, { "epoch": 1.2601031329495527e-13, "grad_norm": 0.5611038208007812, "learning_rate": 7.80572148450817e-05, "loss": 2.3162, "step": 36320 }, { "epoch": 1.260450077644748e-13, "grad_norm": 0.5964344143867493, "learning_rate": 7.803515350287206e-05, "loss": 2.3396, "step": 36330 }, { "epoch": 1.2607970223399434e-13, "grad_norm": 0.5568501949310303, "learning_rate": 7.801309069229531e-05, "loss": 2.3581, "step": 36340 }, { "epoch": 1.2611439670351388e-13, "grad_norm": 0.7028009295463562, "learning_rate": 7.799102641684054e-05, "loss": 2.3201, "step": 36350 }, { "epoch": 1.261490911730334e-13, "grad_norm": 0.6251636743545532, "learning_rate": 7.796896067999697e-05, "loss": 2.3448, "step": 36360 }, { "epoch": 1.2618378564255295e-13, "grad_norm": 0.5662901997566223, "learning_rate": 7.794689348525413e-05, "loss": 2.3329, "step": 36370 }, { "epoch": 1.2621848011207248e-13, "grad_norm": 0.5937387943267822, "learning_rate": 7.792482483610176e-05, "loss": 2.3369, "step": 36380 }, { "epoch": 1.2625317458159202e-13, "grad_norm": 0.6042428612709045, "learning_rate": 7.790275473602982e-05, "loss": 2.3396, "step": 36390 }, { "epoch": 1.2628786905111156e-13, "grad_norm": 0.5750364065170288, "learning_rate": 7.788068318852849e-05, "loss": 2.3611, "step": 36400 }, { "epoch": 1.263225635206311e-13, "grad_norm": 0.5893942713737488, "learning_rate": 7.785861019708824e-05, "loss": 2.3665, "step": 36410 }, { "epoch": 1.2635725799015063e-13, "grad_norm": 0.5856435894966125, "learning_rate": 7.783653576519965e-05, "loss": 2.3358, "step": 36420 }, { "epoch": 1.2639195245967016e-13, "grad_norm": 0.6043999195098877, "learning_rate": 7.781445989635366e-05, "loss": 2.3946, "step": 36430 }, { "epoch": 1.264266469291897e-13, "grad_norm": 0.5712578892707825, "learning_rate": 7.779238259404139e-05, "loss": 2.3735, "step": 36440 }, { "epoch": 1.2646134139870924e-13, "grad_norm": 0.5743290781974792, "learning_rate": 7.777030386175414e-05, "loss": 2.3489, "step": 36450 }, { "epoch": 1.2649603586822877e-13, "grad_norm": 0.5987901091575623, "learning_rate": 7.774822370298347e-05, "loss": 2.3361, "step": 36460 }, { "epoch": 1.265307303377483e-13, "grad_norm": 0.5770807266235352, "learning_rate": 7.77261421212212e-05, "loss": 2.3244, "step": 36470 }, { "epoch": 1.2656542480726785e-13, "grad_norm": 0.6094013452529907, "learning_rate": 7.77040591199593e-05, "loss": 2.3472, "step": 36480 }, { "epoch": 1.2660011927678738e-13, "grad_norm": 0.5993067622184753, "learning_rate": 7.768197470269004e-05, "loss": 2.3646, "step": 36490 }, { "epoch": 1.2663481374630692e-13, "grad_norm": 0.5524348616600037, "learning_rate": 7.765988887290585e-05, "loss": 2.3418, "step": 36500 }, { "epoch": 1.2666950821582645e-13, "grad_norm": 0.5890040993690491, "learning_rate": 7.763780163409944e-05, "loss": 2.3821, "step": 36510 }, { "epoch": 1.26704202685346e-13, "grad_norm": 0.5658707022666931, "learning_rate": 7.761571298976371e-05, "loss": 2.3365, "step": 36520 }, { "epoch": 1.2673889715486553e-13, "grad_norm": 0.5977237820625305, "learning_rate": 7.759362294339179e-05, "loss": 2.3649, "step": 36530 }, { "epoch": 1.2677359162438506e-13, "grad_norm": 0.5630584359169006, "learning_rate": 7.7571531498477e-05, "loss": 2.3251, "step": 36540 }, { "epoch": 1.268082860939046e-13, "grad_norm": 0.5769416093826294, "learning_rate": 7.7549438658513e-05, "loss": 2.3477, "step": 36550 }, { "epoch": 1.2684298056342413e-13, "grad_norm": 0.5734657645225525, "learning_rate": 7.752734442699348e-05, "loss": 2.3528, "step": 36560 }, { "epoch": 1.2687767503294367e-13, "grad_norm": 0.6009817123413086, "learning_rate": 7.750524880741251e-05, "loss": 2.3179, "step": 36570 }, { "epoch": 1.269123695024632e-13, "grad_norm": 0.6449573636054993, "learning_rate": 7.748315180326428e-05, "loss": 2.3614, "step": 36580 }, { "epoch": 1.2694706397198274e-13, "grad_norm": 0.5684632062911987, "learning_rate": 7.746105341804329e-05, "loss": 2.3051, "step": 36590 }, { "epoch": 1.2698175844150228e-13, "grad_norm": 0.5985947251319885, "learning_rate": 7.743895365524418e-05, "loss": 2.3084, "step": 36600 }, { "epoch": 1.2701645291102182e-13, "grad_norm": 0.5995283126831055, "learning_rate": 7.741685251836186e-05, "loss": 2.3571, "step": 36610 }, { "epoch": 1.2705114738054135e-13, "grad_norm": 0.5804901123046875, "learning_rate": 7.739475001089142e-05, "loss": 2.3511, "step": 36620 }, { "epoch": 1.270858418500609e-13, "grad_norm": 0.5820469856262207, "learning_rate": 7.737264613632818e-05, "loss": 2.3647, "step": 36630 }, { "epoch": 1.2712053631958042e-13, "grad_norm": 0.5937743782997131, "learning_rate": 7.735054089816764e-05, "loss": 2.355, "step": 36640 }, { "epoch": 1.2715523078909996e-13, "grad_norm": 0.5627845525741577, "learning_rate": 7.732843429990562e-05, "loss": 2.3402, "step": 36650 }, { "epoch": 1.271899252586195e-13, "grad_norm": 0.6231589913368225, "learning_rate": 7.730632634503805e-05, "loss": 2.3496, "step": 36660 }, { "epoch": 1.2722461972813903e-13, "grad_norm": 0.5827193856239319, "learning_rate": 7.728421703706113e-05, "loss": 2.3567, "step": 36670 }, { "epoch": 1.2725931419765857e-13, "grad_norm": 0.573498547077179, "learning_rate": 7.726210637947127e-05, "loss": 2.3575, "step": 36680 }, { "epoch": 1.272940086671781e-13, "grad_norm": 0.5856126546859741, "learning_rate": 7.723999437576502e-05, "loss": 2.3697, "step": 36690 }, { "epoch": 1.2732870313669764e-13, "grad_norm": 0.5759071111679077, "learning_rate": 7.721788102943927e-05, "loss": 2.3386, "step": 36700 }, { "epoch": 1.2736339760621718e-13, "grad_norm": 0.6075596809387207, "learning_rate": 7.7195766343991e-05, "loss": 2.3631, "step": 36710 }, { "epoch": 1.2739809207573671e-13, "grad_norm": 0.5848450064659119, "learning_rate": 7.717365032291751e-05, "loss": 2.3478, "step": 36720 }, { "epoch": 1.2743278654525625e-13, "grad_norm": 0.5979676246643066, "learning_rate": 7.71515329697162e-05, "loss": 2.3288, "step": 36730 }, { "epoch": 1.2746748101477579e-13, "grad_norm": 0.5719823837280273, "learning_rate": 7.712941428788481e-05, "loss": 2.3447, "step": 36740 }, { "epoch": 1.2750217548429532e-13, "grad_norm": 0.5732263326644897, "learning_rate": 7.710729428092116e-05, "loss": 2.3941, "step": 36750 }, { "epoch": 1.2753686995381486e-13, "grad_norm": 0.6119689345359802, "learning_rate": 7.708517295232337e-05, "loss": 2.3528, "step": 36760 }, { "epoch": 1.275715644233344e-13, "grad_norm": 0.6098507642745972, "learning_rate": 7.706305030558972e-05, "loss": 2.3347, "step": 36770 }, { "epoch": 1.2760625889285393e-13, "grad_norm": 0.5990791916847229, "learning_rate": 7.704092634421872e-05, "loss": 2.3068, "step": 36780 }, { "epoch": 1.2764095336237347e-13, "grad_norm": 0.5890535116195679, "learning_rate": 7.70188010717091e-05, "loss": 2.3557, "step": 36790 }, { "epoch": 1.27675647831893e-13, "grad_norm": 0.5954375863075256, "learning_rate": 7.699667449155977e-05, "loss": 2.3579, "step": 36800 }, { "epoch": 1.2771034230141254e-13, "grad_norm": 0.5613542199134827, "learning_rate": 7.697454660726982e-05, "loss": 2.3655, "step": 36810 }, { "epoch": 1.2774503677093207e-13, "grad_norm": 0.5686646103858948, "learning_rate": 7.695241742233866e-05, "loss": 2.3219, "step": 36820 }, { "epoch": 1.277797312404516e-13, "grad_norm": 0.5891755819320679, "learning_rate": 7.693028694026579e-05, "loss": 2.3684, "step": 36830 }, { "epoch": 1.2781442570997115e-13, "grad_norm": 0.5915035605430603, "learning_rate": 7.690815516455096e-05, "loss": 2.3736, "step": 36840 }, { "epoch": 1.2784912017949068e-13, "grad_norm": 0.5849133729934692, "learning_rate": 7.68860220986941e-05, "loss": 2.378, "step": 36850 }, { "epoch": 1.2788381464901022e-13, "grad_norm": 0.5767281651496887, "learning_rate": 7.686388774619537e-05, "loss": 2.3462, "step": 36860 }, { "epoch": 1.2791850911852976e-13, "grad_norm": 2.4343059062957764, "learning_rate": 7.684175211055517e-05, "loss": 2.3669, "step": 36870 }, { "epoch": 1.279532035880493e-13, "grad_norm": 0.5852108001708984, "learning_rate": 7.6819615195274e-05, "loss": 2.3717, "step": 36880 }, { "epoch": 1.2798789805756883e-13, "grad_norm": 0.5762050747871399, "learning_rate": 7.679747700385265e-05, "loss": 2.3684, "step": 36890 }, { "epoch": 1.2802259252708836e-13, "grad_norm": 0.5887752771377563, "learning_rate": 7.67753375397921e-05, "loss": 2.3505, "step": 36900 }, { "epoch": 1.280572869966079e-13, "grad_norm": 0.5793030858039856, "learning_rate": 7.67531968065935e-05, "loss": 2.3605, "step": 36910 }, { "epoch": 1.2809198146612744e-13, "grad_norm": 0.574053943157196, "learning_rate": 7.673105480775822e-05, "loss": 2.3388, "step": 36920 }, { "epoch": 1.2812667593564697e-13, "grad_norm": 0.6152908205986023, "learning_rate": 7.67089115467878e-05, "loss": 2.3527, "step": 36930 }, { "epoch": 1.281613704051665e-13, "grad_norm": 0.581163227558136, "learning_rate": 7.668676702718407e-05, "loss": 2.3707, "step": 36940 }, { "epoch": 1.2819606487468604e-13, "grad_norm": 0.6011641621589661, "learning_rate": 7.666462125244892e-05, "loss": 2.3661, "step": 36950 }, { "epoch": 1.2823075934420558e-13, "grad_norm": 0.5814068913459778, "learning_rate": 7.664247422608455e-05, "loss": 2.3193, "step": 36960 }, { "epoch": 1.2826545381372512e-13, "grad_norm": 0.5940492749214172, "learning_rate": 7.662032595159333e-05, "loss": 2.3601, "step": 36970 }, { "epoch": 1.2830014828324465e-13, "grad_norm": 0.5870245695114136, "learning_rate": 7.659817643247781e-05, "loss": 2.3489, "step": 36980 }, { "epoch": 1.283348427527642e-13, "grad_norm": 0.5829894542694092, "learning_rate": 7.657602567224075e-05, "loss": 2.408, "step": 36990 }, { "epoch": 1.2836953722228372e-13, "grad_norm": 0.5632487535476685, "learning_rate": 7.65538736743851e-05, "loss": 2.3422, "step": 37000 }, { "epoch": 1.2840423169180326e-13, "grad_norm": 0.6148874163627625, "learning_rate": 7.6531720442414e-05, "loss": 2.3257, "step": 37010 }, { "epoch": 1.284389261613228e-13, "grad_norm": 0.5770079493522644, "learning_rate": 7.650956597983077e-05, "loss": 2.3447, "step": 37020 }, { "epoch": 1.2847362063084233e-13, "grad_norm": 0.6138296723365784, "learning_rate": 7.648741029013902e-05, "loss": 2.3301, "step": 37030 }, { "epoch": 1.2850831510036187e-13, "grad_norm": 0.5736187696456909, "learning_rate": 7.646525337684241e-05, "loss": 2.3486, "step": 37040 }, { "epoch": 1.285430095698814e-13, "grad_norm": 0.5871807336807251, "learning_rate": 7.644309524344489e-05, "loss": 2.3189, "step": 37050 }, { "epoch": 1.2857770403940094e-13, "grad_norm": 0.589050829410553, "learning_rate": 7.642093589345059e-05, "loss": 2.3129, "step": 37060 }, { "epoch": 1.2861239850892048e-13, "grad_norm": 0.5765824317932129, "learning_rate": 7.639877533036379e-05, "loss": 2.4167, "step": 37070 }, { "epoch": 1.2864709297844001e-13, "grad_norm": 0.5702725648880005, "learning_rate": 7.637661355768905e-05, "loss": 2.3406, "step": 37080 }, { "epoch": 1.2868178744795955e-13, "grad_norm": 0.584068775177002, "learning_rate": 7.6354450578931e-05, "loss": 2.3568, "step": 37090 }, { "epoch": 1.2871648191747909e-13, "grad_norm": 0.5850456953048706, "learning_rate": 7.633228639759455e-05, "loss": 2.3535, "step": 37100 }, { "epoch": 1.2875117638699862e-13, "grad_norm": 0.6020603179931641, "learning_rate": 7.631012101718479e-05, "loss": 2.3308, "step": 37110 }, { "epoch": 1.2878587085651816e-13, "grad_norm": 0.5799092054367065, "learning_rate": 7.628795444120697e-05, "loss": 2.297, "step": 37120 }, { "epoch": 1.288205653260377e-13, "grad_norm": 0.6513389945030212, "learning_rate": 7.626578667316651e-05, "loss": 2.3487, "step": 37130 }, { "epoch": 1.2885525979555723e-13, "grad_norm": 0.5978080630302429, "learning_rate": 7.624361771656911e-05, "loss": 2.3679, "step": 37140 }, { "epoch": 1.2888995426507677e-13, "grad_norm": 0.5892207622528076, "learning_rate": 7.622144757492057e-05, "loss": 2.3629, "step": 37150 }, { "epoch": 1.289246487345963e-13, "grad_norm": 0.5771394371986389, "learning_rate": 7.619927625172692e-05, "loss": 2.3378, "step": 37160 }, { "epoch": 1.2895934320411584e-13, "grad_norm": 0.5888023376464844, "learning_rate": 7.617710375049435e-05, "loss": 2.3723, "step": 37170 }, { "epoch": 1.2899403767363538e-13, "grad_norm": 0.6039904356002808, "learning_rate": 7.615493007472923e-05, "loss": 2.381, "step": 37180 }, { "epoch": 1.290287321431549e-13, "grad_norm": 0.5724816918373108, "learning_rate": 7.61327552279382e-05, "loss": 2.3587, "step": 37190 }, { "epoch": 1.2906342661267445e-13, "grad_norm": 0.6237884759902954, "learning_rate": 7.611057921362796e-05, "loss": 2.3474, "step": 37200 }, { "epoch": 1.2909812108219398e-13, "grad_norm": 0.5783429741859436, "learning_rate": 7.608840203530547e-05, "loss": 2.3056, "step": 37210 }, { "epoch": 1.2913281555171352e-13, "grad_norm": 0.5902111530303955, "learning_rate": 7.606622369647789e-05, "loss": 2.3546, "step": 37220 }, { "epoch": 1.2916751002123306e-13, "grad_norm": 0.6177036762237549, "learning_rate": 7.604404420065249e-05, "loss": 2.3321, "step": 37230 }, { "epoch": 1.292022044907526e-13, "grad_norm": 0.5903717875480652, "learning_rate": 7.60218635513368e-05, "loss": 2.3659, "step": 37240 }, { "epoch": 1.2923689896027213e-13, "grad_norm": 0.5844502449035645, "learning_rate": 7.599968175203848e-05, "loss": 2.3794, "step": 37250 }, { "epoch": 1.2927159342979166e-13, "grad_norm": 0.6037951111793518, "learning_rate": 7.597749880626541e-05, "loss": 2.3845, "step": 37260 }, { "epoch": 1.293062878993112e-13, "grad_norm": 0.5706465244293213, "learning_rate": 7.59553147175256e-05, "loss": 2.3505, "step": 37270 }, { "epoch": 1.2934098236883074e-13, "grad_norm": 0.6024826765060425, "learning_rate": 7.59331294893273e-05, "loss": 2.3373, "step": 37280 }, { "epoch": 1.2937567683835027e-13, "grad_norm": 0.5728061199188232, "learning_rate": 7.59109431251789e-05, "loss": 2.3392, "step": 37290 }, { "epoch": 1.294103713078698e-13, "grad_norm": 0.6172255873680115, "learning_rate": 7.588875562858899e-05, "loss": 2.3859, "step": 37300 }, { "epoch": 1.2944506577738935e-13, "grad_norm": 0.5581206679344177, "learning_rate": 7.586656700306633e-05, "loss": 2.3342, "step": 37310 }, { "epoch": 1.2947976024690888e-13, "grad_norm": 0.5677633285522461, "learning_rate": 7.584437725211986e-05, "loss": 2.3461, "step": 37320 }, { "epoch": 1.2951445471642842e-13, "grad_norm": 0.6304829120635986, "learning_rate": 7.58221863792587e-05, "loss": 2.3177, "step": 37330 }, { "epoch": 1.2954914918594795e-13, "grad_norm": 0.598619282245636, "learning_rate": 7.579999438799212e-05, "loss": 2.3904, "step": 37340 }, { "epoch": 1.295838436554675e-13, "grad_norm": 0.5886795520782471, "learning_rate": 7.577780128182963e-05, "loss": 2.3598, "step": 37350 }, { "epoch": 1.2961853812498703e-13, "grad_norm": 0.5760323405265808, "learning_rate": 7.575560706428087e-05, "loss": 2.3896, "step": 37360 }, { "epoch": 1.2965323259450656e-13, "grad_norm": 0.5940855741500854, "learning_rate": 7.573341173885563e-05, "loss": 2.3604, "step": 37370 }, { "epoch": 1.296879270640261e-13, "grad_norm": 0.5653848052024841, "learning_rate": 7.571121530906397e-05, "loss": 2.3977, "step": 37380 }, { "epoch": 1.2972262153354563e-13, "grad_norm": 0.6002959609031677, "learning_rate": 7.5689017778416e-05, "loss": 2.3493, "step": 37390 }, { "epoch": 1.2975731600306517e-13, "grad_norm": 0.5900611281394958, "learning_rate": 7.566681915042213e-05, "loss": 2.3808, "step": 37400 }, { "epoch": 1.297920104725847e-13, "grad_norm": 0.5764289498329163, "learning_rate": 7.564461942859282e-05, "loss": 2.3756, "step": 37410 }, { "epoch": 1.2982670494210424e-13, "grad_norm": 0.6303838491439819, "learning_rate": 7.562241861643882e-05, "loss": 2.3282, "step": 37420 }, { "epoch": 1.2986139941162378e-13, "grad_norm": 0.5626283288002014, "learning_rate": 7.560021671747099e-05, "loss": 2.3393, "step": 37430 }, { "epoch": 1.2989609388114332e-13, "grad_norm": 0.5882997512817383, "learning_rate": 7.557801373520031e-05, "loss": 2.3675, "step": 37440 }, { "epoch": 1.2993078835066285e-13, "grad_norm": 0.5655671954154968, "learning_rate": 7.555580967313807e-05, "loss": 2.331, "step": 37450 }, { "epoch": 1.299654828201824e-13, "grad_norm": 0.6184231638908386, "learning_rate": 7.55336045347956e-05, "loss": 2.3622, "step": 37460 }, { "epoch": 1.3000017728970192e-13, "grad_norm": 0.5957379341125488, "learning_rate": 7.551139832368446e-05, "loss": 2.3506, "step": 37470 }, { "epoch": 1.3003487175922146e-13, "grad_norm": 0.5795299410820007, "learning_rate": 7.54891910433164e-05, "loss": 2.3835, "step": 37480 }, { "epoch": 1.30069566228741e-13, "grad_norm": 0.5756956338882446, "learning_rate": 7.546698269720327e-05, "loss": 2.3599, "step": 37490 }, { "epoch": 1.3010426069826053e-13, "grad_norm": 0.6083265542984009, "learning_rate": 7.544477328885716e-05, "loss": 2.3441, "step": 37500 }, { "epoch": 1.3013895516778007e-13, "grad_norm": 0.5908844470977783, "learning_rate": 7.542256282179027e-05, "loss": 2.396, "step": 37510 }, { "epoch": 1.301736496372996e-13, "grad_norm": 0.580263614654541, "learning_rate": 7.540035129951501e-05, "loss": 2.3497, "step": 37520 }, { "epoch": 1.3020834410681914e-13, "grad_norm": 0.5815404057502747, "learning_rate": 7.537813872554393e-05, "loss": 2.3368, "step": 37530 }, { "epoch": 1.3024303857633868e-13, "grad_norm": 0.576303243637085, "learning_rate": 7.535592510338978e-05, "loss": 2.3357, "step": 37540 }, { "epoch": 1.3027773304585821e-13, "grad_norm": 0.5747520327568054, "learning_rate": 7.533371043656542e-05, "loss": 2.3279, "step": 37550 }, { "epoch": 1.3031242751537775e-13, "grad_norm": 0.6084715127944946, "learning_rate": 7.531149472858396e-05, "loss": 2.3401, "step": 37560 }, { "epoch": 1.3034712198489729e-13, "grad_norm": 0.5791711807250977, "learning_rate": 7.528927798295857e-05, "loss": 2.3992, "step": 37570 }, { "epoch": 1.3038181645441682e-13, "grad_norm": 0.620807945728302, "learning_rate": 7.526706020320266e-05, "loss": 2.3561, "step": 37580 }, { "epoch": 1.3041651092393636e-13, "grad_norm": 0.5852569937705994, "learning_rate": 7.524484139282979e-05, "loss": 2.348, "step": 37590 }, { "epoch": 1.304512053934559e-13, "grad_norm": 0.5759270191192627, "learning_rate": 7.522262155535363e-05, "loss": 2.328, "step": 37600 }, { "epoch": 1.3048589986297543e-13, "grad_norm": 0.6029862761497498, "learning_rate": 7.52004006942881e-05, "loss": 2.3522, "step": 37610 }, { "epoch": 1.3052059433249497e-13, "grad_norm": 0.6137840151786804, "learning_rate": 7.517817881314724e-05, "loss": 2.3492, "step": 37620 }, { "epoch": 1.305552888020145e-13, "grad_norm": 0.6244508028030396, "learning_rate": 7.515595591544524e-05, "loss": 2.3237, "step": 37630 }, { "epoch": 1.3058998327153404e-13, "grad_norm": 0.6120102405548096, "learning_rate": 7.513373200469643e-05, "loss": 2.3503, "step": 37640 }, { "epoch": 1.3062467774105357e-13, "grad_norm": 0.6107417345046997, "learning_rate": 7.511150708441537e-05, "loss": 2.3569, "step": 37650 }, { "epoch": 1.306593722105731e-13, "grad_norm": 0.6533042192459106, "learning_rate": 7.508928115811673e-05, "loss": 2.3368, "step": 37660 }, { "epoch": 1.3069406668009265e-13, "grad_norm": 0.6230975985527039, "learning_rate": 7.506705422931532e-05, "loss": 2.3796, "step": 37670 }, { "epoch": 1.3072876114961218e-13, "grad_norm": 0.6039360761642456, "learning_rate": 7.50448263015262e-05, "loss": 2.3283, "step": 37680 }, { "epoch": 1.3076345561913172e-13, "grad_norm": 0.5897181034088135, "learning_rate": 7.502259737826446e-05, "loss": 2.3328, "step": 37690 }, { "epoch": 1.3079815008865125e-13, "grad_norm": 0.5942196846008301, "learning_rate": 7.500036746304545e-05, "loss": 2.3359, "step": 37700 }, { "epoch": 1.308328445581708e-13, "grad_norm": 0.5833100080490112, "learning_rate": 7.497813655938461e-05, "loss": 2.3551, "step": 37710 }, { "epoch": 1.3086753902769033e-13, "grad_norm": 0.5780715346336365, "learning_rate": 7.49559046707976e-05, "loss": 2.3437, "step": 37720 }, { "epoch": 1.3090223349720986e-13, "grad_norm": 0.6411371231079102, "learning_rate": 7.493367180080018e-05, "loss": 2.374, "step": 37730 }, { "epoch": 1.309369279667294e-13, "grad_norm": 0.5947141051292419, "learning_rate": 7.49114379529083e-05, "loss": 2.3442, "step": 37740 }, { "epoch": 1.3097162243624894e-13, "grad_norm": 0.5940773487091064, "learning_rate": 7.488920313063806e-05, "loss": 2.3179, "step": 37750 }, { "epoch": 1.3100631690576847e-13, "grad_norm": 0.5963879227638245, "learning_rate": 7.486696733750567e-05, "loss": 2.3568, "step": 37760 }, { "epoch": 1.31041011375288e-13, "grad_norm": 0.5829976797103882, "learning_rate": 7.484473057702755e-05, "loss": 2.3119, "step": 37770 }, { "epoch": 1.3107570584480754e-13, "grad_norm": 0.594672441482544, "learning_rate": 7.482249285272026e-05, "loss": 2.3636, "step": 37780 }, { "epoch": 1.3111040031432708e-13, "grad_norm": 0.5791453123092651, "learning_rate": 7.480025416810051e-05, "loss": 2.3721, "step": 37790 }, { "epoch": 1.3114509478384662e-13, "grad_norm": 0.586618185043335, "learning_rate": 7.477801452668513e-05, "loss": 2.3426, "step": 37800 }, { "epoch": 1.3117978925336615e-13, "grad_norm": 0.59465491771698, "learning_rate": 7.475577393199115e-05, "loss": 2.3967, "step": 37810 }, { "epoch": 1.312144837228857e-13, "grad_norm": 0.5891191959381104, "learning_rate": 7.473353238753571e-05, "loss": 2.3247, "step": 37820 }, { "epoch": 1.3124917819240522e-13, "grad_norm": 0.6337776780128479, "learning_rate": 7.471128989683613e-05, "loss": 2.3393, "step": 37830 }, { "epoch": 1.3128387266192476e-13, "grad_norm": 0.6210059523582458, "learning_rate": 7.468904646340989e-05, "loss": 2.3092, "step": 37840 }, { "epoch": 1.313185671314443e-13, "grad_norm": 0.6032281517982483, "learning_rate": 7.466680209077454e-05, "loss": 2.3494, "step": 37850 }, { "epoch": 1.3135326160096383e-13, "grad_norm": 0.5958722829818726, "learning_rate": 7.46445567824479e-05, "loss": 2.3272, "step": 37860 }, { "epoch": 1.3138795607048337e-13, "grad_norm": 0.5783767700195312, "learning_rate": 7.462231054194782e-05, "loss": 2.3619, "step": 37870 }, { "epoch": 1.314226505400029e-13, "grad_norm": 0.6165918111801147, "learning_rate": 7.46000633727924e-05, "loss": 2.3532, "step": 37880 }, { "epoch": 1.3145734500952244e-13, "grad_norm": 0.6009612679481506, "learning_rate": 7.45778152784998e-05, "loss": 2.3559, "step": 37890 }, { "epoch": 1.3149203947904198e-13, "grad_norm": 0.5992047786712646, "learning_rate": 7.455556626258841e-05, "loss": 2.3443, "step": 37900 }, { "epoch": 1.3152673394856151e-13, "grad_norm": 0.6269800066947937, "learning_rate": 7.453331632857667e-05, "loss": 2.3445, "step": 37910 }, { "epoch": 1.3156142841808105e-13, "grad_norm": 0.6006940007209778, "learning_rate": 7.451106547998322e-05, "loss": 2.3369, "step": 37920 }, { "epoch": 1.3159612288760059e-13, "grad_norm": 0.5995170474052429, "learning_rate": 7.448881372032686e-05, "loss": 2.3059, "step": 37930 }, { "epoch": 1.3163081735712012e-13, "grad_norm": 0.588946521282196, "learning_rate": 7.446656105312651e-05, "loss": 2.3242, "step": 37940 }, { "epoch": 1.3166551182663966e-13, "grad_norm": 0.5987786650657654, "learning_rate": 7.444430748190126e-05, "loss": 2.3415, "step": 37950 }, { "epoch": 1.317002062961592e-13, "grad_norm": 0.6117120385169983, "learning_rate": 7.44220530101703e-05, "loss": 2.3619, "step": 37960 }, { "epoch": 1.3173490076567873e-13, "grad_norm": 0.6144567131996155, "learning_rate": 7.439979764145296e-05, "loss": 2.385, "step": 37970 }, { "epoch": 1.3176959523519827e-13, "grad_norm": 0.6778997182846069, "learning_rate": 7.437754137926876e-05, "loss": 2.3487, "step": 37980 }, { "epoch": 1.318042897047178e-13, "grad_norm": 0.5943764448165894, "learning_rate": 7.435528422713735e-05, "loss": 2.3214, "step": 37990 }, { "epoch": 1.3183898417423734e-13, "grad_norm": 0.5998103618621826, "learning_rate": 7.433302618857847e-05, "loss": 2.3555, "step": 38000 }, { "epoch": 1.3187367864375688e-13, "grad_norm": 0.600451648235321, "learning_rate": 7.431076726711207e-05, "loss": 2.3175, "step": 38010 }, { "epoch": 1.319083731132764e-13, "grad_norm": 0.5890810489654541, "learning_rate": 7.428850746625819e-05, "loss": 2.3268, "step": 38020 }, { "epoch": 1.3194306758279595e-13, "grad_norm": 0.6368334293365479, "learning_rate": 7.426624678953703e-05, "loss": 2.3628, "step": 38030 }, { "epoch": 1.3197776205231548e-13, "grad_norm": 0.634029746055603, "learning_rate": 7.424398524046891e-05, "loss": 2.3447, "step": 38040 }, { "epoch": 1.3201245652183502e-13, "grad_norm": 0.6016285419464111, "learning_rate": 7.422172282257434e-05, "loss": 2.319, "step": 38050 }, { "epoch": 1.3204715099135456e-13, "grad_norm": 0.5867778658866882, "learning_rate": 7.41994595393739e-05, "loss": 2.3202, "step": 38060 }, { "epoch": 1.320818454608741e-13, "grad_norm": 0.6044045686721802, "learning_rate": 7.417719539438835e-05, "loss": 2.3406, "step": 38070 }, { "epoch": 1.3211653993039363e-13, "grad_norm": 0.5669294595718384, "learning_rate": 7.415493039113854e-05, "loss": 2.3316, "step": 38080 }, { "epoch": 1.3215123439991316e-13, "grad_norm": 0.5884884595870972, "learning_rate": 7.413266453314553e-05, "loss": 2.3307, "step": 38090 }, { "epoch": 1.321859288694327e-13, "grad_norm": 0.5833523869514465, "learning_rate": 7.411039782393046e-05, "loss": 2.3224, "step": 38100 }, { "epoch": 1.3222062333895224e-13, "grad_norm": 0.589713990688324, "learning_rate": 7.408813026701462e-05, "loss": 2.359, "step": 38110 }, { "epoch": 1.3225531780847177e-13, "grad_norm": 0.580680251121521, "learning_rate": 7.406586186591943e-05, "loss": 2.3546, "step": 38120 }, { "epoch": 1.322900122779913e-13, "grad_norm": 0.5939783453941345, "learning_rate": 7.404359262416645e-05, "loss": 2.3343, "step": 38130 }, { "epoch": 1.3232470674751085e-13, "grad_norm": 0.5916965007781982, "learning_rate": 7.402132254527734e-05, "loss": 2.3291, "step": 38140 }, { "epoch": 1.3235940121703038e-13, "grad_norm": 0.5707841515541077, "learning_rate": 7.399905163277395e-05, "loss": 2.3127, "step": 38150 }, { "epoch": 1.3239409568654992e-13, "grad_norm": 0.6008639931678772, "learning_rate": 7.397677989017825e-05, "loss": 2.3243, "step": 38160 }, { "epoch": 1.3242879015606945e-13, "grad_norm": 0.5888257622718811, "learning_rate": 7.395450732101228e-05, "loss": 2.3011, "step": 38170 }, { "epoch": 1.32463484625589e-13, "grad_norm": 0.6137056946754456, "learning_rate": 7.39322339287983e-05, "loss": 2.3419, "step": 38180 }, { "epoch": 1.3249817909510853e-13, "grad_norm": 0.5802078247070312, "learning_rate": 7.390995971705862e-05, "loss": 2.3447, "step": 38190 }, { "epoch": 1.3253287356462806e-13, "grad_norm": 0.6016493439674377, "learning_rate": 7.388768468931572e-05, "loss": 2.3172, "step": 38200 }, { "epoch": 1.325675680341476e-13, "grad_norm": 0.6227039694786072, "learning_rate": 7.386540884909223e-05, "loss": 2.3588, "step": 38210 }, { "epoch": 1.3260226250366713e-13, "grad_norm": 0.6121275424957275, "learning_rate": 7.384313219991085e-05, "loss": 2.2906, "step": 38220 }, { "epoch": 1.3263695697318667e-13, "grad_norm": 0.5833830237388611, "learning_rate": 7.382085474529448e-05, "loss": 2.3485, "step": 38230 }, { "epoch": 1.326716514427062e-13, "grad_norm": 0.6099114418029785, "learning_rate": 7.379857648876605e-05, "loss": 2.3362, "step": 38240 }, { "epoch": 1.3270634591222574e-13, "grad_norm": 0.5765507221221924, "learning_rate": 7.377629743384872e-05, "loss": 2.3212, "step": 38250 }, { "epoch": 1.3274104038174528e-13, "grad_norm": 0.5964168906211853, "learning_rate": 7.375401758406571e-05, "loss": 2.3336, "step": 38260 }, { "epoch": 1.3277573485126482e-13, "grad_norm": 0.588639497756958, "learning_rate": 7.373173694294042e-05, "loss": 2.3484, "step": 38270 }, { "epoch": 1.3281042932078435e-13, "grad_norm": 0.6069753766059875, "learning_rate": 7.370945551399628e-05, "loss": 2.3468, "step": 38280 }, { "epoch": 1.3284512379030389e-13, "grad_norm": 0.6284986138343811, "learning_rate": 7.368717330075698e-05, "loss": 2.3327, "step": 38290 }, { "epoch": 1.3287981825982342e-13, "grad_norm": 1.5802029371261597, "learning_rate": 7.366489030674618e-05, "loss": 2.3521, "step": 38300 }, { "epoch": 1.3291451272934296e-13, "grad_norm": 0.5950668454170227, "learning_rate": 7.364260653548781e-05, "loss": 2.3193, "step": 38310 }, { "epoch": 1.329492071988625e-13, "grad_norm": 0.6216443181037903, "learning_rate": 7.362032199050584e-05, "loss": 2.3534, "step": 38320 }, { "epoch": 1.3298390166838203e-13, "grad_norm": 0.6183179020881653, "learning_rate": 7.359803667532435e-05, "loss": 2.3698, "step": 38330 }, { "epoch": 1.3301859613790157e-13, "grad_norm": 0.6219282150268555, "learning_rate": 7.35757505934676e-05, "loss": 2.3462, "step": 38340 }, { "epoch": 1.330532906074211e-13, "grad_norm": 0.5666971802711487, "learning_rate": 7.355346374845994e-05, "loss": 2.3049, "step": 38350 }, { "epoch": 1.3308798507694064e-13, "grad_norm": 0.5942696332931519, "learning_rate": 7.353117614382583e-05, "loss": 2.3507, "step": 38360 }, { "epoch": 1.3312267954646018e-13, "grad_norm": 0.570194661617279, "learning_rate": 7.350888778308988e-05, "loss": 2.3421, "step": 38370 }, { "epoch": 1.331573740159797e-13, "grad_norm": 0.586506187915802, "learning_rate": 7.348659866977678e-05, "loss": 2.3558, "step": 38380 }, { "epoch": 1.3319206848549925e-13, "grad_norm": 0.5758358836174011, "learning_rate": 7.346430880741139e-05, "loss": 2.3477, "step": 38390 }, { "epoch": 1.3322676295501878e-13, "grad_norm": 0.6196051836013794, "learning_rate": 7.344201819951863e-05, "loss": 2.3483, "step": 38400 }, { "epoch": 1.3326145742453832e-13, "grad_norm": 0.5798191428184509, "learning_rate": 7.341972684962356e-05, "loss": 2.3678, "step": 38410 }, { "epoch": 1.3329615189405786e-13, "grad_norm": 0.5963371396064758, "learning_rate": 7.33974347612514e-05, "loss": 2.3306, "step": 38420 }, { "epoch": 1.333308463635774e-13, "grad_norm": 0.589914083480835, "learning_rate": 7.337514193792747e-05, "loss": 2.3622, "step": 38430 }, { "epoch": 1.3336554083309693e-13, "grad_norm": 0.6068048477172852, "learning_rate": 7.335284838317713e-05, "loss": 2.3967, "step": 38440 }, { "epoch": 1.3340023530261647e-13, "grad_norm": 0.5919543504714966, "learning_rate": 7.333055410052594e-05, "loss": 2.3336, "step": 38450 }, { "epoch": 1.33434929772136e-13, "grad_norm": 0.5921361446380615, "learning_rate": 7.330825909349954e-05, "loss": 2.3818, "step": 38460 }, { "epoch": 1.3346962424165554e-13, "grad_norm": 0.5831207633018494, "learning_rate": 7.328596336562371e-05, "loss": 2.3444, "step": 38470 }, { "epoch": 1.3350431871117507e-13, "grad_norm": 0.5828667879104614, "learning_rate": 7.326366692042432e-05, "loss": 2.3708, "step": 38480 }, { "epoch": 1.335390131806946e-13, "grad_norm": 0.6164109706878662, "learning_rate": 7.324136976142737e-05, "loss": 2.3211, "step": 38490 }, { "epoch": 1.3357370765021415e-13, "grad_norm": 0.5765037536621094, "learning_rate": 7.321907189215894e-05, "loss": 2.3349, "step": 38500 }, { "epoch": 1.3360840211973368e-13, "grad_norm": 0.6063082218170166, "learning_rate": 7.319677331614526e-05, "loss": 2.372, "step": 38510 }, { "epoch": 1.3364309658925322e-13, "grad_norm": 0.599259614944458, "learning_rate": 7.317447403691267e-05, "loss": 2.3282, "step": 38520 }, { "epoch": 1.3367779105877275e-13, "grad_norm": 0.6293078064918518, "learning_rate": 7.31521740579876e-05, "loss": 2.3539, "step": 38530 }, { "epoch": 1.337124855282923e-13, "grad_norm": 0.6598015427589417, "learning_rate": 7.312987338289657e-05, "loss": 2.3412, "step": 38540 }, { "epoch": 1.3374717999781183e-13, "grad_norm": 0.5850784778594971, "learning_rate": 7.310757201516632e-05, "loss": 2.3274, "step": 38550 }, { "epoch": 1.3378187446733136e-13, "grad_norm": 0.6088405847549438, "learning_rate": 7.308526995832351e-05, "loss": 2.3659, "step": 38560 }, { "epoch": 1.338165689368509e-13, "grad_norm": 0.6090706586837769, "learning_rate": 7.306296721589509e-05, "loss": 2.3587, "step": 38570 }, { "epoch": 1.3385126340637044e-13, "grad_norm": 0.5789343118667603, "learning_rate": 7.304066379140804e-05, "loss": 2.3267, "step": 38580 }, { "epoch": 1.3388595787588997e-13, "grad_norm": 0.5788248777389526, "learning_rate": 7.301835968838945e-05, "loss": 2.3655, "step": 38590 }, { "epoch": 1.339206523454095e-13, "grad_norm": 0.5788350105285645, "learning_rate": 7.299605491036651e-05, "loss": 2.3476, "step": 38600 }, { "epoch": 1.3395534681492904e-13, "grad_norm": 0.5859256982803345, "learning_rate": 7.297374946086653e-05, "loss": 2.3693, "step": 38610 }, { "epoch": 1.3399004128444858e-13, "grad_norm": 0.5772829651832581, "learning_rate": 7.295144334341694e-05, "loss": 2.3636, "step": 38620 }, { "epoch": 1.3402473575396812e-13, "grad_norm": 0.5901744961738586, "learning_rate": 7.292913656154525e-05, "loss": 2.3341, "step": 38630 }, { "epoch": 1.3405943022348765e-13, "grad_norm": 0.6416703462600708, "learning_rate": 7.290682911877908e-05, "loss": 2.3373, "step": 38640 }, { "epoch": 1.340941246930072e-13, "grad_norm": 0.5912882089614868, "learning_rate": 7.288452101864617e-05, "loss": 2.3521, "step": 38650 }, { "epoch": 1.3412881916252672e-13, "grad_norm": 0.6132241487503052, "learning_rate": 7.286221226467436e-05, "loss": 2.3502, "step": 38660 }, { "epoch": 1.3416351363204626e-13, "grad_norm": 0.5642889738082886, "learning_rate": 7.283990286039156e-05, "loss": 2.3051, "step": 38670 }, { "epoch": 1.341982081015658e-13, "grad_norm": 0.5819729566574097, "learning_rate": 7.281759280932584e-05, "loss": 2.3316, "step": 38680 }, { "epoch": 1.3423290257108533e-13, "grad_norm": 0.6032084226608276, "learning_rate": 7.279528211500534e-05, "loss": 2.3456, "step": 38690 }, { "epoch": 1.3426759704060487e-13, "grad_norm": 0.6300930976867676, "learning_rate": 7.277297078095829e-05, "loss": 2.3344, "step": 38700 }, { "epoch": 1.343022915101244e-13, "grad_norm": 0.5742791295051575, "learning_rate": 7.275065881071303e-05, "loss": 2.3493, "step": 38710 }, { "epoch": 1.3433698597964394e-13, "grad_norm": 1.2698904275894165, "learning_rate": 7.272834620779804e-05, "loss": 2.3479, "step": 38720 }, { "epoch": 1.3437168044916348e-13, "grad_norm": 0.5768912434577942, "learning_rate": 7.270603297574181e-05, "loss": 2.3133, "step": 38730 }, { "epoch": 1.3440637491868301e-13, "grad_norm": 0.5719104409217834, "learning_rate": 7.268371911807304e-05, "loss": 2.3554, "step": 38740 }, { "epoch": 1.3444106938820255e-13, "grad_norm": 0.5720154047012329, "learning_rate": 7.266140463832045e-05, "loss": 2.3495, "step": 38750 }, { "epoch": 1.3447576385772209e-13, "grad_norm": 0.5915097594261169, "learning_rate": 7.26390895400129e-05, "loss": 2.3661, "step": 38760 }, { "epoch": 1.3451045832724162e-13, "grad_norm": 0.5737224817276001, "learning_rate": 7.261677382667929e-05, "loss": 2.3363, "step": 38770 }, { "epoch": 1.3454515279676116e-13, "grad_norm": 0.627443253993988, "learning_rate": 7.259445750184872e-05, "loss": 2.3227, "step": 38780 }, { "epoch": 1.345798472662807e-13, "grad_norm": 0.5898643732070923, "learning_rate": 7.257214056905026e-05, "loss": 2.3602, "step": 38790 }, { "epoch": 1.3461454173580023e-13, "grad_norm": 0.6055237054824829, "learning_rate": 7.254982303181318e-05, "loss": 2.3734, "step": 38800 }, { "epoch": 1.3464923620531977e-13, "grad_norm": 0.5925328135490417, "learning_rate": 7.252750489366678e-05, "loss": 2.3277, "step": 38810 }, { "epoch": 1.346839306748393e-13, "grad_norm": 0.5711758732795715, "learning_rate": 7.250518615814052e-05, "loss": 2.3451, "step": 38820 }, { "epoch": 1.3471862514435884e-13, "grad_norm": 0.5757656693458557, "learning_rate": 7.248286682876387e-05, "loss": 2.3355, "step": 38830 }, { "epoch": 1.3475331961387838e-13, "grad_norm": 0.580124020576477, "learning_rate": 7.24605469090665e-05, "loss": 2.3487, "step": 38840 }, { "epoch": 1.347880140833979e-13, "grad_norm": 0.6010006666183472, "learning_rate": 7.243822640257805e-05, "loss": 2.3223, "step": 38850 }, { "epoch": 1.3482270855291745e-13, "grad_norm": 0.602912425994873, "learning_rate": 7.241590531282837e-05, "loss": 2.3232, "step": 38860 }, { "epoch": 1.3485740302243698e-13, "grad_norm": 0.6108682155609131, "learning_rate": 7.239358364334728e-05, "loss": 2.3546, "step": 38870 }, { "epoch": 1.3489209749195652e-13, "grad_norm": 0.5957403779029846, "learning_rate": 7.237126139766484e-05, "loss": 2.3082, "step": 38880 }, { "epoch": 1.3492679196147606e-13, "grad_norm": 0.6349421739578247, "learning_rate": 7.234893857931105e-05, "loss": 2.3529, "step": 38890 }, { "epoch": 1.349614864309956e-13, "grad_norm": 0.6068275570869446, "learning_rate": 7.232661519181612e-05, "loss": 2.3592, "step": 38900 }, { "epoch": 1.3499618090051513e-13, "grad_norm": 0.6126410365104675, "learning_rate": 7.230429123871029e-05, "loss": 2.3418, "step": 38910 }, { "epoch": 1.3503087537003466e-13, "grad_norm": 0.5712219476699829, "learning_rate": 7.228196672352389e-05, "loss": 2.2851, "step": 38920 }, { "epoch": 1.350655698395542e-13, "grad_norm": 0.5937545299530029, "learning_rate": 7.225964164978735e-05, "loss": 2.3132, "step": 38930 }, { "epoch": 1.3510026430907374e-13, "grad_norm": 0.6422864198684692, "learning_rate": 7.22373160210312e-05, "loss": 2.3413, "step": 38940 }, { "epoch": 1.3513495877859327e-13, "grad_norm": 0.6018370389938354, "learning_rate": 7.221498984078603e-05, "loss": 2.3208, "step": 38950 }, { "epoch": 1.351696532481128e-13, "grad_norm": 0.6555159687995911, "learning_rate": 7.219266311258255e-05, "loss": 2.356, "step": 38960 }, { "epoch": 1.3520434771763234e-13, "grad_norm": 0.613046407699585, "learning_rate": 7.217033583995156e-05, "loss": 2.3043, "step": 38970 }, { "epoch": 1.3523904218715188e-13, "grad_norm": 0.5909930467605591, "learning_rate": 7.214800802642386e-05, "loss": 2.3264, "step": 38980 }, { "epoch": 1.3527373665667142e-13, "grad_norm": 0.5715978145599365, "learning_rate": 7.212567967553045e-05, "loss": 2.3266, "step": 38990 }, { "epoch": 1.3530843112619095e-13, "grad_norm": 0.5978307723999023, "learning_rate": 7.210335079080237e-05, "loss": 2.3243, "step": 39000 }, { "epoch": 1.353431255957105e-13, "grad_norm": 0.5726698637008667, "learning_rate": 7.208102137577073e-05, "loss": 2.2846, "step": 39010 }, { "epoch": 1.3537782006523003e-13, "grad_norm": 0.5960017442703247, "learning_rate": 7.205869143396672e-05, "loss": 2.3647, "step": 39020 }, { "epoch": 1.3541251453474956e-13, "grad_norm": 0.6045165657997131, "learning_rate": 7.203636096892165e-05, "loss": 2.3538, "step": 39030 }, { "epoch": 1.354472090042691e-13, "grad_norm": 0.5921467542648315, "learning_rate": 7.201402998416686e-05, "loss": 2.3672, "step": 39040 }, { "epoch": 1.3548190347378863e-13, "grad_norm": 0.5936181545257568, "learning_rate": 7.199169848323385e-05, "loss": 2.3555, "step": 39050 }, { "epoch": 1.3551659794330817e-13, "grad_norm": 0.5788806676864624, "learning_rate": 7.196936646965413e-05, "loss": 2.3375, "step": 39060 }, { "epoch": 1.355512924128277e-13, "grad_norm": 0.5990017652511597, "learning_rate": 7.194703394695932e-05, "loss": 2.3378, "step": 39070 }, { "epoch": 1.3558598688234724e-13, "grad_norm": 0.5808977484703064, "learning_rate": 7.19247009186811e-05, "loss": 2.3339, "step": 39080 }, { "epoch": 1.3562068135186678e-13, "grad_norm": 0.6014850735664368, "learning_rate": 7.190236738835124e-05, "loss": 2.3257, "step": 39090 }, { "epoch": 1.3565537582138631e-13, "grad_norm": 0.583827793598175, "learning_rate": 7.188003335950165e-05, "loss": 2.3129, "step": 39100 }, { "epoch": 1.3569007029090585e-13, "grad_norm": 0.6033966541290283, "learning_rate": 7.185769883566422e-05, "loss": 2.3192, "step": 39110 }, { "epoch": 1.3572476476042539e-13, "grad_norm": 0.6124194264411926, "learning_rate": 7.183536382037094e-05, "loss": 2.3476, "step": 39120 }, { "epoch": 1.3575945922994492e-13, "grad_norm": 0.6067337393760681, "learning_rate": 7.181302831715396e-05, "loss": 2.3497, "step": 39130 }, { "epoch": 1.3579415369946446e-13, "grad_norm": 0.6196133494377136, "learning_rate": 7.179069232954541e-05, "loss": 2.277, "step": 39140 }, { "epoch": 1.35828848168984e-13, "grad_norm": 0.6728718876838684, "learning_rate": 7.176835586107753e-05, "loss": 2.3583, "step": 39150 }, { "epoch": 1.3586354263850353e-13, "grad_norm": 0.5796716809272766, "learning_rate": 7.174601891528267e-05, "loss": 2.3023, "step": 39160 }, { "epoch": 1.3589823710802307e-13, "grad_norm": 0.603195071220398, "learning_rate": 7.172368149569319e-05, "loss": 2.3193, "step": 39170 }, { "epoch": 1.359329315775426e-13, "grad_norm": 0.5923677086830139, "learning_rate": 7.17013436058416e-05, "loss": 2.3294, "step": 39180 }, { "epoch": 1.3596762604706214e-13, "grad_norm": 0.6084728837013245, "learning_rate": 7.16790052492604e-05, "loss": 2.2923, "step": 39190 }, { "epoch": 1.3600232051658168e-13, "grad_norm": 0.6256945133209229, "learning_rate": 7.165666642948226e-05, "loss": 2.3361, "step": 39200 }, { "epoch": 1.360370149861012e-13, "grad_norm": 0.5735234618186951, "learning_rate": 7.163432715003983e-05, "loss": 2.3297, "step": 39210 }, { "epoch": 1.3607170945562075e-13, "grad_norm": 0.592930257320404, "learning_rate": 7.16119874144659e-05, "loss": 2.3247, "step": 39220 }, { "epoch": 1.3610640392514028e-13, "grad_norm": 0.6169461011886597, "learning_rate": 7.158964722629328e-05, "loss": 2.3394, "step": 39230 }, { "epoch": 1.3614109839465982e-13, "grad_norm": 0.5700929164886475, "learning_rate": 7.156730658905491e-05, "loss": 2.3098, "step": 39240 }, { "epoch": 1.3617579286417936e-13, "grad_norm": 0.5852692723274231, "learning_rate": 7.154496550628375e-05, "loss": 2.3423, "step": 39250 }, { "epoch": 1.362104873336989e-13, "grad_norm": 0.6309505701065063, "learning_rate": 7.152262398151285e-05, "loss": 2.3386, "step": 39260 }, { "epoch": 1.3624518180321843e-13, "grad_norm": 0.5778313279151917, "learning_rate": 7.150028201827534e-05, "loss": 2.3747, "step": 39270 }, { "epoch": 1.3627987627273797e-13, "grad_norm": 0.586098313331604, "learning_rate": 7.147793962010439e-05, "loss": 2.3433, "step": 39280 }, { "epoch": 1.363145707422575e-13, "grad_norm": 0.5869436860084534, "learning_rate": 7.145559679053329e-05, "loss": 2.3532, "step": 39290 }, { "epoch": 1.3634926521177704e-13, "grad_norm": 0.5925723314285278, "learning_rate": 7.143325353309535e-05, "loss": 2.3532, "step": 39300 }, { "epoch": 1.3638395968129657e-13, "grad_norm": 0.6502321362495422, "learning_rate": 7.141090985132394e-05, "loss": 2.3477, "step": 39310 }, { "epoch": 1.364186541508161e-13, "grad_norm": 0.5855398178100586, "learning_rate": 7.138856574875256e-05, "loss": 2.3387, "step": 39320 }, { "epoch": 1.3645334862033565e-13, "grad_norm": 0.6245114207267761, "learning_rate": 7.136622122891471e-05, "loss": 2.3223, "step": 39330 }, { "epoch": 1.3648804308985518e-13, "grad_norm": 0.5896978378295898, "learning_rate": 7.1343876295344e-05, "loss": 2.3267, "step": 39340 }, { "epoch": 1.3652273755937472e-13, "grad_norm": 0.5947259068489075, "learning_rate": 7.132153095157408e-05, "loss": 2.3291, "step": 39350 }, { "epoch": 1.3655743202889425e-13, "grad_norm": 0.5851527452468872, "learning_rate": 7.129918520113867e-05, "loss": 2.31, "step": 39360 }, { "epoch": 1.365921264984138e-13, "grad_norm": 0.5945432782173157, "learning_rate": 7.127683904757157e-05, "loss": 2.3366, "step": 39370 }, { "epoch": 1.3662682096793333e-13, "grad_norm": 0.6263712644577026, "learning_rate": 7.125449249440663e-05, "loss": 2.3342, "step": 39380 }, { "epoch": 1.3666151543745286e-13, "grad_norm": 0.5966119170188904, "learning_rate": 7.123214554517774e-05, "loss": 2.2956, "step": 39390 }, { "epoch": 1.366962099069724e-13, "grad_norm": 0.5942675471305847, "learning_rate": 7.120979820341893e-05, "loss": 2.3482, "step": 39400 }, { "epoch": 1.3673090437649194e-13, "grad_norm": 0.6364938020706177, "learning_rate": 7.118745047266417e-05, "loss": 2.3247, "step": 39410 }, { "epoch": 1.3676559884601147e-13, "grad_norm": 0.6292659640312195, "learning_rate": 7.116510235644765e-05, "loss": 2.3456, "step": 39420 }, { "epoch": 1.36800293315531e-13, "grad_norm": 0.6101124882698059, "learning_rate": 7.114275385830345e-05, "loss": 2.3306, "step": 39430 }, { "epoch": 1.3683498778505054e-13, "grad_norm": 0.5919538140296936, "learning_rate": 7.112040498176582e-05, "loss": 2.3393, "step": 39440 }, { "epoch": 1.3686968225457008e-13, "grad_norm": 0.603076159954071, "learning_rate": 7.109805573036905e-05, "loss": 2.3117, "step": 39450 }, { "epoch": 1.3690437672408962e-13, "grad_norm": 0.5949022769927979, "learning_rate": 7.10757061076475e-05, "loss": 2.3356, "step": 39460 }, { "epoch": 1.3693907119360915e-13, "grad_norm": 0.5754596590995789, "learning_rate": 7.105335611713551e-05, "loss": 2.331, "step": 39470 }, { "epoch": 1.369737656631287e-13, "grad_norm": 0.5777242183685303, "learning_rate": 7.103100576236761e-05, "loss": 2.3576, "step": 39480 }, { "epoch": 1.3700846013264822e-13, "grad_norm": 0.6034757494926453, "learning_rate": 7.100865504687827e-05, "loss": 2.3223, "step": 39490 }, { "epoch": 1.3704315460216776e-13, "grad_norm": 0.5884444713592529, "learning_rate": 7.098630397420208e-05, "loss": 2.3345, "step": 39500 }, { "epoch": 1.370778490716873e-13, "grad_norm": 0.5998156070709229, "learning_rate": 7.096395254787365e-05, "loss": 2.3285, "step": 39510 }, { "epoch": 1.3711254354120683e-13, "grad_norm": 0.5992612838745117, "learning_rate": 7.094160077142769e-05, "loss": 2.3194, "step": 39520 }, { "epoch": 1.3714723801072637e-13, "grad_norm": 0.5944038033485413, "learning_rate": 7.091924864839895e-05, "loss": 2.3562, "step": 39530 }, { "epoch": 1.371819324802459e-13, "grad_norm": 0.5947644710540771, "learning_rate": 7.089689618232222e-05, "loss": 2.2984, "step": 39540 }, { "epoch": 1.3721662694976544e-13, "grad_norm": 0.6079123616218567, "learning_rate": 7.087454337673231e-05, "loss": 2.3456, "step": 39550 }, { "epoch": 1.3725132141928498e-13, "grad_norm": 0.6299799084663391, "learning_rate": 7.085219023516416e-05, "loss": 2.3339, "step": 39560 }, { "epoch": 1.3728601588880451e-13, "grad_norm": 0.5909645557403564, "learning_rate": 7.082983676115273e-05, "loss": 2.3478, "step": 39570 }, { "epoch": 1.3732071035832405e-13, "grad_norm": 0.6083903908729553, "learning_rate": 7.080748295823303e-05, "loss": 2.3344, "step": 39580 }, { "epoch": 1.3735540482784359e-13, "grad_norm": 0.6105157732963562, "learning_rate": 7.078512882994008e-05, "loss": 2.334, "step": 39590 }, { "epoch": 1.3739009929736312e-13, "grad_norm": 0.5814316272735596, "learning_rate": 7.076277437980904e-05, "loss": 2.3786, "step": 39600 }, { "epoch": 1.3742479376688266e-13, "grad_norm": 0.5878698229789734, "learning_rate": 7.074041961137504e-05, "loss": 2.3453, "step": 39610 }, { "epoch": 1.374594882364022e-13, "grad_norm": 0.5934957265853882, "learning_rate": 7.071806452817336e-05, "loss": 2.3097, "step": 39620 }, { "epoch": 1.3749418270592173e-13, "grad_norm": 0.6077732443809509, "learning_rate": 7.069570913373916e-05, "loss": 2.3406, "step": 39630 }, { "epoch": 1.3752887717544127e-13, "grad_norm": 0.6104267239570618, "learning_rate": 7.067335343160784e-05, "loss": 2.3319, "step": 39640 }, { "epoch": 1.375635716449608e-13, "grad_norm": 0.6401292085647583, "learning_rate": 7.065099742531473e-05, "loss": 2.3198, "step": 39650 }, { "epoch": 1.3759826611448034e-13, "grad_norm": 0.6074186563491821, "learning_rate": 7.062864111839525e-05, "loss": 2.3476, "step": 39660 }, { "epoch": 1.3763296058399987e-13, "grad_norm": 0.6288160681724548, "learning_rate": 7.060628451438481e-05, "loss": 2.3332, "step": 39670 }, { "epoch": 1.376676550535194e-13, "grad_norm": 0.6080392599105835, "learning_rate": 7.058392761681898e-05, "loss": 2.3671, "step": 39680 }, { "epoch": 1.3770234952303895e-13, "grad_norm": 0.5967344045639038, "learning_rate": 7.056157042923329e-05, "loss": 2.35, "step": 39690 }, { "epoch": 1.3773704399255848e-13, "grad_norm": 0.5775385499000549, "learning_rate": 7.053921295516332e-05, "loss": 2.3086, "step": 39700 }, { "epoch": 1.3777173846207802e-13, "grad_norm": 0.5957332253456116, "learning_rate": 7.051685519814471e-05, "loss": 2.3352, "step": 39710 }, { "epoch": 1.3780643293159756e-13, "grad_norm": 0.602353572845459, "learning_rate": 7.04944971617132e-05, "loss": 2.3167, "step": 39720 }, { "epoch": 1.378411274011171e-13, "grad_norm": 0.6076748967170715, "learning_rate": 7.047213884940444e-05, "loss": 2.3353, "step": 39730 }, { "epoch": 1.3787582187063663e-13, "grad_norm": 0.585232675075531, "learning_rate": 7.044978026475426e-05, "loss": 2.3105, "step": 39740 }, { "epoch": 1.3791051634015616e-13, "grad_norm": 0.5983977317810059, "learning_rate": 7.042742141129846e-05, "loss": 2.3502, "step": 39750 }, { "epoch": 1.379452108096757e-13, "grad_norm": 0.6394805908203125, "learning_rate": 7.040506229257292e-05, "loss": 2.3118, "step": 39760 }, { "epoch": 1.3797990527919524e-13, "grad_norm": 0.5775141716003418, "learning_rate": 7.03827029121135e-05, "loss": 2.3101, "step": 39770 }, { "epoch": 1.3801459974871477e-13, "grad_norm": 0.5889573693275452, "learning_rate": 7.03603432734562e-05, "loss": 2.2995, "step": 39780 }, { "epoch": 1.380492942182343e-13, "grad_norm": 0.5972214341163635, "learning_rate": 7.033798338013694e-05, "loss": 2.3138, "step": 39790 }, { "epoch": 1.3808398868775384e-13, "grad_norm": 0.6107890009880066, "learning_rate": 7.03156232356918e-05, "loss": 2.3139, "step": 39800 }, { "epoch": 1.3811868315727338e-13, "grad_norm": 0.5811108946800232, "learning_rate": 7.02932628436568e-05, "loss": 2.3494, "step": 39810 }, { "epoch": 1.3815337762679292e-13, "grad_norm": 0.5937615036964417, "learning_rate": 7.02709022075681e-05, "loss": 2.3338, "step": 39820 }, { "epoch": 1.3818807209631245e-13, "grad_norm": 0.5811803340911865, "learning_rate": 7.024854133096178e-05, "loss": 2.3492, "step": 39830 }, { "epoch": 1.38222766565832e-13, "grad_norm": 0.562246561050415, "learning_rate": 7.022618021737404e-05, "loss": 2.3545, "step": 39840 }, { "epoch": 1.3825746103535153e-13, "grad_norm": 0.5986263751983643, "learning_rate": 7.020381887034111e-05, "loss": 2.3265, "step": 39850 }, { "epoch": 1.3829215550487106e-13, "grad_norm": 0.627097487449646, "learning_rate": 7.018145729339927e-05, "loss": 2.2955, "step": 39860 }, { "epoch": 1.383268499743906e-13, "grad_norm": 0.6116889715194702, "learning_rate": 7.015909549008477e-05, "loss": 2.3449, "step": 39870 }, { "epoch": 1.3836154444391013e-13, "grad_norm": 0.5716720223426819, "learning_rate": 7.013673346393394e-05, "loss": 2.3262, "step": 39880 }, { "epoch": 1.3839623891342967e-13, "grad_norm": 0.566609799861908, "learning_rate": 7.011437121848314e-05, "loss": 2.3355, "step": 39890 }, { "epoch": 1.384309333829492e-13, "grad_norm": 0.5917127132415771, "learning_rate": 7.00920087572688e-05, "loss": 2.3195, "step": 39900 }, { "epoch": 1.3846562785246874e-13, "grad_norm": 0.5991883873939514, "learning_rate": 7.00696460838273e-05, "loss": 2.2991, "step": 39910 }, { "epoch": 1.3850032232198828e-13, "grad_norm": 0.5951396822929382, "learning_rate": 7.004728320169516e-05, "loss": 2.2831, "step": 39920 }, { "epoch": 1.3853501679150781e-13, "grad_norm": 0.5891526341438293, "learning_rate": 7.002492011440884e-05, "loss": 2.3269, "step": 39930 }, { "epoch": 1.3856971126102735e-13, "grad_norm": 0.6029372811317444, "learning_rate": 7.000255682550487e-05, "loss": 2.3395, "step": 39940 }, { "epoch": 1.3860440573054689e-13, "grad_norm": 0.6291112899780273, "learning_rate": 6.998019333851981e-05, "loss": 2.3778, "step": 39950 }, { "epoch": 1.3863910020006642e-13, "grad_norm": 0.5820087194442749, "learning_rate": 6.995782965699028e-05, "loss": 2.3788, "step": 39960 }, { "epoch": 1.3867379466958596e-13, "grad_norm": 0.6252443194389343, "learning_rate": 6.993546578445289e-05, "loss": 2.3261, "step": 39970 }, { "epoch": 1.387084891391055e-13, "grad_norm": 0.6255847811698914, "learning_rate": 6.991310172444429e-05, "loss": 2.3171, "step": 39980 }, { "epoch": 1.3874318360862503e-13, "grad_norm": 0.6027980446815491, "learning_rate": 6.989073748050114e-05, "loss": 2.3213, "step": 39990 }, { "epoch": 1.3877787807814457e-13, "grad_norm": 0.5994889140129089, "learning_rate": 6.98683730561602e-05, "loss": 2.3229, "step": 40000 } ], "logging_steps": 10, "max_steps": 80000, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 5000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 2.112438593912832e+18, "train_batch_size": 8, "trial_name": null, "trial_params": null }